Professional Documents
Culture Documents
Composites: Part A
journal homepage: www.elsevier.com/locate/compositesa
a r t i c l e i n f o a b s t r a c t
Article history: The use of long natural fibres (LNF) as reinforcement in composite systems for structural applications has
Received 20 January 2017 been steadily growing in the automotive and construction industries as these materials offer sustainabil-
Received in revised form 6 March 2017 ity benefits combined with high specific strength and stiffness. However, the performance of natural
Accepted 9 March 2017
fibres has been questioned by a high variability in their mechanical properties and design data for struc-
Available online 11 March 2017
tural reliability analysis of LNF composites are not yet available. Here, we present a statistical study of the
elastic modulus, strength and failure strain of a comprehensive set of LNF composite systems. We have
Keywords:
found that the variability of LNF laminate properties is similar to that of carbon fibre laminates. We pro-
A. Natural fibres
A. Laminates
vide recommendations to apply the statistical parameters determined here to the design of natural fibre
B. Mechanical properties composite structures. Our findings provide a deeper understanding of LNF composites reliability and are
C. Statistical properties important for the further acceptance of these materials by the industry.
2017 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.compositesa.2017.03.010
1359-835X/ 2017 Elsevier Ltd. All rights reserved.
100 J.P. Torres et al. / Composites: Part A 98 (2017) 99104
probabilistic design methodologies allow to quantify the inherent wider range of LNF reinforced composite laminate systems. Further
risk of failure. However, a significant drawback of this approach understanding of these systems can help in enhancing design and
is the large experimental data that is required to establish input reliability assessment practices in natural fibre composites to
distributions. extend their use in commercial applications.
So far, probabilistic design in composites has been exclusively
used for carbon fibre and glass fibre composites. Conversely, the 2. Experimental
use of natural fibres as reinforcement for structural applications
is only beginning to emerge [22,23]. The variability of plant fibre 2.1. Composite laminate tests
properties is influenced by variables such as chemical composition
[24], plant species, location of fibre in the plant, environmental An extensive experimental program was carried out to deter-
conditions in the cultivation area, and fibre processing methods mine the statistical distributions of mechanical properties. Several
[25]. In addition, the inherent irregularity of shape and cross- material and geometrical configurations were studied. Reinforce-
sectional area in natural fibres further contributes to their variabil- ment fibres include unidirectional flax fabrics FLAXPLY provided
ity. These irregularities create experimental difficulties in the mea- by LINEO [37], and jute and flax woven fabrics provided by Biotex
surement of fibre endogenous parameters such as diameter or [38]. The resin system used in all laminates is a SR-8100 Epoxy
lumen size. This requires the use of techniques such as high reso- resin produced by Sicomin [39], except for Flax-VE-0 which uses
lution dimensional measurement [26], optical, laser or electron an EPOVIAOPTIMUM KRF2000SE vinil-ester matrix provided by
microscopy [27,28], or X-ray computer tomography [10]. Several Polyint Composites Australia [40]. Fibre areal weight and fabric
investigations have quantified the variability in stiffness, strength configuration are shown in Table 1.
and failure strain of natural fibres using the coefficient of variation Composite panels were manufactured by vacuum assisted resin
C V (i.e. the ratio of standard deviation to mean value). Virk et al. transfer moulding (VARTM) on a rectangular tool plate. Before
[10] studied the fracture behaviour of different sets of 50 single infusion, the fibres were dried in an oven at 60for 6 h. After infu-
jute fibres and found maximum variabilities of 31% in stiffness, sion panels were left for 24 h at ambient temperature and post-
46% in strength, and 35% in strain to failure. Baley tested flax fibres cured at 60for 8 h under a 80 kPa vacuum. Rectangular specimens
under uniaxial tensile loading and found a variability of 28% in of 250 mm 25 mm dimensions were cut from the panels using
stiffness, 36% in strength, and 26% in strain to failure [29]. Adusu- waterjet cutting. Fibre weight before infusion and panel weight
mali et al. [30] also studied the tensile behaviour of flax fibres and after manufacturing were measured to calculate composite fibre
found a scatter 48% in stiffness, 36% in strength, and 14% in strain volume fraction according to V f wf qc =qf , where V f is the fibre
to failure. Other publications regarding the variability of single volume fraction, wf is the fibre weight fraction, qc is the composite
fibres include the work of Eichhorn and Young for hemp fibres density and qf is the fibre density. An important simplification is
[31], Xue et al. for kenaf fibres [32], and Bodros and Baley for sting- made in this calculation by assuming the composite has no voids.
ing nettle fibres [33]. However, the statistical aspects of the However, as we will show later, the conclusions reached in this
mechanical response resulting from assembling these fibre sys- paper are still valid for different volume fraction values. Further
tems into a composite laminate has not been analysed in these details including number of specimens, individual specimen geom-
investigations. Additional work by Virk et al. [34,28] and Ander- etry and testing results are detailed in the complementary Data in
sons et al. [35] have studied the strength and failure strain distri- Brief article [41]. As is the typical case in conventional VARTM pro-
butions of flax fibres using modified two-parameter Weibull cesses, fibre volume fraction and sample thickness were only par-
statistics. However, the statistical aspects of the mechanical tially controlled through the vacuum pressure level and the
response resulting from assembling natural fibre systems into a number of fibre plies [42].
composite laminate has not been analysed in these previous inves- Uniaxial tensile tests were carried out in an electromechanical
tigations (it must be noted nevertheless, that a previous study by INSTRON 5584 frame using a 30 kN load cell following the proce-
Bader et al. on the strength of carbon fibres [36] had found a higher dures depicted in [43]. Tensile specimens were supported with
weibull modulus (i.e. lower variability) of carbon fibre composite hydraulic grips with a gripping force that prevented both specimen
compared to single carbon fibres). sliding and premature failure at the grips. Engineering strain was
A previous work by the authors has studied the statistical measured using an optical extensometer. Elastic modulus was
aspects of single flax fibres and unidirectional flax reinforced com- determined from the initial slope of the experimental stress-
posites to conclude that when long natural fibres are used as rein- strain curves in the strain range 0.0010.003 mm/mm. Strength
forcements in composite laminates, the variability of the resulting and strain at failure were measured at the point corresponding
composite is significantly lower than that of the constituent fibres. to the maximum axial load in the stress-strain curves. Raw data
The present work follows up to assess the statistical behaviour of a files for all test results can be found in Ref. [41].
Table 1
Testing program for tensile tests.
that each ply in the Flax-CP laminates is non-woven (i.e. has min-
imum crimp) while plain and 2 2 twill weaves have the highest different material systems does not show a clear correlation to
crimp. On the other hand, the failure strain behaviour among the the fibre architecture.
102 J.P. Torres et al. / Composites: Part A 98 (2017) 99104
To compare the variability in elastic modulus, strength and fail- unidirectional 90 flax laminate presented strength and strain to
ure strain, we have examined the coefficient of variation C V corre- failure C V values slightly above 10%. Since the observed failure
sponding to the experimental distributions. The C V value is an mode for this system was matrix cracking, we speculate that this
appropriate comparative measure between the different materials greater variability can be attributed to a higher sensitivity of
and properties as it is a dimensionless ratio, normalized by the matrix failure initiation and sudden propagation to surface defects
property mean value, whose absolute magnitude can vary greatly. and porosity.
The results are shown in Fig. 2 and Table 2. Lower C V values imply a To compare the effect of matrix material in the composite vari-
smaller expected scattering in the corresponding variable. The ability we analysed the results for epoxy (Flax-0 ) and vinyl ester
short-fibre mat system (Flax-Short) shows the highest variability (Flax-VE-0 ) matrix composites. A comparison of their C V values
for all 3 properties. This observation can be attributed to the inho- indicates that the vinyl ester resin introduces a moderately higher
mogeneity of the mat reinforcement geometry, i.e., the uneven dis- scatter in the composite properties. Since manufacturing variables
tribution of the local mass/unit area values which gives place to a such as tool and bagging quality, vacuum level in bag, operator
non-homogeneous fibre volume fraction over the test specimen skills and layup tools can be considered constant throughout our
volume. study, the lower C V of Flax-0 compared to Flax-VE-0 suggests
With the exception of the Flax-Short system, the variability in that the fibre/resin wettability properties [49] can have a minor
elastic modulus for all materials is between 5% and 10%, which is effect on the composite part variability.
considerably lower than the reported values for single flax fibres To assess the effect of different reinforcement fibres in the com-
in the range of 2848% [29,30]. The same considerations apply posite variability, we compared the C V results for flax (Flax-0 ) and
for the scatter in strength and strain to failure. These observations carbon fibre (Carbon-0 ) reinforced unidirectional laminates.
are of great importance for the understanding of LNF reinforced Results show that the variability in the flax reinforced composite
composites reliability as the variability of their mechanical proper- is indeed lower than that of the carbon fibre laminate. Here again,
ties is significantly lower to what was expected from the previous since manufacturing variables have been kept constant, we suggest
studies of single natural fibres. that For the woven fabric the variability may arise due to differences in reinforcement per-
composites, strength systematically exhibits a lower variability meability, resin/fibre wettability and resin content [13]. Most
( 5%) than strain to failure. This suggests that strength may be important is the observation that variability in natural fibre com-
a more reliable predictor of failure in these material systems. The posite laminates is indeed in the same order of magnitude to that
of carbon fibre composites. This argument was proposed in a pre-
vious investigation by the authors, however we have now tested its
validity for two material systems under the same manufacturing
and testing conditions.
Table 2
Mean, standard deviation rsd and coefficient of variation C V values for elastic moduli, strength and fracture strain.
Material Modulus (GPa) Strength (MPa) Fracture strain (%)
Mean rsd CV Mean rsd CV Mean rsd CV
Flax-0 18.6 1.0 5.54 283.4 13.0 4.6 1.87 0.09 4.9
Flax-90 4.1 0.2 5.50 26.1 0.8 2.9 0.89 0.09 9.7
Flax-CP 8.5 0.6 6.92 147.6 8.3 5.6 2.15 0.10 4.6
Flax-Twill 6.4 0.5 8.0 90.7 3.3 3.6 2.42 0.13 5.6
Jute-Twill 6.4 0.6 9.1 63.1 3.6 5.8 1.21 0.11 9.1
Jute-Plain 6.4 0.5 7.9 66.7 3.5 5.3 1.33 0.11 8.4
Jute-Satin 8.2 0.6 7.8 102.8 3.2 5.1 1.70 0.11 6.9
Flax-VE-0 15.1 1.3 8.4 269.4 18.8 7.0 2.05 0.12 5.9
Carbon-0 120.5 11.0 9.1 1171.2 100.4 8.6 0.95 0.10 10.5
Flax-Short 4.8 0.5 11.7 41.7 6.3 15.0 1.18 0.21 17.7
J.P. Torres et al. / Composites: Part A 98 (2017) 99104 103
Table 3
Weibull distribution parameters for strength and fracture strain.
Fig. 3. (a) Stress-strain curves for 86 samples of the Jute-Plain system. (b), (c) and (d): statistical distributions together with their probability distribution function fit: (b)
elastic modulus (normal), (c) strength (weibull), (d) failure strain (weibull). (For interpretation of the references to colour in this figure legend, the reader is referred to the
web version of this article.)
using rule of mixture based models to determine elastic modulus [16] Gary P, Riskalla M. Development of probabilistic design methodology for
composite structures. Tech rep. DTIC Document; 1997.
and strength from fibre and matrix properties, and using the deter-
[17] Long M, Narciso J. Probabilistic design methodology for composite aircraft
mined values as the lm values in the analysis described above (Eqs. structures. Tech rep. DTIC Document; 1999.
(2) and 4). [18] Shiao M. Evaluation of the probabilistic design methodology and computer
code for composite structures. Tech rep. DTIC Document; 2001.
[19] Pellissetti M, Schuller G. On general purpose software in structural reliability
4. Conclusions an overview. Struct Safety 2006;28(1):316.
[20] Thacker BH, Riha DS, Fitch SH, Huyse LJ, Pleming JB. Probabilistic engineering
analysis using the nessus software. Struct Safety 2006;28(1):83107.
We have analysed the statistical nature of the mechanical prop- [21] Tvedt L. Probanprobabilistic analysis. Struct Safety 2006;28(1):15063.
erties of several natural fibre reinforced composites. The variability [22] Khalfallah M, Abbs B, Abbs F, Guo Y, Marcel V, Duval A, Vanfleteren F,
of elastic modulus, strength and failure strain using the coefficient Rousseau F. Innovative flax tapes reinforced acrodur biocomposites: a new
alternative for automotive applications. Mater Des 2014;64:11626.
of variation C V was assessed. Our results show that the variability [23] Niaounakis M. Biopolymers: processing and products. William Andrew; 2014.
of long natural fibre composites is in the same order of magnitude [24] Legras A, Kondor A, Heitzmann M, Truss R. Inverse gas chromatography for
to that of carbon fibre composites manufactured with a compara- natural fibre characterisation: identification of the critical parameters to
determine the brunaueremmettteller specific surface area. J Chromatogr A
ble process. In addition, our results indicate that short fibre com-
2015;1425:2739.
posites have a higher scatter than long fibre composites. Most [25] Charlet K, Baley C, Morvan C, Jernot J, Gomina M, Brard J. Characteristics of
importantly, we observe that the variability of long natural fibre herms flax fibres as a function of their location in the stem and properties of
composites is significantly lower to that observed in previous stud- the derived unidirectional composites. Compos Part A: Appl Sci Manuf
2007;38(8):191221.
ies of single natural fibres. [26] Legras AMD. Manufacture and characterisation of short fibre biocomposites.
We suggest that the statistical parameters of the fitted probabil- PhD Thesis. The University of Queensland. doi:http://dx.doi.org/10.14264/uql.
ity distribution functions can be used to model the scatter in sim- 2016.352.
[27] Xu XW, Jayaraman K. An image-processing system for the measurement of the
ilar material systems with the aim of determining design dimensions of natural fibre cross-section. Int J Comput Appl Technol 2009;34
allowables and perform probabilistic design analyses. These results (2):11521.
can assist in the development of design guidelines and reliability [28] Virk AS, Hall W, Summerscales J. Tensile properties of jute fibres. Mater Sci
Technol 2009;25(10):128995.
assessment practices for natural fibre composites and extend their [29] Baley C. Analysis of the flax fibres tensile behaviour and analysis of the tensile
use in commercial applications. stiffness increase. Compos Part A: Appl Sci Manuf 2002;33(7):93948.
[30] Adusumali R-B, Reifferscheid M, Weber H, Roeder T, Sixta H, Gindl W.
Mechanical properties of regenerated cellulose fibres for composites.
Acknowledgments Macromolecular symposia, vol. 244. Wiley Online Library; 2006. p. 11925.
[31] Eichhorn S, Young R. Composite micromechanics of hemp fibres and epoxy
This investigation was partially carried out under the Coopera- resin microdroplets. Compos Sci Technol 2004;64(5):76772.
[32] Xue Y, Du Y, Elder S, Wang K, Zhang J. Temperature and loading rate effects on
tive Research Centre for Advanced Composite Strutures Australia
tensile properties of kenaf bast fiber bundles and composites. Compos Part B:
Project 1.1 Plantfibre bio-composites. The authors would also Eng 2009;40(3):18996.
like to acknowledge the contribution of Sarah Schaber. This work [33] Bodros E, Baley C. Study of the tensile properties of stinging nettle fibres
was also supported by an Endeavour Postgraduate Fellowship (Urtica dioica). Mater Lett 2008;62(14):21435.
[34] Virk AS, Hall W, Summerscales J. Multiple data set (mds) weak-link scaling
granted by the Australian Government. analysis of jute fibres. Compos Part A: Appl Sci Manuf 2009;40(11):176471.
[35] Andersons J, Sparnin E, Joffe R, Wallstrm L. Strength distribution of
References elementary flax fibres. Compos Sci Technol 2005;65(3):693702.
[36] Bader MG, Pickering KL, Buxton A, Rezaifard A, Smith PA. Failure
micromechanisms in continuous carbon-fibre/epoxy-resin composites.
[1] Directive 2000/53/EC of the European parliament and of the council of 18
Compos Sci Technol 1993;48(14):13542.
September 2000 on end-of life vehicles. Off J Euro Commun.
[37] LINEO - advanced flax. <http://www.lineo.eu/products>.
[2] Kanari N, Pineau JL, Shallari S. End-of-life vehicle recycling in the european
[38] Biotex high performance natural composites. <http://compositesevolution.
union. JOM 2003;55(8):159. http://dx.doi.org/10.1007/s11837-003-0098-7.
com/biotex/>.
[3] Aylward P. A natural fit: advances in biobased fibers. Compos Manuf 2014;30
[39] Technical data sheet - SR 8100 epoxy system for injection and infusion.
(6):806.
<http://www.sicomin.com/datasheets/product-pdf94.pdf>.
[4] Saxena M, Sharma A, Pappu A, Haque R, Wankhede S. Composite materials
[40] Vinyl ester resins - polyint composites. <http://www.ccpcomposites.com.au/
from natural resources: recent trends and future potentials. INTECH Open
vinyl-ester>.
Access Publisher; 2011. [41] Torres JP, Vandi L-J, Veidt M, Heitzmann MT. Statistical behavior of the tensile
[5] La Mantia F, Morreale M. Green composites: a brief review. Compos Part A:
properties of natural fibre composites. Data in Brief (submitted for
Appl Sci Manuf 2011;42(6):57988.
publication). http://dx.doi.org/10.1016/j.dib.2017.03.043.
[6] Ku H, Wang H, Pattarachaiyakoop N, Trada M. A review on the tensile
[42] Li J. Modeling, design and control of vacuum assisted resin transfer molding
properties of natural fiber reinforced polymer composites. Compos Part B: Eng
(VARTM) for thickness variation reduction. ProQuest; 2006.
2011;42(4):85673. http://dx.doi.org/10.1016/j.compositesb.2011.01.010. [43] ASTM. Standard, D3039, standard test method for tensile properties of
[7] Koronis G, Silva A, Fontul M. Green composites: a review of adequate materials
polymer matrix composite materials. American Society for Testing Materials;
for automotive applications. Compos Part B: Eng 2013;44(1):1207. http://dx.
2003.
doi.org/10.1016/j.compositesb.2012.07.004.
[44] Optimization and root finding. <https://docs.scipy.org/doc/scipy-0.18.1/
[8] Fan M, Weclawski B. Chapter 6 - long natural fibre composites. In: Fan M, Fu F,
reference/optimize.html>.
editors. Advance high strength natural fibre composites in
[45] SciPy 0.16.0 Reference guide; July, 2015. <http://docs.scipy.org/doc/scipy/
construction. Woodhead Publishing; 2017. p. 14177. http://dx.doi.org/
reference/tutorial/stats.html>.
10.1016/B978-0-08-100411-1.00006-6. [46] NIST/SEMATECH e-Handbook of Statistical Methods; April, 2002. <http://
[9] Faruk O, Bledzki AK, Fink H-P, Sain M. Biocomposites reinforced with natural
www.itl.nist.gov/div898/handbook/eda/section3/eda3668.htm>.
fibers: 20002010. Prog Polym Sci 2012;37(11):155296.
[47] Dai S, Cunningham P, Marshall S, Silva C. Influence of fibre architecture on the
[10] Virk AS, Hall W, Summerscales J. Failure strain as the key design criterion for
tensile, compressive and flexural behaviour of 3d woven composites. Compos
fracture of natural fibre composites. Compos Sci Technol 2010;70(6):9959.
Part A: Appl Sci Manuf 2015;69:195207.
[11] Mallick PK. Fiber-reinforced composites: materials, manufacturing, and [48] Hsiao H, Daniel I. Effect of fiber waviness on stiffness and strength reduction of
design. CRC Press; 2007.
unidirectional composites under compressive loading. Compos Sci Technol
[12] Jones RM. Mechanics of composite materials, vol. 193. Washington, DC: Scripta
1996;56(5):58193.
Book Company; 1975.
[49] Van de Velde K, Kiekens P. Wettability of natural fibres used as reinforcement
[13] Robinson P, Greenhalgh E, Pinho S. Failure mechanisms in polymer matrix
for composites. Die Angew Makromol Chem 1999;272(1):8793.
composites: criteria, testing and industrial applications. Elsevier; 2012.
[50] Wolfram Research Gamma Function Evaluation; 2010.
[14] Chamis CC. Probabilistic composite design. ASTM Special Tech Publ
<http://functions.wolfram.com/webMathematica/FunctionEvaluation.
1997;1242:2342.
jsp?name=Gamma>.
[15] Chamis CC. Probabilistic design of composite structures. Glenn Research
Center, Cleveland, Ohio.