- Email: [email protected]
Mechanical behaviour analysis of the interface in single hemp yarn composites: DIC measurements and FEM calculations
Polymer Testing 52 (2016) 1e8
Contents lists available at ScienceDirect
Polymer Testing journal homepage: www.elsevier.com/locate/polytest
Test method
Mechanical behaviour analysis of the interface in single hemp yarn composites: DIC measurements and FEM calculations lie Perrier*, Fabienne Touchard, Laurence Chocinski-Arnault, David Mellier Ame Institut Pprime, CNRS-ISAE-ENSMA-Universit e de Poitiers UPR 3346, D epartement Physique et M ecanique des Mat eriaux, ENSMA 1 av Cl ement Ader, 86961 Futuroscope, Chasseneuil, France
a r t i c l e i n f o
a b s t r a c t
Article history: Received 16 February 2016 Accepted 25 March 2016 Available online 28 March 2016
The present work aims to investigate the local deformation mechanisms around a yarn in an ecocomposite. Different hemp yarn orientations and two types of epoxy resin were tested. Full-field measurements were realised with the digital image correlation technique on specific single yarn composites, either on the face of the specimens, or on the edge. The tensile tests were performed under an optical microscope to give sufficient precision, and a numerical model was developed. The experimental results showed high heterogeneities in strain fields which increase with the applied stress level. The comparison with the underlying microstructure and the numerical model enabled us to study the influence of the yarn on the mechanical behaviour. The local constitutive behaviour of the different constituents of the specimens could be approached by these analyses. These results constitute a complete and original database on hemp/epoxy interface mechanical behaviour. © 2016 Elsevier Ltd. All rights reserved.
Keywords: Single yarn composite Hemp Digital image correlation Local strain measurement
1. Introduction Plant fibres are good candidates for the replacement of some synthetic fibres in semi-structural applications, thanks to their good specific properties comparable to those of glass fibres [1e4]. However, one of the main issues of using these fibres is the incompatibility between the hydrophilic fibres and the hydrophobic matrix. Indeed, this induces a decrease in interfacial characteristics compared to glass/polymer composites [5]. The highest strength level of a composite material is achieved when the ability to transfer stress across the fibre-matrix interface is high, i.e. when adhesion is the best. Thus, a fine characterisation of the interface between plant fibre and polymer matrix is required. Many techniques exist to quantify the strength of a fibre/matrix interface in eco-composites, such as fragmentation test [6,7], pullout test [5,8] or its variant, the microdroplet test [9,10]. These techniques allow measuring the interfacial shear strength (IFSS) but, in order to improve the understanding of the deformation mechanisms involved at the interface, the whole strain fields also have to be investigated. In the present work, deformation mechanisms around a hemp yarn in epoxy resin are studied thanks to a
* Corresponding author. E-mail address: [email protected] (A. Perrier). http://dx.doi.org/10.1016/j.polymertesting.2016.03.019 0142-9418/© 2016 Elsevier Ltd. All rights reserved.
full-field strain measurement method. The digital image correlation (DIC) method is a non-contact fullfield displacement measurement technique. First developed by Sutton et al. [11], this technique has been widely used for several applications, including determination of stress-strain behaviour [12], residual stress measurement [13] or observation of crack growth [14]. It has also been successfully used at the ply scale to measure local strain fields in composites made with epoxy matrix reinforced with woven synthetic [15e17] or natural fibres [18]. In these studies, strain distribution at the surface of composites could be compared with the underlying microstructure, showing the influence of the fabric architecture. In woven or braided composites, yarns are close and interfere with each other. To avoid such influence, single yarn composites are considered in this paper. Their mechanical properties are studied and longitudinal, transverse and shear strain fields are measured by digital image correlation under an optical microscope, which, as far as we know, has never been published in literature. The single hemp yarn composites used in this study enable the determination of local strains at the yarn scale, with comparison with the underlying microstructure. Furthermore, composites made with two types of epoxy resins are compared, one being partially bio-based. A finite element model is also developed, with the aim of using the numerical stress values at some specific points in the composite together with experimental strains to determine local behaviour of
2
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
the different composite components. 2. Materials and methods 2.1. Materials The composite materials studied were made of a single hemp yarn embedded in an epoxy matrix. Hemp yarns, which were not treated, were made of hemp fibres with an average diameter of 13 ± 5 mm [19]. Those yarns were produced with twist level of 324 tpm (yarn surface twist angle of 11 ) and a linear density of 83 tex. Besides the irregular cross-section, the hemp yarns had an apparent diameter of about 300 mm [7]. Two epoxy resins were used: a fully synthetic epoxy resin, Epolam 2020 (Axson technologies), with a density of 1.10 g/cm3 after curing [19], and a partially bio-based resin, Greenpoxy 56 (from Sicomin, France), containing 56% of bio-based carbon atoms, with a density of 1.18 g/cm3. Composite plates were manufactured at Pprime Institute by contact moulding in a specific mould [20]. The plates made with Epolam 2020 were cured with the following cycle: 24 h at ambient temperature, 3 h at 40 C, 2 h at 60 C, 2 h at 80 C and 4 h at 100 C. The plates with Greenpoxy 56 were cured 24 h at 40 C, 16 h at 60 C, 8 h at 80 C and 30 min at 95 C. The curing cycles were chosen in order to reach a maximum degree of crosslinking for each resin. Glass transition temperatures, measured by differential scanning calorimetry, were found to be equal to 89 ± 1 C for fully synthetic resin and 83 ± 1 C for partially bio-based resin. Dumbbell samples 53 mm long were cut from these 2 mm thick plates in two different directions, in such a way that the yarn was oriented at 90 or 45 with respect to the tensile direction (Fig. 1). 2.2. Digital image correlation Plane strain fields on each specimen surface were obtained by digital image correlation. The principle of this technique is based on a unique random pattern, recorded twice, once before loading the sample and once when the sample is deformed. The first picture taken at the initial state is the reference picture which is compared to a second one taken at a deformed state. The first picture is divided into small sub-windows, each one being a measurement point characterised by its grey level distribution. The principle consists in seeking, for each sub-window of the reference image, the most similar one in the deformed image (in terms of spatial distribution of grey levels), using a correlation function. This was performed thanks to the correlation software OpenDIC [21]. In order to match sub-windows uniquely and accurately, the object surface must have a random speckle pattern which deforms together with the object. The speckle pattern can be a natural feature of the object or an artificial feature. In this study, specimen were prepared by laying a mixture of paint and particles of 200 nm
Fig. 1. Single yarn composite samples with hemp yarn oriented at 90 or 45 in regard to the tensile direction.
diameter. This made it possible to obtain a fine random pattern on the material surface. The thickness of the paint layer was about 33 mm. With the aim of performing measurements with high spatial resolution, tests were done with a micro tensile tester placed under an optical microscope (Fig. 2a). Pictures taken with a 5 Megapixel camera covered an area of 846 709 mm2. In this experimental configuration, 1 pixel was equal to 0.345 mm. The sub-windows used for correlation had a linear size of 21 mm, with the same distance between two points to avoid their overlapping. Strain fields were measured either on the specimen face (in this case, specimens were polished to reduce the thickness of the resin between the yarn and the surface), or on the edge where the yarn reaches the surface, as seen in Fig. 2b. On the face or edge which was not involved in the DIC measurement, macroscopic strain was measured by a video extensometer. For that purpose, dumbbell samples were marked with two dots made with a marker pen, spaced approximately 10 mm apart. 2.3. Finite element modelling In order to provide a better understanding of strain mechanisms involved in the studied single yarn composites, a numerical simulation of tensile tests was developed. The single yarn composite specimen was modelled using the commercial finite element software Abaqus, with the yarn oriented at 90 or 45 (Fig. 3). The geometry of the model was based on the experimental specimen dimensions, and the yarn was considered as a homogeneous cylinder of constant diameter inserted in a hole of the same diameter, in the central axis of the specimen. The tensile test was reproduced by submitting one extremity of the specimen to a displacement in direction x, and by clamping the other end. The two materials, epoxy matrix and hemp yarn, were simultaneously submitted to the same displacement. To reduce the processing time, the mesh, composed of linear elements 8-node bricks C3D8, was refined only in the yarn and around it. In this model, the yarn was considered as perfectly bonded to the resin: nodes at the interface between the matrix and the yarn were coincident. The yarn constitutive behaviour was considered as orthotropic linear elastic. The Young's modulus of the yarn was determined from tensile tests of single hemp yarns impregnated with epoxy resin [20], and the Poisson's ratio from an analytical model of a hemp yarn [7]. Elasto-plastic behaviour based on experimental results was entered point by point for the epoxy resin. The initial Young's modulus and Poisson's ratio of the resin were determined experimentally by tensile tests. To evaluate the effect on measured strain of the speckled paint layer, a numerical calculation has been performed with an additional layer 33 mm thick at the surface of the specimen, with the same mesh size. The elastic behaviour of this layer was defined with a Young's modulus of 700 MPa and a Poisson's ratio of 0.39, which corresponds to the type of paint used. Fig. 4 shows the longitudinal strain field measured on the edge of the sample with an applied displacement of 0.53 mm for the two models (with and without speckled paint). The strain values along the horizontal dotted lines are plotted in Fig. 4c. The values calculated with the paint layer fit well those without speckled paint, except close to the yarn/matrix interface. At this location, in the model without paint, the strain values are driven by the abrupt geometrical transition between the two different materials, whereas they are smoothed in the model with the paint layer. This comparison illustrates the effect of the speckled paint; its addition can cause a smoothing of the data if the material contains abrupt transitions in the properties of its constituents, but has very small influence apart from this
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
3
Fig. 2. a) Experimental setup, b) DIC measurement locations on the edge or on the face of a specimen.
Fig. 5. Experimental stress-strain curves for single yarn/Epolam (SYE) and single yarn/ Greenpoxy (SYG) samples for both yarn orientation (90 and 45 ) obtained with video extensometer. Fig. 3. Modelling of both yarn orientations, a) 90 or b) 45 in regard to the tensile direction x.
Fig. 4. Longitudinal strain fields obtained by FEM on the edge of a 90 single hemp yarn composite sample a) without speckled paint, b) with speckled paint, c) strain values along dotted lines.
mismatch zone. 3. Results and discussion 3.1. Macroscopic data The measurement with video extensometer enabled us to obtain stress-strain curves of single hemp yarn samples with Epolam resin (SYE) or Greenpoxy resin (SYG) and with yarn
oriented at 90 or 45 (Fig. 5). The tensile machine was stopped while the pictures for DIC were recorded, as can be seen on each curve. The results show that the hemp yarn orientation has no significant influence on the Young's modulus at this scale (for example, for hemp/Epolam samples, E ¼ 2557 ± 257 MPa for 90 oriented hemp yarn and 2836 ± 374 MPa for 45 oriented one). Specimen with 45 oriented yarn presents a higher elongation at break than with a yarn oriented at 90 . In this last configuration, the hemp yarn is loaded in the transverse direction in which it has low
4
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
resistance, as fibres separate easily from each other, leading to failure of the sample. Averaged macroscopic experimental results for single 90 and 45 yarn composites and neat resin samples are presented in Table 1. 3.2. Comparison of experimental results and numerical model 3.2.1. Hemp/Epolam specimen with yarn oriented at 90 Full-field strain measurements on the samples were obtained by Digital Image Correlation. Fig. 6 shows longitudinal strain maps measured on the edge (Fig. 6a) and on the face (Fig. 6b) of a single yarn composite, with the yarn tilted of 90 . For the different applied stress levels, the same colour code has been used; in this way the evolution of strain is easily shown. Micrographs were taken before the speckled paint addition: we can see in Fig. 6a the yarn section and the hemp yarn visible by transmitted light in Fig. 6b. Particular attention was paid to measure strain at these locations. The measurements show that strain distributions are highly heterogeneous, with their values increasing with applied stress level. On the edge of the specimen, strains are localised inside the section of the yarn (represented by the dotted circle) and, for 2.31% of macroscopic strain, maximum strain value measured by DIC is nearly 17%. On the face, high strain values are located along the underlying yarn, and the maximum longitudinal strain is found to be 14.2% for a macroscopic strain of only 1.39%. There is a large gap between the macroscopic and local strain, showing strong heterogeneities in the strain distribution caused by the embedded yarn. In order to get a better understanding of strain mechanisms involved in the tested samples, strain fields were compared with results of the finite element model. Because of the perfect interface definition in the model, this constitutes a qualitative comparison between the numerical and experimental results. First, comparison has been performed for sample edge (Fig. 7). On this side, even if the modelled yarn is homogeneous, we can already see a good correlation between the experimental and numerical strain fields. The experimental maximum strain measured on the longitudinal strain field is 12.77%, whereas the corresponding macroscopic strain is 1.95%: at the surface of the sample, the yarn highly concentrates strains. In longitudinal and transverse strain fields, high strain levels are localised on the hemp yarn section, and rather on the yarn-matrix interface for the shear strain. This is consistent with the fact that this interfacial area is sensitive to shear, as shown with the finite element model in the corresponding picture. The comparison between DIC measurements and FEM calculation has also been performed for sample face behaviour. Here again, the numerical longitudinal strain fields correspond quite well to the experimental ones (Fig. 8). Two stripes with high strain levels are found on each side of the underlying yarn. These are directly linked with the strain measured on the edge. The experimental transverse and shear strain fields are less marked, because of the low strain levels which are measured. Indeed, the highest strain value is 6.07% for εxx, whereas it is not higher than 2.36% in the other strain fields.
Therefore, in that case the comparison with the model is more difficult, but it seems that the maximum transverse strain is located directly above the underlying yarn on the surface of the 90 oriented yarn sample. For specimens with 90 oriented yarn and Greenpoxy matrix, the measured strain fields are similar to those of samples with Epolam resin, both on the edge and on the face. 3.2.2. Hemp/Greenpoxy specimen with 45 oriented yarn For 45 oriented yarn samples, strain fields are also qualitatively the same for specimens with Epolam or Greenpoxy matrices; thus only the results with the partially bio-based resin are described here. For those samples, the strain distributions are rather different from those with the 90 oriented yarn ones, as we can see in Fig. 9 on a sample edge. Large strain values are measured in the longitudinal field (9.51%), as well as in the shear strain field (6.36%), whereas the transverse strain values are lower (3.4% maximum in absolute value). In all strain fields, high strain values are concentrated at the left of the yarn section. At this location, the yarn/ matrix interface is highly subjected to strain since the yarn extends into the specimen in the (1, 1, 0) direction. The numerical model also shows here good correspondence with experimental results, even if the areas of high strain levels are less extended. The measurements made on the face of a specimen with the yarn oriented at 45 give us strain fields such as seen in Fig. 10. The longitudinal strain field values are lower above the underlying yarn than on both sides of it. The transversal strain is also higher in the resin than above the yarn (1.44% vs 0.39%). In contrast, the shear strain is higher at the yarn location. The yarn prevents large strain at the surface above its location compared to the strain measured in the resin, as seen in longitudinal and transverse strain fields, but it facilitates the shear strain, unlike the 90 oriented yarn specimen. 3.3. Experimental strain evolution along lines The evolution of the measured strain can be more accurately described by plotting on the same graph the strain values for different stress levels of the specimen during a tensile test. Fig. 11a shows the longitudinal strain values along a horizontal line A (see diagram on the right) for the 90 oriented yarn/Epolam specimen, for different applied stress levels. We can see that the strain increases progressively and, significantly, more in the vicinity of the yarn. In the tested configuration, strain heterogeneities appear between applied stress levels of 18 MPa (mean strain value of 0.63%) and 27 MPa (maximum strain value of 2.04%). The evolution of longitudinal strain for hemp/Greenpoxy is plotted in Fig. 11b. The appearance of variations in strain occurs for this sample between 24 MPa (mean strain value of 0.87%) and 31 MPa (with maximum value of 2.60%). For the same applied stress of 31 MPa, the maximum strain along the line for hemp/Epolam specimen is equal to 4.73%, whereas it is only 2.46% for the hemp/Greenpoxy sample. This
Table 1 Mechanical characteristics of single 90 and 45 hemp yarn/Epolam and hemp yarn/Greenpoxy composites and their components submitted to tensile tests.
Hemp yarn/Epolam Hemp yarn/Greenpoxy Epolam Greenpoxy Hemp yarn impregnated with epoxy resin [20]
Yarn orientation
E (MPa)
smax (MPa)
90 45 90 45 \ \ \
2557 ± 257 2836 ± 374 2845 ± 200 2565 ± 333 3112 ± 273 2509 ± 105 El ¼ 23000 ± 3000 Et ¼ 1264 ± 154
45 56 42 47 62 49
± ± ± ± ± ±
7 5 14 4 6 3
srupt (MPa) 45 55 42 46 57 48 601
± ± ± ± ± ± ±
6 6 15 4 3 3 79
εrupt (%) 2.65 3.57 1.92 4.66 7.25 4.34
± ± ± ± ± ±
0.66 0.31 0.67 1.16 3.12 0.54
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
5
Fig. 6. Evolution of longitudinal strain fields with applied stress for single hemp yarn/Epolam sample with 90 oriented yarn. DIC measurements are performed a) on the sample edge, b) on the sample face.
Fig. 7. Qualitative comparison of strain fields (εxx, εzz and εxz) obtained by DIC at 40 MPa applied stress and FEM on the edge of a 90 oriented single hemp yarn/Epolam sample.
Fig. 8. Qualitative comparison of strain fields (εxx, εyy and εxy) obtained by DIC at 49 MPa applied stress and FEM on the face of a 90 oriented single hemp yarn/Epolam sample.
demonstrates better strength of the hemp/Greenpoxy sample in
the interfacial zone than for the sample made with hemp and fully
6
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
Fig. 9. Qualitative comparison of strain fields (εxx, εzz and εxz) obtained by DIC at 38 MPa applied stress and FEM on the edge of a 45 oriented single hemp yarn/Greenpoxy sample.
Fig. 10. Qualitative comparison of strain fields (εxx, εyy and εxy) obtained by DIC at 46 MPa applied stress and FEM on the face of a 45 oriented single hemp yarn/Greenpoxy sample.
Fig. 11. Longitudinal strain values along line A (see diagram) measured by DIC at different applied stress levels on the face of 90 oriented yarn specimens a) a single hemp yarn/ Epolam sample, b) a single hemp yarn/Greenpoxy sample.
synthetic epoxy matrix. Same analysis has been made on the strain measured on the edge of the 90 oriented specimens. The comparison of these measurements for the two resins is shown in Fig. 12. As for the measurements on the face, we observed that the appearance of strain heterogeneities is progressive and localised at the yarn section. At 40 MPa applied stress for hemp/Epolam and 41 MPa for
hemp/Greenpoxy, the maximum longitudinal strain values are 12.33% and 7.41%, respectively. Here again, the strains values are lower for the hemp/Greenpoxy specimen. Concerning the shape of the strains curves, one can see larger waves for the hemp/Greenpoxy specimen than for the hemp/Epolam one, certainly due to the larger section of the yarn (389 mm versus 332 mm along the line B, respectively).
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
7
Fig. 12. Longitudinal strain values along line B (see diagram) measured by DIC at different applied stress levels on the edge of 90 oriented yarn specimens a) a single hemp yarn/ Epolam sample, b) a single hemp yarn/Greenpoxy sample.
3.4. Towards the interface local behaviour The objective of using the DIC full-field technique is the determination of the local mechanical behaviour. This enables us to characterise the local behaviour of each constituent of the composite (resin and yarn) and of the yarn/resin interface. For example, in Fig. 13, local stress-strain curves for some points at specific locations (in the yarn section, at the yarn/matrix interface and in the matrix away from the yarn) are plotted. Stress values correspond to the applied stress levels, and strains correspond to the longitudinal strain values measured by DIC. This type of curve allows us to follow local strain evolution anywhere in the studied area and gives an approach to the local constitutive behaviour at the chosen points. Thus, Fig. 13 shows that strain values are higher in the yarn than in the resin, as observed on the strain fields. It appears that the higher the applied stress, the higher the gap between the maximum and minimum strain values, which illustrates again the heterogeneity in material deformations. For example, for an applied stress of 40 MPa, the strain value at the point in the yarn is almost fifteen times greater than that in the resin for the hemp/Epolam sample. At the interface, the studied points have intermediate behaviour, strain values are 47% and 45% below those measured in the middle of the yarn cross section for the hemp/Epolam and the hemp/Greenpoxy specimen, respectively. The macroscopic strain, measured by video extensometer, takes into account every components of the composites. As yarn and interface exhibit lower apparent stiffness than the resin one, the apparent stiffness of the whole composite is a bit lower than for bulk resin. Indeed, we can see that the apparent stiffness of each component varies greatly. In the resin, it is higher than in the yarn and in the interfacial zone. The apparent stiffness in the yarn, which represents its transverse apparent stiffness, is found between 825 MPa (with Epolam matrix) and 1574 MPa (with Greenpoxy matrix). This is not so far from the
data determined for the impregnated yarn of 1264 MPa (Table 1). At the points of the interface, the estimated apparent stiffnesses have intermediate values, with 1171 MPa for the hemp/Epolam sample and 1869 MPa for the hemp/Greenpoxy one. With no specific treatment applied to the hemp yarn for the manufactured composites with the two polymer matrices, these results show that the interface adhesion quality is better with the partially bio-based Greenpoxy resin than with the fully synthetic Epolam one. These results are the first step for determining local stress-local strain curves. To replace the applied stress values in Fig. 13 by the local stress, and thus to obtain the local behaviour of each component, comparison of the finite element modelling with the experimental results is needed in order to make an iterative optimization of the interface parameters. This will be the next step of this work.
4. Conclusions This study deals with full-field measurements on single hemp yarn composites, with different yarn orientations (90 or 45 ). Two types of epoxy resin have been used: a fully synthetic one (Epolam 2020) and a partially bio-based one (Greenpoxy 56). Tensile tests coupled with the digital image correlation method have been performed under an optical microscope. In-plane longitudinal, transverse and shear strain fields have been obtained at the scale of the yarn either on the face or on the edge of the specimens. For a better understanding of strain mechanisms involved in such composites, a numerical model has been developed, considering, in a first step, perfect adhesion between the yarn and the matrix. The experimental DIC results show strong heterogeneity in strain fields which develops with the applied stress. A comparison with the underlying microstructure and with the numerical model has been realised. It made possible the analysis of the influence of
Fig. 13. Stress-strain curves obtained by DIC at different locations on the edge of 90 oriented yarn sample a) a single hemp yarn/Epolam specimen and b) a single hemp yarn/ Greenpoxy specimen.
8
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
the yarn presence on the longitudinal, transverse and shear mechanical behaviour. This analysis enabled us to approach the local behaviour at specific locations on the edge of the samples. Local strain values in each component of the composites, i.e. in the yarn, at the yarn/matrix interface and in the matrix, have been plotted versus the applied stress, giving a first step towards the determination of local behaviour. These results constitute a complete and original database on hemp/epoxy interface mechanical behaviour. It is essential for determining the local constitutive behaviour of the interface, which is needed for precise finite element calculations. Acknowledgments gion Poitou-Charentes for its financial The authors thank the Re support. References [1] D.U. Shah, Developing plant fibre composites for structural applications by optimising composite parameters: a critical review, J. Mater. Sci. 48 (2013) 6083e6107. [2] D.B. Dittenber, H.V.S. GangaRao, Critical review of recent publications on use of natural composites in infrastructure, Compos. Part Appl. Sci. Manuf. 43 (2012) 1419e1429. [3] S. Goutianos, T. Peijs, B. Nystrom, M. Skrifvars, Development of flax fibre based textile reinforcements for composite applications, Appl. Compos. Mater 13 (2006) 199e215. [4] M.C. Symington, W.M. Banks, D. West, R.A. Pethrick, Tensile testing of cellulose based natural fibers for structural composite applications, J. Compos. Mater. 43 (2009) 1083e1108. [5] T.-T.-L. Doan, H. Brodowsky, E. M€ ader, Jute fibre/epoxy composites: surface properties and interfacial adhesion, Compos. Sci. Technol. 72 (2012) 1160e1166. [6] F.G. Torres, M.L. Cubillas, Study of the interfacial properties of natural fibre reinforced polyethylene, Polym. Test. 24 (2005) 694e698. [7] C. Guillebaud-Bonnafous, D. Vasconcellos, F. Touchard, L. Chocinski-Arnault, Experimental and numerical investigation of the interface between epoxy
matrix and hemp yarn, Compos. Part Appl. Sci. Manuf. 43 (2012) 2046e2058. [8] T.H.D. Sydenstricker, S. Mochnaz, S.C. Amico, Pull-out and other evaluations in sisal-reinforced polyester biocomposites, Polym. Test. 22 (2003) 375e380. [9] S.J. Eichhorn, R.J. Young, Composite micromechanics of hemp fibres and epoxy resin microdroplets, Compos. Sci. Technol. 64 (2004) 767e772. [10] H.U. Zaman, M. Beg, Preparation, structure, and properties of the coir fiber/ polypropylene composites, J. Compos. Mater 48 (2014) 3293e3301. [11] M.A. Sutton, S.R. McNeill, J. Jang, M. Babai, The effect of subpixel image restoration on digital image correlation estimates, Opt. Eng. 27 (1988) 870e877. [12] M.A. Bahlouli, S.M. Guil, S. Ahzi, M. Laberge, Stressestrain response of biomaterials by a digital image correlation method: application to tecoflex, J. Mater. Sci. Technol. 20 (2004) 114e116. [13] J.D. Lord, D. Penn, P. Whitehead, The application of digital image correlation for measuring residual stress by incremental hole drilling, Appl. Mech. Mater 13e14 (2008) 65e73. [14] J. Lambros, J. Abanto-Bueno, Investigation of crack growth in functionally graded materials using digital image correlation, Eng. Fract. Mech. 69 (2002) 1695e1711. , L. Sabatier, Interlaminar [15] P. Lecomte-Grosbras, B. Paluch, M. Brieu, G.D. Saxce shear strain measurement on angle-ply laminate free edge using digital image correlation, Compos. Part Appl. Sci. Manuf. 40 (2009) 1911e1920. [16] G. Nicoletto, G. Anzelotti, E. Riva, Mesoscopic strain fields in woven composites: experiments vs. finite element modeling, Opt. Lasers Eng. 47 (2009) 352e359. [17] D.S. Ivanov, S.V. Lomov, A.E. Bogdanovich, M. Karahan, I. Verpoest, A comparative study of tensile properties of non-crimp 3D orthogonal weave and multi-layer plain weave E-glass composites. Part 2: comprehensive experimental results, Compos. Part Appl. Sci. Manuf. 40 (2009) 1144e1157. [18] A. Perrier, D.S. de Vasconcellos, F. Touchard, L. Chocinski-Arnault, D. Mellier, Full-field measurement at the weave scale in hemp/epoxy composite using digital image correlation, Polym. Polym. Compos 23 (2015). [19] D.S. de Vasconcellos, F. Touchard, L. Chocinski-Arnault, Tensionetension fatigue behaviour of woven hemp fibre reinforced epoxy composite: a multiinstrumented damage analysis, Int. J. Fatigue 59 (2014) 159e169. [20] A. Perrier, R. Ecault, F. Touchard, M. Vidal Urriza, J. Baillargeat, L. ChocinskiArnault, et al., Towards the development of laser shock test for mechanical characterisation of fibre/matrix interface in eco-composites, Polym. Test. 44 (2015) 125e134. , F. Bridier, P. Bocher, An open source software for the [21] N. Vanderesse, M. Lagace measurement of deformation fields by means of digital image correlation, Microsc. Microanal. 19 (2013) 820e821.
Contents lists available at ScienceDirect
Polymer Testing journal homepage: www.elsevier.com/locate/polytest
Test method
Mechanical behaviour analysis of the interface in single hemp yarn composites: DIC measurements and FEM calculations lie Perrier*, Fabienne Touchard, Laurence Chocinski-Arnault, David Mellier Ame Institut Pprime, CNRS-ISAE-ENSMA-Universit e de Poitiers UPR 3346, D epartement Physique et M ecanique des Mat eriaux, ENSMA 1 av Cl ement Ader, 86961 Futuroscope, Chasseneuil, France
a r t i c l e i n f o
a b s t r a c t
Article history: Received 16 February 2016 Accepted 25 March 2016 Available online 28 March 2016
The present work aims to investigate the local deformation mechanisms around a yarn in an ecocomposite. Different hemp yarn orientations and two types of epoxy resin were tested. Full-field measurements were realised with the digital image correlation technique on specific single yarn composites, either on the face of the specimens, or on the edge. The tensile tests were performed under an optical microscope to give sufficient precision, and a numerical model was developed. The experimental results showed high heterogeneities in strain fields which increase with the applied stress level. The comparison with the underlying microstructure and the numerical model enabled us to study the influence of the yarn on the mechanical behaviour. The local constitutive behaviour of the different constituents of the specimens could be approached by these analyses. These results constitute a complete and original database on hemp/epoxy interface mechanical behaviour. © 2016 Elsevier Ltd. All rights reserved.
Keywords: Single yarn composite Hemp Digital image correlation Local strain measurement
1. Introduction Plant fibres are good candidates for the replacement of some synthetic fibres in semi-structural applications, thanks to their good specific properties comparable to those of glass fibres [1e4]. However, one of the main issues of using these fibres is the incompatibility between the hydrophilic fibres and the hydrophobic matrix. Indeed, this induces a decrease in interfacial characteristics compared to glass/polymer composites [5]. The highest strength level of a composite material is achieved when the ability to transfer stress across the fibre-matrix interface is high, i.e. when adhesion is the best. Thus, a fine characterisation of the interface between plant fibre and polymer matrix is required. Many techniques exist to quantify the strength of a fibre/matrix interface in eco-composites, such as fragmentation test [6,7], pullout test [5,8] or its variant, the microdroplet test [9,10]. These techniques allow measuring the interfacial shear strength (IFSS) but, in order to improve the understanding of the deformation mechanisms involved at the interface, the whole strain fields also have to be investigated. In the present work, deformation mechanisms around a hemp yarn in epoxy resin are studied thanks to a
* Corresponding author. E-mail address: [email protected] (A. Perrier). http://dx.doi.org/10.1016/j.polymertesting.2016.03.019 0142-9418/© 2016 Elsevier Ltd. All rights reserved.
full-field strain measurement method. The digital image correlation (DIC) method is a non-contact fullfield displacement measurement technique. First developed by Sutton et al. [11], this technique has been widely used for several applications, including determination of stress-strain behaviour [12], residual stress measurement [13] or observation of crack growth [14]. It has also been successfully used at the ply scale to measure local strain fields in composites made with epoxy matrix reinforced with woven synthetic [15e17] or natural fibres [18]. In these studies, strain distribution at the surface of composites could be compared with the underlying microstructure, showing the influence of the fabric architecture. In woven or braided composites, yarns are close and interfere with each other. To avoid such influence, single yarn composites are considered in this paper. Their mechanical properties are studied and longitudinal, transverse and shear strain fields are measured by digital image correlation under an optical microscope, which, as far as we know, has never been published in literature. The single hemp yarn composites used in this study enable the determination of local strains at the yarn scale, with comparison with the underlying microstructure. Furthermore, composites made with two types of epoxy resins are compared, one being partially bio-based. A finite element model is also developed, with the aim of using the numerical stress values at some specific points in the composite together with experimental strains to determine local behaviour of
2
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
the different composite components. 2. Materials and methods 2.1. Materials The composite materials studied were made of a single hemp yarn embedded in an epoxy matrix. Hemp yarns, which were not treated, were made of hemp fibres with an average diameter of 13 ± 5 mm [19]. Those yarns were produced with twist level of 324 tpm (yarn surface twist angle of 11 ) and a linear density of 83 tex. Besides the irregular cross-section, the hemp yarns had an apparent diameter of about 300 mm [7]. Two epoxy resins were used: a fully synthetic epoxy resin, Epolam 2020 (Axson technologies), with a density of 1.10 g/cm3 after curing [19], and a partially bio-based resin, Greenpoxy 56 (from Sicomin, France), containing 56% of bio-based carbon atoms, with a density of 1.18 g/cm3. Composite plates were manufactured at Pprime Institute by contact moulding in a specific mould [20]. The plates made with Epolam 2020 were cured with the following cycle: 24 h at ambient temperature, 3 h at 40 C, 2 h at 60 C, 2 h at 80 C and 4 h at 100 C. The plates with Greenpoxy 56 were cured 24 h at 40 C, 16 h at 60 C, 8 h at 80 C and 30 min at 95 C. The curing cycles were chosen in order to reach a maximum degree of crosslinking for each resin. Glass transition temperatures, measured by differential scanning calorimetry, were found to be equal to 89 ± 1 C for fully synthetic resin and 83 ± 1 C for partially bio-based resin. Dumbbell samples 53 mm long were cut from these 2 mm thick plates in two different directions, in such a way that the yarn was oriented at 90 or 45 with respect to the tensile direction (Fig. 1). 2.2. Digital image correlation Plane strain fields on each specimen surface were obtained by digital image correlation. The principle of this technique is based on a unique random pattern, recorded twice, once before loading the sample and once when the sample is deformed. The first picture taken at the initial state is the reference picture which is compared to a second one taken at a deformed state. The first picture is divided into small sub-windows, each one being a measurement point characterised by its grey level distribution. The principle consists in seeking, for each sub-window of the reference image, the most similar one in the deformed image (in terms of spatial distribution of grey levels), using a correlation function. This was performed thanks to the correlation software OpenDIC [21]. In order to match sub-windows uniquely and accurately, the object surface must have a random speckle pattern which deforms together with the object. The speckle pattern can be a natural feature of the object or an artificial feature. In this study, specimen were prepared by laying a mixture of paint and particles of 200 nm
Fig. 1. Single yarn composite samples with hemp yarn oriented at 90 or 45 in regard to the tensile direction.
diameter. This made it possible to obtain a fine random pattern on the material surface. The thickness of the paint layer was about 33 mm. With the aim of performing measurements with high spatial resolution, tests were done with a micro tensile tester placed under an optical microscope (Fig. 2a). Pictures taken with a 5 Megapixel camera covered an area of 846 709 mm2. In this experimental configuration, 1 pixel was equal to 0.345 mm. The sub-windows used for correlation had a linear size of 21 mm, with the same distance between two points to avoid their overlapping. Strain fields were measured either on the specimen face (in this case, specimens were polished to reduce the thickness of the resin between the yarn and the surface), or on the edge where the yarn reaches the surface, as seen in Fig. 2b. On the face or edge which was not involved in the DIC measurement, macroscopic strain was measured by a video extensometer. For that purpose, dumbbell samples were marked with two dots made with a marker pen, spaced approximately 10 mm apart. 2.3. Finite element modelling In order to provide a better understanding of strain mechanisms involved in the studied single yarn composites, a numerical simulation of tensile tests was developed. The single yarn composite specimen was modelled using the commercial finite element software Abaqus, with the yarn oriented at 90 or 45 (Fig. 3). The geometry of the model was based on the experimental specimen dimensions, and the yarn was considered as a homogeneous cylinder of constant diameter inserted in a hole of the same diameter, in the central axis of the specimen. The tensile test was reproduced by submitting one extremity of the specimen to a displacement in direction x, and by clamping the other end. The two materials, epoxy matrix and hemp yarn, were simultaneously submitted to the same displacement. To reduce the processing time, the mesh, composed of linear elements 8-node bricks C3D8, was refined only in the yarn and around it. In this model, the yarn was considered as perfectly bonded to the resin: nodes at the interface between the matrix and the yarn were coincident. The yarn constitutive behaviour was considered as orthotropic linear elastic. The Young's modulus of the yarn was determined from tensile tests of single hemp yarns impregnated with epoxy resin [20], and the Poisson's ratio from an analytical model of a hemp yarn [7]. Elasto-plastic behaviour based on experimental results was entered point by point for the epoxy resin. The initial Young's modulus and Poisson's ratio of the resin were determined experimentally by tensile tests. To evaluate the effect on measured strain of the speckled paint layer, a numerical calculation has been performed with an additional layer 33 mm thick at the surface of the specimen, with the same mesh size. The elastic behaviour of this layer was defined with a Young's modulus of 700 MPa and a Poisson's ratio of 0.39, which corresponds to the type of paint used. Fig. 4 shows the longitudinal strain field measured on the edge of the sample with an applied displacement of 0.53 mm for the two models (with and without speckled paint). The strain values along the horizontal dotted lines are plotted in Fig. 4c. The values calculated with the paint layer fit well those without speckled paint, except close to the yarn/matrix interface. At this location, in the model without paint, the strain values are driven by the abrupt geometrical transition between the two different materials, whereas they are smoothed in the model with the paint layer. This comparison illustrates the effect of the speckled paint; its addition can cause a smoothing of the data if the material contains abrupt transitions in the properties of its constituents, but has very small influence apart from this
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
3
Fig. 2. a) Experimental setup, b) DIC measurement locations on the edge or on the face of a specimen.
Fig. 5. Experimental stress-strain curves for single yarn/Epolam (SYE) and single yarn/ Greenpoxy (SYG) samples for both yarn orientation (90 and 45 ) obtained with video extensometer. Fig. 3. Modelling of both yarn orientations, a) 90 or b) 45 in regard to the tensile direction x.
Fig. 4. Longitudinal strain fields obtained by FEM on the edge of a 90 single hemp yarn composite sample a) without speckled paint, b) with speckled paint, c) strain values along dotted lines.
mismatch zone. 3. Results and discussion 3.1. Macroscopic data The measurement with video extensometer enabled us to obtain stress-strain curves of single hemp yarn samples with Epolam resin (SYE) or Greenpoxy resin (SYG) and with yarn
oriented at 90 or 45 (Fig. 5). The tensile machine was stopped while the pictures for DIC were recorded, as can be seen on each curve. The results show that the hemp yarn orientation has no significant influence on the Young's modulus at this scale (for example, for hemp/Epolam samples, E ¼ 2557 ± 257 MPa for 90 oriented hemp yarn and 2836 ± 374 MPa for 45 oriented one). Specimen with 45 oriented yarn presents a higher elongation at break than with a yarn oriented at 90 . In this last configuration, the hemp yarn is loaded in the transverse direction in which it has low
4
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
resistance, as fibres separate easily from each other, leading to failure of the sample. Averaged macroscopic experimental results for single 90 and 45 yarn composites and neat resin samples are presented in Table 1. 3.2. Comparison of experimental results and numerical model 3.2.1. Hemp/Epolam specimen with yarn oriented at 90 Full-field strain measurements on the samples were obtained by Digital Image Correlation. Fig. 6 shows longitudinal strain maps measured on the edge (Fig. 6a) and on the face (Fig. 6b) of a single yarn composite, with the yarn tilted of 90 . For the different applied stress levels, the same colour code has been used; in this way the evolution of strain is easily shown. Micrographs were taken before the speckled paint addition: we can see in Fig. 6a the yarn section and the hemp yarn visible by transmitted light in Fig. 6b. Particular attention was paid to measure strain at these locations. The measurements show that strain distributions are highly heterogeneous, with their values increasing with applied stress level. On the edge of the specimen, strains are localised inside the section of the yarn (represented by the dotted circle) and, for 2.31% of macroscopic strain, maximum strain value measured by DIC is nearly 17%. On the face, high strain values are located along the underlying yarn, and the maximum longitudinal strain is found to be 14.2% for a macroscopic strain of only 1.39%. There is a large gap between the macroscopic and local strain, showing strong heterogeneities in the strain distribution caused by the embedded yarn. In order to get a better understanding of strain mechanisms involved in the tested samples, strain fields were compared with results of the finite element model. Because of the perfect interface definition in the model, this constitutes a qualitative comparison between the numerical and experimental results. First, comparison has been performed for sample edge (Fig. 7). On this side, even if the modelled yarn is homogeneous, we can already see a good correlation between the experimental and numerical strain fields. The experimental maximum strain measured on the longitudinal strain field is 12.77%, whereas the corresponding macroscopic strain is 1.95%: at the surface of the sample, the yarn highly concentrates strains. In longitudinal and transverse strain fields, high strain levels are localised on the hemp yarn section, and rather on the yarn-matrix interface for the shear strain. This is consistent with the fact that this interfacial area is sensitive to shear, as shown with the finite element model in the corresponding picture. The comparison between DIC measurements and FEM calculation has also been performed for sample face behaviour. Here again, the numerical longitudinal strain fields correspond quite well to the experimental ones (Fig. 8). Two stripes with high strain levels are found on each side of the underlying yarn. These are directly linked with the strain measured on the edge. The experimental transverse and shear strain fields are less marked, because of the low strain levels which are measured. Indeed, the highest strain value is 6.07% for εxx, whereas it is not higher than 2.36% in the other strain fields.
Therefore, in that case the comparison with the model is more difficult, but it seems that the maximum transverse strain is located directly above the underlying yarn on the surface of the 90 oriented yarn sample. For specimens with 90 oriented yarn and Greenpoxy matrix, the measured strain fields are similar to those of samples with Epolam resin, both on the edge and on the face. 3.2.2. Hemp/Greenpoxy specimen with 45 oriented yarn For 45 oriented yarn samples, strain fields are also qualitatively the same for specimens with Epolam or Greenpoxy matrices; thus only the results with the partially bio-based resin are described here. For those samples, the strain distributions are rather different from those with the 90 oriented yarn ones, as we can see in Fig. 9 on a sample edge. Large strain values are measured in the longitudinal field (9.51%), as well as in the shear strain field (6.36%), whereas the transverse strain values are lower (3.4% maximum in absolute value). In all strain fields, high strain values are concentrated at the left of the yarn section. At this location, the yarn/ matrix interface is highly subjected to strain since the yarn extends into the specimen in the (1, 1, 0) direction. The numerical model also shows here good correspondence with experimental results, even if the areas of high strain levels are less extended. The measurements made on the face of a specimen with the yarn oriented at 45 give us strain fields such as seen in Fig. 10. The longitudinal strain field values are lower above the underlying yarn than on both sides of it. The transversal strain is also higher in the resin than above the yarn (1.44% vs 0.39%). In contrast, the shear strain is higher at the yarn location. The yarn prevents large strain at the surface above its location compared to the strain measured in the resin, as seen in longitudinal and transverse strain fields, but it facilitates the shear strain, unlike the 90 oriented yarn specimen. 3.3. Experimental strain evolution along lines The evolution of the measured strain can be more accurately described by plotting on the same graph the strain values for different stress levels of the specimen during a tensile test. Fig. 11a shows the longitudinal strain values along a horizontal line A (see diagram on the right) for the 90 oriented yarn/Epolam specimen, for different applied stress levels. We can see that the strain increases progressively and, significantly, more in the vicinity of the yarn. In the tested configuration, strain heterogeneities appear between applied stress levels of 18 MPa (mean strain value of 0.63%) and 27 MPa (maximum strain value of 2.04%). The evolution of longitudinal strain for hemp/Greenpoxy is plotted in Fig. 11b. The appearance of variations in strain occurs for this sample between 24 MPa (mean strain value of 0.87%) and 31 MPa (with maximum value of 2.60%). For the same applied stress of 31 MPa, the maximum strain along the line for hemp/Epolam specimen is equal to 4.73%, whereas it is only 2.46% for the hemp/Greenpoxy sample. This
Table 1 Mechanical characteristics of single 90 and 45 hemp yarn/Epolam and hemp yarn/Greenpoxy composites and their components submitted to tensile tests.
Hemp yarn/Epolam Hemp yarn/Greenpoxy Epolam Greenpoxy Hemp yarn impregnated with epoxy resin [20]
Yarn orientation
E (MPa)
smax (MPa)
90 45 90 45 \ \ \
2557 ± 257 2836 ± 374 2845 ± 200 2565 ± 333 3112 ± 273 2509 ± 105 El ¼ 23000 ± 3000 Et ¼ 1264 ± 154
45 56 42 47 62 49
± ± ± ± ± ±
7 5 14 4 6 3
srupt (MPa) 45 55 42 46 57 48 601
± ± ± ± ± ± ±
6 6 15 4 3 3 79
εrupt (%) 2.65 3.57 1.92 4.66 7.25 4.34
± ± ± ± ± ±
0.66 0.31 0.67 1.16 3.12 0.54
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
5
Fig. 6. Evolution of longitudinal strain fields with applied stress for single hemp yarn/Epolam sample with 90 oriented yarn. DIC measurements are performed a) on the sample edge, b) on the sample face.
Fig. 7. Qualitative comparison of strain fields (εxx, εzz and εxz) obtained by DIC at 40 MPa applied stress and FEM on the edge of a 90 oriented single hemp yarn/Epolam sample.
Fig. 8. Qualitative comparison of strain fields (εxx, εyy and εxy) obtained by DIC at 49 MPa applied stress and FEM on the face of a 90 oriented single hemp yarn/Epolam sample.
demonstrates better strength of the hemp/Greenpoxy sample in
the interfacial zone than for the sample made with hemp and fully
6
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
Fig. 9. Qualitative comparison of strain fields (εxx, εzz and εxz) obtained by DIC at 38 MPa applied stress and FEM on the edge of a 45 oriented single hemp yarn/Greenpoxy sample.
Fig. 10. Qualitative comparison of strain fields (εxx, εyy and εxy) obtained by DIC at 46 MPa applied stress and FEM on the face of a 45 oriented single hemp yarn/Greenpoxy sample.
Fig. 11. Longitudinal strain values along line A (see diagram) measured by DIC at different applied stress levels on the face of 90 oriented yarn specimens a) a single hemp yarn/ Epolam sample, b) a single hemp yarn/Greenpoxy sample.
synthetic epoxy matrix. Same analysis has been made on the strain measured on the edge of the 90 oriented specimens. The comparison of these measurements for the two resins is shown in Fig. 12. As for the measurements on the face, we observed that the appearance of strain heterogeneities is progressive and localised at the yarn section. At 40 MPa applied stress for hemp/Epolam and 41 MPa for
hemp/Greenpoxy, the maximum longitudinal strain values are 12.33% and 7.41%, respectively. Here again, the strains values are lower for the hemp/Greenpoxy specimen. Concerning the shape of the strains curves, one can see larger waves for the hemp/Greenpoxy specimen than for the hemp/Epolam one, certainly due to the larger section of the yarn (389 mm versus 332 mm along the line B, respectively).
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
7
Fig. 12. Longitudinal strain values along line B (see diagram) measured by DIC at different applied stress levels on the edge of 90 oriented yarn specimens a) a single hemp yarn/ Epolam sample, b) a single hemp yarn/Greenpoxy sample.
3.4. Towards the interface local behaviour The objective of using the DIC full-field technique is the determination of the local mechanical behaviour. This enables us to characterise the local behaviour of each constituent of the composite (resin and yarn) and of the yarn/resin interface. For example, in Fig. 13, local stress-strain curves for some points at specific locations (in the yarn section, at the yarn/matrix interface and in the matrix away from the yarn) are plotted. Stress values correspond to the applied stress levels, and strains correspond to the longitudinal strain values measured by DIC. This type of curve allows us to follow local strain evolution anywhere in the studied area and gives an approach to the local constitutive behaviour at the chosen points. Thus, Fig. 13 shows that strain values are higher in the yarn than in the resin, as observed on the strain fields. It appears that the higher the applied stress, the higher the gap between the maximum and minimum strain values, which illustrates again the heterogeneity in material deformations. For example, for an applied stress of 40 MPa, the strain value at the point in the yarn is almost fifteen times greater than that in the resin for the hemp/Epolam sample. At the interface, the studied points have intermediate behaviour, strain values are 47% and 45% below those measured in the middle of the yarn cross section for the hemp/Epolam and the hemp/Greenpoxy specimen, respectively. The macroscopic strain, measured by video extensometer, takes into account every components of the composites. As yarn and interface exhibit lower apparent stiffness than the resin one, the apparent stiffness of the whole composite is a bit lower than for bulk resin. Indeed, we can see that the apparent stiffness of each component varies greatly. In the resin, it is higher than in the yarn and in the interfacial zone. The apparent stiffness in the yarn, which represents its transverse apparent stiffness, is found between 825 MPa (with Epolam matrix) and 1574 MPa (with Greenpoxy matrix). This is not so far from the
data determined for the impregnated yarn of 1264 MPa (Table 1). At the points of the interface, the estimated apparent stiffnesses have intermediate values, with 1171 MPa for the hemp/Epolam sample and 1869 MPa for the hemp/Greenpoxy one. With no specific treatment applied to the hemp yarn for the manufactured composites with the two polymer matrices, these results show that the interface adhesion quality is better with the partially bio-based Greenpoxy resin than with the fully synthetic Epolam one. These results are the first step for determining local stress-local strain curves. To replace the applied stress values in Fig. 13 by the local stress, and thus to obtain the local behaviour of each component, comparison of the finite element modelling with the experimental results is needed in order to make an iterative optimization of the interface parameters. This will be the next step of this work.
4. Conclusions This study deals with full-field measurements on single hemp yarn composites, with different yarn orientations (90 or 45 ). Two types of epoxy resin have been used: a fully synthetic one (Epolam 2020) and a partially bio-based one (Greenpoxy 56). Tensile tests coupled with the digital image correlation method have been performed under an optical microscope. In-plane longitudinal, transverse and shear strain fields have been obtained at the scale of the yarn either on the face or on the edge of the specimens. For a better understanding of strain mechanisms involved in such composites, a numerical model has been developed, considering, in a first step, perfect adhesion between the yarn and the matrix. The experimental DIC results show strong heterogeneity in strain fields which develops with the applied stress. A comparison with the underlying microstructure and with the numerical model has been realised. It made possible the analysis of the influence of
Fig. 13. Stress-strain curves obtained by DIC at different locations on the edge of 90 oriented yarn sample a) a single hemp yarn/Epolam specimen and b) a single hemp yarn/ Greenpoxy specimen.
8
A. Perrier et al. / Polymer Testing 52 (2016) 1e8
the yarn presence on the longitudinal, transverse and shear mechanical behaviour. This analysis enabled us to approach the local behaviour at specific locations on the edge of the samples. Local strain values in each component of the composites, i.e. in the yarn, at the yarn/matrix interface and in the matrix, have been plotted versus the applied stress, giving a first step towards the determination of local behaviour. These results constitute a complete and original database on hemp/epoxy interface mechanical behaviour. It is essential for determining the local constitutive behaviour of the interface, which is needed for precise finite element calculations. Acknowledgments gion Poitou-Charentes for its financial The authors thank the Re support. References [1] D.U. Shah, Developing plant fibre composites for structural applications by optimising composite parameters: a critical review, J. Mater. Sci. 48 (2013) 6083e6107. [2] D.B. Dittenber, H.V.S. GangaRao, Critical review of recent publications on use of natural composites in infrastructure, Compos. Part Appl. Sci. Manuf. 43 (2012) 1419e1429. [3] S. Goutianos, T. Peijs, B. Nystrom, M. Skrifvars, Development of flax fibre based textile reinforcements for composite applications, Appl. Compos. Mater 13 (2006) 199e215. [4] M.C. Symington, W.M. Banks, D. West, R.A. Pethrick, Tensile testing of cellulose based natural fibers for structural composite applications, J. Compos. Mater. 43 (2009) 1083e1108. [5] T.-T.-L. Doan, H. Brodowsky, E. M€ ader, Jute fibre/epoxy composites: surface properties and interfacial adhesion, Compos. Sci. Technol. 72 (2012) 1160e1166. [6] F.G. Torres, M.L. Cubillas, Study of the interfacial properties of natural fibre reinforced polyethylene, Polym. Test. 24 (2005) 694e698. [7] C. Guillebaud-Bonnafous, D. Vasconcellos, F. Touchard, L. Chocinski-Arnault, Experimental and numerical investigation of the interface between epoxy
matrix and hemp yarn, Compos. Part Appl. Sci. Manuf. 43 (2012) 2046e2058. [8] T.H.D. Sydenstricker, S. Mochnaz, S.C. Amico, Pull-out and other evaluations in sisal-reinforced polyester biocomposites, Polym. Test. 22 (2003) 375e380. [9] S.J. Eichhorn, R.J. Young, Composite micromechanics of hemp fibres and epoxy resin microdroplets, Compos. Sci. Technol. 64 (2004) 767e772. [10] H.U. Zaman, M. Beg, Preparation, structure, and properties of the coir fiber/ polypropylene composites, J. Compos. Mater 48 (2014) 3293e3301. [11] M.A. Sutton, S.R. McNeill, J. Jang, M. Babai, The effect of subpixel image restoration on digital image correlation estimates, Opt. Eng. 27 (1988) 870e877. [12] M.A. Bahlouli, S.M. Guil, S. Ahzi, M. Laberge, Stressestrain response of biomaterials by a digital image correlation method: application to tecoflex, J. Mater. Sci. Technol. 20 (2004) 114e116. [13] J.D. Lord, D. Penn, P. Whitehead, The application of digital image correlation for measuring residual stress by incremental hole drilling, Appl. Mech. Mater 13e14 (2008) 65e73. [14] J. Lambros, J. Abanto-Bueno, Investigation of crack growth in functionally graded materials using digital image correlation, Eng. Fract. Mech. 69 (2002) 1695e1711. , L. Sabatier, Interlaminar [15] P. Lecomte-Grosbras, B. Paluch, M. Brieu, G.D. Saxce shear strain measurement on angle-ply laminate free edge using digital image correlation, Compos. Part Appl. Sci. Manuf. 40 (2009) 1911e1920. [16] G. Nicoletto, G. Anzelotti, E. Riva, Mesoscopic strain fields in woven composites: experiments vs. finite element modeling, Opt. Lasers Eng. 47 (2009) 352e359. [17] D.S. Ivanov, S.V. Lomov, A.E. Bogdanovich, M. Karahan, I. Verpoest, A comparative study of tensile properties of non-crimp 3D orthogonal weave and multi-layer plain weave E-glass composites. Part 2: comprehensive experimental results, Compos. Part Appl. Sci. Manuf. 40 (2009) 1144e1157. [18] A. Perrier, D.S. de Vasconcellos, F. Touchard, L. Chocinski-Arnault, D. Mellier, Full-field measurement at the weave scale in hemp/epoxy composite using digital image correlation, Polym. Polym. Compos 23 (2015). [19] D.S. de Vasconcellos, F. Touchard, L. Chocinski-Arnault, Tensionetension fatigue behaviour of woven hemp fibre reinforced epoxy composite: a multiinstrumented damage analysis, Int. J. Fatigue 59 (2014) 159e169. [20] A. Perrier, R. Ecault, F. Touchard, M. Vidal Urriza, J. Baillargeat, L. ChocinskiArnault, et al., Towards the development of laser shock test for mechanical characterisation of fibre/matrix interface in eco-composites, Polym. Test. 44 (2015) 125e134. , F. Bridier, P. Bocher, An open source software for the [21] N. Vanderesse, M. Lagace measurement of deformation fields by means of digital image correlation, Microsc. Microanal. 19 (2013) 820e821.