Coupled Influence of Strain Rate and Heterogeneous Fibre Orientation On The Mechanical Behaviour of Short-Glass-Fibre Reinforced Polypropylene

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Mechanics of Materials 100 (2016) 186–197

Contents lists available at ScienceDirect

Mechanics of Materials
journal homepage: www.elsevier.com/locate/mechmat

Coupled influence of strain rate and heterogeneous fibre orientation


on the mechanical behaviour of short-glass-fibre reinforced
polypropylene
D. Notta-Cuvier a,∗, M. Nciri a,b, F. Lauro a, R. Delille a, F. Chaari a, F. Robache a, G. Haugou a,
Y. Maalej c
a
LAMIH UMR CNRS/UVHC 8201, UVHC Le Mont Houy, 59313 Valenciennes Cedex 9, France
b
National Engineering School of Sfax (ENIS), L2MP, B.P. W3038, Sfax, Tunisia
c
University of Tunis El Manar, ENIT, MAI (LR11ES19), 1002 Tunis, Tunisia

a r t i c l e i n f o a b s t r a c t

Article history: Short-fibre-reinforced thermoplastic polymers (SFRT) are appealing materials for use in technical applica-
Received 24 March 2016 tions, due to high rigidity/density ratio, in particular. However, SFRT have complex anisotropic mechan-
Revised 7 June 2016
ical behaviour, because of specificities of thermoplastic matrix behaviour and heterogeneous reinforce-
Available online 1 July 2016
ment characteristics, in particular in terms of fibre orientation. For instance, matrix behaviour can be
Keywords: viscoelastic and/or viscoplastic (i.e. sensitive to strain rate) pressure sensitive, damageable, etc. Moreover,
Short-fibre-reinforced thermoplastic mechanisms of load transmission from matrix to fibres, at fibre/matrix interface, are still not very well
Strain-rate sensitivity known, although they play a crucial role in SFRT mechanical behaviour. For instance, there is no infor-
Anisotropy mation about their eventual strain-rate sensitivity. In this study, the 3D microstructure of an injection-
Micro-computed tomography moulded short-fibre-reinforced polypropylene is reconstructed using micro-computed tomography. The
tensile macroscopic behaviour of the composite is then studied under different angles of loading with
respect to injection flow direction and different displacement rates, in quasi-static and dynamic range
(from 1 mm/min to 1 m/s). The anisotropy of the composite behaviour can therefore be analysed not only
in terms of dependence on macroscopic loading direction, but also on actual fibre distribution of orien-
tation relatively to tensile direction, at the microscopic scale. In an original way, coupled influence of
fibre orientation and strain rate is then studied. One of the most significant results is that fibre/matrix
interfacial behaviour, which governs load transmission from matrix to fibres, does not show strain rate
sensitivity for the considered composite and strain-rate range (5.10−4 to 50 s−1 ).
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction impact, it is primordial to characterise and model the strain-rate


sensitivity of SFRT behaviour for a wide range of strain rate, from
The use of short-fibre-reinforced thermoplastics (SFRT) is be- quasi-static to dynamic loading conditions.
coming more and more appealing, especially for technical appli- As stated before, thermoplastic polymers, such as polypropy-
cations, because of their interesting mechanical properties, such lene (PP), may have complex mechanical behaviour, possibly vis-
as high stiffness/density ratio. However, the mechanical behaviour coelastic and/or viscoplastic, i.e. sensitive to strain rate, with
of SFRT is complex, in particular because of the inherent com- non-isochoric and pressure dependent plastic flow (Balieu et al.,
plexity of thermoplastic matrix behaviour and heterogeneous re- 2013), etc… Obviously, the mechanical behaviour of composite
inforcement characteristics, for instance in terms of fibre distribu- material is likely to be directly impacted by matrix behaviour
tions of length and orientation. To allow an increasing use of SFRT specificities. In particular, SFRT may present strain-rate sensi-
in technical applications, it is therefore necessary to better under- tivity, as proven for instance by Mouhmid et al. (2006) for a
stand and model SFRT mechanical behaviour under a wide range of short-glass-fibre reinforced PA6,6, Reis et al. (2013) for a short-
loading types. For instance, when aiming at manufacturing techni- glass-fibre reinforced polyurethane, Schoßig et al. (2008) for
cal parts subjected to severe loading conditions, such as crash or glass-fibre-reinforced polypropylene and polybutene-1 or again
Fitoussi et al. (2013) for a glass-fibre-reinforced ethylene-propylene

copolymer. However, studies of Mouhmid et al. (2006) and
Corresponding author.
E-mail address: delphine.notta@univ-valenciennes.fr (D. Notta-Cuvier).
Reis et al. (2013) are restricted to the low strain rates (up

http://dx.doi.org/10.1016/j.mechmat.2016.06.013
0167-6636/© 2016 Elsevier Ltd. All rights reserved.
D. Notta-Cuvier et al. / Mechanics of Materials 100 (2016) 186–197 187

to 5 × 10−2 s−1 and 2 × 10−3 s−1 , respectively). On the contrary, (Notta-Cuvier et al., 2015a; Sato et al., 1991) and fibre breakage.
Schoßig et al. (2008) and Fitoussi et al. (2013) have studied the Lastly, some other phenomena must be taken into account if con-
tensile behaviour of short-fibre-reinforced thermoplastics under sidering vegetal fibres, in particular the debonding of fibrils in fibre
dynamic loading conditions (strain rate up to 174 s−1 and 200 s−1 , bundles (Charlet and Béakou, 2011; Notta-Cuvier et al., 2015b).
respectively). Both studies have highlighted the increase of com- The present study is part of work initiated a few years ago, hav-
posite stiffness and strength with strain rate. However, strain-rate ing as global objective the proposal of an original model of SFRT
dependency is not analysed together with the influence of the mechanical behaviour that takes all specificities of these materi-
orientation of short fibres. Indeed, Schoßig et al. (2008) do not als into account (Notta-Cuvier et al., 2013, 2014, 2015a, 2015b).
give any information concerning fibre orientation, whereas Fitoussi In particular, current aim is to extend the range of application of
et al. present results of tensile tests performed only in injection the existing model to the prediction of the mechanical response of
flow direction (the material is considered as orthotropic). SFRT technical parts under severe loading conditions, such as crash
In addition to specificities of matrix behaviour, SFRT are also and impact. The model’s constitutive laws must therefore be en-
characterised by complex reinforcement characteristics, in particu- riched in order to take the strain rate dependency (i.e. viscoelastic-
lar in terms of fibre orientation. For example, in the frequently en- ity and/or viscoplasticity) of SFRT into account. To achieve this aim,
countered case of a thin injection-moulded part, short fibres tend a better understanding of composite behaviour under dynamic ten-
to orient into distributed in-plane orientations around a preferen- sile test is needed. In particular, a crucial issue is to determine
tial direction, which may be different depending on the location in whether the evolution of SFRT behaviour with strain rate, as high-
the part. For instance, fibres’ preferential direction of orientation is lighted for example by Fitoussi et al. (2013) and Schoβ ig et al.
generally transverse and parallel to injection flow direction in core (2008), is only due to the strain-rate sensitivity of the thermoplas-
and shell layers of the injected part, respectively (Bernasconi et al., tic matrix (e.g. Balieu et al., 2013; Dasari et al., 2003; Epee et al.,
2007; Friedrich, 1985; Vincent et al., 1997). Thin skin zones are 2011; Zeng et al., 2010), or if strain rate also has an influence on
also generally encountered, where fibres follow in-plane random mechanisms of load transmission from matrix to fibres at the in-
orientation (Bernasconi et al., 2007). It can be noticed that this terface. Obviously, involved phenomena are also strongly depen-
layered structure results from the combination of shear flow and dent on local fibre orientation. Therefore, influence of strain rate
fountain flow in injection moulded process (Hull and Clyne, 1996). on SFRT behaviour must be analysed together with microstructure
Thickness of layers and local distributions of orientation depend on heterogeneity, in terms of fibre distributions of orientation.
the properties of composite constituents (matrix viscosity, fibre as- In this framework, this article presents results of quasi-static
pect ratio, etc), fibre content (Doghri and Tinel, 2006; Kammoun and dynamic tensile experimental tests on injection-moulded
et al., 2011) as well as process conditions (injection speed, mould- specimens of short-glass-fibre-reinforced polypropylene (30 wt.%
melt temperature gradient, etc). Determining and/or predicting the of glass fibres). Different strain rates (from about 5 × 10−4 to
exact distribution of fibre orientation in SFRT is therefore a very 50 × s−1 ) and different angles of loading with respect to injection
challenging issue. This issue however deserves consideration since flow direction (from 0° to 90°) are considered. As a basis of com-
SFRT mechanical behaviour obviously strongly depends on fibre parison, similar tests are also performed on unreinforced PP ma-
orientation. Indeed, load transmission from matrix to fibres, and so trix (Section 2.2). An extensive campaign of microstructure anal-
on composite strength and stiffness, increase as fibre angle with ysis using 3D reconstruction of micro-tomographic scans is also
respect to loading direction decreases (Notta-Cuvier et al., 2013). achieved (Section 2.3). Results enable to investigate the anisotropy
Yet, due to complex distributions of fibre orientation, the mechani- of SFRT behaviour not only through the variation of loading angle
cal behaviour of SFRT does not depend only on the angle of loading but also based on the actual local fibre distributions of orientation
but on local orientation of fibres with respect to load direction. It is with respect to macroscopic loading (Section 3). Coupled influence
therefore essential to take into account fibre distributions of orien- of local fibre orientations and strain rate on composite behaviour
tation in the development of accurate models of SFRT mechanical is then analysed (Section 4), in order to improve the understand-
behaviour. ing of the strain-rate dependency of composite behaviour and then
In addition to complex matrix behaviour and reinforcement enrich behaviour models. In particular, it is investigated whether
characteristics, composite behaviour also naturally depends on the the load transmission from matrix to fibre is strain-rate sensitive
properties of fibre/matrix interface, which govern load transmis- or if composite strain-rate sensitivity can be attributed to matrix
sion from matrix to fibres. Maximal load transmission capacity de- viscous overstress only.
pends on the initial adhesion between fibres and matrix, which
itself depends on the nature of constituents, as well as possible
physical and/or chemical treatments (e.g. fibre sizing, compatibi- 2. Material and experiments
lization grafting (Arbelaiz et al., 2005; Pracella et al., 2006; Tiwari
et al., 2012)). Among available theories, shear-lag model (Bowyer 2.1. Materials
and Bader, 1972; Notta-Cuvier et al., 2013) gives a satisfying pre-
diction of load transmission. It relies on the determination of in- Polypropylene (PP) was supplied by Lyon dell Basell (commer-
terfacial shear strength (IFSS) in order to compute the average fi- cial name: Moplen HP500N). Based on supplier information, Mo-
bre axial stress. Among common methods of identification of IFSS, plen HP500N is a homopolymer for injection moulding appli-
there are pull-out or microbond tests (Yue et al., 1995; Zhandarov cations, with a MFR equal to 12 g.(10 min)−1 and a density of
and Mäder, 2014) or single fibre fragmentation tests (Awal et al., 0.9 g.cm−3 . Composite PP-30 wt.% glass fibre (hereafter called PP-
2011)). An important point is that those tests of interface be- 30GF) was supplied by Albis (commercial name: ALTECH PP-H A
haviour characterisation are all performed at very low displace- 2030/159 GF30 CP). PP30GF is a PP homopolymer reinforced by
ment rate. Consequently, there is no information in the literature short-glass fibre with chemical coupling, characterised by a MFR
concerning interface behaviour under high-strain-rate loading. of 2 g.(10 min)−1 and a density of 1.12 g.cm−3 . Both materials were
Finally, even if it is outside the scope of the present arti- kindly provided by Decathlon.
cle, it is worth noting that behaviour of SFRT may be affected PP and PP-30GF plates were injection moulded following the
by several coexisting damage phenomena, such as matrix duc- process conditions prescribed by the suppliers. The mould was
tile damage, whose anisotropy depends on local fibre orienta- filled in by using a unique injection gate. Injected plates are
tion (Notta-Cuvier et al., 2014), fibre/matrix interfacial debonding 200 mm-edge-squares with a thickness of about 2.5 mm. An illus-
188 D. Notta-Cuvier et al. / Mechanics of Materials 100 (2016) 186–197

(a) For quasi-static tests (b) For dynamic tests


Fig. 1. Geometry of specimens for tensile tests (in mm).

tration of the plate geometry, including injection gate and end- During all tensile tests of unreinforced PP, true in-plane strains
filling zone, is shown in Fig. 2 (Section 2.2). were computed using displacement field measurements by digital
image correlation (DIC) (Sutton et al., 2009). Surface of specimens
was painted in white and covered with sprays of black paint in or-
2.2. Tensile tests der to draw a random pattern. This random pattern was divided
into sub-pixel zones (facets), each of them being characterised by
Quasi-static tensile tests were performed at room tempera- a unique signature in grey level, which allows their tracking by
ture using Instron E30 0 0 electromagnetic jack (3 kN load cell sen- DIC software (VIC2D software here). This tracking enables the de-
sor). The geometry of quasi-static tensile specimens follows ISO527 termination of in-plane displacements of facet centres, with re-
norm (Fig. 1a), with a cross-section of about 13 mm2 . Quasi-static spect to the reference image (first image, recorded at an unde-
tests were carried out at displacement rates of 1 and 10 mm.min−1 . formed stage), and so on the computation of in-plane strains at
Those displacement rates correspond to average strain rates of the centre of each facet. Rigid body motion tests were used in or-
about 5.55 × 10−4 and 5.55 × 10−3 s−1 , respectively, since the re- der to analyse the sensitivity of the signal/noise ratio to size of
gion of interest (ROI) is 30 mm high. In the following, the average facets. A compromise between noise level and spatial resolution
strain rate which is computed as the ratio of displacement rate by was achieved for facets of 20 × 20 pix2 (step size, i.e. distance be-
ROI height will be referred to as engineering strain rate. tween two points of displacement measurement, equals to 10 pix).
Dynamic tensile tests were carried out at room temperature us- It can be noted that this facet size in pixels corresponds to ac-
ing Instron 65/20 hydraulic jack (i.e. 65 kN load cell sensor, maxi- tual facets of 1 × 1 mm2 and 1.96 × 1.96 mm2 for quasi-static and
mum speed 20 m.s−1 ). For these experiments on PP and reinforced dynamic tests, respectively.
PP, a piezoelectric load cell, calibrated in the range 0–5 kN, with Due to noise, the accuracy of DIC techniques may be reduced
a precision of 2.5 N, was fixed on the rigid frame of the jack. A at low strain levels (significant noise/strain ratio). Since tensile be-
specific set-up for dynamic test, developed in LAMIH, was used to haviour of PP-30GF was expected to be more brittle than that
clamp the specimen. This set-up prevents specimen loading as long of unreinforced PP, and therefore limited to low strain level, DIC
as imposed test velocity is not reached and stabilised. Geometry of techniques were not used during tensile tests of PP-30GF compos-
specimens is specially designed to fit this set-up (Fig. 1b). Length ite. Instead, axial displacements were measured by optical exten-
of the ROI is reduced compared to quasi-static tensile specimen in sometry, i.e. non-contact elongation measurement based on mo-
order to reach higher strain rates, and now equals 20 mm. In this tion tracking of black-and-white transition lines (targets). During
study, displacement rates of 10, 100 and 1000 mm.s−1 were im- quasi-static tests, optical extensometer ZS16D (CCD line scan sen-
posed (engineering strain rate of about 0.5, 5 and 50 s−1 ). sor - Rudolf GmbH) with a precision of 3 μm over 50 mm was
Tensile specimens were cut in the injection-moulded plates by used. It allows the tracking of up to 10 colour transitions. Here, the
water jet at different angles θ with respect to the injection flow di- elongation of 3 black-painted areas of 3 mm height, spaced apart
rection (IFD). Fig. 2 shows examples of tensile specimen locations by 3 mm with the middle area centred in the ROI, was followed
in PP-30GF plates. Head of composite specimens are localised at a (Fig. 3). Axial strain, ε , can therefore be computed simultaneously
minimum of 50 mm from the injection gate and end-filling zone. over different “gauge” lengths by the ratio of measured axial elon-
The aim is to avoid placing specimen’s ROI in areas where fibre gation between two given colour transitions, L, by the initial
orientation is too chaotic. In the central part of the plate, fibres distance between them, L0 , with ε = L/L0 . The axial strain was
globally tend to follow distributed orientation around IFD, with first computed over the three 3 mm-height areas, which allowed
some variation depending on specimen location on the plate (cf establishing that the axial strain in the central part of PP-30GF
Section 3.1). specimens remained homogeneous during quasi-static tensile tests.
For both quasi-static and dynamic tensile tests, and for all ma- Axial strain considered in the following analysis was then com-
terials, engineering axial stress, σ (hereafter simply referred to as puted over the whole 15 mm height tracked area (i.e. L0 = 15 mm).
axial stress) was computed as the ratio of load, F, measured by the This way, the “gauge” length, L0 , is close to that used for the ex-
load cell sensor, by the initial cross section area at the centre of ploitation of DIC measurements, performed during tests of unre-
the ROI, S0 , so that σ = F/S0 .
D. Notta-Cuvier et al. / Mechanics of Materials 100 (2016) 186–197 189

Fig. 2. Location of tensile specimens in injection-moulded plates.

Table 1
Tensile tests specifications.

Measurement technique Displacement rate Engineering strain-ratea (s−1 ) Frequency of acquisition (Hz) Camera image resolutionc (pix2 )

Load Displacementb

PP DIC 1 mm.min−1 5.55 10−4 20 0.5 1392 × 1040


10 mm.min−1 5.55 10−3 100 5 1392 × 1040
10 mm.s−1 0.5 50 0 0 600 128 × 528
100 mm.s−1 5 50,0 0 0 60 0 0 128 × 528
1 m.s−1 50 50 0,0 0 0 25,0 0 0 128 × 384
PP-30GF composite Optical extensometry 1 mm.min−1 5.55 10−4 100 10
10 mm.min−1 5.55 10−3 10 0 0 10
10 mm.s−1 0.5 10,0 0 0 10,0 0 0
100 mm.s−1 5 10 0,0 0 0 10 0,0 0 0
1 m.s−1 50 10 0 0,0 0 0 10 0 0,0 0 0
a
Engineering strain-rate is the ratio of displacement rate by ROI length (30 and 20 mm, respectively in QS and dynamic tests).
b
Extensometer frequency of acquisition or camera frame rate.
c
In quasi-static tests, 1 pix ≈ 50 μm; in dynamic test, 1 pix ≈ 98 μm.

Table 2
Coordinates of scanned volumes in PP-30GF plates and fibre average component Ayy
of Advani-Tucker orientation matrix.

Specimen Angle θ Location of the Average value of


designation between centre of the Ayy in the whole
specimen axis scanned volumea scanned volumeb
and IFD (mm)

0–1 0° (16;100) 0.7109


0–4 (55;100) 0.6438
0–7 (94;100) 0.6146
0–8 (107;100) 0.6355
0–11 (146;100) 0.6327
0–14 (185;100) 0.6827
20–6 20° (93.26;99) 0.5813
20–7 (107.1;99) 0.6106
45–4 45° (88.79;96.32) 0.5335
45–12 (111.21;96.32) 0.5337
Fig. 3. Colour transitions for displacement tracking by optical extensometry on
60–5 60° (98.53;113.17) 0.5201
quasi-static tensile specimens.
60–13 (101.47;113.17) 0.5175
90–4 90° (53;99) 0.4428
90–11 (147;99) 0.4271
inforced PP. During dynamic tests of PP-30GF, optical extensome-
ter 200XR (Rudolf GmbH - precision of 5 μm over 50 mm), which a
In the system of coordinates of injected plate, (X0 ,Y0 ) (cf Fig. 2). Y0 axis is IFD
allows higher frequency of acquisition than ZS16D, was used. The direction.
b
With y the specimen axis (cf Section 3.1).
tracked zone covered the entire specimen’s ROI (i.e., “gauge” length
of 20 mm).
All technical specifications of tensile tests are given in Table 1. 5 2.3. X-ray micro-computed tomography
specimens of PP and PP-30GF with an angle of cutting θ = 0°, 20°,
45°, 60° and 90° were tested at each quasi-static displacement rate Prior to tensile testing, 15 ISO527-type specimens of PP-30GF
(giving a total of 30 tests per displacement rate). Under dynamic (listed in Table 2), cut at all values of angle θ with respect to IFD,
loading conditions, 5 specimens of PP and 3 specimens of PP-30GF were selected to be analysed by X-ray micro-computed tomogra-
at each angle θ were tested at each displacement rate (i.e. 20 tests phy (μ-CT), in order to determine the distributions of fibre orien-
per displacement rate). tation.
190 D. Notta-Cuvier et al. / Mechanics of Materials 100 (2016) 186–197

(a) 3D reconstruction
(b) Example of 2D image before thresholding
(quarter of the whole scanned volume)
Fig. 4. Reconstruction of PP-30GF microstructure by micro-computed tomography.

Fig. 5. Definition of cutting angle, θ , and fibre angle of orientation with respect to specimen axis, ψ .

High-resolution microtomograph Skyscan 1172 (Bruker Micro fibres present a very low angle with respect to (x, y) plane (i.e.
CT) was used. Voltage of 30 kV and current of 40 μA were used. plane orthogonal to specimen thickness). In the following, fibres
The rotation step was 0.4° (scan over 360°). A spatial resolution, are therefore assumed to have in-plane orientation. Slices of 3D-
i.e. voxel size, of 3.87 μm was attained with these parameters. The scan in the (x, y) plane, at regularly spaced positions in thickness
investigated volume was centred on the specimen ROI. Its height direction, were extracted, in the form of 2D grey-scale images. Im-
(in y-axis) was about 5 mm while it covered the whole specimen age segmentation was performed via grey-level thresholding in or-
width (x-axis, ≈ 5 mm) and thickness (z-axis, ≈ 2.5 mm). The loca- der to distinguish fibres (i.e. structures to be detected) from ma-
tions of the scanned volumes in the composite plate are given in trix material (ambient medium), using Fiji tools of Image J soft-
Table 2 (see also Fig. 2 for the definition of X0 and Y0 axes - the ware (Schindelin et al., 2012), based on Otsu method. The grey-
meaning of Ayy will be given in Section 3.1.). level threshold was fixed “manually” after a trial and error process,
Data acquired by μ-CT allowed the 3D reconstruction of PP- and validated by comparing the whole volume fraction of detected
30GF microstructure (Fig. 4a). Sets of 2D greyscale images, cor- fibres with theoretical fibre volume fraction in the composite ma-
responding to slices of the 3D view, can then be extracted in all terial. It can be noted that similar fibre contents were computed
directions (Fig. 4b, e.g.) and analysed using Fiji image processing in all investigated sub-volumes and for all the specimens, thus
package (available in Image J software - Schindelin et al., 2012), as indicating good fibre dispersion throughout each scanned volume
described in next Section 3.1. and more generally throughout the whole injected plate. The same
grey-level threshold was applied to the whole set of 2D images. Fiji
3. Results tools of structure detection and analyses (Schindelin et al. 2012)
were applied to 2D images in order to identify and count fibres
3.1. Distributions of fibres’ orientation identified by microtomography and determine their characteristics (length, diameter and orienta-
analyses tion). It is worth noting that mean fibre length and radius were
of about 750 μm and 18 μm, respectively. The orientation of each
A first analysis of the 3D reconstruction of PP-30GF microstruc- detected fibre is characterised by fibre angle of orientation with
ture resulting from μ-CT has revealed that the vast majority of respect to specimen axis (angle ψ - Fig. 5).
D. Notta-Cuvier et al. / Mechanics of Materials 100 (2016) 186–197 191

Fig. 6. Distributions of fibre orientation in scanned volume of specimens cut at different plate locations and angles θ with respect to IFD (location of specimens in the
injected plate are given in Table 2).

The analysis of orientation of detected fibres reveals that IFD entation depends on specimen location in the plate (cf Table 2 -
appears in all cases as the preferential orientation of fibres, which Section 2.3 and Fig. 7 for the particular case θ = 0°). For instance,
means that a majority of fibres are oriented along IFD but not distribution curve tends to be sharper around IFD direction, i.e. the
all fibres. Preferential orientation of fibre is therefore equal to ± θ fraction of fibres that are oriented close to IFD increases near plate
with respect to specimen axis, where θ is the cutting angle of edge, as illustrated by specimens 0–1 and 0–14. Specimens 0–7
the specimen (cf Section 2.2 and Fig. 5). More precisely, orienta- and 0–8, on the one hand, and 0–4 and 0–11, on the other hand,
tions are distributed around preferential orientation with a fairly present similar distributions of orientation, thus indicating a quite
large scatter, as can be seen in Fig. 6 for different cutting angles, good symmetry of material properties with respect to plate median
θ , between specimen axis and IFD. In order to plot histograms in line in Y0 direction.
Fig. 6, numbers of fibres characterised by an angle ψ with respect Evolution of fibre orientation through plate thickness (z di-
to specimen axis such that ψ ∈ [α ; α + 2◦ ] were established, for rection) is also investigated. For instance, Fig. 8 shows the av-
α varying from –90 to 88°. Densities were then simply defined erage value of the absolute value of ψ computed for fibres de-
as the ratio of these numbers by the total number of fibres de- tected in 2D images (in (x; y) plane), extracted at different posi-
tected in the whole scanned volume. For clarity, continuous lines tion through material thickness, for 3 specimens cut in IFD direc-
are plotted by extrapolation between points separated by step of tion (i.e. θ = 0°). It highlights the well-known skin-shell-core phe-
2°. Histograms in Fig. 6 also reveal that fibre distribution of ori- nomenon (Friedrich, 1985; Bernasconi et al., 2007, e.g.). In fact,
192 D. Notta-Cuvier et al. / Mechanics of Materials 100 (2016) 186–197

Fig. 7. Example of location of specimens in the injected plate: case θ = 0°.

ented along specimen axis and decreased when fibre direction de-
viates from specimen axis direction (i.e. when |ψ | increases). The
use of Advani–Tucker orientation matrix component (Advani and
Tucker, 1987) allows an unbiased computation of average orienta-
tion when aiming at analysing the influence of fibre orientation on
composite mechanical behaviour. First, it appears that the average
value of Ayy decreases when the cutting angle, θ , between spec-
imen axis and IFD, increases. This result is logical since fibres are
preferentially oriented along IFD in the injected plate. Thus, the av-
erage value of Ayy ranges from 0.61 to 0.71 for θ = 0° and decreases
up to 0.43–0.44 for θ = 90°.
In summary, micro-computed tomography analyses revealed
that short-glass fibres embedded in PP matrix are actually follow-
ing in-plane distributions of orientation. Preferential orientation is
globally along the injection flow direction if considering orienta-
tion averaged over the scanned volumes. However, some variations
are noticed, in particular in the specimen skins, where fibres tend
to orient randomly, and core, where they tend to have higher an-
Fig. 8. Evolution of average fibre orientation in specimen thickness - Specimens 0–
gle of orientation with respect to IFD. In addition, μ-CT analyses
1, 0–4 and 0–7 (θ = 0°).
revealed that distribution of fibre orientation slightly varies de-
pending on specimen location in the injected plate. In the fol-
lowing paragraph, specimens cut at different angles with respect
mean angle of fibre orientation tends to increase in plate skins, to IFD are subjected to tensile loading, under quasi-static and dy-
where fibres tend to orient randomly (i.e. theoretical mean an- namic test conditions. The influence of microstructure heterogene-
gle equal to 45°), due to faster cooling near mould walls, then ity on PP-30GF mechanical behaviour can thus be investigated over
decreases in shell layers, where fibres are preferentially oriented a large range of strain rate.
along IFD, and increases again in core layer where fibres tend the-
oretically to orient transversally to IFD. As stated in the intro- 3.2. Strain-rate sensitivity of PP matrix and PP-30GF composite for
duction section, this layered structure results from the combina- different fibres’ distributions of orientation
tion of shear flow and fountain flow in injection moulded process
(Hull and Clyne, 1996). In the present case, it is interesting to note Fig. 9 shows axial tensile behaviour of unreinforced PP, for all
that the thickness of the core layer where fibre orientation notice- investigated displacement rates. Axial strains are extracted from
ably deviate from IFD tends to be higher in intermediate areas of DIC data (average over all facets of the ROI). Since tensile tests
the plate (where X0 is close to 1/4 or 3/4 of the plate width - of PP showed very good reproducibility at all test speeds, only
Fig. 7), as illustrated by specimen 0–4 in Fig. 8, than near plate one behaviour curve per displacement rate is presented. Strain-
edge or in the central part (specimens 0–1 and 0–7, respectively). rate sensitivity of PP is obvious, with an increase of stiffness and
In addition, average fibre angle of orientation tends to increase strength with the engineering strain rate simultaneously with a
when the thickness of the core layer decreases (Fig. 8). It explains drop in axial strain at break.
the secondary peaks that are observed at ψ = ± 45° on global fibre Composite PP-30GF shows quite brittle behaviour at all dis-
orientation distribution in the whole scanned volume of specimens placement rates, as can be seen in Fig. 10, for instance for cut-
0–4 and 0–11, peaks that are not detected in specimens 0–1, 0–7, ting angles θ = 0° and 90° (the same trends are noticed for other
0–8 and 0–14 (Fig. 6a). values of θ ). As for unreinforced PP, composite axial stress and ap-
Finally, values of Advani–Tucker orientation matrix compo- parent rigidity both increase with the engineering strain rate. Val-
nent Ayy averaged in whole scanned volumes are presented in ues of composite strain at break remain low whatever the displace-
Table 2 (Section 2.3). Here, Ayy is simply equal to cos(ψ )2 , where ment rate, and there is no marked trend of evolution with strain-
ψ is the angle of fibre orientation with respect to specimen axis, rate (Fig. 11). Interestingly, composite strain at break globally in-
y. Then, average value of Ayy is equal to 1 when fibres are all ori- creases when increasing the angle of cutting, θ , at all displacement
D. Notta-Cuvier et al. / Mechanics of Materials 100 (2016) 186–197 193

Fig. 9. Axial tensile behaviour of PP matrix at different displacement rates.


Fig. 11. PP30-GF strain at break for different cutting angles and engineering strain
rates (Crosses stand for average values; vertical bars indicate standard deviation).

rates (except for unexplained low values for the case θ = 60°, at
gle ψ ∈ [0°;10°[∩[170°;180°] (that corresponds to Ayy > 0.9698) in
1 mm/min (5 × 10−4 s−1 ) - Fig. 11). This may be explained by the
scanned volume of specimen 0–14, to be compared to 33 and 30%
fact that matrix behaviour takes a more important part in compos-
in the cases of specimen 0–8 and 0–11, respectively. Then, with
ite behaviour when θ increases, because the proportion of fibres
higher average value of Ayy , indicating that fibres are globally ori-
that are oriented closely to loading direction decreases. As a con-
ented more closely to loading direction, tensile stress level at a
sequence, strain at break increases while stress level decreases (as
given strain of specimen 0–14 is higher than that of other spec-
also noticed here). However, the increase of strain at break with
imens (specimens 0–10 and 0–9 were not analysed by X-ray mi-
angle of cutting remains low (e.g. about 4 to 6% from θ = 0° to
crotomography). As demonstrated before, specimen cutting angle
θ = 90° at 1 mm/min, to be compared to more than 20% for unre-
with respect to IFD, θ , has a direct influence on mean fibre orien-
inforced PP). This may indicate that composite damage is mostly
tation with respect to specimen axis. More precisely, average value
governed by fibre/matrix interface debonding, which seems to be
of Ayy increases when θ decreases. Therefore, composite stiffness
activated at a lower strain level than matrix ductile damage and
and strength are logically improved when θ decreases (Figs. 13
only slightly delayed at high angle of cutting (i.e. at higher av-
and 14, for displacement rate of 1 mm.min−1 and 100 mm.s−1 , re-
erage angle between fibres and loading direction) . This is how-
spectively). As previously stated, tensile strain at break of PP-30GF
ever just an assumption at this stage and failure analysis is outside
composite is slightly influenced by specimen angle of cutting θ but
the scope of this article. Thorough work is currently being done
behaviour of PP-30GF remains quite brittle in all cases and at all
to better understand damage and failure mechanisms in PP-30GF
test speeds.
composites, based in particular on SEM and microtomographic ob-
servations and in situ tests, as well as on tensile tests on notched
specimens that are more appropriate for the analysis of failure be- 4. Coupled influence of strain-rate and fibre orientation on
haviour. reinforcement efficiency
Fig. 10 reveals a significant scattering of PP-30GF behaviour at
a given test speed. Fig. 12 illustrates this scattering for the ex- In this study, reinforcement efficiency is defined as the compos-
ample of θ = 0°, at 10 mm.min−1 . In fact, scattering is directly re- ite gain in tensile stress, at a given tensile axial strain, compared
lated to the difference observed between distributions of fibre ori- to unreinforced matrix. In Section 3, experimental results showed
entation. Indeed, microtomography analysis of specimen 0–14 (cut that reinforcement efficiency is strongly dependent on fibre ori-
close to plate edge) revealed that average value of component Ayy entation. Indeed, the load that is potentially transmitted from the
of Advani–Tucker orientation matrix (where axis y is specimen PP matrix to a given high-rigidity glass fibre increases when the
axis, i.e. loading direction) in the whole scanned volume is equal component Ayy of its orientation matrix increases, that is to say
to 0.6827 (cf Table 2), that is higher than for specimens 0–8 and when the angle ψ between that fibre and the loading direction de-
0–11 (about 0.63). To go further, 40% of fibres are oriented with an- creases (Notta-Cuvier et al., 2013, 2014, 2015a, 2015b). For instance,

Fig. 10. Tensile axial behaviour of PP-30GF - θ = 0° and θ = 90°, for all displacement rates.
194 D. Notta-Cuvier et al. / Mechanics of Materials 100 (2016) 186–197

Fig. 12. Tensile axial behaviour of PP-30GF - θ = 0° for quasi-static displacement rate of 10 mm.min−1 .

Fig. 13. Tensile axial behaviour of PP-30GF at displacement rate of 1 mm.min−1 for all values of θ .

difference between average value of Ayy of only 0.05 (Fig. 12 and


Table 2).
Strain-rate sensitivity of tensile strength and rigidity of unre-
inforced PP and PP-30GF composite has also been proven thanks
to quasi-static and dynamic tensile tests at different displacement
rates (engineering strain rate varying from 5 × 10−4 s−1 to 50 s−1 ).
Since behaviour of mineral fibres is generally considered as insen-
sitive to strain-rate (Reder et al., 2003), it is then interesting to
analyse whether strain-rate sensitivity of composite is merely the
consequence of additional viscous stress in PP matrix or if strain
rate has also an influence on interfacial behaviour that governs
load transmission from PP to glass fibres. In other words, it should
be investigated whether efficiency of PP short-glass-fibre reinforce-
ment depends on coupled influence of fibre orientation and strain-
rate.
A quantitative evaluation of reinforcement efficiency through-
Fig. 14. Tensile axial behaviour of PP-30GF at displacement rate of 100 mm.s−1 for
out the tensile test can be given by the relative gap between the
all values of θ .
volume energy of deformation of composite and unreinforced ma-
trix. Volume energy of deformation, E, up to any value of axial
strain, ε max , is given by the integral (1) of behaviour curve, σ vs.
in the particular case of a fibre aligned with loading direction (i.e. ε.
Ayy = 1), the fraction of load applied to the composite that is trans-  ε=εmax
mitted to fibres through PP matrix is maximal (applied load is in- E (εmax ) = σ dε (1)
tegrally transmitted to fibres in case of perfect fibres/matrix adhe- ε=0
sion). For that reason, even small variations of local fibre distribu- where σ is the axial engineering stress of unreinforced PP or PP30-
tions of orientation may result in a noticeable variation of compos- GF composite. ε is the axial tensile strain, obtained by optical ex-
ite axial stress, for the same angle of loading θ , as demonstrated tensometry for the composite and by DIC technique (average over
in Section 3. For instance, at a value of composite axial strain of all facets within the ROI) for PP (see Section 2.2).
3.5%, engineering axial stress of specimen 0–14 is 5 MPa higher To avoid bias due to the scattering of axial strain at break, areas
(about 7.5% of tensile strength) than that of specimen 0–11 for a are computed up to ε max = 0.03. Finally, reinforcement efficiency,
D. Notta-Cuvier et al. / Mechanics of Materials 100 (2016) 186–197 195

Table 3
Estimation of fibre average component Ayy of Advani–Tucker orientation matrix, in central volume of some dynamic specimens.

Designation of dynamic Angle θ between Test displacement rate Location of the centre Corresponding Assumed average value
specimen specimen axis and IFD of the specimena (mm) quasi-static specimenb of Ayy in the whole
scanned volumec

D0-1 0° 10 mm.s−1 (184;104) 0–14 0.6827


D0-3 1 m.s−1 (58;104) 0–4 0.6438
D0-5 10 mm.s−1 (100;104) Between 0–7 and 0–8 0.6251
D0-7 1 m.s−1 (142;104) 0–11 0.6327
D0-10 100 mm.s−1 (97;95) 0–7 0.6146
D45-3 45° 10 mm.s−1 (116.26;98.65) 45–12 0.5337
D45-9 10 mm.s−1 (88.39;99.65) 45–4 0.5335
D60-11 60° 100 mm.s−1 (103.80;112.21) 60–13 0.5175
D90-3 90° 10 mm.s−1 (143;104) 90–11 0.4271
D90-8 100 mm.s−1 (57;104) 90–4 0.4428
a
In (X0 ,Y0 ) system of coordinates (cf Fig. 2). Y0 axis is IFD direction.
b
Quasi-static specimen whose centre is located very close to that of dynamic specimen in the plate’s system of coordinates.
c
With y the specimen axis (i.e. loading direction).

age value of Ayy decreases, since load transmission to fibres de-


creases when Ayy decreases. More interestingly, the same trend of
evolution can be observed for all testing speeds, including for dy-
namic tests corresponding to strain rate of 0.5 and 5 s−1 . Average
values of Ayy in the two specimens loaded at 50 s−1 are too close
to each other to allow the extraction of a reliable trend of evolu-
tion. However, it seems to be similar than for other strain rates.
These results suggest that reinforcement efficiency is not strain-
rate sensitive. In other words, fibre behaviour as well as mecha-
nisms of load transmission at fibre/matrix interface do not seem
to be influenced by strain rate. Finally, all or almost increase of
tensile strength of PP-30GF composite with loading rate can be at-
tributed to strain-rate sensitivity of PP matrix.
These results are also insightful for the modelling of mechanical
behaviour of SFRT. Indeed, numerous models are currently devel-
oped for SFRT (Kammoun et al., 2011; Miled et al., 2013; Modniks
Fig. 15. Influence of fibre orientation and engineering strain rate on reinforcement and Andersons, 2013; Notta-Cuvier et al., 2013, 2014, 2015a, 2015b)
efficiency. but very few of them (only work by Miled et al. (2013) to authors’
knowledge but without experimental validation at high strain rate)
are dedicated to the modelling of strain-rate sensitivity of these
E, is defined by:
materials, mainly due to lack of experimental results. In this frame-
 
  EC Āyy , u˙ , εmax − EM (u˙ , εmax ) work, this study first demonstrates that gain in composite ten-
E Āyy , u˙ , εmax = (2) sile stress at high strain rate results from matrix viscous over-
EM (u˙ , εmax )
stress only. An important consequence for modelling purpose is
where EC and EM are respectively the volume deformation energy that short-glass fibre behaviour can be modelled in the same way
computed for PP-30GF composite, for a given displacement rate, (generally in linear elasticity) whatever the strain rate. It is worth
u˙ , and mean component of orientation matrix in loading direction, noting that this result can probably be extended to other mineral
Āyy , and for PP matrix, for the same displacement rate. fibres but not to vegetal fibres that can have viscoelastic behaviour
It is worth noting that the true axial strain field that is com- (Placet et al., 2014). Another important result is that load trans-
puted during PP tensile tests using DIC technique always remains mission from matrix to fibres does not seem to be influenced by
homogeneous over the whole ROI at least up to ε max , for all spec- strain rate. Then, parameters of interfacial behaviour, such as in-
imens and all displacement rates. Using a value of axial strain av- terfacial shear strength, that are identified at very low test speed
eraged over all facets of the ROI therefore does not introduce any (by pull-out or microbond tests (Yue et al., 1995; Zhandarov and
error in the computation of EM . Mäder, 2014) or single fibre fragmentation tests (Awal et al., 2011))
In order to analyse both influence of strain rate and fibre orien- remain valid for the modelling of SFRT behaviour at higher loading
tation on reinforcement efficiency, fibre orientations in specimens rates.
for dynamic tests must be known. These specimens have not been
analysed using X-ray microtomography. However, since a good re- 5. Conclusions and future work
peatability of distributions of orientation from one plate to an-
other has been observed, average fibre orientation in the centre In this study, unreinforced polypropylene (PP) and injection-
of a given dynamic specimen was assumed to be the same as in moulded short-glass-fibre reinforced polypropylene (PP-30GF) are
the quasi-static specimen cut at the same location in the plate and subjected to tensile loadings, under several quasi-static and dy-
with the same cutting angle θ . Estimated average values of com- namic loading rates, and different loading directions for the com-
ponent Ayy of fibre orientation matrix, with axis y the loading di- posite. Simultaneously, composite microstructure is thoroughly
rection, are listed in Table 3. analysed thanks to an extensive campaign of X-ray microtomog-
The evolution of the reinforcement efficiency versus the average raphy. Scans reveal a strong heterogeneity of the microstructure,
value of Ayy is plotted in Fig. 15. At a given value of engineering in terms of fibre orientation. In fact, short-glass fibres follow in-
strain rate, reinforcement efficiency logically decreases when aver- plane distributions of orientation around a preferential orientation
196 D. Notta-Cuvier et al. / Mechanics of Materials 100 (2016) 186–197

in shell and core layers of the injected composite plates, with pref- Acknowledgments
erential orientation parallel to the injection flow direction (IFD) in
shell layers and highly angled with respect to IFD in core layer. The present research work has been supported by the Interna-
Thin skin zones are also detected where fibres follow in-plane tional Campus on Safety and Intermodality in Transportation, the
random orientation. In addition, fibre local orientations depend Franco-Tunisian joint comity for university cooperation (CMCU) the
on location on the plate (distance from injection gate and plate Nord Pas-de-Calais Region, the European Community, the Regional
edge). Delegation for Research and Technology, the Ministry of Higher
The mechanical behaviour of both the unreinforced PP ma- Education and Research and the National Centre for Scientific Re-
trix and PP-30GF composite is strain-rate sensitive, with an in- search (CNRS).
crease of tensile strength and stiffness with strain rate. How- Authors are also sincerely grateful to Antoine Giboire (De-
ever, if matrix strain at break clearly decreases with the strain cathlon SA) who very kindly provided us with the materials used
rate, there is not marked trend of evolution of composite strain in this study and made Decathlon injection-moulding facilities
at break as a function of strain rate and the composite remains available to us.
quite brittle at all investigated displacement rates. This is likely
to reveal that damage of PP-30GF composite is dominated by
References
debonding at fibre/matrix interface. This phenomenon seems to
initiates at relatively low strain level and to be only slightly Advani, S.G., Tucker, C.L., 1987. The use of tensors to describe and predict fiber ori-
delayed when angle between fibre axis and loading direction entation in short fiber composites. J. Rheol. 31, 751–784.
increases. Arbelaiz, A., Fernández, B., Ramos, J.A., Retegi, A., Llano-Ponte, R., Mondragon, I.,
2005. Mechanical properties of short flax fibre bundle/polypropylene compos-
Combined analyses of micro-computed tomography scans and ites: Influence of matrix/fibre modification, fibre content, water uptake and re-
tensile tests at several loading direction highlight the strong de- cycling. Composites Sci. Technol. 65, 1582–1592.
pendence of macroscopic mechanical behaviour of PP-30GF com- Awal, A., Cescutti, G., Ghosh, S.B., Müssig, J., 2011. Interfacial studies of natural fi-
bre/polypropylene composites using single fibre fragmentation test (SFFT). Com-
posite on local fibre orientation. This dependence appears stronger
posites 42, 50–56.
than a simple effect of loading angle. Indeed, a variation of only a Balieu, R., Lauro, F., Bennani, B., Delille, R., Matsumoto, T., Mottola, E., 2013. A
few degrees of the average angle of fibre orientation with respect fully coupled elastoviscoplastic damage model at finite strains for mineral filled
semi-crystalline polymer. Int. J. Plasticity 51, 241–270.
to loading direction, which corresponds to a variation of a few
Bernasconi, A., Davoli, P., Basile, A., Filippi, A., 2007. Effect of fibre orientation on the
hundredth of the average value of the component of fibre orien- fatigue behaviour of a short glass fibre reinforced polyamide-6. Int. J. Fatigue 29,
tation matrix with respect to loading direction, can lead to a vari- 199–208.
ation of about 7.5% of the composite tensile strength. Therefore, an Bowyer, W.H., Bader, M.G., 1972. On the re-inforcement of thermoplastics by imper-
fectly aligned discontinuous fibres. J. Mater. Sci. 7, 1315–1321.
accurate prediction of composite mechanical response cannot be Charlet, K., Béakou, A., 2011. Mechanical properties of interfaces within a flax bundle
based only on the consideration of macroscopic loading direction, - Part I: Experimental analysis. Int. J. Adhesion Adhesives 31, 875–881.
assuming that all fibres are oriented along injection flow direction, Dasari, A., Rohrmann, J., Misra, R.D.K, 2003. Microstructural evolution during tensile
deformation of polypropylenes. Mater. Sci. Eng. A 351, 200–213.
but must also takes local fibre distribution of orientation into ac- Doghri, I., Tinel, L., 2006. Micromechanics of inelastic composites with misaligned
count. inclusions: Numerical treatment of orientation. Comput. Methods Appl. Mech.
The reinforcement efficiency is defined here as the relative gap Eng. 195, 1387–1406.
Epee, A.F., Lauro, F., Bennani, B., Bourel, B., 2011. Constitutive model for a
between the energy of deformation (up to an axial strain of 3%) of semi-cristalline polymer under dynamic loading. Int. J. Solids Struct. 48, 1590–
the composite and of the unreinforced PP matrix, under the same 1599.
loading conditions. Thorough analysis of reinforcement efficiency Fitoussi, J., Bocquet, M., Meraghni, F., 2013. Effect of matrix behavior on the damage
of ethylene-propylene glass fiber reinforced composite subjected to high strain
demonstrates that load transmission in PP-short glass fibres com-
rate tension. Composites 45, 1181–1191.
posite is not strain-rate sensitive, at least up to 50 s−1 . Therefore, Friedrich, K., 1985. Microstructural efficiency and fracture toughness of short
strain-rate sensitivity of the composite can be attributed to matrix fiber/thermoplastic matrix composites. Composite Sci. Technol. 22, 43–74.
Hull, D., Clyne, T.W., 1996. An Introduction to Composite Materials, 2nd ed Cam-
viscous overstress only.
bridge solid state science series.
This study therefore provides valuable information on the high- Kammoun, S., Doghri, I., Adam, L., Robert, G., Delannay, L., 2011. First pseudo-grain
strain-rate behaviour of short-glass-fibre-reinforced polypropylene, failure model for inelastic composites with misaligned short fibers. Composites
in relation to local reinforcement properties, which is crucial for 42, 1892–1902.
Miled, B., Doghri, I., Brassart, L., Delannay, L., 2013. Micromechanical modeling of
the extension of behaviour models of SFRT to cases of high strain- coupled viscoelastic–viscoplastic composites based on an incrementally affine
rate loadings. In particular, it is now possible to consider that formulation. Int. J. Solids Struct. 50, 1755–1769.
mechanisms of load transmission remain the same whatever the Modniks, J., Andersons, J., 2013. Modeling the non-linear deformation of a short-
-flax-fiber-reinforced polymer composite by orientation averaging. Composites
strain rate. A very important consequence is that the value of 54, 188–193.
the interfacial shear strength that is identified at very low strain Mouhmid, B., Imad, A., Benseddiq, N., Benmedakhène, S., Maazouz, A., 2006. A study
rate remains valid for the computation of composite dynamic be- of the mechanical behaviour of a glass fibre reinforced polyamide 6,6: Experi-
mental investigation. Polymer Testing 25, 544–552.
haviour. Notta-Cuvier, D., Lauro, F., Bennani, B., Balieu, R., 2013. An efficient modelling
Yet, this study does not give information about the strain- of inelastic composites with misaligned short fibres. Int. J. Solids Struct. 50,
rate dependency of mechanisms of damage and failure in SFRT. 2857–2871.
Notta-Cuvier, D., Lauro, F., Bennani, B., Balieu, R., 2014. Damage of short-fibre rein-
In particular, future works will have to investigate whether vari-
forced materials with anisotropy induced by complex fibres orientations. Mech.
ation of local mechanical state in the matrix induced by strain- Mater. 68, 193–206.
rate sensitivity may influence the initiation and evolution of dam- Notta-Cuvier, D., Lauro, F., Bennani, B., 2015a. Modelling of progressive fibre/matrix
debonding in short-fibre reinforced composites up to failure. Int. J. Solids Struct.
age mechanisms, such as interfacial debonding or matrix micro-
66, 140–150.
cracking. In that framework, extensive campaigns of experimen- Notta-Cuvier, D., Lauro, F., Bennani, B., Nciri, M., 2015b. Impact of natural variability
tal testing and observations are currently taking place or are of flax fibres properties on mechanical behaviour of short-flax-fibre-reinforced
planned soon. Among them, there are classical tensile tests on polypropylene. J. Mater. Sci. 51, 2911–2925.
Placet, V., Cissé, O., Lamine Boubakar, M., 2014. Nonlinear tensile behaviour of el-
notched specimens but also in-situ tensile tests in microtomo- ementary hemp fibres. Part I: Investigation of the possible origins using re-
graph, as well as comparisons of damaged microstructures after peated progressive loading with in situ microscopic observations. Composites
quasi-static and dynamic loadings (microtomograph and SEM), for 56, 319–327.
Pracella, M., Chionna, D., Anguillesi, I., Kulinski, Z., Piorkowska, E., 2006. Function-
instance. alization, compatibilization and properties of polypropylene composites with
Hemp fibres. Composites Sci. Technol. 66, 2218–2230.
D. Notta-Cuvier et al. / Mechanics of Materials 100 (2016) 186–197 197

Reder, C., Loidl, D., Puchegger, S., Gitschthaler, D., Peterlik, H., Kromp, K., Khatibi, G., Tiwari, S., Bijwe, J., Panier, S., 2012. Optimization of surface treatment to enhance
Betzwar-Kotas, A., Zimprich, P., Weiss, B., 2003. Non-contacting strain measure- fiber-matrix interface and performance of composites. Wear 274-275 326–334.
ments of ceramic and carbon single fibres by using the laser-speckle method. Vincent, M., Devilers, E., Agassant, J.-F., 1997. Fibre orientation calculation in injec-
Composites 34, 1029–1033. tion moulding of reinforced thermoplastics. J. Non-Newtonian Fluid Mech. 73,
Reis, J.M.L., Chaves, F.L., da Costa Mattos, H.S., 2013. Tensile behaviour of glass fibre 317–326.
reinforced polyurethane at different strain rates. Mater. Des. 49, 192–196. Yue, C.Y., Looi, H.C., Quek, M.Y., 1995. Assessment of fibre–matrix adhesion and in-
Sato, N., Kurauchi, T., Sato, S., Kamigaito, O., 1991. Microfailure behaviour of ran- terfacial properties using the pull-out test. Int. J. Adhesion Adhesives 15, 73–
domly dispersed short fibre reinforced thermoplastic composites obtained by 80.
direct SEM observation. J. Mater. Sci. 26 (14), 3891–3898. Zeng, F., Le Grognec, P., Lacrampe, M.F., Krawczak, P., 2010. A constitutive model for
Schindelin, J., Arganda-Carreras, I., Frise, E., 2012. Fiji: an open-source platform for semi-crystalline polymers at high temperature and finite plastic strain: Applica-
biological-image analysis. Nat. Methods 9 (7), 676–682. tion to PA6 and PE biaxial stretching. Mech. Mater. 42, 686–697.
Schoßig, M., Bierögel, C., Grellmann, W., Mecklenburg, T., 2008. Mechanical behavior Zhandarov, S., Mäder, E., 2014. An alternative method of determining the local in-
of glass-fiber reinforced thermoplastic materials under high strain rates. Polym. terfacial shear strength from force-displacement curves in the pull-out and mi-
Testing 27, 893–900. crobond tests. Int. J.Adhesion Adhesives 55, 37–42.
Sutton, M.A., Orteu, J.-J., Schreier, H.W., 2009. Image correlation for shape, mo-
tion and deformation measurements. Basic Concepts, Theory and Applications.
Springer Edition.

You might also like