Examination of the failure of 3D woven composites

Examination of the failure of 3D woven composites

Available online at www.sciencedirect.com Composites: Part A 39 (2008) 273–283 www.elsevier.com/locate/compositesa Examination of the failure of 3D ...

1MB Sizes 0 Downloads 114 Views

Available online at www.sciencedirect.com

Composites: Part A 39 (2008) 273–283 www.elsevier.com/locate/compositesa

Examination of the failure of 3D woven composites J.P. Quinn, A.T. McIlhagger *, R. McIlhagger Engineering Composites Research Centre, School of Electrical and Mechanical Engineering, University of Ulster, Shore Road, Newtownabbey, Co. Antrim, Northern Ireland, United Kingdom Received 17 January 2005; received in revised form 22 October 2007; accepted 23 October 2007

Abstract The manufacture of advanced 3D woven textiles as reinforcements has been possible for number of years using standard textile weaving machinery with minimal modifications. Orthogonal structures represent one of the more straightforward structures in terms of tow path complexity, yet also represent a structure where the advantages of a high proportion of straight tows bound together by a binder tow result in a composite with high performance and additionally reduced sensitivity to interlaminar shear. The results presented in this paper are from an examination of the failure of 3D woven composites in tension and an examination of the strain distribution using electronic speckle pattern interferometery (ESPI). The study reports the performance of 3D woven specimens, loaded in tension and shows increased strain in the area where the binder tow enters the subsurface layers of the fabric structure to bind the layers together. The ESPI technique is used to illustrate how the magnitude of the strains induced in the test specimen under load, are distributed across the surface of the specimen.  2007 Elsevier Ltd. All rights reserved. Keywords: A. Carbon fibre; A. Fabrics/textiles; B. Mechanical properties; B. Physical properties; E. Resin transfer moulding

1. Introduction Composite materials have shown significant advantages over their metallic counter-parts, in certain applications, in terms of their high strength/weight ratio, high stiffness/ weight ratio and the ease of fabrication of complex shaped components [1]. 3D woven preforms, when matched with liquid moulding processes such as resin transfer moulding (RTM), can offer significant advantages in terms of production over the traditional prepreg autoclave-manufacturing route. In addition to this, the design of 3D woven structures facilitates the placement of specific amounts of fibre in each of the principal directions according to the ultimate application and desired mechanical performance of the composite component. The flexibility of design that can be incorporated into the production of 3D woven textile composites puts the onus on the designer to develop the textile structure to accept the component loads as effi*

Corresponding author. Tel.: +44 2890366670. E-mail address: [email protected] (A.T. McIlhagger).

1359-835X/$ - see front matter  2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.compositesa.2007.10.012

ciently as possible. As a result of this, there must be an understanding of the effects of design parameters such as binder density, etc. not only on the final mechanical performance of the composite but also on the processing route selection and conditions. Generally, it is recognised that the advantages of 3D woven reinforcements lie in the ability of the structure to sustain large strains to failure in compression and compression after impact. However, reported values show that 3D woven reinforcements also have mechanical advantages in terms of tensile strength [2], flexural strength [3] and interlaminar shear strength [4] over their 2D woven counterparts. With research showing increased mechanical properties of 3D woven composites in a number of areas, understanding of the failure of these advanced composites is required. It is envisaged that comprehension of the failure mechanisms of these materials will accelerate the design of improved structures, since those micro-structural features that degrade performance can potentially be eliminated through design, while those that enhance performance can be incorporated [5].

274

J.P. Quinn et al. / Composites: Part A 39 (2008) 273–283

2. Structural loading of composite materials Tensile testing has traditionally been used in the characterisation of metals where the properties are generally isotropic. Loading of components in any application e.g. aerospace, automotive, sports goods, is generally never in tension alone and almost always involves a form of more complex loading where the component will experience a combination of forces, including flexure, compression tension and shear. Characterisation of the performance of 3D woven composites under specific static loading conditions is much more difficult as the material is not isotropic. Generally a range of tests is conducted on the composite to assess performance for specific applications and the performance of 3D woven composites have been examined by a number of authors [6–10]. Cox [6] in particular has completed extensive work on the failure mechanisms associated with 3D woven composites, in this case mostly angle interlock fabrics. The author concluded that a kink band formation occurs prior to the failure of the samples and that the main reason for failure in 3D wovens lies in the geometrical irregularity associated with the complex architectures of weave. Callus and Leong [8,9] have concentrated more on the examination of orthogonal structures. The authors have indicated the formation of cracks at low strain in the areas of resin concentration. These areas exist in the region of the structure where the binder enters and exits at the surface. They conclude that the tensile strength of the 3D woven structures was really independent of the weave style i.e. orthogonal, angle interlock or layer to layer, assuming the results are normalised with respect to fibre volume fraction in the direction of loading, more importantly the authors noted that the potential improvement in properties will be as a consequence of controlled tension during the manufacturing stages. The overall main conclusions, drawn from the work on tensile failure of composites, follow the argument that the failure of 3D woven reinforcements is based around matrix cracking. This takes place in the region of the binder tow that passes from top to bottom of an orthogonal weave structure (see Fig. 1). In addition to this straightening of the stressed tows under the uniaxial tensile load is reported to initiate failure. In almost all cases of reported results [6–10] the

Fig. 1. Basic interlinked 3D woven structure.

work has concentrated on a number of specific fabric architectures and reported values for strength and elastic modulus which will be unique to the studied fabrics. It is therefore difficult to compare the results from previous studies in the area to the present results. 3. Advantageous properties of 3D woven composites 3D woven reinforcements offer a distinct advantage over other composites in that the amount of fibre in each direction can be controlled, tailoring the modulus and in part, strength of the final component. Hence, assuming a direct relationship between the volume of fibre under load and the ultimate strength, the designer can tailor the fabric structure to the loading environment. The introduction of fibres in each of the three principal directions also allows the design of many more types of structure. On 3D wovens Vandeurzen et al. [11] state that ‘‘the fabric architecture can be altered to meet the specific mechanical properties required by the composite application. This is achieved by changing the weave pattern, fibre type, tow type, tow count, etc. The number of possible architectures is unlimited’’. 4. Textile design 2D weave fabrics used for lamination have nominally 50% fibre in the warp (X) direction and 50% in the weft (Y). This, when combined with an appropriate resin system and laminate configuration, results in a high strength composite component; however, the laminate construction makes the composite weak under (interlaminar) shear loads. The 3D textile designs used for the work are based on the concept of having high percentages of fibre in the warp and weft directions with a small percentage of the warp tows used to bind the layers together and increase the interlaminar shear strength (ILSS) of the final composite. Fig. 1 shows a cross-sectional representation of a basic interlinked 3D woven structure. 5. Methodology The textile preform design used in this study design is a 3D representation of the 2D satin fabric structure. Satin fabrics have a specific design in which the tows pass under and over one another in set amounts according to the harness number of the design. The harness number used in the designation (typically 4, 6 and 8) is the total number of tows crossed and passed under, before the pattern repeats. The 2D satin weaves are very flat, have good wet out and a high degree of drape (the ability to form around a complex curvature). The low crimp gives good mechanical properties. Satin weaves allow tows to be woven in the closest proximity and can produce fabrics with a close ‘tight’ weave. The 3D representation of the satin fabric used for this work has seven layers of warp and six layers of weft. Two of the warp tows are used as binders. In addition to

J.P. Quinn et al. / Composites: Part A 39 (2008) 273–283

Binder

275

[13]. Ten specimens from each of the composites were cut parallel to the warp axis. The specimen dimensions were length 300 mm · width 30 mm · thickness 2.8 mm. The specimens were held within the tensile test apparatus with specially treated grips to promote failure of the sample within the gauge length area. 7.1. Electronic speckle pattern interferometery (ESPI)

Weft

Warp

Repeat

Fig. 2. SAT 4 fabric.

this basic design, altering the harness number in each case creates the family of fabrics. The SAT 4 shown in Fig. 2 fabric has two warp binder tows that pass over three weft yarns before passing through the thickness of the fabric and repeating the pattern. 6. Fabric manufacture and composite processing A conventional Jacquard Dataweave power loom with a flexible rapier method of weft insertion was drawn in with warp ends fed from a 600-bobbin creel. 3D structures consisting of seven warp and six weft layers were manufactured with continuous filament carbon fibre tows (12k). The warp set measured 2 ends/cm/layer and 2 weft picks/cm/layer. A beat-up reed with 1 dent/10 mm pitch spaced the yarns. Preform dimensions were restricted by the Jacquard harness capacity of 24 ends/cm and weaving size of 1000 mm length · 500 mm width. Each perform was impregnated using epoxy resin matrix. The matrix was a two-part epoxy system supplied by Ciba Polymers; LY564 resin mixed with HY2954 hardener. The resin was mixed by hand and degassed in a vacuum chamber for 1 h at 30 C to remove moisture and air suspended in the resin. Reinforcement was positioned in an aluminium tool. The tool was held under vacuum at 75 C and the degassed resin injected by peripheral gating arrangement, into the tool at a supply pressure of 750 mbar absolute. When filled, the tool temperature was raised to 100 C at approximately 0.64 C/min and the composite cured for 99 min before the composite part was cooled and demoulded. After each 3D woven structure was completed the fabrics were trimmed to dimensions of 420 · 350 mm and processed by an RTM process to manufacture a composite plaque.

ESPI is a non-contact 3D displacement measurement system. Speckle interferometry uses a high resolution CCD-camera to take the image of the object under investigation while it is illuminated with laser light. The resulting speckle pattern is stored as reference image on a personal computer. When the object is deforming e.g. under tensile load the speckle pattern will change and the comparison with the reference pattern will show correlation fringes representing the displacement of the object [16]. The obtained fringes then correspond to inplane displacements. The derivation of the displacement values leads to the strain distribution at each position of the measuring area. In order to investigate the strain experienced during the tensile testing of the 3D woven samples a full field strain measurement technique was selected. Samples were loaded in a universal test machine (Zwick Z100) that incorporated a mounting plate to site the ESPI camera. Fig. 3 shows a schematic of the experimental set-up. Samples mounted in the universal tester were loaded in stages up to a maximum of 7 kN representing an engineering strain of approximately 0.29% (note the graph has peaks over the 0.5% strain value). At each loading stage an image was recorded by the ESPI apparatus. At stage one with the load at zero an image taken will be the baseline image that all other images are referenced to. Subsequently at the second stage of loading the image recorded will show some deformation of the sample, and when it is compared to the reference image captured in stage one a patterned surface plot of the strain (extension/original length) can be produced. This process was repeated each time the sample was loaded to

Universal Testing Machine

Moving Crosshead

Sample Camera

7. Mechanical testing The Royal Aerospace Establishment CRAG standards [12] were used for this research. These standards have been used in previous research at the Engineering Composites Research Centre and so provide comparative data. The tensile testing was performed to CRAG standard 302

Base

Mounting Platform

Fig. 3. ESPI testing apparatus schematic.

276

J.P. Quinn et al. / Composites: Part A 39 (2008) 273–283

another level e.g. 1500–3000, etc. The aim of this approach was to try to isolate areas of increased strain under the relatively low loading conditions. Electronic speckle pattern interferometery (ESPI) is a non-contact laser optic measurement device that allows the measurement of strain over relatively large areas. The ability of the system to measure strain over large areas, 256 · 256 points, compared to strain gauges (1 point) makes the system ideal for 3D woven composites, where the repeating pattern of the design can be greater in size than the traditional strain gauges used for strain measurement in many applications. 8. Results Tensile tests for the determination of tensile strength and tensile modulus were completed in accordance with CRAG Method 302 [12]. Analysis of the results from testing 10 specimens of each weave structure yielded results for tensile strength and modulus, as well as strain to failure. The specimens were all tested to destruction and, in each case, a proportional relationship between the stress and strain was observed, with no visible plastic deformation occurring during the test. The tests were stopped at the maximum stress value so that the failed specimens could be examined for assessment of the failure in each case. Fig. 4 shows a stress/strain graph for each of the three fabric structures and a plain weave 2D laminate as a baseline fabric. From examination of the graph the samples behave in a linear elastic manner. The strain to failure values for the plain weave sample is significantly higher than the 3D woven samples. It is hypothesised that the failure mechanisms developed in the 3D structure by the presence of a binding tow have a considerable effect on strain to failure. 9. Tensile modulus One of the main areas that differentiates’ the fabric architectures from one another is the binder region. The SAT 4 design has a higher binder density than the other samples. On a visual basis this difference in binder density

Fig. 5. SAT 4 fabric structure showing float length.

manifests itself as a distinct difference in float length on the fabric surface. The SAT 8 fabric has a much longer float length than the SAT 4 and SAT 6 fabrics. Figs. 5–7 show the face of the SAT 4, SAT 6 and SAT 8 fabrics used in the study. The float lengths are marked on each. The float length in each of the structures is a direct consequence of the binder density. A long float length in the structure imparts a larger percentage of nominally straight fibre in the warp direction and a reduced amount of binder tow crimp. When the fabrics are loaded in this direction, as in a tensile test, the expected response would be that the more fibre in the direction of loading the higher the sustainable stress. The tensile modulus of the samples tested in this work was deduced from the slope of the stress/strain curve (Fig. 4). Results of the tensile modulus in the warp direction for each of the fabrics studied are shown in Table 1. For comparative purposes, a 200 g/m2 12-layer 2D carbon fibre plain weave fabric processed the same as the 3D woven samples is also shown. In addition to the calculation of the tensile modulus for each of the samples tested the fibre volume fraction was calculated according to the density method [14].

600

Stress (MPa)

500 400

SAT 4 SAT 6

300

SAT 8 200

Plain Weave(2D)

100 0 0

0.005

0.01

Fig. 4. Stress vs. strain for tensile test of fabric structures.

J.P. Quinn et al. / Composites: Part A 39 (2008) 273–283

Fig. 6. SAT 6 fabric structure showing float length.

277

30–40% region. Higher performance composites have a fibre content generally over 55%. The fibre volume fraction must be taken into account when comparing tensile results between the 3D weave structures and a plain weave sample. Therefore, if the results are normalised with respect to fibre volume fraction a more accurate comparison of properties in the tested samples is achieved. Table 2 shows the results normalised to a fibre volume fraction of 55%. The normalised results show the 3D woven fabric structures have modulus values comparable to that for a 12layer plain weave RTM processed composite. The lower values of tensile modulus in the 3D woven composite can be explained by the fabric architecture. The amount of crimp in the warp and weft stuffers is quite low (1%), however, the very high (19% max) levels of crimp in the binder tows (note, the high crimp factor is realised by the path of the binder from top to bottom of the fabric structure) of the 3D textiles adversely affects the tensile modulus by reducing the efficiency of loading in the fibre direction. In addition to this it is hypothesised that tow misalignment plays a part in the lower modulus values of the 3D structures. 3D woven reinforcements are a complex architecture of tow paths that can be rearranged in numerous ways, under compression. This movement of the tows introduces misalignment and hence there is a reduction in the strength of the composite. Examination of the 3D woven composites exclusively, shows that, as expected, the decrease in binder density from SAT 4 to SAT 8 and hence increase in the amount of ‘‘straight’’ tows within the warp sections of the structure resulted in an increase, albeit by a small percentage, in the modulus of the composite. 10. Tensile strength

Fig. 7. SAT 8 fabric structure showing float length.

The results show the stiffness of the plain weave sample to be higher than the 3D woven samples. The strength of any composite material is greatly influenced by the ratio of fibre to resin in the component. Typically, hand lay-up composite components have fibre volume fractions in the Table 1 Measured values of modulus for 3D woven and plain weave composite specimens Fabric structure

Tensile modulus (GPa)

Standard deviation (GPa)

CV (%)

Fibre volume fraction (%) (CV)

SAT 4 SAT 6 SAT 8 Plain weave

43.21 45.57 47.30 56.25

2.55 1.53 2.15 1.91

5.90 3.37 4.55 3.39

47.3(4.04) 48.1(5.48) 49.3(3.98) 53.9(4.99)

The tensile strength of a composite material is the maximum force the sample will withstand under a uniaxial load. Compromises in tensile strength of composite material are introduced when the tows under load are not unidirectional e.g. chopped strand mat. Fabric preforms, whether 2D or 3D have reduced tensile strength when compared to a unidirectional material where all the fibre lies in the direction of the applied load and have minimal or no crimp. The undulations caused by tows passing over and under one another cause the fabric to become crimped. Theoretically, 3D weaving offers an opportunity to arrange a number of tows in an uncrimped fashion in the warp and weft direction with the binder tows holding the structure together. The straight tows are the warp and weft stuffer tows in the 3D structure. However, as mentioned previously, in reality the effects of layer compaction, complexity of the fabric architecture and tension of the binder tows all induce some level of crimp into the structure. The calculated tensile strength results for the fabric structures used for this work are presented in Table 3. The tensile strength of the composite material was calculated as per CRAG Method 302. From Table 3, it is

278

J.P. Quinn et al. / Composites: Part A 39 (2008) 273–283

Table 2 Table of values of tensile modulus normalised to 55% fibre volume fraction Fabric structure

Tensile modulus (GPa) normalised

SAT 4 SAT 6 SAT 8 Plain weave

50.56 52.22 53.10 57

Table 3 Measured values of tensile strength for 3D woven and plain weave composites Fabric structure

Tensile strength (MPa)

Standard deviation (MPa)

CV (%)

Fibre volume fraction (%) (CV)

SAT 4 SAT 6 SAT 8 Plain weave

319.0 344.9 380.8 648

27.7 34.5 16.2 5.7

8.7 10 4.2 3.8

47.3(4.04) 48.1(5.48) 49.3(3.98) 53.9(4.99)

observed that the results follow a similar trend to the modulus results, the tensile strength of the 3D woven composite samples increased as the amount of binder in the sample was reduced. This can be accounted for by the increase in the amount of aligned in the direction of the loading; hence there was more fibre to resist fracture resulting in a higher tensile strength. Comparing the 3D woven reinforcements to the plain weave samples the tensile strength is quite low at approximately 50–60% of the plain weave strength. Normalising the results with respect to fibre volume fraction, the differences in tensile strength between the 2D plain weave and the 3D woven are only reduced marginally to 66–75 % of the plain weave tensile strength. The norma-

lised maximum tensile strength results, especially in the case of the SAT 8 structure, compare favourably with the work of Tan et al. [15], who reported a maximum tensile strength of 466 MPa for a similar structure. However, it should be noted that comparison of results obtained with work by other authors is very difficult as the weave patterns studied, while similar in some cases, are not entirely the same. The mechanical properties of 3D reinforced composites are very dependent on the weave structure and manufacturing process. Therefore, it is more meaningful to compare the mechanical results with more traditional materials such as plain weave laminates, which 3D wovens have the potential to replace in advanced composite components.

11. Failure mechanisms in tension Failure of the samples tested for this work occurred in each case within the gauge length. A number of points are noteworthy from examination of the broken samples. All samples had failure of the binder tows visible on the surface of the structure. Fig. 8 shows the binder tow at the edge of a SAT 6 sample broken on the upper surface. The failure at the binder region of the structure occurred where the binder tow entered the fabric. A schematic of the idealised fabric structure is shown in Fig. 8 which shows the areas of failure for the binder. It is hypothesised that increased localised strain on this area of the structure, brought about by the presence of a resin rich area at this point, caused the failure. The binder tow passing through the structure of the fabric caused an increased amount of resin in the binder area. The binder tow pinches the weft yarns below leaving a

Fig. 8. Fracture at the binder turn and subsequent path of fracture along binder (magnification 30·).

J.P. Quinn et al. / Composites: Part A 39 (2008) 273–283

Binder failure area

Binder failure area Warp stuffer

Warp

Weft

Fig. 9. Schematic showing the typical binder path through the structure and the areas of failure.

depression at the surface of the fabric. During the mould filling process the resin filled this depression, and subsequently cured, to form the resin rich area. Fig. 9 shows a fractured sample, where the fracture has occurred at the resin rich area at the top of the binder and has then moved, as the force on the sample increased, along the direction of the binder through the structure. It is hypothesised that the presence of the resin rich area in the structure may be alleviated in some respect with careful selection of the binder tow count and type and control of the binder tension during weaving. For example, the fabric structures presented in this work were woven with 12k warp weft and binder tows; replacement of the binder tows with two tows of cross-section equivalent to the 12k tow (e.g. two 6k tows) or a single 6k tow may nest a little better and hence reduce the amount of pinching experienced by the weft stuffer tows in the fabric structure (Fig. 10). Leong et al. [9] examined binder tow path in a weft bound composite structure and it’s effects on tensile properties. It is interesting to note that the authors concluded that the binder tow path had a large influence on the tensile strength and strain to failure of the sample. One reason put forward for this was the collimation of the weft tows and the subsequent resin rich areas left as a result of the movement of the bound warp tows. This observation can also be seen in the warp bound samples used in this study, how-

279

ever, not to the same degree. Fig. 11 shows the resin rich area in a SAT 6 composite. Similar micrographs taken of the processed SAT 4 and 8 composites showed similar effects. Failure of the composite around the binder area was not limited to the binders at the edge of the specimen, rather, the binder tows across the surface of the structure failed also. This agrees with the concept that the binder tow is a stress raiser in the structure. Where the binder tow failed in the structure there was a large degree of surface fibre pullout, in cases where one binder tow failed at the main fracture site there were also tows pulled from the structure at the next point of binding. Fig. 12 shows a binder breakage on the surface of a SAT 8 fabric. In the centre of the photograph a large broken binder tow can be seen, below this the binder tow adjacent to it can be seen also to be broken. This pattern of failure was observed in all 3D woven samples fractured during testing. Previous work by Cox et al. [6] noted a large amount of warp stuffer tow pullout during tensile failure. This was found in the samples tested, however, not to the degree described in the literature. In this case the samples failed in a brittle manner with no residual strength after the maximum strain was achieved. The warp tows on the surface, separated in main fracture area rather than pulling out from the matrix, whereas the warp tows within the inner layers of the composite tended to pullout of the matrix. Fig. 13 shows the pullout of warp tows within the structure. The amount of fibre pullout in a composite structure affects the strain to failure of the structure. The average maximum strain values are shown in Table 4 for the samples tested. Values for ultimate strain are lower than some of the reported values from the work of other authors [6,9] and it is most likely that the reduction in the amount of fibre pullout during failure can account for this. However, the fabric structures used by others in the field have varying tow weights and types, this would explain the difference in

Fig. 10. Failure originating from resin rich area (magnification 30·).

280

J.P. Quinn et al. / Composites: Part A 39 (2008) 273–283

Fig. 11. Microsection showing the resin rich area around the binder tow within a 3D woven composite (magnification 30·).

Fig. 12. Failure of the binder at the next binding point adjacent to the main failure area.

results by different authors. The strain to failure of the 3D woven specimens was significantly lower than the plain weave sample. One reason for the reduced strain to failure values was the initial failure of the 3D woven structure at the binder area. This initial failure weakened the structure and hence the crack propagated across the structure, whereas the uniform plain weave structure failed in a more dramatic fashion with a more uniform distribution of strain across the sample width. 12. Examination of binder tows using ESPI As discussed previously the region of the 3D woven structure where the binder tow enters the layers of the fabric to accomplish the binding is a potential area of weak-

ness in the 3D structure. Accumulation of resin caused by the collimation of the weft tows and pinching of the binder tows as they pass through the structure have been proposed as reasons for the weakness in this region of the composite. To examine this, electronic speckle pattern interferometry (ESPI) technique was used to identify any small changes in deformation of a uniaxially loaded sample in the binder tow region and hence infer the strain at that point. The ESPI technique provides the opportunity to map the strain over a large surface area. 3D woven structures are made up of repeating patterns and hence the areas of strain concentration would be repeated over the surface of the specimen. In addition to this, unlike conventional plain, satin or twill weave 2D fabrics the repeat size for a 3D woven structure can be very large (in some cases over

J.P. Quinn et al. / Composites: Part A 39 (2008) 273–283

281

Fig. 13. Warp tow pullout within the layers of the composite.

Table 4 Test results of tensile strain for 3D woven and plain weave test specimens Fabric structure

Average maximum strain (mm/mm)

Standard deviation

CV (%)

Fibre volume fraction (%) (CV)

SAT 4 SAT 6 SAT 8 Plain weave

0.00723 0.00743 0.00791 0.01

0.00043 0.00091 0.00057 0.00042

5.9 12 7.2 4.2

47.3(4.04) 48.1(5.48) 49.3(3.98) 53.9(4.99)

a number of cm2). The more traditional method of strain measurement such as strain gauges, measure localised strain and hence, the potential size of the repeat in a 3D woven structure makes measurement of strain over the area of the repeat difficult using strain gauging. Without the application of numerous gauges, measurement of strain over a number of repeat units is not possible. Fig. 14 shows a strain map taken from a uniaxially loaded specimen of the SAT 4 fabric. Results from tests performed on three specimens from each of the weave structures have shown that as expected a concentrated area of deformation, and hence strain, exists within the 3D

Areas of increased strain

woven composite structure. The strain concentration area was found to be in the same region regardless of the fabric structure. From the strain map the areas of varying strain are shown as a change in color over the area of the strain map. Fig. 15 shows the strain map across the face of the SAT 4 sample and a photograph of the face of the SAT

Direction of Strain Measurement

Fig. 15. Strain map and photograph correlation for increased strain and binder frequency.

Binder tows

2.02 1.80 1.59 1.37

Strain(x10-3)

Strain Direction

1.16 0.95 0.73 0.52 0.30 0.09

Fig. 14. Strain map of the SAT 4 composite under uniaxial loading (2.5 kN).

282

J.P. Quinn et al. / Composites: Part A 39 (2008) 273–283

Strain Distribution in a 3D woven Composite

% Strain

2 1.5

Loading Phase Loading Phase Loading Phase Loading Phase

1 0.5

4 3 2 1

0 -26

-15 -3 9 Distance from Sample centre

21

Fig. 16. Graph of strain along a warp binder tow over the several repeats of the 3D woven SAT 4 samples.

4 sample. It is observable from the regularity of the areas of increased strain that they can be attributed to the binder entrance and exit within the fabric structure. The sample in Fig. 15 was loaded with 1500 N of force equivalent to a stress value of approximately 17 MPa, (failure occurred around 300 MPa). Magnitudes of strain for each of the specimens tested were not recorded; rather, it is more beneficial to note the variation in strain in this region. The average strain across the surface of the specimen had a very large range with a maximum value occurring at the binder tow point. Plotting a graph of the strain along a profile of the sample area indicates more clearly the variation in strain. Fig. 16 shows the graph of profiles taken from a composite sample loaded in stages of 1500 N at a time to a maximum value of 7 kN. The graph, Fig. 16, clearly shows the variation of strain over the length of the sample. A variation in strain over a woven sample is not unexpected as the structure of the weave is not totally uniform. However, the peaks in the relative strain shown in Fig. 15 occurred with the same frequency as the binder tows indicating a direct correlation between the point at which the binder tow entered the structure and an area of potential failure. 13. Conclusions The ESPI technique has highlighted the areas of increased strain within the 3D woven composite studies. Previous work [7–10] has indicated that the failure of 3D woven structures is a complex mixture of a number of variables including the resin rich areas at the binder tow and throughout the structure, however, no previous study has measured the distribution of strain caused by the resin rich areas. Further study in this area has the potential to highlight any changes weaving parameters such as binder type and binder tension would have on the strain distribution within a 3D woven composite. However, correlation with standardised techniques for strain measurement where possible would serve to increase the confidence of measurements taken using this technique. Tensile tests completed on composites manufactured via RTM using 3D woven reinforcements have shown comparable values of modulus with 2D plied lay-up test specimens. Tensile strength of

the 3D woven composite is somewhat lower then the 2D plain weave plied lay-up. It is hypothesised that the reduction in tensile strength comes about for a number of reasons:  Crimp, due to the binder tows, added to the structure by the weaving process reduced the efficiency of the structure.  Resin rich areas created by the collimation of the weft yarns, a result of uncontrolled binder tow tension [9].  Shearing of the binder tow as it entered and exited the fabric.  Large binders induced resin rich areas around the binder turn, creating areas of concentrated strain. Strain mapping using the ESPI system has shown large variations in the strain over the repeat of a test sample subjected to staged loading. The magnitude of the values generated for strain from these tests require further verification, however, the trend of results provides important information showing the characteristics of the composite material under tensile loading. It is suggested that a more in depth study of the behaviour of 3D composite specimens using the ESPI system would be beneficial in understanding the failure behaviour of samples under stress. Future development of this work would include, correlations made using traditional strain gauging and the ESPI system, this would provide more confident measurement of strain with the ESPI technique. Possible applications for the systems would also include use for verification of predicted strains from finite element models. However, such a study is outside the remit of this investigation. References [1] Quinn JP, Hill BJ, McIlhagger R. An integrated design system for the manufacture and analysis of 3-D woven preforms. Compos Pt A: Appl Sci Manuf 2001;32(7):911–4. [2] Guess TR, Reedy ED. Comparison of interlocked fabric and laminated fabric kevlar 49/epoxy composites. J Compos Technol Res 1985;7:136. [3] Adanur S, Tam CA. On machine interlocking of 3D laminate structures for composites. Compos Pt B 1997;28(B):497–506. [4] Matthews ST, Hill BJ, McIlhagger AT, McIlhagger R. Investigation into through the thickness yarns in carbon fibre epoxy composites. In:

J.P. Quinn et al. / Composites: Part A 39 (2008) 273–283

[5]

[6]

[7]

[8]

[9]

[10]

Proceedings of the sixth international conference on automated composites; 1999. p.195–201. Chou S, Chen H, Chen H. Effect of weave structure on mechanical fracture behaviour of three dimensional carbon fibre fabric reinforced epoxy resin composites. Compos Sci Technol 1992;45:23–35. Cox BN, Dadkhah MS, Morris WL, Flintoff JG. Failure mechanisms of 3D woven composites in tension, compression, and bending. Acta Metall Mater 1994;42(12):3967–84. Cox Brian N, Dadkhah Mahyar S, Morris WL. On the tensile failure of 3D woven composites. Compos Pt A: Appl Sci Manuf 1996;27(6): 447–58. Callus PJ, Mouritz AP, Bannister MK, Leong KH. Tensile properties and failure mechanisms of 3D woven GRP composites. Compos Pt A 1999;30:1277–87. Leong KH, Lee B, Herzberg I, Bannister MK. The effect of binder path on the tensile properties and failure of multilayer woven CFRP composites. Compos Sci Technol 2000;60:149–56. Chou S, Chen H, Chen H. Effect of weave structure on mechanical fracture behaviour of three dimensional carbon fibre fabric reinforced epoxy resin composites. Compos Sci Technol 1992;45:23–35.

283

[11] Vandeurzen Ph, Ivens J, Verpoest I. A three-dimensional micromechanical analysis of woven-fabric composites: I. Geometric analysis. Compos Sci Technol 1996;56(11):1303–15. [12] Curtis PT, editor. CRAG Standard Test Methods for reinforced plastics royal aerospace establishment; 1988. [13] Curtis PT, editor. CRAG Standard Method 302 ‘‘Method of test for the tensile strength and modulus of multidirectional fibre reinforced plastics royal aerospace establishment’’; 1988. [14] Determination of the composite fibre volume fraction with the density method using liquid displacement. Internal Engineering Composite Research Centre Document, University of Ulster; 1998. [15] Ping Tan, Liyong Tong, Steven GP, Takashi Ishikawa. Behaviour of 3D orthogonal woven CFRP composites. Part 1: Experimental investigation. Compos Pt A 2000;31:259–71. [16] Ettemeyer A. Noncontact and whole field strain analysis with a laseroptical strain sensor. In: Proceedings of the third international congress on experimental mechanics; 1996.