You are on page 1of 10

Composites Science and Technology 64 (2004) 719728

www.elsevier.com/locate/compscitech

Dynamic damage initiation of composite beams subjected


to axial impact
Zheng Zhang, Farid Taheri*
Department of Civil Engineering, Dalhousie University, 1360 Barrington Street, Halifax, NS, Canada B3J 1X1
Received 30 January 2003; received in revised form 18 June 2003; accepted 24 July 2003

Abstract
The dynamic damage behavior of carbon-epoxy laminated beams, having initial geometric imperfections, subject to an axial
impulse was investigated numerically and experimentally. The study focused on investigating the damage initiation and damage
mechanism in the beams impacted axially by a moving mass. The dynamic equilibrium equations were developed based on the
Timoshenko beam assumption with the consideration of beams transverse inertia, transverse shear deformation and the cross
sections rotational inertia eects. The Higher-Order Shear Deformation Theory was adopted to model the nonlinear distributed
shear strain across the beam thickness. The von-Karman StrainDisplacement nonlinear relationship was used to model the
deformation of the beam. Hashins failure criteria was used to predict the damage of beams. The experiments were conducted using
a horizontal linear bearing impact setup. Scanning Electron Microscopy results showed that delamination and matrix crack were
the primary damage mechanisms in the beams. Eect of ber angle, lay-up sequence and initial geometric imperfection on critical
energy of damage initiation was also investigated.
# 2003 Elsevier Ltd. All rights reserved.
Keywords: B. Impact behaviour; C. Damage mechanics; C. Delamination; B. Modeling; D. Fractography

1. Introduction
Due to their high specic stiness and strength, berreinforced plastic (FRP) laminated composites have been
widely used in industrial applications such as aerospace,
automobile, shipbuilding, marine and civil infrastructures. However, their susceptibility to damage
resulted from mechanical, physical and chemical factors,
greatly degrades their stiness, strength and durability.
Impact in particular is one of the most signicant sources
of damage that can cause matrix crack, delamination and
ber breakage. Often damages generated in FRP are
undetectable to the naked eyes, therefore, it is particularly important to understand the damage mechanism
(including initiation and progression) in FRP.
Although a great number of investigations have considered FRP impact characterization, most of such
works [110] have considered on the damage due to
transverse impact. Choi et al. [1] conducted analytical
* Corresponding author. Tel.: +1-902-494-3935; fax: +1-902-4846635.
E-mail address: farid.taheri@dal.ca (F. Taheri).
0266-3538/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compscitech.2003.07.002

and experimental investigations of damage initiation of


graphite/epoxy composite plates subjected to transverse
line-loading impact, in which the eects of laminate lay-up
sequence and impactors mass were the focus. Pavier and
Clarke [3] proposed an experimental technique which
could be used to replicate the damage of composite laminates, which was transversely impacted by a drop-weight.
Zhou [4] conducted tests on thick glass-ber-reinforced
laminates under transverse impact by a at-ended impactor. Post-impact damage mechanism was investigated by
visual inspection and ultrasonic C-scanning techniques.
Sohn et al. [5] performed drop-weight impact damage test
of carbon-ber/epoxy composite and used several characterization techniques, such as cross-section fractography, scanning acoustic microscopy, scanning electron
microscopy (SEM), to observe and assess the damage due
to impact. Park and Zhou [6] investigated the transverse
impact response and damage in composite laminates by
obtaining time history results of contact force, displacement and energy absorption on a three-point bend xture
in a split Hopkinson pressure bar. With the photo elastic
stress coating technique, Franz [7] established an experimental method for investigating the dynamic response

720

Z. Zhang, F. Taheri / Composites Science and Technology 64 (2004) 719728

and damage behavior of composite plates due to impact


load. Necib and Mili [8] experimentally investigated the
dynamic behavior of various E-glass/epoxy laminates
subjected to a drop weight impact on the transverse side
of the specimens. Luo et al. [9] proposed an approach to
evaluate the impact damage initiation and propagation
in composite plates, both experimentally and numerically. The plates were impacted by a controlled drop
weight transversely and the damage and propagation
were inspected with an optical microscope and X-ray
chamber. Olsson [10] suggested an analytical model for
prediction of impact damage initiation and growth in
composite laminates. Critical loads and energies for
damage initiation and growth were discussed.
Most of the experimental and analytical works have
been based on transverse impact of composite laminates
due to a drop weight. Only a few investigations have
been reported on the response of composite laminates
subject to axial or in-plane impact.
Axially load bearing, slender structural components are
commonly found in various structures. These components
may easily buckle when subjected to static or dynamic
loads [11]. Due to manufacturing induced factors, many
composite laminates bear initial imperfections and/or
voids. When such a laminate is subjected to an axial or inplane static or dynamic load, regardless of whether the
component undergoes buckling or not, it may experience
damage in the form of delamination, ber breakage and
matrix crack if certain stress or strain components exceed
the limiting criterion during the loading period.
Using a falling weight impact system, Hsiao and Daniel
[12] investigated the strain rate eect on the compressive
and shear behavior of carbon/epoxy composite laminates. Bogdanovich and Friedrich [13] predicted the
initial failure and ply-by-ply failure processes of composite laminates under dynamic loading. Abrate [14] introduced an energy-balance model, spring-mass model and a
full model to simulate the impact in composite structures.
It can be seen that damage characterization in axial
composite members, subjected to impulse loading, has not
been thoroughly investigated. The scarcity of such experimental data thus motivated the present investigation.
The purpose of this paper is to therefore experimentally investigate the damage initiation mechanism and
type of damage in ber-reinforced composite laminated
beams subjected to axial impulse. The objective is also
to predict and simulate such damage mechanism by
computational simulations.

the analytical model developed by the authors [16] by


including the damping eects in the dynamic equilibrium equations. We considered a n-layer FRP composite laminated beam with one end impacted by a pulse
load as shown in Fig. 1. The cross section of the beam
was rectangular with width b and thickness h. The
length of the beam is L. The initial geometric imperfection is w0 x, which is dened as the initial displacement
of the beam in Z-direction as a function of location x.
With the Timoshenko beam assumption, the equilibrium equations were constructed for the FRP laminated
beam. The formulation includes the axial, lateral and
rotational inertias of the beam, the shear deformation
through thickness of the beam, and the damping eects in
axial, lateral and rotaional directions as follows:
I1

@2 u @Nx
@u
 CL

2
@t
@t
@x

 

@w0
@
N


x
@2 w @V
@w
@x

I1 2
 CT
@t
@x
@t
@x

I2

@2  @Mx
@
 V  CR

@t2
@t
@x

1a

1b

1c

where ux; t, an unknown, is the axial displacement of


the beam in x-direction; wx; t, an unknown, is the lateral displacement of the beam in z-direction; x; t, also
an unknown, is the rotation of the cross section of the
beam about the y-axis;
h=2
Nx
x dz
2a
h=2

are the axial force per unit beam width, x is the axial
stress of each lamina;
h=2
x zdz
2b
Mx
h=2

are the bending moment per unit beam width;

2. Analytical considerations and damage models


2.1. Dierential equations of motion
To investigate the dynamic behavior of composite
laminated beams subject to axial impulse, we upgraded

Fig. 1. Schematics of a FRP beam being impacted by a moving mass.

721

Z. Zhang, F. Taheri / Composites Science and Technology 64 (2004) 719728

h=2
V

xz dz

h=2

are the translational inertia per unit beam width, and 


is the mass density of the lamina.
h=2
z2 dz
4b
I2
h=2

are rotaional inertia per unit beam width;


CL ,CT ,CR are damping factors in axial, lateral direction
and rotation of cross section of beam, respectively.

2.2. Boundary and initial conditions


The boundary conditions for Eqs. (1a)(1c) can be
pinned-end or clamped-end supports or any reasonable
combinations with the provision that the impacted end
is allowed to move freely along the axial direction of the
beam. The impact can be due to a known impulse force
time history F(t), to a known mass M with initial velocity V0 of the impact body, or other types of impact
contact model. For our current problems, we will only
consider the xed end support conditions [i.e., both the
left end (x=0) and right end (x=L) of the beam are
xed, but the beam can roll in x direction at the right
end]. Thus, the following boundary conditions are
established,
@w
0
@x

u0

at

at

x 0; x L

x0

5a

5b

During the impact period, i.e., when Nx x L < 0


ut Ut

at

is the shear force per unit beam width across the crosssection of beam, and x is the shear stress of each
lamina;
h=2
dz
4a
I1

Vtdt;

Ut

h=2

x L

5c

in which,Vt is the velocity of the moving mass during


the impact period and can be represented by:
t
b
Vt V0 A1; 1 "x jxL dt
7
M
0
The beam is assumed to be at rest initially. Therefore,
the initial conditions of the current problem are
wxjt0 w0 x

8a

@w0
@x

8b

xjt0 
@w
j 0
@t t0

8c

@
j 0
@t t0

8d

ujt0 0x < L

8e

@u
j 0x < L
@t t0

8f

2.3. Constitutive relationship


The strain and stress relationship for the kth layer of
the laminated beam is,
8
9 2
< x =
Q 11
4
y
Q 21
:
;
xy
Q 61


yz
xz


Q 44
Q54

Q 45
Q 55



9
38
Q 16 < "x =
Q 26 5 "y
:
;

xy
Q 66

yz

xz

9a


9b

where Q ij are the transformed stiness terms.


For the slender laminated beam, we can assume
"y 0,
xy 0, and
yz 0, then Eq. (9a) and (9b)
simplify to:
x Q 11 "x

10a

and
xz Q 55
xz

During the post-impact period, the right end of the


beam is free to move in the axial direction (i.e.,
Nx x L 0), while the right ends displacement of
beam can be obtained through zero neutral axis strain.
In Eq. (5c), Ut is dened as the displacement of the
moving mass [15],

Q 12
Q 22
Q 62

10b

2.4. Second order shear deformation theory applied to


the current problem
As demonstrated by the authors in their earlier work
[16], higher order shear deformation theory (HSDT)

722

Z. Zhang, F. Taheri / Composites Science and Technology 64 (2004) 719728

provides more accurate solution for the case under the


investigation than that of the rst order shear deformation theory when evaluating the accurate distribution of
shear strain across the thickness. To predict the damage
initiation of composite laminated beams, one also needs
to obtain accurate strain and stress distributions across
the beam thickness when subject to axial impact. Using
Reddys higher order shear deformation theory [17],
and considering the initial geometric imperfection w0 x
of the beam, the displacement eld of the composite
beam can be dened as follows [16],


@w @w0
4z3

ux x; z; t ux; t z  
 2
@x
@x
3h

11a

uz x; z; t wx; t  w0 x

11b

From the above equations, one can see that the


transverse shear strain distribution through the thickness of the beam is parabolic. Not only the transverse
shear strain distribution is nonlinear, but HSDT
assumes that the relationship between the axial strain
and displacement, and the deection of the beam is also
nonlinear.
The von Karman straindisplacement relationship
was then used to model the nonlinear relationship,
 


@ux 1 @w 2 1 @w0 2
"x


12a
2 @x
@x 2 @x

xz

@ux @uz

@z
@x

12b

With the help of Eqs. (2)(4) and (10)(12), Eq. (1)


could be rewritten in the non-dimensional form and
solved using nite dierence method (FDM). One
could then evaluate the dynamic behavior, (i.e., time
history results of axial displacement, transverse
deection and rotation of cross section) of the FRP
laminated beam. Moreover, with help of the von
Karman straindisplacement relationship and the
constitutive relationship, the strain and stress time
history results of each layer of the beam can be
evaluated. Refs. [11] and [16] provide detailed
description and verication of the analytical model
and the FDM solution, and their integrity in predicting post impulse response of such beams.
2.5. Damage criterion
Three failure modes in the forms of ber failure,
matrix failure and delamination can be expected to
occur in the composite laminated components subject
to the dynamic load. For the one-dimensional laminated beams subject to axial impulse, we will not

consider the matrix failure along the width of the beam


and, therefore, only ber failure and through the thickness delamination of the beam will be considered in the
simulation of laminated beams subject to axial impulse
in this investigation.
2.6. Fiber failure
2.6.1. Tension mode

As Hashin [18] stated, both tensile stress, 11


and
shear stress, 13 of the bers aect the failure of

bers. The ber failure occurs if the tension stress 11


and shear stress 13 of bers satisfy the following
equations,
 2 

11
13 2
T

5 1:0
13
Ff
SLT
SL
in which, FfT is the failure index of ber in tension
mode, the SL is the tension strength of lamina and SLT
is the shear strength of the lamina in the 13 plane.
2.6.2. Compressive mode
Under compressive status, ber is assumed to fail in
microbuckling and kinking of bers with the matrix.
Rosen [19] stated that bers under axial compression
buckle in a shear mode provided that volume fraction of
bers is higher than a certain limit (i.e., 60% for carbon/epoxy). The shear and compressive stresses both
contribute to the compressive failure of bers. The
deviatoric strain energy theory, also known as the Tsai
Hill criterion [20], is similar to Hashins failure criteria
for bers in tension, that is:
  2 


13 2
FfC 11

5 1:0
14
SL
SLT

is the compressive stress, FfC is the failure
in which, 11
index of ber in compressive status, the SL is the compressive strength of ber.

2.7. Delamination
Delamination occurs in laminates due to the normal
and inter-laminar shear stresses. For the current onedimensional laminated beam subject to axial impulse, as
described by the dierential equations, the normal and
through-the-thickness stresses are ignored. The failure
criteria can then be described by [18],


31 2
Fd
5 1:0
15
STL
in which, Fd is the failure index of delamination and STL
is the through-the-thickness shear strength of the
lamina.

Z. Zhang, F. Taheri / Composites Science and Technology 64 (2004) 719728

3. Experimental setup
3.1. Experimental setup
Drop weight setup is a common mean for applying
impact load for dynamic testing, as reported in several
experimental works [110]. The disadvantage is that it is
dicult to eliminate the rebound impact. To overcome
this problem, a new impact setup was designed to perform the axial impulse, as shown in Fig. 2. The major
components of the system include a pendulum, a guiding tube with linear bearings along its full length, specimen support xtures and an impactor. The linear
bearings were horizontally xed on a strand. The linear
bearing, xture and specimen were calibrated horizontally so that the impact contact would be collinear.
When the pendulum hits the impactor, it forces the
impactor to travel through the tube on the bearings. By
this system the impactor does not rebound and impact
the specimen more than once. Two non-contact optical
sensors were used to monitor the velocity of the impactor. The rst optical sensor also acted as a trigger for
the data acquisition setup to collect the data. The extant
of damage in the specimens was inspected by visual
method and scanning electron microscopy (SEM).
3.2. Specimen preparation
The material used in this study was TENAX/R6376
carbon ber/epoxy pre-preg by HEXCEL. R6376 is a
high performance tough matrix formulated for the fabrication of primary aircraft structures. It oers high
impact resistance and damage tolerance. The laminate
was cured 2 h at temperature 175  C and 700 kN/m2
pressure with heating rate of 2  C to 5  C per min in an
oven, as recommended by the manufacturer. The laminate was then cut into beams. The mechanical properties of the laminate and geometric properties of beams
are listed in Table 1.

Fig. 2. Experimental setup.

723

Table 1
Physical and mechanical properties of carbon-ber/epoxy laminated
beams
Specimen length (mm)
Specimen width (mm)
Specimen thickness (mm)
Imperfection magnitude W0 (mm)
Longitudinal modulus E11 (GPa)
Transverse modulus E22 (GPa)
In-plane shear modulus G12 (GPa)
Major Poissons ratio (v12)
Density (Kg/m3)
Longitudinal tensile strength S+
L (MPa)
Longitudinal compressive strength S
L (MPa)
Transverse tensile strength S+
T (MPa)
Transverse compressive strength S
T (MPa)
In-plane shear strength SLT (MPa)
Through-the-thickness shear strength STL (MPa)

300
20
1.6
Factor thickness
1.18E+02
5.54E+00
4.77E+00
0.27
1512
1094.8
712.9
26.44
84.33
84.42
65.36

4. Results and discussion


To investigate the damage initiation behavior of carbon/epoxy composite laminated beams with initial geometric imperfections subject to axial impulse of a
moving mass, several groups of beams with dierent
lay-up congurations were fabricated and tested, and
also analyzed numerically.
The lay-up
congurations


 
in
this
study
were
0

22:5

45

67:5

12
3 s
3 s,



 3 s
0=903 s and 0=902 =02 s . The support condition of
the beam is shown in Fig. 1. The initial geometric
shapes of the beams were either in sinusoidal or random
forms [21,22]. Fig. 3 shows the typical initial geometric
imperfection forms. The imperfection values were
obtained using a measuring gauge after the beams were
mounted on the test xture (jig), before the impact tests
were conducted. The damping coecient of the beam is
approximately 0.04, as measured by a GrindoSonic
MK5 Industrial instrument [23].
4.1. Strain records
Fig. 4 shows typical strain values recorded by two
strain gages mounted on beams top and bottom surfaces, at the mid-span of the beam. When a geometrically imperfect beam is impacted axially, it would
vibrate in both axial and transverse directions. From
Fig. 4, we can see that both top and bottom surfaces
experience compressive strain at the beginning of the
event, because the beam is deformed axially due to the
impact compressive load. As the stress wave propagates
along the beam, the beam experiences a combined
bending and axial compressive loading, in which the top
surface of the beam is in compressive state, while the
bottom is in tension. Due to the damping eect, the
vibration diminishes gradually. From this gure, we can
see the maximum strain or stress that occurred during
the rst transverse vibration cycle. Considering the

724

Z. Zhang, F. Taheri / Composites Science and Technology 64 (2004) 719728

Fig. 5. Typical delamination and matrix crack damage.


Fig. 3. Initial out of plane geometric imperfection shapes of tested
specimens.

Fig. 4. Typical strain records.

dynamic pulse-buckling phenomenon [11], the damage


may occur during pre- and/or post-buckling phases.
4.2. Damage investigation
The damage mechanism of axially impacted FRP
beams diers from the damage of FRP due to transverse impact in which the damages, including delamination, ber breakage and matrix crack mainly occurs
around the contact area of the impacted region [35].
Under axial impact, the damage occurs at a location
along the beam or along the whole length of the beam.
Damage location through the beam thickness varies
depending on lay-up congurations. When impacted
axially, the beams with initial geometric imperfections
deform transversely and axially. The maximum tension
and/or compression strains occur at the outer plies.
According to the higher order shear deformation theory, the shear strain through beam thickness varies in
parabolic distribution and the maximum occurs on the
beams mid-plane. This phenomenon was demonstrated
in Fig. 5. Fig. 5 represents a typical demalination and
matrix damage. Initial voids due to manufacturing are
not negligible, and delmination and matrix crack occurs
usually on the weak regions.
For all the

 tested
 specimens, except those with 67:53 s and 012 lay-ups,
delamination and matrix cracking dominated the
damage mechanism. For most of the beams with
67:53 s lay-up, no delamination and matrix crack
were observed after impact, but the deformed shape was
maintained due to the plastic deformation of matrix, as
shown in Fig. 6(a) and (b). The details of such mechanism and prediction method will be presented elsewhere.

Fig.
6. Initial
and deformed beam shape of one of the beams with


67:53 s lay-up.



For those beams with 012 lay-up, matrix cracking
occurred normal to the ber direction, and the beam
was split into two or more parts.


4.2.1. Damage analysis of beams with 22:53 s lay-up
of beams, having lay up sequence of
 Three groups

22:53 s had three dierent lengths of 105, 128 and
148 mm. Each group had 4 specimens and each specimen was impacted with dierent impact energy.
Damage, mostly in the form of delamination, was
observed between plies 6 and 7. Some of the delaminations extended along the whole length of the beams but
was limited between the two restrained ends. Fig. 7(a)
and (b) show the time history results of axial displacement and lateral deection at station 0.7 of the beam
(0.7 indicates 7/10L distance from the cantilever support
end of the beam), obtained though our numerical analyses for one of the specimens. Fig. 7(c) shows the lateral deformed shape at time 0.4ms compared with the
initial shape of the beam. Fig. 8(a) and (b) show the
failure indices of ber breakage for each layer, and
delamination failure for each inter-laminar interface at
station 0.7 of the beam, obtained though Eqs. (13)(15).
From the gures, we can see that all ber failure indices
are less than 1.0, which indicates that the bers do not
break; some of the delamination failure indices are
however grater than 1.0, indicating that delamination
would have occurred. Further analysis of Fig. 8(b)
indicated that even though failure indices of interface 5

Z. Zhang, F. Taheri / Composites Science and Technology 64 (2004) 719728

725

Fig. 7. Numerical results of time displacements. (a) Time history of


axial displacement at station 0.7 of the beam. (b) Time history of
deection at station 0.7 of the beam. (c) Deformed shape and initial
shape of beam.

6, and 78 were greater than 1.0, but visual observation


did not conrm such a mechanism. This was because the
interface 67, which was at the beams mid-plane, met
the delamination criterion before the other interfaces. In
this case the other interfaces did not have the opportunity to delaminate, even though the failure factors were
greater than 1.0.


4.2.2. Damage analysis of beams with
453 s lay-up

Unlike in beams with 22:53 s lay-up in which the
delamination extended along the whole length of beam
and dominated the damage mechanism, the delamination in these beams occurred at several discrete stations
along the beam length and was predominantly accompanied by matrix cracking. Fig. 9 is a typical illustration
of delamination and matrix crack in those beams.
Fig. 10 is another example of this beam group, showing
a major delamination between the ply 6 and 7, which
was also connected to another delamination (between
ply 3 and 4) by a matrix crack, which was facilitated
though an initial void. Fig. 11 illustrates a typical
matrix cracking of this group.


In summary,
 most of the delaminations in 22:53 s
and 453 s beams occurred at the beams mid-span,
extending
 through
 the beams thickness. The specimens
having 0=903 s lay-up, however, exhibited dierent
failure patterns, as follows.

Fig. 8. Numerical results of the failure Indies. (a) Failure index for
ber breakage. (b) Failure index for delamination.

Fig. 9. SEM
showing typical delamination and matrix
 fractograph

crack of a 453 s specimen.

726

Z. Zhang, F. Taheri / Composites Science and Technology 64 (2004) 719728



4.2.3. Damage analysis of beams with 0=903 s lay-up
In this group of beams, the two 90 plies placed at
the beams mid-plane did not contribute much in
carrying the axial force. During the transverse deformation stage, outer plies endured most of the tensile
and compressive loads. The 3rd, 5th, 8th and 10th 0
plies would have endured the maximum shear strain.
Therefore, it is reasonable to see that delaminations
would occur between plies 12, 1112, 56 and 78,
as was the case. Moreover, the 56 interface delamination was connected to the 78 by a matrix crack
running through plies 6 and 7. Fig. 12(a) is an example of delamination between plies 1 and 2. Fig. 12(b)

shows the connection of the two delaminations, as


described above.


4.2.4. Damage analysis of beams with 0=902 =02 s
lay-up
Four specimens were tested within this group. Most
delaminations occurred at the mid-plane and delamination dominated the damage mechanism. Similarly,
delamination modes were very similar to those discussed
earlier.
4.3. Location of damage initiation
When a damage criterion is satised, the damage (ber
breakage, delamination and or matrix crack) would initiate due to excessive stresses. According to the visual
inspection and the associated numerical analyses of the
test specimens, delamination and matrix cracks are the
most dominating damage mechanisms observed in the
imperfect beams that were subject of our investigation.

Fig. 10. SEM fractograph showing typical interfacing of two


delaminations.

Fig. 11. SEM fractograph of a typical matrix crack.

Fig. 12. SEM fractographs. (a) delamination between ply 1 and 2. (b)
delamination between plies 5 and 6 connected to delamination
between plies 7 and 8 by a matrix crack.

Z. Zhang, F. Taheri / Composites Science and Technology 64 (2004) 719728

Once the delamination and or matrix crack occurred,


the energy absorbed by the beam was partly released by
generation of damage, and then, whether the delamination would propagate, it would depend on the amount
of energy absorbed by the beam. This phenomenon will
be discussed later, in another document. In here, we
discuss the initiation location of delamination, that is,
location along the beam and through inter-laminar
interfaces. Visual and microscope inspection of the tested specimens suggested that most delamination occurred between the supports. Numerical analysis results for
all the tested specimens indicated that initial geometric
imperfection has a signicant eect on the delamination
initiation position along the beam. Fig. 13(a) and (b) are
histograms showing the density of delamination based
on their relative distance along the beam, and their
interface location. From Fig. 13, one can see that beside
the initiation positions at station 0.0 (xed end) and 1.0
(impacted end) along the beam, most damage took
place around stations 0.3 to 0.4 (near to the xed end)
and 0.7 (near to the impacted end). The delamination
positions through the thickness were at inter-laminar 4,
5 and 6. The distribution of shear strain through the
beam thickness was maximum at the beam mid-span,
and minimum at the top and bottom surfaces of the
beams. Due to the dierence of ply orientation, however, the center plies did not necessarily experience a
maximum stress; therefore, a certain amount of delaminations occurred at the 4-5 interface.

727

Fig. 14. Variation of the critical energy for damage initiation as a


function of the axial stiness and slenderness ratio.

Observation of the experimental results also indicated


that beam having relatively large imperfection (i.e.,
W0 =h > 0:2), failed mostly at their ends; otherwise, failure occurred along the length of the beams.
4.4. Critical energy for damage initiation
When addressing the damage initiation of FRP laminated beams subject to axial impulse load, one should
consider the critical energy used for damage initiation,
an unavoidable phenomenon. The angle ply laminates
with relatively large axial and bending stiness will
endure greater axial and transverse deformation resistance when subject to impact, and thus would endure
larger stresses; the reverse is true for laminates which
have smaller axial and bending stiness. Moreover, the
larger the slenderness ratios, the larger the energy
required to initiate the damage. This can be seen from
Fig. 14, which was constructed based on the inspection
of the experimental results. The axial stiness on the
x-axis is normalized relative to that of the zero degree
(uniaxial) layers.

5. Conclusion
Dynamic damage behavior of FRP laminated slender
beams, having various lay-ups, subjected to axial impulse
was investigated experimentally and numerically. Several
factors (such as the beams axial transverse inertia, cross
sections rotational inertia, the non-uniform distribution
of shear stress across the beam cross-section, damping
eect, the nonlinear strain-displacement relationship
between axial strain and transverse displacement) were
accounted by our numerical solution.
Our experimental investigation subjected carbon/
epoxy laminated beams with dierent lay-up and slenderness ratio to axial impact load. The following conclusions could be drawn from the results obtained
through our numerical and experimental analyses:
Fig. 13. Histograms from all specimens of failure occurrence in the
FRP beams. (a) Delamination initiation location. (b) Lamina interface
number.

1. Delamination and matrix cracking were the


dominating damage mechanisms in the carbon/

728

Z. Zhang, F. Taheri / Composites Science and Technology 64 (2004) 719728

epoxy laminated beams subjected to axial


impulse.
2. The density and length of delamination(s)
depend on the lay-up sequences.
3. The delaminations were mainly concentrated at
stations 0.40.7 (along the beam length), and at
interface of layers 4 and 5. The damage
mechanism was also strongly dependent on nature of the initial geometric imperfections.
4. The critical energy for damage initiation varied
with the lay-up and slenderness ratio.

[7]

[8]
[9]

[10]
[11]

[12]

Acknowledgements

[13]

The nancial support of NSERC in the form of an


operating grant to the second author in support of this
work is gratefully acknowledged.

[14]

References

[16]

[1] Choi HY, Wang HS, Chang GK. Eect of laminate conguration
and impactors mass on trhe initial impact damage of graphite/
epoxy composite plates due to line-loading impact. J Compos
Mater 1992;26:80527.
[2] Christoforou AP, Yigit AS. Normalized impact response and
damage in a thin thin composite laminate supported by a rigid
substrate. J Compos Mater 1994;28:155373.
[3] Pavier MJ, Clarke MP. Experimental techniques for the investigation of the eects of impact damage on carbon-ber composites. Compos Sci Technol 1995;55:15769.
[4] Zhou G. Damage mechanisms in composite laminates impacted
by a at-ended impactor. Compos Sci Technol 1995;54:26773.
[5] Sohn MS, Hu XZ, Kim JK, Wlker L. Impact damage characterization of carbon ber/epoxy exompoasite with multi-layer reinfotrcement. J Compos B 2000;31:68191.
[6] Park SW, Zhou M. Time-resolved impact response and damage

[15]

[17]
[18]
[19]
[20]
[21]
[22]

[23]

of ber-reinforced composite laminates. J Compos Mater 2000;


34:879904.
Franz T. Photoelastic study of the mechanic behavior of orthotropic composite plates subject to impact. J Compos Struct 2001;
54:16978.
Necib B, Mili F. Impact behavior of cross-ply laminated composite plates under low velicities. J Compos Struct 2001;51:23744.
Luo RK, Green ER, Morrison CJ. An approach to evaluate the
impact damage initiation and propagation in composite plates. J
Compos B 2001;32:51320.
Olsson R. Analytical prediction of large mass impact damage in
composite laminates. J Compos A 2001;32:120715.
Zhang Z, Taheri F. Numerical studies on dynamic pulse buckling
of FRP composite laminated beams subject to an axial impact. J
Compos Struct 2002;56:26977.
Hsiao HM, Daniel IM. Strain rate behavior of composite materials. J Compos B 1998;29:52133.
Bogdanovich A, Friedrich K. Initial and progressive failure analysis of laminated composite structures under dynamic loading. J
Compos Struct 1994;27:43956.
Abrate S. Modeling of impacts on comosite structures. J Compos
Struct 2001;51:12938.
Ari-Gur J, Weller T, Singer J. Experimental and theoretical
studies of columns under axial impact. Int J Solids Struct 1982;
18(7):61941.
Zhang Z, Taheri F. Dynamic pulsebuckling and postbuckling of
composite laminated beam using higher order shear deformation
theory. J Compos B 2003;34:3918.
Reddy JN. A simple higher-order theory for laminated composite
plates. J Appl Mech 1984;51:74552.
Hashin Z. Failure criteria for unidirectional ber composites. J
App Mech 1980;47:32934.
Rosen BW. Mechanics of composite strengthening, Ch. 3. In:
Fiber composite materials, ASM, Metal Park, OH, 1965.
Hill R. A theory of the yielding and plastic ow of anisotropic
metals. Proceedings of the Royal Society 1948:193.
Lindberg HE, Florence AL. Dynamic Pulse Buckling Theory and
Experiment. Dordrecht: Martinus Nijho Publishers; 1987.
Kenny S, Pegg N, Taheri F. Dynamic elastic buckling of a slender beam with geometric imperfection subject to an axial impulse.
Finite Elements in Analysis and Design 2000;35:22746.
J.W. Lemmens, Inc., Operating instructions for the GrindoSonic
MK5 Industrial instrument. Avaialble: http://www.grindosonic.com.