Vous êtes sur la page 1sur 42

Damage Tolerance of Laminated

Composites Containing an Open Hole and


Subjected to Compressive Loadings:
Part I-Analysis
FU-KUO CHAtiG AND LARRYB. LESSARD
Department of Aeronautics atid Asrrotiaiitics
Starford Utiisersity
Starford, CA 94305
(Received April 24, 1989)

ABSTRACT: An analytical investigation was performed to study the damage in lami-


nated composites containing an open hole and subjccted to compressive loading. A pro-
gressive damage model was developed during the inxstigation to predict the extent and the
failure modes of the internal damage in the laminates as a function of the applied load and
to simulate the in-plane response of the laminates from initial loading to final collapse.
The model consists of a stress analysis and a failure analysis. Stresses and strains inside
the laminates were calculated by a nonlinear finite element analysis which is based on
finite deformation theory with consideration of material and geometric nonlinearities. The
types and extent of damage in the material were predicted by a failure analysis which in-
cludes a set of proposed failure criteria and material degradation models.
Numerical results from the model were compared with the data which were obtained
during the investigation and are presented in a companion paper [l]. Good agreements
were found between the predictions and the test results. A computer code was developed
based on the model which can be used as a tool for sizing and designing composite plates
containing holes and subjected to compression.

1. INTRODUCTION
composite technologies have increased the mechanical
R ECENT ADVANCES IN
performance laminated composites significantly.
of consequence, com-
posite structures are required in more sophisticated designs carry heavier and
As a
to
heavier loads as primary load-carrying elements [2,3]. Therefore, it becomes in-
creasingly important that the behavior of composite structures subjected to com-
pression be predictable and compression failure mechanisms in composites be
fully understood.
Laminated composite plates containing holes or cutouts have been studied very
extensively in the literature [4-6].Howvcver, most studies concentrated on tensile

2 Joirrtinl of COWOSITE Vol. 25- Jatiiimy 1991


MATERIALS,
0021-9983/91/01 0002-42 S6.0010
0 1991 Technoniic Publishing Co., Inc.

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Damage Tolerance of hmirtaied Conposiies: P0r1 I-Analjsis 3

loading only. Relatively fewer studies were conducted on the compression failure
of composites containing holes and cutouts 17-18]. Due to the nature and coni-
plexity of composites, the failure mechanisms in laminated composites due to
compression are quite different from those resulting from tensile failure. Accord-
ingly, in order to accurately predict compression failure, modeling should ac-
count for the different failure modes in laminated composites specifically due to
compression.
The objective of this investigation was to study, both analytically and ex-
perimentally, the in-plane compression failure mechanisms and the response of
laminated composites containing an open hole. The response of the laminates due
to local damage as a result of stress concentrations \ n s the primary concern. The
effects of ply orientation, stacking sequence, and geometry on the strength and
response of the laminates were also studied.
The focus of this study is to develop a progressive damage model for predicting
the types and extent of in-plane damage in composite plates as a function of ap-
plied load, and for evaluating the residual strength and stiffness of composites at
a given load. An extensive expcrimental program was also performed by the
authors in 111 to examine the failure mechanisms in the laminates and to generate
data for verifying the analytical model. Based on the model, a computer code
was developcd which can be used as a design tool for sizing and designing com-
posite plates containing an open hole and subjected to compression.

2. STATEhlENT OF PROBLEM
Consider a laminated composite plate, made of layers of continuous fibers eni-
bedded in an organic matrix, containing an open hole located at the center of the
plate, as shown in Figure 1. The ply orientation of the laminate can be selected
arbitrarily, but it must be symmetric with respect to the middle plane of the plate.
The laminate is subjected to an in-plane compressive load only, as shown in
Figure 1. No out-of-plane loading, bending or torsion is applied. The dimensions
of the plate are chosen such that the laminate will not buckle globally due to ge-
onietry.
The load is applied incrementally. During loading, the plate is only allowed to
deform in its own plane until it collapses at which point the plate cannot sustain
any additional load. It is desired to obtain the following information:
1. The types and size of damage inside the laminate at a given load
2. The in-plane response of the plate
3. The compressive strength of the plate
4. The residual stiffness and residual strength of the plates after a prescribed
load

3. ANALYTICAL AIODEL
Since the major interest of this research is in-plane compression failure in lam-
inated composites due to material damage, rather than geometrically related

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


4 Fu-KUO CHAKG B. LESSARD
A N D LARRY

Figure 1. Illustration of the basic geometry of the problem showing length L, width W, edge
distance E, thickness H, hole diameter D and applied load P.

buckling, out-ofiplane deformations of the composite plates are not considered in


the analysis. Hence, the problem becomes that of two-dimensional plane stress.
The proposed analytical model can be divided into t\vo parts: the stress analysis
and the failure analysis. In the following, a detailed description of each analysis
will be given.

3.1 Stress Analysis


Due to the inherent nature of the materials, the laiiiinated composites can re-
spond linearly or nonlinearly to loading, depending upon the ply orientation of
the laminates and the loading direction [13,19-211; thus, material nonlinearity
must be considered in the stress analysis. Furthermore, local damage in the mate-
rial can cause substantial degradation in mechanical properties and result in non-
linear material and geometric response. Therefore, the development of the stress
analysis is based on the theory of finite elasticity to take into account the material
and geometric nonlinearities.

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Daniagc Toleratice of htnirmted Cotnposiles: Pcirt I-Atialyis 5

Following the method of the updated Lagrangian formulation [22-241, thc


variational form of the equation of equilibrium can be written as:

where "+ASjkare the Piola-Kirchhoffs Stresses at the next step I I 1, "+.'Ejkare +


the incremental Green's strains from step I I to 11 +
1, and .G' is the surface trac-
tion term:

which contains n+lr.,


the surface traction acting on surface "Sin configuration ti.
The total stresses "+,'Sjkcan be expressed as [22,23]

where "ajk are the Cauchy stresses in the current configuration IZ and "+,!ASjkarc
the incremental Piola-Kirchhoffs stresses from "V to ""Vin terms of the "Vcon-
figuration.
Substituting Equation (3) into Equation (l), we obtain

1." "',!AsjhG"+,'Ejk(i"II

The incremental Green strains


-k "Cl,rS"+,'Ejkd"t'- n+l.G'= 0

can be written as [22,23]


(4)

where

and

the common strains in small scale deformation, "+,!Vjk is the rotation ten-
" + . ' E ~ ~are
sor associated with large deformation, and "+'Ailjis the incremental displacement
vector.

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


6 FU-KUO CHANG AND LARRY
B. LESSARD

Equations (5-7) are combined with Equation (4) to yield

It was assumed that in each step from configuration ti to configuration tl $. 1 ,


the changes in stresses were small enough so that the stress-strain relations could
be treated as linear during the deformation. Hence the incremental Piola-
Kirchhoffs stress in Equation (8) could be expressed as [22,23]

where ,,Cjhlm,
the reduced moduli, were given by

Here hi is the thickness of the ith layer, and .g;,is the transformed reduced stiff-
ness matrix for the ith layer [6].The subscriptsj, k, f, and tti are related to rand
s as follows

j+tk-r=3 1f tt1- s = 3

The conversion of indices shown in Equation (11) greatly simplifies the number
of tensor multiplications required. The result is that the stresses and strains are
both reduced from 2 x 2 symmetric tensors to 3 x 1 vectors. The material
property is reduced from a 2 x 2 x 2 x 2 4th order tensor to a 3 x 3 tensor.
The transformed reduced stiffness matrix must be recalculated at each incre-
ment because of material nonlinearity. Thus the reduced moduli ,,cjk/,,, also de-
pend on each increment. The components of ,,Pjscan be related to properties of
a unidirectional layer, but, first, the layer properties with their inherent non-
linearity must be established. For graphite fiber-reinforced, organic-matrix
composites, material nonlinearity is primarily due to nonlinear shear-stress/
shear-strain relation in each unidirectional layer [19]. Therefore, the material
nonlinearity is introduced through the shear stress-shear strain relation at the
ply-by-ply level. Different composite ply orientations show different degrees of
nonlinearity, for example a [ &45], layup exhibits a strong nonlinear behavior,
while a [0/90], layup behaves very linearly.

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Daniage Tolernrice of hniirinted Composites: Pan I-Adysis 7

On a ply-by-ply level, the in-plane stress-strain relations can be expressed as

where x and yare coordinates parallel and normal to the fibers, respectively. The
material nonlinearity, that appears at the ply level, is in the shear-stress/shear-
strain function. The nonlinear relationship advanced by Hahn and Tsai [19], suc-
cessfully used previously by Chang et al. [6,21], was chosen and has the form

Y, = (&) a,, + ady


where GI, is the initial ply shear modulus and a is the nonlinear paranieter of the

:;:
material that has to be determined experimentally.
From Equations (12) and (13), the on-axis stiffness of a unidirectional layer of
composite material can be deduced as:

0
01

EX Eyvx
1 - VXVY 1 - vp,

Exvy EY

1 -- v,vy 1 - YIVY

0 0

where cxy is the nonlinear shear modulus obtained by taking the partial deriva-
tive of Equation (13), with respect to yxy.

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


8 Fu-Kuo CHANG
AND LARRY
B. LESSARD

Note that cry will depend on the current value of the shear stress in that layer,
i.e., “a,.
The components of ,,&, i.e., the “off-axis” material propcrties can be related
to ,,&, the “on-axis” material properties, by tensor transformation as follows:
-
nQF = ‘Ci(nQij)qs (16)

where T,iis a 3 x 3 transformation matrix [ 2 5 ] .

which is ready for finite element formulation.

3.2 Failure Analysis


Due to the complex nature of composites, failure modes in laminated compos-
ites are strongly dependent on geometry, loading direction, and ply orientation.
Considering only a unidirectional composite, there are basically five different in-
plane failure mechanisms: matrix tensile cracking, matrix compression, fiber
breakage, fiber-matrix shearing, and fiber buckling. All of the mechanisms, with
the exception of fiber breakage, can induce compression failure in laminated
composites, and are shown schematically in Figure 2. The inclusion of matrix
tensile cracking mode, which Seems incorrect intuitively, will be proven to be an
important failure mode in multidirectional laminated composites subjected to
compression. Furthermore, for laminated composites, delamination failure may
also occur at the interfaces of t w neighboring layers with different ply orien-
tations.
For laminated composites containing holes, the failure phenomena become
even more complicated because of the introduction of stress concentrations near
the hole boundary. Thereforc, in order to accuntely analyze compression failure
of laminated composites containing holes, the proposed failure analysis must not
only be able to predict failure modes in the laminated composites, but can also
evaluate the reduction of the material properties within the damaged area. In this
investigation, failure in laminated composites is predicted by a set of proposed
failure criteria for each mode of failure, and the material properties within the
damaged area were evaluated according to the proposed property degradation
models.
Most traditional failure criteria [26] are unable to distinguish failure modes in
each layer. Hashin [n] proposed a set of failure criteria for predicting failure of
unidirectional composites based on each failure mode. Recently, Chang et al. [6]
modified these failure criteria to predict failure of laminated composites contain-
ing a hole and subjected to tensile loading. The modified failure criteria take into
account the nonlinear shear deformation in each layer. Only three failure modes
were considered: matrix cracking, matrix compression, and fiber breakage.

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Damage Tolernrice of Lntiiiiiated Composites: Part I-Aiioljsis 9

MATRIX TENSILE h1ATRI.X


CRACKIKG CO;\lPRESSION

FIBER-M ATRIX FIBER BUCKLING


SHEARING
Figure 2. The four important failure modes for the implane compressionproblem, viewed at
the ply level.

Therefore, in this investigation, the failure criteria previously developed by


Chang et al. [6] will be extended to include the other critical failure modes (fiber
buckling and fiber-matrix shearing failures) for modeling in-plane failure of lam-
inated composites containing holes and subjected to compression. In addition,
these failure criteria will be modified to take into account the effect of ply cluster-
ing (plies with the same ply orientation grouped together) on the strength of lami-
nates.
It is well known that, from experiments, in sirii ply strengths of each layer in
a laminate can differ significantly from the measured ply properties data [28,29],
because of the lamination effect [28-471. Two major effects haw been observed
in the literature: (1) as the thickness of a group of identical plies increases, its

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


10 FU-KUO CHANCAND LARRYB. LESSARD

strength will decrease [28,29,3446] and (2) the ply orientation ofan adjacent ply
will strengthen a ply according to the "adjacent ply constraint" effect [29,32-341.
Generally speaking, interlaminar shear and interlaminar normal stresses are
greatly influenced by both the ply thickness effect and the adjacent ply constraint
effect. These effects influence stresses in the inner regions of a plate as well as at
the free edge [20,31,33,46,47].Physically, the inherent weakening effects could
be attributed to the following: as a group of plies becomes thicker, a crack in the
group can propagate to a larger extent before being arrestcd by an adjacent ply.
Also, the ply angle of that adjacent ply will determine its effectiveness in arrest-
ing such a crack.
Laminates with weaker ply strength distributions are more prone to fail prema-
turely by matrix cracking and delamination, resulting in lower overall laminate
strength. In general, laminates with alternating ply orientation (no plies with the
same ply orientation stacked together) are stronger than laminates with clustered
ply orientation (plies with the same ply orientation stacked together) [28-471, be-
cause the latter fails prematurely by matrix cracking and delamination. There-
fore, in order to predict failure in laminated composites, accurate ply strength
distributions in the laminate should be applied in the failure criteria chosen.
However, all the current failure criteria are based on the ply strength properties
measured from standard unidirectional ply tests.
Accordingly, in this investigation, a set of failure criteria incorporated with iiz
sitii ply strength distributions will be proposed to predict in-plane failure and the
corresponding mode of failure in laminated composites containing holes. Once
failure occurs in the composites, the material properties within the damaged area
suffer some degree of degradation. The amount of property reduction strongly
,depends on the mode of failure. Therefore, based on the predicted modes of fail-
ure, property degradation models will be proposed to evaluate the reduction of
material properties within the damaged area.

3.3 Zn sitit Ply Strength Distributions

3.3.1 PLY TRANSVERSE TENSME STRENGTH


It is well known that the transverse tensile strength, Y,,of unidirectional plies
in a laminate strongly depends on the thickness of the plies in the laminate and
the ply orientation of surrounding plies [29]. Generally speaking, the transverse
tensile strength is a function of ply thickness and the neighboring ply orientation.
Figure 3 shows a typical distribution of the ply transverse tensile strength as a
function of neighboring ply orientation and the ply thickness. As the thickness of
90 layers in the [0/90] laminates increases, the strength of 90" plies (associated
with first cracking) decreases and approaches to the transverse strength of a uni-
directional laminate of ply layup [90n]..
This phenomenon has drawn significant attention among the composites com-
munity. Numerous researchers have proposed approaches to analyze the prob-
lem. One approach has been to use a Weibull strength distribution [29,39,40],
while others have used shear lag theory combined with fracture mechanics [35],
or fracture mechanics only [30]. Some researchers have also pointed out that

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Darnage Tolerutice of hrtiiriated Cornposires: Pun I-Atmljsis 11

18 -l
T300 I 9 3 4
16- 0 DATA OF FLACCS b. KURAL

0
8
0

# OF CLUSTERED 90’ PLIES (n)


Figure 3. The variation of transverse tensile strength versus ply clustering in [ 0/9OJs lay-
ups. The model uses Equation (4.1) with A = 1.3 and B = 0.7 for T300/934 graphite/epoxy
composites.

strengths tend to vary roughly as 1/d#, where N is the number of consecutive


(clustered) plies of the same ply angle [37,38].
Here, a simple “decaying” type distribution, which modifies the I/dm relation-
ship, is introduced to evaluate the ply transverse tensile strength in a laminate as
a function of layer thickness and neighboring ply orientation, i.e.,

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


12 Fu-Kuo CHANG B. LESSARD
AND LARRY

where

Y, = in
sirrr transverse tensile strength of a laminate, function of both A0
and N
A0 = minimum ply angle change bet\reen the ply under consideration and its
neighboring plies (above and below)
N = number of consecutive (clustered) plies of the same ply angle
E = transverse tensile strength of a [90,], laminate (11 2 6 )
A , B = material parameters-the values of A and B have to be determined from
expcriment

The major difference of Equation (18) from thc others is that it accounts for ply
orientation. By comparing the test data [29J shown in Figure 3, the results of
calculations based on the model fit reasonably \veil with the data (using A = 1.3
and B = 0.7 for T300/934 graphitekpoxy).
3.3.2 PLY SHEAR STRENGTH
Similar behavior was also observed from experiments for the ply shear strength
distribution in laminated composites by investigators in the litemture [45,46,48].
An extensive experimental study on the shear strength distribution in composites
was reported by Chang et al. [45,46]. The in sirii ply shear strength in a laminated
composite strongly depended on the thickness of the ply in the laminate and ply
orientation. Figure 4 shows a typical distribution of the ply shear strength

2-

0 . . . . 1 . . . . I . . . '

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Damage Elerarice of htriitiflted Composites: Pmt I-Atialysis 13

measured by a rail shear or an Iosipescu fixture. It clearly indicates that a signifi-


cant reduction in ply shear strength (by as much as fifty percent) can be made by
grouping plies with the same ply orientation together in the laminate. Although
the degree of reduction in shear strength is different from that in transverse ten-
sile strength, both reductions show a similar trend.
Hence, an expression similar to Equation (18) for transverse tensile strength
was proposed for evaluating the shear strength distribution in a laminate, i t . ,

where
S, = in sitii ply shear strength of a laminate, function of both A0 and N
A0 = minimum ply angle change between the ply under consideration and its
neighboring plies (above and below)
N = number of consecutive (clustered) plies of the same ply angle
S: = shear strength of unidirectional composites with niore than eight plies
C, D = inaterial parameters-the \dues of Cand D have to be determined from
experiment
Again, the proposed distribution fit the data very well as shown in Figure 4 (using
C = 2.0 and D = 1.0 for T300/!376 graphite/epoxy).
3.3.3 FIBER BUCKLING OR KINKING STRENGTH
Fiber buckling is a very important compression failure mode because it fre-
quently leads to catastrophic failure of a structure in compression. Fiber buckling
strength of a unidirectional composite is normally measured experimentally
by performing a uniaxial compression test. The load distribution is uniform
across the specimen width. However, for composite structures subjected to
nonuniformly distributed loads, the local fiber buckling strength in a composite
can be affected significantly by haw the load was distributed over the region. For
instance, consider a unidirectional composite subjected to either a uniformly dis-
tributed load or a nonuniform, and more concentrated load as shown in Figure 5.
Schematically, for the composite subjected to constant loading, all the fibers will
deform consistently once the buckling load is reached; thus, the fiber-matrix-
fiber interaction is negligible. However, for the composite subjected to a concen-
trated load, the fibers subjected to the highest load tend to buckle earlier than the
others which, in this situation, seem to serve ;is lateral supporters to the highly
loaded fibers, in order to prevent them from buckling. Hence, the fiber-matrix-
fiber interaction effect appears when the structure is subjected to nonuniform
loadings and, therefore, has to be taken into account when evaluating in sirti fiber
buckling strength of composites due to stress concentrations.
An analytical model has been developed by the authors to study the effect of
load distribution on the fiber buckling strength of unidirectional composites and
was presented in detail elsewhere [49]. Basically, the model is based on the en-
ergy principle which considers the fiber bending energy, matrix shearing energy,
and the matrix extension energy. The fiber buckling strength X , as a function of

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


14 Fu-Kuo CHANG
A N D LARRY
B. LESSARD

FIBER BUCKLING
UNIFORM LOADING NON-UNIFORhl LOADING

BEFORE AFTER B E ~ R E AFTER


NO FIBER INTERACTION FIBER INTERACTION
Figure 5. Simulation of the buckling of microfibers due to uniform and nonuniform loading
conditions.

load distribution can be determined from the model. In this investigation, the
model was adopted to determine the iiz sitii fiber buckling strength X, in the
notched composites. For example, the fiber buckling strength on the hole bound-
ary near the stress concentrations of a T300/976 [(0/90)6],composite plate with
a diameter of 0.25 inches was determined to be about 1.78 times higher than the
fiber buckling strength X: measured from constant loadings [49]. The calculated
irz sitii fiber buckling strength X , from the model will be incorpomted with the
failure criteria for predicting damage in composite structures.
3.4 Failure Criteria
A set of failurecriteria incorporated with iu sitii ply strength distributions were
proposed to predict failure and the mode of failure of each ply in laminated com-
posites.
3.4.1 MATRIX E N S I L E CRACKING
For predicting tensile matrix cracking failure (a, 2 0), the failure criterion
has the form [6]

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Damage Tolerance of hininated Cottposiies: Part I-Aiinljsis 15

where uyand uryare the transverse tensile stress and shear stress in each layer, re-
spectively. -y2y is the ultimate shear strain and Y, is the transverse tensile strength
as defined by the in sirii ply strength distribution [Equation (IS)].
By introducing the ply shear stress-shear relationship of Equation (13) into
Equation (20), we obtain [5,6]:

a,, + -3a d y
-
+ 2Gxy
s:3
4
= e:f.
2);(
- + -as:
2G,, 4

where Gxyis the ply in-plane shear modulus, a is the shear nonlinearity parame-
ter, and S, is the ply shear strength as defined by the in siru ply strength distribu-
tion [Equation (19)]. Note that for laminates with linear elastic behavior ( a = 0 ) ,
Equation (21) can be reduced to

2);( + (z)2 = Pif.

which is the two-dimensional form of the tensile matrix criterion proposed by


Hashin [27].
In summary, the matrix tensile cracking failure criterion states that when, in
any one of the plies in a laminate, the stresses u, and a,, satisfy Equation (20)
(with e,,, > l), matrix cracking occurs in that layer.
3.4.2 MATRIX COMPRESSION FAILURE
For predicting matrix compression failure (a, c 0 ) , a new criterion was pro-
posed which has a similar form to that of the tensile matrix cncking criterion.
The criterion can be expresscd as:

(;)2.

1..
or

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


16 FU-Kuo CHAtiG AND LARRYB. LESSARD

where Y, is the in-plane matrix compressive strength of a unidirectional ply. For


linearly elastic laminates, Equation (24) can be reduced similarly to

($2 + ($)2 = 4,-

The matrix compression failure criterion states that when, in any one of the
plies in a laminate, the stresses a, and us, satisfy the criterion (en,- > I), that
layer fails by matrix compression.
3.4.3 FIBER BUCUING FAILURE
For predicting fiber buckling (kinking or microbuckling) (u= < 0), the failure
criterion is based on the fiber buckling strength theory proposed by Lessard and
Chang [49]. The criterion can be expressed as:

where X , is a function of load distribution and mechanical properties of fibers


and matrix. Refcrence [49] elaborates on the factors that influence the value
of x,.
The fiber buckling failure criterion states that when, in any one of the plies
in a laminate, the stress satisfies the criterion (cb > I), that layer fails in fiber
buckling.
3.4.4 FIBER-MATRIX SHEARING FAILURE
For predicting fiber-matrix shearing failure (us < 0), the criterion can be ex-
pressed as

or

0,
-
2GY
+ -(Yo:,
3
4
= eX
s:
- 3
(?)2 +

2GY
+ -as:
4

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Dntnage Tolernrice of Lnrninnted Conyosites: POI?I-Anoljsis 17

Once again, for laminates with linear elastic behavior (a = 0), Equation (28)
can be reduced to:

The fiber-matrix shearing failure criterion states that when in any one of the
plies in a laminate, the combined stresses or (ax< 0) and u,,satisfy the criterion
of Equation (27) with c,~ > 1, that layer fails by fiber-matrix shearing.
3.5 Property Degradation Models
Material degradation within the damaged area was evaluated based on the
mode of failure predicted by the failure criteria. Therefore, the residual stiff-
nesses of composites strongly depend on the mode of failure in each layer. The
property degradation models for each layer are proposed as follows.
3.5.1 MATRIX TENSILE FAILURE
For matrix tensile cracking failure in a layer, the transverse modulus E, and
Poisson's ratio v, are reduced to zero, i.e., the matrix can no longer carry any
load in tension. However, the longitudinal modulus Ex, and the shear strcss-
strain relations are unchanged, i.e., in the failed layer, the in-plane properties are

Ex Eyvr '
1 - vxuy 1 - u*uy
-[Ex 0 0
O
Exv, EY
1 - "UY 1 - VIVY

3.5.2 MATRIX COMPRETSION FAILURE

' Ex E, v,
1 - u,uy

1
Exv,.
- vxuy
1

1
-

-
E
Y
uxuy

v,uy
-[::]

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


18 B.
FU-KUO CtlANG AND LARRY LESSARD

3.5.3 FIBER BUCKLING FAILURE (KINKING O R MICROBUCKLING)


Fiber buckling in a layer is a catastrophic mode of failure and when it occurs,
the material in that region cannot sustain any more load. Thus the material prop-
erties for the failed layer and all other layers are reduced to zero.
For a, < 0 and eb > 1

3.5.4 FIBER-MATRIX SHEARING FAILURE


In fiber-matrix shearing failure mode, the material can still carry load in the
fiber direction and in the matrix direction, but shear loads can no longer be car-
ried. This is modeled by reducing the shear property and the Poisson's ratios, v,.,
and v,, to zero.
For a, < 0 and e,, > 1

Ex
1 - VKVY

Exvy
1 - v,vy

4. NURERICAL RIETHOD AND COhIPUTER CODE


The stress analysis and the failure analysis developed in this study were im-
plemented into a finite element code. Basically, stresses and strains are calculated
at each incremental step, and evaluated by the failure criteria to determine the oc-
currence of failure and the mode of failure. Mechanical properties in the dam-
aged area are reduced appropriately, according to the property degradation
models. Stresses and strains will then be recalculated to determine any additional
damage as a result of stress redistributions at the same load. This procedure will
continue until no additional damage is found, and the next increment is then pur-
sued. The final collapse load is determined when the plates cannot sustain any
additional load.
During the course of the study, damage growth was found to be sensitive to the
finite element mesh used. The problem of mesh sensitivity to darnage growth is

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Daiiiuge Tolerance of Lnniinuted Composites: Part I-Aita [pis 19

similar to the problem of crack propagation in isotropic materials. For those


problems, adaptive finite element meshes are often used which always keep ele-
ment size small in the region of the crack tip for better results. Therefore, using
a constant mesh for the progressive failure analysis of laminated composites with
arbitrary ply orientations becomes inadequate. In this study, a mesh generator
was developed which can automatically create proper meshes depending on the
ply orientation, because damage growth in composites depends strongly on the
ply orientation. For example, in a [0/90], layup, the mesh was generated with a
rectangular area of dense elements near the hole boundary, as shown in Figure
6(a), because damage ivas expected to be propagating along 0 and 90 degrees
measured from loading direction. However, for [( f30)& layups, the mesh \vas
generated with elements emanating from the hole area along 130" lines as
shown in Figure 6(b). Details of the finite element analysis can be found in [25].
5. VERIFICATION AND CO&IPARISONS
In order to verify the proposed analytical model, numerical results were gener-
ated from the model to compare with the test data obtained in [l]. The material

(a) mesh used for (b) mesh used for


[(0/90)6], modeling [(+30/-30)61, modeling
Figure 6. Two types of meshes generated by the "directional mesh" parameters of the
PDHOLEC computer code.

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


20 FU-KUO CflAfiG AND LARRY
B. LESSARD

properties of T300/976composites used in the calculations are listed in Table 1.


The comparisons between the predictions and the data are presented in the fol-
lowing for each selected ply orientation.

Internal damage and the corresponding failure modes predicted by the model
for a [(0/90),], laminate containing an open hole are graphically presented at dif-
ferent load levels in Figure 7. The model predicted that fiber-matrix shearing
emanated from thc hole boundary near the stress concentrations, and then propa-
gated vertically along the loading direction (parallel to the 0" fibers). The dam-
age was quite confined to a local area until fiber buckling was initiated in the 0"
plies. Fiber buckling mode then quickly propagated horizontally in the direction
normal to the loading direction. It is believed that fiber buckling could have trig-
gered delaminations and caused the total failure of the structure. The specimen
collapsed at the load of about 8500 lbf. This predicted'damage growth pattern is
very similar to what was observed in tests [l]. Figure 8 shows the X-radiographic
result of a test to a prescribed load of about 95% of the total failure load. The
confined matrix cracks near the hole boundary that are shown in Figure 8 agreed
with the prediction very well. The calculated load-shortening curve of the

Table 7 . The material constants and parameters required for the use of
PDHOLEC code with T300/976graphite/epoxy material used as an example.
~ ~~ ~ ~ _ _ _

Moduli Parameters Symbol (units)

Longitudinal modulus E, (Msi) 22.7


Transverse modulus E,. (Msi) 1.88
Shear modulus G,,.(Msi) 1.01
Poisson's ratio vx (-1 0.23
Material nonlinearity parameter (I (psi)-3 0.8 x 10-14

Strength Parameters Symbol (units)

Longitudinal tension XI (ksi) 220


Longitudinal compression XE (ksi) 231
Transverse tension YP (ksi) 6.46
Transverse compression Y, (ksi) 36.7
Longitudinal shear, cross ply S,, (ksi) 15.5

Micromechanical Parameters Symbol (units)


~ ~~

Fiber modulus (Msi)


€1 34.6
Matrix tensile modulus Em (ksi) 510
Matrix shear modulus Gm(ksi) 200
Fiber diameter tt (in) 0.0003
Fiber volume fraction VI (-1 0.66
Kink length ratio (L/t,) " (-1 30

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Damage Elerarice of Lnmitiated Composites: Rirt I-Atidpis 21

T300 I976
[(01~0)~1s
D = 0.25 In
WID = 4.0
MATRIX TENSION

= MATRIX COMP.
FIBER BUCKLING
FlBEWMATRlX
SHEARING

4773 I b 6026

8033 8492 COLLAPSE

Figure 7. Mode of failure for a [(0/90)& layup by progressive damage modeling.

[(0/90)a], laminate was also compared with the test results (see Figure 9). The
curve again agreed with the test results very well. The curve was fairly straight,
indicating that the specimen responded linearly up to the final failure. The pre-
dicted failure load also agreed with the data within 10%.
Numerical results were also generated for the specimens with other configura-
tions. Since the results were quite similar to the ones shown in Figures 7 and 8,
the additional results for different configurations will not be presented. However,
the predicted collapse loads for specimens with different dimensions are shown
in Figure 10. Again, the predicted failure loads as a function of width-to-diameter
ratio agreed very well with the data.

5-2 [(k 30)61,


A very surprising result was found for [(j=30),], laminates. The predicted in-
ternal damage and failure modes as a function of the applied load are presented
in Figure 11. The code predicted that there was only very limited damage near the
hole boundary in the laminate until the applied load reached the final failure load.
At the final failure stage, internal damage immediately expanded from the hole
boundary along 30" angles to the free edges, resulting in a total catastrophic fail-

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


T300/976 [(0190)6]s D = 0.25 in, WID = 4

I BEFORE TEST AFTER TEST


1
Figure 8. X-Radiographs of a [(0/90)& specimen, before and after testing to 95% of final
failure load.

EXTENSOMETER MEASUREMENT A E (in)


Figure 9. Load-shortening curve for a [(0/90),].specimen for two independent tests with
results from the model superimposed.

22

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Damage %!erarice of hntiitafed Contposites: Part I-Analjsis 23

--
c
h
v) 'O

-? T
a*- ' :-
60

al 50 -

D@piiJ
X
E-
40 -
crl
-
z 30 D 5 0.25 in

crl 20-
5
3 Lo-- ANALYSIS
2E A
* O DATA
,
$ v
2 3 5 6 7

WIDTH/DIAMETER RATIO (WID)


Figure 70. Strength versus geometry for [(0/90)& layups at various width/diarneter ratios.

ure. By carefully examining the predicted failure mode, it was found that matrix
cracking caused by matrix tensile failure was the dominating failure mode. This
can explain why this type of specimen failed very suddenly in a clear fracture
mode during the mechanical tests [l]. The prediction coincided with the experi-
ments very well as shown by the X-radiographic result in Figure 12. It should be
noted that this damage pattern is totally different from that of [(0/90),], laminates.
The predicted load-shortening curve of the [( f30)s], laminate also agreed with
the data very well (see Figure 13). The predicted failure loads as a function of
width-to-diameter ratio were also presented in Figure 14. The predicted results
also agreed fairly well with the data.

For [( &45)& laminates, the predicted internal damage and corresponding fail-
ure modes are presented in Figure 15 at different loading stages. Based on the
model, it can be seen that damage initiated from the hole boundary near the
stress concentrations and propagated in a direction 45" from that normal to the
loading direction. The failure modes are dominated by matrix compression and
fiber-matrix shearing, which are indicated by different types of shading in the
figure. The predicted damage pattern in the [( i45),], laminate is very similar to
the X-radiographic results of the specimen in Figure 16, in which the specimen
was loaded up to 95% of the final collapse load.
Figure 17 shows the calculated load-shortening curve of the specimen, com-
pared with the experimental results measured fiom an extensometer. The curve
followed the.experimenta1 data very closely. The predicted collapse load was
about 3000 lbf, which agreed with the experimental data very well. It is worth

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


24 Fu-Kuo CHAKGAND LARRYB. LESSARD

noting that the load-shortening curve for the [( *45)J, specimen is nonlinear. It
is believed that the nonlinearity of the curve is attributed to the nonlinearities of
the material and the local damage.
The effect of the width-to-diameter ratio on the strength of the laminates is also
predicted by the model and compares \Ye11 with the data, as shown in Figure 18.

5.4 [(O/ 45/90),],


Predicted damage progression of a [(O/ &45/90),],laminated composite as a

1300 I 976 MATRIX TENSION


[(+30/-30)6]S MATRIX COMP.
D = 0.25 In FIBER BUCKLING
FIBEiUMATRIX
SHEARING

3183 Ib 4358 COLLAPSE

Figure 11. Mode of failure for a [( f 30)& layup by progressive damage modeling.

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


I T3001976 [(+30/-30)& D = 0.25 in, WID = 6

I BEFORE TEST AFTER TEST

Figure 12. X-Radicgraphs of a [( 3015, specimen, before and after testing to final failure
load.

5000

4000
- 0
ANALYSIS
DATA

-n
h

. - 8

0.000 0.002 0.004 0.006 0.008


EXTENSOMETER MEASUREMENT A E (in)
Figure 13. Load-shorteningcurve for a [(k 30)& test specimen with results from the model
superimposed.

25

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


F- 0
ANALYSIS
DATA

F
D = 0.25 in
H=0.135
(thickness)

[ (30/-30)6]s

"
0 2 3 4 5 6 7
u
WIDTH/DIAMETER RATIO (WID)
Figure 14. Strength versus geometry for [(+ 30)& layups at various wdth/diameter ratios.

I
T300 I976
[(+45/-45)61S
D = 0.25 In
WID = 4.0
MATRIX TENSION
MATRIX COMP.
FIBER BUCKLING
FlEERlMATRlX
SHEARING

2704 Ib 2895

3005 2978 COLLAPSE

Figure 15. Mode of failure for a [(+-45)& layup by progressive damage modeling.

26

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


I T300/976 [(+45/45)& D = 0.25 In, W/D 5

I BEFORE TEST AFFER TEST


Figure 16. X-Radiographs of a [(k 45)& specimen, before and after testing to 95% of final
failure load.

4000
pimq
-ANALYSIS

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


B. LESSARD
FU-KUO CHAh’G A K D LARRY

60 -
.
-
- 0
ANALYSIS
DATA
D = 0.25 in

50

40 -
30 .-
20 I -
0

Figure 18. Strength versus geometry for [(+-45)& layups at various width/diameter ratios.

function of the applied load is shown in Figure 19. Damage was first predicted at
the applied load near 70% of the maximum load. As the load continues to in-
crease, damage by fiber-matrix shearing gradually expands in the direction nor-
mal to that of loading. Finally, the specimen collapsed due to fiber buckling, initi-
ating from the hole boundary and propagating immediately into the specimen.
Figure 20 shows the X-radiographs of the specimen, before and after a test of up
to 95% of the maximum load. The predictions in Figure 19 coincide with the ex-
perimental results. The prediction of fiber buckling mode in 0 degree plies was
also confirmed by experiment by observation through a microscope lens [l]. It is
believed that, like the [(0/90),],, the fiber buckling failure in [(0/*45/90)3],
would trigger multiple delaminations leading to collapse of the composites. This
prcdiction is consistent with the experimental observation that was recently per-
formed by Wass [50].The comparisons between the experiment and the predic-
tions on load-shortening and on failure strength are presented in Figures 21 and
22, respectively.

5.5 [(Ol+45)<Ip
The composites with this layup behave similarly to [(O/ ~t445/90),], and
[(0/90),], laminates. Final failure was triggered by fiber buckling. Figure 23
shows the predicted damage progression as a function of the applied load, and
Figure 24 shows the X-radiographs of the test results. The load vs extensometer
curve is shown in Figure 25, and the predicted failure loads, as compared with
experimental data, are shown in Figure 26.

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


T300 I 9 7 6
[(0/+45/-45/9O)JS
D = 0.25 l f l
WID = 4.0
MATRIX TENSION
a
= MATRIX COMP.
FIBER BUCKLING
FIBEIUMATRIX
SHEARING

4431 Ib 5398

5590 COLLAPSE

Figure 79. Mode of failure for a [(O/+ 45/90)& layup by progressive damage modeling.

5.6 Effect of Clustering: [06/9061r vs [(0/90)61s


The effect of ply clustering (plies with the same ply orientation grouped to-
gether) on the strength and damage of the notched laminates \!as also studied.
For example, considering [06/906], “alternating” laminates, the predicted damage
growth pattern in terms of the applied load is shown in Figure 27. Unlike
[(0/90)6]1 laminates, damage in [06/906]slaminates, dominated by fiber-matrix
shearing and matrix compression, initiates from the hole boundary much earlier,
and grows much faster in the direction parallel to the loading direction. The pre-
dicted damage size was much more extensive than that predicted in [(0/90)6].
laminates. Such extensive darnage \vould most likely induce delaniinations, be-
cause delaminations can be initiated by in-plane failures such as matrix cracking,
fiber matrix shearing, and fiber buckling. In this case, delamination buckling
may play an important role in determining failure load of the plates.
Figure 28 shows the X-radiograph of a [06/906], specimen after a test to 90%
of the total failure load. Due to thermal residual stress effect, extensive matrix
cracks were found in the specimen after fabrication. The specimen was then
loaded up to about 90% of the total failure load. From the X-radiograph of the
specimen taken after the test, it can be seen that there were no more matrix

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


T3001976 [(0/+45/-45/90)3], D = 0.5 in, WID = 3

BEFORE TEST AFTER TEST

Figure 20. X-Radicgraphs of a [(O/k 45/90)3]s


specimen, before and after testing to 95% of
final failure load.

0.000 0.002 0.004 0.006 0.008


EXTENSOMETER MEASUREMENT AE (in)
Figure 2 1. Load-shortening curve for a [(O/+ 45/90)Jsspecimen for two independent tests
with results from the model superimpased.

30

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


2
W
40 -
cz
30 -

Fl
D = 0.25 in

2
W 20

-
v)

v)
W
2 10 - ANALYSIS
0 DATA
D
$f
[ (0/45/-45/90)JS

E
n
5 "2
n .
3 4 5 6 7

WIDTH/DIAMETER RATIO (WID)


Figure 22. Strength versus geometry for [(O/+_ 45/90)& layups at various width/diarneter
ratios.

T300 I976
[(0/+45/-45)65
D = 0.25 in
W/D = 4.0
MATRIX TENSION
MATRIX COMP.
FIBER BUCKLING
FIBEWMATRIX
SHEARING

5343

6157
.~ ----
6583 COLLAPSE

Figure 23. Mode of failure for a [(O/+ 45)& layup by progressive damage modeling.
31

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


T300/976 [(01+45/-45)& D = 0.25 in, W/D= 3

I BEFORE TEST AFTER TEST

Figure 24. X-Radiographs of a [(O/+ 45)& specimen, before and after testing to 95% of
final failure load.

0.000 0.002 0.004 0.006 0.008 0.010


EXTENSOMETER MEASUREMENT A E (in)
Figure 25. Load-shortening curve for a [(O/k 45)& specimen for two independent tests
with results from the model superimposed.

32

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


33

,IT300 I 976/
60 -
50 - 6
40 -
30 -

20 -
lo
- ANALYSIS

U
2 3 4 5 6 7

WIDTHlDIAhlETER RATIO (WID)


Figure 26. Strength versus geometry for [(O/i 45)4]s layups at various width/diameter
ratios.

cracks introduced up to that load level, except for those matrix cracks emanating
from the hole boundary in the direction parallel to the loading direction. Exten-
sive delaminations were found in this specimen, possibly due to the extensive
matrix cracks observed in the X-radiograph before testing. The initial matrix
cracks were not included in the model.
For the final failure load, an interesting observation can be made. Figure 29
shows that find failure loads for [06/906]$specimens were higher than predicted,
and unexpectedly higher than [(0/90)6],specimens. This phenomenon could pos-
sibly be explained by the fact that the [06/906], specimen has lost a significant
amount of the stress concentrations due to full length matrix cracks coupled with
extensive delaminations near the hole boundary, which resemble the configura-
tion shown in Figure 30. Under this situation, the specimen would fail at a higher
load and thus explain the result shown in Figure 29.

6. PAFURIETRIC STUDY
Based on the comprehensive comparison between the analytical predictions
and the test results, it is clear that the proposed analytical model is capable of
modeling in-plane compression failure. This section will serve as an example of
the possible uays that useful numerical results can be generated from the com-
puter code.

6.1 Effect of Layup Sequence and Width-to-Diameter Ratio


Figure 31 shows the effect of layup sequence and geometry (width-to-diameter)

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


1
T300 I976 MATRIX TENSION
[O6/90ds
D = 0.25 In
W/D = 4.0
= MATRIX COMP.
FIBER BUCKLING
FIBEWMATRIX
SHEARING

3937 Ib 5681 8023

Figure 27. Mode of failure for a [Od90& "clustered" layup by progressive damage
modeling.

34

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Donrage i%lerarice of hriiiiiared Cotnposites: Part I-Analysis 35

I BEFORE TEST AFTER TEST

Figure 28. X-Radiographs of a [Od906], specimen (“clustered” layup), before and after
testing to 90% of final failure load.

on the strength of [(&to),],and [306/-306], composite plates containing an


open hole. Each curve shows the diminishing effect of the hoIe as the plate width
becomes large, and also shows the critical width region ( W / D 5 4) where the
ultimate strength is quite sensitive to the width ratio. The difference between the
two curves shows the effect of the clustering of the layup sequence. Similar data
have been reported by Herakovich [36] for tension testing of the same layup but
without a hole. Since both layups show little signs of damage before final failure,
it would be logical to select [( *30)a], layups in practical design.
However, design based solely on the strength concept of composites may result
in unsafe and potentially dangerous structures. For instance, although [06/906].
composites can carry higher static loads than [(0/90)6]1composites, internal dam-
age in [06/906]s composites grew much more extensively than that in [(0/90),],
composites, starting in the earlier loading stages. Such extensive internal damage
should not be accepted for design, because it could potentially l a d to failure of
the structures under other loading conditions, such as fatigue. .Therefore, it is
also important to know the size and the types of internal damage in composites
as a function of the applied load. For instance, by selecting the maximum damage
area in [(0/90)6], composites as a maximum allowable damage for this type of
material, the “design” strengths of [06/906], composites corresponding to the

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


1 2 3 4 5
\VIDTIUDIAhlETER RATIO (\V/D)
Figure 29. Strengh versus geometry for [06/90& "clustered" layups at various width/di-
ameter ratios, with comparison to [(OILXI)& "alternating" layups.

36

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


ii a a

[(0/90)6]5 "ALTERSATIKG" LAYUP [Od906]5 "CLUSTERED" LAYUP


0 deg. CRACKS C O W L E D WITH
DELI~IISATION
FAILURE LOAD = P FAILURE LO.4D 1.4 P -
Figure 30. PossibB explanation for [0d906]s
"clustered" layups having higher final failure
load than [(0/90)6]s
"alternating" layups.

- H = 0.135
60 (thickness)

10
L
-
.
-
-
I Im
ANALYSIS-ALTERNATLYC
-
ANALYSIS CLUSTERED
n
"
2 3 4 5 6 7

WIDTH/DIAMETER RATIO (WlD)


[(+30/-30)(Js & [(+30,/-3O~>Js
(ALTERNATIXG AND CLUSTERED LAYUP)
Figure 37. Parametric study showing the effects of clustering on the failure load of [* 30Is
layups.

37

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


38 FU-KUO CHAKG
AND LARRYB. LESSARD

allowable damage are plotted in Figure 32 as a function of width-to-diameter


ratio. Clearly, based on the damage concept, [(0/90)6],composites should be
selected in design because they can carry a much higher load than that of
[0,/906], composites within the allowable damage range. *

6.2 Effect of Loading Direction


Often, a design is made and subsequently, the loading conditions imposed
upon the design are changed. Rather than redesign the component, it would be
useful to observe the effect that a design change w u l d have upon the component.
Figure 33 shows what would happen to a layup if the direction of loading would
change by a certain angle. In the model, the loading direction remains the same
while ply orientation is merely rotated by the desired angle (in this case, by 20").
Since the problem is no longer left-right symmetric, a full mesh is used. It is ob-
vious from Figure 33 that the failure mode depends on the loading direction.
The variation in final failure modes are shown in Figure 34 as the loading
direction on a [( 3=45)6],layup is rotated from 0" to 90" off-axis. It is interesting
to observe how the final failure mode changes as the loading direction varies.
Fiber buckling failure is observed only when the layup is rotated by 45", which
is the same as a [(0/90)6],layup. Figure 34 shows the significant changes of final
failure mode resulting from an off-axis load.

7. CONCLUSIONS
A progressive damage model has been developed to study the compression fail-

60 -
50 -

40 -
"CLUSTERED"
30 - LLIILIIIIIIII

2 3 4 5 6 I
WIDTWDIAMETER RATIO W/D
Figure 32. Parametric study showing the effects of clustering on a new possible failure
design criterion for [Od906]s"clustered" layups and [(0/90)& "alternating" layups.

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


MATRIX TENSION FIBER BUCKLING
+251-65]0 WID = 4 MATRIX COMP. l?BFIBER'MATRIX

3029 Ib 3372

3617 COLLAPSE

Figure 33. Progressive damage analysis showing the effects rotating the loading direction
by 20 degrees ("off-axis" loading) for [(* 45)& layup.

39

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


c.
0

a,
s

40

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Darnnge Elernrice of hminnted Contposites: Purl I-Aiinljsis 41

ure in a laminated composite plate with a hole and to predict internal material
damage in the composites. The model can provide the following information:
1. The types and size of damage at a given load
2. The residual stiffness and strength of the laminates
3. The final collapse load
4. The complete response of the laminates from initial loading to final cata-
strophic failure
In addition, the effects of ply orientation, ply clustering, and geometry, on the
strength and response were also studied. However, delamination and delamina-
tion-induced buckling were not considered in thc model. The computer code
based on the model can be used as a design tool for sizing and designing compos-
ite plates containing holes and subjected to compression.

8. ACKNO\VLEDGEhIEKTS
The support of this research by the National Science Foundation's Presidential
Young Investigator Award and a Charles Lee Po\wll Foundation research grant
is gratefully acknowledged. Dr. Larry Lessard would also,like to thank the
F.C.A.R. (Fonds pour la Formation de Chercheurs et L'aide a la Recherche) of
Quebec for their support through a scholarship.

REFERENCES
1. Lessard, L. B. and F. K. Chang. 1991. "Damage Tolerance of Laminated Composites Containing
an Open Hole a d Subjected to Compressive Loadings: Part 11-Experiment," Journal of Coni-
posite hfuterials, 25(1):43-63.
2. Stenberg, R. 1988. "Carbon Fibre Composite on the Viggen Aircraft," Corrtposite Strirctures,
10(1):75-81.
3. Steinbeg, hi. A. 1986. "hiaterids for Aerospace," Sn'entiJic Arrwricari, 255(4):66-R.
4. Awerbuch, J. A. and hl. S. hladhukar. 1985. "Notchcd Strength of Compsite Laminates: Pre-
dictions and Experiments-A Review," Joitrrinl of Reirlforced Plastics arid Composites,4:3- 159.
5. Chang, F. K . and K. Y. Chang. 1987. Tost-Failure Analysis of Bolted Composite Joints in Ten-
sion or Shear-Out hlode Failure," Jortniol of Corriposire hfa/erinls,21:809-833.
6 . Chang, F. K. and K. Y. Chang. 1987. "A Progressive Damage hlodel for Laminated Composites
Containing Stress Concentrations," Jorrrrial of Corrlpnrire hfmeriols, 21:8%855.
7. Ericson, K., hi. Persson, L. Carlsson and A. Gustavsson. 1981. 'On the Prediction of the Initia-
tion of Delamination in a [OiSOl, Laminate with a Circular Hole," Journal of Corrtposite hfote-
rials, 18:495-506.
8. Nuismer. R. J. and J. D. Labor. 1979. -Application of the Alerage Stress Failure Criterion: Part
11-Compression," Joirrrinl of Cotripmite hfaterials, 13:49-60.
9. Rhodes, hl. D., hl. hi. hlikulas, Jr. and P. E. hlcGoum. 1983. "Effects of Orthotropy and Width
on the Compression Strength of Graphite-Epoxy Panels with Holes," AIM Jourrinl, 22(9):
1283-1292.
10. Gurdal, 2. and R. T. Haftka. 1986. "A Compressive Failure hlodel for Anisotropic Plates with
a Cutout under Compressive and Shear Loads," presented at 27rh Structures, Stnrctirral Djriarn-
ics and hfaterials Conference. Sari kitoriio. TX. hfay 19-21, 1956, pp. 680-688.
1 1 . Sohi, hl. hl., H. T. Hahn and J. G. Wliams. 1985. "The Effect of Resin Toughness and hlodulus
on Compressive Failure hiodes of Quasi-Isotropic GraphitdEpoxy Laminates," Torrgherzed Com-

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


42 Fu-KuoCHAW,AND LARRYB. LESSARD

posires, N h f STP 937. N. J. Johmton, ed. American Society for Testing and hlaterials
(hfarch):37-60.
12. Starnes, J. H. Jr. and J. G.\Villiams. 1982. 'Failure Characteristicsof Graphite-Epoxy Structural
Components Loaded in Compression," in hfechanirsof Composite hfaterialsr Recent Advances.
2. Hashin and C. T. Herakovitch, eds. IllTAhf Synpsiiim on hfeclranics of Coniposire bfate-
rials, Blacksbiirg, YA, Aitgiisr 16-19, 1952. pp. 283-306.
13. Shuart, hl. J. and J. G. Williams. 1983. "Compression Failwe of 45-Dominated Laminates with
a Circular Hole or Impact Damage," presented at 25111Srntditres. Sinrcriiral Dyamics and bfa-
rerials Corference, Pa fin Springs, CA, May 1446. I%4, pp. 399-409.
14. Chang. F. K., L. Lessard and J. hl. Tang. 1988. "Compression Response of Laminated Compos-
ites Containing an Open Hole," SAhfPE Qiiarierly. 19(4):46-51.
15. Guynn, E. G. and W. L. Bradley. "A Detailed Investigation of the hlicromechmisms of Com-
pressive Failure in Open Hole Lamimtes," Journal of Composite hfarerials (submitted).
16. Guynn, E. G.a d \V. L. Bradley. 1989. "hleasurements of the Stress Supported by the Crush
Zone in Open Hole Composite Laminates Loaded in Compression,"Joitrnal OfRPinforced Plas-
rics mrd Cornposires. 8(2):133-139.
17. Burns, S. W., C. T. Herakovich and J. G. William. 1987. 'Failure hlcdes for Compression
Loaded Angle-Ply Plates with Holes," Proceedings of 4111International Cofferelice on Composite
Srrucrures. Srorlnnd, UK.,Jitlj 27-29. 1987, pp. 2.175-2.191.
18. Gilletta, D. and R. Girard. 1987. "Degradation hlodels in Finite Element Analysis of htul-
tila~eredComposite Structures," Proceedings of 4Ill Irirernational Conference on Cornposire
Srrucritres. Scorland. UK..July 27-29. 1987, pp. 1555-1.563.
19. Hahn, H. T. 1973. "Nonlinear Bchavior of Laminated Composites," Journal of Cornposire hfare-
rials, 7:257-Zll.
20. Pipes, R. B., B. E. Kaminski and N. J. Pagano. 1973. 'Influence of the Free Edge upon the
Strength of Angle-Ply Laminates," Analysis of ihe Tesr Merliods for High hfdiihrs fibers and
Cornposires, M h f STP 521. J. hi. Whitney, ed. American Society for Testing and hlaterials,
pp. 218-228.
21. Chang, F. K., R. A. Scott and G. S. Springer. 1983. "Failure Strength of Nonlinearly Elastic
Composite Laminates Containing a Pin Loaded Hole," Joitrnal of Coinpsire hfarerials, 18:
463-477.
22. Bathe, K . J., E. Ramm and E. L. Wilson. 1975. 'Finite Element Formulations for h r g e Defor-
mation Dynamic Analysis," International Joiirnal for Niunerical Aferliods in Engineering, 9:
353-386.
23. Bathe. K. J. 1982. Finire Elemei~Procediires in Engineering Annljsis. Engleaood Cliffs, NJ:
Prentice-Hall.
24. Fung, Y. C. 1965. Foiindarions of Solid Mechanics. Englewood Cliffs, NJ Prentice-Hall.
25. Lessard, L. B. 1989. 'Compression Failure in Laminated Composites Containing an Open Hole,"
Thesis Dissertation, Stanford University.
26. Nahas, hl. N. 1986. "Survey of Failure and Post-Failurc Theories of Laminated Fiber-Reinforced
Composites," Joiunal of Composites TEclinologj & Reseatrli. 8(4):138- 153.
27. Hashin, 2. 1980. "Failure Criteria for Unidirectional Fiber Composites," Joitrnal of Applied hfe-
cl~anics.47:329-334.
28. Crossman, F. W,IV. J. Warren and A. S. D. Wang. 1983. "Influence of Ply Thickness on
Damage Accumulation and Final Fracture," in 1983Adrnnces in Aerospace Srrucrures, hlaterials
mid Dynamics,AD-Vo[ 06. U. Yuceoglu et al., eds. American Society of Mechanical En,'oineers,
pp. 215-226
29. Flaggs, D. L. and hl. H. Kural. 1982. "Experimental Determination of the In Sirii Transverse
Lamina Strength in GraphitelEpoxy Laminates," Joiunal of Conposire hfarerials, 16:103-116.
30. Nuismer, R. J. and S.C. Tan. 1982. "The Role of hIatrix Cracking in the Continuum Constitutive
Behavior of a Damaged Composite Ply," in hfeclianicsof Composite hfaateriafs:Recent Advances.

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015


Damage Tolemrice of Lmiitiated Composites: Par1 I-Atzafjsis 43

2. Hashin and C. T. Herakovitch, eds. 1UTithf S~niposiirin011 hlechairics of Composite hlore-


rials. BlacXsbttrg, VA, Aitgitsr 16-19. 1982, pp. 437-418.
3 I. Lagace, P. A. 1986. -Notch Sensitivity and Stacking Sequence of Laminatd Composites," Corn-
posite hfaterinls: Tesring mid Design (Seventh Conference).N h f STP 893. J. hl. Whitney, ed.
American Society for Testing and hlaterials, pp. 161-176.
32. Stinchcomb, W. W.,K. L. Reifsnider, P. Yeung and J: hlasters. 1981. "Effect of Ply Constraint
on Fatigue Damge Development in Composite hlaterial Laminates," Fatigite of Fibroits Coni-
posite hfaterials, AWhf STP R 3 . K. N. Lauraitis, ed. American Society for Testing and hlate-
rials, pp. 61-81.
33. Herakovich, C. T. 1981. "On the Relationship beta.een Engineering Properties and Delamination
of Composite hlaterials," Joitriid of Composite hfarerials, 15:336-318.
31. Talreja, R. 1985. "Transverse Cncking and Stiffness Reduction in Composite Laminates," Joitr-
rial of Compo;ite hfaterials, 19:355-375.
35. Garrett, K. \V. and J. E. Bailey. 1977. 'hlultiple Transverse Fracture in 90 Cross-Ply Laminates
of a Glass Fibre-Reinforced Polyester," Joirrrial of hkiteriol Science, 12:157- 168.
36. Herakovich, C. T. 1982. "Influence of Layer Thickness on the Strength of Angle-Ply Laminates,"
Joirrnal of Coiiipsire hfaterials, 16:216-227.
37. Crossman, F. W.and A. S. D. Mng. 1982. "The Dependyce of Transverse Cncking and Dc-
lamination on Ply Thickness in GraphitelEpoxy Laminates," Darnage in Coniposite hlateriols.
AWhI STP 775. K. L. Reifsnider, ed. American Society for Testing and hlaterials, pp. 118-139.
38. OBrien, T. K. 1982. "Characterization of Delamination Onset and Growth in a Composite Lami-
nate," Daimge in Coniposite hfaterials.AWhf57P 775. K. L. Reifsnider, d.American Society
for Testing and hlaterisls, pp. 110-167.
39. \Vang, A. S. D., H. hliller and P. C. Chou. 1982. *A Theory for Transverse Cracks in Composite
Laminates," in IW2 Arlvances iii Aerospace Strtrcntres and hfaterials.AD-l'olO3. R. hl. Lauren-
son and U. Yuceoglu, cds. American Society of hlechanical Engineers, pp. 51-60.
40. Peters, P. \V. hl. 1981. "The Strcngth Distribution of 90 Plies in 0/90/0 Graphite-Epoxy Lami-
nates," Joitnial of Coniposire hfarerials, 18:545-556.
41. Kistner, hl. D., J. hl. Whitney and C E. Browning. 1985. 'First-Ply Failure of GnphitelEpoxy
Laminates," in Rtcenr Adlances in Co~nposiresin rlre United Stores arid Japan. AWhf STP 864.
J. R. Vinson and hl. Taya, eds. American Society for Testing and hlatcrials, pp. 44-61.
42. Kim, K. S. and C. S. Hong. 1986. "Delamination Grmvth in Angle-Ply Laminated Composites,"
Jottriial of Coinposite dfaterials, 20:423-438.
43. Sun, C. T. and K. C. Jen. 1987. "On the Effect of hlatrix Cracks on Laminate Strength," Joitrrinl
of ReinJorced Pla~ricsand Composites, 6:208-222.
44. Han, Y. hl. and H. T. Hahn. 1988. "A Simplified Analysis of Transverse Ply Cncking in Cross-
Ply Laminates," Coniposites Science arid Pclinologj, 31:165-177.
45. Chang, F. K., J. hl. Tang and D. G. Rterson. 1987. "The Effect of Testing hlethods on the Shear
Strength Distribution in Lnminatcd Composites," Joitrtlnl of Reinforced Plastics arid Coniposites,
6:301-318.
46. Chang, F. K. and hi. H. Chen. 1987. "TIC I n Sirit Ply Shear Strength Distributions in Graphitel
Epoxy Laminated Composites," Jottrnol of Coinpositr hfaterials, 21:708-733.
47. PJgano, N. J. and R. B. Pipes. 1971. "The Influence ofStacking Sequence on Laminate Strength,"
Jotrrnal of Conipsite hfaterials, 5 3 - 5 7 ,
48. Pindera, hl. J., C Choksi, J. S. Hiddeand C. T. Herakovich. 1987. "A hlethodology for Accurate
Shear Characterization of Unidirectional Composites," Joirrnol of Composite hfaterials, 21:
1164-1181.
49. Lessard, L. B. and F. K. Chang. 1991. 'Effect of Load Distribution on the Fiber Buckling
Strength of Unidirectional Composites," Journal of Coinposire hfarerials, 25(1):65-87.
50. Was, A., C. Babcock and IV. Knauss. 1989. -An Experimental Study of the Initiation and Propa-
gation of Damage in Compressively Loaded Composite Laminates in the Presence of a Circular
Cutout," Proceedings of the AIAA/AShfE/ASCE/AHS/ASC30th Stnrcrtires, Stntcritral Dynamics
and hlaterials Coifererice. Part I t . hfobile. AL. April 1989, pp. IOOO-1011.

Downloaded from jcm.sagepub.com at UNIV OF PITTSBURGH on August 18, 2015

Vous aimerez peut-être aussi