Beruflich Dokumente
Kultur Dokumente
Materials
http://jcm.sagepub.com
Published by:
http://www.sagepublications.com
On behalf of:
American Society for Composites
Additional services and information for Journal of Composite Materials can be found at:
Email Alerts: http://jcm.sagepub.com/cgi/alerts
Subscriptions: http://jcm.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations (this article cites 28 articles hosted on the
SAGE Journals Online and HighWire Press platforms):
http://jcm.sagepub.com/cgi/content/refs/39/4/323
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
INTRODUCTION
HE MECHANISMS THAT lead to failure in composite materials are not yet fully understood, especially for matrix or fiber compression. The inadequate understanding of
the mechanisms, and the difficulties in developing tractable models of the failure modes
explains the generally poor predictions by most participants in the World Wide Failure
Exercise [17] (WWFE). The results of the WWFE round-robin indicates that the predictions of most theories differ significantly from the experimental observations, even when
analyzing simple laminates that have been studied extensively over the past 40 years [3].
The uncertainty in the prediction of initiation and progression of damage in composites
suggests a need to revisit the existing failure theories, assess their capabilities, and to develop
323
324
C. G. DAVILA ET AL.
new theories where necessary. The present paper describes a newly developed set of six
phenomenological criteria for predicting failure of unidirectional FRP laminates.
For simplicity, the development was based on plane stress assumptions. In the following
section, the motivation for developing nonempirical failure criteria to predict matrix
cracking under in-plane shear and transverse tension or compression is established by
examining several existing criteria. Next, new matrix failure criteria for matrix tension and
compression are proposed. Then, a new fiber kinking failure criterion for fiber compression
is developed by applying the matrix failure criteria to the configuration of the kink. The
resulting set of six failure criteria is denoted LaRC03 and are summarized in the Appendix.
Finally, examples of calculated failure envelopes are presented. The predicted results are
compared to the results of other available failure criteria and with experimental results.
STRENGTH-BASED FAILURE CRITERIA
Strength-based failure criteria are commonly used with the finite element method to
predict failure events in composite structures. Numerous continuum-based criteria have
been derived to relate internal stresses and experimental measures of material strength to
the onset of failure. In the following sections, the Hashin criteria [8,9] are briefly reviewed,
and improvements proposed by Sun et al. [10], and Puck and Schurmann [5,6] over
Hashins theories are examined.
Hashin Criteria 2D (1980)
Hashin established the need for failure criteria that are based on failure mechanisms. In
1973, he used his experimental observations of failure of tensile specimens to propose two
different failure criteria, one related to fiber failure and the other to matrix failure. The
criteria assume a quadratic interaction between the tractions acting on the plane of failure.
In 1980, he introduced fiber and matrix failure criteria that distinguish between tension
and compression failure. Given the difficulty in obtaining the plane of fracture for the
matrix compression mode, Hashin used a quadratic interaction between stress invariants.
Numerous studies conducted over the past decade indicate that the stress interactions
proposed by Hashin do not always fit the experimental results, especially in the case of
matrix or fiber compression. It is well known, for instance, that moderate transverse
compression ( 22<0) increases the apparent shear strength of a ply, which is not well
predicted by Hashins criterion. In addition, Hashins fiber compression criterion does not
account for the effects of in-plane shear, which reduce significantly the effective
compressive strength of a ply. Several researchers have proposed modifications to
Hashins criteria to improve their predictive capabilities. Modifications proposed by Sun
and Puck are described below.
IMPROVED CRITERIA FOR MATRIX COMPRESSION
Sun et al. [10] proposed an empirical modification to Hashins 1973 criterion for matrix
compression failure to account for the increase in shear strength due to compressive 22.
The Sun criterion is:
2
2
22
12
L
1
1
YC
S 22
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
325
Figure 1. Fracture of a unidirectional lamina subjected to transverse compression and in-plane shear.
T
S T T n
2
2
L
L
1
S L n
where T and L are the shear stresses acting on the fracture plane defined in Figure 1. The
direct contribution of 22 has been eliminated by assuming that fracture initiation is
independent of the transverse compressive strength. Internal material friction is
characterized by the coefficients L and T, which are determined experimentally.
A key element to Pucks proposal is the calculation of the angle of the fracture plane, ,
shown in Figure 1. Puck determined that matrix failures dominated by in-plane shear
occur in a plane that is normal to the ply and parallel to the fibers ( 0 ). When the
transverse compression rises, the angle of the fracture plane jumps to about 40 , and
increases with compression to 53 2 for pure transverse compression. In the WWFE,
Pucks predicted failure envelopes correlated reasonably well with the test results [6].
However, Pucks semi-empirical approach uses several material parameters that are not
physical and may be difficult to quantify without considerable experience with a particular
material system.
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
326
C. G. DAVILA ET AL.
eff
0=53
T
n
2 2= -YC
T
B
0= 530
-Yc
-1
tan(20)
eff=0.378 Y c
n
Figure 2. Mohrs circle for uniaxial compression and the effective transverse shear.
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
327
where the terms T and L are referred to as coefficients of transverse and longitudinal
influence, respectively, and the operand hxi 12x jxj: Matrix failure under compression
loading is assumed to result from a quadratic interaction between the effective shear
stresses acting on the fracture plane. The failure index for a failure mode is written as an
equality stating that stress states that violate the inequality are not physically admissible.
The matrix failure index (FIM) is
LaRC03#1
FIM
T 2 L 2
eff
effL 1
ST
Sis
where ST and SisL are the transverse and longitudinal shear strengths, respectively. The
subscript M indicates matrix failure. The subscript is indicates that for general laminates
the in situ longitudinal shear strength rather than the strength of a unidirectional laminate
should be used. The constraining effect of adjacent plies substantially increases the
effective strength of a ply. It is assumed here that the transverse shear strength ST is
independent of in situ effects. The calculation of the in situ strength SisL is discussed in a
later section.
The stress components acting on the fracture plane can be expressed in terms of the inplane stresses and the angle of the fracture plane, (see Figure 1).
n 22 cos2
T 22 sin cos
12 cos
Using Equations (3) and (5), the effective stresses for an angle of the fracture plane
between 0 and 90 are
T
eff
22 cos sin T cos
L
cos j12 j L 22 cos
eff
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
328
C. G. DAVILA ET AL.
1
tan 20
Puck [5] determined that when loaded in transverse compression, most unidirectional
graphiteepoxy composites fail by transverse shear along a fracture plane oriented at
0 53 2 . Therefore, the coefficient of transverse influence is in the range
0.21 T 0.36. Note that if the fracture plane were oriented at 0 45 , the coefficient
of transverse influence would be equal to zero.
The transverse shear strength ST is difficult to measure experimentally. However,
substituting Equation (9) into Equation (7) gives an expression relating the transverse
shear strength to the transverse compressive strength:
cos 0
S T Y C cos 0 sin 0
tan 20
10
For a typical fracture angle of 0 53 gives ST 0.378 Y C, as was shown in Figure 2.
The transverse shear strength is often approximated as ST 0.5 Y C, which would imply
from Equations (7) and (9) that the fracture plane is at 0 45 and that T 0. Using this
approximation, Hashins 1980 two-dimensional criterion for matrix failure in compression
becomes identical to his 1973 criterion.
The coefficient of longitudinal influence, L, can be determined from shear tests with
varying degrees of transverse compression. In the absence of biaxial test data, L can be
estimated from the longitudinal and transverse shear strengths, as proposed by Puck:
L
T
SL cos 20
T ) L C
L
S
S
Y cos2 0
11
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
329
12 , MPa
=52
100
=45
=40
40
80
=30
45
=0
60
40
52
20
0
-140
-120
-100
-80
-60
-40
-20
22 , MPa
Figure 3. Matrix failure envelopes for a typical unidirectional E-glassepoxy lamina subjected to in-plane
transverse compression and shear loading.
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
330
C. G. DAVILA ET AL.
0.3
T300/944
[25/90n]s
[252/-252/902]s
[908]s
Onset of
delamination
0.2
[0/90n/0]
Thin Thick
0.1
0
0
0.4
0.8
1.2
1.6
2.0
3 (T)
3 (T)
2a0
1 (L )
2 a 0L
2a0
It has a length 20 across the thickness of a ply, t. Physically, this crack represents a
distribution of matrixfiber debonds that may be present in a ply as a consequence of
manufacturing defects or from residual thermal stresses resulting from the different
coefficients of thermal expansion of the fibers and of the matrix. Therefore, the slit crack is
an effective crack, representing the macroscopic effect of matrixfiber debonds that occur
at the micromechanical level [16].
Plane stress conditions are assumed. The transverse tensile stress 22 is associated with
Mode I loading, whereas the in-plane shear stress 12 is associated with Mode II loading.
The crack represented in Figure 5 can grow in the 1-(longitudinal, L)-direction, in the
3-(transverse, T )-direction, or in both directions.
The components of the energy release rate for the crack geometry represented in Figure 5
were determined by Dvorak and Laws [19]. For mixed-mode loading, the energy release
rate for crack growth in the T and L directions, G(T ) and G(L), respectively, are given by:
a0 2 0 2
2
I 22 22 2II 044 12
2
a0 2 0 2
2 0 2
II
GL
44 12
4 I 22 22
GT
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
12
331
where it can be observed that the energy release rate G(L) for longitudinal propagation is a
function of the transverse slit size a0 and that it is not a function of the slit length in the
longitudinal direction aL0 .
The parameters i, i I, II in Equation (12) are the stress intensity reduction coefficients
for propagation in the transverse direction, and the parameters i, i I, II are the
reduction coefficients for propagation in the longitudinal direction. These coefficients
account for the constraining effects of the adjoining layers on crack propagation: the
coefficients are nearly equal to 1.0 when 20 t, and are less than 1.0 when a0 t.
Experimental results [18] have shown an increase in the in situ transverse tensile strength
of [ /90n]s, 0 ,30 ,60 , laminates for increasing stiffness of adjoining sublaminates
. This implies that the value of the parameter i decreases with increasing stiffness of
adjoining sublaminates. Considering that a transverse crack can promote delamination
between the plies, Dvorak and Laws [19] suggested that the effective value of i can be
larger than obtained from the analysis of cracks terminating at the interface, and suggested
the use of i i 1.
The parameters 0jj are calculated as [19]:
022 2
044
1
2
21
E2 E1
13
G12
The Mode I and Mode II components of the energy release rate can be obtained for the
T-direction using Equation (12) with i 1:
a0 0 2
2 22 22
a0 0 2
GII T
2 44 12
GI T
14
GIc T
15
where YisT and SisL are the in situ transverse tensile and shear strengths, respectively.
For propagation in the longitudinal direction, the Mode I and Mode II components of
the energy release rate are:
a0 0 2
4 22 22
a0 0 2
GII L
4 44 12
GI L
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
16
332
C. G. DAVILA ET AL.
17
Having obtained expressions for the components of the energy release rate and fracture
toughness, a failure criterion can be applied to predict the propagation of the slit crack
represented in Figure 5. Under the presence of both in-plane shear and transverse tension,
the critical energy release rate Gc depends on the combined effect of all microscopic energy
absorbing mechanisms such as the creation of new fracture surface. Relying on
microscopic examinations of the fracture surface, Hahn and Johannesson [22] observed
that the fracture surface topography strongly depends on the type of loading. With
increasing proportion of the stress intensity factor KII, more hackles are observed in the
matrix, thereby indicating more energy absorption associated with crack extension.
Therefore, Hahn proposed a mixed mode criterion written as a first-order polynomial in
the stress intensity factors KI and KII. Written in terms of the Mode I and Mode II energy
release rates, the Hahn criterion is
s
GI i
GI i
GII i
g
1,
1 g
GIc i
GIc i GIIc i
i T,L
18
where the material constant g is defined identically from either Equation (14) or (17) as:
2
GIc
022 YisT
g
19
A failure index for matrix tension can be expressed in terms of the ply stresses and in situ
strengths YisT and SisL by substituting either Equations (14), (15) or (16), (17) into the
criterion in Equation (18) to yield:
LaRC03#2
FIM
2 2
22
22
12
1 g T g T L 1
Yis
Yis
Sis
20
The criterion presented in Equation (20), with both linear and quadratic terms of the
transverse normal stress and a quadratic term of the in-plane shear stress, is similar to the
criteria proposed by Hahn and Johannesson [22], Liu and Tsai [23] (for transverse tension
and in-plane shear), and Puck and Schurmann [5]. In addition, if g 1 Equation (18)
reverts to the linear version of the criterion proposed by Wu and Reuter [24] for the
propagation of delamination in laminated composites:
GI
GII
1
GIc GIIc
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
21
333
Furthermore, using g 1, Equation (20) reverts to the well-known Hashin criterion [8]
for transverse matrix cracking under both in-plane shear and transverse tension, where the
ply strengths are replaced by the in situ strengths:
FIM
22
YisT
2 2
12
L 1
Sis
22
23
It can be observed from Equations (23) that the in situ strengths of thick plies YisT and
are functions of the toughnesses GIc(T ) and GIIc(T ) of the material and the size of the
material flaw, 2a0. Therefore, the in situ strengths for thick plies are independent of the ply
thickness, as has been observed by Dvorak and Laws [19] and Leguillon [25], and as was
shown in Figure 4.
SisL
12, Sis
2a0
22, Yis
Figure 6. Geometry of slit crack in a thick embedded ply subjected to tension and shear loads.
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
334
C. G. DAVILA ET AL.
In the case of thin plies, crack defects can only grow in the longitudinal (L) direction,
or trigger a delamination between the plies. The in situ strengths can be calculated from the
components of the fracture toughness as:
s
8GIc L
t022
s
8GIIc L
L
Sis
t044
YisT
Thin ply:
24
p
where it can be observed that the in situ strengths are inversely proportional to t. The
toughnesses GIc (L) and GIIc (L) can be assumed to be the values measured by standard
Fracture Mechanics tests, such as the double cantilever beam test (DCB) for Mode I and
the end-notch flexure test (ENF) test for Mode II. Using Equations (23) and (24), Dvorak
and Laws [19] obtained a good correlation between the predicted and experimentally
obtained in situ strengths of both thick and thin 90 plies in [0/90n/0] laminates, as shown
in Figure 4.
FAILURE OF UNIDIRECTIONAL LAMINATES
The fracture of a unidirectional specimen is taken as a particular case of a thick ply [19].
The defect size is 2a0, as for thick embedded plies. However, in the absence of constraining
plies, the critical initial slit crack is located at the surface of the laminate. For tensile
loading, the crack can be located at the edge of the laminate, which increases the energy
release rate when compared with a central crack. In the case of shear loading, there is no
free edge, so the crack is a central crack, as shown in Figure 8. Crack propagation for
unidirectional specimens subjected to tension or shear loads is obtained from the classic
solution of the free edge crack [26].
GIc T 1:122 a0 022 Y T 2
25
22, YisT
12, SisL
22, Y T
2a0
2a0
12, SL
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
335
where YT and SL are the material tensile and shear strengths as measured from
unidirectional laminate tests. Note that the in situ strengths for thick plies can be obtained
as a function of the strength of unidirectional laminates by substituting Equations (25)
into (23):
p
YisT 1:12 2Y T
p
SisL 2S L
Thick ply :
26
The failure criterion for unidirectional plies under in-plane shear and transverse tension
is represented in Equation (20). The toughness ratio g for a thick laminate can also be
calculated in terms of the unidirectional properties by substituting Equations (25) into
Equation (19) to yield:
T 2
0
GIc
2 22 Y
1:12 0
g
GIIc
44 S L
27
FIF
"11
1
"T1
28
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
336
C. G. DAVILA ET AL.
Several authors [32,33] have considered that misaligned fibers fail by the formation
of a kink band when local matrix cracking occurs. Potter et al. [34] assumed that
additional failure mechanisms that may occur under uniaxial longitudinal compression, such as crushing, brooming, and delaminations, are essentially triggered by matrix
failure.
In the present approach, the compressive strength X C is assumed to be a known
material property that can be used in the LaRC03 matrix damage criterion (Equation (4))
to calculate the fiber misalignment angle that would cause matrix failure under uniaxial
compression.
CALCULATION OF FIBER MISALIGNMENT ANGLE
The imperfection in fiber alignment is idealized as a local region of waviness, as shown
in Figure 9. The ply stresses in the misalignment coordinate frame m shown in Figure 9 are
m
cos2 11 sin2 22 2 sin cos j12 j
11
m
22
sin2 11 cos2 22 2 sin cos j12 j
m
12
29
2
30
where SisL is the in situ longitudinal shear strength defined in Equation (23) for
thick plies and in Equation (24) for thin plies. Solving for C leads to the quadratic
equation:
2
tan
L
SisL
Sis
L
C
tan
0
XC
XC
22
11
31
m
22
m
11
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
337
1
1
1 4 SisL =X C L SisL =X C
A
2 SisL =X C L
32
Note that if L were neglected and were assumed to be a small constant angle,
Equation (30) would give C SisL =X C , which is the expression derived by Argon [30] to
estimate the fiber misalignment angle.
The total misalignment angle can be decomposed into an initial (constant)
misalignment angle 0 that represents a manufacturing imperfection, and an additional
rotational component R that results from shear loading. The angles 0 and R can be
calculated using small angle approximations and Equations (29)
m
12
11 22 j12 j
G12
G12
m
12
XC C
0
C
R
C
0
XC
) 1
G12 X C
G12
33
where the absolute value of the shear | 12| is used because the sign of is assumed to be
positive, regardless of the sign of the shear stress. Recalling that 0 R, it is now
possible to solve Equations (33) for in terms of C.
j12 j G12 X C C
G12 11 22
34
Fiber compression failure by the formation of a kink band is predicted using the stresses
from Equation (29) and the failure criterion for matrix tension or matrix compression. For
m
< 0, the criterion is the Mohr-Coulomb criterion given in
matrix compression 22
T
0. The criterion for fiber kinking becomes:
Equation (4), with 0 and eff
m
m
j12
j L 22
1
FIF
SisL
LaRC03#4
35
For fiber compression with matrix tension, the transformed stresses of Equation (29) are
substituted into the matrix tensile failure criterion given in Equation (20) to yield the
following criterion for fiber kinking:
LaRC03#5
m 2 m 2
m
22
FIF 1 g T g 22T 12L 1
Yis
Yis
Sis
36
It is interesting to note that for 22 0, the fiber failure criterion in Equation (35) becomes
FIF
11 j12 j
L 1
XC
kS
with k
1
>1
1 2 2L
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
37
338
C. G. DAVILA ET AL.
The linear interaction between 11 and 12 in Equation (37) is identical to the form used by
Edge [7] for the WWFE. For T300/914C, Edge suggests using an empirical value of
k 1.5. However, Edge also indicates that other researchers have shown excellent
correlation with experimental results with k 1. Using the WWFE strength values of
X C 900 MPa, SL 80 MPa, Y C 200 MPa, an assumed fracture angle in transverse
compression of 53 , and Equations (11) and (32), we get: L 0.304, C 5.3 . With the
approximation C , Equation (37) gives k 1.07.
LaRC03#6
FIM
mT
effT
S
2 mL 2
effL 1
Sis
38
mT
mL
and eff
are defined as in Equation (6), but in terms
where the effective shear stresses eff
of the in-plane stresses in the misalignment frame, which are defined in Equations (29).
m
mT
22
cos sin T cos
eff
m
mL
L m cos
eff
cos 12
22
39
mT
mL
and eff
are functions of
As for all matrix compressive failures herein, the stresses eff
the fracture angle , which must be determined iteratively.
VERIFICATION PROBLEMS
Example 1
All of the failure modes represented by the six LaRC03 criteria (Equations (4), (20),
(28), (35), (36) and (38), and summarized in the Appendix) can be represented as a failure
envelope in the 11 22 plane. An example is shown in Figure 10 for a 0 E-glass/MY750
epoxy lamina. Pucks WWFE analysis results [6] are shown for comparison. In the figure,
there is good agreement between LaRC03 and Puck in all quadrants except biaxial
compression, where LaRC03 predicts an increase of the axial compressive strength with
increasing transverse compression. Testing for biaxial loads presents a number of
complexities, and experimental results are rare. However, Waas and Schultheisz [35]
report a number of references in which multiaxial compression was studied by superposing
a hydrostatic pressure in addition to the compressive loading. For all materials considered,
there is a significant increase in compressive strength with increasing pressure. In
particular, the results of Wronski and Parry [36] on glass/epoxy show a strength increase
of 3.3 MPa per MPa of hydrostatic pressure, which gives an actual strength increase of
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
339
22, MPa
50
FIF, LaRC03 #5
FIM, LaRC03 #2
22m =0
1500
1000
11, MPa
500
500
1000
1500
FIF, LaRC03 #4
FIF, LaRC03 #3
50
Puck
FIM, LaRC03 #6
100
11=Yc
150
FIM, LaRC03 #1
4.3 MPa per MPa of applied transverse biaxial stress. More recently, Sigley et al. [37]
found 3271% increase in compressive strength per 100 MPa superposed pressure. The
results of Wronsky et al. is comparable to the 4.3 MPa/MPa increase in compressive
strength for the plane stress failure envelope in Figure 10.
Example 2
A comparison of results from various failure criteria with the experimental results in
the 22 12 stress plane obtained from the WWFE [38] is shown in Figure 11. It can be
observed that within the positive range of 22, all the quadratic failure criteria and
LaRC03 give satisfactory results. Since the Maximum Stress Criterion does not prescribe
interactions between stress components, its failure envelope is rectangular. The most
interesting behavior develops when 22 becomes compressive. Hashins 1973 criterion
gives an elliptical envelope with diminishing 12 as the absolute value of compressive 22
increases, while the experimental data shows a definite trend of shear strength increase as
22 goes into compression.
The envelope for Hashins 1980 criteria was calculated using a transverse strength
obtained from Equation (10), and it provides a modest improvement in accuracy
compared to the 1973 criterion. Of the criteria shown in Figure 11, Sun (Equation (1)),
LaRC03#1 (Equation (4)), and Puck (Equation (2)) capture the shear strength increase at
the initial stage of compressive 22. The failure envelope for Suns criterion (Equation (1))
was calculated using L 0.336 from Equation (11). The results indicate a significant
improvement over Hashins criteria. An even better fit would have been achieved using a
higher value for L. Pucks envelope, which was extracted from the 2002 WWFE [6],
appears to be the most accurate, but it relies on fitting parameters based on the same test
data. The LaRC03 curve uses the stiffnesses and strengths shown in Table 1, an assumed
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
340
C. G. DAVILA ET AL.
12 , MPa
WWFE test 02
100
Puck 02
LaRC03 #1
Sun 96
Max. stress
50
Hashin 73
Hashin 80
Hashin
LaRC03 #2
-100
-150
-50
22 , MPa
50
Figure 11. Failure envelopes and WWFE test data [2] for unidirectional composite E-Glass/LY556.
E22
G12
12
YT
YC
SL
17,700
5830
0.278
36
138
61
0 53 , and no other empirical or fitting parameter. In the case of matrix tension, Pucks
predicted failure envelope is nearly identical to LaRC03#2.
Example 3
Cross-ply Laminates
Shuart [39] studied the compression failure of [ ]s laminates and found that for
< 15 , the dominant failure mode in these laminates is interlaminar shearing; for
15<< 50 , it is in-plane matrix shearing; and for >50 , it is matrix compression. Fiber
scissoring due to matrix material nonlinearity caused the switch in failure mode from
in-plane matrix shearing to matrix compression failure at larger lamination angles. The
material properties shown in Table 2 were used for the analysis. The angle 0 was assumed
to be a typical 53 . For other lamination angles, the fracture angle was obtained by
searching numerically for the angle that maximizes the failure criterion in Equation (4).
The in situ shear strength did not need to be calculated since it is given by Shuart [39] as
SisL 95:1 MPa.
The results in Figure 12 indicate that the predicted strengths using LaRC03 criteria
correlate well with the experimental results. The compressive strength predicted using
Hashin 1973 criteria is also shown in Figure 12. For <20 , the Hashin criteria result in an
overprediction of the failure load because the criterion does not account for the effect of
in-plane shear on fiber failure. For lamination angles near 70 , the use of the Hashin
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
341
127,600
E22
G12
12
XC
YC
SLis
11,300
6000
0.278
1045
244
95.1
Hashin 73
LaRC03
Test, Shuart 89
=00
=44
0=530
criteria results is in an underprediction of the failure load because the criteria do not
account for the increase in shear strength caused by transverse compression. All of these
effects are represented by the LaRC03 criteria, which results in a good correlation between
the calculated and experimental values.
CONCLUSIONS
The results of the recently concluded World Wide Failure Exercise indicate that the
existing knowledge on failure mechanisms needs further development. Many of the
existing failure models could not predict the experimental response within a tolerable limit.
In fact, differences of up to an order of magnitude between the predicted and experimental
values were not uncommon. In this paper, a new set of six phenomenological
failure criteria is proposed. The criteria for matrix and fiber compression failure are
based on a Mohr-Coulomb interaction of the stresses associated with the plane of
fracture. Failure envelopes for unidirectional laminates in the 11 22 and 22 12 stress
planes were calculated, as well as the compressive strength of cross-ply laminates.
The predicted failure envelopes indicate good correlation with experimental results. The
proposed criteria, referred to as LaRC03, represent a significant improvement over
the commonly used Hashin criteria, especially in the cases of matrix or fiber failure in
compression.
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
342
C. G. DAVILA ET AL.
APPENDIX: SUMMARY OF LARC03 FAILURE CRITERIA
Matrix Cracking
Matrix tension, 22 0
2 2
22
22
12
FIM 1 g T g T L
Yis
Yis
Sis
11 < Y C
mT 2 mL 2
FIM effT effL
S
Sis
11 Y C
T 2 L 2
FIM eff
effL
T
S
Sis
Fiber Failure
Fiber tension, 11 0
FIF
"11
"T1
m
22
0
m
m 2 m 2
FIF 1 g 22T g 22T 12L
Yis
Yis
Sis
Required Unidirectional Material Properties: E1, E2, G12, 12, XT, XC, YT, YC, SL,
GIc(L), GIIc(L)
Optional Properties: 0, L
The in situ strengths for a thin embedded ply are:
s
s
8G
L
8GIIc L
1
221
Ic
T
L
0
and Sis
where 22 2
Yis
E2 E1
t022
t044
and
044
1
G12
Toughness ratio g is obtained from fracture mechanics test data or from the unidirectional
properties:
2
GIc L
0 Y T
. Otherwise, g 1:122 22
For a thin embedded ply, g
GIIc L
044 S L
The effective shear stresses for matrix compression are
(
T
eff
22 cos sin T cos
L
cos j12 j L 22 cos
eff
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
343
T
1
tan 20
and
L
where
C
j12 j G12 X
G12 11 22
and
C tan1 @
1
1
1 4 SisL =X C L SisL =X C
A
2 SisL =X C L
REFERENCES
1. Soden, P.D., Hinton, M.J. and Kaddour, A.S. (1998). A Comparison of the Predictive
Capabilities of Current Failure Theories for Composite Laminates, Composites Science and
Technology, 58: 12251254.
2. Soden, P.D., Hinton, M.J. and Kaddour, A.S. (2002). Biaxial Test Results for Strength and
Deformation of a Range of E-glass and Carbon Fibre Reinforced Composite Laminates: Failure
Exercise Benchmark Data, Composites Science and Technology, 62: 14891514.
3. Hinton, M.J., Kaddour, A.S. and Soden, P.D. (2002). A Comparison of the Predictive
Capabilities of Current Failure Theories for Composite Laminates, Judged against Experimental
Evidence, Composites Science and Technology, 62: 17251797.
4. Hinton, M.J. and Soden, P.D. (1998). Predicting Failure in Composite Laminates: The
Background to the Exercise, Composites Science and Technology, 58: 10011010.
5. Puck, A. and Schurmann, H. (1998). Failure Analysis of FRP Laminates by Means of Physically
Based Phenomenological Models, Composites Science and Technology, 58: 10451067.
6. Puck, A. and Schurmann, H. (2002). Failure Analysis of FRP Laminates by Means of Physically
Based Phenomenological Models, Composites Science and Technology, 62: 16331662.
7. Edge, E.C. (1998). Stress-based Grant-sanders Method for Predicting Failure of Composite
Laminates, Composites Science and Technology, 58: 10331041.
8. Hashin, Z. and Rotem, A. (1973). A Fatigue Criterion for Fiber-reinforced Materials, Journal of
Composite Materials, 7: 448464.
9. Hashin, Z. (1980). Failure Criteria for Unidirectional Fiber Composites, Journal of Applied
Mechanics, 47: 329334.
10. Sun, C.T., Quinn, B.J. and Oplinger, D.W. (1996). Comparative Evaluation of Failure Analysis
Methods for Composite Laminates DOT/FAA/AR-95/109.
11. Salencon, J. (2001). Handbook of Continuum Mechanics: General Concepts, Thermoelasticity,
Springer, New York.
12. Kawabata, S. (1982). Strength of Epoxy Resin under Multiaxial Stress Field, ICCM-IV
Conference, Tokyo, Japan.
13. Boehler, J.P. and Raclin, J. (1985). Failure Criteria for Glass-fiber Reinforced Composites
Under Confining Pressure, J. Struct. Mech., 13: 371392.
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
344
C. G. DAVILA ET AL.
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
345
37. Sigley, R.H., Wronski, A.S. and Parry, T.V. (1992). Axial Compressive Failure of Glass-fibre
Polyester Composites under Superposed Hydrostatic Pressure: Influence of Fibre Bundle Size,
Composite Science and Technology, 43: 171183.
38. Soden, P.D., Hinton, M.J. and Kaddour, A.S. (1998). Lamina Properties, Lay-up
Configurations and Loading Conditions for a Range of Fibre-reinforced Composite
Laminates, Composites Science and Technology, 58: 10111022.
39. Shuart, M.J. (1989). Failure of Compression-loaded Multidirectional Composite Laminates,
AIAA Journal, 27: 12741279.
Downloaded from http://jcm.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on September 25, 2007
2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.