Vous êtes sur la page 1sur 14

Chinese Journal of Aeronautics, (2017), 30(1): 447460

Chinese Society of Aeronautics and Astronautics


& Beihang University
Chinese Journal of Aeronautics
cja@buaa.edu.cn
www.sciencedirect.com

Stress analysis and damage evolution in individual


plies of notched composite laminates subjected
to in-plane loads
Hu Junshan a, Zhang Kaifu a,*, Cheng Hui b, Liu Ping a, Zou Peng a, Song Danlong a

a
The Ministry of Education Key Lab of Contemporary Design and Integrated Manufacturing Technology,
Northwestern Polytechnical University, Xian 710072, China
b
School of Mechanical Engineering, Northwestern Polytechnical University, Xian 710072, China

Received 23 February 2016; revised 5 April 2016; accepted 5 August 2016


Available online 21 December 2016

KEYWORDS Abstract This work aims to investigate local stress distribution, damage evolution and failure of
Damage evolution; notched composite laminates under in-plane loads. An analytic method containing uniformed
Failure mode; boundary equations using a complex variable approach is developed to present layer-by-layer stres-
Failure strength; ses around the notch. The uniformed boundary equations established in series together with confor-
Layer-by-layer stresses; mal mapping functions are capable of dealing with irregular boundary issues around the notch and
User-defined subroutine at infinity. Stress results are employed to evaluate the damage initiation and propagation of notched
composites by progressive damage analysis (PDA). A user-defined subroutine is developed in the
finite element (FE) model based on coupling theories for mixed failure criteria and damage mechan-
ics to efficiently investigate damage evolution as well as failure modes. Carbon/epoxy laminates
with a stacking sequence of [45/0/60/90]s are used to investigate surface strains, in-plane load
capacity and microstructure of failure zones to provide analytic and FE methods with strong val-
idation. Good agreement is observed between the analytic method, the FE model and experiments
in terms of the stress (strain) distributions, damage evaluation and ultimate strength, and the layer-
by-layer stress components vary according to a combination effect of fiber orientation and loading
type, causing diverse failure modes in individuals.
2016 Chinese Society of Aeronautics and Astronautics. Production and hosting by Elsevier Ltd. This is
an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

* Corresponding author. Carbon fiber reinforced polymer (CFRP) composites are


E-mail address: zhangkf@nwpu.edu.cn (K. Zhang). widely used in all fields of aerospace, automobile, electronic
Peer review under responsibility of Editorial Committee of CJA. power, and mechanical engineering for several advantages
offered over metals, ceramics, and plastics. These include low
density, robust specific strength, low thermal expansion, corro-
sion resistance and designable characteristics for lightweight,
Production and hosting by Elsevier

http://dx.doi.org/10.1016/j.cja.2016.10.022
1000-9361 2016 Chinese Society of Aeronautics and Astronautics. Production and hosting by Elsevier Ltd.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
448 J. Hu et al.

efficient structures. Cutouts, especially circular and elliptical This literature mostly concentrated on stress distribution in
shapes, commonly appear in composite structural components anisotropic plates as well as laminated composites, which are
due to requirements for stability, maneuverability, low weight equivalent anisotropic plates, thus the stresses around cutouts
optimization and accessibility of other systems. This is often were simplified to be uniform throughout, regardless of thick-
the case in aircraft where composite structures such as wing ness. It should be noted that composite laminates contain sev-
spars and thin walls are drilled for electronic wires and hydrau- eral plies which possess dissimilar properties due to different
lic pipes1 or to facilitate assembly operations.2 Other instances fiber orientations, so the stress distribution in individual plies
occur not as part of the initial design of essential structures but should be thoroughly investigated.
due to material defects or unexpected damage during a service The above stress research methodologies and conclusions
cycle.3 These are serious and dangerous cases since they are not establish a theoretical basis for damage and failure of anisotro-
supposed to be there. Whatever their reasons for being, the pic plates. As for failure analysis, the damage propagation pro-
integrity and continuity of fiber and matrix in the composite cess is often depicted based on progressive damage analysis
are destroyed; this makes the notch region the weakest part (PDA) utilizing numerical and experimental methods.18 Lapc-
of the structure and causes serious local stress concentrations zyk and Hurtado19 proposed an anisotropic damage model
in the vicinity. As a consequence, structural capacity is reduced suitable for predicating failure and post-failure behavior in
and uncertain reactions to external loads may occur and lead fiber-reinforced materials. The plane stress formulation is used
to unexpected fractures in service. In order to avoid potential and the response of the undamaged material is assumed to be
safety hazards, accurate analysis of local stress levels and dam- linearly elastic. The evaluation law is based on fracture energy
age evolution of notched laminates are of great significance to dissipation and implemented in a finite element code. Zahari
the utilization of the material and lay the foundation for engi- and EI-Zafrany20 developed a progressive analysis algorithm
neering applications. based on Tsai-Hill failure to model the non-linear material
The determination of stress fields in notched anisotropic behavior and capture the compressive response of woven
plates has been the focus of many scholars for a long time. glass/epoxy composite plates via non-linear finite element anal-
Muskhelishvili4 first introduced complex potential theory to ysis. Rakesh et al.21 introduced a generic finite element model
the isotropic elastic plates and successfully obtained an accu- to investigate the failure of unidirectional glass fiber reinforced
rate solution of stress distribution. Analytical solutions for plastic (UD-GFRP) composite laminates with drilled holes
the stress distribution around holes of different shapes in ani- under tensile testing and compared the results with experimen-
sotropic plates were given by Lekhnitskii5 using series meth- tal work done earlier. Effects of joint geometry and stacking
ods. Savin6 presented a much simpler approach by sequence on the bearing strength and damage mode were
conformal mapping of Cauchy integrals. Gao7 used a biaxial investigated by Ondurucu et al.22 and specimens were exam-
loading factor together with an arbitrary orientation angle pre- ined for failure modes using a scanning electron microscope
viously used to solve the problem of a plate with biaxial load- (SEM). Kim et al.23 employed rosette strain gauges to measure
ing at infinity, avoiding the superposition of a solution with the strain around the joint hole during insertion of stainless
two uniaxial loading problems. The stress analysis research steel pins into glass fiber reinforced plastic (GFRP) specimens,
above established the base for later study of notched compos- and results were compared with the finite element method. Sat-
ite laminates. Ukadgaonker et al. developed a general solution apathy et al.24 presented a modified fiber failure fatigue model
for stress around oval holes,8 triangular holes9 with rounded to characterize the behavior of laminated composites with a
corners10 and irregular shaped holes11 in cross-ply and angle- central circular hole under in-plane fatigue loading. Martins
ply orthotropic plates under in-plane loading by superposition et al.25 studied the influence of diameter and thickness on fail-
of two stage solutions for boundary conditions. The stresses ure pressure during tube burst tests and employed the progres-
were employed together with a series of failure criteria, which sive failure analysis using a damage model by a user subroutine
were an extension of the Von Mises criterion for quadratic (UMAT) implemented in ABAQUS software to understand
interaction, to calculate the first ply failure (FPF) strength. the behavior of composite tubes under internal pressure. Lee
Similar solutions around a rectangular hole in an infinite iso- et al.26 proposed an evaluation method for the progressive fail-
tropic and anisotropic plate were given by Pan et al.12 and ure of composite laminates built upon Puck failure criterion by
Rao et al.,13 respectively, using the complex variable method. implanted UMAT to efficiently analyze the progressive failure
Sharma14 suggested general solutions for determining the phenomenon in glass/carbon fiber-reinforced composite lami-
stress distribution around polygonal holes and investigated nates. Compression tests were performed by Aljibori et al.27
the effect of hole geometry and loading pattern on the stress on 16 fiber-glass laminated plates with and without circular
concentration factor. Batista15 used a modified solution to cut-outs to investigate the effects of varying the centrally
solve problems of stress distribution around polygonal holes located circular cut-out sizes and fiber angle-ply orientations
of complex geometry in an infinite plate subjected to uniform on the ultimate load. Similar experiments were also conducted
loading at infinity. Remeepazhand and Jafari16 also studied the by Abu et al.28 to investigate the influence of cut-outs on multi-
central polygonal hole problem in composite plates using layer Kevlar-29/epoxy composite laminated plates.
two simple equations with parameters k, c, n and w controlling These remarkable works contribute a lot to the failure anal-
the size, shape and bluntness of corners. Toubal et al.17 exper- ysis of notched anisotropic composites whose layups are quasi-
imentally investigated the tensile strain field of composite isotropic, cross-ply, and angle-ply. At the same time, numeri-
plates in the presence of stress concentration caused by geo- cal and advanced experimental methods for progressive failure
metrical cutouts consisting of circular holes by Electronic have greatly developed. However, the failure criteria employed
Speckle Pattern Interferometer (ESPI). The stress obtained in in the literature were linear expressions; whereas composite
experiments is consistently lower than the analytical and laminates in special application are not quasi-isotropic
numerical models. lay-up, composite materials accumulate damage in shear,
Individual plies of notched composite laminates 449
8 r
leading to nonlinear stress-strain relations. The relationship >
> r1 k 1 k  1 cos 2b
> x
> 2
between stress distribution and damage evolution thus needs < r
further investigation. r1 k 1  k  1 cos 2b 2
>
>
y
2
In the present study, an analytical model for layer-by-layer >
: s1 r k  1 sin 2b
>
stress around notches in individual plies of general composite xy
2
laminates is investigated using a complex variable approach
that contains uniformed boundary functions. The functions where r1 1 1
x , ry and sxy are x-direction, y-direction and shear

established in series together with conformal mapping func- loading components along xoy axes, respectively. Through
tions are capable of dealing with irregular boundary issues the variation of parameters k and b, the uniformed external
both around the notch and at infinity. A stress solution is loading form Eq. (2) covers several load conditions, namely
employed to evaluate the progressive damage and failure of uniaxial (k 0), biaxial (k 0, b0) and shear stress
notched laminates based on coupling theories for mixed failure (k 1, b = p/4 or 3p/4), as well as their combinations.
criteria and damage mechanics. A user-defined subroutine is In states of generalized plane stress for laminates, according
then developed in ABAQUS to simulate the damage evolution to Lekhnitskii,5 the generalized Hookes law is valid when
under tensile and compressive loading. Experiments for strain analysis is limited to the linear elastic range. Hookes law in
and failure are conducted using a 3D digital speckle measure- terms of mean stresses rx;y and mean strains ex;y is given by:
ment system and microscopy study to validate the proposed 2 3 2 32 3
ex a11 a12 a16 x
r
analysis method. 6 7 6 76 7
4 ey 5 4 a12 a22 a26 54 r y 5 or ex; y aij  r
 x; y 3
2. Analysis for local stress distribution cxy a16 a26 a66 sxy

2.1. Problem configuration of notched laminates where ex , ey and cxy are x-direction, y-direction and shear mean
strain components, respectively. r  y and sxy are x-direction,
x , r
y-direction and shear mean stress components, respectively. aij
A composite laminate with a traction free notch under inplane i; j 1; 2; 6 are effective coefficients of the compliance matrix
loading is investigated in Fig. 1. The inplane loading is in the for symmetric multilayered composite plates.
form of stress resultants in an arbitrary coordinate system x0 oy0 In the absence of body force, the mean stress components
rotated clockwise by an angle b from the global coordinate sys- x; y can be expressed in terms of Airys stress function Ux; y:
r
tem xoy, as shown in Fig. 1(a). D is the diameter of the circle
that encloses the notch. It can be written in terms of Gaos 8
>
> @2U
loading conditions7 using stress loading r1 >
> x 2
r
x0 ; y0 as follows: > @y
8 1 >
>
<
< ry0 r
>
r
y
@2U 4
r1 >
> @x2
x0 kr 1 >
>
>
: s1 0 >
>
: sxy @ U
2
x0 y0
>
@x@y
where r1 1 1 0 0
x0 , ry0 and sx0 y0 are x -direction, y -direction and shear
loading components, respectively, r is applied stress in y0 direc- In order to ensure the integrity and continuity during defor-
tion, and k is biaxial loading factor defined by the ratio of r1 x0
mation, the mean strains ex; y should meet the equilibrium
to r1
y0 .
equation:
The notched plate is projected to a unit circle on f plane @ 2ex @ 2ey @ 2cxy
under normalized loading in Fig. 1(b) with an angle h to 2 5
@y2 @x @x@y
denote hole position. The equivalent loading r1 x; y along xoy
axes corresponding to original loading along x0 oy0 axes is By substituting Eqs. (3) and (4) into Eq. (7), a forth order
obtained by transformation of axis: differential equation is obtained:

Fig. 1 Problem configuration and solution of a notched composite under in-plane loading.
450 J. Hu et al.

@4U @4U @4U @4U Through substituting Eqs. (9) and (11) into Eq. (4), the
a22  2a26 3 2a12 a66 2 2  2a16 parameters B, B0 and C0 are obtained by the following equa-
@x4 @x @y @x @y @x@y3
tion group:
@4U 2 2   32 3 2 1 3
a11 0 6 21 l22 l
l1 l 22 i l22  l
22 B rx
@y4
6 76 0 7 6 0:5r1 7
The general solution of bi-harmonic Eq. (6) is determined 4 1 1 0 54 B 5 4 y 5 12
by roots of the following characteristic equation:  1 l2 l
l1 l 2 il2  l 2 C0 s1
xy

a11 l4  2a16 l3 2a12 a66 l2  2a26 l a22 0 7


2.2.2. Boundary conditions at the notch edge
In general, it can be proved that Eq. (7) has four distinct
roots, which are always complex conjugates in pairs: At the notch edge, which is traction free, the notch area on the
8 z plane is projected onto the exterior of a unit circle on f com-
> l1 a1 ib1 plex plane using the conformal mapping method (Fig. 1). The
>
>
< l a ib unit circle is defined by the Euler formula f eih in Fig. 1(b).
2 2 2
8
>
> 
l a1  ib1
This opens up the possibility of a simple general approach to
>
:
1

2 a2  ib2 determine the stress functions of the actual notch contour


l
independently.
where a1 , a2 , b1 , b2 are real numbers and b1 > 0, b2 > 0. l1 , l2 The mapping function for the elliptic notch in anisotropic
and their conjugates are complex parameters of a plane stress cases from z plane to f plane is defined by:
state. They are determined by effective compliance coefficients      
R 1 1
aij and can be used to describe the degree of material zp xp f 1 m1 f i1  m1  f lp
anisotropy. 2 f f
Through the integral operation of Eq. (6), the stress func- p 1; 2 13
tion Ux; y can be expressed by the real part of two complex
where R a b=2 is a constant for size of the hole, and
functions F1 z1 and F2 z2 : m1 a  b=a b is defined to control the ellipticity.
Ux; y 2ReF1 z1 F2 z2 z1 ; z2 2 Sh 9 Parameters a and b are major and minor axes of the elliptic
hole, respectively. These are typically: (1) m1 0 or a b: cir-
where zp x lp y p 1; 2 are anisotropic complex coordi-
cular holes; (2) 0 < m1 < 1 or ab: elliptical holes; (3) m1 ! 1
nates, and Sh represents the boundary conditions of the lami- or a  b: blunted cracks.
nate in Fig. 1(a). So, the problem is reduced to finding two By the projection process and taking Eq. (13) into the uni-
complex analytical stress functions that satisfy the boundary formed boundary Eq. (10) on f plane, the boundary equations
conditions at the notch edge as well as infinity. in f plane are written as:
8
2.2. Determination of analytic stress functions > X
n
< uf uz1 c lnx1 f Bx1 f Ak fk
>
:
k1
Pn
Here, in the issue of plane stress state, the uniformed boundary 0 0 0 k
wf wz2 c lnx2 f B iC x2 f k1 Bk f
equations uz1 and wz2 are developed in the form of deriva-
14
tives of F1 and F2:
8 where parameters Ak , Bk k 1; 2; . . . are constants of stress
dF1 X1 an
>
> uz1 c ln z1 Bz1 functions. The resultant stress boundary conditions are deter-
< dz1 n1 zn
mined by mean resultant forces XS in x-direction and YS in
X1 bn z1 ; z2 2 Sh
1
>
> dF y-direction:
: wz2 2 c0 ln z2 B0 iC0 z2
dz2 n1 zn
2
( R
2Rel1 uf l2 wfS S X ds
10 R 15
2Reuf wfS  S Y ds
where parameters c and c0 are complex variables of the notch
boundary. Parameters B, B0 and C0 are loading condition con- It is seen that the boundary of the notch is free of traction,
tents. Parameters an and bn are series coefficients determined which indicates that the mean resultant forces on the notch are
by the boundary conditions Sh around the cutouts and at zero:
(
infinity. X0
16
Y0
2.2.1. Remote boundary conditions
Parameters c and c0 are confirmed to be zero, and the log-
At the infinite plate edge (jzj ! 1), the mean stress compo- arithmic terms drop out. The rest of the parameters in series
nents reach the applied stress loading of the normalized terms are obtained by substituting Eqs. (14) and (16) into
boundary condition: Eq. (15):
8
>  x r1
lim r 8
>
> x!1 x
> l2 K1 K2 K3 K4 m1 l2 K1 K2 K3 K4
< < A1
> l1 l2
y r1
lim r 11
> y!1 y
B  l1 K1 K2 K3 K4 m1 l1 K1 K2 K3 K4 17
>
> lim s s1 >
> 1 l1 l2
: :
y
x!1; y!1 xy
Ak Bk 0; k 2; 3; 4 . . .
Individual plies of notched composite laminates 451

where K1 , K2 , K3 and K4 are complex constants. They are where rL , rT and sLT are longitudinal, transverse and shear
determined by: stress components in principal directions of the kth ply, respec-
8 tively. m and n are parameters in stress transformed matrix Tk
> R
>
> K1 B1 il1 B0 iC0 1 il2 k
>
> 2 decided by fiber orientation angle of the kth ply. Qij
>
>
>
> R i; j 1; 2; 6 is the transformed reduced stiffness matrix for
< K2 B1  il1 B0 iC0 1  il2
2 18 the kth layer. Eq. (22) can be written in contracted notation as:
>
> R 0 0
>
> K l B1 il l B iC 1 il k k
>
>
3
2 1 1 2 2
rL; T Tk  Qij  aij  r
x;y 23
>
>
> R
: K4 l B1  il l B iC 1  il
0 0
2 1 1 2 2

3. Analysis for damage and failure


2.3. Local stresses around notches of individual plies
As mentioned before, one of the main purposes of determining
local stress distribution in individual plies around the notch is
By substituting the analytic stress functions into Eq. (4), the
to investigate the damage and failure of the perforated lami-
mean stress components around the hole in global coordinates
nate accurately. In progressive damage analysis (PDA) the-
(Cartesian coordinates in Fig. 1) are obtained through the fol-
ory,31 composites are materials characterized by complicated
lowing equations:
8 failure phenomena that interact. The onset of failure in an
 0 0 
> 2 u f 2 w f
> individual ply generally does not lead to the structures col-
>
> x
r 2Re l l
>
>
1 0
x f 2 0
x2 f lapse, and this may still not occur until the failure has spread
>
>  0 1 
< u f w f 0 to each of the multiple plies, indicating that some plies fail
y 2Re 0
r 0 19 before others. Since the notch region is the weakest part of a
>
> x1 f x2 f
>
>   composite plate, the damage initiates around the notch.
>
> u0 f w0 f
>
> sxy 2Re l1 l2 0
: 0
x1 f x2 f
3.1. Failure criterion
According to the classical laminate theory29,30, laminate
strains can be written in terms of the midplane strains e0x;y Most failure criteria are extended from the von Mises criterion
and midplane curvatures jx;y as follows: for quadratic interaction.32 They allow the failure of a whole
ply to be checked, but ignore the distinct failure modes of com-
2 3 2 0 3 2 3
ex ex jx posites. Given this, the mixed failure criteria, which combines
6 e 7 6 e0 7 6 7 Hanshin criteria and maximum criteria, differs greatly from
4 y 5 4 y 5 z4 jy 5 or ex;y e0x;y zjx;y 20
them. It not only allows distinguishing between failure in ten-
cxy cxy
0
jxy sion compression as well as fiber and matrix, but also takes
nonlinear shear stress-strain behavior into consideration. The
where ex , ey and cxy are x-direction, y-direction and shear mixed failure criteria33 are given by:
strain components in z thickness layer, and e0x , e0y and c0xy are  2 
x-direction, y-direction and shear strain components in the rT 2s2 G12 3as4LT
emt LT2 24
midplane. jx , jy and jxy are x-direction, y-direction and tor- Yt 2Sc =G12 3aS4c
sional curvatures in midplane, respectively. 
 2
For the symmetric laminates studied here, it can be proven rT 2s2LT G12 3as4LT
that the coupling stiffness coefficients Bij i; j 1; 2; 6 are emc  25
Yc 2S2c G12 3aS4c
zero, which indicates the resultant and moment terms are
uncoupled. Hence the curvature of midplane jx;y is equal to  2 
rL 2s2 G12 3as4LT
zero. So, the strain at an arbitrary point of the notch edge eft LT2  26
throughout its thickness is the same as midplane strain and Xt 2Sc G12 3aS4c
the mean strain:  2
rL
ex; y e0x; y ex; y 21 efc 27
Xc
The stress components in the kth ply in principal direction
k  2 
rL; T are calculated by the following equation modified from rL 2s2LT G12 3as4LT
efs  28
Stress-Strain Equation and the stress transformation of axis: Xc 2S2c G12 3aS4c
2 3 2 2 32 3
rL m n2 2mn Q11 Q12 Q16 where emt , emc , eft , efc and efs are failure indices that denote
6 7 6 76 7 matrix tensile, matrix compression, fiber tensile, fiber compres-
4 rT 5 4 n2 m2 2mn 5 4 Q12 Q22 Q26 5
sion and the fiber-matrix shear-out, respectively. The parame-
sLT k mn mn m2  n2 k Q16 Q26 Q66 k ter a is a material constant that is determined experimentally
2 32 3
a11 a12 a16 x
r according to Chang and Lessard34 and commonly takes the
6 76 7 value of 2.44  108 MPa3 in the Users Manual of Abaqus
 4 a12 a22 a26 54 ry 5
a16 a26 a66 sxy 6.11. Once an index exceeds 1.0, the corresponding failure
occurs. The terms Xt , Xc are tensile and compressive strengths
22 in the fiber direction and Yt , Yc are tensile and compressive
452 J. Hu et al.

strengths in the traverse direction. Sc is the in-plane shear properties are given in Table 2. The specimen configuration
strength. for the open hole tensile (OHT) test is illustrated in Fig. 2
according to ASTM D5766M-02a Standard.36 The open hole
3.2. Material property degradation rules compressive (OHC) test followed ASTM D 6484/6484M-
04.37 The specimen geometry was similar to that used for the
From the phenomenological viewpoint, the defect initiation/ OHC test. The aspect ratio met the condition W=D P 4 so
propagation phenomenon for a material is strongly related that the external boundary effect was negligible.38
to the degradation of the materials capacity, and the most vis- In the FE model, the 3D eight-node layered continuum
ible aspect is the degradation of the materials stiffness. There- shell element was employed for the laminate. The mesh was
fore, it is possible to describe the progressive failure of a structured using linear elements with reduced-integration
composite laminate by numerically representing the correla- (SC8R), and the notch region was highly refined by partition
tion between an increase in the materials internal damage to capture the high near-hole stress gradients. A planar mesh
and a decrease in the materials stiffness. The Camanho and size of 0.188 mm  0.275 mm in (in x-y plane) in the vicinity
Matthews degradation rules35 (Table 1) are adopted because of the hole was used while a coarser mesh was used away from
they are closest to the real conditions according to the research the zone of the interest to computation cycles. In the thickness
of Chang.34 Parameters E1 , E2 , G12 and t12 are the longitudinal direction, a single layer of elements was used to model each
Youngs modulus, transverse Youngs modulus, shear modu- ply. The approximate element size was 0.206 mm 
lus and Poisons ratios of the lamina. E01 , E02 , G012 and t012 are 0.188 mm  0.375 mm. Successive space discretization was
material properties after degradation. compared, varying the element size surrounding the hole to
verify that the final mesh size was enough to provide accurate
3.3. Progressive failures of notched laminates results without expensive computational cost. A total of 92,345
eight-node brick elements were employed to model the speci-
men. The specimen was encastre on one side and applied ten-
After using Eq. (23) to obtain the stress distribution around
sile or compressive stress loading on the other side. Two lines,
the notch in each individual ply, the mixed failure criteria is
between which the length was L-2S, were marked to define the
employed to check the failure state around the notch edge.
elongation corresponding to the part of the specimen that was
On the occurrence of some sort of failure, the failing ply
not griped. The mesh and boundary conditions (BCs) are
should be degraded according to Table 1, which causes redis-
shown in Fig. 3.
tribution of stresses around the notch. Thus the failure state
The progressive model is implemented in ABAQUS by
of the notch edge will be rechecked. The procedure of stress
applying a separate subroutine called user-define field
redistribution and failure recheck will continue until all plies
(USDFLD).39 In the analysis, the material properties depend
fail or fiber failure occurs, causing sharp decline of carrying
on six field variables (FVs). The first five FVs from FV1 to
capacity. So, the location of damage initiation and final failure
FV5 represent failure indexes corresponding to emt , emc , eft ,
strength can be obtained during the analysis. Failure strength
efc and efs . The last one, FV6, denotes the shear damage
here refers to the maximum permissible remote stress loading
parameter. The algorithm is illustrated in Fig. 4. In the begin-
applied.30
ning, the FVs and the solution-dependent state variables
(SDVs) are set to 0 and the material properties are equal to
4. Numerical implementation their initial values. Then the load is increased gradually; For
each load increment, several iterations are necessary before
The stress distribution, damage and strength of notched spec- the analysis converges to an equilibrium state. At the end of
imens were also investigated using FE analysis implemented in each increment, stresses, failure indices and shear damage
the commercial software ABAQUS 6.11/standard. A compos- parameter are computed at the integration points. The value
ite laminate with a general stacking sequence of [45/0/60 of the shear damage parameter is directly assigned to the cor-
/90]s was chosen as the investigated object, and the material responding FVs and used to degrade the shear modulus. The
values of the failure indices are stored as SDVs. Once an
SDV exceeds 1.0, the corresponding FV is set to 1 and main-
Table 1 Property degradation rules of Camanho and tains this value until the end of the analysis. Meanwhile, the
Matthews. material properties are automatically reduced according to
Failure mode Property degradation rules the degradation rules in Table 1. With each increasing applied
load, the procedure is repeated until the program terminates
Matrix tensile failure
  E02 0:2E2 ; m012 0; G012 0:2G12
k
due to excessive element distortion. Since equilibrium will
rL > 0 not be reestablished after degradation of properties, small load
Matrix compression failure
  E02 0:4E2 ; m012 0; G012 0:4G12 increments should be set to obtain an accurate solution.40
k
rL < 0
Fiber tensile failure E01 0:07E1 ; m012 0 5. Experimental investigation
 
k
rT > 0
Fiber compression failure E01 0:14E1 ; m012 0 The damage and failure behavior of notched composite lami-
k
(rT < 0)
nates is also investigated on a Bairoe electronic universal test-
Fiber-matrix shear-out failure m012 0; G012 0 ing machine with a load capacity of 100 kN. The tests were run
  in displacement control at a rate of 2 mm/min. Loading
k
rL < 0
stopped after dropping 30% from the peak load. A non-
Individual plies of notched composite laminates 453

Table 2 Material properties of carbon-epoxy T300/3526.


E1 (GPa) E2 (GPa) t12 G12 (GPa) Xt (MPa) Xc (MPa) Yt (MPa) Yc (MPa) Sc (MPa)
133 10.4 0.29 4.14 1830 872 37.7 137 69.7

Fig. 2 Specimen configuration of composite for open hole test.

Fig. 3 Mesh and BCs of the FE model.

Fig. 5 Test machine with digital speckle measurement system


and specimen processing.

Fig. 4 Algorithm for ABAQUS user-defined subroutine. is based on a 3D digital image correlation (DIC) system with
binocular stereo vision technique that together perform full-
contact measurement method, 3D digital speckle measurement field 3D measurements of surface strain and out-of-plane
system (DSMS), was used to investigate the strain field in the deformation of the joints, as shown in Fig. 5(a). The principle
vicinity of the hole in the notched composite plate. The system is to track an applied surface pattern during loading by
454 J. Hu et al.

continuously taking digital images of the surface.41 Two strains fade much more sharply along the radial direction at
charge coupled device (CCD) cameras and two light emitting the points where the strain concentrations are the most serious
diode (LED) sources are required for measurement. To ensure at the edge of the hole, so the obtained strains from ring belt
data synchronization, DSMS was connected with the testing near these points are apparently smaller than those from the
machine by a synchronous data transmission device. hole edge.
The CFRP laminate was fabricated through hand lay-up The small error between analytical and FE results is due to
method using resin prepregs, and specimens were processed the algorithm in the FE method, which is a numerical and
with a water jet cutter to avoid defects caused by cutting heat. approximate solution as compared to an analytical solution.
The material properties and specimen geometry are the same Actually, the strains under tension and compression possess
as the FE model in Section 5. Holes were drilled and reamed the opposite distribution law due to the tensile and compres-
with carbide-tipped tools to attain fine surface quality, .then sive deformation state of specimen in either the x or y axial
surfaces of the specimens were polished with sandpaper (first direction. For strains in the x axial direction, maximum strain
using a grain size of 800 then a grain size of 2000) to ensure occurs at the 90 point where by pass load goes through net
a smooth surface. All the specimens were cleaned with alcohol tension plane and causes larger strain than any other part of
to remove grease and dust before the test. A white background the specimen. As for strains in the y axial direction, extreme
was painted on the testing surface of the specimens using spray points occur at regions around 0 and 90 due to transverse
paint, and later small black spray paint droplets were subtly contracting under tension (Fig. 6(b))or transverse stretching
applied with an appropriate density to make a high contrast- under compression (Fig. 7(b)). Extreme points at 45 and
pattern, as shown in Fig. 5(b). Images of speckle pattern were 135 are caused by shear effect.
recorded at a frequency of 5 Hz. The VIC 3D software calcu-
lated the accumulated strain by comparing the subsequent 6.2. Stress distribution around the hole in individual layers
images with the first image, which was taken as a reference.
The layer-by-layer stress distributions under tensile and com-
6. Results and discussion
pressive loading are provided separately in Figs. 8 and 9 to give
clear insight into the mechanical behavior of notched compos-
6.1. Experimental verification of the strain distribution around ite laminates. Good agreement is observed between theory and
the notch FEA, thus providing confidence in the accuracy of the present
results. Stress components in individual plies are shown to be
The strain distributions around the notch on the surface of the highly non-uniform throughout the laminate thickness. The
specimen under tensile and compressive loading are presented normalized longitudinal stresses in an individual layer have a
in Figs. 6 and 7, respectively. The test was repeated three times, much larger magnitude than the transverse and tangential
and the strains in the vicinity of the open hole were extracted stresses, reflecting the specific characteristic of anisotropic
by DSMS (Fig. 5(b)) to verify the validation of the theory and material in which carbon fibers bear most of the load.
FE model. Since the strains around the notch are center sym- It is seen that the stress concentration level in a layer
metric, only those whose angular position h ranges from 0 to depends on the relative angle between the fiber orientation
180 are revealed. and loading direction. The same dependence exists for concen-
The experimental results show relatively good trending tration regions. For longitudinal stress, the smaller the angle is,
alongside the proposed analytical solution and FE method the higher concentration level found near the region where the
with acceptable amplitude errors. There are two possible rea- fiber direction is tangential to the hole perimeter. This can
sons for the errors: Either the open hole is a singularity at explain why the 0 layer exhibit the highest stress peaks of
the surface of the specimen, or the fragmentary grid on the 8.65 near 91.8 and is the most effective in carrying the axial
hole boundary cannot be used to calculate the strain field.42 load while the 90 layer has the smallest concentration level of
In fact, the strains obtained by DSMS are the same as for 2.25 at 0. The transverse stresses in each layer exhibit peaks
the ring belt, which is a small offset from the hole edge. The at locations where the fiber direction deviates most from the

Fig. 6 Strain distribution around circular notch of composite laminates under tensile loading.
Individual plies of notched composite laminates 455

Fig. 7 Strain distribution around circular notch of composite laminates under compressive loading.

Fig. 8 Stress distribution around circular notch in individual layers of composite laminates under tensile loading.

Fig. 9 Stress distribution around circular notch in individual layers of composite laminates under compressive loading.

loading direction or where the fiber direction is normal to the steps (IS). Some of the important damage initiation locations
notch boundary. So the 90 layer shows the largest concentra- were also predicted using the analytical method in section
tion of 0.62 near 90 while the 0 layer has a level of 0.18 3.3. A microscopy study that investigated micro-scale failures
near 0. Actually, there is adverse distribution rule between occurring in the vicinity of the hole using scanning electron
longitudinal and transverse stresses. For tangential stress influ- microscope (SEM) is shown in Fig. 12 to validate the damage
enced by the shear effect, higher degrees of concentration are mode simulated in the FE model.
found in layers whose fiber orientation possesses an angle in Fig. 10 shows the failure mode (i.e., matrix tensile, matrix
the loading direction. compress, fiber tensile, fiber compress, and fiber-matrix shear)
in different ply orientations under tensile loading. Matrix ten-
6.3. Damage initiation and propagation of notched composite sile failure, which onset first at 92.72 in the 90 layer, was the
laminates dominant failure mode in early increment steps. This damage,
though possibly not affecting tensile strength much, was
Damage initiation and propagation was evaluated by compar- caused by transverse stress concentration and spread to other
ing screenshots of various failure modes in different increment layers with increasing loading. Fiber tensile failure occurred
456 J. Hu et al.

Fig. 10 Damage initiation and propagation of notched composite laminates under tensile loading.

Fig. 11 Damage initiation and propagation of notched composite laminates under compressive loading.
Individual plies of notched composite laminates 457

at 88.71 in the 0 layer and was accompanied by matrix com- as the FE model simulated in Fig. 11. Slight delamination is
pressive failure in later increment steps. As loading increased, also found in the experiment because excessive load destroyed
fiber damage propagated across the net tension plane causing the compatibility of deformation among layers. This is beyond
dramatic decrease in load capacity. Matrix-fiber shear failure the elastic range and not considered in the analytical solution.
was observed due to excessive transverse shrinkage in late To further support the progressive failure results from the
loading increments. The failure mode under compressive load- FE model, strain concentration and crack growth path on
ing where matrix compressive failure initiation was observed at the specimen surface were captured by DSMS to compare with
88.4 in the 90 layer and propagates in early increment steps is the damage of the 45 layer (surface layer) in the composites,
presented in Fig. 11. Fiber compressive failure was found and the results are listed in Figs. 13 and 14. For OHT testing,
along with matrix-fiber shear in the 0 layer before specimen the initial damage points induced by matrix tensile failure
buckling deformation. The initiation location of fiber compres- (Fig. 13(c)) matched very well with the zones of strain concen-
sive damage occurred where fiber direction was normal to the tration (Fig. 13(a)), and the damage propagation path follows
hole perimeter. Tensile failure rarely appeared in the process. the crack profile. Fiber tensile failure (fiber tensile rupture) is
Microscopy investigation of damaged specimen form tests also observed in both test results (Fig. 13(b)) and the FE
was conducted to back up the progressive failure results from results (Fig. 13(d)). As for OHC testing, the initial damage
FE model. In the microscopic morphology, matrix crack points of failure (Fig. 14(c)) were in accordance with the zones
(Fig. 12(a)) and fiber breakage (Fig. 12(b)), which corre- of strain concentration around the hole (Fig. 14(a)). The fiber-
sponded to matrix tensile failure and fiber tensile failure from matrix shear-out failure in the 45 layer (Fig. 14(d)) resulted in
FE results in Fig. 10, were observed as common micro-scale surface cracking of the specimen (Fig. 14(b)) because they pos-
failure modes in OHT specimens. Matrix crush (Fig. 12(c)), sessed the same growth path. The above analysis provided suf-
fiber buckling (Fig. 12(d)) and matrix-fiber shear (Fig. 12(e)) ficient credibility to the validation of the FE model for
were dominant failure modes in compressive specimens, just progressive failure analysis.

Fig. 12 Scanning electron microscope images of damage types in the vicinity of the notch.

Fig. 13 Comparison of failure results between experimental test and FE model under tensile loading.
458 J. Hu et al.

Fig. 14 Comparison of failure results between experimental test and FE model under compressive loading.

6.4. Failure strength captured by the numerical models, indicating that the compos-
ite can be regarded as linear-elastic material despite its brittle-
The load-displacement plots of experimental and FE simula- ness. The stiffness of the specimen (slope of the curve) from the
tion results for the notched composite laminates are shown test is smaller than that from FE results due to the defects of
in Figs. 15 and 16. The slope of the linearity in the experimen- the composite laminates and the degradation rules for the
tal tests (i.e. elastic specimen stiffness) is seen to be very well material properties. The OHC test shows slight nonlinear

Fig. 15 Load-displacement curves of notched composite laminates under tensile test.

Fig. 16 Load-displacement curves of notched composite laminates under compressive test.


Individual plies of notched composite laminates 459

behavior at the start of the curve because the specimen slightly damage occurs and builds until matrix damage causes
buckled under compressive loading. Since the OHC specimens ruptures. While fiber buckling occurs in most plies, fiber
were not ruptured, compressive failure is much more moderate breakage is found in the ply where the fiber orientation
than in the OHT test. The nonlinear behavior observed near is parallel with loading direction.
the peak load is associated with severe damage in the speci- (4) Load-displacement plots exhibit a long linear state, indi-
mens leading to fracture. cating that the composites can be regarded as linear-
In the tensile test, there is a protuberant point at a load of elastic material despite of their brittleness. Matrix dam-
247 MPa, which is mainly caused by the initiation of fiber ten- age evolution does not affect tensile strength much while
sile damage around the notch in the 0 layer along with serious fiber damage initiation may cause sudden shocks in
matrix tensile failure in all plies. With each load increment, the plots. CFRP laminates possess better tensile load capac-
fiber tensile damage spreads across the tension plane at 0 ity than compressive ones.
through thickness into 90, 60,+45 layers in the vicinity
of the hole, resulting in drastic fluctuations in the curve. The Acknowledgements
experimental ultimate strength was 484.6 MPa with an error
of 7.7% compared with the FE result of 524.9 MPa. In the This work was sponsored by the National Natural Science
compressive test, the main damage mode in an earlier state Foundation of China, with three different programs (No.
was matrix compressive failure. Fiber buckling failure occurs, 51275410, No. 51305349 and No. 51305352) that supports
especially in the 0 layer in the net tension plan, causing a the present work financially. The authors would like to
quick shock in the load capacity curve. As the load increased, acknowledge the editors and the anonymous reviewers for
buckling failure extended beyond the surroundings of the their insightful comments.
notch and caused bending of the specimen. Since the specimens
were not ruptured, the compressive failure is much more mod- References
erate than in the tensile test. There was a difference of 5.3%
between the test strength of 206.1 MPa and FE result of 1. Ostergaard MG, Ibbotson AR, Roux OL, Prior AM. Virtual
217.5 MPa. testing of aircraft structures. CEAS Aeronaut J 2011;1(1
4):83103.
7. Conclusions 2. Singh I, Bhatnagar N. Drilling-induced damage in uni-directional
glass fiber reinforced plastic (UD-GFRP) composite laminate. Int
J Adv Manuf Tech 2006;27(910):87782.
This research investigates the layer-by layer stress components, 3. Senthil K, Arockiarajan A, Palaninathan R, Santhosh B, Usha
and the initiation and propagation of damage and failure in KM. Defects in composite structures: Its effects and prediction
notched composite laminates by analytic method and FE methods a comprehensive review. Compos Struct
model, both of which are validated by experimental tests. 2013;106:13949.
There is good agreement between the analytic method and 4. Muskhelishvili NI. Some basic problems of the mathematical theory
the FE model in terms of the stress (strain) distributions, dam- of elasticity. Noordhoff P. Ltd., 1953, 123129.
age initiation and location of failure modes, and experiments 5. Lekhnitskii SG. Anisotropic plates. New York: Gordon and
Breach; 1968. p. 415.
on strain by DSMS and microscopy study by SEM provide
6. Savin GN. Stress concentration around holes. New York: Pergamon
them with strong suitability and validation. The proposed ana-
Press; 1961. p. 16974.
lytic fundamentals present a dependable and manageable alter- 7. Gao XL. A general solution of an infinite elastic plate with an
native to evaluate stress level and damage initiation quickly, elliptic hole under biaxial loading. Int J Pres Ves Pip 1996;67
while the FE model simulates the damage evolution process (1):95104.
thoroughly. Several conclusions can be drawn from the 8. Ukadgaonker VG, Awasare PJ, Gade SV. Stress analysis of door
research: and window of a passenger aircraft. Aeronaut J 2000;104
(1039):40914.
(1) The stress components are all non-uniform and periodic 9. Ukadgaonker VG, Rao DKN. Stress distribution around trian-
in different plies and caused by the anisotropy of the gular holes in anisotropic plates. Compos Struct 1999;45
(3):17183.
laminates. The magnitude of longitudinal stress is much
10. Ukadgaonker VG, Rao DKN. A general solution for stresses
larger than that of transverse stress and tangential stress
around holes in symmetric laminates under inplane loading.
due to the high elastic modulus of the fiber contrasting Compos Struct 2000;49(3):33954.
with the matrix. 11. Ukadgaonker VG, Kakhandki V. Stress analysis for an ortho-
(2) The layer-by-layer stress distributions around cutouts tropic plate with an irregular shaped hole for different in-plane
are affected by fiber orientation and loading direction. loading conditions Part 1. Compos Struct 2005;70(3):22574.
The high stress level of rL mainly occurs where the fiber 12. Pan ZX, Cheng YS, Liu J. Stress analysis of a finite plate with a
direction is tangential to the hole perimeter. The concen- rectangular hole subjected to uniaxial tension using modified stress
tration of rT is observed where the fiber direction is nor- functions. Int J Mech Sci 2013;75:26577.
mal to the notch boundary. The sLT is more influenced 13. Rao DKN, Babu MR, Reddy KRN, Sunil D. Stress around
square and rectangular cutouts in symmetric laminates. Compos
by shear effect and possesses more extreme points than
Struct 2010;92(12):284559.
rL and rT .
14. Sharma DS. Stress distribution around polygonal holes. Int J
(3) Matrix dominant failure is common in early states of in- Mech Sci 2012;65(1):11524.
plane loading. It initiates at the hole edge in the ply 15. Batista M. On the stress concentration around a hole in an infinite
whose fiber orientation is perpendicular to the load plate subject to a uniform load at infinity. Int J Mech Sci 2011;53
direction and then propagates to other plies. Fiber (4):25461.
460 J. Hu et al.

16. Remeepazhand J, Jafari M. Stress analysis of perforated compos- 30. Muc A, Rys J, Latas W. Limit load-carrying capacity for spherical
ite plates. Compos Struct 2005;71(34):4638. laminated shells under external-pressure. Compos Struct 1993;25
17. Toubal L, Karama M, Lorrain B. Stress concentration in a (14):295303.
circular hole in composite plate. Compos Struct 2005;68(1):316. 31. Zhang JY, Liu FR, Zhao LB, Chen YL, Fei BJ. A progressive
18. Liu PF, Zheng JY. Recent developments on damage modeling and damage analysis based characteristic length method for multi-bolt
finite element analysis for composite laminates: a review. Mater composite joints. Compos Struct 2014;108:91523.
Des 2010;31(8):382534. 32. Pietropaoli E. Progressive failure analysis of composite structures
19. Lapczyk I, Hurtado JA. Progressive damage modeling in fiber- using a constitutive material model (USERMAT) developed and
reinforced materials. Compos Part A Appl Sci 2007;38 implemented in ANSYS. Appl Compos Mater 2012;19(3
(11):233341. 4):65768.
20. Zahari R, EI-Zafrany A. Progressive failure analysis of composite 33. Ribeiro ML, Vandepitte D, Tita V. Damage model and progres-
laminated stiffened plates using the finite strip method. Compos sive failure analyses for filament wound composite laminates. Appl
Struct 2009;87(1):6370. Compos Mater 2013;20(5):97592.
21. Rakesh PK, Singh I, Kumar D. Failure prediction in glass fiber 34. Chang FK, Lessard LB. Damage tolerance of laminated compos-
reinforced plastics laminates with drilled hole under uni-axial ites containing an open hole and subjected to compressive
loading. Mater Des 2010;31(6):30027. loadings. 1. Analysis. J Compos Mater 1991;25(1):243.
22. Ondurucu A, Esendemir U, Tunay RF. Progressive failure 35. Camanho PP, Matthews FL. A progressive damage model for
analysis of glass-epoxy laminated composite pinned-joints. Mater mechanically fastened joints in composite laminates. J Compos
Des 2012;36:61725. Mater 1999;33(24):224880.
23. Kim SY, He B, Shim CS, Kim D. An experimental and numerical 36. ASTM International. D5766/D5766M-02a Standard test method
study on the interference-fit pin installation process for cross-ply for open hole tensile strength of polymermatrix composite lami-
glass fiber reinforced plastics (GFRP). Compos Part B-Eng nates. 2007.
2013;54:15362. 37. ASTM International. D6484/D6484M-04 Standard test method for
24. Satapathy MR, Vinayak BG, Jayaprakash K, Naik NK. Fatigue open-hole compressive strength of polymermatrix composite lam-
behavior of laminated composites with a circular hole under in- inates. 2004.
plane multiaxial loading. Mater Des 2013;51:34756. 38. Wu HC, Mu B. On stress concentrations for isotropic/orthotropic
25. Martins LAL, Bastian FL, Netto TA. Reviewing some design plates and cylinders with a circular hole. Compos Part B-Eng
issues for filament wound composite tubes. Mater Des 2003;34(2):12734.
2014;55:2429. 39. Hibbit, Karlssonand Sorensen Inc. ABAQUS/Standard version
26. Lee CS, Kim JH, Kim SK, Ryu DM, Lee JM. Initial and 6.11, users manual. 2011.
progressive failure analyses for composite laminates using Puck 40. Huang ZM. Inelastic and failure analysis of laminate structures by
failure criterion and damage-coupled finite element method. ABAQUS incorporated with a general constitutive relationship. J
Compos Struct 2015;121:40619. Reinf Plast Comp 2007;26(11):113581.
27. Aljibori HSS, Chong WP, Mahlia TMI, Chong WT, Edi P, 41. Ekh J, Schon J, Melin LG. Secondary bending in multi fastener,
Alqrimli H, et al. Load-displacement behavior of glass fiber/epoxy composite-to-aluminium single shear lap joints. Compos Part B-
composite plates with circular cut-outs subjected to compressive Eng 2005;36(3):195208.
load. Mater Des 2010;31(1):46674. 42. Vieille B, Aucher J, Taleb L. Overstress accommodation in
28. Talib AR Abu, Ramadhan AA, Rafie ASM, Zahari R. Influence notched woven-ply thermoplastic laminates at high-temperature:
of cut-out hole on multi-layer Kevlar-29/epoxy composite lami- numerical modeling and validation by digital image correlation.
nated plates. Mater Des 2013;43:8998. Compos Part B-Eng 2013;45(1):290302.
29. Jones RM. Mechanics of composite materials. 2nd ed. Philadel-
phia: Taylor & Francis; 1999. p. 10921.