You are on page 1of 13

Available online at www.sciencedirect.

com

Computational Materials Science 42 (2008) 679691 www.elsevier.com/locate/commatsci

Mechanical modelling of monolament technical textiles


Valter Carvelli *, Carola Corazza, Carlo Poggi
Department of Structural Engineering, Technical University (Politecnico) of Milan, Piazza Leonardo da Vinci 32, 20133 Milano, Italy Received 11 June 2007; received in revised form 3 October 2007; accepted 16 October 2007 Available online 28 November 2007

Abstract The mechanical behaviour of textiles is often of great importance in several industrial applications. This is the case for textile reinforcement used in composites or for technical textiles. In composites manufacturing and in particular in the processes of forming and consolidation of composites, properties as in-plane shear and tensile behaviour are, together with the transversal compaction of the yarns, the most important features. In the case of monolament technical textiles used in the screen printing industry, mechanical properties such as the response of the fabric to contemporary actions of in-plane tensile and local bending are considered of primary interest. In all the cases the experimental characterisation is not sucient and must be integrated with analytical or numerical models of the fabrics. This paper presents a numerical model to evaluate the mechanical properties of dry monolament technical textiles. At the same time an experimental campaign was undertaken both to evaluate the main mechanical features and to calibrate the numerical model. Several monolament textile geometries were examined and the tests were carried out on single bres, to determine their tensile and friction properties, and on specimens cut from the textile to determine both the axial and biaxial behaviour. The macroscopic mechanical features of the monolament textiles are obtained at the mesoscale level with a model of the representative volume (RV) assuming the periodicity in the textile geometry. The resulting macroscopic mechanical properties of the textile are used in the simulation of a structural cylindrical component for screen printing applications. The proposed numerical model appears to be a tool for the design of various applications of technical textiles where the mechanical aspects are essential. 2007 Elsevier B.V. All rights reserved.
PACS: 02.70.Dh; 62.20.-x Keywords: Monolament textiles; Mechanical properties; Experimental results; Finite element model

1. Introduction Man-made bres account for about 65% of all bres used in textile applications and technical bres are today a very important area in this sector. Since 1960, particularly in industrial economies, technical bres have become the fastest growing sector of the bres industry. Among the various technical bres, the following can be reminded: high tenacity yarns of polyester; yarns and bres of aramid and polyamide; polypropylene yarns and staple bres for uses outside the traditional spinning technologies, as, for

Corresponding author. Tel.: +39 02 2399 4354; fax: +39 02 2399 4369. E-mail address: valter.carvelli@polimi.it (V. Carvelli).

example, nonwovens. Today, nanotechnology based bres, are another rapid developing reality. All these bres are used to produce technical textiles for several industrial applications where their technical properties are an essential aspect and, in many cases, the mechanical behaviour is the most important factor. Examples are given by air-bags, protective clothes, lter fabrics, paper-making machines, clothing, conveyor belts and textiles for screen printing, that are the subject of this study. In all these cases the experimental results are not sucient for a complete mechanical characterisation and must be integrated with analytical or numerical models. Dierent experimental methods are available and some of them are reported in standards while others are still object of research, e.g. for biaxial [15] and shear tests [1]. The

0927-0256/$ - see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.commatsci.2007.10.003

680

V. Carvelli et al. / Computational Materials Science 42 (2008) 679691

response of these last two tests play an important role in the forming of three-dimensional composite parts. The results of the experimental characterisation are essential also for the validation of numerical models. In the last two decades several models for textile composite reinforcements were published in the literature while only few were proposed for monolament technical textiles. These models are based both on analytical energy methods (see e.g. [610]) or nite element theory (see e.g. [1114]) and applied mainly to textiles subject to axial or biaxial loads. Models for shear behaviour are more complicated since they must also include particular aspects such friction between bres and yarns, bending and variable yarn crosssections [1520]. This paper examines the behaviour of monolament technical textiles, typical in the screen printing industry, and presents both the results of an experimental campaign and a numerical modelling. In this eld, the deformability and strength of the textile are essential aspects that deserve adequate attention since the quality of the nal print is strongly aected by the shape regularity of the textile cells. This obviously depends upon several factors like the initial deformations imposed to the fabric during the installation of the textile into the supporting frame, the irregular strains produced in the vicinity of the boundaries and the local strains produced by the tool to spread the ink on the textile. Therefore, a full knowledge of the mechanical properties of the textile is mandatory for modelling all the above aspects. The mechanical behaviour of the monolament textiles is here studied by means of numerical analysis of a representative volume (RV) following the approach implemented and described in [21] for textile composite materials. The geometry of the RV model was reproduced on the basis of the measures obtained from microscopic digital pictures and with the assumption of regular distribution of the bres. With the proposed numerical model it is possible to produce a ctitious homogeneous material equivalent, from the mechanical point of view, to the original textile. The macroscopic equivalence is obtained from the features of the textile at the mesoscale on the basis of the homogenization theoretical background [22]. This ctitious homogeneous material may be used in the FE model of textile structures. Two dierent monolament polyester plain-weave textiles were examined in this work. For both textiles, an experimental investigation (Section 2) was carried out in accordance to international standards (where available), and the nite element model of the representative volume was used for the numerical predictions (Section 3). Finally, the macroscopic mechanical features predicted with the FE analyses of the textile RV were used to simulate a textile cylinder often used in screen printing applications (Section 4). 2. Geometric and mechanical properties of the textiles The rst part of this study was dedicated to the denition of the geometry of the textiles here considered. Several

samples of the textiles were observed with a microscope to measure both the geometric features and the permanent deformations generated in the bres by the manufacturing process. In fact, after the weaving process, the bres assume a crimped form that depends on their longitudinal and transversal stiness. At the crossover points, the bres are subject to evident ovalization and the contact zone is a function of this deformability. It should be pointed out that the examined textiles were subject to a nal thermal treatment aimed to produce bres adhesion at the crossovers and, as a consequence, local permanent deformations were generated. Two monolament technical textiles were investigated. Both textiles present a typical plain-weave geometry formed interlacing polyester bres with the same geometry in two perpendicular directions. The geometric features reported in Table 1 are the nominal geometric values provided by the producer. The fabrics dier in the bre diameters and in the number of bres per unit length. The nominal dimensions of the free zone in each cell are equal to 32 and 97 lm for textiles A and B, respectively. It should be reminded that these geometric properties are important in the screen printing process since inuence the dimensions of the ink drop. 2.1. Geometric properties of the fabrics An optical microscope was used to measure the geometry of the yarns and the distance between the bres (see Fig. 1). Table 2 reports the average values of the geometrical measurements that were estimated processing the digital images. A scanning electronic microscope (SEM) was used to estimate the ovalization of the bre cross-sections measuring the ratio between the maximum (Dmax) and minimum (Dmin) diameters. The measurement were taken from SEM images of the transversal section of the bres but only the sections where the bre axis was orthogonal to the plane of the picture were used. Fig. 2 shows two bundles of bres for a qualitative evaluation. It looks evident that the ovalization of the bres is more pronounced for the bres in the warp direction where the ratio of the maximum and minimum diameters is greater than 30% (Table 3). This fact reects the inuence of the production technique: the weft bres are stretched and kept straight during the weaving process while the bres in the warp direction are bent on the weft ones [1]. The ovalization is more pronounced in the vicinity of the crossovers, this is due both to the local deformation
Table 1 Nominal features of the monolament technical textiles Textile Direction Nominal bres diameter (lm) 34 64 Number of bres per cm 150 62

A B

Warp and weft Warp and weft

V. Carvelli et al. / Computational Materials Science 42 (2008) 679691

681

Fig. 1. Cross-section of textile (a) A and (b) B in the warp direction.

Table 2 Geometrical features of the textiles (see Fig. 1) Textile A B Direction Warp Weft Warp Weft k (lm) 24.3 28.5 47.8 53.9 p (lm) 67 67 161 161

Table 3 Determination of bre ovalization (Dnom, Dmax and Dmin are the nominal, maximum and minimum diameters, respectively) Textile A B Direction Warp Weft Warp Weft Dnom (lm) 34 34 64 64 Dmax (lm) 44.92 42.67 73.22 77.83 Dmin (lm) 34.01 37.97 55.81 65.04 Dmax/Dmin 1.32 1.12 1.31 1.20

of the bres mentioned above and to the nal thermal treatment of the textile applied to produce the adhesion of the orthogonal bres. It was not possible, at the current phase of the research, to evaluate the two eects separately. As a consequence, dierent values of the ovalization were assumed for the bres in warp and weft directions, respectively but in both cases this value was kept constant along the axis of the bre. The average values of the measurements of the elliptical cross-section (Table 3) were used in the geometry of the numerical model. Another aspect that can be deduced from Fig. 1a is that the bending rigidity of the bres is not negligible since the bre is not straight in the free zone between the contact points. This aspect is less evident in Fig. 1b, where the free zone is larger. However, as shown in Section 3, it is not necessary to determine the bending rigidity of the bres for the adopted numerical model. Several geometrical models to describe the bre geometry and the contact points of an elementary volume are known from the literature starting from the rst attempt of Peirce [23] based on two circular paths connected by a straight line. More recently lenticular shapes have been used to represent bre cross-sections (see e.g. [10]). This permits to model also the attening of the yarns in the contact zone.

2.2. Friction between bres For completeness the friction coecient between bres was determined experimentally even if this may be irrelevant when axial and biaxial loadings are applied to a plain-weave textile in the principal directions. In fact, the relative sliding of the bres is negligible in these cases. On the contrary, the friction between bres at the crossovers is an important aspect both in the manufacturing process of the textile and when cyclic loadings or shear deformations are applied. This factor could be taken into account in the numerical analysis assuming a friction model at the contact zones between bres. In this case it becomes necessary to determine the static friction coecient but this is not immediate. On the contrary, there are several methods to determine the dynamic friction coefcient. In our case this was measured according to the Twisted strand method, a standard test proposed in the ASTM D3412-01 (Standard Test Method for Coecient of Friction, Yarn to Yarn), assuming a wrap angle of 900. The dynamic friction coecient was evaluated by the relation

Fig. 2. Textile A. (a) Overview of the textile; (b) cross-section of the bres in the warp direction and (c) weft direction.

682

V. Carvelli et al. / Computational Materials Science 42 (2008) 679691

l ln

  T2 h1 ; T1

where T1 is the mean input tension, T2 the mean output tension and h the wrap angle. The average value of the friction coecient, obtained from ve tests, is listed in Table 4 together with the relevant coecient of variation. Since a correlation between the dynamic and the static friction coecients for the specic material under consideration is unknown, to authors knowledge, some tentative values were used in the analyses starting from the value in Table 4 as further described in Section 3. 2.3. Mechanical properties of bres and textiles The experimental activities were carried out both on single bres and textile specimens. Tensile tests were performed on the bres while both uniaxial and biaxial tensile tests were carried out on specimens of the textiles. The tractions on bres were performed using a uniaxial testing machine MTS Synergie 200H equipped with a load cell with maximum capacity of 100 N. The uniaxial and biaxial tensile tests on textiles were carried out using a specialised device described in Section 2.3.2. 2.3.1. Tensile tests of the bres A preliminary study was performed to evaluate the different behaviour of the bres before and after the weaving process. For this reason, both virgin bres and bres extracted from the textiles were tested. As explained in the previous chapter, the bres are subject to permanent deformations during both the weaving and the thermosetting processes. These produce a variability of the shape of the cross-section and, probably, a local
Table 4 Results from the friction coecient tests Fibrebre average dynamic friction coecient Coecient of variation (%) 0.112 3.2

detrimental eect of the mechanical properties at the crossovers. The last aspect is not easy to investigate experimentally but would deserve attention in future developments of the research. For these reasons in what follows the tensile

Fig. 4. (a) Biaxial tensile device; (b) set of hinges for clamping and (c) cruciform specimen for biaxial traction.

800 Fibres 34m 600

800 Fibres 64m 600

Stress [MPa]

400

Stress [MPa]
warp weft

400

200

200

warp weft

0 0 10 20 30

0 0 10 20 30 40

Strain [%]

Strain [%]

Fig. 3. Experimental tensile tests of bres extracted from textiles. Stress vs. strain of the bres with diameter (a) 34 lm and (b) 64 lm.

V. Carvelli et al. / Computational Materials Science 42 (2008) 679691

683

properties of the bres extracted from the textiles are used in the numerical analyses as average constitutive behaviour. The specimen preparation and the tensile tests were carried out in accordance to the standards ASTM D 3822-96 (Standard Test Method for Tensile Properties of Single Textile Fibres). Several bres extracted from warp and weft directions were tested for each type of textile. Some tests were carried out at dierent stretch rates in order to evaluate the sensitivity of the polymer to the strain rates but the results were insensitive to the testing speed if this was contained below 500 mm/min. Finally, a constant displacement rate of 150 mm/min, as suggested in the standards, was applied during the test. Some typical stressstrain curves of the bres are reported in Fig. 3. The stress was evaluated using the nominal cross-section, while the strain was recorded during the experiments on a base length of 200 mm. All the bres show a non-linear response. The bres in the weft direction provide a linear behaviour up to a stress equal to 60% of the ultimate capacity. On the contrary the warp bres present a non-linear behaviour even at lower levels of the stress. This is due to the permanent deformations produced by the weaving process that are more evi-

dent in the warp bres. The initial stiness of the weft bres was higher in both cases.

145

60

145mm 25

25mm

15

r=20 15

markers

clamping zones

Fig. 5. Geometry of the specimen for biaxial test.

2.3.2. Biaxial tensile tests of the textiles The biaxial tensile tests were performed using an homemade device equipped with two independent orthogonal axes (see Fig. 4a). Along each direction the load is applied by a pair of beams that can translate parallel to the edges of the specimen that is placed in the centre of the device. Each beam can slide on two rails (one per direction) by means of two screw jacks. Two independent servo-motors are linked to the screws in order to apply the load with a controlled speed in the range 1280 mm/min. The clamping system consists of four sets of hinges on each specimen sides to permit the transversal displacements in the direction orthogonal to the load and to avoid transversal contraction (see Fig. 4b). Each clamping system is equipped with four load cells of 1.25 kN each. The maximum load per direction is 5 kN. The specimen shape is cruciform (see Fig. 4c). A typical example of the geometry is depicted in Fig. 5 (see e.g. [3,5]). A set of markers was reported in the central part of the specimen at a proper distance from the clamped edges. The displacements of the markers during the loading process were recorded by means of a digital camera at regular intervals during the test. The post processing of the digital images is carried out by a specic software that provides the maps of the displacements in the central area of the specimen. Biaxial tensile tests were performed on textile A and B with a constant speed rate of 15 mm/min in each load direction. Fig. 6a and b report typical diagrams of the biaxial load and the corresponding strains as obtained from the optical measurements. The positions of the external markers are depicted in Fig. 7 in the undeformed state and at failure (strain %8.5%) for a specimen of textile A. The comparison of the marker positions during the test shows the correct deformation of the specimen centre zone. The strain

145mm

60

145

500 Textile A 400

500 Textile B 400

Force [N]

300 200 100 0 0 3 6 9 12 15 warp weft

force [N]

300 200 100 0 0 3 6 9 12 15 warp weft

Strain [%]

Strain [%]

Fig. 6. Experimental biaxial tensile test. Force vs. strain of: (a) textile A and (b) textile B.

684

V. Carvelli et al. / Computational Materials Science 42 (2008) 679691

components distribution at failure are depicted in Fig. 8a, b and c. These pictures represent the GreenLagrange strain components assuming bi-quadratic displacement functions tting the displacement measurements at the marker points as proposed in [24]. The values of the shear strain compo-

200 Textile A

Weft direction [pixels]

100 0% 8.5% -100

nent are negligible, this conrms the parallelism of the applied loads and the warp and weft bres directions. Another important mechanical property is the ultimate strength of the textile subject to biaxial loadings. The ultimate domain of each textile can be obtained applying different ratio of loads in the warp and weft directions. The domains are expressed in terms of the stresses, estimated using the nominal cross-section of the bres, in the warp and weft directions (Fig. 9). In the same gure a TsaiHill limit curve (see e.g. [25]), is reported for a plane stress case and assuming null shear stress (r12 = 0). This may be written Ar2 Br2 Cr1 r2 1; 1 2 2

where indexes 1 and 2 refer to the weft and warp directions, respectively, while the coecients have the following expression: 1 A 2 ; r1u 1 B 2 ; r2u 1 C
r2 1ub r2 1u

-200 -200

r2 2ub r2 2u

-100

100

200

r1ub r2ub

Warp direction [pixels]


Fig. 7. Experimental biaxial tensile test of textile A. Position of the external markers in the undeformed state (0%) and at failure (strain %8.5%).

where r1u and r2u are the ultimate stresses in a uniaxial traction in the weft and warp direction, respectively; r1ub and r2ub are the ultimate stresses in a biaxial traction in the weft and warp directions.

15

15

Weft direction [mm]

5 0 -5 -10 -15 -15 -10 -5 0 5 10 15

Weft direction [mm]

10

10 5 0 -5 -10 -15 -15 -10 -5 0 5 10 15

Warp direction [mm]


15

Warp direction [mm]

Weft direction [mm]

10 5 0 -5 -10 -15 -15

-10

-5

10

15

Warp direction [mm]


Fig. 8. Experimental biaxial tensile test of the textile A. Maps of the strain components at failure in the specimen centre zone: (a) strain in the warp direction; (b) strain in the weft direction; (c) shear strain in the textile plane.

V. Carvelli et al. / Computational Materials Science 42 (2008) 679691

685

600

600

experimental Tsai-Hill

experimental Tsai-Hill

Stress weft [MPa]

400

Stress weft [MPa]

400

200

200

Textile A
0 0 200 400 600 0 0

Textile B
200 400 600

Stress warp [MPa]

Stress warp [MPa]

Fig. 9. Failure domain of: (a) textile A and (b) textile B.

Table 5 TsaiHill failure domain parameters r1u (MPa) Textile A Textile B 528 473 r2u (MPa) 488 441 r1ub (MPa) 338 340 r2ub (MPa) 242 250

These TsaiHill curves seems to t the experimental data with a good level of approximation assuming the parameters detailed in Table 5 extracted from the uniaxial and biaxial tensile tests. 2.3.3. Uniaxial tensile tests of the textiles The uniaxial tensile tests were carried out in accordance to the standards ASTM D 5035-95 (Standard Test

500

Textile A
400

Force [N]

300 200 100 0 -10 -8 -6 -4 -2 0 10 20 30 40 warp weft

Transv.strain [%]
500

Longit.strain [%]

Textile B
400

Force [N]

300 200 100 0 -10 -8 -6 -4 -2 0 10 20 30 40 warp weft

Transv.strain [%]

Longit.strain [%]

Fig. 10. Uniaxial tensile test in the warp and weft directions. Force vs. Longitudinal and Transversal Strain of textiles A (a) and B (b).

686

V. Carvelli et al. / Computational Materials Science 42 (2008) 679691

Method for Breaking Force and Elongation of Textile Fabrics, Strip Method), using the same biaxial device described above. Rectangular strips were cut parallel to the principal directions, 50 mm wide and 300 mm length. For textile A, other specimens with the bres oriented 45 to the load direction were tested. A loading rate of 150 mm/min was used during the tests. The tensile force vs. strain curves reported in Figs. 10 and 11 were produced using the digital images recorded during the tests. The textiles exhibit a non-linear response and a dierent behaviour in the weft and the warp directions with a higher elastic stiness in the weft one. When the bres are oriented 45 to the load direction the typical non-linear shape of the force vs. shear strain curve is obtained (see Fig. 11) in which two distinct regions are identied [1]. The initial slope of the curves in Fig. 11 is not negligible due to the adhesion of the bres at the crossovers after the thermal treatment. 3. The nite element textile model A three-dimensional nite element model was implemented to predict the mechanical behaviour of monolament technical textiles. The use of this model is complementary to
500

the experimental tests since this permits both to study some local phenomena (like stress concentrations) and to determine the inuence of some geometrical features and material mechanical parameters on the overall mechanical behaviour of the textile. The numerical model is based on the main assumption of regular distribution of the bres (see Fig. 12a). Because of the periodicity in the textile, the study can be limited to the representative volume (RV). From the numerical results of the RV analysis, the macroscopic properties of the equivalent material are evaluated by averaging the quantities on the RV volume [22]. The representative volume was modelled using the geometrical quantities measured on the specimens of the two textiles (see Tables 2 and 3). The longitudinal shapes of the bres in the RV were dened on the basis of the microscope measurements (Table 2) that were tted with a trigonometric function. The transversal cross-section of the bres in the RV were assumed with elliptic shape adopting the data of Table 3. As explained in Section 2.1, this shape was kept constant in the bre longitudinal direction. In order to reproduce the textile periodicity (hypothesis of regular distribution of the bres in the textile), the following kinematic boundary conditions were imposed to the nite element model of the RV [21]. These are written in term of the displacement vector u (see Fig. 12b): uA uB ExA xB ; uC uD ExC xD ; 4a 4b

Textile A
400

Force [N]

300 200 100 0 0 20 40 60 80

where A and B are two points of a weft bre on the crosssections x1 = 0 and x1 = 1 corresponding in the periodicity;
Table 6 RambergOsgoods parameters for the bre constitutive behaviour used in the nite element models Fibres of textile A Weft E (MPa) a n r0 (MPa) 4600 0.544 3.254 697 Warp 3000 0.132 5.405 681 Fibres of textile B Weft 2900 0.185 5.77 565 Warp 1850 0.008 19.228 523

Shear strain [%]


Fig. 11. Uniaxial tensile test with bres oriented 45 to the load direction. Force vs. Shear strain of textile A.

RV

Fig. 12. (a) Representative volume (RV) in the textile; (b) some geometrical parameters of the RV.

V. Carvelli et al. / Computational Materials Science 42 (2008) 679691

687

C and D are two points of a warp bre on the cross-sections x2 = 0 and x2 = 2 corresponding in the periodicity. E represents the macroscopic strain tensor (see e.g. [21]). Two of the imposed kinematic conditions are referred to the interaction between bres: one enforces no penetration

between bres according to the exponential model implemented in the FE code [26]; the other imposes the Coulomb friction model for brebre contact. The nite element mesh of the RV of textile B has 6612 nodes and 26880 four nodes tetrahedral elements (C3D4)

800

800

Fibres 34m
600 experimental 600

Fibres 64m
experimental

Stress [MPa]

400 analytical 200 warp weft 0 0 10 20 30

Stress [MPa]

400 analytical 200 warp weft 0 0 10 20 30 40

Strain [%]

Strain [%]

Fig. 13. Comparison between the experimental tensile tests of the bres and the analytical RambergOsgood tting. Stress vs. strain of the bres with diameter (a) 34 lm and (b) 64 lm.

500 400

Textile A warp

numerical experimental

Force [N]

300 200 100 0 -10 -8 -6 -4 -2 0 10 20 30 40

Transv. strain [%]


500

Longit. strain [%]


numerical experimental

Textile A weft
400

Force [N]

300 200 100 0 -10 -8 -6 -4 -2 0 10 20 30 40

Transv. strain [%]

Longit. strain [%]

Fig. 14. Comparison between numerical simulations and experimental uniaxial tensile tests along (a) warp and (b) weft directions. Force vs. Strain of textile A.

688

V. Carvelli et al. / Computational Materials Science 42 (2008) 679691

while the RV of textile A has 6380 nodes and 25920 four nodes tetrahedral elements (see Fig. 17). The three-dimensional numerical simulation of a monolament textile requires the constitutive behaviour of the bres to be as close as possible to the real one. The analytical RambergHosgood non-linear constitutive model was used to t the experimental data. In the uniaxial state, the RambergHosgoods equation is  n1 ! jrj Ee r 1 a ; 5 r0 where E is Youngs modulus (dened as the initial slope of the stressstrain curve), a the yield oset (variation coecient of the initial stiness), r0 yield stress and n is the hardening exponent for the plastic (non-linear) term. The E, a, r0 and exponent n were obtained tting the experimental results; their values are reported in Table 6 for the bres constitutive behaviour used in the numerical predictions. The comparison between the experimental results and the analytical interpolations are given in Fig. 13.

3.1. Applications and comparisons The validation of the proposed nite element approach was conducted comparing the numerical results to the experimental observations for the two textiles both in the elastic range and up to failure. The ultimate failure was simulated assuming a strain based failure criterion, i.e. a bre fails when the average longitudinal strain in a crosssection reaches the ultimate experimental strain. Even if this is a simplied criteria, the obtained results were satisfactory. Future developments of the model could take into account more rened criteria based on strain or stress components interaction. As previously mentioned, the considered textiles presented bres adhesion at the crossovers due to the nal thermal treatment but this aspect was not initially included in the numerical model. It was preferred to produce a more general model to consider also the friction between bres. A preliminary parametric study was carried out to understand the inuence of the static friction coecient on the mechanical response of the textile but, as it is shown in Section 2.2, only the dynamic friction coecient was determined experimentally. For this reason, the static friction

500 400
Textile B warp

Force [N]

300 200 100 0 -10 -8 -6 -4 -2 0 10 20 30 40

numerical experimental

Transv. strain [%]


500

Longit. strain [%]

Textile B

400

weft

Force [N]

300 200 100 0 -10 -8 -6 -4 -2 0 10 20 30 40

numerical experimental

Transv. strain [%]

Longit. strain [%]

Fig. 15. Comparison between numerical simulations and experimental uniaxial tensile tests along (a) warp and (b) weft directions. Force vs. Strain of textile B.

V. Carvelli et al. / Computational Materials Science 42 (2008) 679691

689

500

Textile A
400 experimental

warp weft

500

Textile B
400 experimental

Force [N]

300 200 100 0 0 3 6 9 12 15 numerical

Force [N]

300 200 numerical 100 0 0 3 6 9 warp weft 12 15

Strain [%]

Strain [%]

Fig. 16. Comparison between numerical simulations and experimental biaxial tensile tests along warp and weft directions. Force vs. strain of textile (a) A and (b) B.

coecient was initially assumed equal to the dynamic and then increased with regular steps up to a value ten times greater than the dynamic friction coecient. The analyses were carried out on the RV models loaded both in uniaxial and biaxial tension. In conclusion, the results were insensitive to dierent values of the static friction coecient. This conrm the initial expectations based on the consideration that the friction should not be inuent in a plain-weave textile subject to traction in the direction of the bres. On the contrary it is believed that the friction has a relevant inuence in the simulation of the shear loading [1] but this case was not examined in this work. Figs. 14 and 15 show the numerical macroscopic mechanical response of the RV subject to uniaxial traction. Both the longitudinal (load direction) and the transversal strains are predicted. For textile B a satisfactory agreement with the experiments is obtained (Fig. 15). On the contrary a disagreement between numerical and experimental results is observed for textile A (Fig. 14). This could be probably improved with a better simulation of the constitutive behaviour of the bres. The results of the biaxial simulations are reported in Fig. 16a and b in terms of force vs. strain curves of both textiles A and B and compared to the experimental values. The numerical curves are in good agreement with the experimental data, in particular in the weft direction. From the FE model is possible to estimate also the local deformations of the bres and, as a consequence, the deformation of the free space in the unit cell (see e.g. Fig. 17). This kind of observations are useful to the technical textile industry to evaluate the inuence of the material properties and the geometry of the bres to the textile shape. In particular, in a screen printing process the size of the ink drops are inuenced by the shape of the unit cell of the textile and this is determined by the tension applied to the frame where the textile is xed. Therefore the model seems to be adoptable as a design tool to optimise the bre dimensions, the geometry and the material properties of monolament technical textiles.

4. Mechanical properties of a textile cylinder To verify the applicability of the proposed model to practical industrial cases, the mechanical response of a textile component employed in a screen printing technique was investigated. The cylindrical textile component is made with the textile A analyzed in the previous section. The

Fig. 17. Undeformed (white) and deformed (black) geometry of textile A during the biaxial loading.

Fig. 18. Textile cylindrical component.

690

V. Carvelli et al. / Computational Materials Science 42 (2008) 679691

Fig. 19. Finite element model of the textile cylindrical component.

Table 7 Elastic mechanical properties of the homogenized textile EC = 2972 MPa EZ = 4456 MPa mCZ = 0.32 mZC = 0.31 G = 115 MPa

model of the whole cylinder was made using shell nite elements and assuming an homogeneous orthotropic material with the mechanical properties obtained from the numerical analysis of the RV. The geometry of the textile structure is represented in Fig. 18. Two metallic rings are placed at the top and bottom of the cylinder to give the regular shape to the textile. The two rings are linked by two longitudinal bars. The cylinder was made from a rectangular piece of textile that was wrapped around the cylindrical frame. The overlapping was parallel to the cylinder axis. The cylinder has a length of 700 mm and a diameter equal to 201 mm. The nite element model of the cylindrical component was made with 3-noded shell elements using a commercial nite element code [26]. The imposed constraints are (see Fig. 19): clamping in the bottom ring zone; no radial displacement in the top ring zone; imposed displacement along the axial direction (10 mm) at the top circumference.

The numerical analysis is limited to the elastic behaviour of the cylinder subject to axial tension. The homogenized mechanical properties of the textile, as obtained by the numerical analyses of the RV (Fig. 14), are listed in Table 7. EZ and EC represent the elastic stiness in the axial and circumferential direction, respectively, and mCZ and mZC the Poissons coecients. The initial shear modulus G was set equal to the one obtained by the tensile tests of the specimen with bres at 45 to the load direction (see Fig. 11). To verify the inuence of the longitudinal overlapping to the global response of the cylinder, two limit situations were simulated: perfect sticking (continuous surface) and unbounded textile boundaries in the overlapping zone (discontinuous surface) (see Figs. 18 and 19). In Fig. 20 the contour plot of the displacement magnitude is depicted on the deformed shape for both conditions. The maximum radial displacement, recorded in the middle cross-section, was equal to 6.68 mm for the cylinder with a continuous surface and 8.62 mm considering a discontinuous surface (see Fig. 21). These results are astride the experimental radial displacement measured by the producer that was equal to 8 mm. The pictures in Fig. 21 represent the displacement vectors of the middle cross-section for the two cases studied. Fig. 21a does not show the expected axisymmetry of the displacement eld on the perfect cylinder. This may be

Fig. 20. Contour plot of the displacement magnitude on the deformed shape: (a) perfect sticking in the overlap zone and (b) unbounded textile boundaries in the overlap zone.

V. Carvelli et al. / Computational Materials Science 42 (2008) 679691

691

Fig. 21. Displacements of the middle transversal cross-section: (a) perfect sticking in the overlap zone and (b) unbounded textile boundaries in the overlap zone.

due to an irregular distribution of the elements during the automatic generation of the FE mesh. The axial symmetry is slightly violated in the middle part of the cylinder and is perfectly satised in the top and bottom zone. The value of the stresses were not a critical issue of this analysis and are not reported. In fact the maximum von Mises stress was equal to 55 MPa that is negligible if compared to the ultimate strength of the bres (see Fig. 3). 5. Conclusions Experimental and numerical experiences to study the mechanical properties of monolament plain-weave textiles were presented. The numerical modelling is based on the nite element method. Uniaxial and biaxial tensile tests were simulated and compared to the experimental data. The comparison of the mechanical behaviour of the two textiles and of the textile component considered is satisfactory. It should be underlined that the proposed numerical approach was implemented using a commercial nite element code. The industrial designers could easily adopt similar procedures to study and optimize technical textiles varying the mechanical and geometrical properties according to the specic application. Further numerical developments of the instigation could include: the implementation of a non-constant ellipticity of the bre cross-section along the axial direction, as highlight in the microscopic observations; dierent weaving patters (e.g. twill); numerical simulation of in-plane shear loading. Acknowledgement The authors wish to acknowledge the nancial support of SAATIPRINT S.p.A. (Appiano Gentile, Italy). References
[1] A.C. Long, Design and manufacture of textile composites, Woodhead Publishing Ltd., 2005.

[2] Y. Luo, I. Verpoest, Composite Part B 33 (2002) 197203. [3] J.S. Welsh, D.F. Adams, Composite Part A 33 (2002) 829839. [4] E.J. Sturrock, C. Boote, G.E. Attenburrow, K.M. Meek, Journal of Materials Science 39 (2004) 24812486. [5] M. Geiger, W. Hunatter, M. Merklein, Journal of Materials Processing Technology 167 (2005) 177183. [6] S. Kawabata, M. Masako Niwa, H. Kawai, Journal of Textile Institute 64 (1973) 2146. [7] S. Kawabata, M. Masako Niwa, H. Kawai, Journal of Textile Institute 64 (1973) 4761. [8] S.V. Lomov, G. Huysmans, I. Verpoest, Textile Research Journal 71 (2001) 534543. [9] S.V. Lomov, T. Truong Chi, I. Verpoest, T. Peeters, D. Roose, Ph. Boisse, A. Gasser, International Journal of Forming Processes 6 (2003) 413442. [10] T.V. Sagar, P. Potluri, J.W.S. Hearle, Computational Materials Science 28 (2003) 4962. [11] A. Gasser, P. Boisse, S. Hanklar, Computational Materials Science 12 (2000) 720. [12] P. Boisse, K. Buet, A. Gasser, J. Launay, Composites Science and Technology 61 (2001) 395401. [13] P. Boisse, A. Gasser, G. Hivet, Composite Part A 32 (2001) 13951414. [14] M. Tarfaoui, J.Y. Drean, Textile Research Journal 71 (2001) 790795. [15] S. Kawabata, M. Masako Niwa, H. Kawai, Journal of Textile Institute 64 (1973) 6185. [16] T.M. McBride, J. Chen, Composites Science and Technology 57 (1997) 345351. [17] U. Mohammad, C. Lekakou, L. Dong, M.G. Bader, Composites Part A 31 (2000) 299308. [18] L. Liu, J. Chen, X. Li, J. Sherwood, Composites Part A 36 (2005) 105114. [19] S.V. Lomov, I. Verpoest, Composites Science and Technology 66 (2006) 919933. [20] P. Badel, E. Vidal-Salle, P. Boisse, Computational Materials Science 40 (2007) 439448. [21] V. Carvelli, C. Poggi, Composite Part A 32 (2001) 14251432. [22] S. Nemat-Nasser, M. Hori, Micromechanics: Overall Properties of Heterogeneous Solids, Elsevier Science Publishers, 1999. [23] F.T. Peirce, Journal of Textile Institute 28 (1937) 4596. [24] V. Quaglini, C. Corazza, C. Poggi, Composites Part A (2007), doi:10.1016/j.compositesa.2007.07.008. [25] I.M. Daniel, O. Ishai, Engineering mechanics of composite materials, Oxford University Press, 1994. [26] ABAQUS Standard Users Manual, rel. 6.4 (2003), Hibbit, Karlsson & Sorensen Inc.