Vous êtes sur la page 1sur 12

Composite Structures 28 (1994) 61-72

' ~ ; ' "': L

Impact response of laminated composite plates:


Prediction and verification
H. V. Lakshminarayana, R. Boukhili & R. Gauvin
Department of Mechanical Engineering, Ecole Polytechnique de Montreal, Canada
Methods and procedures for predicting the impact response of laminated composite plates using a commercial finite element system are described. Results of
element evaluation, procedure verification and a correlation study are presented and discussed. The need for a hybrid experimental-numerical approach
and combined geometric and material nonlinear finite element analysis is identified. A methodology for the prediction of delamination (onset and growth) is
outlined.

1 INTRODUCTION

while few studies have included nonlinear effects.


Almost all of them make use of special-purpose
programs. Such programs, valuable in their own
right, are difficult to access, are not well documented and hence are difficult to use by the
practitioners. (2) Assessment of convergence and
accuracy of the deflection history, stress distributions and damage zones calculated by the Finite
Element Method (FEM) have not been given
explicit attention. (3) Experimental investigations
that document the response recorded during the
impact test are rather limited. The majority of
them provide a qualitative/quantitative description of the accumulated damage but not its
growth. This information is vital to perform correlation studies. (4) Few investigators have considered full-scale components. The majority of the
reported studies use generic structures such as
beams, circular plates, square plates and cylindrical panels. There is a real danger in extrapolating
conclusions drawn from studies on test specimens
to real life components.
It is more appropriate to use commercial FEM
systems. They are widely distributed, well documented and user friendly. However, there is a
need to verify the accuracy of material models,
finite elements and analysis procedures in such
systems before using them for the intended application, namely, prediction of the impact response
of composite plates. The predictability issue itself
demands a correlation study. This, in fact, is the
aim and scope of the present study.
A brief description of the specific methods and
procedures used is given in the next section. In

Laminated composite plates are easily damaged


by impacts, especially those normal to the plane of
the laminate. Systematic study of such problems
can be divided into three distinct areas: structural
mechanics, damage mechanics and residual
strength prediction. A study in that order will
provide a mechanistic basis for both design and
assessment of damage tolerance.
Reliable and accurate prediction of the impact
response of multilayered anisotropic plates covering a wide range of parameters is the focus of this
study. Impact response means contact force history, deformation history, surface strain history,
stress distribution across the laminate thickness
(including interlaminar stresses), nature and
extent of damage and stiffness and strength loss
associated with that damage. Parameters significantly influencing the impact response include
impact velocity/energy, material system, ply
orientation and stacking sequence, plate geometry
(shape, size and thickness), wall construction
(solid laminate, integrally stiffened, and sandwich), support conditions, pre-stress state and
initial curvature (curved panels).
A review of previous work is available in a
monograph L2 and two review papers, 3,4 which
provide a background to the present study. A
critical assessment of this vast literature, with
particular reference to the focus of the present
study, can be summarized as follows. (1) Finite
element analysis of linear and transient structural
behaviour has been the focus of intense research,
61

Composite Structures 0263-8223/94/S07.00 1994 Elsevier Science Limited, England. Printed in Great Britain

62

H. V. Lakshminarayana, R. Boukhili, R. Gauvin

succeeding sections, results of element evaluation,


procedure verification and correlation study are
presented and discussed. The presentation concludes by identifying directions for further work.

2 M E T H O D S AND P R O C E D U R E S

Numerical results for this study were generated


using ABAQUS -- a general-purpose finite element code with emphasis on nonlinear applications. ~ This program is capable of modelling
multilayered anisotropic materials. It provides
elements suitable for dynamic analysis of composite plates and shells taking into account bendingmembrane coupling and transverse shear
deformation effects. Among these, the S8R, an
isoparametric quadrilateral plate/shell element, is
employed in the present study. This element has
eight nodes and six engineering degrees of freedom at each node. The user can specify within
each element an arbitrary number of layers, each
with its own thickness, ply orientation and orthotropic elastic properties. The formulation uses a
value of 5/6 for the shear correction factors as
default. However, the user has the option to use
any other value through independent input of
transverse shear stiffness. The element output
includes membrane stress resultants, bending
stress resultants and transverse shear-stress
resultants, either at the nodes or at the four integration points. The user can also request ply-byply stresses at the integration points and at a
maximum of 3 section points within each ply. It
should be noted that the formulation of S8R does
not ensure the continuity of interlaminar normal
and shear stresses at the interfaces between plies.
ABAQUS provides two procedures for calculating the response of structures subjected to
impulsive loads. They are modal analysis and
dynamic analysis by direct implicit integration of
the equations of motion. The modal analysis is
limited to linear transient behaviour. Its convergence and accuracy is dependent on the number of
natural modes considered in the analysis. The
pulse shape and the duration (t~) for which the
contact force acts (in comparison with the period
T~ of the fundamental mode of free vibration of
the structure) strongly influence the response calculated by the modal method. For complex pulse
shapes and for (t,,/Ti)'~ 1, a very large number of
modes is required for convergence. Unfortunately, accurate determination of the eigenvalues
and eigenfunctions associated with higher modes

places very heavy demands on computational


resources by the FEM, Incidentally, eigenvalue
extraction in ABAQUS is done using the subspace iteration method. The transient response
calculation by the direct integration of the equations of motion is applicable to linear as well as
nonlinear structural behaviour. When applying
this step-by-step method, the time step At should
be selected with caution, because a system of nonlinear algebraic equations must be solved at each
time increment. This is done in ABAQUS iteratively by using Newton's method or, if preferred,
the quasi-Newton's method. This time stepping,
nonlinear equation-solving procedure is computationally expensive. The principal advantage of this
procedure is that it is unconditionally stable,
which means that there is no mathematical limit
on the size of the time increment that can be used.
In practice, At should be small enough to adequately define the history of excitation, its value
being chosen on the basis of the shortest period
which corresponds to the highest natural mode
likely to contribute to the response.
A recommended methodology(' for determining the contact force history is to measure the
local contact stiffness of the composite plate in
static tests and use this in conjunction with a finite
element model. This approach, however, could
not be considered truly predictive since it requires
fabrication of the plate and static indentation tests
for every impactor under consideration. Alternatively, it can be measured during impact tests using
suitable instrumentation and data acquisition system. This approach is applicable to the nonlinear
response also.
The nature and extent of impact-induced
damage is estimated by first calculating the
dyamic stresses and their spatial distribution in
the laminate and incorporating these in appropriate failure criteria. Multiple matrix cracks, delaminations and fiber breaks are the failure modes
observed after impact tests. These failure modes
and complex interactions between them complicate the prediction of damage. The tensor polynomial failure criterion proposed by Tsai and Wu 7
is employed in the present study to calculate the
failure index (FI) given by
F I = k ' l o , + F2o~ + F~o~ + F11o~ + F2:o!
+ F(~,o~ + _F~2o 102

where k~j are the strength tensors; 7 o~, o 2 are the


lamina stresses in the fiber direction and transverse direction respectively; and o~, is the in-plane

Impact responseof laminated compositeplates


shear stress. Loci of points at which FI = 1 define
the damage zone. The failure modes are identified
using the maximum stress criteria. 7
The stiffness and strength loss due to impactinduced damage are not at present truly predictable. A suggested approach involves flexure tests. ~
Evaluation of the accuracy of the S8R element
for the analysis of composite plates in general is
presented in the next section. We identify three
distinct procedures: (1) modal analysis (procedure
# 1); (2) linear and transient response analysis
(TRA) (procedure #2); and (3) nonlinear and
transient response analysis (NLTRA) (procedure
# 3). The convergence and accuracy of each one
of these is verified in a section entitled procedure
verification. Finally, a critical assessment of their
predictability with particular reference to the
impact response of laminated composite plates is
presented in Section 5.

3 ELEMENT EVALUATION
Application of the full set of test problems proposed in Ref. 9 to the S8R element is not pre-

I I

6
_

W=
5111111

Ill Ill

W(O,O) E,h~ 10s


Pa"
I

4 PROCEDURE VERIFICATION

CPT Solution

-1

10
Fig. 1.

sented here. The results showed that a particular


problem, a homogeneous, anisotropic, clamped,
square plate under uniform pressure, provided an
intensive measure of element performance. This
problem, shown as the insert in Fig. 1, was chosen
to evaluate the combined effect of material anisotropy and shear deformation on the accuracy of
the S8R element. Numerical results were obtained
for a unidirectional laminate, made of a highmodulus graphite/epoxy composite (EI/E 2= 40,
E2 = 5.17 GPa (0.75 x 106 psi), Gi2--- 3.10 GPa
(0"45 x 10 ~' psi), G23/GI2=0"8, vl2=0.25 ), for a
ply orientation of 45 and for various values of the
thickness ratio (h/a). Computed results are compared with the converged solutions given in Ref.
9. The agreement is very good for displacements
as well as stress resultants provided (h/a)> 0.01.
The effect of h/a on the predicted central deflection is shown in Fig. 1, which also has the classical
plate theory (CPT) solution for comparison. For a
given thickness ratio, the inaccuracy associated
with the omission of transverse shear deformation
effects is obtained from this figure. Obviously, the
accuracy of the S8R element deteriorates for (hi
a)<0"01, indicating that the formulation is not
free from the so-called shear-locking phenomenon. There is scope to improve the performance of the SSR element for applications to thin
and highly anisotropic situations. An approach to
accomplish this is demonstrated in Ref. 10.

II

IIIIII

63

-2

10
h/a

-3

10

Effect of (h/a) on the central deflection predicted


using the S8R element (8 x 8 mesh, whole plate).

The chosen elements and procedures should be


capable of providing converged and accurate
solutions to standard test problems for which
alternative solutions are available for comparison.
The specific problem used in this study is a crossply laminated, graphite/epoxy composite, square
plate with all edges simply supported and subjected to impact at its center, which was investigated by Sun and Chen. ~
The geometry, ply orientation/stacking sequence, finite element mesh and contact force
history (corresponding to impact by a 12.7 mm
diameter steel ball at velocity V0 -- 3 m/s) used in
the computations are shown in Fig. 2. Representative displacement time histories calculated using
procedure # 2 (linear and transient response by
direct implicit integration) are given in Fig. 3.
These results are in excellent agreement with
those reported in Ref. 11, in spite of the larger
time step (6t=0.05 ms) used. No significant

64

H. V. Lakshminarayana, R. Boukhili, R. Gauvin

P(t)

w(t)
P(t)

Pm a x

h = 5.1 m m
a = 203.2 mm

0
90

Pr.~x= 27.4 Kg
to = 0.2 ms

I
I
I

90'

0
0

',\

t 1 = 0.25 t o

90

0
90

t
t

Contact Force History

Ply Orientation/Stacking S e q u e n c e

/;

j/-

/
i

Finite Element Model (Quarter Plate)

Fig. 2.

Test problem for procedure verification.

improvement is noticed using a finer mesh and a


smaller time step, indicating that a converged and
accurate solution has been obtained. Figure 3 also
indicates that two pulse shapes (triangular and
half sine-wave) produce almost identical results.
Since the half sine-wave pulse can be represented
by an analytical expression, it is ideally suited to
create benchmarks for the impact response of
composites.
For the same problem, Fig. 4 shows displacement-time histories calculated using procedure
# 1 (modal analysis). There is no sign of convergence as the number of modes used is increased
from 10 to 15. These results do not show any
comparison with the reference solution in Ref. 11.
Not only is the displacement history different, the
maximum amplitudes differ by one order of

magnitude. This observation is substantiated by


the findings of a recent round robin study.~2 It is
noted that whenever the duration of impact is very
small (in comparison with the period of the fundamental mode of free vibration of the target), the
number of modes to be considered for convergence may be so large as to be computationally
inefficient. It appears that reliable prediction of
the impact response is outside the domain of the
modal analysis.
For the problem specified in Fig. 2, a comparison of the displacement histories calculated
using procedure # 3 (nonlinear and transient
response by direct implicit integration) and procedure # 2 is made in Fig. 5. The two solutions are
in fact identical because at this low load level the
resulting deflections are very small (in comparison

Impact response of laminated composite plates


0.24

0.22

0.22

0.2
0.18

0.2

0.16

0.18

0.14

0.16

0.12

"- 0.14

o.1

"*"= 0.12
0
0.1
v0

0.08
I~1 0.06
5
G)

II

0.04

I~:

0.02

c-

0GI -0,02

Pmax= 27.4 Kg
to= 0.2 ms
t, = 0.25 t o

(3.
.~_
-o.04

-o.os

006

0.08

0.04

E 0.02

o,

65

P(t

oo -0.02

=~

-0.04

\,

-0.08
-0.1
-0.12
0

11

0.5

; f/

-o ~
~_o

' i
(;t~

0.5

-0.06

-0,08

I, t l

!1 Prnax= 27.4 Kg ' 1


1, to= 0"2,ms

-0.1

1.5

1.5

Time t(miliseconds)

Time t(miliseconds)
Fig. 3.

\ fltl=

Central displacement history calculated using procedure # 2.

1.8

1.8

N u m b e r of m o d e s = 1 0

1.6

1.6

1.4

1.4

1.2

1.2
t-

d o.8

"~. o.e
o
<:5 o.6

0.4

II

0.2

15

I~

o,

~
e-

!P(t)

0.6
0.4

o.2
0

Pr..x
P(t)

e~

e~

w,t,

._~
a

P(t) = P S i n

(nt/to)

Pmax= 27.4 Kg

to

t = 0.2 ms
o

Time t(miliseconds)
Fig. 4.

Time t(miliseconds)

Central displacement history calculated using procedure # 1.

66

H. V. Lakshminarayana, R. Boukhili, R. Gauvin


0.24

0.24

0.22

Procedure # 2
6t = 0.05 ms

0.22
0.2
0.18

0.18

0.16
e.-

0.16

0.14

0.14

C~ 0.12
0
0.1

o.12

II

0.1

0.08

0.08

15
.. 006
E

0.04

0.02

Procedure # 3
6t = 0.05 ms

0.2

P(t)

to= 0.2 ms

_
g 0.04

Pmax

0.02

..........

~.

~.
0
.oo
E3 -o.o2

o?,,J,,2t

-0.04
-0.06
-0.08

-0.1
0

0.5

Time
Fig. 5.

"o

1.5

t(miliseconds)

-0.02
P(t) = Pmeeqin (rrt/t o)

-0.04 --

Pm~ = 27.4 Kg
VO 3 m/s
O= 0.2 ms

-0.06 I
-0.08

-0.1 J
0

0.5

1.5

T i m e t(miliseconds)

Comparison of displacement-time histories calculated by procedures # 2 and # 3 (impact velocity = 3 m/s).

with the thickness h) and hence nonlinear effects


do not show up.
Impact velocity is a very important parameter
controlling the response. Figures 6 and 7 provide a
comparison of displacement time histories calculated using procedure # 2 and procedure # 3 at
V~=10 m/s and V0=30 m/s, respectively.
Obviously, the responses predicted by the two
procedures do not agree with one another. Both
the displacement-time history as well as the maximum amplitudes predicted differ rather significantly. Basically, the load levels are such that
resulting deflections are of the order of the plate
thickness and therefore procedure # 3, which
includes nonlinear effects, is more appropriate.
Unfortunately, the accuracy of the numerical solutions presented in Figs 6 and 7 could not be verified due to the nonavailability of reference
solutions.
Procedure # 3 is therefore more appropriate
for numerical solutions of impact tests. In fact, for
complex structures, this procedure may provide
more cost-effective information than experimentation. However, there is a need to further validate
the predictability of this procedure for the impact

response. This aspect is addressed in the next


section.
5 CORRELATION STUDY
The predictability aspect of the procedures verified in the previous section, with particular reference to the impact response of composite plates is
assessed using a bench mark. Bench marks are
fully specified and standard problems, which
resemble instances found in practical applications, and for which reference solutions have been
obtained using both analytical/numerical and
experimental methods.
The bench mark chosen for this study was
created by Aggour and Sun. j3 It consists of a
cross-ply laminated, E-glass/epoxy composite
(Ej=38"6 GPa (5"6x106 psi), E2=I0.34 GPa
(1.5106 psi), G2=4"14 GPa (0.6x106 psi),
v,2 =0.25) square plate with all edges clamped.
The geometric parameters, ply orientation/stacking sequence, and the finite element discretization
used in the computations are shown in Fig. 8. The
contact force history corresponds to an impact by

Impact response of laminated composite plates

67

1
0.9

0.9 f

Procedure #2
6t = 0.05 ms

0.8

0.7

0.7

0.6

0.6

0.5

~,
~d

t-

0.5

0.4

0.4

0.3

C) 0.3
O
0.2

0.2

II 0.1

0.1
0

e(9

-0.1

-0.2

E -0.1
(9 -0.2

~.

-0.3

~ . -0.3

i'~

-(I.4

u)
~,~ -0.4

-0.5

-0.5

-0.6

-0.6

-0.7

-0.8

Procedure # 3
5t = 0.05 ms

0.8

0.5

~W(t) ~
Ih
P(t)= PmaxSin(~t/t 0)
V 0 = 10 m / s
P m ~ = 9 1 . 2 Kg

-0.7

to

~-~

1.5

-0.8
0

t 0= 0.2

0.5

Time t(miliseconds)
Fig. 6.

ms

1.5

Time t(miliseconds)

Comparison of displacement-time histories calculated by procedures # 2 and # 3 (impact velocity = 10 m/s).

Procedure #2 : LTRA

Procedure # 3 : NLTRA

1.8

6t = 0.05 ms

1.6
1.4
1.2

J::

1
0.8

c:) 0.6

o
0

II

O.4

II 0.2

,4..I

t5

0[

(9

P(t)

o,

c
(9 -0.2

(9 -0.4
o
0 -0,6

P(t) = Pm=Sin (nl/t o )


Prr,= = 364.8 Kg

-2

-1.2

Vo= 30 m/s
t -- 0.2 ms

-1.4

Vo= 30 m/s
Pro= = 364.8 Kg
to= 0.2 ms

-3
0

0.5

Time t(miliseconds)
Fig. 7.

a-~ -0.8
-1

1.5

-1.6

0.5

1.5

Time t(miliseconds)

Comparison of displacement-time histories calculated using procedures # 2 and # 3 (impact velocity = 30 m/s).

68

H.V. Lakshminarayana, R. Boukhili, R. Gauvin


z [ P(t)

--.

a
P(t)

P ,x ....

\
/

'\

'\
',,

ho o

/iiiiiiiiiiiiiiii

9o!

I!iii!i!~iiii!iiii!iiiiiiliiiiiiiiiii!iiiiii!i
ilili
~i!i!i!i!~!i!iiiii~ii~iililiiililili9o
i!iiiiiJo

\\\

~_t

t
Contact Force Histor~
P(t) = Pm~xSin(nUt o )
P~x = 310.1 Kg
to= 0.25 ms

Ply Orientation/Stacking Sequence


a = 139.7 mm
h=4.1

mm
I

Finite Element Model (Quater Plate).

Fig. 8. Testproblemfor correlation study.

a steel cylinder (diameter 9.5 mm and length 25.4


mm) at a velocity E, = 22.6 m/s.
The calculated central displacement history
using procedure # 3 is presented in Fig. 9. Results
using procedure # 2 and test data taken from Ref.
13 are also included in the same figure to enable a
three-way correlation. Results obtained using
linear finite element analysis by Aggour and Sun ~3
are in close agreement with those obtained in the
present study using procedure # 2. Furthermore,
the responses calculated using procedure # 3 and
procedure # 2 are identical, indicating that nonlinear effects are negligible. It is gratifying to note
that computed results closely follow the central
deflection measured during an impact test for
t-<0"5t0, where to is the duration of impact. How-

ever, for higher values of t there is considerable


difference between the computed and measured
results. Therefore, to achieve a satisfactory correlation between the two sets of results over a larger
duration, further refinement of the procedures
used appears to be essential.
Basically, the effects of impact-induced damage
has to be included in the analysis. More specifically, a combined geometric and material nonlinear dynamic finite element analysis is
indispensable for this purpose. However, this is
more easily said than actually performed, because
a priori knowledge of the nature and extent of
impact-induced damage and accurate constitutive
modelling of failed plies and damaged laminates
will be required after each time step. At present,

Impact response of laminated compositeplates

69

1
Procedure # 2

0.9

6t = 0.025 ms

0.8

0.6

Testdata

0.4

~,

0.4

0.3

0.3

oc

o.1

~ -0.1

0{

~ -0.1

~ -0.2

~ -0.2

-0.3

-o.g

i~ -0.4

-0.5

-0.5

-0.6

-0.6

-0.7

-0.7

-0.8

-0.3

-0.8
0.5

l
0

Time t(miliseconds)
Fig. 9.

Comparison of measured and predicted central displacement histories.

1.8

Procedure # 2

l'k
1.6

1.4

1.61.4-

VO= 40.0 m/s


[]

Vo-- 2226 m/s

...

0.6

Vo= 40.0 m/s

Vo= 30.0 m/s

0.8

Vo= 226 m/s

0,6
oo-

ji

0.4

/'//

0.2

II
-0.2

.,..,
e.- -0.2 1

(1) -0.4

E -o.4 ~

-0.6

\'~ "t~r~
\\

(:~ -0.6

-o8

~.

-0.8

-1

i~

-1

-1.2

-1.2

-1.4

~N

-1.6

-1.4
-1.6

-1.8
0.5

Time t(miliseconds)
Fig. 10.

Procedure # 3

1.2

Vo= 30.0 m/s

0.8

0
~

0.5

Time t(miliseconds)

1.8

0.2

0.1

c5 0.4

Studyl

Testdata

0.5

0.2

Prese~

0.7
0.6

0.5

c~

#3

6t = 0.05 ms

0.8

Present Study

0.7

Procedure

0.9

-1.8
0.5

Time t(miliseconds)

Comparison of predicted displacement-time histories by linear and nonlinear analysis procedures.

70

H. V. Lakshminarayana, R. Boukhili, R. Gauvin


1.6

Procedure #2
V = 40 m/s

1.4

= 0~25 ms

t=

Procedure #3
V = 40 m/s

14
,2

o2oms

I t = 0z15 ms

x::

[t = o

, ~f
i

ms/

/t = 0.10 msl

"~'I~ 0.8!
..

0.6 I

.6

0.4

0,4

0.2
O"I

-02

0.1

0.2

0.3

0.4

05

x/a
Fig. 11.

-02

0.1

0.2

0.3

0.4

0.5

x/a

Comparisonof predicted displacementdistributionsby linear and nonlinearanalysisprocedures.

the former cannot be truly calculated while information on the latter is not available. The progressive failure finite element analysis outlined is in
fact a topic for further research.~4
Reliable prediction of displacement history as a
function of impact velocity (Fig. 10) constitutes
only a fraction of the impact response story. Spatial distribution of displacement (Fig. 11 ), stress
distribution across the laminate thickness (Fig. 12)
and damage growth (Fig. 13) are in fact even more
important. The F E M in general and A B A Q U S in
particular is able to provide such results. Indeed a
major problem is to find experimental techniques
that can provide data that are equally detailed to
validate the results presented here.

6 CONCLUDING REMARKS
A hybrid experimental-numerical approach is
necessary to predict the impact response of laminated composite plates covering a wide range of
the parameters involved. Experimental determination of contact force history is essential if the
structural behaviour is nonlinear. Nondestructive
test methods are indispensable for characterisation of damage. Post-impact tests are needed to
determine residual stiffness and strength. The
combined geometric and material nonlinear finite
element analysis capability required for this purpose is available in commercial finite element
systems. However, accurate constitutive models

for composite laminates with multiple matrix


cracks, fiber breaks and delaminations are not yet
available.
Delaminations are often the primary, life-limiting failure modes. The task of developing
methods and data for predicting the onset and
growth of delamination due to impact was not
considered in the present study. Accurate evaluation of interlaminar normal and shear stresses a
verified interface criteria and test methods to
measure material properties associated with such
a criterion are the prerequisites for prediction of
delamination onset. It is to be noted that an
appropriate finite element model for this purpose
should be based on a lamination theory that
enforces the continuity of interlaminar normal
and shear stresses at ply interfaces,~5 and these are
not yet implemented in commercial FEM systems
such as ABAQUS. For the prediction of delamination growth, the fracture mechanics concept ~6is
indispensable. This procedure involves calculation of energy release rates associated with delamination growth, the use of a mixed-mode fracture
criterion and test methods to measure interlaminar fracture toughness. A drastically different
finite element modelling approach is necessary for
the numerical determination of energy release
rates. These are topics deserving further research.
This study sets the stage for confident application of a commercial FEM system ABAQUS for
numerical simulation of impact tests conducted
on composite plate and shell structures.

Impact response of laminated compositeplates


Procedure #3

71

z/h

at t = 0 . 2 ms, V o = 4 0 m / s
__

X = 1022.3 MPa

0'

iii:!ii:ii:i!i:!i:i:i:ii:ii:!i:i:i

90'

?::::..:.v....:.v..::.v.~..`..v:.......v.....v.v...v.`....:..:......`.:.v..::...

I1

~ : ~ : ~ : ~ : ~ : ~ : ~

90'

~.........................................................
:~:~:~:~:~::~:~:~:~

90'

~::..`.:......`.:.v`.::~.:~.....:.v:.....v.....v`.`.`.`.:~.~.`.:::......::.v..`.`.`..
i..:.v..`.:....::.`....:::......`.:............:.v...v....:..`.........v.....v:...v::.v
:::~::~::~:~:~:~:~::~:~:~:~::~:~:~::~::~:~:~::~:~:~:~:~
90'
O'

--

O*

-1

01/X
O
0

Y = 27.3 MPa

!iiii:iiiiiii:ii:i:iiiii:iii:i:ii!i!:iii(.:.~...~.~.~.~.~::..~..~..~...~..~...~......`..~.~.::..`.9o'
~.`.`.~`~...~.`..~
i!i!:!i:!:!i:i!:i!i!hi!i:!i!i!i:!i!:!i
9O'

~.~.):.~.~.~.)):~)~.)))~.~.~`~.~.~.33)&:.:.:.~.))3~`~.~.)~.)~.)90'
)~.:.:.:.~.~.333~:.~
"

0'
0
q

-10

lO

~/y
z/h
S = 40.9 MPa

ii:iiiii:iii:!i!ih:i:i:i!iii:i!:i:!i!

9o'

:......:.....v:.v...v.......v~..`:..:.v...~`v.......v.v......:..::..:.v....`....`....:
~0'
0

.1

otis
Fig. 12.

Stress distribution across laminate thickness.

ACKNOWLEDGEMENTS
t =0.05 ms
~]

t = 0.15 ms

t=O.2Oms

900

i 0

This research was supported by the Natural


Sciences and Engineering Research Council of
Canada and la Direction de la Recherche de
l'Ecole Polytechnique de Montr6al.

REFERENCES

V0= 40 m/s Procedure #3


Fig. 13.

Growth of impact damage.

1. Greszczuk, L. B., Damage in composite materials due to


low-velocity impact. In h n p a c t D y n a m i c s , ed. J. A.
Zukas et al. John Wiley & Sons, New York, USA, 1982.
2. Zukas, J. A., Numerical simulation of impact phenomena. In I m p a c t D y n a m i c s , ed. J. A. Zukas et al. John
Wiley & Sons, New York, USA, 1982.

72

H. V. Lakshminarayana, R. Boukhili, R. Gauvin

3. Abrate, S., Impact on laminated composite material.


Applied Mechanics Reviews, 44 ( 1991 ) 155-90.
4. Cantwell, W. J. & Morton, J., The impact resistance of
composite materials -- Review. Int. J. Composites, 22
( 1991 ) 347-62.
5.
ABAQUS
-A general purpose finite element code,
users manual, version 4-7. Hibbit, Carlson and Sorensen Inc.. Providence, RI, USA.
6. Yan, T. M. & Sun, C. T., Use of statistical indentation
laws in the impact analysis of laminated composite
plates. J. Appl. Mech., 52 (1985)6-12.
7. S. W. Tsai, & Hahn, H. T., brtroduction to Composite
Materials. Technomic, Lancaster, PA, USA, 1982.
8. Nemes, J. A. & Simmonds, K. E., Low-velocity impact
response of foam-core sandwich composites. J. C~mp.
Mat., 26 (1992) 500-19.
9. kakshminarayana. H. V. & Sridhara Murthy, S., A shear
- - flexible triangular finite element model for laminated
composite plates. Int. J. Nttmer. Meth. In Eng., 20
11984) 591-623.
10. Lakshminarayana, H. V. & Kailash, K., A shear -deformable curved shell element of quadrilateral shape.

Computers and Structures, 33 (1989) 987-1001.


11. Sun, C. T. & C h e n . J. K., On the impact of initially
stressed composite laminates. J. (~mp. Materials. 19
(1985) 490-504.
12. Ewins, D. J. & lmregun, M., A survey to assess structural
dynamic response prediction capabilities. In Quulit3,
Assurance in FEM Technology, ed. J. Robinson. Robinson Associates, UK, 1987.
13. Aggour, H. & Sun, C. T., Finite element analysis of laminated composite plate subjected to circularly distributed
central impact loading. Computers & Structures, 28
(1988) 729-36.
14. Yener, M. & Wolcott, E., Damage assessment analysis of
composite pressure vessels subjected to random impact
loading. J. Pressure Vessel Technolog3', 111 (1989)
124-9.
15. kec, C. Y. & kiu, D., lnterlaminar stress continuity
theory for laminated composite analysis. AIAA ,/.. 29
1 9 9 ( ) 2010-12.
16. Friedrich. K. (ed.). Application of kracture Mechanics to
('omposite Materials. Elsevier Applied Science Publishers, Amsterdam, The Netherlands, 1989.

Vous aimerez peut-être aussi