Thin-Walled Structures 119 (2017) 103–113

Contents lists available at ScienceDirect

Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws

Full length article

Compression strength of natural fibre composite plates and sections of flax, MARK
jute and hemp
M.R. Bambach
School of Civil Engineering, The University of Sydney, NSW 2006, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: Natural fibres such as flax, jute and hemp have been utilised for thousands of years, however have only recently
Natural fibre composites been considered for fibre-resin composites. A major motivation for such an application is their superior sus-
Flax tainability attributes compared with traditional building materials. Population rise continues to place increasing
Jute demands for new infrastructure. Meanwhile, public concerns about the environment, climate change, energy
Hemp
consumption and greenhouse gas emissions, place increasing demands for the use of sustainable materials in
Plates
infrastructure.
Channels
Compression While there is a wealth of knowledge in the materials aspects of natural fibre composites, relatively few
Design studies have investigated their potential for structural applications. This paper presents an experimental and
analytical study of natural fibre composite plates and channel sections consisting of flax, jute and hemp fibres
and subjected to pure compression. The intrinsic mechanical properties are shown to be relatively modest.
However, the buckling and post-buckling responses are shown to be stable, the ultimate condition is reached in a
stable and predictable manner, and failure ensues in a gradual and ductile process. These characteristics show
promise for the use of natural fibre composite sections in light structural applications such as in the residential
and light commercial markets. Additionally, the analytical effective width mechanics model shows promise for
use as a design technique for estimating their compression strength.

1. Introduction mechanical properties compared with their synthetic fibre counterparts

Natural fibres such as flax, jute and hemp have been used by hu-
mans for thousands of years, with early records of textile use of flax
dating back at least 7000 years in Egypt [1]. In recent decades, the use
of natural fibres in composite materials has gained substantial interest,
where these fibres may be combined with thermoset or thermoplastic
polymers to create natural fibre composites, which have been particu-
larly identified for their sustainability attributes [2–7]. Favourable
sustainability properties of natural fibres such as flax, jute and hemp
include: renewable resources that absorb CO2 whilst retuning O2 to the
environment; short growth cycle times (eg the sow to harvest cycle for
flax is 100 days, compared with around 15–20 years for radiata pine);
low energy production; recyclable; biodegradable; and low hazard
manufacturing and composite handling and working.
Much of the previous research has focused on the materials aspects,
including fibre processing techniques, composite fabrication meth-
odologies, matrix materials and their effects on the mechanical prop-
erties [8–13]. While applications for natural fibre composites could be
similar to carbon and glass fibre composites, research has indicated that
composites consisting of natural fibres have comparably low intrinsic
[2–13]. As a result, identifying structural applications such as those in
civil infrastructure have thus far been limited [14]. Nonetheless, nat-
ural fibre composites have intrinsic strength which could potentially be
harnessed for structural applications so long as the mechanical beha-
viour is well understood, reliable and predictable.
Some previous studies have investigated flax fibre composites for
use as energy absorbing components in crashworthiness applications
[8,15], and semi-structural applications. The latter includes such ap-
plications as: flax fibre confinement for concrete columns [16]; flax
fibre skin foam core [17], flax fibre skin balsa core [18] and jute fibre
skin foam core [19] sandwich panels; flax fibre wind turbine blades
[20]; bamboo fibre boat hulls [21]; and wood flour sheet piling [22]. To
the authors knowledge no studies have specifically examined flax, jute
and hemp fibre composites for use as primary structural components for
civil infrastructure applications. For such applications, an important
basic load case is compression to resist gravity loads. The aim of this
study was to characterise the intrinsic compression behaviour of natural
fibre composites. To establish these compression characteristics, the
two fundamental components of flat plates and plain channel sections
were considered.

E-mail address: mike.bambach@sydney.edu.au.

http://dx.doi.org/10.1016/j.tws.2017.05.034
Received 23 November 2016; Received in revised form 29 May 2017; Accepted 31 May 2017
0263-8231/ © 2017 Elsevier Ltd. All rights reserved.
M.R. Bambach Thin-Walled Structures 119 (2017) 103–113

Table 1
Composite plate specimens and test results (subscript P refers to predicted values).

Fibre Layers Fibre weight Width Thickness EC fuC Fbuckle Fultimate λ FbuckleP Fbuckle/FbuckleP
(g/m2) (mm) (mm) (MPa) (MPa) (N) (N) (N)

Flax 2 800 152.8 1.6 4199 38.5 306 2294 5.55 435 0.70
Flax 2 800 152.7 1.6 4199 38.5 218 2231 6.57 435 0.50
Flax 3 1200 154.0 2.5 4199 38.5 1300 5275 3.34 1547 0.84
Flax 3 1200 153.2 2.5 4199 38.5 1293 5081 3.34 1556 0.83
Flax 4 1600 153.8 3.4 4199 38.5 2612 8800 2.78 4141 0.63
Flax 4 1600 154.2 3.4 4199 38.5 2906 9181 2.64 4129 0.70
Jute 2 800 153.2 1.6 3523 40.2 175 1825 7.50 364 0.48
Jute 2 800 153.1 1.6 3523 40.2 187 1995 7.26 364 0.51
Jute 3 1200 152.8 2.5 3523 40.2 1081 4238 3.77 1391 0.78
Jute 3 1200 153.0 2.5 3523 40.2 1218 3863 3.55 1389 0.88
Jute 4 1600 152.9 3.5 3523 40.2 3025 9394 2.65 3652 0.83
Jute 4 1600 152.5 3.5 3523 40.2 2975 8656 2.67 3662 0.81
Hemp 3 861 152.4 2.0 2649 33.9 313 3688 5.67 498 0.63
Hemp 3 861 152.7 2.0 2649 33.9 401 3138 5.02 497 0.81
Hemp 4 1148 153.6 2.7 2649 33.9 1031 6319 3.66 1238 0.83
Hemp 4 1148 152.6 2.7 2649 33.9 987 6313 3.73 1247 0.79
Hemp 6 1722 152.7 3.6 2649 33.9 2656 10,400 2.65 3125 0.85
Hemp 6 1722 153.4 3.6 2649 33.9 2962 10,063 2.51 3109 0.95
Mean: 0.74

2. Methods
2.1. General
Plate tests consisted of flat plates simply supported on all four sides,
subjected to uni-directional compression. Section tests consisted of
plain channel sections, being a simple combination of flat stiffened
(web) and unstiffened (flange) plate elements, subjected to pure com-
pression. Thin plate buckling involves the development of membrane
strains in the two principle orthogonal directions, longitudinal and
transverse, thus at a minimum fibres are required in these two principle
directions. In this study the simplest fibre orientation appropriate for
such load consisting of multiple layers of biaxial [0/90] woven fabric
was chosen. A standard bulk commercial laminating epoxy resin with a
room temperature cure was selected, and the hand layup technique
with vacuum bag curing was used for fabrication. This study therefore
attempted to characterise the compression strength of natural fibre
composite structures in their simplest form, and no attempt was made
to optimise these basic characteristics, such as could be achieved with;
complex fibre layouts, different matrix types, automated fibre layups,
autoclave fabrication, complex plate element stiffener arrangements,
etc.
2.2. Materials
Three different natural fibres were investigated in the present study;
flax, jute and hemp. The flax and jute fabrics were commercially pro-
duced for fibre-resin composite fabrications by Composites Evolution;
Biotex Flax 400 g/m2 2 × 2 Twill weave and Biotex Jute 400 g/m2 2 ×
2 Twill weave. Nominal density, tensile strength and modulus values
for the flax were; 1.5 g/cm3, 500 MPa and 50 GPa, and for the jute
were; 1.46 g/cm3, 400 MPa and 40 GPa. The hemp fabric was not
produced specifically for fibre-resin composite applications, however
was recommended as the most appropriate fabric for working with
resins by the manufacturer, and was a plain weave 287 g/m2 pure
unbleached hemp fabric with density 1.48 g/cm3. The commercial bulk
laminating epoxy resin Kinetix R240 with H160 (medium) hardener
was used for all composite fabrications, with density 1.1 g/cm3.
Compression and tension material tests of the neat epoxy resin and
fabricated fibre-resin composites were undertaken in accordance with
ISO 604 [23] and ISO 527 [24], respectively. The fibre-resin composite
material specimens were cut from the fabricated plate specimens. Fibre
volume fractions were calculated using the mass of fabric prior to
fabrication, the mass of composite after fabrication, and the constituent
densities.
2.3. Plate specimens
Flax, jute and hemp fibre-resin composite plates were fabricated
with nominal geometries of 150 mm width and 450 mm length. A
length of three times the width was chosen since stiffened plates are
known to nominally buckle in square half-wavelengths, and having
three half-wavelengths of buckle minimises end effects in plate tests
[25]. Three different plate thicknesses were fabricated for each of the
different fibre types; for the flax and jute composites 2, 3 and 4 fibre
layers were fabricated. Since the hemp fabric had a lower aerial mass,
and in order to create hemp composites with approximately the same
mass of fibre as the flax and jute composites for comparison purposes,
the hemp plates were fabricated with 3, 4 and 6 layers.
The plates were fabricated with a hand layup technique whereby
each layer of fabric was wetted out with resin using a paint brush and
roller. The fabrics were laid flat and a full vacuum (approximately
100kPa) was then set over the plate and held during a cure time of a
minimum of 15 h in a constant temperature room at 23 °C. The plates
were fabricated with approximate dimensions of 250 mm width and
1200 mm length, then following curing were trimmed to size and cut
into two specimens using a guillotine. The measured plate geometries
are summarised in Table 1. Exemplar plate specimens are shown in
Fig. 1.
2.4. Plate test rig
The plates were tested with simple supports on all four sides (stif-
fened plates). The ends of the plates were seated in circular steel rods
with a 6 mm groove to accommodate plates up to 6 mm in thickness.
Shim plates were used to firmly seat the plates with thickness less than
6 mm. The grooved plate-seat rod was housed in a split needle bearing,
allowing the rod to rotate freely in the solid steel bearing block (Fig. 2).
Each bearing block was 60 mm in width, and each grooved plate-seat
rod was 20 mm in width, allowing differential rotation across the plate
width. The upper bearings were fixed to the loading platen of the test
machine, and the lower bearings were fixed to the rigid base of the
machine. The plate end bearings have been used and validated pre-
viously by the author, and further details are provided in [26]. Simple
supports along the plate longitudinal edges were achieved with Teflon
half-rounds that were lightly clamped to each side of the plate edge

104
M.R. Bambach
Fig. 1. Exemplar composite plate specimens, from left; flax, jute and hemp.
Fig. 2. Plate testing rig (side view).
(Fig. 2). The Teflon half-rounds were supported by steel angle sections
containing slotted holes for adjustment, and bolted to a vertical steel
channel section for support. This method has been used and validated
previously by the author in [27]. The split needle bearings undergo a
small amount of bedding-in as load is applied to the plate, therefore the
need for specific plate edge preparations prior to testing was precluded
[26]. Five displacement transducers were located longitudinally at
75 mm intervals along the plate centre-width, to measure the out-of-
plane plate buckling deformations. The plates were loaded in dis-
placement control.
2.5. Channel specimens
Flax, jute and hemp fibre-resin composite channels were fabricated
with nominal geometries of web depth 100 mm, flange width 40 mm
and 300 mm length. A length of three times the web depth was chosen
since webs (stiffened plates) are known to nominally buckle in square
half-wavelengths, and having three half-wavelengths of buckle mini-
mises end effects. Three different channel thicknesses were fabricated
for each of the different fibre types, the same thicknesses as those for
105
Thin-Walled Structures 119 (2017) 103–113
the plate specimens; 2, 3 and 4 fibre layers for the flax and jute chan-
nels, and 3, 4 and 6 fibre layers for the hemp channels.
The channels were fabricated with a hand layup technique whereby
each layer of fabric was wetted out with resin using a paint brush and
roller. The fabrics were laid over a mandrel, being a steel square hollow
section of outside dimension 100 mm and corner radius of 8 mm,
thereby all channels had an internal web depth of 100 mm. A full va-
cuum was then set over the channel and held during a cure time of a
minimum of 15 h in a constant temperature room at 23 °C. The chan-
nels were fabricated with approximate dimensions of flange width
70 mm and length 850 mm, then following curing were trimmed to size
and cut into two specimens using a water jet cutter. The measured
channel geometries are summarised in Table 2. Exemplar channel
specimens are shown in Fig. 3.
2.6. Channel tests
The thin channels were tested in pure compression between fixed
load platens. In order to reduce possible end effects such as eccentric
loading or stress concentrations resulting from the channel ends not
being completely flat and/or parallel, prior to testing 2 mm thick steel
end plates were bonded to both ends of the channels using fast cure
Araldite K219 epoxy. This resulted in the channel ends being essentially
fixed ended, since end rotations were precluded. This method has been
used and validated previously by the author in [28]. Additionally, the
channels were seated on a spherical bearing which when unlocked al-
lowed free rotations in all directions. The test procedure involved pla-
cing the channel on the unlocked spherical bearing and applying a
small load, then locking the spherical bearing such that rotations were
then restrained. The restraint condition was thereby fixed end re-
straints. Five displacement transducers were located around the chan-
nels to measure the out-of-plane buckling deformations, including three
transducers located longitudinally at 75 mm intervals along the web
centre-width and one transducer at the edge of each flange at the mid-
length. The channels were loaded in displacement control.
3. Results
3.1. Materials
The fabrication technique generated consistent fibre volume frac-
tions for both the plate and channel specimens. For all specimens the
mean and standard deviations of the fibre volume fractions were;
42.5% and 2.1% for flax, 44.3% and 2.2% for jute and 40.2% and 2.5%
for hemp, respectively. The slightly lower fibre volume fraction for
hemp resulted from the more open weave in the hemp fabric. The
measured material properties are summarised in Table 3 and exemplar
tension stress-strain curves are shown in Fig. 4. It is evident from these
data that the flax and jute composites had similar tensile elastic moduli
(ET) with the flax showing a slightly higher value (6.0 GPa and 5.2 GPa,
respectively). While the flax composite was slightly stiffer, it was
slightly less strong than the jute composite, with ultimate tensile
stresses (fuT) of 45.4 MPa and 52.1 MPa, respectively. The hemp com-
posite had lower values of both tensile elastic modulus and strength of
3.8 GPa and 35.3 MPa, respectively. The compression material prop-
erties showed similar characteristics, where the flax and jute compo-
sites had quite similar properties and the hemp composite slightly lower
values. Compared to the tension properties, the elastic moduli and
strength in compression (EC, fuC) were on average 22% lower for all
composites (Table 3).
3.2. Plate tests
The plate test results are summarised in Table 1. The plates typically
demonstrated a linear elastic response until buckling, a post-buckling
response characterised by a reduced stiffness and increasing out-of-
M.R. Bambach Thin-Walled Structures 119 (2017) 103–113

Table 2
Composite channel specimens and test results (dimensions refer to outside dimensions, subscript P refers to predicted values).

Fibre Layers Fibre weight Web Flanges Thick-ness EC fuC Fbuckle Fultimate λ FbuckleP Fbuckle/FbuckleP FultimateP Fultimate/FultimateP
(g/m2) (mm) (mm) (mm) (MPa) (MPa) (N) (N) (N) (N)

Flax 2 800 104.1 39.8 1.7 4199 38.5 949 2683 3.45 1527 0.62 3807 0.70
Flax 2 800 104.0 39.8 1.7 4199 38.5 982 2922 3.39 1526 0.64 3808 0.77
Flax 3 1200 105.5 41.2 2.5 4199 38.5 3898 6898 2.06 4439 0.88 7377 0.94
Flax 3 1200 105.9 41.4 2.5 4199 38.5 3667 6381 2.13 4402 0.83 7370 0.87
Flax 4 1600 107.5 41.3 3.3 4199 38.5 8933 11,023 1.58 11,406 0.78 12,950 0.85
Flax 4 1600 107.4 41.2 3.3 4199 38.5 9287 11,655 1.55 11,463 0.81 12,954 0.90
Jute 2 800 104.0 41.2 1.8 3523 40.2 825 2996 3.86 1299 0.64 3864 0.78
Jute 2 800 104.1 41.1 1.8 3523 40.2 1027 2967 3.46 1306 0.79 3864 0.77
Jute 3 1200 106.1 42.0 2.8 3523 40.2 4414 7678 2.10 5142 0.86 8731 0.88
Jute 3 1200 105.9 42.0 2.8 3523 40.2 4253 7611 2.14 5136 0.83 8733 0.87
Jute 4 1600 107.1 41.7 3.5 3523 40.2 7655 9555 1.79 11,259 0.68 13,791 0.69
Jute 4 1600 107.3 41.8 3.5 3523 40.2 7514 10,078 1.81 11,211 0.67 13,785 0.73
Hemp 3 861 104.8 41.8 2.2 2649 33.9 1031 1985 3.53 1799 0.57 4489 0.44
Hemp 3 861 104.6 42.1 2.2 2649 33.9 985 2207 3.62 1769 0.56 4487 0.49
Hemp 4 1148 106.2 41.2 2.7 2649 33.9 3205 4663 2.24 3812 0.84 6807 0.69
Hemp 4 1148 106.0 41.0 2.7 2649 33.9 2748 4903 2.41 3849 0.71 6812 0.72
Hemp 6 1722 109.0 41.5 3.9 2649 33.9 10,780 11,879 1.47 12,231 0.88 13,611 0.87
Hemp 6 1722 108.9 41.7 3.9 2649 33.9 11,687 13,262 1.41 12,088 0.97 13,607 0.97
Mean 0.75 Mean: 0.77

Fig. 3. Exemplar composite channel specimens, from

left; flax, jute and hemp (for each fibre type one of
each thickness is shown, thinnest to thickest from
left).

Table 3
Material properties of the composites and the epoxy resin (neat).

ET fuT εuT EC fuC

(MPa) (MPa) (%) (MPa) (MPa)

Kinetix R240 4362 33.1 0.8 3588 105.3

Flax 6030 45.4 3.1 4199 38.5
Jute 5184 52.1 1.6 3523 40.2
Hemp 3826 35.3 1.8 2649 33.9

plane buckling deflections in multiple half-waves, and attainment of the

ultimate load characterised by the localisation of the buckling into one
of the half-waves. The post-buckling and ultimate load responses were
both very stable, and in no cases was an unstable brittle fracture mode
characterised by sudden failure and loss of load carrying capacity evi-
dent. Exemplar axial force – axial displacement plots are provided in
Fig. 5, while a plot with photos of the deformed shape at different load
levels is shown in Fig. 6. Due to the tight tolerances in the plate test rig Fig. 4. Exemplar measured tension stress-strain curves of the neat epoxy and the fibre-
only a small portion of the unloading curve could be captured. resin composites.
The buckling modes consisted of between two and four half-waves
of buckling, with no discernible relationship between material or di- specimens are shown in Fig. 7. The buckling loads (Table 1) were
mensions and the number of half-waves. The particular buckled shape quantified using plots of the load versus out-of-plane buckling dis-
was likely a response to the geometric and/or material imperfections of placement squared methodology [29], as used previously and validated
the individual specimen. Exemplar deformed shapes at the ultimate by the author [26]. The maximum values from the five transducers was
condition for the three different fibres and their thinnest and thickest used for this analysis. Due to the different number of half-waves that

106
M.R. Bambach
Fig. 5. Exemplar composite plate test results; (a) flax, (b) jute, (c) hemp.
developed in the different plate specimens, the five transducers were
not necessarily located in the position of maximum out-of-plane dis-
placement in the plate, therefore these values provide upper bounds to
the buckling loads.
107
Thin-Walled Structures 119 (2017) 103–113
Fig. 6. Test result for composite plate consisting of 4 layers of flax
fibre.
3.3. Channel tests
The channel test results are summarised in Table 2. The channels
typically demonstrated a similar response to the plates, including a
linear-elastic response followed by buckling, stable post-buckling, at-
tainment of the ultimate load and then stable unloading. The post-
M.R. Bambach
Fig. 8. Exemplar composite channel test results; (a) flax, (b) jute, (c) hemp.
buckling and ultimate load responses were both very stable, and in no
cases was an unstable brittle fracture mode characterised by sudden
failure and loss of load carrying capacity evident. Exemplar axial force –
axial displacement plots are provided in Fig. 8, while a plot with photos
of the deformed shape at different load levels is shown in Fig. 9. The
channels tests were continued until the load reduced to at least 80% of
the ultimate value, and the response demonstrated a very stable, gra-
dual unloading mechanism.
The buckling modes consisted of two or three half-waves of buck-
ling, with no discernible relationship between material or dimensions
and the number of half-waves. The particular buckled shape was likely
a response to the geometric and/or material imperfections of the in-
dividual specimen. Buckling occurred with approximately the same
half-wavelength in both the web and the flanges. Exemplar deformed
shapes at the ultimate condition for the three different fibres and their
thinnest and thickest specimens are shown in Fig. 10. Since fewer
108
Thin-Walled Structures 119 (2017) 103–113
Fig. 7. Exemplar composite plate specimens
at the ultimate condition, from left; flax, jute
and hemp (for each fibre type one of each
thickness is shown, thinnest to thickest from
left).
transducers were used in the channel tests (one per flange and three in
the web), and since the flange transducer was frequently not in the
position of maximum out-of-plane displacement (since many channels
buckled in two half-waves rather than three), the buckling behaviour
was less well characterised by the transducers than with the plate
specimens. The channel buckling loads were therefore quantified using
the change in section stiffness methodology, whereby the point of linear
elastic axial stiffness bifurcation was identified and the corresponding
load taken as the section buckling load. Previous work on steel channel
sections by the author has indicated the validity of this methodology in
comparison with using the buckling deformation values [30]. Since the
stiffness reduction occurs following buckling, these values provide
upper bounds to the section buckling loads.
M.R. Bambach
3.4. Failure mechanisms
In both the composite plates and channels failure was precipitated
by the localisation of the buckling into one of the half-waves, thereby
reaching large out-of-plane displacements in that half-wave, resulting
in matrix failures along the critical fold lines. Exemplar failure me-
chanisms are shown in Fig. 11, where the mechanisms in the plate and
the channel web resemble the classic flipped disk mechanism, while
that in the channel flange resembles the inclined yield line mechanism.
It is interesting to note the similarities between these mechanisms and
yield lines that form in the plastic collapse mechanisms commonly
observed in thin steel plates and channels [31].
4. Design
4.1. General
The plates and channels were fabricated with a biaxial [0,90] woven
fabric and were therefore not strictly quasi-isotropic (given that
+/−45 layers were not included). Nonetheless, for the purposes of the
present loading case of pure compression resulting in buckling induced
109
Thin-Walled Structures 119 (2017) 103–113
Fig. 9. Test result for composite channel consisting of 4 layers of
jute fibre.
membrane strains in the two principle orthogonal directions, and
without significant shear stresses nor shear buckling, an isotropic
theory was considered potentially appropriate [32]. Additionally, in-
dividual specimens had the same amount of fibre in the 0 and 90 di-
rections, and since the elements were thin the through-thickness
properties were not especially important. The in-plane compression
buckling and strength theory used for thin steel structures prone to out-
of-plane buckling was assessed for the potential to be used for the de-
sign of natural fibre composite plates and channel sections.
4.2. Buckling
The theoretical buckling stress (fcr) of flat thin isotropic plates is
given by Eq. (1), where E is elastic modulus, t is the plate thickness, b is
the plate width, υ is Poisson's ratio and k is the elastic buckling coef-
ficient (4 for stiffened plates and 0.43 for unstiffened plates) [33]. The
buckling stresses of the plates were calculated using the compression
material properties in Table 3, and the results are tabulated in Table 1.
Fig. 12a compares the experimental and theoretical plate buckling
stresses. From these data it is evident that the theory over-estimates all
the experimental buckling values (by 26%), which is expected since the
Fig. 10. Exemplar composite channel spe-
cimens at the ultimate condition, from left;
flax, jute and hemp (for each fibre type one
of each thickness is shown, thinnest to
thickest from left).
M.R. Bambach
plates contain geometric and material imperfections while the theory
assumes perfect plate properties. Similar results were found for the
channels (25%), as shown in Table 2 and Fig. 12b, where the lower of
the theoretical buckling stresses calculated for the web and flanges was
taken as the buckling stress for the section.
kπ 2E t 2
fcr = ⎛ ⎞
12(1 − υ2) ⎝ b ⎠
4.3. Strength
The Winter strength formulae [34] used for the design of thin steel
sections in several international cold-formed steel design codes [35,36]
are given by Eqs. (2) and (3), where λ is the plate or section slenderness,
fuC is the material ultimate stress in compression, ρ is the plate or sec-
tion effectiveness ratio (the ratio of the effective area to the gross area
of the plate or section) and A is the plate or section area. For steel the
ultimate stress is taken as the tensile yield stress, however for the
purposes of the present composite materials the ultimate compression
stress was used (Table 3). The plate or section strength is then calcu-
lated as the effective area multiplied by the ultimate compression stress
Fig. 12. Comparisons of experimental and theoretical buckling stresses; (a) composite plates, (b) composite channels.
Thin-Walled Structures 119 (2017) 103–113
Fig. 11. Exemplar failure mechanisms in
flax composite plates (left) and channels
(right).
(fuC).
fuC
λ=
fcr (2)
0 . 22
Aeffective 1−
ρ= = λ
A λ (3)
(1)
These formulae are based on the effective width principle, which
states that the post-buckling strength is obtained when the unbuckled
portions of the plate elements reach the ultimate stress of the material,
while the buckled portions are assumed ineffective in carrying load
[34]. This mechanics model is a reasonable assumption for the com-
posite plates in the present study. The theory is based on two calcula-
tions: first, the buckling stress must be established, and from this value
the slenderness is calculated; second, the post-buckling strength is es-
tablished from the known value of slenderness using the Winter equa-
tion. While the basic form of the Winter equation is given by Eq. (3), it
is well known that this strength curve typically needs to be empirically
adjusted according to the specific material and/or geometric and/or
loading conditions at hand. For this purpose plate experiments are ty-
pically employed [37]. In the present study, the experimental buckling
110
M.R. Bambach
Fig. 13. Comparisons of experimental and theoretical strengths; (a) composite plates, (b) composite channels.
stresses in the plate tests were used to calculate the experimental plate
slenderness values, and these values were plotted against the effec-
tiveness ratio (Aeffective/A) established from the experimental plate
strength. In this calculation it was assumed that 5 mm of plate width on
either side of the Teflon half-round longitudinal supports did not
buckle. The results are shown in Fig. 13a, where it is evident that the
traditional Winter Eq. (3) over-estimates many plate strengths. A re-
duced Winter effective width Eq. (4) was established simply by making
empirical adjustments to the Winter equation until an approximate
lower bound to the test results was achieved (Fig. 13a) [37].
1.5
0 . 22
⎛1 − λ ⎞
ρ=⎜ ⎟
λ (50 MPa, 8 GPa). Costa and D’Almeida [9] tested unidirectional jute
⎝ ⎠
The empirical strength Eq. (4) derived from the plate tests was then
compared with the channel sections for validation. As with the plates,
the experimental channel buckling stresses were used to calculate the
experimental channel slenderness values, and these values were plotted
against the effectiveness ratio established from the experimental
channel strengths. The results are shown in Fig. 13b, where it is evident
that in accordance with the plate tests, the traditional Winter equation
over-predicted the strengths of the channels, while the empirical Eq. (4)
provides a more realistic estimate.
Having established an appropriate strength equation for the natural
fibre composite channel sections, this equation was used to calculate
design values for the channel compression strengths. In the design
procedure the buckling stresses of the web and flanges are those the-
oretical values calculated from Eq. (1), from which the slenderness
values of the web and flanges are derived from Eq. (2), and following
which the strength of the channel section is established from Eq. (4).
These values are compared with the test strengths in Table 2, where
they over-predict the test strengths on average by 23%. This stems from
the fact that the theoretical buckling stress values over-predict the ex-
perimental values on average by 25% (Table 2). As demonstrated in
Fig. 13b, when the experimental buckling stress values are used the
modified strength Eq. (4) predicts the strength reasonably well, and
when these values are employed in place of the theoretical values from
Eq. (1) the over-prediction of the section strength reduces to 7%.
However, in a design situation these values would not be known a-
priori, therefore improvements to the prediction of the buckling stresses
are required in order to improve the theoretical strength predictions.
5. Discussion
In the present study the intrinsic mechanical properties of the nat-
ural fibre composites were shown to be relatively low, with tensile
Thin-Walled Structures 119 (2017) 103–113
strengths between 35.3 MPa and 52.1 MPa and tensile moduli between
3.8 GPa and 6.0 GPa. In the literature a wide variety of mechanical
properties for natural fibre composites have been documented, the
values of which may vary due to the broad range of possible fibre
treatments, fabric types, resin types, fibre laying processes, fabrication
and curing techniques, etc. For example Pickering et al. [5] reviewed
many published values of various fibres, resins and fibre volume frac-
tions, reporting that unidirectional hemp composites had (tensile
strength, tensile moduli) values between (77 MPa, 10 GPa) and
(165 MPa, 17 GPa), while unidirectional flax composites had values
between (40 MPa, 7 GPa) and (321 MPa, 29 GPa). They also noted
biaxial hemp values of (62 MPa, 7 GPa) and woven jute values of
(4)
epoxy resin composites with fibre volume fractions of 10%, noting
values of (50.3 MPa, 4.02 GPa). Hargitai et al. [10] tested unidirec-
tional hemp polypropylene composites with fibre mass fractions of
40%, finding values of (56 MPa, 5.9 GPa). Oksman [11] tested uni-
directional flax epoxy resin composites with fibre volume fractions of
32%, noting values of (132 MPa, 15 GPa), and high performance Artic
flax with fibre volume fractions of 42% with values of (280 MPa,
35 GPa). Van de Weyenberg et al. [12] tested unidirectional flax epoxy
resin composites with fibre volume fractions of 40%, finding values of
(133 MPa, 28 GPa). Weclawski et al. [8] tested unidirectional hemp
polyester resin composites with fibre volume fractions of 30%, finding
tension values of (128.9 MPa, 8.9 GPa) and compression values of
(53 MPa, 3.83 GPa). In comparison with unidirectional composites,
woven biaxial composites will have material properties much less than
the unidirectional values (for the same volume of fibres), since half of
the fibres contribute little to the unidirectional strength, and the weave
further reduces the fibre effectiveness compared with unidirectional
layers. Notwithstanding this, the mechanical properties of the compo-
sites fabricated in the present study are towards the lower end of pre-
viously published values.
Importantly, while the mechanical properties of natural fibre com-
posites are relatively low, the buckling and post-buckling responses
were shown to be stable, the ultimate condition was reached in a stable
and predictable manner, and the failure ensued in a gradual and ductile
process. These stable compression characteristics imply an inherent
suitability for structural applications. The relatively modest compres-
sion strengths lend themselves to light structural applications, poten-
tially in the residential and light commercial markets. For comparison
purposes, design compression loads for Australian domestic stud wall
columns were extracted from AS1720 [38], using standard conditions
(moderate wind loads, factored dead and live loads, 2.7 m wall height,
sheet roof, plasterboard claddings, etc). For 0.6 m stud spacings, one
and two storey house loads are 4.5 kN and 9 kN respectively, while for
111
M.R. Bambach
Fig. 14. Comparisons of experimental and theoretical plate strengths disaggregated by
fibre type.
1.2 m stud spacings 9 kN and 15 kN respectively, per stud column.
These results indicate that the natural fibre composite channels devel-
oped in the present study, with maximum compression strengths of
between 10.1 kN and 13.3 kN (for the thickest specimens of each fibre
type), could potentially be structurally appropriate for such applica-
tions. This is especially true in light of the fact that the channel
strengths could potentially be substantially enhanced following opti-
misation procedures.
For comparison purposes with traditional building materials in light
structural applications such as residential buildings, the four layer
composite jute channels (107 × 42 × 3.5 mm) are compared with
exemplar thin steel channel studs and solid rectangular timber studs
typical to this application. A typical light steel stud in structural wall
framing in Australia is a 90 × 37 mm channel with flange and web
stiffeners, fabricated from G450 steel of thickness 0.55 mm. A typical
light timber stud is a 70 × 35 mm solid rectangular member from
seasoned radiata pine softwood of grade F7. For the composite jute
channel the compression stiffness and ultimate stress are 3.5 GPa and
40 MPa, while the nominal values for the steel channel are 200 GPa and
450 MPa, and the nominal values for the timber stud are 7.9 GPa and
15 MPa. The cross-section area and un-factored short column maximum
compression load values for the composite jute channel are 614 mm2
and 10 kN, while the values for the steel channel are 97 mm2 and 14 kN
(incorporating the effects of local buckling), and the values for the
timber stud are 2450 mm2 and 37 kN.
Another attribute that potentially indicates the suitability of natural
fibre channels for light structural applications is the predictable nature
of their compression strength. Within the limitations of the small
sample in the present study, the effective width mechanics model
showed promise for estimating the section strength. For new types of
fibre-resin composites the strength equations may need to be empiri-
cally modified, however this can be easily achieved with plate tests
such as those undertaken in the present study. Indeed, this metho-
dology was used to develop the strength equations in current interna-
tional specifications for the design of thin steel sections [35,36] such as
those undertaken by the author [37,39], which are routinely used for
designing thin steel stud wall columns.
It should be noted that while the effective width method shows
promise for natural fibre composites, the effects of geometric and ma-
terial imperfections on the buckling stress requires further research.
The theoretical buckling stress values substantially exceeded the ex-
perimental values, which led to a substantial over-prediction of the
theoretical section strength. Additionally, the strength equations
themselves could certainly be improved once more data becomes
112
Thin-Walled Structures 119 (2017) 103–113
available; in particular it may be beneficial to develop strength equa-
tions for different types of fibres. Indeed, disaggregating the plate data
by fibre type in Fig. 14 indicates that while the flax and jute fibre
composites follow a similar trend, the hemp fibre composites demon-
strate a higher strength curve. Further plate data is required to accu-
rately discern these trends and improve theoretical predictions.
As noted previously, optimisation of natural fibre composite
channel sections would potentially result in substantially larger com-
pression strengths that would increase their suitability for light struc-
tural applications. Material and/or geometric optimisation was not
considered in the present study, and accordingly the present results
might be considered a lower bound to what could be achieved with
optimisations in the areas of, for example; advanced fibre layouts, fibre
treatments, different resins, machine layups, autoclave curing regimes
and geometric considerations. Indeed simple geometric optimisation
with the present composite materials such as the inclusion of flange
edge, flange interior and/or web longitudinal stiffeners could see sig-
nificant strength enhancements, in the same manner as they are em-
ployed in thin steel sections to delay the onset of buckling. One could
also add more fibre layers. Clearly, there are many other factors that
would also need to be addressed before natural fibre composites could
be introduced into buildings, for example; column buckling, other
loading types, connections, fire performance, environmental exposure
and durability, etc., which were outside the scope of the present study.
6. Conclusions
Increasing populations and increasing awareness of environmental
concerns are leading a drive towards more sustainable structural ma-
terials for infrastructure applications. The present study has demon-
strated the reliable and predictable, albeit modest, mechanical beha-
viour of natural fibre composite structures of flax, jute and hemp in
pure compression. With appropriate attention to fibre and resin design
and structural geometry, natural fibre composites may prove a viable
alternative to traditional building materials in the future. In conclusion,
these initial findings of structural properties and design methods show
promise for residential and light commercial building applications,
however further research is necessary.
References
[1] J. Summerscales, N.P.J. Dissanayake, A.S. Virk, W. Hall, A review of bast fibres and
their composites. Part 1 – fibres as reinforcements, Composites A 41 (2010)
1329–1335.
[2] J. Summerscales, N.P.J. Dissanayake, A.S. Virk, W. Hall, A review of bast fibres and
their composites. Part 2 – composites, Composites A 41 (2010) 1336–1344.
[3] H. Ku, H. Wang, N. Pattarachaiyakoop, M. Trada, A review on the tensile properties
of natural fibre reinforced polymer composites, Composites B 42 (2011) 856–873.
[4] L.B. Yan, N. Chouw, K. Jayaraman, Flax fibe and its composites – a review,
Composites B 56 (2014) 296–317.
[5] K.L. Pickering, M.G.A. Efendy, T.M. Le, A review of recent developments in natural
fibre composites and their mechanical performance, Compos.: Part A 83 (2016)
98–112.
[6] O. Faruk, A. Bledzki, H. Fink, M. Sain, Biocomposites reinforced with natural fibres:
2000 – 2010, Progress. Polym. Sci. 37 (2012) 1552–1596.
[7] M. Dicker, P. Duckworth, A. Baker, G. Francois, M. Hazzard, Green composites: a
review of material attributes and complementary applications, Compos. Part A 56
(2014) 280–289.
[8] B.T. Weclawski, M. Fan, D. Hui, Compressive behaviour of natural fibre composite,
Composites B 67 (2014) 183–191.
[9] Costa FHMM, J.R.M. D’Almeida, Effect of water absorption on the mechanical
properties of sisal and jute fibre composites, Polym. Plast. Technol. Eng. 38 (5)
(1999) 1081–1094.
[10] H. Hargitai, I. Racz, R.D. Anandjiwala, Development of hemp fiber reinforced
polypropylene composites, J. Therm. Comp. Mat. 21 (2008) 165–174.
[11] K. Oksman, High quality flax fibre composites manufactured by the resin transfer
moulding process, J. Reinf. Plast. Comp. 20 (7) (2001) 621–627.
[12] I. Van de Weyenberg, J. Ivens, A. De Coster, B. Kino, E. Baetens, Verpoest. Influence
of processing and chemical treatment of flax fibres on their composites, Comp. Sci.
Tech. 63 (2003) 1241–1246.
[13] A. Lefeuvre, A. Bourmaud, C. Morvan, C. Baley, Elementary flax fibre tensile
properties: correlation between stress- strain behaviour and fibre composition, Ind.
Crops Prod. 52 (2014) 762–769.
M.R. Bambach
[14] D. Dittenber, H. GangRao, Critical review of recent publications on use of natural
composites in infrastructure, Composites A 43 (2012) 1419–1429.
[15] L.B. Yan, N. Chouw, Crashwortiness characteristics of flax fibre reinforced epoxy
tubes for energy absorption application, Mat. Des. 51 (2013) 629–640.
[16] L.B. Yan, N. Chouw, Behaviour and analytical modelling of natural flax fibre re-
inforced polymer tube encased coir fibre reinforced concrete composite column, J.
Comp. Mater. 47 (17) (2013) 2133–2148.
[17] K. Mak, A. Fam, C. McDougall, Flexural behavior of sandwich panels with bio-FRP
skins made of flax fibers and epoxidized pine-oil resin, J. Compos. Constr. 19 (6)
(2015) 04015005.
[18] A.L. Duigou, J.M. Deux, P. Davies, C. Baley, PLLA/flax mat/balsa bio-sandwich
manufacture and mechanical properties, Appl. Compos Mater. 18 (2011) 421–438.
[19] N. Uddin, R.R. Kalyankar, Manufacturing and structural feasibility of natural fiber
reinforced polymeric structural insulated panels for panelized construction, Int J.
Polym. Sci. 96 (2011) 35–49.
[20] D.U. Shah, P.J. Schubel, M.J. Clifford, Can flax replace E-glass in structural com-
posites? A small wind turbine blade case study, Compos. Part B 52 (2013) 172–181.
[21] S. Corradi, T. Isidori, M. Corradi, F. Soleri, Composite boat hulls with bamboo
natural fibres, Int J. Mater. Prod. Technol. 36 (1) (2009) 73–89.
[22] D. Alvarez-Valencia, H.J. Dagher, R.A. Lopez-Anido, W.G. Davids, D.J. Gardner, A.
C. Center, Behavior of natural-fiber/thermoplastic sheet piling. ACMA-composites
and polycon conference, January 2009, Tampa, Florida, USA, 2009.
[23] ISO 604:2002. Plastics – Determination of compressive properties.
[24] ISO 527-4:1997. Plastics – Determination of tensile properties. Part 4: Test condi-
tions for isotropic and orthotropic fibre-reinforced plastic composites.
[25] T. Burns, P. Bezkorovainy, M.R. Bambach, K.J.R. Rasmussen, Numerical modelling
of anisotropic stainless steel plates, J. Constr. Steel Res. 59 (11) (2003) 1345–1362.
[26] M.R. Bambach, K.J.R. Rasmussen, Tests of unstiffened plate elements under com-
bined bending and compression, J. Struct. Eng., ASCE 130 (10) (2004) 1602–1610.
113
Thin-Walled Structures 119 (2017) 103–113
[27] M.R. Bambach, Experiments of edge-stiffened plates in uniform compression, Thin-
Walled Struct. 49 (2) (2011) 343–350.
[28] M.R. Bambach, Axial capacity and crushing of thin-walled metal, fibre-epoxy and
composite metal-fibre tubes, Thin-Walled Struct. 48 (6) (2010) 440–452.
[29] K.R. Venkataramaiah, J. Roorda, Analysis of local plate buckling data. in:
Proceedings of the Sixth international specialty conference on cold-formed steel
structures, 1982.
[30] M.R. Bambach, Photogrammetry measurements of buckling modes and interactions
in channels with edge stiffened flanges, Thin-Walled Struct. 47 (5) (2009) 485–504.
[31] N.W. Murray, P.S. Khoo, Some basic plastic mechanisms in the local buckling of
thin-walled steel structures, Int. J. Mech. Sci. 23 (12) (1981) 703–713.
[32] M.R. Bambach, H.H. Jama, M. Elchalakani, Axial capacity and design of thin-walled
steel SHS strengthened with CFRP, Thin-Walled Struct. 47 (10) (2009) 1112–1121.
[33] P.S. Bulson, The Stability of Flat Plates, Chatto and Windus, London, 1970.
[34] G. Winter, Strength of Thin Steel Compression Flanges, Transactions 112 ASCE,
1947, pp. 527–576.
[35] AS/NZS4600. Australian/New Zealand Standard. Cold-Formed Steel Structures,
Standards Australia, Sydney, 2005.
[36] AISI S100−12. North American Specification for the Design of Cold-Formed Steel
Structural Members. American Iron and Steel Institute, 2012.
[37] M.R. Bambach, K.J.R. Rasmussen, Effective widths of unstiffened elements with
stress gradients, J. Struct. Eng. ASCE 130 (10) (2004) 1611–1619.
[38] AS 1720.3:2016. Australian Standard for Timber structures, Part 3: Design criteria
for timber-framed residential buildings. Standards Australia, 2016.
[39] M.R. Bambach, K.J.R. Rasmussen, Design provisions for sections containing un-
stiffened elements with stress gradient, J. Struct. Eng. ASCE 130 (10) (2004)
1620–1628.