FractureMechanicsVol. 34,No. 1,pp.

43-54, 1989
Engineering 0013-7944/89 $3.00 + 0.00
Printed in Great Britain. 0 1989 Maxwell Pergamon Macmillan plc.

DELAMINATION OF TAPERED COMPOSITE

STRUCTURES-f
JOHN C. FISHS and SUNG W. LEE4
Center for Rotorcraft Education and Research, Department of Aerospace Engineering,
University of Maryland, College Park, Maryland, U.S.A.

Abstract-The delamination of tapered composite laminates with multiple internal ply drop steps is
investigated. Both experimental testing of glass-epoxy coupon specimens and finite element modeling of
the tapered region are conducted. The average stress concept is applied to out-of-plane stresses from a
thr~-dimensional finite element model. ~lamination failure criteria are then used to predict strength
based on ply failure and interply resin failure. Strength predictions based on interply resin failure using
a stress averaging distance of one ply thickness are found to correlate well with the experimental results.
Delamination in the tapered laminates is due to interlaminar shear failure of the interply resin surrounding
the last ply drop step.

NOMENCLATURE
stress averaging distance
strength tensors
average laminate ply thickness
indices for stress and strength tensors
interply resin layer thickness
through-the-thickness shear strength
local right-handed Cartesian coordinate system with x-axis along the sublaminate length, y-axis along the
sublaminate width and z-axis along the thickness direction of the sublaminate
global right-handed Cartesian coordinate system with X-axis along the laminate length, Y-axis along the
laminate width and Z-axis along the thickness direction of the laminate
distance along laminate length (X-direction) from end of last ply drop step
through-the-thickness compressive strength
through-the-thickness tensile strength
stress tensors
out-of-plane stresses
out-of-plane stresses averaged over one or more finite elements

INTRODUCTION
MANY AIRCRAFT structural components made from composite materials inco~orate taper in their
design. Introducing taper in composite structures may permit structural tailoring which is
unattainable with flat laminates. A composite flexbeam of a helicopter rotor hub is one example
of the application of taper to the design of a composite structure. A flexbeam must be stiff at the
hub end, where large structural displacements are undesirable, yet flexible at the blade end, where
large flapping motions of the blade must be managed. However, plies must be internally dropped
along the length of the laminate to produce the taper. The introduction of these structural
discontinuities within the laminate produces out-of-plane stresses, which may cause delamination
of the composite structure,
Although delamination in composite laminates has been studied extensively in recent years,
most of the work has focused on the free-edge problem[l-71, while little work has been conducted
on delamination resulting from the te~ination of internal plies[8,9]. Reference [8] presented a
failure analysis of eight-ply quasi-isotropic graphite-epoxy laminates containing up to three
dropped 0” plies in a single ply drop step. The analysis assumed a state of generaiized plane
deformation[lO] for which the width-wise normal strain was set to zero, and the finite element

tModified version of a paper presented at the 29th AIAA/ASME/ASCE/AHS Structures, Structural Dynamics and
Materials Conference, Williamsburg, VA, 18-20 April (1988).
SRotorcraft Fellow. Currently Research Engineer, McDonnell Douglas Helicopters, Mesa, Arizona, U.S.A.
§Associate Professor.

43
44 JOHN C. FISH and SUNG W. LEE

method was applied to determine the thr~-dimensional stress state. However, no experimental
analysis was performed to correlate with the analytical results, In ref. [9], sixteen-ply graphite-epoxy
laminates, containing four contiguous plies terminated at the midplane, were studied. Strain
components associated with width-wise stretching and bending were computed from a shell finite
element model of the laminates and then prescribed to a second finite element model which was
based on generalized plane deformation assumptions. The failure analysis underestimated the
experimental failure load by 33%.
If large changes in laminate thickness are desired, the internally dropped plies must be
gradually terminated in a series of steps to attain maximum tensile strength. This study develops
a methodology for the prediction of delamination onset in tapered composite laminates containing
multiple ply-drop steps. The inff uence of the sublaminate stiffnesses above and below the plydrop
steps are investigated. Both ex~~mental testing of glass-epoxy coupon specimens and finite
element modeling of the tapered region are conducted.

EXPERIMENTS

Specimens
The test specimens were made from SP-25OS29 glass-epoxy uni-directional prepreg tape
manufactured by 3M. The tapered coupons were IO-in. long, 1.5-in. wide and were tapered from
28 plies to 16 plies through a taper angle of approximately 5.7”. Three pairs of symmetric two-ply
steps were used to produce the taper (as shown in Fig. 1). Two different tapered laminates were
studied and will be referred to as laminates A and B. Laminate A consisted of a (OJ( f45),), thick
section and a (O,/( i 45)& thin section while laminate B consisted of a ((+45),/O& thick section
and a (( ~45)~~O~)~thin section. Both laminates dropped three pairs of 1-45 plies internally.
In addition to the tapered specimens, flat coupons representing the thin sections of the tapered
lay-ups were tested. The inplane dimensions of the flat specimens were identical to those of the
tapered specimens.
Fiberglass tabs, approximately 2 in. in length, were adhesively bonded to the ends of all test
specimens to prevent grip damage during the tension tests. Finally, to aid in visual observation of
damage development during the tests, black India ink was swabbed on to the edges of the test
specimens.

Set-up and tests

Strain gages were bonded to the center of the thin section of the tapered laminates to measure
the average axial strain in that section of the laminates.
The tensile tests were performed with a 24 kip screw-driven test machine. Load and strain were
continuously monitored using an x-y pfotter with the crosshead rate set to 0.03 infmin. Visual
inspection for damage was conducted by viewing the blackened edges while a lamp was used to

28 PLIES

Fig. 1. Edge view of tapered geometry.

 Delamination of tapered composite structures 45

SH + RI

A---- 245H __3/+ 20H +20H -+20H -‘ii-- 120H- &A7

I
* Y
X
Fig. 2. Geometry of tapered finite element model.

pass light through the translucent material. Damage development was observed and recorded using
a camera with a high magnification lens.

ANALYSIS

Finite element model

A three-dimensional finite element model is used to calculate the stress state in the ply drop
region of the tapered laminate. Due to the symmetry of the problem only one-quarter of the tapered
section needs to be modeled. The inter-ply resin region surrounding the ply drops was modeled,
as well as the laminae and is shown by the shaded area in Fig. 2. The dimensions of the model,
in terms of the ply thickness, H and the resin layer thickness, R, and the global coordinate system
are also shown in Fig. 2. The resin layer thickness between plies was assumed to be ten per cent
of the ply thickness. The model contains 433 eight-node hexahedral elements and 7 six-node
pentahedral elements for a total of 2,916 global degrees-of-freedom. The hexahedral elements are
based on an assumed stress hybrid formulation[l l] and provide more accurate stresses than linear
displacement elements. The pentahedral elements are based on the assumed displacement formu-
lation.
Each of the plies and the resin layer between the plies is modeled through the thickness by
one element. However, the resin rich region ahead of the ply drop steps is modeled by as many
as four elements through the thickness. Each element spans the half-width of the laminate and the
mesh becomes very fine along the length of the laminate in the vicinity of the last ply drop step,
where the maximum interlaminar stresses are located. An edge view of the finite element mesh
surrounding the last ply drop step is shown in Fig. 3. The local coordinate systems for the
sublaminates above and below the ply drops, and the coordinate, X’, representing the distance
along the laminate length from the last ply-drop step, are also shown in the figure.

Fig. 3. Edge view of fine mesh surrounding last ply-drop step.

 JOHN C. FISH and SUNG W. LEE

Table 1. SP-25OS29 alass-enoxv unidirectional taoe

Elastic moduli EL = 7.0 x lo6 psi
E,=2.1 x 106psi
Shear moduli G,, = 0.8 x lo6 psi
Poisson’s ratio vLT= 0.26
Longitudinal
tensile strength 260 ksi
Longitudinal
compressive strength 145 ksi
Transverse
tensile strength 6.2 ksi
Transverse
compressive strength 29 ksi
In-plane shear strength 14 ksi

Each ply is assumed to be homogeneous and orthotropic. The material properties for the
unidirectional prepreg tape, as specified by the manufacturer, are presented in Table 1. The
out-of-plane properties are assumed to be equivalent to the transverse properties of the com-
posite. Furthermore, the out-of-plane shear moduli are assumed to be equal to the in-plane shear
modulus.
The material properties for the resin are presented in Table 2.

Failure criterion
Since delamination is an out-of-plane failure mode, out-of-plane stresses should be the primary
contributors to any delamination strength criteria. Therefore, a modification of the Tsai-Wu failure
criterion[ 121 will be developed as a candidate to predict the delamination strength of composite
structures.
The basic assumption of the Tsai-Wu strength criterion is that there exists a surface in
stress-space which can be expressed in scalar form as
Fia, + Fppj = 1 (1)
where F, and F, (i, j = 1,2,3, . . . , 6) are strength tensors and ci and oj are stress tensors equivalent
to *X.X,CTYY’(T,_,. . . ) oxv, respectively. If delamination is considered an out-of-plane failure mode,
independent of the inplane stress state, the inplane stresses can be uncoupled from the out-of-plane
stresses, reducing the failure criteria to
F3a3+ F4a4+ Fsas+ F,,a: + 2F,,a,a, + 2F,,a,a, + F,a: + 2F,,a,a, + F,,a: = 1. (2)
Furthermore, if strength is considered to be independent of the sign of the shear stress, all
components containing a linear shear stress term may be excluded, leaving
F,a, + F,,a: + F,a: + F,,a: = 1. (3)
Finally, if the two out-of-plane shear strengths are assumed to be equal (Fs5= F,), the failure
criterion can be expressed as
F,a, + F,,a: + F,(a: + a:) = 1. (4)
The strength parameters can be obtained in terms of the out-of-plane strengths, using
experimental tests in tension, compression and shear. Thus, switching to the X, y, z coordinate

Table 2. Resin properties

Elastic modulus 0.57 x 106psi
Shear modulus 0.15 x 106psi
Poisson’s ratio 0.37
Tensile strength 9.0 ksi
 Delaminationof taperedcompositestructures 47

system, the modified Tsai-Wu failure criterion for predicting delamination strength can be
represented by
ZC-ZT
(5)
ZC ZT
where ZT, ZC and S are the through-the-thickness tensile, compressive and shear strengths,
respectively.
In addition to this quadratic delamination failure criterion, the maximum stress failure
criterion will be considered to predict strength when the out-of-plane stress state is dominated by
a single stress component.
For out-of-plane ply failure, the through-the-thickness tensile and compressive strengths are
assumed to be equal to the transverse tensile and compressive strengths. In addition, the
through-the-thickness shear strength is assumed to be equivalent to the inplane shear strength.
For interply resin failure, the resin shear strength is assumed to be equal to the resin tensile
strength and the resin compressive strength is assumed to be equal to the transverse compressive
strength of the composite.
The composite and resin strengths are given in Tables 1 and 2, respectively.

RESULTS AND DISCUSSION

Experimental
Delamination in the tapered laminates was easily detected by a large drop in load, as well as
both audible and visual indications. The onset of delamination was initially unstable, growing
through the width of the laminate across the ply drop steps (as shown in Fig. 4). Subsequent to
the through-the-width growth, the delamination grew in a stable manner along the length of the
laminate, mostly into the thin section. Delamination in the majority of the laminate A specimens
initiated in the region above the last ply drop step as shown in Fig. 5. However, some of the
laminate A specimens delaminated in the region below the last ply drop step. All of the laminate
B specimens delaminated in the region below the Iast ply drop step (as shown in Fig. 6).
Table 3 presents the average laminate axial stresses at delamination onset from the tensile tests
of the tapered specimens. The mean strengths for laminate A and laminate B are 81.6 ksi and
75.6 ksi, respectively.
The experimentally determined flat laminate strengths are given in Table 4. The mean strengths
are 109 ksi and 59.9 ksi for the thin section representatives of laminates A and B, respectively. Thus,
tapering reduces the strength of laminate A by 25% and apparently increases the strength of
laminate B by 26%. The apparent increase in strength may be due to a beneficial stress state created
by the taper or to greater end effects on the thin section of the tapered laminate.

Failure is assumed to occur when the average stresses over some distance, 4, force the failure
criterion above unity. This stress averaging concept [5,6,7, 13,143 is applied to the out-of-plane

Table 3. Tapered laminatestrengths(ksi) Table 4. Flat laminatestrengths(ksi1

Specimen LaminateA LaminateB Specimen LaminateA LaminateB
1 80.6 74.3 1 105 59.7
2 87.1 84.7 2 110 57.2
3 71.1 68.1 3 I12 63.2
4 83.9 76.2 4 107 61.1
5 80.2 72.7 5 - 58.3
6 82.8 73.9 Mean 109 59.9
7 85.7 79.1
Mean 81.6 75.6
48 JOHN C. FISH and SUNG W. LEE

stresses obtained from the finite element model. The average out-of-plane stresses due to the
tapered geometry may be expressed as

(6)

(7)

(8)

The smallest averaging distance considered is one-half of a ply thickness (do = 0.5H) which
is the dimension, in the x-direction, of the smallest finite element. Stress averaging distances of up
to 2H are investigated.
Throughout the remainder of this section, out-of-plane stresses in the plies adjacent to the ply
drop steps and in the resin surrounding the ply drops are examined for a laminate axial stress of
1 ksi. The failure criterion are then applied to predict the strength of laminates A and B. The
modified Tsai-Wu failure criterion is used to predict strength, unless mentioned otherwise.
The out-of-plane stresses in the adjacent 0” ply above the ply drops in laminate A are plotted
in Fig. 7 with respect to the distance in the laminate length direction from the last ply-drop step,
X’. The stress state is dominated by the shear stress, oXz,which is at a nearly constant value of
approximately -0.2 ksi for much of the tapered region. The out-of-plane normal stress, crrr, is
tensile throughout, but nearly goes to zero at the end of the last ply-drop step. The magnitude of
the shear stress, a,,>, is negligible compared to the other two stresses.
Figure 8 shows the out-of-plane stresses in the adjacent +45” ply below the ply drops in
laminate A. It is apparent that the magnitudes of the stresses, all of which remain below 0.05 ksi,
are much lower than those in the 0” ply above the ply drops. Therefore, out-of-plane ply failure
of laminate A is more likely to occur in the 0” ply above the ply drops.
The out-of-plane stresses for laminate B in the -45” and 0” plies above and below the ply
drops are shown in Figs 9 and 10, respectively. The stresses follow the same trends as those for
laminate A, attaining larger magnitudes in the ply above the ply drops (-45”). However, in general
the magnitudes of the stresses are less than those for laminate A.
The laminate A strength predictions for out-of-plane failure of the plies above and below the
ply-drop steps are shown in Fig. 11 for different stress averaging distances. The mean of the
experimentally determined strengths is shown by the dashed line in the figure. The laminate strength
is underestimated by approximately 40%, if strength is based on failure of the ply below and ply
drops, and overestimated by as much as 124%, if strength is based on failure of the ply above the
ply drops. Since the out-of-plane stress distributions (with respect to X’) for these plies are nearly
constant over a large region, the ply failure predictions are relatively insensitive to the stress
averaging distances investigated.

-.3J.““....‘.““‘...r....l....r....J
25 20 15 10 5 0 -5 -10
-.3’..‘.‘..“‘..“‘..“‘““‘“.““‘.)
25 20 15 10 5 0 -5 -1
XI/H XI/H
Fig. 7. Laminate A ply stresses above last ply-drop step. Fig. 8. Laminate A ply stresses below last ply-drop step.
 Delamination of tapered composite structures 49

Fig. 4. Front view of across-the-width delamination.

 Delamination of tapered composite structures 51

r:‘~
25 20 15 10
X)‘H ’
0 -5 -10
~~2~ X1/H 5
0

Fig. 9. Laminate B ply stresses above last ply-drop step. Fig. 10. Laminate B ply stresses beiow iast ply-drop step.

The laminate B strength, predictions for out-of-plane ply failure are shown in Fig. 12. The
laminate strength is overestimated by no less than 39% and 71% for ply failure above and below
the ply drops, respectively. In addition, the lowest ply failure prediction for laminate B is greater
than the lowest ply failure prediction for laminate A, implying that laminate B should be stronger
than laminate A.
Thus, predicting strength based on out-of-plane failure of the plies provides poor correlation
with the experimental data, not only yielding large percentage errors but incorrectly ranking the
strengths of the laminates as well.
Next, strength based on failure of the interply resin s~rounding the ply drops is investigated.
The interlaminar stresses in the resin layer above the ply drops and below the ply drops are shown
in Figs 13 and 14, respectively, for laminate A. Above the ply drops (Fig. 131, a,, is highly
compressive and becomes tensile just ahead of the last ply drop step. The interlaminar shear stress,
u XZYattains large magnitudes near the end of the ply drop step. However, it quickly decreases to
small values throughout most of the region ahead of the drops, until the root of the taper
(X//H = 20) is approached, where it increases again. As with the ply stresses, o,,~,is negligibly small.
Below the ply drops (Fig. 14), crzzis not as highly compressive (compared to above the ply drops)
and attains larger tensile values ahead of the drops. In addition, the sign of cXzbecomes reversed,
as expected, but the magnitude is nearly the same as above the drops.
The resin stresses in laminate B above and below the ply drops are shown in Figs 15 and 16,
respectively. The magnitudes of the stresses above the drops are considerably less than the
ma~tudes of the stresses below the drops. Below the drops large values of o,, are attained at the
end of the last ply drop step.

200* 200 -
-*---- EXPERIMENTAL MEAN ------
EXPERIMENTM. MEAN
.0 . pLy#@/E mp ....o..., PLY ABOVE STEP
-O- PLY BELOW 9EP 4 PLY BELOW STEP

01
I I I
3 1 1.5 2

do/H 4,/H
Fig. 11. Laminate A ply failure strength predictions using Fig. 12. Laminate B ply failure strength predictions using
modi!ied Tsai-Wu criterion. modified Tsa-Wu criterion.
 52 JOHN C. FISH and SUNG W. LEE

-.151"""""".""'."'."""""'~ _,,5t,, , . I ,,,, I I, i ,L #I.. I I,,,,,... , .,,“&J

25 20 15 10 ( 5 0 -5 -10 25 20 15 10, 5 0 -5 -10

X/H X/H

Fig. 13. Laminate A interply resin stresses above last Fig. 14. Laminate A interply resin stresses below last ply-
ply-drop step. drop step.

The laminate A strength predictions based on interply resin faihtre above and below the
ply-drop steps are shown in Fig. I? for different stress averaging distances. For averaging distances
of fess than about one ply thickness (C&/H< 11,faiiure is predicted to occur in the interply resin
above the ply drops. For Iarger averaging distances, failure is predicted to occuf below the ply
drops. However, the predictions are closer to the experimentally determined strength for the smaller
averaging distances considered. Therefore, failure of laminate A should occur in the interply resin
above the ply-drop steps.
The laminate B strength predictions for interply resin failure are shown in Fig. 18. Failure of
the resin above the ply drops greatly overestimates the strength of the laminate for all averaging
distances considered. However, failure of the interply resin below the ply drops provides an
excellent prediction of laminate strength for a stress averaging distance of 1.5 ply thicknesses.
In order to develop a valid approach to predicting strength using the average stress concept,
the same stress averaging distance must be applicable for both laminates, The strength predictions
for the laminates are compared in Fig. 19, For a stress averaging distance of ~2423= 0.5, both
predictions underestimate the laminate strengths. For a stress averaging distance of C&/H= 1.5,
both predictions overestimate the laminate strengths. With a stress averaging distance of one ply
thickness, the strength of laminate A is overestimated by approximately 8% and the strength of
laminate B is underestimated by about the same amount. Therefore, a stress averaging distance
of one ply thickness appears to be appropriate for predicting laminate strength.
Since the interlaminar shear stress, c,,, dominates the interlaminar stress state in the interply
resin, the maximum shear stress failure criterion will be considered as an additional means of
predicting the strength of these laminates.

Fig. 15. Laminate B interply resin stresses above last piy- Fig. 15, Laminate I3 interply resin strasses beiow Iast piy
drop step. drop step.
 Delamination of tapered composite structures 53

250
------ MPERIMENML ~pfj
,(,.....”
* ABM sTEp
,.,. .A’
,,...,..”
-n-BELowmp /......’
h
‘3 200 -
,,,,,,,,,, ~ ,,,,,,,,,,,,...........~..~....
..........,....O.“““”
5
...
z ,(,.....”
: 150 - ,,,,,,.,,....../.“”
0
&

a
3 loo-

oi
.5 1
do/H
Fig. 17. Laminate A interply resin failure strength predic-
1.5 2
50 ’
.5

Fig. 18. Laminate B interply resin failure strength predic-

I
1
do/H
I
1.5

tions using modified Tsai-Wu criterion. tions using modified Tsai-Wu criterion.

The laminate strength predictions using the maximum shear stress failure criterion are shown
in Fig. 20. With a stress averaging distance of one ply thickness, the strengths of laminates A and
B are predicted to be 83.8 ksi and 76.0 ksi, respectively. This yields errors of +2.7% and +0.5%
for laminates A and B, respectively. Thus, only the interlaminar shear stress, aXr,need be considered
to accurately predict the strength of these tapered laminates.
In addition, it is interesting to note that the laminate A strength prediction for interply resin
failure below the ply drops is close to the strength prediction for failure above the ply drops
(91.2 ksi vs 83.8 ksi). This implies that failure may occur below the ply drops in laminate A if local
imperfections, such as voids, are larger or more numerous in that region, compared to above the
ply drops. However, the laminate B strength prediction for interply resin failure below the ply drops
is much less than the strength prediction for failure above the ply drops (76.0 ksi vs 159 ksi). This
means that failure of laminate B is not likely to occur above the ply drops. Thus, the results
correlate with the experimental observation of failure at both locations in laminate A and only
below the ply drops in laminate B.

SUMMARY AND CONCLUSIONS

The strength of tapered composite structures may be accurately predicted if delamination is
based on failure of the interply resin surrounding the internal ply drops. The maximum shear stress
failure criterion, using an interlaminar stress averaging distance of one ply thickness, provides
consistent and accurate delamination onset predictions for the laminates investigated. Experimental
observations suggest that the analysis correctly predicted the failure locations for the two laminates.
The failure of the tapered laminates is due to the interlaminar shear stress, axr, and occurs at the
120 b 120 -
-.- A EXP. MEAN
------- B EXP. MEAN
~..+~-
=-I A ABOVE
“B % y STEP,,,,,,,,,,,,,,,,_,,,,.,,,,,,,,,,,,,,,,,,,,,_,,.,,
o” (..,..................
......,...,..,..““““” ...+- A ABOVE STEP
h -a- B BELOW STEP h -Cl- B BELOW STEP
‘B ‘I
*loo - ~100 -

ii
ki
,.,...”
ti ,,/’ ./
.... I

I I I I
60 60
.5 1 1.5 2 .5 1 1.5 2
do/H do/H
Fig. 19. Comparison of interply resin failure strength pre- Fig. 20. Comparison of interply resin failure strength pre-
dictions using modified Tsai-Wu criterion. dictions using maximum shear stress criterion.
54 JOHN C. FISH and SUNG W. LEE

last ply-drop step. If the stacking sequence is arranged such that the stiffest (extensionally) plies
lie above the ply drop steps, more of the load carried by the dropped plies can be transferred to
the sublaminate above the dropped plies, reducing the peak interlaminar shear stress in the
laminates.

Acknowledgements-This work was sponsored by the U.S. Army Research Office (Contract No. DAAG-29-83K002). The
program monitor was Dr Robert Singleton. The authors would like to thank Dr T. Kevin O’Brien of the U.S. Army
Research and Technology Laboratories (AVSCOM), NASA Langley Research Center and Dr Wen S. Chan of Bell
Helicopter Textron for their helpful discussions during the initial phases of this research. The computer funds were provided
by the Computer Science Center of the University of Maryland.

REFERENCES
HI A. H. Puppo and H. A. Evensen, Interlaminar shear in laminated composites under generalized plane stress. J. Compos.
Mater. 4, 204-220 (April 1970).
PI R. Byron pipes and N. J. Pagano, Interlaminar stresses in composite laminates under uniform axial extension.
J. Compos. Mater. 4, 538-548 (October 1970).
131 A. S. D. Wang and F. W. Crossman, Some new results on edge effect in symmetric composite laminates. J. Compos.
Mater. 11, 92-107 (January 1977).
141 T. K. O’Brien, Characterization of delamination onset and growth in a composite laminate, in Damage in Composite
Materials, ASTM STP 775, pp. 140-167. American Society for Testing and Materials (1982).
151 R. Y. Kim and S. R. Soni, Experimental and analytical studies on the onset of delamination in laminated composites.
J. Compos. Mater. 18, 70-80 (January 1984).
161 Som R. Soni and Ran Y Kim, Delamination of composite laminates stimulated by interlaminar shear, in Composite
Materials: Testing and Design (Seventh Conference), ASTM STP 893, pp. 286307. American Society for Testing and
Materials, Philadelphia (1986).
Paul A. Lagace, Delamination in composites: is toughness the key? SAMPE J. 22,53360 (November/December 1986).
;:; B. L. Kemp and E. R. Johnson, Response and failure analysis of a graphite/epoxy laminate containing terminating
internal plies. Proc. AIAAIASMEIASCEIAHS 26th Structures, Structural Dynamics and Materials Conf. Part I,
Orlando, Florida, AIAA Paper No. 85-0608, 1517 April 1985, pp. 13-24.
[91 J. M. Curry, E. R. Johnson and J. H. Stames, Jr, Effect of dropped plies on the strength of graphite-epoxy laminates.
Proc. AIAAIASMEIASCEIAHSIASC 29th Structures, Structural Dynamics and Materials Conference, Part I,
Monterey, California, AIAA Paper No. 87-0874, 6-8 April 1987, pp. 737-747.
1101S. G. Lekhnitskii, Theory of Elasticity of an Anisotropic Body. MIR Publishers, Moscow (1981). English translation
of the revised 1977 Russian edition.
M. J. Loikkanen and B. M. Irons, An 8-node brick finite element. Int. J. numer. Meth. Engng 20, 523-528 (1984).
;t:; S. W. Tsai and E. M. Wu, A general theory of strength for anisotropic materials. J. Compos. Mater. 5, 58-80 (1971).
1131J. M. Whitney and R. J. Nuismer, Stress fracture criteria for laminated composites containing stress concentrations.
J. Compos. Mater. 8, 253-265 (July 1974).
1141 R. J. Nuismer and J. M. Whitney, Uniaxial failure of composite laminates containing stress concentrations in Fracture
Mechanics of Composites; ASTM STP 593, 117-142. American Society for Testing and Materials (1975).

(Received 7 June 1988)