Polymer Testing 12 (1993) 15-29

Problems Associated with the Compression Testing of

Fibre Reinforced Plastic Composites

J. F. H a r p e r , N. A. Miller & S. C. Y a p

IPTME, Loughborough University of Technology, Loughborough, Leics., UK

(Received 10 March 1992; accepted 21 April 1992)

ABSTRACT
Compression testing of thin and slender specimens must take into
consideration the inherent instability of such specimens. Four different
procedures which address this problem of instability (tendency to
buckle) were used to characterise a fibre reinforced plastic (FRP). It is
demonstrated that the procedure adopted influences the results obtained.
Use of the Celanese antibuckling guide recommended in A S T M
standards gave rise to high values of compressive strength whereas
complete edge support of specimens gave rise to high secant modulus
values. Other guides gave results with much lower coefficients of
variation. Possible explanations for these findings are presented and
recommendations for use are made.

1 INTRODUCTION

The increasing use of fibre reinforced plastic (FRP) composities to

replace the more conventional engineering materials requires a greater
understanding of those factors which influence their behaviour in a
given situation. Whilst some aspects, e.g. tensile behaviour, have been
examined by many and are well understood, there are others which
have received relatively little attention and are correspondingly less
well understood. Compressive behaviour is one such case and there is
presently considerable interest in the generation of more data and
understanding in this area. However, compressive data are not as
15
Polymer Testing 0142-9418/93/$06.00 t~) 1993 Elsevier Science Publishers Ltd, England.
Printed in Northern Ireland
16 J. F. Harper, N. A. Miller, S. C. Yap

readily obtained for FRP composites as for other materials. Bearing in

mind the types of application and the physical characteristics of FRP
composites, compression testing of cubes or squat cylinders (as are used
in other materials) would be inappropriate. More relevant mechanical
tests are carried out on thin specimens. This presents few problems
when examining the tensile behaviour but does present a situation
which needs careful consideration by those interested in compressive
behaviour.
1.1 Difficulties in compressive testing
Hofer & Rao ~ give three reasons for the paucity of compressive
strength data. They are the high performance of FRP composites under
compressive load, testing difficulties involving specimen preparation
such as end splitting, buckling of specimens, load alignment problems
and the fact that specially designed test equipment is required.
Lee 2 suggested that three particular points must be considered in the
design of a compression testing system. They are brooming at the end
of the specimen, axial misalignment and specimen buckling.
Brooming is overcome most easily by designing a system in which the
ends of the specimen are gripped. If free-standing specimens between
two fiat platens are used, then the ends of the test specimen must be
perfectly fiat and square. This problem does not arise if the specimens
are suitably gripped in wedge grips.
Axial misalignment is another problem which can lead to an
underestimate of the compressive properties as reported by Curtis &
Morton. 3 Any misalignment has the tendency to induce buckling in the
test specimen which introduces a bending moment causing strain
divergence on either face. This type of situation must be avoided if the
failure is to be of a purely compressive nature. It is therefore crucial
that axial alignment is obtained by suitable system design. Often
packing shims are inserted into the testing machine to ensure that the
platens or grips are accurately located.
Specimen buckling can be overcome by ensuring that the 'column
length' is less than the critical Euler buckling length4 or by supporting
the 'column' using an anti-buckling guide.
There is considerable scatter in the compressive properties reported
by many authors. ~-7 The scatter in compressive data can be caused by a
number of factors including non-uniformity of stress application, poor
specimen design and different failure modes. The various modes that
are suspected of inducing a compressive failure include fibre symmetric
and non-symmetric microbuckling, fibre compressive failure and
delamination.
 Compression testing of fibre reinforced plastic components 17

1.2 Influence of test methods

Currently the interpretation of compression data must take into account

the methods of test and special fixtures designed to restrict or prevent
buckling. 8 Thus compression data may only represent the behaviour of
specimens subjected to particular test procedures and may not be truly
representative of the compression response of the material. Indeed, it is
seldom possible to be certain that the compressive strength is being
determined. 9
At present, there are no generally accepted international standard
test methods for obtaining the ultimate compressive strength of a FRP
composite laminate. Many authors ~t~-14simply used anti-buckling guides
to provide effective restraint against gross buckling. This has led to the
use of a variety of techniques for axial compression measurements.~'7'~5
The Celanese test method (ASTM D3410-87, Procedure m ) , 16 u s e s
split conical collet grips that fit into matching sleeves, which in turn fit
into a cylindrical shell. However, the design is such that specimen
slippage, buckling and wall friction can be problems. 1'17 If the test
specimen thickness is not within the specification for the jig, then the
grips may not seat properly on a cone-to-cone surface contact arrange-
ment. Instead, contact may occur along a pair of lines on opposite sides
at each end of the specimen. This can cause the seats at each end of the
specimen to shift laterally relative to each other, resulting in contact
with the cylindrical shell. This promotes conditions of high frictional
stress which affect the result obtained.
Lee 2 reported that because of the very short gauge length of the
Celanese specimen, a certain amount of Poisson expansion was
restricted by the grips and hence a lower energy mode of failure (e.g.
splitting) was suppressed. Supported by Woolstencroft et al., 15 they
rejected the Celanese test method as the results obtained showed a
combination of low mean failure stress and a high coefficient of
variation. However, Schulte TM obtained satisfactory results using the
Celanese jig for compression testing.
Other compression testing fixtures such as the sandwich beam, the
TEl (Texaco Experiment Inc.) compression fixture, the Royal Aircraft
Establishment (RAE) fixture, the Composite Research Advisory Group
(CRAG) jig, the sandwich stabilised fixture (which incorporated some
of the best features of both the sandwich beam and the TEl fixture),
the Narmco test 303 jig, the Federal test fixture, the IITRI (Illinois
Institute of Technology Research Institute) compression fixture, the
ASTM D695 and D3410 test fixtures and the modified Celanese jig are
discussed in detail by various a u t h o r s , l'12"lSAvAg'2°'22
18 J. F. Harper, N. A. Miller, S. C. Yap

2 EXPERIMENTAL WORK

The material used in this work was Fibredux 914C-833-40 prepreg

which is commercially manufactured by Ciba-Giegy. The
manufacturer's instructions were followed with respect to storage,
temperature, equilibration and exposure to air before use.

2.1 Laminate lay-up

The 914C-833-40 woven fabric prepreg was cut into sheets of the
required size with a sharp pair of scissors. The stacking configuration of
(0/90, + 4 5 / - 4 5 , + 4 5 / - 4 5 , 0/90)2s was symmetrical about the centre
ply. The off-axis fibre orientation in the lay-up was accomplished by the
use of suitable templates to meet the required angle tolerances. A total
of 8 layers were stacked to produce a cured composite laminate with a
nominal thickness of 2 mm.
The uncured laminate was placed in a steel frame of 2 mm thickness
(nominal thickness of the laminate) and then laid up between porous
and non-porous absorbent materials. Resin bleed-out was enabled by
the use of a perforated releasable membrane placed in contact with the
laminate and backed up with an absorbent material. These materials,
laid symmetrically on both sides of the laminate, permitted the escape
of air and volatiles as well as the bleed-out of excess resin.
The compression moulding process was used for all the mouldings as
its use is widely practised, especially for making laboratory specimens.
Each laminate was cured with a breathing or 'bumping of the press',
where the pressure was applied and released momentarily before
moulding under the curing pressure.

2.2 Moulding conditions

Laminates of dimensions 150 x 200mm were moulded under three

different curing pressures of 653 k N / m 2, 1307 k N / m 2 and 1960 k N / m 2,
at a curing temperature of 170 °C for 1 h to produce laminates of three
different qualities. Curing in each case was followed by cooling to
120 °C before releasing the pressure. Each of the laminates was then
postcured at 190 °C for 4 h and conditioned for at least 48 h at room
temperature in a desiccator.

2.3 Mechanical testing

All mechanical testing was carried out at ambient temperature at a test

speed of 1 mm/min. The Celanese jig was loaded between the flat
 Compression testing of fibre reinforced plastic components 19
platens of the test machine. All other test procedures involved location
of the test specimen in the test machine using hydraulically assisted
wedge grips. Each of the four different test fixtures used in this work is
described in detail below.
The first anti-buckling guide, as shown in Fig. 1, enables the edges of
the specimen to be completely supported. This guide, named ABG-1, is

LOAD

L_ 32

"3 8 ~

HEX HEAO
8-TAP M8

^-A SECTION

Fig. 1. Completeedge supported anti-bucklingguide (ABG-1).

20 J. F. Harper, N. A. Miller, S. C. Yap

based on the R A E method 19 with the following modifications:

(1) a window was made on each side of the guide to allow access to
strain gauges placed on the specimen;
(2) the body of the guide was of smaller dimensions to allow the use
of smaller specimens; and
(3) the guide was modified to be freestanding on the bottom grip,
rather than fixed in the grip with the specimen, leaving only an
unsupported region (5 mm) at the top.
The R A E jig allows for 5 mm of unsupported specimen at each end of
the gauge length. These unsupported regions allow longitudinal com-
pression of the specimen without the grip fixture coming into contact
with the guide. Specimen failure is required to be within the guide and
the chances of such behaviour are improved when there is only one
unsupported region.
The second anti-buckling guide, named ABG-2, is a modification of
ABG-1. The design includes an extra large window on each side of the
guide as shown in Fig. 2. The depth of the windows (column length L)
was 15 mm as determined using Euler critical buckling theory for a
column which is fixed at both ends with a safety factor of 2-5.
The design of the third jig, named the C R A G jig and shown in Fig.
3, is also based on the R A E method. 19 The following modifications
were made to suit the configuration of the specimens used in this work:
(1) the body of the jig was of smaller dimensions to allow the use of
smaller specimens; and
(2) the gripped part of the jig was eliminated to allow direct
gripping of the two ends of the specimen protected with
aluminium end tabs.
The fourth fixture, named the Celanese jig, consists of split conical
collet grips that fit into matching sleeves. These in turn fit into a
cylindrical shell as shown in Fig. 4. This design is based on ASTM
D3410-8716 modified to take account of differing specimen thickness.

3 RESULTS AND DISCUSSION

The results of the mechanical tests are listed in Tables 1-3. The effect
of the different jigs on the measured static compressive strength is
shown in Fig. 5. The Celanese jig gave the highest static compressive
strengths. However, it gave lower secant modulus values than the
complete edge supported ABG-1 but similar failure strains as shown in
Figs 6 and 7, respectively.
 Compression testing of fibre reinforced plastic components 21

.
I 32

L 38 I
15. IL15
--It

k____
i

K
0

V
I
I
I
I
1
I

B-B SECTION

SPECIMEN
f

HEX HEA0
i
8-TAP M8

A-A SECTION

Fig. 2. Partial edge supported anti-buckling guide ( A B G - 2 ) .

3.1 Static compressive strength

The fact that the Celanese jig gives significantly higher compressive
strength suggests that either the jig, the specimen design or a
combination of the two have some influence on the compressive
behaviour of the material.
When a specimen is gripped in the Celanese jig, a triaxial stress field
22 J. F. Harper, N. A. Miller, S. C. Yap

LOAD

i -I
I

_2

SPECIMEN

4-TAP M4

LOAD
Fig. 3. CRAG anti-bucklingjig.

is established in the specimen between the grip faces due to the

compressive clamping force of the grip. The influence of this stress field
extends not only into the material in the grip region but also into the
material within the gauge length. When the uniaxial compressive force
is applied to the Celanese jig (and the specimen) during testing,
another triaxial stress field is superimposed on the field already existing
in the specimen. This is accompanied by an increase in the cross-
sectional area of the specimen (Poisson effect). Such increases are
constrained in the regions near the grips. It is clear that the region of
the specimen that is under the influence of only the applied compressive
force is somewhat less than the gauge length of the specimen. 21
The design of the Celanese jig specimen with a gauge length of
 Compression testing of fibre reinforced plastic components 23
LOAD

•° . ; "~

Specimen-.___.__ \
.-,.,

x\
\
,..•o
• • °

\ C0ltet
- - - - - - - - Grip
•°°. \
• .
• . .
o
\
• •. "

• . °-
\
.•° '\
• o.• . \
,..°.

, • °

__.__.___ Eytindricat
••'. °.

• o
Shett
• ,. o

Tapered
Steeve • o •

l
C
,
,

".°•.

•
-o,"
•
.
°

.
°
.

°.
.

°
I ~Locaf
Dowieo
rsn
.°. •

Fig. 4. Celanese anti-buckling jig.

TABLE 1
Compressive Strength Data

Test jig Moulding pressure

653 k N / m 2 1 307 k N / m 2 1 960 k N / m 2

MCS ~ SD b MCS SD MCS SD

Celanese jig 442 35.30 447 24-37 490 44.14

ABG-1 403 12.63 416 9.55 434 15-52
ABG-2 369 23.38 396 16.81 396 20.40
C R A G jig 376 21.55 384 8-56 387 9.91

a MCS = Mean Compressive Strength (MPa)

b SD = Standard Deviation
24 J. F. Harper, N. A. Miller, S. C. Yap

TABLE 2
Secant Modulus at 0.25% Strain

Test jig Moulding pressure

653 k N / m 2 1 307 k N / m 2 1 960 k N / m 2

MCS a SD b MCS SD MCS SD

Celanese jig 45-2 3.55 47.6 3.14 46.4 2.49

ABG-1 47.7 1.60 49.1 0.46 49-8 0-84

"MCS = Mean Compressive Strength (MPa)

b SD = Standard Deviation

TABLE 3
Failure Strain Data

Test jig Moulding pressure

653 k N / m 2 1 307 k N / m 2 1 960 k N / m 2

MCS a SD b MCS SD MCS SD

Celanese jig 0.91 0.027 0.93 0-104 0.90 0.208

ABG-1 0.89 0.027 0.87 0.098 0-93 0.029

a MCS = Mean Compressive Strength (MPa)

b SD = Standard Deviation

500

~ 400

ee CRAG Jig
[] ABG-2
¢/)
[] ABG-I

100

i I

653 1307 1960

CuringPressure(kN/rrrS2)
Fig. 5. Influence of jig design on static compressive strength.
 Compression testing of fibre reinforced plastic components 25

50

II////
48
[ ] CelaneseJig
! ABC_,-I

!"
42

40
653 1307 1960
Curing Pressure (kN/m^2)

Fig. 6. Secant modulus vs curing pressure.

12.5 mm is a compromise between the need to eliminate buckling of the

specimen and to minimise grip effects. It is possible that a large
proportion of the short gauge length is not solely under the influence of
the uniaxially applied compressive force. In contrast, for tests using the
ABG-1, ABG-2 and the C R A G jig, the 110ram long specimen is
subjected to less constraint.
Another source of error is related to the construction of the Celanese
jig. Split collet assemblies are used to grip the specimen. Each grip
assembly is located inside a tapered conical sleeve and there should be
a matching fit between the grip and its sleeve. This is only achieved if
the collet, when holding the specimen, is perfectly circular in cross-
section. Therefore, the thickness of the gripped part of the specimen
(i.e. the end tabs, adhesive and the specimen) must be closely
controlled. Any deviation from the required thickness results in a
mismatch between the collet and its enclosing sleeve.
When the thickness of the specimen (t) is too large, the diameter A is

1.00
0.98
0.96
0.94

j 0.92
r,/2 0.90 I~ Celanese Jig
_=
[] ~G-I
0.88
0.86
0.84
0.82
0.8C
653 1307 1960
Curing Pressure(kNlr:2)

Fig. 7. Failure strain vs curing pressure.

26 J. F. Harper, N. A. Miller, S. C. Yap

~ t grip

Fig. 8. Plan view of celanese jig.

greater than diameter B (Fig. 8). On fitting the sleeve to the grip,
contact between the two occurs at each end of diameter A. This leads
to higher frictional forces than would be present if the diameters A and
B were equal.
Conversely, when the specimen thickness is too small, diameter B
becomes greater, which again leads to increased frictional forces and a
reduction in the gripping force across diameter A as there is a tendency
for the grips to be forced open.
It is apparent that the development of forces within such a system
depends crucially on the specimen thickness. The results of tests carried
out using this particular method showed a greater scatter than those of
any other method. There is a possibility that the small variation in
thickness (+0.05 mm) is responsible. However, the specimen thickness
is only of vital importance because the cross-section of the grip is
circular. It is in fact the inherent design of the jig which dictates the
precise thickness of the specimen. This is, at best, an unnecessary
restriction. If tapered wedge grips were used (as in ASTM D3410-87,
Procedure B) or the British Aerospace modification, then similar
problems would not arise. Clearly the Celanese jig method (ASTM
D3410-87, Procedure A) is less than satisfactory and an alternative
procedure or method should be used wherever possible.
The other three anti-buckling guides, namely ABG-1, ABG-2 and
the CRAG jig, gave similar compressive strength values but all of them
were lower than those obtained with the Celanese jig. However, the
ABG-1 gives higher values than the other two. Although a grip effect
exists in the region of the specimen near the grips, its influence is not
significant because of the long gauge length. When the uniaxial force is
applied to the specimen, the stress field established in the gauge length
causes the specimen to fracture under axial compression.
Both the A B G - 2 and CRAG jigs gave similar compressive strength
 Compression testing of fibre reinforced plastic components 27

values for each laminate tested. The ABG-2 design included two
windows in the middle of the guides. The gauge length of the specimen
within the windows is free from any constraint from the guide. At the
same time, buckling of the specimen is also prevented. The same
principle applied for the C R A G jig which also prevents buckling
without restricting the specimen from axial compression under the
applied force. This indicates that the area of unsupported gauge length
within a guide may not be critical as long as buckling of the specimen
during compression is prevented.
3.2 Secant modulus
The calculated secant moduli 23 of specimens supported by the ABG-1
were higher than those obtained using the Celanese jig but the failure
strain of the ABG-1 was lower. However, there is considerable scatter
in both sets of results. This is because the small windows on the ABG-1
caused constraint on the area of the specimen around the windows. To
determine the secant modulus, the strain on the specimen was
measured using strain gauges on both sides of the specimen and
because gauged areas were constrained in the ABG-1, the strain is
lower in this region than it would be in an unconstrained situation. This
lower measured strain leads to a high value for the secant modulus.
In the case of the Celanese jig, the gauge length of the specimen is
free from any contact with the jig. Therefore, the calculated secant
modulus is more representative of the material in the region of the
strain gauge. In addition, the gauge length of the Celanese jig specimen
is so short that the site of failure is very likely to be located near to the
strain gauges (length of sensitive element of strain gauge is 5 mm). In
contrast, the gauge length of the specimen used for the three other
types of jig is so long that the strain gauges may be far away from the
failure site. This will result in a registered strain somewhat lower than
that at the failure site. This effect was shown in the tests on specimens
prepared under 1960 k N / m 2 curing pressure. Four of the six specimens
used for failure strain measurement in the ABG-1 failed within or near
the gauged area. Therefore, this set of specimens gave high failure
strain values compared with a similar set of specimens tested using the
Celanese jig.

4 CONCLUSIONS
The compressive strength, secant modulus and failure strain results
obtained differed depending upon the type of test jig used. Therefore,
the results must be considered in the light of the test procedure used.
28 J. F. Harper, N. A. Miller, S. C. Yap

The Celanese jig gave high static compressive strength measurements

due to the high degree of support offered by the grips and possible
friction effects.
The complete-edge supported guide, ABG-1 provides lateral con-
straint on the specimen during testing. This constraint on the specimen,
particularly in the area around the window, gave rise to higher secant
modulus values than for the Celanese jig although the elimination of
grip effects led to a lower failure stress.
The partial-edge supported guide, ABG-2, and the C R A G jig gave
similar strength values with low coefficients of variation because there
was less constraint on the specimen during testing.
Considerable care needs to be taken when interpreting results
recorded when using the Celanese jig (ASTM D3410-87, Procedure A).
Consideration should be given to excluding this method from the
standards or including the modified form of the jig.

REFERENCES

1. Hofer, K. E. & Rao, P. N., A new static compression fixture for advanced
composite materials. J. of Testing and Evaluations, 5 (4) (1977) 278-83.
2. Lee, R. J., Compressive strength of aligned fibre reinforced thermoplastics
laminates. Composites, 15 (1) (1987) 35-9.
3. Curtis, P. T. & Morton, J., The effect of fibre surface treatment on the
compressive strength of the CFRP laminates. RAE Technical Report
82047, April 1982.
4. Timoshenko, S. P. & Gere, J. M., Mechanics of Materials. Van Nostrand
Reinhold, 1972, pp. 323-43.
5. Jeffery, M. R., Sourour, J. A. & Schultz, J. M., Fatigue behaviour of
thermosetting polyester matrix sheet moulding compounds. Polymer
Composites, 3 (1) (1982) 18-28.
6. Purslow, D. & Childs, R., Autoclave moulding of carbon fibre-reinforced
epoxies, Composites 17 (2) (1986) 127-36.
7. Sinclair, J. H. & Chamis, C. C., Compressive behaviour of unidirectional
fibrous composites. In Compressive Testing of Homogeneous Materials and
Composites, ed. R. Chait & P. Papirno. ASTM STP 808, 1983, pp.
153-74.
8. Highsmith, A. L., Stinchcomb, W. W. & Reifsnider, K. L., Effect of
fatigue-induced defects on the residual response of composite laminates.
In Effect of Defects in Composite Materials. ASTM Symposium Proceed-
ings, San Francisco, 1982, pp. 194-216.
9. Harris, B., The strength of fibre composites. Composites, 3 (4) (1972)
152-67.
10. Curtis, P. T. & Moore, B. B., A comparison of the fatigue performance of
woven and non-woven CFRP laminates in reverse axial loading.
International Journal of Fatigue, 9 (2) (1987) 67-78.
 Compression testing of fibre reinforcedplastic components 29
11. Gerharz, J. J., Mechanisms of fatigue damage and fatigue testing. In
Practical Consideration of Design, Fabrication and Tests for Composite
Materials. AGARD-LS-124, Sept 1982, pp. 7.1-7.20.
12. Kan, H. P. & Ratwani, M. M., Compression fatigue behaviour of fibre
composites. SAMPE Quarterly, 11 (4) (1980) pp. 10-14.
13. Morton, J., Damage characteristics in notched carbon fibre composites
subjected to fatigue loading-environmental effects. J. of Composite
Materials, 22 (1988) 657-73.
14. Ryder, J. T. & Walker, E. K., Effect of compression on fatigue properties
of a quasi-isotropic graphite/epoxy composite. In Fatigue of Elementary
Composite Materials, ed. K. L. Reifsnider & K. N. Lauraitus. ASTM STP
636, 1977, pp. 3-26.
15. Woolstencroft, D. H., Curtis, A. R. & Harescheugh, R. J., A comparison
of test techniques used for the evaluation of the unidirectional compressive
strength of CFRP. Composites, 12 (4) (1981) 275-80.
16. ASTM, Compressive properties of unidirectional or crossply fibre resin
composites. ASTM D3410-87, 1987.
17. Adsit, N. R., Compression testing of graphite/epoxy. In Compression
Testing of Homogeneous Materials and Composites. ed. R. Chait & R.
Papirno. ASTM STP 808, 1983, pp. 175-86.
18. Schulte, K., Mechanical Properties of Polymer Composites. 3rd European
Conference on Composite Materials (France), March 1989.
19. Curtis, P. T., CRAG test methods for the measurement of the engineering
properties of fibre reinforced plastics. RAE Technical Report 88012, Feb
1988.
20. Shockey, P. P. & Waddoups, M. E., Strength and modulus determination
of composite materials within sandwich beams. Report FZM 4691, Sept
1966.
21. Yap, S. C., The effect of moulding conditions on the static and dynamic
compressive strength of prepreg moulded materials. PhD thesis, Lough-
borough University of Technology, June 1991.
22. Harper, J. F., Problems encountered in testing fibre reinforced polymer
composites. Cellular Polymers, 9 (6) (1990) 462-73.
23. Harper, J. F. & Heumann, T. O., The strain dependence of elastic
modulus in unidirectional composites. In Composites Evaluation, ed. J.
Herriot. Butterworth Scientific Ltd., 1987, pp. 189-193.