Documente Academic
Documente Profesional
Documente Cultură
Composites
Abstract
Two experimental methods for determining the inter-laminar shear strength (ILSS),
are compared: the short-beam-shear (SBS) and the double-lap-shear (DLS) test
method. Specimens with a constant ply angle for all layers are considered. The
experimental results show a significant difference (up to 50%) in the obtained ILSS.
A finite element analysis shows that both test methods underestimate the real ILSS
and demonstrate that the standardized ILSS-evaluation procedures are more or less
valid for the SBS test, but require modifications in the case of the DLS test. Acoustic
the real ILSS from the DLS results. The real ILSS cannot be obtained from the SBS
test without an extended analysis. It is, however, possible to determine bounds for
Emission
the most critical failure mechanism of laminates. The resistance against de-
inter-laminar shear strength is defined as the shear stress at rupture where the
rial combinations. The main methods for the determination of the ILSS are
the short-beam-shear test (ASTM D-2344-00), the four-point shear test, the
Iosipescu test (ASTM D-5379-98), the tensile test (ASTM D-3518-94), and the
double notch shear test (ASTM D-3846-94). A new apparatus for the deter-
mination of the ILSS of flat and curved composite material is the compression
shear device (CSD) (see Rosselli and Santare [1], Schneider et al. [2]).
properties could be an ILSS test method, which allows to measure the static
ILSS as well as the fatigue strength under inter-laminar stresses. The above
mentioned Iosipescu test method has been examined extensively and was found
to be a highly effective and reliable method for predicting the static ILSS
(Adams and Lewis [3], Schneider et al. [2]). However, measuring the fatigue
2
Unlike the Iosipescu shear test, the double notch shear test method does not
of the drawbacks of the double notch (single-lap) shear test is the bending
eliminate this unwanted loading but the fixation might have an influence on
symmetric shape and therefore avoids the above drawback (Rosenkranz et al.
[5]).
This paper reports investigations of the DLS specimen and its suitability for an
accurate assessment of the static ILSS. Based on these results it can be decided
Results obtained from the double-lap-shear test are compared with those ob-
the SBS-method (e.g. Adams and Lewis [6], Adams and Lewis [3], Schneider
et al. [2]) and described certain problems with the determination of the ILSS.
We review the SBS-method as well and find it suitable (in combination with
simple finite element calculations) to predict upper and lower bounds for the
true ILSS. Such bounds are especially useful for establishing the quality of a
The first delamination failure loads (FDF-loads) are determined from acoustic
emission experiments. The FE-analyses are based on these FDF-loads and the
3
Failure Models
Kim and Soni [7] and simplified by Brewer and Lagace [8] is applied.
2 2 2
σ13 σ23 σ33
IL 2 + IL 2 + IL 2 =1 σ33 > 0
R13 R23 R33
2 2
σ13 σ23
IL 2 + IL 2 =1 σ33 < 0 . (1)
R13 R23
directions, see Figure 1 and Figure 2). Furthermore, this quadratic delamina-
tion failure hypothesis is based on the assumption that only tensile stresses
the strength values for delamintion are independent of the fiber directions of
IL IL
adjacent layers, R13 = R23 (see Puck [9]), the stresses or strengths may be de-
the interface. Due to the contamination of the interface during the fabrication
process the inter-laminar strength values used in Eq. (1) are in general smaller
For the prediction of ply failure (i.e. intra-laminar failure) the Tsai-Wu failure
m
In order to answer the question, how far away the stress state σij (m . . .
layer under consideration) is from the critical stress state m σij∗ (i.e. how large
is the multiplier which must be applied to the external loads in order to just
reach the critical stress state at the considered position), a safety factor m λ for
4
the individual layers or interfaces is introduced. Hence, the failure condition
consists of m plies, the safety factor for the laminate λlam for a given position
The risk parameter of the laminate E lam , as defined by Puck [9], Skrna-Jakl
1
E lam = . (4)
λlam
The risk parameter for delamination can also be obtained from the above
equations. Note that the risk parameter and the failure index, which is e.g.
defined as the sum of the left hand term of Eq. (1), are not the same. From a
a failure index.
gradients occurs in the critical region near the notch due to the notch geometry.
numerically, first delamination failure loads are required and have to be de-
termined, e.g., from acoustic emission experiments. The average stress failure
analysis may then be used to predict delamination (Brewer and Lagace [8])
and ply failure (de Azevedo Soriano and de Almeida [12]) of regions showing
high stress gradients. A critical length X avg from the free edge must be chosen
for the evaluation of average stresses and σijavg can be calculated by integrating
5
the stress value σij over X avg . These stress values are then used as stress input
ultimate loads Fult . The inter-laminar shear strength is evaluated from Fult
IL SBS 3Fult
(R13 )exp = , (5)
4bt
(where a parabolic shear stress distribution over the beam thickness is as-
sumed). t is the beam thickness and b the width of the SBS-specimen. In the
IL DLS Fult
(R13 )exp = , (6)
2bL
where 2bL is the overlapping area. This formula is similar to the standardized
The geometry of the SBS- and the DLS-specimens is shown in Figure 1 and
are kept constant. Effects of the notch width, of the notch depth (undercut) or
of the specimens thickness are not investigated. Shokrieh and Lessard [4] found
that those effects do not play a major rule. The SBS-specimen is analyzed for
different widths b. The span length is chosen smaller (span/t=3) than rec-
6
mode would be ply failure under the load introduction for the recommended
span length.
The finite element analysis code MSC-Nastran is used for all numerical inves-
Three dimensional linear elastic finite element models are employed to obtain
is justified for the DLS-specimen if, first delamination loads are applied. In
local failure occurs before complete failure. A comparison of linear and non-
Examples of meshes for the SBS- and the DLS-specimen are shown in Figure 3
material it is sufficient to model only one quarter of the DLS- and the SBS-
with other element types and integration schemes as well as with finer mesh
Experimental Procedures
tria) consists of 20 woven fabric glass fiber plies with a [0]20 -lay-up in epoxy
7
are the local and 1, 2, 3 are the global coordinate axes. The subscripts “C”
and “T” denote the compressive and the tensile loading direction. The values
with an MTS 810 TestStar II Material Testing System under static loading
conditions with a cross head speed of 0.5 mm/min (DLS tests) and of 1.3
speed of 0.5 mm/min was chosen from the tensile test standard ASTM D-
specimen geometry was investigated on four or five samples. For more details
regarding the test procedure see our work Rosenkranz et al. [5]. As mentioned
ASTM D-2344-00 (span/t=5) because the critical global failure mode would
be ply failure under the load introduction for the recommended span length.
This fact can be proved by the formula for the bending stress of a beam, which
span 2 RlT b t
< (7)
t 3Fult
Short-Beam-Shear Specimen
FE-results on other SBS-geometries look similar. The first local failure, ac-
8
determined ultimate load. This local failure does not affect the global fail-
stresses in the cross section between the support and the load introduction.
The shear stress distributions are plotted in Figure 6 for the “inside” and
“outside” region. These results are compared with those obtained from ASTM
bution being constant over the width of the specimen (analytical 2D solution).
It can be seen that the shear stresses which are predicted by the numerical
analyses in the inside region are approximately equal to the analytical result.
However, they are significantly higher at the outside region, although for the
considered laminate no classical “free edge effect” appears. This stress concen-
show additional stresses (σi2 6= 0, see Reckling [14]) and produce ILS-stress
concentrations at the free edge, which should be distinguished from the “free
factor f as follows
(σ13 )3D
f= , (8)
(σ13 )2D
where (σ13 )2D and (σ13 )3D stands for the shear stress obtained from 2D and
concentration factors within the critical region (between support and load
9
introduction in the middle of the specimen) is shown in Figure 7. The ILS-
stress concentration factors are drawn over the distance from the free edge for
a stress concentration appears at the free edge and approaches 1.21, when b
the stress is equal to the shear stress obtained from an analytical 2D solution.
Only the specimen with b=3 mm shows a different behavior. In that case, the
much smaller than b/2, which leads to a smaller difference in the ILS-stresses
between the inside and outside region. This shows that a decreasing width
temperature is noted. As found above the “real ILSSs” are higher than the
assume that the ILSSs are equal to the maximum shear stresses in the critical
region. Then the modified ILSS is obtained by multiplying the data by the
be 53 MPa for 293 K and 80 MPa for 77 K. Slight derivations occur only for
b/t = 1.5, but this specimen type showed large confidence intervals and should
be discarded.
Based on the above investigations two statements regarding the ILSS obtained
10
(1) The experimental ILSS represents a lower bound for the real ILSS. The
the non-linear material behavior leads, after the occurance of first delam-
within the critical region) to the inside region, and therefore this leads to
higher shear stresses at the ultimate load level. For this reason the real
(2) The maximum ILSS, which is obtained from the numerical analyses, rep-
resents an upper bound for the ILSS, if we assume that for F = Fult the
ing in the critical region. In the real structure a load F < Fult leads to the
first local delamination and, therefore, the real ILSS-values are smaller
approximately 15 % (293 K) and 20 % (77 K), respectively. The real ILSS lies
Double-Lap-Shear Specimen
Figure 9 shows the computed delamination risk parameter of one of the in-
vestigated DLS-models for F = Fult . The critical region occurs near the notch
(not around the hole), i.e. crack initiation starts at the notch for F < Fult .
The evaluation of the average delamination risk parameter within the over-
lapping zone leads to a value of approximately 50-60%. Hence, from this point
11
the “standardized” analysis method (Eq. (6)) which is based on mean shear
stresses within the overlapping zone for F = Fult . The problem of the double-
lap-shear test comes from the fact that the standard procedure describes a
The average stress analysis method (Brewer and Lagace [8]) was applied to
obtain the unknown ILSS-value. A priori, only the experimental first delami-
nation (FDF) loads are known from acoustic emission testing. The averaging
length X avg and the “real ILSS” are unknown. An ILSS evaluation consisting
(2) Linear FE-analysis and computation of the stress state σij within the
rewritten as:
IL avg avg
R13 (X avg ) = F (σ13 , σ33 IL
, R33 ), (9)
avg avg
where σ13 and σ33 are the averaged shear and tensile stresses over
The local ILSSs are shown in Figure 10 1 . The averaging length has a sig-
1 For these results the mesh shown in Figure 4 is refined. The critical region is
12
nificant influence only for X avg < 0.34 mm. For X avg > 0.34 mm all curves
with L = 4 mm and the other DLS-specimens comes from the influence of the
stress concentration due to the hole on the stress field in the notch region.
This influence is not observed in the case of the two other specimens. The
proper averaging distance for the present material system is found to be 0.34
IL
mm. This corresponds to the point on the R13 (X avg )-curves, where the ILSS
of all three curves first meet. Based on this investigation the “real ILSS” is
the “real ILSS”. As mentioned above, the SBS results provide bounds for the
“real ILSS”.
In Table 2 the “real ILSSs” are compared with the experimental SBS and DLS
type did not lead to reliable ILSSs. It can be seen that both tests underesti-
mate the real ILSS, but that especially the DLS-specimens shows a significant
ployed.
Conclusions
double-lap-shear test is not well suited for predicting the ILSS, if standard-
ized analysis methods are employed. Acoustic emission results and numerical
13
short-beam-shear test has the advantage of a simpler test arrangement, but it
provides only ILSS-bounds which can be useful if the quality of other ILSS-
tests is evaluated.
Acknowledgements
This work has been carried out within the Association EURATOM-OEAW.
References
[1] Rosselli, F., Santare, M. H., 1997. Comparison of short beam shear (SBS)
and interlaminar shear device (ISD) tests. Composites Part A 28A, 587–
594.
[2] Schneider, K., Lauke, B., Beckert, W., 2001. Compression shear test
[3] Adams, D. F., Lewis, E. Q., 1997. Experimental assessment of four com-
posite material shear test methods. J. Testing & Evaluat. 25, 174–181.
49–64.
[5] Rosenkranz, P., Humer, K., Weber, H. W., Pahr, D. H., Rammerstorfer,
F. G., 2001. Static and dynamic scaling experiments on double lap shear
14
[6] Adams, D. F., Lewis, E. Q., 1995. Experimental study of three- and
four-point shear test specimens. J. Comp. Tech. & Research 17, 341–349.
[7] Kim, R. Y., Soni, S. R., 1983. Experimental and analytical studies on
70–80.
[8] Brewer, J. C., Lagace, P. A., 1988. Quadratic stress criterion for initiation
[10] Tsai, S. W., Wu, E. M., 1971. A general theory of strength for anisotropic
Verlag, Düsseldorf.
[13] Xie, M., Adams, D. F., 1996. A nonlinear finite element analysis for
223.
[14] Reckling, K. A., 1967. Plastizitätstheorie und ihre Anwendung auf Fes-
15
L=4/6/8 2
3
2
2 t=4
1
b=10 70
Symmetry
Planes
Symmetry Plane
Symmetry Plane
16
1.00+00
9.33-01
8.67-01
8.00-01
7.33-01
6.67-01
6.00-01
5.33-01
4.67-01
4.00-01
3.33-01
2.67-01
2.00-01
Z
1.33-01
Y
X 6.67-02
4.47-08
4
FE-model: inside
FE-model: outside
Analytical solution
Distance in 3-direction [mm]
inside Fult
2
support
3 outside
2
1
1
0
0 10 20 30 40 50 60 70 80
ILS-stress [MPa]
17
1.2 SBS: b=3mm
SBS: b=6mm
ILS-stress concentration factor f SBS: b=8mm
SBS: b=10mm
1.1
0 1 2 3 4 5
Distance from free edge [mm]
80
ILSS [MPa]
T=77 K
60
T=293 K
40
20
1 1.5 2 2.5 3 3.5
Width-to-specimen thickness ratio b/t
K and 293 K
18
1.00+00
9.33-01
8.67-01
8.00-01
7.33-01
6.67-01
6.00-01
5.33-01
4.67-01
4.00-01
3.33-01
2.67-01
2.00-01
Z
1.33-01
Y
X
6.67-02
4.47-08
19
120
DLS: L=4mm
DLS: L=6mm
DLS: L=8mm
100
local ILSS
80
T=77 K
60
T=293 K
40
0.15 0.2 0.25 0.3 0.35 0.4
Distance from notch [mm]
Fig. 10. Local ILSS based on numerical DLS investigations and experimental FD-
F-loads
Fig. 11. Comparison of the different ILSS values for 77 K and 293 K
20
Table 1
Table 2
IL and the experimental SBS- and DLS-ILSS for 293
Comparison of the real ILSS R13
SBS DLS
Width b IL
R13 IL )SBS
(R13 Difference Length L IL
R13 IL )DLS
(R13 Difference
exp exp
21