Sie sind auf Seite 1von 10

Journal of Composite

Materials
http://jcm.sagepub.com/

Calculation of Edge Stresses in Multi-Layer Laminates by


Sub-Structuring
A.S.D. Wang and Frank W. Crossman
Journal of Composite Materials 1978 12: 76
DOI: 10.1177/002199837801200106
The online version of this article can be found at:
http://jcm.sagepub.com/content/12/1/76

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: http://jcm.sagepub.com/content/12/1/76.refs.html
Downloaded from jcm.sagepub.com at Bibliotheek TU Delft on September 5, 2014

>> Version of Record - Apr 1, 1978


What is This?

Downloaded from jcm.sagepub.com at Bibliotheek TU Delft on September 5, 2014

Calculation of Edge Stresses in


Multi-Layer Laminates by Sub-Structuring
A. S. D. WANG

Drexel University
Philadelphia, Pennsylvania 19104
AND

FRANK W. CROSSMAN

Lockheed Palo Alto Research Laboratory


Palo Alto, California

(Received November 3, 1977)

ABSTRACT
A

calculating interlaminar stresses is accomlaminas as a quasi-homogeneous lamina.


Numerical examples show the accuracy and the limitations of the smearing
technique.

simplified

method for

plished by smearing

two

or more

field in composite laminates that have edge


be calculated on the basis of anisotropic elasticity
[1 ] . The material in the lamina comprising the laminate is assumed to be homogeneous and anisotropic even though the lamina material is usually a fiber-reinforced system. This approach is apparently justified if the thickness of the lamina is
several orders of magnitude larger than the diameter of the fiber. Under service
load, a three dimensional stress field is induced in general, and significant stress
concentration occurs when free edges are present in the laminate.
In order to evaluate the nature of the stress concentration, a three-dimensional
elasticity boundary-value problem must be solved. Unfortunately, even with a long
finite-width tensile coupon whose stress field is governed essentially by a twodimensional system of equations, it defies any rigorous closed-form solution.
HE THREE DIMENSIONAL stress

Tboundaries can theoretically

or numerical approaches such


finite-element
have to be used. The
and
methods,
[6, 7]
[8-10]
are
often
for
the
evaluation of local
methods
approximate analytical
inadequate

Rather,

as

some

approximate analytical methods [2-S]

finite-difference

Downloaded from jcm.sagepub.com at Bibliotheek TU Delft on September 5, 2014

76

stress

concentration, and their procedures become extremely tedious when

multi-

e.g. the finite element

procedure,
layer laminate is involved. Numerical methods,
provide the option of placing highly populated grid-work near regions of possible
stress concentration [8]. However, one major difficulty in the numerical approach
is that the number of nodes in the grid-work can be excessive if laminates of large
number of layers are to be examined, and the limits of available computational
capacity can easily be exceeded.
In this paper, a simple scheme is suggested, which calculates the stress field in
laminates of any number of layers without increasing the computational requirement. While the scheme is essentially a compromise which had to be made in
approaching a very difficult problem, it is hoped that the results obtained therein
would still give the desired accuracy. The scheme is derived within the premise of
Saint-Venants principle which states, in essence, that the stress at a point distance
from the boundary does not distinguish the actual boundary force distribution
from its static equivalence. Thus, for a multi-layer laminate, the stress distribution
along a particular laminar interface should remain unaltered if layers at some
distance from this interface are replaced by a statically equivalent body~.
As an example, let us consider the 6-layer laminate shown in Figure 1. If the
stress distribution along the interface between layer 5 and 6 is desired, one wishes
to &dquo;smear&dquo; layers 1, 2, 3 and 4 into a single uniform layer having &dquo;statically
equivalent&dquo; properties such that the stresses along the specified interface remain
unchanged. The problem then can be treated as one of 3-layer laminate as depicted
in the same Figure l. Similarly, stresses in any other layer or on any other layerinterface can be calculated by sub-structuring the laminate with fewer number of
effective layers. In this manner, the number of laminar interfaces are greatly reduced facilitating finite element calculations with a minimal grid size requirement.

Figure 1. A schematic illustration of a 6-layer laminate


(left) with first four layers smeared (right).

Two practical questions now remain2 : (1) What constitutes a &dquo;statically equivalent&dquo; replacement body? (2) What is considered a proper distance between the

1
By equivalence, it is meant that the internal stresses acting on the surface which separates
the replacement body from the laminate will remain statically equivalent in the sense of
Saint-Venant.
2
The plausibility of Saint-Venant principle has been argued. Exceptions have been cited
where the premise of Saint-Venants principle becomes invalid. See discussions in, e.g.
Sokolnikoff [ 11].

Downloaded from jcm.sagepub.com at Bibliotheek TU Delft on September 5, 2014

77

replacement body and the interface of

interest

so

that the

computed

stresses

are

accurate?

Unfortunately, no quantitative generalization can be made in answering these


questions, as the Saint-Venants principle itself is a scheme of approximation to a
mechanically complicated situation. For laminates, the &dquo;replacement body&dquo; may be
a sub-laminate whose &dquo;bulk&dquo; or statically equivalent property is described reasonably well e.g. by, a simple laminated- plate theory [12]. As regard to determining
the proper distance separating the replacement body and the interface of interest, it
is another matter of practicability. Depending on the degree of accuracy desired,
the distance can be made as large as is practically possible, or it can be made as
small as the thickness of the layer adjacent to the specified interface. Generally
speaking, the larger the distance, the better the approximation.
In what follows, we shall demonstrate, by means of numerical examples, that the
sub-structure scheme just described is very effective and yields satisfactory results
for the class of problems considered herein. The scheme, as it seems plausible, could
have far-reaching implication in a truly three-dimensional stress field calculation,
such as the stress field around a hole. In this case, the problem of increased computer storage requirements and calculations time is especially acute [13, 14] . With
the sub-structuring scheme, one has the option to populate only a chosen interface
with fine grids near the free edge region. In the &dquo;replacement body,&dquo; any stress
singularity will be averaged out; hence a fine grid-work will not be necessary there.
NUMERICAL EXAMPLES
We shall consider as illustrations two quasi-isotropic composite laminates in the
form of a finite-width tensile coupon. The stacking sequences of the two laminates
are respectively,
(45/-45/0/90)~ and (90/0/45/-45),; and their width (2b) to
thickness (2t) ratio is taken as 4 to 1. The stress fields in these laminates loaded by
uniaxial tension have previously been solved by a finite element method [8] using
material data for a typical graphite-epoxy system. In order to achieve an accurate
prediction of the stress field in regions of steep stress gradients, a great number of
nodes had to be assigned along each of the laminar interfaces of the laminate.
To illustrate the use of the sub-structuring scheme, we shall solve the same two
problems by considering one interface at a time. The nodal density along the
chosen interface will remain the same as in the original problem in order to provide
a consistent comparison; but in the &dquo;smeared&dquo; or &dquo;replacement&dquo; layer only the
&dquo;averaged&dquo; elastic constants were used to calculate stresses in that layer. We then
compare the stresses along the chosen interface as they are computed by the two
different procedures. The material properties of the replacement layer are calculated by means of a simple lamination theory [12] which does not include any
free-edge effect.
Figure 2 shows for the (45/-45/0/90)s laminate the normal-to-thickness stress

78

Downloaded from jcm.sagepub.com at Bibliotheek TU Delft on September 5, 2014

Figure 2

az-distribution along 90/90 interface for (45/~5/0/90JSs


with various

layers smeared.

(90/90) interface (the center-line of the laminate cross-section). oz is


here
against the logarithmic distance from the free edge in order to show
plotted
the highly localized stress concentration near the free edge (for details, refer to (8]).
In this figure, az computed without sub-structuring is shown in solid line; when the
(45/-45) layers are &dquo;smeared,&dquo; az is shown in dashed line; when the (0/90) layers
are &dquo;smeared,&dquo; oz is shown by the line connecting circled dots. Finally, when the
(45/-45/0) layers are &dquo;smeared,&dquo; az is shown by isolated triangles. It is seen that Qz
computed in all cases agree extremely well, except when (0/90) are smeared. Even
in this case, the discrepancy occurred only within the free-edge region. This is of
course expected, for by smearing the (0/90) layers, fine details inside the layers are
actually averaged out. On the other hand, edge effects are absent inside the laminate ; and the stress field in that region can be described well by the lamination
theory which is the basis of the smearing procedure.
Significant interlaminar shear stress concentration was found present along the
(45) interface [8]. Figure 3 displays this shear stress, TXZ, computed by the
sub-structuring scheme and plotted against the logarithmic distance from the free
edge. It is seen that with the (0/90) layers smeared, TxZ is everywhere close to the
value calculated without smearing. On the other hand, when the (45/-45) layers
are smeared, the distortional mechanism between the two layers (the so-called
scissoring effect) is removed resulting in a zero T xz everywhere in the field. The
effect is the same if we smear the (45/-45/0) layers.
oz along the

Downloaded from jcm.sagepub.com at Bibliotheek TU Delft on September 5, 2014

79

Figure 3.

T xz-distribution along 451-45 interface for (45/~5/0/90Jss


with various layers smeared.

clear that we must not smear the two adjacent layers if the stresses on a
interface
are to be calculated. In order to gain more insight about the
particular
stress field in the entire laminate with or without sub-structuring, let us examine
Figure 4 which shows a through-thickness distribution of QZ near the free edge of
the laminate. Again, the solid line depicts az computed without sub-structuring; the
dashed line is for the case when the (45/-45) layers are smeared. These two lines
agree well inside the 0 and the 90 layers as expected, but depart from each other
inside the (45) layers. On the other hand, when the (0/90) layers are smeared, az
(circled dots) is accurate inside the (45) layers, but is averaged out inside the
(0/90) layers. Finally, when the (45/-45/0) layers are smeared (black dots), the
stress a., inside the 90 layer is still very accurate, and is averaged out inside the
smeared layer.
A similar through-thickness distribution of TxZ (near free edge) is shown in
Figure 5. As was mentioned earlier, the smearing of the (45) layers removes the
distorsional effect and TxZ is zero in the field. The smearing of the (0/90) layers
does not seem to have any effect on the accuracy of the TxZ stress even within the
smeared layer. This may be due to the fact that the laminate preserves the monoclinic symmetry of the individual layers; the smearing of the (0/90) layers does not
alter this symmetry property.
As a second example, let us examine the stresses in the (90/0/45/-45)s laminate.
It is

80

now

Downloaded from jcm.sagepub.com at Bibliotheek TU Delft on September 5, 2014

Figure 4. Through-thickness distribution of a,

(45/-~5/0/90JS

with various

near free edge of


layers smeared.

Figure 5. Through-thickness distribution of T xz

near

free edge of

(451-4510190)s with various layers smeared.


Through-thickness distributions of oz and TXZ, both near the free edge region, are
displayed in Figure 6 and Figure 7 respectively. Results similar to the previous
example are again obtained.
Both examples demonstrated the applicability of the sub-structuring scheme just

Downloaded from jcm.sagepub.com at Bibliotheek TU Delft on September 5, 2014

81

Figure 6. Through-thickness distribution of az near free edge of


(90101451-45)s with various layers smeared.

Figure 7. Through-thickness distribution of TxZ near free edge of


(90/0/45/^45Js~ with various layers smeared.

presented. Application of this approach would be especially useful in the calculation of a fully three dimensional stress field such as that in a laminate structure
containing holes or cutouts.
Caution should be exercised, however, in the use of the suggested scheme, for

82

Downloaded from jcm.sagepub.com at Bibliotheek TU Delft on September 5, 2014

the method would be only as accurate (and perhaps less) as the specific numerical
procedures used. In the present illustrations, for instance, a standard constant-strain
finite element method is used; the accuracy of the results obtained therein is still

subject

to

scrutiny.
ACKNOWLEDGMENT

This study was supported under the Lockheed Independent Research Program.
Discussions with Dr. N. J. Pagano on this topic are gratefully acknowledged.

REFERENCES
," Holden-Day
Lekhnitsky, "Theory of Elasticity of An Anisotropic Elastic Body
(1963).
2. N. J. Pagano, "On the Calculation of Interlaminar Normal Stresses in Composite Laminates," J. Composite Materials, Vol. 8 (1974), p. 65.
3. R. B. Pipes and N. J. Pagano, "Interlaminar Stresses in Composite Laminates - An Approximate Elasticity Solution," J. Appl. Mechanics, Vol. 41 (1974), p. 668.
4. S. Tang and A. Levy, "A Boundary Layer Theory - Part II: Extension of Laminated Finite
Strip," J. Composite Materials, Vol. 9 (1975), p. 42.
5. P. W. Hsu, "Interlaminar Stresses in Composite Laminates — A Perturbation Analysis,"
Ph.D. Thesis, VPI and SU (1976).
6. R. B. Pipes and N. J. Pagano, "Interlaminar Stresses in Composites Under Uniform Axial
Extension," J. Composite Materials, Vol. 4 (1970), p. 538.
7. N. J. Pagano and R. B. Pipes, "The Influence of Stacking Sequence on Laminate Strength,"
J. Composite Materials, Vol. 5 (1971), p. 55.
8. A. S. D. Wang and F. W. Crossman, "Some New Results on Edge Effects in Symmetric
Composite Laminates," J. Composite Materials, Vol. 11 (1977), p. 92.
9. A. S. D. Wang and F. W. Crossman, "Edge Effects on Thermally Induced Stresses in
Composite Laminates," J. Composite Materials, Vol. 11 (1977), p. 300.
10. F. W. Crossman and A. S. D. Wang, "Stress Field Induced by Transient Moisture Sorption in
Finite-Width Composite Laminates," J. Composite Materials, Vol. 12 (1978), p. 2.
Mathematical Theory of Elasticity," McGraw Hill (1956), p. 89.
11. I. S. Sokolnikoff, "
12. R. M. Jones, "Mechanics of Composite Materials," Scripta (1975).
13. R. M. Barker, J. R. Dana and C. W. Pryor, "Stress Concentrations Near Holes in Laminates," J. Eng. Mech., ASCE, June (1974), p. 477.
14. E. F. Rybicki and D. W. Schmueser, "Three-dimensional Finite-Element Stress Analysis of
Laminated Plates Containing a Circular Hole," AFML-TR-76-92 (1976).
1. S. G.

83

Das könnte Ihnen auch gefallen