MECHANICAL TESTS 34

Yu.M. Tarnopol’skii and VI L. Kulakov

34.1 STRUCTURAL HIERARCHY OF FIBROUS The test results are used to calculate the
COMPOSITES properties of materials with more complex
configurations of fiber arrangement and of
Fibrous composites are inhomogeneous mate-
hybrids, i.e. materials with different reinforce-
rials with multiple levels of structural scale.
ments in the same lay-up.
The three levels of structural scale can be
For multilayered composites, the largest
arranged in a hierarchy. The characteristic
structural scale is the thickness of the lami-
dimensions for the three levels are: fiber diam-
nated plate that is equal to the sum of stacked
eter, lamina thickness and plate thickness. The
laminae and interleaves. The theory of lami-
most appropriate test methods and structural
nated plates allows one to determine the
analysis techniques are different for each level
properties of the plate using the properties of
in the hierarchy. Test objectives and associated
the monolayers and their stacking sequence. If
problems are also different for each level.
the monolayers are part of a structural ele-
The smallest scale is the diameter of the
ment, then the highest level of scale is the
reinforcing fiber. The properties of the rein-
characteristic size of the object. The dimen-
forcing fiber and polymer matrix and their
sions of structural elements typically exceed
interaction are studied in the field of microme-
chanics. the thickness of the plate by several times. The
The second level scale is the thickness of the properties of components are measured by tra-
unidirectional lamina. Macromechanics ditional mechanical and physical test
describes the properties of a monolayer under methods’.
loading at an angle to the fiber direction. A The results of the analyses or tests at the
monolayer is defined as a flat or curved ele- first level are used as input for the analyses at
ment of material composed of a polymeric the second level. The same relationship holds
matrix and reinforcement of the same type and for the second and third levels. Upon convert-
orientation throughout the layer. It is the basic ing each scale to a continuum at the next
structural element of laminated and fibrous higher scale, it is important that for each level
composites. The characterization of monolay- under consideration the number of elements
ers by mechanical test methods is given be sufficient, e.g. through the width and thick-
particular emphasis in this chapter since test- ness of the lamina or through the thickness of
ing of anisotropic materials is a relatively the plate, so the transition from a discrete
novel and seldom studied field of mechanics. medium to a continuum is performed without
great error. This progression up the hierarchy
eventually leads to solutions of real life prob-
lems involving structural elements or
prototypes.
Handbook of Composites. Edited by S.T. Peters. Published
in 1998 by Chapman & Hall, London. ISBN 0 412 54020 7
 Problems of composite testing 779

34.2 PROBLEMS OF COMPOSITE TESTING systems of coordinates are introduced: the

axes of elastic symmetry in the material (1, 2,
34.2.1 HISTORY 3) and axes of loading (x, y, z for flat speci-
The first reference to mechanical testing of a mens; 8, z and Y for ring and tubular
structural material for an engineering applica- specimens). It is preferable to use methods in
tion is dated July 4,1662. The objective of the which the x, y, z axes (or 8, z, Y ) coincide with
tests was to compare the tensile strengths of the 1 , 2 , 3 axes.
cords made of Riga and Dutch yarns. The The majority of laminated and fibrous com-
stronger in this contest was the thinner cord posites exhibit low interlaminar shear and
manufactured in Riga. In the years that fol- transverse tension strengths. Shear strength is
lowed, equipment and methods for testing characterized by the relations between E x / G x Z
engineering materials, particularly metals, (shear stiffness) and a;/zx; (shear strength).
have reached a high degree of perfection and Transverse tension and compression strengths
consistency. The appearance of composites perpendicular to the fibers are determined by
the relations E x / E Z , axt”/a;, where Ex
ax’u/az’u,
and their ever-expanding use has once again
made it necessary to improve mechanical test and EZ are the moduli of elasticity in the x and
methods. Although significant progress has z directions; GxZ is interlaminar shear modulus;
been made, there are vast differences in their
a,“ and a; are strengths in the x and z direc-
tions; zxzuis shear strength in xz plane. The x
maturity. The methods differ primarily in the
degree to which they minimize extraneous and y axes are located in the fiber lay-up (rein-
stresses and strains. Although test methods forcement) plane, the z-axis is perpendicular
tend to become more complex as their accu- to this plane; the (t) and (c) designate tension
and compression, respectively.
racy increases, economics must be considered
in their selection. Factors such as complexity
of specimen preparation, amount of material 34.2.3 UNIQUE REQUIREMENTS OF
required, and the requirements for specially COMPOSITE TESTING
designed equipment must be considered.
The anisotropy and unique structural proper-
ties of composite materials cause serious
34.2.2 DETERMINABLE VALUES difficulties. For example, a large number of
strength and elastic properties must be deter-
The purpose of mechanical tests is to deter- mined for complete characterization of the
mine the strength and elastic properties of a material. Since the number of determinable
material. However, only loads, displacements characteristics depends on the state of stress
and strains can be measured in a mechanical and the degree of anisotropy23, one should
test. The theory of elasticity for an anisotropic select the loading methods for which the
body is used to determine the desired proper- experimentally determinable values are most
ties of composites from these measurable simply related to the material characteristics.
quantities. It should be remembered that The selection of techniques for analyzing the
advanced fibrous composites with unidirec- data is critical as well as the determination of
tional, laminated or spatial fiber lay-ups are their range of validity. Since the composite
inhomogeneous, essentially anisotropic mate- analysis techniques are based on the theory of
rials. The customary terms, i.e. tension, elasticity for an anisotropic body, it is neces-
compression, shear and bending, are meaning- sary to consider the error in treating an
less without specification of the direction of inhomogeneous anisotropic medium as a con-
the load and its relationship to the axes of elas- tinuous anisotropic medium. For example, the
tic symmetry of the material. Therefore, two number of structural elements (fibers, lamina,
780 Mechanical tests

etc.) must be sufficiently large to support this of fiber waviness can cause the measured val-
approximation43r6. ues of longitudinal modulus of elasticity and
Once the general test method has been strength to be considerably lower than those of
selected, the details of the loading and the materials with ideally straight fibers. Fiber
sample geometry must be selected. For fibrous waviness also influences on the coefficient of
composites, the principal difficulties lie in the thermal expansion in the fiber direction. The
generation of a uniform stress field in a repre- modulus of elasticity perpendicular to the
sentative volume of material, i.e. the fiber direction and the in-plane shear modulus
elimination of end and edge effects. This is dif- are not significantly affected by fiber wavi-
ficult even for the most simple types of tests. ness.
The difficulties increase with increasing All of the aforementioned unique testing
degree of anisotropy, i.e. materials reinforced requirements apply to composites of a fibrous
by high-modulus or high-strength fibers and laminated structure. Additional difficul-
(boron, carbon and organic fibers). ties arise when spatially reinforced composites
End effects are primarily influenced by the are tested because the transverse strength and
method of fastening and loading of the speci- stiffness are derived from a rigid framework
men, the length of the grip section, and the rather than from a compliant matrix.
fiber orientation. The region involved in end
effects extends in the direction of the greatest
34.2.4 SUMMARY TABLES
stiffness of the material and increases with the
anisotropy of the material. The most common methods of testing com-
Edge effects are primarily influenced by the posites in tension, compression, torsion and
size and shape of the specimen, the fiber ori- bending are described in Tables 34.1-34.5 . The
entation and the angle of specimen cutting. If high performance test fixtures designed
strength anisotropy is present, improper load- specifically for composite testing, their
ing and fastening can lead to changes in the description and recommended applications
failure mode and the resulting strength value. are given in Reference 7.
A most important considerationis the selec-
tion of the specimen width. The width must be
34.3 TEST SPECIMENS
large enough to avoid the effect of cut fibers at
free edges which is important for specimens of The important relationships between fabrica-
off -angle, angle-ply and cross-ply materials. tion methods, test methods and required
Edge effects are manifested as interlaminar specimen shapes are shown in Fig. 34.1.
stresses at the free edges of the specimen, the Specimens for mechanical testing are classified
direction and magnitudes of which depend on as flat specimens (bars and plates), rings (com-
the fiber lay-up. plete and segments) and tube@.
Material quality also causes unique require- The specimens and test methods in Fig. 34.1
ments for testing composites. Quality cannot are used to characterize the monolayer. Flat
be ignored during testing because the material monolayers can be characterized with speci-
and structure are formed simultaneously. In mens that have a different fiber lay-ups but the
addition, composites are extremely sensitive same general, flat, long, narrow shape. To ade-
to mechanical and thermal history. Structural quately characterize wound monolayers, it is
imperfections, in particular porosity, waviness necessary to use both rings and tubes. Ring
and misalignment of fibers, require special specimens of a unidirectional fiber lay-up are
attention. The presence of porosity affects the used to assess characteristics in the fiber direc-
measurement of polymer matrix dominated tion. Tubular specimens with a 90" wind angle
properties, e.g. shear strength. Small amounts are used to measure properties perpendicular
 Test specimens 781
x
Ir
2h"
.. I1
782 Mechanical tests
h
w
W
h
9
t l
 Test specimens
..
%
J . 4
784 Mechanical tests
0
0
m
2
&
m
d'
0
4
r3
a-
d
2
v)
.r(
X
T
N
Y
3
0
%
v
.3
aJ
8
t
cn
m
U
Y3
.
m
h
.r(
aJ
Y
U
2
d
8
aJ
Y
6
 Test specimens 785
I m u
c\
II
%
bn
@
786 Mechanical tests

t
Fig. 34.1 Methods of material fabrication and respective specimens for testing.

to the fiber direction. Tubular specimens with methods, loading types, and failure modes are
different balanced fiber lay-ups (fiber angles all different for the two test methods (Table
are symmetric with respect to the longitudinal 34.1).
specimen axis) are employed to assess shear A uniaxial tension or compression test
characteristics and to study complex states of specimen has several functionally different
stress. parts: two loading sections, a gage section,
The specimen shape, to a great extent, and two transition sections. The loading sec-
tions provide a means of fastening the
depends on the objective of the test: verification
of scientific hypotheses, engineering specifica- specimen in the testing machine. They receive
tion of the material, or quality control of the and transmit the external loads to the gage
materials. The most rigorous requirements are section of the specimen. In the gage section,
imposed on specimens of the second group. deformations are measured and stresses are
calculated according to the geometrical
dimensions and external load. The transition
34.4 TENSION AND COMPRESSION
sections serve to attenuate stress-strain per-
turbations in the loading section to isolate
34.4.1 TENSION TESTING OF FLAT SPECIMENS
them from the gage section. The specimen
In spite of any analogy that may be drawn dimensions that are recommended in the
between loading of flat composite specimens existing standards meet these requirements.
in tension and compression, only the calcula- The specimen dimensions (length, width and
tion relationships (taking into account the sign thickness) specified in standards as a function
of the stresses and strains) are common to both of the type of fiber lay-up are shown in Table
cases. The specimen shape and size, gripping 34.1 (Methods (a) and (b)).
 Tension and compression 787

The greatest technical challenge in tension forces acting on the specimen ends (Table 34.1,
testing of composites, especially unidirec- Method (c)),it is impossible to achieve a suffi-
tional composites, is the reliable transmission ciently uniform stress distribution over the
of tensile forces from the grips to the speci- faces of the specimen. As a result, premature
men. This is generally performed through the local failure of the specimen occurs. When the
use of friction forces. Tabs bonded to the spec- loading is achieved by shear forces acting on
imen improve the efficiency of load the sides of the specimen as specified by
transmission considerably. The tabs should be ASTM D3410 (Method (d)), the stress distrib-
made of a material that has a much lower ution in the specimen is also non-uniform,
modulus of elasticity and a higher total elon- especially when flat wedge grips are used.
gation than the respective characteristics of the The best method is a combination of the two
specimen material. Tabs have been made of methods in which normal forces are applied
fiberglass reinforced composites, aluminum at the ends and shear forces are applied along
and wood veneers. The thickness of tabs the faces of the specimen grip section
should be between 1.5t and 4t, where t is the (Method (e)).
specimen thickness. The tabs must have a In compression testing of unidirectional
large enough area that the ultimate shear load composites in the fiber direction, three basic
capacity of the bond between the tabs and the modes of failure are observed: buckling of the
specimen is greater than the breaking load of reinforcing fibers, transverse cracking of the
the specimen gage section. matrix, and shearing of reinforcing fibers at a
The mode of failure in tension depends on 45" angle without local buckling of the rein-
the relationship between the external load forcement. Transverse cracking is caused by
and the reinforcing fibers and on the type of differences in the Poisson's ratios of the mate-
reinforcement lay-up. When unidirectional rial components and by a non-uniform
composites are loaded in the reinforcement transverse strain distribution along the speci-
direction, they fail by breakage of the rein- men length. Materials reinforced at an angle to
forcing fibers. This is accompanied by the specimen's longitudinal axis fail in shear
transverse cracks, longitudinal shear cracks without crumpling at the end faces because all
and delamination of the polymer matrix. of the shear load is taken up by the matrix. The
Increasing the angle between the load and the aforementioned basic modes of failure can be
reinforcing fibers causes the mode of failure accompanied by a series of other phenomena:
to change gradually from shear and splitting inelastic and non-linear deformation of the
of the polymer matrix parallel to the fiber reinforcing fibers and matrix, delamination,
direction to pure transverse tensile cracking surface peeling, overall buckling and crushing
of the polymer matrix. The failure mode of of the end faces. Failures with different combi-
composites with a balanced angle-ply rein- nations of these phenomena can make the
forcement depends on the angle of the fiber determination of the failure mode very diffi-
lay-up. cult.
In compression testing, great care must be
taken to ensure stability of the specimen, espe-
34.4.2 COMPRESSION TESTING OF FLAT
cially in the gage section. Buckling of the
SPECIMENS
specimen side face is not always detectable
The main problem in compression testing of and will cause erroneous strain measure-
flat specimens is the selection of a loading ments. Special test fixtures are used to prevent
method that ensures compressive failure. overall buckling of the specimen.
When the loading is achieved by normal
788 Mechanical tests

34.4.3 TENSION TESTING OF RINGS The stress concentration problem is elimi-

The most popular means of tension testing nated in tests that use uniform internal
rings uses a half-disk loading device (Table pressure generated by the use of a compliant
34.2, Methods (a) and (b)). This is because the ring (Method (d)) or by a hydraulic system
test is easy to perform and the data is easy to (Method (e)). The disadvantages of the com-
analyze. However, it has several significant pliant ring test method are the need for
disadvantages: the strain distribution over the periodic calibration of the loading element
specimen circumference is non-uniform, fric- and the need for very careful preparation of
tion between the specimen and the half-disks the specimen surface. The disadvantage of the
has a strong effect on the results, and there is a hydraulic test technique is the need for expen-
stress concentration in the specimen at the gap sive hydraulic equipment.
between the two half-disks because of a varia-
tion in the radius of curvature of the specimen.
In thin-walled rings, the stress concentration 34.4.4 COMPRESSION TESTING OF RINGS
takes the form of a slight increase in radial ten- Radial compression of rings is accomplished by
sile stresses that causes a minor effect on the external pressure (Table 34.2).The analog of the
test results. However, in thicker specimens, half-disk tension test is the simplest compres-
the resultant interlaminar shear stresses can be sion test method (Method (c)). The primary
high enough to cause failure at lower loads difference is that in the compression test, it is
than failure due to the circumferential stresses possible to reduce the stress concentration in
alone. The error increases with increases in the the specimen at the split line. The best results
relative specimen thickness t / R , the degree of are obtained with a semi-circular housing that
anisotropy, and the ultimate strain of the mate- has a locking arrangement that prevents radial
rial. Since this test method yields erroneously growth at the split-line (Fig. 34.2). There are
low strength values, it can only be used for also compression analogs for the compliant
qualitative comparison of composites. ring (Method (f)) and the hydraulic system
Corrections have been suggested but have not (Method (g)).In the compliant ring method, the
been used in practice. compliant ring is the elastic foundation of the

Fig. 34.2 Typical interlocking features for ring compression fixture.

 Shear 789

specimen and to a certain extent it prevents gage section where the state of stress is the
buckling of the specimen. The external pressure most uniform. However, edge effects have
may also be applied by mechanical devices considerable influence on the shear strength.
such as multiple cam. Therefore, it is better to bond the specimen to
The primary difficulty in compression test- the rail links than to use mechanical fasteners.
ing of rings by external pressure is the The stress distribution is not affected by the
selection of a relative thickness, t / R , to reduce loading direction, i.e. along the diagonal or
secondary loading effects. Depending on the parallel to the sides of specimen gage section.
relative thickness of specimen, t / R , and the Measuring the shear properties by tensile
degree of anisotropy of the material E,/G,, loading of an anisotropic strip is distinguished
three different failure modes have been by its apparent simplicity (Table 34.1, Method
observed. Thin-walled rings fail by buckling, (f)). The strip can have one of several different
thick-walled rings fail by biaxial compression, fiber lay-ups. This method is not used to deter-
and optimum thickness rings fail by circum- mine in-plane shear strength because it yields
ferential compression. Analysis of the test data low values. A state of pure shear is not assured
must consider the radial as well as circumfer- even with a +45" lay-up.
ential stresses. A similar test method involves tensile load-
Delamination of the inner layers of the ring ing of a strip of a unidirectional material cut at
makes it difficult to correctly calculate the an angle, 8, to the reinforcing fibers. The opti-
compressive strength. This delamination is mum angle is the one for which the relative
often noisy. The occurrence of this failure shear strain y 1 2 / ~is
x maximized and the shear
mechanism also depends on the relative thick- stress rI2reaches its critical value. This angle
ness of the ring. Delamination of helical depends on the anisotropy of the elastic and
windings can lead to unwinding of the speci- strength properties of the material tested. For
men. Hoop wound rings can fail by advanced composites, the optimum angle is 10
layer-by-layer delamination. to 15". Because the stress ratios are very sensi-
tive to changes in the angle, rigid tolerances,
4 degree, are set on the specimen cut angle,
34.5 SHEAR
the strain gage angle, and the direction of
loading. In order to ensure that the stress state
34.5.1 IN-PLANE SHEAR
is uniform, relatively narrow strips, L/w = 14
Shear properties, especially shear strengths, to 16, are used.
are difficult to measure. The simple and eco- The in-plane shear modulus is often mea-
nomical rail shear test is often used for this sured by twisting a square plate with four
purpose (Table 34.3, Methods (a) and (b)).The point loading (Method (c)). The wide accep-
extent of edge effects and the uniformity of the tance of this method may be attributed to the
shear stress distribution over the specimen simplicity of its calculations. However, the
width depends on the length-to-width ratio of experiments should be performed with utmost
the specimen gage section L/w and on the care. This method is only applicable for small
relation of elastic constants G,/E of the mate- deflections, wp < O.lt, on plates made of mate-
rial. Edge effects are negligible tor L/w > 10. rials which are uniform in thickness and
Edge effects cannot be eliminated for materials orthotropic along the specimen axes. Test
with v, = vyxG -1. Thus, this method cannot be results for several different materials have
used for such materials. The elastic constants shown that the P-wp relationship remains lin-
obtained by the rail shear test are less sensitive ear up to wp/t z 1. However, in practice, the
to the relative ratio, L/w, since the measure- deflection, w should be limited to 0.5t to pre-
P'
ments are taken in the center of the specimen vent instability. Only the initial linear section
790 Mechanical tests

of the P-wp curve should be used to determine 34.6 TORSION

the shear modulus. The optimum range of rel- Torsional loading of thin-walled tubes is a
ative plate thickness, L / t , is determined by standard test for measuring in-plane shear
two conditions: the contribution of transverse modulus and strength (Table 34.4, Method (c)).
shear to the deflection at small values of L / t In this test, the stresses are distributed uni-
and the possible loss of stability at large values formly around the circumference and along
of L / t . The limits of L / t are given for BFRP in the length of the specimen. The shear strains
Table 34.3. However, tests run on GFRP (glass are practically constant through the thickness
fiber reinforced plastic), CFRP (carbon fiber of the specimen wall. In torsion, the definition
reinforced plastic) and BFRP (boron fiber rein- of 'a thin-walled tube' is a function of the
forced plastic) with different fiber lay-ups have degree of material anisotropy EJE,, which can
shown that reliable data can be obtained at L / t vary over a wide range. The disadvantages of
> 15. The specimen must be flat and of constant this method are that it requires relatively large
thickness because the calculated shear modu- specimens, special test fixtures, inserts to pre-
lus is related to t3. The distance from the point vent buckling of some specimens and wound
of support or load application to the corners of specimens or specimens of special configura-
the plate should not exceed 2t. Experimental tions, e.g. materials in which the fiber lay-up is
evaluation of these three shear test methods parallel to the specimen axis.
has shown that they all yield comparable val- The results obtained by torsional shear tests
ues of in-plane shear modulus9. compare favorably with results obtained by
The successful application of the double V- test methods using flat specimens.
notch or Iosipescu shear test (Method (d)) to Torsional loading of split rings is also used
all types of fiber lay-ups is well known'O. It is to measure shear moduli (Method (b)). If the
invaluable for testing spatially reinforced com-
sample size limitations indicated in Table 34.4
posites since these materials do not possess are followed, bending effects are negligible.
planes of low shear strength. All other shear
strength test methods use this characteristic of
laminated composites to induce shear failure 34.6.1 INTERLAMINAR SHEAR
first and therefore are useless for spatially
Good estimates of interlaminar shear proper-
reinforced materials. In-depth investigations ties, especially for spatially reinforced
have shown that the Iosipescu method and its
materials, have been obtained by torsion test-
modifications yield good results in shear tests
ing of rods with a circumferential notch
of carbon-carbon composites reinforced along
(Method (a)).The specimens can be tested with
three mutually perpendicular directions (3-D)
or without a central bore. The important geo-
and four principal diagonals of cube of 4-D".
metric parameters of the notch are the relative
Spatially reinforced composites are less sensi-
width, Lp/d, diameter, d, and wall thickness, t.
tive to the dimensions of the notches and gage
It has been shown that within a range of L / d
length than laminated composites. The distrib-
= 0.2-1.0, the length of the notch does not afPect
ution of shear stresses is essentially uniform
the measured shear strength, tnu.The gage sec-
throughout the gage section of 3-D and 4-D
tion diameter can be increased from 5 mm to 15
materials cut at a 90" angle with a total notch
mm (0.6 in) without affecting rnu. However,
depth equal to a half of the specimen thick- increasing the diameter beyond 15 mm (0.6 in)
ness. Moreover, it is possible to prevent stress causes a sharp drop in the measured strength.
concentrations at the notch tips. Specimens
with extra side notches yield the best shear
strength data.
 Bending 791

34.7 BENDING for measuring interlaminar shear strength.

However, refined analysis has shown that the
34.7.1 THREE-POINT BENDING state of stress in a short bar of anisotropic
The most popular type of bend test is the material is significantly different from the
three-point bend test (Table 34.5, Methods (a) state of stress predicted by isotropic theory2J2.
and (b)).The four-point and five-point bend The shear stresses through the thickness of a
tests are less popular in spite of their consider- relatively short anisotropic bar have a para-
able technical advantages over the three-point bolic distribution only in the middle of the
bend test. span. At the loading points, the distribution of
Theoretically, the moduli of elasticity of shear stresses through the thickness of the
homogeneous materials in tension, compres- specimen has peaks near the surface directly
sion, and bending are the same, i.e. Exf = E; = beneath the loading points. In relatively short
E:. However, due to imperfections, the state of anisotropic bars, there are no planar regions of
stress in bending and differences in the fiber constant maximum shear stress. Moreover, on
lay-up through the thickness of the material, relatively short bars (L/w 4 5), the compres-
the bending modulus of elasticity E: can differ sive transverse stresses from the load
somewhat from Exf or E;. This difference is application points can extend over the entire
emphasized by the superscript ’b’. length of the specimen and can exceed mean
The formulae used to determine the elastic shear stress by up to a factor of 15. These com-
bending constants, E: and GxZb, from the deflec- pressive stresses constrain crack opening at
tion of a bar at its midspan must take into sample delamination and result in an apparent
consideration the effect of interlaminar shear. increase of interlaminar shear strength. As a
The effect of interlaminar shear can be result of these deviations from the ideal para-
neglected for large values of relative span bolic stress distribution, the experimentally
length L / t . For highly anisotropic materials, the determined interlaminar strength appears to
relative span length must be greater than 40. decrease with increasing relative span width.
When determining bending strengths, fail- Therefore, shear testing of relatively short bars
ures limited by the normal strength, ., ””,
and can provide only a qualitative comparison of
failures limited by the shear strength, zxF, different composites.
must be separated. Unlike those of isotropic Interlaminar shear strength can also be
materials, the two strengths for composite measured by three-point bending of curved
materials can differ by an order of magnitude. segments. The shear strength is calculated
The shear stresses can have a considerable with the same equations used for bending of
effect on the failure. prismatic bars. However, the additional inter-
laminar normal stresses unique to curved
Failure due to normal stresses occurs by
fracture of the extreme outer layers in com- beams must be taken into consideration. The
pression or tension. Failure due to shear normal stresses act over the entire length of
stresses occurs by delamination approximately the specimen. The sign of the stress depends
at the midplane of the specimen. Laminated on the orientation of the specimen. In the case
materials can fail by a violent debonding of the of segments loaded with their convexity
compressed outer layer. Very short bars experi- upwards (center load applied to the outer
ence a third failure mode. They fail by diameter of the specimen) the stresses are ten-
crumpling and shearing which is accompanied sile, 0. When the convexity is downwards, the
:

by an apparent increase in shear strength. radial stresses are compressive, a;. In the for-
Three-point bend testing of short bars or mer case, shear and tensile radial stresses
ring segments is the most widely used method combine to decrease the apparent shear
strength. In the latter case, the compressive
792 Mechanical tests

radial stresses impede the growth of the shear and 11),and the Arcan Test (mixed Modes I and
delamination crack and raise the apparent 11). The geometry of the specimens, prepara-
interlaminar shear strength. For accurate tion methods, and analysis procedures have
determination of the interlaminar shear been described in detai113J4.
strength, r&bu,segment dimensions must be Advanced composites, especially those
selected so that the normal circumferential reinforced with carbon and aramid fibers,
stresses, as, and normal radial stresses, or,are have highly anisotropic thermophysical prop-
negligible compared to the shear stresses. erties. This is reflected in the thermomechanical
behavior of the structures fabricated with these
materials. The combination of a polymer
34.7.2 BENDING OF RINGS
matrix having a high coefficient of thermal
Bending of complete rings by diametrically expansion and fibers having a negative coeffi-
opposed loads (Method (c)) is used to deter- cient of thermal expansion allows the
mine elastic and strength properties of fabrication of composites with extremely low
composites. Reliable results are obtained if the thermal expansion. This property of compos-
relative specimen thickness, t l r , is properly ites is used in the fabrication of structures
selected. The acceptable range of relative which are stable over wide temperature
thickness for determining the shear modulus, ranges. Low thermal expansion is commonly
Go,: is based on the material anisotropy, achieved in two directions and the process can
E,b/G,b. The shear modulus is calculated from be extended to materials which are spatially
the load-diametral deflection data using the reinforced in three directions (3-D) or along
same equations for three point bending of four diagonals of a cube (4-D)15.These com-
prismatic bars with a correction factor for the posites possess a thermal expansion coefficient
fraction of deflection induced by shear which is both isotropic and very low.
stresses. When the test is used to determine Measurement of the thermal expansion
the interlaminar shear strength, rOpU,the rela- coefficients of carbon and aramid composites,
tive specimen thickness must be chosen to especially in the reinforcement direction, is not
ensure failure by shear delamination at the a trivial experiment. It must employ modem
specimen mid-radius rather than failure by dilatometer and interferometer methodsI4.
normal circumferential stresses, a?.
34.9 STRUCTURAL TESTING
34.8 SPECIAL TESTS
The unique challenges of designing and test-
Laminates can fail by interlaminar delamina- ing composite parts are due to the fact that the
tion along specific planes. Toughness data is as material and its micro- and macro-structures
important for characterization and failure pre- are created at the same time as the part. The
diction of composites as strength and stiffness design of critical structures must include the
data. Cracks in composites can propagate by design of the material and must consider the
Mode I, (crack faces opening normal to the unique behavior of composite materials which
crack plane), by Mode I1 (crack faces sliding in is influenced by processing techniques, actual
their planes), or by Mixed-Mode (combination service loading, and environmental condi-
of Mode I and Mode 11).The interlaminar frac- tions. Structural testing should start with tests
ture tests include the Double Cantilever Beam of small-scale models fabricated by the same
Test (Mode I), the Edge Delamination Test manufacturing process as the full-scale struc-
(Mode I), the End Notched Flexure (Mode 11), ture, followed by tests of prototype parts,
the Notched Three-Rail Shear Test (Mode 11), specimens cut from structural elements, and
the Cracked Lap Shear Test (mixed Modes I finally, full-scale tests9J9.
 References 793

REFERENCES 11. Greszczuk L.B., Shear Modulus Determination

of Isotropic and Composite Materials. ASTM
1. Lubin, G. (ed.). Handbook of Composites, New Special Technical Publication, 1969,460: 140-9.
York: Van Nostrand Reinhold, 1982. 12. Uemura M., Problems in Mechanical Testing
2. Tamopol‘skii, Yu.M. and Kincis, T.Ya, Static Test Methods of Advanced Composite Materials. In
Methods for Composites. New York: Van Proc. 10th Tsukuba General Symp., 1990, pp.
Nostrand Reinhold, 1985. 43-54.
3. Witney, J.M., Daniel, I.M. and Pipes, R.B., 13. Pagano, N.J. (ed.), Interlaminar Response of
Experimental Mechanics of Fiber Reinforced Composite Materials, (Composites Materials
Composite Materials, Rev. Ed., Society for Series; 5) Amsterdam: Elsevier, 1989.
Experimental Mechanics. Englewood Cliffs, NJ: 14. Carlsson, L.A. and Pipes, R.B., Experimental
Prentice-Hall, 1984. Characterization of Advanced Composite Materials.
4. Kelly. A. (ed.). Concise Encyclopedia of Composite Englewood Cliffs, NJ: Prentice-Hall, 1987.
Materials, Oxford: Pergamon Press, 1989. 15. Tarnopol’skii, Yu.M., Zhigun, I.G., and
5. Tarnopol’skii, Yu.M. and Vasiliev. V.V. (eds.). Polyakov, V.A., Spatially Reinforced Composites.
Structural Composites. A Handbook, Moscow: Lancaster: Technomic, 1992.
Mashinostroyenie, 1990. 16. Mechanical Testing of Advanced Fibre Composites,
6. Tsai, S.W., Theory of Composites Design, Dayton, Imperial College of Science and Technology,
Paris and Tokyo: Think Composites, 1992. London: University of London, 1992.
7. W y o m i n g Test Fixtures. High Performance Test 17. Sims G.D., Nimmo, W., Johnson, A.F. and
Fixtures. Product Catalog, Laramie: Wyoming Ferriss, D.H., Analysis of Plate-Twist Test for In-
Test Fixtures Inc, 1993. Plane Shear Modulus of Composite Materials.
8. Peters, S.T., Humphrey, W.D. and Foral, R.F., Teddington: National Physical Laboratory, 1992.
Filament Winding Composite Structure Fabrication. 18. Lee, S. and Munro, M., Evaluation of in-plane
Covina: SAMPE, 1991. shear test methods for advanced composite
9. Nikolaev, V.P., Panfilov, N.A., Popov, V.D., and materials by the decision analysis technique.
Sinitsyn, E.N., Analysis of the Failure Composites, 1986,17(1), 13-22.
Mechanism of Large-Scale Structures. Mech. 19. Nikolayev, V.P., Popov, V.D. and Sborovskii,
Composite Mater., 1993,29(2):203-11. A.K., Strength and Reliability of Wound
10. Pinderra, M.J., Gurdal, Z.C., Hidde, J.S. and Fiberglass Reinforced Plastics. Leningrad:
Herakovich, C.T., Mechanical and Thermal Mashinostroyeni ye, 1983.
Characterization of Unidirectional Aramid/
Epoxy. Report CCMS-86-08, VPI-E-86-29,
Virginia Polytechnic Institute and State
University, Blacksburg, VA, 1987.