You are on page 1of 11

Composites: Part A 36 (2005) 279–289

www.elsevier.com/locate/compositesa

The effect of hydrostatic pressure on the mechanical properties


of glass fibre/epoxy unidirectional composites
P.J. Hinea,*, R.A. Ducketta, A.S. Kaddourb, M.J. Hintonc, G.M. Wellsd
a
IRC in Polymer Science and Technology, University of Leeds, Leeds LS2 9JT, UK
b
QinetiQ, Future Systems Technology Division, Farnborough, Hampshire GU14 0LX, UK
c
QinetiQ, Future Systems Technology Division, Fort Halstead, Sevenoaks, Kent TN14 7BP, UK
d
DSTL, Farnborough, Hampshire GU14 0LX, UK

Abstract
An investigation has been carried out to show the effect of hydrostatic pressure up to 850 MPa on the mechanical properties of pure epoxy
and unidirectional (UD) E glass/epoxy lamina. Longitudinal tension, transverse compression and 108 off-axis tension tests were carried out
on the composites and compression tests were carried out on the resin. Strain gauges were used to measure the strains during tests at pressures
up to 500 MPa. From the longitudinal tensile tests on UD lamina, the longitudinal tensile Young’s modulus was found to increase slightly
with increasing pressure, while the longitudinal tensile strength was found to fall, with a change in failure mode from axial splitting to a
transverse break at elevated pressures. From the transverse compression tests on UD lamina, the transverse compressive Young’s modulus
and transverse compressive strength were both found to increase markedly with pressure. A knee in behaviour was seen in the transverse
compressive strength at a pressure of about 300 MPa: below this pressure the failure appeared flaw dominated, while above this pressure the
failure appeared yield dominated. The in-plane UD lamina shear strength, determined from the 108 off-axis tension tests, showed a similar
behaviour to the transverse compression tests, with a transition in the failure behaviour at a pressure of around 300 MPa.
Compression tests on pure epoxy samples showed that the compressive Young’s modulus increased significantly with increasing pressure
whilst the compressive strength showed a modest increase with increasing pressure.
By using a micromechanical model and the measured pressure dependence of the epoxy resin, it was found that the pressure dependent
increase in the transverse modulus for a UD lamina could be predicted accurately.
q 2004 Elsevier Ltd. All rights reserved.

Keywords: A. Polymer matrix composites; B. Mechanical properties; High pressure

1. Introduction remains the need to obtain a better understanding of the


strength under complex loading. This need has been
Fibre reinforced plastics materials are increasingly recently highlighted through a well-publicised activity,
becoming potential candidates for replacing conventional known as the World-Wide Failure Exercise (WWFE),
materials in a variety of sub-sea, naval, automobile, which was organised and co-ordinated by Hinton et al.
aerospace and other applications. Triaxial loadings may be [10,11]. The exercise was aimed at:
encountered in many situations, such as bolted joints, thick
blades, underwater structures. Hence, for a better utilisation † establishing the current level of maturity of theories for
of composites, it is essential that the mechanical failure predicting the failure response of fibre reinforced plastic
behaviour under triaxial loading is adequately understood. laminates.
Although some work has been published describing the † closing the knowledge gap between theoreticians and
behaviour of composites under triaxial stress [1–9] there design practitioners in this field.
† stimulating the composites community into providing
design engineers with more robust and accurate failure
* Corresponding author. Tel.: C44-113-343-3827.
E-mail address: p.j.hine@leeds.ac.uk (P.J. Hine). prediction methods, and the confidence to use them.

1359-835X/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compositesa.2004.06.004
280 P.J. Hine et al. / Composites: Part A 36 (2005) 279–289

(c) transverse compression tests. Pure epoxy specimens were


In the exercise, invited participants, including leading
made in the form of cylindrical bars of 9 mm diameter and
designers and academia, employed a wide range of predic-
13.5 mm long.
tion methodologies to solve challenging test cases. The
Table 1 shows the dimensions for the specimens used to
methodologies focussed mainly on in-plane biaxial loading
characterise the lamina behaviour. The dimensions were
of composites—that being a sufficiently controversial area to
largely chosen based on the available arrangement inside
start with. Little attention was devoted to triaxial test cases in
the cavity of the high pressure rig and the capacity of the rig
the WWFE because of a lack of experimental data under
(see below for details of the apparatus). Figs. 1 and 2 show
triaxial loading with a good pedigree and also a lack of
schematics of the specimens used and the loading
confidence in the current failure theories even under in-plane
loading, let alone any more generalised 3D stress states. configurations. The specimens used for longitudinal and
The present paper aims to fill a gap in the experimental off-axis tension had aluminium tabs bonded at each end
data under triaxial stress by presenting experimental results using a cold-cured epoxy resin adhesive.
on the failure of E-glass/MY750 epoxy resin unidirectional The fibre volume fraction of each of these panels was
(UD) composite subjected to longitudinal tension, trans- measured using a burn off test, and the values obtained are
verse compression and in-plane shear under superposed shown in Table 1: five measurements were made for each
hydrostatic pressure. This grade of material has been well panel. In the thinner laminates, the fibre volume fraction
characterised over the last 25 years through extensive showed significant scatter. The fibre volume fraction of the
uniaxial and biaxial loading experiments at QinetiQ and thicker panel was very consistent.
UMIST [12]. Hence, the present results serve to extend the
characteristics of this material under triaxial loads. 2.2. High pressure apparatus

The tests were carried out using a high pressure testing


2. Experimental details facility [13,14] developed at Leeds University, Fig. 3. The
specimen to be tested is located within a cylindrical steel
2.1. Material details vessel, into which a fluid (50/50 mixture of castor oil and
brake fluid) is introduced via a pressure intensifier. The
Tests were carried out on pure epoxy resin and E glass fibre pressure in the vessel is controlled by the switching of a
reinforced epoxy composite laminates manufactured by solenoid valve which is in line with the intensifier. Force is
QinetiQ (formerly known as the Defence Research and applied to the sample through a pull rod which exits the
Evaluation Agency DERA). The E-glass fibre reinforcement vessel through a specially designed seal (a Morrison seal).
was Silenka 051L, 1200 tex and the epoxy resin system was The other end of the rod is enclosed in another vessel (also
Ciba-Geigy MY750/HY917/DY063 mixed in weight pro- with a Morrison seal) to balance the end load on the rod due
portions of 100:85:2. The curing cycle was 2 h at 90 8C to the hydrostatic pressure.
followed by 1.5 h at 130 8C and 2 h at 150 8C, see Ref. [12] for The load during each test is measured using external
more details. The UD composite panels were made by filament strain gauged bars. These were calibrated against a
winding using a flat mandrel, where the mandrel was in the calibrated testing machine before the testing programme
form of a thin rectangular box. The assembly of the composites commenced. The measured load, therefore, includes the
and the mandrel was then cured in an oven. Fibre orientation friction between the pull rod and the two Morrison seals.
measurements carried out on the cured composites, using The seal friction is measured before and after the test and
image analysis, gave an average fibre angle to the reference this is subtracted from the measured loads as described in
axis of 1.48 (hcos2qiZ0.9994, where q is the angle between a Section 3.
fibre and the reference axis). Samples of different thicknesses For tensile testing (axial tensile and 108 off-axis tests),
were used to carry out three types of test: (a) longitudinal the specimen (with aluminium end tabs) is held in self-
(along the fibre) tensile tests, (b) 108 off-axis tensile tests, and tightening grips which are tightened by hand to a set torque
Table 1 before the test. For compression tests (transverse com-
Details of the composite laminates made for this study pression), a small compression cage is used, where the
sample sits between two metal surfaces which apply the
Test type Axial tension 108 Off-axis Transverse
tension compression compressive load as the machine operates in tension.
The pressure fluid used is an equal mixture of castor oil
Nominal sample 0.6!5! 0.7!5! 5!5!20 mm
dimensions 50 mm 50 mm and brake fluid. Normally, the sample is protected from any
Fibre volume 58G6 38G6 60G1 effects of the pressure fluid by encasing it in a rubber sheath.
fraction (%) For this programme, as with previous tests on composites
Properties Strength, Strength Strength, [8], the samples were protected by painting with a rubber
measured modulus, modulus,
solution (Copydex) and left to dry overnight. This rubber
failure strains failure strains
coating has the additional purpose of protecting the machine
P.J. Hine et al. / Composites: Part A 36 (2005) 279–289 281

Fig. 1. Schematic diagrams of the testing geometries: (a) longitudinal and off-axis tension; (b) transverse compression; (c) pure epoxy resin in compression.

from any debris which might come from a failed composite in an oven at 75 8C for 4 h after an overnight cold cure. It
sample. was found that this system could survive around eight high
The test procedure involved filling the vessel, pressurising pressure tests, up to a pressure of 500 MPa and therefore no
to the desired pressure and then waiting 10 min for the effects strains were measured above this pressure. When failure of
of any adiabatic compression heating to be dissipated before the epoxy coated plug did occur, it was due to one of the
carrying out the test. As samples are pressurised significant copper wires failing, as the plug tried to extrude through
dimensional changes occur which can lead to significant the outer hole, and no leakage was encountered through the
sample loads if the movement of the samples is restricted. For epoxy coated plug.
the tensile tests, the sample holder allows this movement to A simple data acquisition system was used to record the
occur without applying any load to the sample. For the force, testing pressure and sample strain using a Picoscope
compression tests, a small load is initially applied to the 12 bit data logger. All the tests were carried out at a nominal
sample to stop it falling over when the vessel is filled. As crosshead speed of 2 mm/min.
the pressure in the vessel is increased and the sample shrinks,
the crosshead is moved manually to keep this small preload 3. Results
constant.
Various strain gauges were used to monitor the strain in A typical experimental data set, in this case for a
the majority of the tests. For the longitudinal tensile tests, transverse compression test carried out at a pressure of
the strain gauge used was C930519K, Measurements Group,
UK and the adhesive used was M-Bond. For the transverse
compression tests, and the atmospheric pure epoxy test,
post-yield strain gauge used was YFLA-2 (Tokyo Sokki
Kennyujo Co Ltd, UK), capable of measuring large strains,
and the adhesive used was cyanoacrylate. For the pure
epoxy tests, only the atmospheric pressure test was strain
gauged. No strain gauges were used in the tests carried out
on the 108 off-axis specimens.
In order to get the strain gauge wires out of the vessel, a
method that was developed by Pugh [15] using an epoxy
resin conical metal plug, was employed in the present work.
The diameter of the plug was 5 mm at the thick end, and 1.1
mm at the thin end: the plug was 8 mm long. The layer of
epoxy coating was in the order of 0.5 mm. The epoxy used Fig. 2. State of stresses under three types of loading: (a) longitudinal
was a standard slow curing Araldite, which was post-cured tension, (b) transverse compression and (c) tension of 10 off-axis lamina.
282 P.J. Hine et al. / Composites: Part A 36 (2005) 279–289

Fig. 3. Photograph and a schematic diagram of the high pressure testing facility.

500 MPa on a composite lamina (Fig. 1b), is shown in had to be determined from the machine crosshead displace-
Fig. 4. It is seen that after pressurisation and stabilisation ment and the compliance of the machine at atmospheric
(before a time of w20 s), there is an applied compressive pressure (0.1 MPa). Fig. 5 shows a comparison of the real
strain on the sample due to dimensional changes caused by sample compliance, measured using a strain gauge on the
the imposed hydrostatic pressure: as described above sample, and the measured sample compliance determined
the samples are all free to contract during pressurisation. from the crosshead movement. It is seen that the compliance
Once the crosshead begins to move (at around 25 s), an as measured by the machine is both much larger than
increase in force is seen due to the seal friction. Once the the real compliance and also non-linear, reinforcing the
sample has broken the crosshead is allowed to continue to necessity of using strain gauges where possible. As the
move, so a second measure of the seal friction can be measured sample compliance from crosshead movement is
taken. An average of the seal friction before and after the non-linear, we have made the arbitrary choice to compare
test is then determined and subtracted from the overall
measured force.
The failure stresses (s) presented in the following tables
and graphs were computed from the applied Force (F) and the
cross-sectional area (A) of the specimens, i.e. sZF/A. In
addition to the applied stress, the specimens were subjected
to a hydrostatic pressure component (KP) along the loading
direction. Hence, the net stress acting in the loading direction
will be (sKP) for the case of an applied external tensile stress
(a and b in Fig. 2) and K(sCP) for an applied external
compressive stress (transverse compression, c in Fig. 2 and
pure epoxy compressive tests). The stresses in a direction
perpendicular to the loading direction are those generated by
the pressure and they are compressive (ZKP).

3.1. Compression tests on pure epoxy

In the compression tests on epoxy carried out above Fig. 4. A typical set of experimental data for force, pressure and strain vs
atmospheric pressure, no strain gauges were used to time (1 VZ10 kN for the force signal, Z1.38% for the strain signal and Z
measure the strains in the specimens and hence the strains 1 GPa for the pressure signal).
P.J. Hine et al. / Composites: Part A 36 (2005) 279–289 283

Fig. 5. A comparison of the real and measured sample compliance for a pure
epoxy sample in compression.

the slopes of these two curves between 1500 and 3000 N.


This is a compromise between a reduction in the non-
linearity of the compliance and the necessity to keep below
the onset of permanent deformation in the epoxy sample. In
this region we obtained values of the measured compliance
of 2.336!10K7 m/N, the real compliance (strain gauge)
of 6.568!10K8 m/N, giving a machine compliance of
1.679!10K7 m/N. We now use this value of the machine
compliance to correct the measurements made under
pressure, using the same sample as used above, as further Fig. 6. Epoxy resin properties vs pressure (a) compression modulus,
experiments showed the sample compliance to differ (b) compressive strength.
between samples, due, no doubt to differing contact between
These results are not unexpected as it is well known that
the samples ends and the testing machine, and other such
polymer modulus and strength increase significantly with
effects. Tests were carried out at 100, 200, 300, 400, 500 and
applied hydrostatic pressure as the internal free volume
600 MPa.
is decreased and so molecular mobility is also decreased
The results obtained in this way are shown in Fig. 6a and
[6,16,17].
b for modulus and strength, respectively, and are also
tabulated in Table 2. There is a clear, gradual, increase in
modulus up to a pressure of 400 MPa, after which the 3.2. Lamina tests
modulus increases rapidly. It is likely that as the sample
stiffens, the measured compliance gets closer to the machine Three types of lamina tests were carried out: (a)
compliance, making the errors involved in the difference longitudinal (along the fibres) tension, (b) transverse
increase. Of course we are also making the assumption that (perpendicular to the fibres) compression and (c) 108 off-
the machine compliance is independent of pressure. axis tension. The results from these tests are summarised in
However, with these caveats, we can approximate the
Table 2
dependence of modulus with pressure up to 400 MPa by a The variation of the epoxy resin modulus and strength with pressure
straight line as shown in Fig. 6a. The intercept of the
Pressure Measured epoxy resin Measured epoxy compres-
regression line was fixed at the value at atmospheric
(MPa) compression modulus sive yield strength (MPa)
pressure (0.1 MPa), as this is the most accurate, determined (GPa)
using strain gauges.
0.1 3.27 119G2.8
For the compressive strength, the results showed a 100 4.19 136G2.7
monotonic increase with pressure, with a value at atmos- 200 4.64 151G3.0
pheric pressure (0.1 MPa) of 119G3 MPa and a gradient of 300 4.70 177G3.5
0.2 MPa/MPa, determined from a linear regression analysis. 400 5.9 198G4.0
The samples all failed by yield.
284 P.J. Hine et al. / Composites: Part A 36 (2005) 279–289

Table 3 Table 5
Longitudinal tensile results on UD lamina at various pressure values Experimental data obtained from tensile tests on 108 off-axis UD lamina
coupons at various pressure values
Pressure Tensile mod- Tensile strength Failure strain
(MPa) ulus E11 (GPa) (GPa) (%) Pressure (MPa) Shear stress at
failure, t12 (MPa)
0.1 43.5G1.7 1.26G0.07 3.05G0.09
100 40.6G1.8 1.05G0.05 2.68G0.06 0.1 40.1G3.2
150 40.6G0.6 1.13G0.02 2.84G0.04 150 68.1G1.8
200 43.9G1.5 1.21G0.03 2.79G0.02 300 78G8
250 42.5G1.0 1.24G0.04 2.95G0.04 500 80G5
500 44.5G0.6 1.14G0.01 2.59G0.02
860 0.90G0.07
For the axial strength, the general trend is that the strength
decreases as the pressure increases. This fall in the
Tables 3–5 and shown in Figs. 7–14. The state of stresses in longitudinal strength has been observed by other workers in
the three types of tests is shown schematically in Fig. 2. the field [1,2,9] and by these authors as part of a previous
study on UD carbon fibre/epoxy composites [8]. One possible
3.2.1. Axial tension mechanism for this fall in strength is an increased localisation
Axial tensile tests (1 direction) were carried out at a of damage as the pressure increases, making individual flaws
variety of pressures between atmospheric (0.1 MPa) and more important, although it cannot be discounted that the fibre
860 MPa and a summary of the results from all the tests is strength itself could fall with increasing pressure as we found
shown in Table 3. Although the force/time and strain/time in the previous experiments on carbon fibre/epoxy compo-
curves were often non-linear, the calculated stress–strain sites and individual carbon fibre bundles [8].
curves were all predominantly linear up to failure. Typical examples of failed samples are shown in Fig. 8:
Fig. 7 shows the variation of the axial modulus (E11) and this shows axial splitting at atmospheric pressure and a clean
the axial strength (s1T) with applied hydrostatic pressure. break at all elevated pressures. The top sample shows a
The modulus possibly shows a small increase with pressure, typical failure at atmospheric pressure, while the lower
although there is some scatter, which we partly attribute to sample shows the clean break seen with all the tests at
the variations in the fibre volume fraction described earlier. elevated pressures.
To elucidate this effect, a range of pressure tests could in
3.2.2. Transverse compression tests
principle be carried out using a single specimen in order to
Transverse compression tests were carried out over
reduce sample variability.
a similar range of pressures to the axial tensile tests.
Table 4a Typical stress–strain curves for these tests are shown in
Transverse compression results on UD lamina at various pressure values

Pressure Transverse Transverse compressive Failure


(MPa) compressive strength, s2C (MPa) strain (%)
modulus,
E22 (GPa)
0.1 14.4G2.7 107G15 0.83G0.28
100 14.5G0.7 147G22 1.68G0.03
150 18.5G2.6 184G1.0 2.4G0.1
200 19.9G2.2 226G6.0 2.82G0.42
250 19.4G6 250G3.5 3.75G0.3
500 20.1G4 319G5.0 3.7G0.3
860 417G48

Table 4b
Transverse compressive Young’s modulus results from a single UD lamina
sample taken at 0.5% strain

Pressure (MPa) Transverse compressive


modulus, E22 (GPa)
0.1 13.5
100 17.6
200 16.8
300 18.3
400 20.4
500 21.8
600 23.7
Fig. 7. Longitudinal tensile modulus (E11) and strength (s1T) vs pressure.
P.J. Hine et al. / Composites: Part A 36 (2005) 279–289 285

Fig. 8. Typical examples of failed tensile samples: top atmospheric


pressure, bottom all other pressures.

Fig. 10. Transverse compression stress–strain results for a single sample vs


Fig. 9a, and a typical failed sample in Fig. 9b. A summary
pressure.
of the results from the transverse compression tests is
shown in Table 4. non-linear behaviour at atmospheric pressure and the failure
Fig. 9a shows typical stress–strain curves at various stresses was also higher.
pressure values. The curves show non-linear response and With increasing pressure, initial modulus, failure stress
the degree of non-linearity increases with increasing and failure strain increased considerably. For all of the
pressure. At atmospheric pressure there was a difference specimens, the initial modulus was measured between
in behaviour compared to the previous study on standard strains of 0.1 and 0.4%. Initial modulus values determined
size samples [12]. In the previous work this material showed in this way were found to show significant scatter
(Table 4a), although the trend was clearly a significant
increase in the transverse compression modulus with
increasing hydrostatic pressure.
Most of the samples failed as shown, by shear on a plane
at w408 to the side face, which did not change significantly
with pressure. It was observed that at atmospheric pressure
the specimen failure was generated at the bottom of

Fig. 9. Transverse compression results (a) typical stress–strain curves,


(b) typical failed sample. Fig. 11. Transverse compressive strength (s2C) vs pressure.
286 P.J. Hine et al. / Composites: Part A 36 (2005) 279–289

Fig. 12. Transverse compression modulus (E22) vs pressure.

the samples, while the majority of failures under pressure


were generated from the top corner. Fig. 14. Measured and predicted transverse compressive modulus (E22) vs
pressure.
To improve the modulus measurements, tests at a variety
of pressures were carried out on a single sample taking the Fig. 11 shows the results for the dependence of the
sample to w0.5% strain at each pressure. The results of this transverse compressive failure strength (s2C) with pressure.
experiment are shown in Fig. 10 and Table 4. The scatter in The measured data are shown by the filled squares, and
the modulus values (E22) was now much reduced (Table 4b), indicates a clear increase in the transverse compressive
although the overall increase in modulus with pressure failure stress with increasing pressure, with a transition in
mirrored the results from Fig. 9a. As with the axial tensile behaviour at around 300 MPa.
tests, measurements of modulus could only be made up to For the upper region the slope is w0.27, which is a
500 MPa as above this pressure the epoxy coated plug was typical value for the pressure dependence of yield for a
found to fail. polymer [18]. If the samples were failing according to the
Mohr–Coulomb criterion [19], the failure plane would be
expected to be on a plane at 418 to the side face independent
of pressure, which is in excellent agreement with the
measured value of 408. We would suppose that in the lower
region the samples are failing before reaching a true yield
point due to the presence of flaws, or due to the small size of
the samples. As the pressure is increased the flaws become
less effective in the increasingly compressive environment
and so the samples get closer to the true yield point. This can
be seen in the stress–strain curves shown in Fig. 9a, where
the samples tested at higher pressures show a flat top
characteristic of a yield point. Extrapolating the upper curve
back to the Y axis gives a plausible value for the true
atmospheric transverse compressive strength in the absence
of flaws of 182 MPa.
It is worth noting here that while the final strength and
strains increased with increasing pressure, the secant
modulus at failure was observed to decrease with increasing
pressure (Fig. 12).

3.2.3. 108 off-axis tensile tests


A 108 off-axis tensile test is often used to measure the
interlaminar shear strength of UD fibre composites [20]. It
can easily be shown that if the applied stress along the
Fig. 13. 108 tensile tests: (a) typical sample failure (b) shear strength (t12) samples axis is sx, then the interlaminar shear stress on the
vs pressure. 108 plane, t12, is given by
P.J. Hine et al. / Composites: Part A 36 (2005) 279–289 287

t12 Z 12 sx sin 208 Z 0:171sx (1) Table 7


Predicted Young’s modulus changes with pressure for a UD lamina
Two tests were carried out at each of four pressures, Pressure Predicted axial Predicted transverse Predicted shear
atmospheric pressure, 150, 300 and 500 MPa, and the (MPa) tensile modulus, compression modulus, G12
failure load, and hence the failure stress, was measured E11 (GPa) modulus, E22 (GPa) (GPa)
for each sample. Sample strain was not measured in these 0.1 44.1G2 14.7G1.6 4.11G0.4
tests. A typical failed sample is shown in Fig. 13a, 100 44.4G2 16.9G1.7 4.76G0.4
while the variation of the shear stress with applied 200 44.6G3 19.0G1.9 5.39G0.5
hydrostatic pressure is shown in Fig. 13b. The samples all 300 44.9G3 21.0G2.0 6.00G0.6
400 45.2G3 22.9G2.2 6.58G0.6
failed on the 108 plane, as expected. The shear stress 500 45.4G3 24.6G2.3 7.13G0.6
initially increased rapidly with pressure and then levelled 600 45.7G3 26.3G2.5 7.66G0.6
off at around 300 MPa, mirroring the same transition
as seen in the transverse compression results (Fig. 11).
a modulus of 72.5 GPa and a Poisson’s ratio of 0.2 were
A summary of the 108 off-axis tests results is shown
used, while for the epoxy resin the modulus was as
in Table 5.
measured above in Section 3.1 and a value of Poisson’s
ratio of 0.35 was assumed. An average fibre volume
4. Discussion fraction from the axial tensile and transverse compression
tests of 59G4 was used.
Table 6 presents a summary of the UD lamina properties The results calculated using the theory of Wilczynski
measured at atmospheric pressure. The data are compared are shown in Table 7. The results show that the axial
with those using standard sized samples [7]. The agreement tensile modulus (E11) is predicted to increase only slightly
between these two sets of data is very good, apart from the with pressure, because it is fibre dominated: additionally
shear tests where the volume fraction of the panels was the difference between the upper and lower limits (due to
different. volume fraction variation from specimen to specimen) is
greater than the expected change due to the epoxy resin
4.1. Prediction of the effect of pressure on elastic constants modulus increase with pressure. This confirms the
experimental results for axial tensile modulus shown in
The present results describing the effect of pressure on Fig. 7.
the modulus of the pure resin epoxy are in accordance with The transverse modulus is predicted to increase much
previously documented results reported by Wronski [21]. more significantly with pressure, because it is much more
The current results display a linear increase in properties dependent on matrix properties. Interestingly, the increase
with pressure. in composite modulus is greater than the absolute change in
In order to determine the effects of pressure on the epoxy modulus with increasing pressure. Fig. 14 shows
composite elastic constants, the pressure dependence of a comparison of the measured transverse compression
the pure epoxy resin can be used in conjunction with modulus with that predicted by modelling based on the
suitable micro-mechanics relation. The model adopted epoxy resin pressure dependence. It is seen that there is
here is that developed by Wilczynski [22], which we have excellent agreement between the measured values and the
found in previous work to give very good predictions of predictions, confirming the importance of matrix properties
UD lamina properties. The phase (constituent) properties for the transverse direction.
used for modelling were as follows. For the glass fibres As the effect of pressure on the composite axial and
Table 6 transverse modulus is predicted well by the values for the
6 A comparison of the values measured in this programme with the pure epoxy sample and the micromechanical model, we can
previously published values [12] also predict the effect of pressure on the shear modulus, as
Property Measured Typical data
shown in Table 7. It is assumed that the resin shear modulus
[12], VfZ60% follows the same trend as that of the resin Young’s modulus
VfZ59% VfZ38% and that Poisson’s ratio is pressure independent. The
Longitudinal modulus, 43.5G1.7 45.6
increase of the composite shear modulus, G12, with pressure
E11 (GPa) is predicted as an almost linear increase, and is substantial as
Transverse compressive 14.4G3 16.2 this composite property is also very dependent on the
modulus, E22 (GPa) behaviour of the epoxy resin.
Longitudinal tensile 1.26G0.1 1.28
strength, s1T (GPa)
Transverse compressive 182G5 145 4.2. Strength prediction
strength, s2C (MPa)
Shear strength, t12 40.1G3.2 73
(MPa)
A number of failure criteria have emerged recently from
the WWFE. Some of these criteria [23–35] have the
288 P.J. Hine et al. / Composites: Part A 36 (2005) 279–289

Table 8 compression tests, with a transition in the failure


A list of the theories represented in the world-wide failure exercise [14–28] behaviour at a pressure of around 300 MPa.
Authors Ability to Theory designation † The increase in the transverse modulus with pressure was
tackle triaxial predicted using a micromechanical model and the
stresses measured pressure dependence of the epoxy resin.
Chamis CC, Gotsis PK and Yes Chamis(1), Chamis(2)
Minnetyan L
Hart-Smith LJ Hart-Smith(1)
Hart-Smith LJ Hart-Smith(2)
Eckold GC Eckold
Edge EC Edge
Acknowledgements
McCartney LN Yes McCartney
Puck A and Schürmann H Yes Puck The present work was carried as a part of the UK CRP
Wolfe WE and Butalia TS Wolfe programme, sponsored by MoD. The authors would like to
Sun CT and Tao JX Yes Sun(L), Sun(NL) thank the sponsors for their financial support. The authors
Zinoviev P, Grigoriev SV, Zinoviev
Labedeva OV and Tairova LR
would also like to thank Keith Norris (Leeds University) for
Tsai SW and Liu K-S Yes Tsai his technical assistance in specimen preparation and
Rotem A Rotem instrumentation.
Cuntze R and Freund A Yes Cuntze
Bogetti T, Hoppel C, Harik V, Yes Bogetti
Newill J and Burns B
Mayes SJ and Hansen AC Yes Mayes
References
Huang Z-M Yes Huang
Hart-Smith LJ Hart-Smith(3)
[1] Parry TV, Wronski AS. The effect of hydrostatic pressure on the
tensile properties of pultruded CFRP. J Mater Sci 1985;20:2141–7.
potential to be extended to cater for triaxial state of stresses. [2] Parry TV, Wronski AS. The tensile properties of pultruded GRP tested
Table 8 provides a list of the World Wide Failure Exercise under superposed hydrostatic pressure. J Mater Sci 1986;21:4451–5.
theories and an indication is given where the theory is [3] Parry TV, Wronski AS. The effect of hydrostatic pressure on
transverse strength of glass and carbon fibre-epoxy composites.
capable of handling three-dimensional state of stresses.
J Mater Sci 1990;25:3162–6.
All these theories have been recently bench-marked [4] Sigley RH, Wronski AS, Parry TV. Tensile failure of pultruded glass-
against test data under biaxial loading. Some of these were polyester composites under superimposed hydrostatic pressure.
observed to be mature enough to capture various aspects Compos Sci Technol 1991;41:395–409.
related to biaxial test data, such as predicting mode of [5] Shin ES, Pae KD. Effects of hydrostatic pressure on the torsional shear
behaviour of graphite/epoxy composites. J Compos Mater 1992;26:
failure and final strength.
462–85.
The application of these theories to three-dimensional [6] Hoppel CPR, Bogetti TA, Gillespie JW. Literature-review—effects of
state of stresses has not been carried out so far. This will be hydrostatic-pressure on the mechanical-behavior of composite-
the subject for future considerations as, undoubtedly, this materials. J Thermoplast Compos Mater 1995;8:375–409.
would establish the capability of current failure criteria to [7] Zinoviev PA, Tsvetkov SV. Mechanical properties of unidirectional
organic-fiber-reinforced plastics under hydrostatic pressure. Compos
predict failure under triaxial loading.
Sci Technol 1998;58:31–9.
[8] Hine PJ, Duckett RA, Van Schepdael LJJM, Hamoen JR, Van Den
5. Conclusions Oever MJA, Van Den Berg RW. In Fifth International Conference on
Deformation and Fracture of Composites, IOM Communications Ltd,
London, UK; 1999.
An investigation has been carried out to determine the [9] Zinoviev PA, Tsvetkov SV, Kulish GG, van den Berg RW, Van
effects of hydrostatic pressure on some of the mechanical Schepdael L. The behavior of high-strength unidirectional composites
properties of E-glass/epoxy materials, using an already under tension with superposed hydrostatic pressure. Compos Sci
existing high pressure cell capable of exerting up to Technol 2001;61:1151–61.
[10] Hinton MJ, Kaddour AS, Soden PD. Failure criteria in fibre reinforced
850 MPa. A number of conclusions may be drawn:
polymer composite. Compos Sci Technol J 1998;58 [special issue].
† The longitudinal modulus was observed to increase [11] Hinton MJ, Kaddour AS, Soden PD. Failure criteria in fibre reinforced
polymer composite: Part B: Comparison between theories and
slightly with increasing pressure. experiments. Compos Sci Technol J 2002;62 [special issue].
† The longitudinal strength was found to fall with increas- [12] Soden PD, Hinton MJ, Kaddour AS. Lamina properties, lay-up
ing pressure, with a change in failure mode from axial configurations and loading conditions for a range of fibre-reinforced
splitting to a transverse break at elevated pressures. composite laminates. Compos Sci Technol 1998;58:1011–22.
† The transverse compression modulus and strength were [13] Sweeney J, Duckett RA, Ward IM. The fracture-behavior of oriented
polyethylene at high-pressures. Proc R Soc Lond Ser A, Math Phys
found to increase markedly with pressure with a knee at Eng Sci 1988;420:53–80.
around 300 MPa pressure. [14] Tsui SW, Duckett RA, Ward IM. Influence of molecular-weight and
† The shear strength determined from the 108 off-axis branch content on the fracture-behavior of polyethylene. J Mater Sci
tension tests showed a similar behaviour to the transverse 1992;27:2799–806.
P.J. Hine et al. / Composites: Part A 36 (2005) 279–289 289

[15] Pugh HLD. The mechanical behaviour of materials under pressure. [27] McCartney LN. Predicting ply crack formation and failure in
London: Elsevier; 1970. laminates. Compos Sci Technol 2002;62:1619–32.
[16] Ainbinder SB. Effect of hydrostatic pressure on mechanical properties [28] Puck A, Schürmann A. Failure analysis of FRP laminates by means of
of plastics. Mekh Polim 1965;1:65–75. physically based phenomenological models. Part B. Compos Sci
[17] Spitzig W. Richmond. Polym Eng Sci 1979;19:16. Technol 2002;62:11633–72.
[18] Rabinowitz S, Ward IM, Parry JSC. J Mater Sci 1970;5:29–39. [29] Sun CT, Tao J, Kaddour AS. Prediction of failure envelopes and
[19] Coulomb CA. Mem Math Phys 1773;7:343. stress-strain behavior of composite laminates: comparison with
[20] Chamis CC, Sinclair JH. Ten-deg off-axis test for shear properties in experimental results. Compos Sci Technol 2002;62:1672–82.
fiber composites. Exp Mech 1977;17:339–46. [30] Kuraishi A, Tsai SW, Liu K. A progressive quadratic failure criterion,
[21] Wronski AS, Pick M. Pyramidal yield criteria for epoxides. J Mater Part B. Compos Sci Technol 2002;62:1682–96.
Sci 1977;12:28–34. [31] Zinoviev PA, Lebedeva OV, Tairova LP. Coupled analysis of
[22] Wilczynski AP. A basic theory of reinforcement for unidirectional experimental and theoretical results on the deformation and failure
of laminated composites under a plane state of stress. Compos Sci
fibrous composites. Compos Sci Technol 1990;38:327–30.
Technol 2002;62:11711–24.
[23] Gotsis PK, Chamis CC, Minnetyan L. Application of progressive
[32] Bogetti TA, Hoppel CPR, Harik VM, Newill JF, Burns BP. Predicting
fracture analysis for predicting failure envelopes and stress–strain
the nonlinear response and progressive failure of composite laminates.
behaviors of composite laminates: a comparison with experimental
Compos Sci Technol 2004;64:329–42.
results. Compos Sci Technol 2002;62:1545–60.
[33] Cuntze RG, Freund A. The predictive capability of failure mode
[24] Eckold GC. Failure criteria for use in the design environment. Part B.
concept-based strength criteria for multidirectional laminates. Com-
Compos Sci Technol 2002;62:1561–70. pos Sci Technol 2004;64:343–77.
[25] Edge EC. Theory v. experiment comparison for stress based Grant– [34] Mayes SJ, Hansen AC. Composite laminate failure analysis using
Sanders method. Compos Sci Technol 2002;62:1571–90. multicontinuum theory. Compos Sci Technol 2004;64:379–94.
[26] Hart-Smith LJ. Comparison between theories and test data concerning [35] Huang ZM. A bridging model prediction of the tensile strength of
the strength of various fibre-polymer composites. Compos Sci composite laminates subjected to biaxial loads. Compos Sci Technol
Technol 2002;62:1591–618. 2004;64:395–448.

You might also like