XML Template (2013) [16.5.

2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

JOURNAL OF
COMPOSITE
Original Article M AT E R I A L S
Journal of Composite Materials
0(0) 1–17
! The Author(s) 2013
Analysis of classical and first-order shear Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
deformable cracked orthotropic plates DOI: 10.1177/0021998313487756
jcm.sagepub.com

András Szekrényes

Abstract
The Kirchhoff and Mindlin plate theories are applied in this study to calculate the stresses and the energy release rates in
delaminated orthotropic composite plates. A novel double-plate system is developed with the imposition of the kine-
matic continuity constraints in the interface plane. The governing equations of the system were derived in both cases. As
a demonstrative example a simply-supported plate subjected to a point force was analyzed using Lévy plate formulation
and the problem was solved by a state-space model. The distribution of the stress resultants and the interlaminar
stresses in the uncracked part were also determined. Moreover, the distributions of the mode-II and mode-III energy
release rates along the crack front were calculated by the J-integral. The 3D finite element model of the plate was
created providing reference data for the analytical model. The results show that the displacement and stress fields
obtained from the Kirchhoff and Mindlin theories are quite similar, but in the case of the energy release rates, transverse
shear effect is necessary to consider to obtain reasonably good agreement between the analytical and numerical results.

Keywords
Laminated plate, delamination, J-integral, Mixed-mode II/III fracture, Classical laminated plate theory, first-order
plate theory

Introduction possible acceptable tests under mode-III,16–18 mixed-

Linear plate theories play important role in the design mode II/III22 and mixed-mode I/III30 equally involve
of composite,1,2 steel3 and sandwich structures4 in the plate-type specimens with rectangular delamination
engineering life. The interlaminar fracture is one of the surface.
primary failure modes in composite and sandwich In general, an analytical solution can be easily devel-
structures during transverse and sudden impact load- oped for beam-type specimens; in contrast, for plate-
ing, as well. The fracture behavior of laminated mater- type samples the analytical solutions – mainly because
ials is characterized by the energy release rate (ERR).5 of the complex boundary conditions – are not available.
The critical value of this global parameter (critical Therefore, without any exception, the data reduction is
energy release rate – CERR) is the limit value to pro- always made by the finite element method (FEM)
portionate the structures against delamination failure. incorporating the virtual-crack closure technique
Within the scope of linear elastic fracture mechanics,5 (VCCT)19 and cohesive zone model (CZM)30 applica-
the basic failure modes include mode-I (opening mode), tions. For delaminated plates, Davidson et al.35 applied
mode-II (in-plane shear) and mode-III (anti-plane shell elements with interface constraints to calculate the
shear), respectively. The fracture resistance under ERRs, Sankar and Sonik36 performed similar
these modes is determined through standard or non-
standard tests. The significant part of the literature Department of Applied Mechanics, Budapest University of Technology
deals with mode-I,6,7 mode-II8–10 and mixed-mode and Economics, Hungary
I/II10–12 cases. However, in the last decades, more and
more emphasis was put on the investigation of mode- Corresponding author:
András Szekrényes, Department of Applied Mechanics, Budapest
III.13–20 fracture, later even the combined mode University of Technology and Economics, Budapest, Mu00 egyetem rkp 5,
II/III,21–29 I/III30,31 and I/II/III32–34 conditions were Building MM, H-1111, Hungary.
taken into account. The latest developments for Email: szeki@mm.bme.hu

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

2 Journal of Composite Materials 0(0)

computations. The crack tip force method (CTFM)37 is where  takes ‘‘top’’ and the upper sign for the top
a similar solution to the VCCT, utilizing the crack tip plate element in Figure 1(a) and  takes ‘‘bot’’ and
forces to calculate and separate the ERRs. However, its the lower sign for the bottom plate element of the dela-
result does not differ from that of a VCCT analysis. minated portion in Figure 1(b). Moreover, u0 and v0 are
The main problem of the FE models is that a 3D the constant parts of the displacements, x and y are
model is necessary to construct and the VCCT is avail- the rotations of the cross section about the y and x
able only in few FE packages (e.g. in ABAQUS and axes, respectively. In this paper, only mixed-mode
NASTRAN, but not available in ANSYS and II/III problems are considered, i.e. the crack opening
COSMOS/M) as a built-in command. In accordance is zero.
with this overview, we can also conclude that there Therefore, the transverse deflection, w ¼ w(x,y) is the
are no analytical solutions available for delaminated same for both the top and bottom plates. Applying the
composite plate specimens as possible data reduction basic equations of linear elasticity49 for an ortho-
schemes. tropic plate we obtain the strain and stress fields,
The aim of this paper is to present the application of then by integrating the stresses over the thickness, the
classical laminated plate theory (Kirchhoff plate theory stress resultants in terms of strains and shear strains
or CLPT)2,38–40 and first-order shear deformable plate become:2
theory (Mindlin plate theory or FSDT)2–4 to calculate
the stresses and ERRs in delaminated plates with sym-    ( )
metrical lay-up and straight delamination front under
fNg ½A ½0 f"ð0Þ g
¼ ,
mixed-mode II/III condition. First, the displacement fMg ½0 ½D f"ð1Þ g
field is formulated by imposing the interface con- Qx ¼ kA55 xz , Qy ¼ kA44 yz , ð2Þ
straints. Second, the basic equations of linear elasticity
are applied to derive the strain and stress fields in elastic
orthotropic composite plates. The present formulation where {N}T ¼ {Nx, Ny, Nxy} is the vector of in-plane
does not incorporate arbitrarily defined parameters normal and shear forces, {M}T ¼ {Mx, My, Mxy} is
(e.g. shear compliances,41–46) and it is shown that the the vector of bending and twisting moments, Qx, Qy
developed model is physically consistent with the equa- are the shear forces and k ¼ 5/6 is the shear correction
tions of linear elasticity. As an example, a simply-sup- factor.2,3 Moreover, {"(0)} and {"(1)} are the vectors of
ported plate subjected to a point force is analyzed constant and linear strains, finally  xz and  yz are the
applying the state-space model.2 The distribution of shear strains. The stress resultants can be calculated
the interlaminar stresses is calculated, moreover, the from the integration of the stresses in the through-
J-integral47,48 is utilized to determine the distribution thickness direction:2
of the mode-II and mode-III ERRs along the crack
front. An FE model is also created and the numerical    
Zt=2 Zt=2
results are compared to those obtained by analysis. The N 1
¼  dz, Q ¼ z dz, ð3Þ
relatively good agreement obtained shows the useful- M z
ness of plate theories in delamination analysis. t=2 t=2

FSDT- Mindlin plate theory, general where  and  take the symbols x and y. The stiffness
parameters are defined as:1,2
formulation
We assume that a delaminated plate with straight crack
Nl Zkþ1
z
front lies in the x-y plane and the crack front is parallel X ðkÞ
Aij , Dij ¼ Cij ð1, z2 Þdz, ð4Þ
to the y axis. The plate theory presented in this section k¼1 z
is utilized to capture the displacement and stress fields k

in the delaminated portion of an elastic laminated

orthotropic plate with symmetric lay-up presented in where the reference plane is the local midplane of the
Figure 1. The thickness of the top and bottom plates top and bottom plates, respectively (refer to Figure 1).
is t. The assumed displacement field based on FSDT for The vectors of strains are written as:
elastic plates can be written as:2
8 9 8 ð0Þ 9 8 ð1Þ 9
> " > > > " > > " > ( ) ( ð0Þ )
< x = < x > = < x >
> = xz xz
u ðx, y, zÞ ¼ u0 ðx, yÞ þ x ðx, yÞ  z, ð0Þ ð1Þ
ð1Þ "y ¼ "y þ z "y , ¼ , ð5Þ
v ðx, y, zÞ ¼ v0 ðx, yÞ þ y ðx, yÞ  z, >
: >
; >
> >
> >
> >
> yz ð0Þ
yz
xy : ð0Þ ; : ð1Þ ;
xy xy

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

Szekrényes 3

where
9 8 8 9
>
> >
> @u0>
> @x > >
8 ð0Þ 9 >
> >
>8 ð1Þ 9 > > >
>
> " > >
> >
>> " >@x >
> @x >
> ( ð0Þ ) ( )
>
< x >
= > >
= >
<< x >
= > < @ >
=
ð0Þ ð1Þ @v0 y xz x þ @w
@x
"y ¼ , "y ¼ , ¼ ð6Þ
>
> > > > > >@x > @x > ð0Þ
yz y þ @w
: ð0Þ > ; > >
>
>
>
>
>
: ð1Þ
>
; >
>
>
>
>
> @y
xy >
>
> @u0 @v0 >
>
>
xy >
>
> @x @y > >
>
: þ ; : þ ;
@y @x @y @x
Formulating the total potential energy for the elastic
plate subjected to transverse load and deriving the @ 2 y @ 2 y @ 2 x
D22 þ D66 2 þ ðD12 þ D66 Þ
Euler-Lagrange equations, we obtain the equilibrium @y 2 @x @x@y
equations in the following forms:2,3  
@w
 kA44 y þ ¼ 0, ð12Þ
@y
@Nx @Nxy @Nxy @Ny
þ ¼ 0, þ ¼ 0, ð7Þ    
@x @y @x @y @x @2 w @y @2 w
kA55 þ þ kA44 þ ¼ q ð13Þ
@Mx @Mxy @Mxy @My @x @x2 @y @y2
þ  Qx ¼ 0, þ  Qy ¼ 0,
@x @y @x @y
@Qx @Qy where Aij and Dij are stiffness parameters and q ¼ q(x,y)
þ þq¼0 ð8Þ is the external load of the plate.
@x @y

Taking the stress resultants back into equation (7)

and (8), we obtain:
FSDT with interface constraint
The FSDT utilized in this section is based on an
@ 2 u0 @ 2 u0 @ 2 v0 assumed displacement field including interface con-
A11 2 þ A66 2 þ ðA12 þ A66 Þ ¼ 0, ð9Þ straints to formulate the model of the uncracked
@x @y @x@y
region of a delaminated orthotropic composite plate.
@2 v0 @ 2 v0 @ 2 u0 The mathematical form of in-plane displacement com-
A22 þ A 66 þ ðA12 þ A66 Þ ¼ 0, ð10Þ ponents are the same as those given by equations (1);
@y2 @x2 @x@y
however, we have to ensure the displacement continuity
@2 x @ 2 x @ 2 y between the top and bottom plate elements shown by
D11 þ D66 þ ðD 12 þ D66 Þ Figure 1. The interface constraint equations of the
@x2 @y2 @x@y
  uncracked portion are
@w
 kA55 x þ ¼ 0, ð11Þ  
@x utop z¼ t ¼ ubot jz¼2t , vtop z¼ t ¼ vbot jz¼2t ð14Þ
2 2

(a) (b)
Delaminated portion Z, z Undelaminated portion Z, z Delaminated portion

Reference plane
vtop (x,y,z) utop (x,y,z) top plate
t
y x
Interface

Y X
ubot (x,y,z) Delamination
Delamination vbot (x,y,z)
t
y x
Reference plane
bottom plate

Figure 1. Deformations of the top and bottom plate elements.

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

4 Journal of Composite Materials 0(0)

These conditions make it possible to express u0 and uncracked portion are the local midplanes (see
v0 in terms of the rotation parameters: Figure 1), and consequently the stiffness parameters
t t in equation (4) for the delaminated and uncracked
u0 ¼  x , v 0 ¼  y ð15Þ plate portions are the same. The different deformations
2 2
of the two portions are included through the interface
Taking these back into equation (1), we obtain the constraint conditions in equation (14).
displacement field satisfying the interface constraint
conditions:
CLPT – Kirchhoff plate theory

t t

utop ðx, y, zÞ ¼ x þ z , vtop ðx, y, zÞ ¼ y þ z ð16Þ The Kirchhoff plate theory (CLPT) can be considered
2 2 as a special case of FSDT, when the rotations become:2
@w @w
For the bottom plate, similar expressions can be x ¼  , y ¼  ð24Þ
derived. Due to the symmetric lay-up with respect to @x @y
the x-y plane, we analyze only the top plate in the i.e., in accordance with the Kirchhoff hypothesis, the
sequel. Based on equation (6), the strains and shear normal of the cross section is parallel to the derivative
strains become: of the deflection.
8 9
>
> @x >
>
8 ð1Þ 9 >
> >
> 8 ð0Þ 9 8 ð1Þ 9 8 9
> " > >
> @x >
> > "x > > "x > ( ð0Þ ) > @w >
< x >
> = >
< >
= >
< >
= t<> >
= < x @x >
>  þ =
@y xz
"ð1Þ
y ¼ , "ð0Þ
y ¼ "ð1Þ
y , ¼ ð17Þ
>
> > > > @x > > 2> > ð0Þ
yz > @w >
: ð1Þ >; > >
>
>
>
> : ð0Þ >
> ; : ð1Þ >
> ; >
: y þ >
;
xy >
>
> @x @y >
>
>
xy xy @y
: þ ;
@y @x
In this case, the system of equilibrium equations con-
sists of three equations only:
Delaminated portion
  For the delaminated portion, the constant strains
@Mx @Mxy t @Nx @Nxy are the same as those in equation (6), the linear
þ þ þ  Qx ¼ 0, ð18Þ
@x @y 2 @x @y strains are
  8 ð1Þ 9
@Mxy @My t @Nxy @Ny
þ þ þ  Qy ¼ 0, ð19Þ < "x =  T
@x @y 2 @x @y @2 w @2 w @2 w
"ð1Þ
y ¼  2  2 2 ð25Þ
: ; @x @y @x@y
 xyð1Þ
@Qx @Qy
þ þq¼0 ð20Þ
@x @y
The shear strains  xz and  yz are apparently zero.
Calculating the stress resultants (equation (2)) in The resulting equilibrium equations are given by equa-
terms of the displacement parameters and taking them tion (7); moreover, the three equations in equation (8)
back into the equilibrium equations (equations (18)– reduce to the one below:
(20)), the following system of equations is obtained:
@ 2 Mx @2 Mxy @2 My
2 2 2 þ2 þ ¼ q ð26Þ
@ x @ x @ y @w @x 2 @x@y @y2
a^ 1 þ a^ 2 2 þ a^ 3 x þ a^ 4 þ a^ 5 ¼ 0, ð21Þ
@x2 @y @x@y @x
Taking the displacement parameters back into equa-
@ 2 x @ 2 y @2 y @w tions (6) and (26) we obtain:
b^1 þ b^2 2 þ b^3 2 þ b^4 y þ b^5 ¼ 0, ð22Þ
@x@y @x @y @y
@4 w @4 w @4 w
@x @y @2 w @2 w D11 þ 2ðD12 þ D66 Þ þ D 22 ¼q ð27Þ
c^1 þ c^2 þ c^3 2 þ c^4 2 þ c^5 q ¼ 0 ð23Þ @x4 @x2 @y2 @y4
@x @y @x @y
where the coefficients depend on the stiffness param- Along with equations (9) and (10) (these still hold in
eters and the thickness of the plate, and are collected the case of CLPT) in this case, we have a three-para-
in Appendix 1. It is important to highlight that the meter displacement field, while the system consists of
reference planes of the top and bottom plates of the three PDEs.

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

Szekrényes 5

Undelaminated portion Example – simply-supported plate, Lévy

The substitution of equation (24) into equation (16)
plate formulation
yields the following displacement field for the undela- In this section, we apply the state-space model2 to solve
minated portion: the system of equations for a delaminated plate sub-
jected to a point force, shown in Figure 2. The govern-
@w t
@w t

ing PDE system is different for the delaminated and

utop ðx, y, zÞ ¼  þ z , vtop ðx, y, zÞ ¼  þz
@x 2 @y 2 uncracked parts; therefore, the state-space models are
ð28Þ developed separately. In accordance with Lévy plate
formulation, the displacement components and the
In this case, the strain field can be obtained by equa- external load for simply-supported (Mindlin and
tion (25) and the second of (17), and consequently the Kirchhoff) plates1–3 can be written as:
displacement field involves a single parameter, the
deflection of the plate, w(x,y). The equilibrium equa- 8 9 8 9
>
> u0 ðx, yÞ >
> >
> U0n ðxÞ sin y >
>
tions by equations (18)–(20) reduce to a single equation: >
> > > >
< v0 ðx, yÞ >
= X1 >< V0n ðxÞ cos y >
= X1
  x ðx, yÞ ¼ Xn ðxÞ sin y , q ¼ Qn sin y
@ 2 Mx @2 Mxy @2 My t @2 Nx @2 Nxy @2 Ny >
>
>  ðx, yÞ
>
>
> n¼1
>
>
> Y ðxÞ cos y
>
>
> n¼1
þ2 þ þ þ2 þ >
: y >
; >
: n >
;
@x2 @x@y @y2 2 @x2 @x@y @y2 wðx, yÞ Wn ðxÞ sin y
¼ q ð29Þ ð31Þ

Finally, by calculating the stress resultants from where  ¼ n/b. Apparently, in the case of Kirchhoff
equations (2) and (3) we obtain the governing PDE of plate theory x and y are not applicable.
the deflection of undelaminated portion:
  FSDT- Mindlin plate theory
A11 t2 @4 w ðA12 þ A66 Þt2
ðD11 þ Þ þ 2 D12 þ D66 þ Delaminated portion. Taking the solutions above back
4 @x4 4
4
 2
 4 into the system of governing equations of the delami-
@ w A22 t @ w
þ D22 þ ¼q ð30Þ nated portion (equations (9)–(13)), we can derive the
@x2 @y2 4 @y4
state-space model, in a general form it becomes:

As it can be seen from equation (30), the transform- ðd Þ

Z0 ¼ Tðd Þ Zðd Þ þ F ðd Þ ð32Þ
ation of the local bending stiffness is involved with t/2
distance from the local reference plane.1 In the next where the superscript ‘‘(d)’’ refers to the delaminated
section, we solve a simply-supported plate subjected plate portion and Z is the state vector:
to a point force by the state-space model.2 The displace-
 T
ment and stress fields are calculated and the J-integral Zðd Þ ¼ U0n U00n V0n V00n Xn X0n Yn Y0n Wn W0n
is utilized to calculate the ERR distributions along the
ð33Þ
delamination front.

Simply-supported
Z 2
Simply-supported
1
1q
Simply-supported
Q0
1a Delamination

2t
2d0 c
xQ
yQ
a Y
X
b
t
Simply-supported
Simply-supported

Figure 2. Simply-supported plate subjected to point force.

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

6 Journal of Composite Materials 0(0)

moreover, T is the system matrix, F contains the The expanded state-space model becomes:
inhomogeneity of the governing PDE system. In an
expanded form equation (32) becomes:

2 3 2 32 3 2 3
U00n 0 1 0 0 0 0 0 0 0 0 U0n 0
6 00 7 6 76 7 6 7
6 U0n 7 6 f1 0 0 f2 0 0 0 0 0 0 76 U00n 7 6 0 7
6 7 6 76 7 6 7
6 V0 7 6 0 0 0 1 0 07 6 7 6 7
6 0n 7 6 0 0 0 0 76 V0n 7 6 0 7
6 00 7 6 76 0 7 6 7
6 V0n 7 6 0 g1 g2 0 0 07 6 7 6 7
6 7 6 0 0 0 0 76 V0n 7 6 0 7
6 0 7 6 76 7 6 7
6 Xn 7 6 0 0 0 0 0 1 0 0 0 0 76 Xn 7 6 0 7
6 7 6 76 7þ6 7
6 X00 7 ¼ 6 0 0 0 0 h h3 7 6 0 7 6 7 ð34Þ
6 n7 6 1 0 0 h2 0 76 Xn 7 6 0 7
6 0 7 6 76 7 6 7
6 Y 7 60 0 0 0 0 07 6 7 6 7
6 n7 6 0 0 1 0 76 Yn 7 6 0 7
6 00 7 6 76 0 7 6 7
6 Yn 7 6 0 0 0 0 0 j1 j2 0 j3 0 76 Yn 7 6 0 7
6 7 6 76 7 6 7
6 W0 7 6 0 0 0 0 0 17 6 7 6 7
4 n5 4 0 0 0 0 5 4 Wn 5 4 0 5
W00n 0 0 0 0 0 k1 k2 0 k3 0 W0n k4 Qn

where the coefficients in the 10  10 system matrix T are

collected in Appendix 2. The general solution of equa- 2 3 2 32 3 2 3
X0n 0 1 0 0 0 0 Xn 0
tion (32) is2
6 X00 7 6 d^ 0 0 d^ 0 d^ 76 X0 7 6 0 7
2 3 6 n7 6 1 2 3 76 n 7 6 7
Zx 6 7 6 76 7 6 7
T ðd Þ
x ðd Þ
6 Y0 7 6 0 0 0 1 0 0 76 Yn 7 6 0 7
Zðd Þ ðxÞ ¼ e 4Kðd Þ þ e T
F ðd Þ ð Þd 5 6 n7 6 76 7 6 7
6 00 7 ¼ 6 76 7þ6 7
6 Yn 7 6 0 e^1 e^2 0 e^3 0 76 Y0n 7 6 0 7
x0 6 7 6 76 7 6 7
6 0 7 6 76 7 6 7
¼ G ðxÞK ðd Þ ðd Þ
þ H ðxÞðd Þ
ð35Þ 4 Wn 5 4 0 0 0 0 0 1 54 Wn 5 4 0 5
W00n 0 f^1 f^2 0 f^3 0 W0n f^4 Qn
where K is the vector of constants, H is the vector of
ð38Þ
particular solutions.
which involves a 6  6 system matrix with coefficients
Uncracked portion. It has been shown that because of defined in Appendix 1. Considering the fact that there is
the kinematic constraints, two of the five displacement no external load at the uncracked plate portion the
parameters can be eliminated. Utilizing equations solution of equation (38) is
(21)–(23), the state-space model can be derived as:
Tðud Þ x
Zðud Þ ðxÞ ¼ e Kðud Þ ¼ Gðud Þ ðxÞKðud Þ ð39Þ
0 ðud Þ ðud Þ ðud Þ ðud Þ
Z ¼T Z þF ð36Þ
Kirchhoff plate theory
where the superscript ‘‘(ud)’’ refers to the undelami-
nated region. The state vector in this case is Delaminated portion. In this case, the system matrix is an
8  8 matrix. Without going into details, based on
 T
Zðud Þ ¼ Xn X0n Yn Y0n Wn W0n ð37Þ equations (9), (10) and (27), the state-space model
takes the form of:
2 3
2 3 2 32 3 0
U00n 0 1 0 0 0 0 0 0 U0n
6 7
6 U00 7 6 f1 0 0 f2 0 0 0 07 6 0 7 6 0 7
6 0n 7 6 76 U0n 7 6 7
6 0 7 6 76 7 6 7
6 V0n 7 6 0 0 0 1 0 0 0 0 76 V0n 7 6 0 7
6 7 6 76 7 6 7
6 V00 7 6 0 g1 g2 0 0 0 0 07 6 0 7 6 0 7
6 0n 7 6 76 V0n 7 6 7
6 0 7¼6 76 7þ6 7 ð40Þ
6 Wn 7 6 0 0 0 0 0 1 0 0 76 Wn 7 6 0 7
6 7 6 76 7 6 7
6 W00 7 6 0 0 0 0 0 0 1 07 6 0 7 6 0 7
6 n 7 6 76 Wn 7 6 7
6 000 7 6 76 7 6 7
4 Wn 5 4 0 0 0 0 0 0 0 1 54 W00n 5 6 0 7
4 Qn 5
WIV
n
0 0 0 0 4 dc13 0 22 cc41 0 W000
n c1
where the coefficients are defined in Appendix 2.

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

Szekrényes 7

Undelaminated portion. From equations (30) and (31), we then by highlighting the differences between the two
obtain the following state-space model: theories we continue with FSDT.
2 3
2 0 3 0 1 0 0 2 3 2 3
Wn Wn 0 Kirchhoff plate theory. For the present problem, 28 con-
6 7
6 W00n 7 6 0 0 1 0 76 0 7 6 0 7 ditions need to be formulated. The B.C.s for simply-
6 000 7 ¼ 6 0 0 0 17 6 Wn00 7 þ 6 7
4 Wn 5 6 ^ 74 Wn 5 4 0 5 supported edges are (6 conditions)
4 b3 a^4 5 1
WIV
n 4 0 22 0 W000
n a^1 Qn
a^ 1 a^1 ð1aÞ
Wð1aÞ ð1aÞ ð1aÞ
n ðaÞ ¼ 0, V0n ðaÞ ¼ 0, nxn ðaÞ ¼ 0, mxn ðaÞ ¼ 0
ð41Þ t ð2Þ
where the coefficients are defined in Appendix 1. Wð2Þ ð2Þ
n ðcÞ ¼ 0, mxn ðcÞ þ nxn ðcÞ ¼ 0 ð44Þ
2
The continuity conditions between regions ‘‘1’’ and
Boundary and continuity conditions
‘‘2’’ are (6 conditions)
The elements of the state vectors in equations (36) and
(41) can be referred to as: t 0ð2Þ t
Uð1Þ ð1Þ
0n ð0Þ ¼  Wn ð0Þ, V0n ð0Þ ¼  Wn ð0Þ
ð2Þ
X ðd Þ ðd Þ X ðud Þ ðud Þ 2 2
Zðd Þ
i ¼ Gij Kj þ Hðd Þ ðud Þ
j , Zi ¼ Gij Kj þ Hðud Þ
Wð1Þ ð2Þ 0ð1Þ 0ð2Þ
j n ð0Þ ¼ Wn ð0Þ, Wn ð0Þ ¼ Wn ð0Þ,
j j
t ð2Þ
ð42Þ mð1Þ ð2Þ
xn ð0Þ ¼ mxn ð0Þ þ nxn ð0Þ
2
In accordance with Figure 2, we have four different t
plate portions. The point force causes singularity in the qxn ð0Þ  mxyn ð0Þ ¼ mxn ð0Þ þ n0ð2Þ
ð1Þ ð1Þ 0ð2Þ
ð0Þ
2 xn
PDEs, therefore a plate portion loaded by a constant t ð2Þ

 2 mð2Þ
xyn ð0Þ þ nxyn ð0Þ ð45Þ
line force was applied, the length d0 was a very small 2
value compared to the plate dimensions. In this case,
Qn ¼ 2q0/b  sin(y0) [2]. Thus, the four parts are The continuity and boundary conditions for the Mx
denoted by ‘‘1a’’, ‘‘1q’’, ‘‘1’’ for the delaminated por- bending moment and Vx ¼ Qx þ qMxy/qy (effective or
tion and ‘‘2’’ for the undelaminated region. Kirchhoff shear force3), respectively, must consider
Consequently, the state-space models for the delami- the fact that the in-plane normal and shear forces pro-
nated portion are utilized for the ‘‘1a’’, ‘‘1q’’ and ‘‘1’’ duce concentrated moments about the global X and Y
portions, while the model for the undelaminated por- axes of the uncracked plate portion. This effect – which
tion was used to capture region ‘‘2’’. The boundary is included also in equations (18) and (19) – is demon-
conditions (B.C.s) are formulated through the displace- strated in Figure 3 and was considered in the continuity
ment parameters and the stress resultants. The latter conditions. Thus, the continuity conditions between
ones can be expressed in the following forms: regions ‘‘1q’’ and ‘‘1’’ are (8 conditions)
8 9 8 9
>
> Nx >> >
> nxn >
> 8 9 8 9 Uð1qÞ ð1Þ ð1qÞ ð1Þ
>
> > > > 0n ðx01 Þ ¼ U0n ðx01 Þ, V0n ð0Þ ¼ V0n ð0Þ,
< Ny > = X 1 >< nyn >= < Nxy = X 1 < nxyn =
Mx ¼ mxn sin y, Mxy ¼ mxyn cos y Wð1qÞ ð1Þ
n ðx01 Þ ¼ Wn ðx01 Þ,
>
> >
> >
n¼1 >
>
> : ; n¼1 : ;
>
> M y >
> >
> m yn >
> Q y qyn Wn0ð1qÞ ðx01 Þ ¼ W0ð1Þ ð1qÞ ð1Þ
n ðx01 Þ, nxn ðx01 Þ ¼ nxn ðx01 Þ,
: ; : ;
Qx qxn
nð1qÞ ð1Þ
xyn ðx01 Þ ¼ nxyn ðx01 Þ,
ð43Þ
mð1qÞ ð1Þ ð1qÞ ð1qÞ
xn ðx01 Þ ¼ mxn ðx01 Þ, qxn ðx01 Þ  mxyn ðx01 Þ
i.e. nxn is the function coefficient in the Fourier series of
Nx, etc. To define the B.C.s we start with CLPT, and ¼ qð1Þ ð1Þ
xn ðx01 Þ  mxyn ðx01 Þ, ð46Þ

(a) Z, z (b) Z, z
Nx Nx
t
t 2 1 1
2 X tN tN Y
Nxy Nxy y 2 xy 2 xy
x
x y
X Nx Y Nx
Nxy Nxy
1 1 1
tN tN tN
1 2 xy 2 xy 2 x
tN
2 x

Figure 3. The effect of in-plane normal and shear forces of the top (a) and bottom (b) plates on the global moment equilibrium of
the system.

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

8 Journal of Composite Materials 0(0)

where x01 ¼ xQ – d0. Moreover, 8 continuity conditions

can be formulated between regions ‘‘1a’’ and ‘‘1q’’ at
x02 ¼ xQ þ d0, these are similar to those in equation n
(46), therefore they are not presented. Totally, the C
x2
x2
system involves 28 equations and the same number of
unknowns. P
A x3
Mindlin plate theory. The B.C.s given by equation (44) are
complemented with further two conditions (resulting in Crack tip
8 conditions): Crack front
Cγ

ð2Þ
Yð1aÞ
n ðaÞ ¼ 0, V0n ðaÞ ¼ 0 ð47Þ
Figure 4. Reference system for the J-integral.
The continuity conditions between regions ‘‘1’’ and
‘‘2’’ are (8 conditions):

t ð2Þ t ð2Þ
Uð1Þ ð1Þ ð2Þ ð1Þ
0n ð0Þ ¼ Xn ð0Þ, Xn ð0Þ ¼ Xn ð0Þ, V0n ð0Þ ¼ Yn ð0Þ, where W is the strain energy density:
2 2
Yð1Þ ð2Þ ð1Þ ð2Þ 0 0
n ð0Þ ¼ Yn ð0Þ, Wn ð0Þ ¼ Wn ð0Þ, Wn ð1Þð0Þ ¼ Wn ð2Þð0Þ,
t ð2Þ t ð2Þ Z"ij
mð1Þ ð2Þ ð1Þ ð2Þ
xn ð0Þ ¼ mxn ð0Þ þ nxn ð0Þ, mxyn ð0Þ ¼ mxyn ð0Þ þ nxyn ð0Þ
1
2 2 W¼ ij d"ij ¼ ðx "x þ y "y þ xy xy þ xz xz þ yz yz Þ
2
ð48Þ 0
ð51Þ
The latter two ones consider the effect presented in
Figure 4. The continuity conditions between regions Moreover, based on Figure 4, nk is the outward
‘‘1q’’ and ‘‘1’’ are (10 conditions) normal vector of the contour C,  ij is the stress
tensor, ui is the displacement vector, A is the area
enclosed by the contour C. The separation of the
Uð1qÞ ð1Þ ð1qÞ ð1Þ
0n ðx01 Þ ¼ U0n ðx01 Þ, Xn ðx01 Þ ¼ Xn ðx01 Þ, modes is possible by using a direct method50,51:
Vð1qÞ ð1Þ
0n ðx01 Þ ¼ V0n ðx01 Þ
pffiffiffiffiffiffiffiffiffiffi
Yð1qÞ ð1Þ ð1qÞ ð1Þ
n ðx01 Þ ¼ Yn ðx01 Þ, Wn ðx01 Þ ¼ Wn ðx01 Þ, J1 ¼ JI þ JII þ JIII , J2 ¼ 2 JI JII , J3 ¼ JIII ð52Þ
Wn0ð1qÞ ðx01 Þ ¼ W0ð1Þ
n ðx01 Þ
nð1qÞ
xn ðx01 Þ ¼ nxn ðx01 Þ, nð1qÞ
ð1Þ
xyn ðx01 Þ ¼ nð1Þ
xyn ðx01 Þ,
In our problem, x1 ¼ x, x2 ¼ z and x3 ¼ y, respect-
ively. For the calculation, we apply a zero-area path
mð1qÞ
xn ðx01 Þ ¼ mð1Þ
xn ðx01 Þ, mð1qÞ
xyn ðx01 Þ ¼ mð1Þ
xyn ðx01 Þ ð49Þ around the crack tip.36 This way the surface integral
in equation (50) becomes zero. The layerwise stress–
strain relations in laminated composite plates are1,2
Further, 10 continuity conditions can be formulated
between regions ‘‘1a’’ and ‘‘1q’’ at x02 ¼ xQ þ d0, and so 2 3ðkÞ 2 3ðkÞ 2 3
totally the system involves 36 equations. Based on the x C11 C12 0 "x
6 7 6 7 6 7
continuity conditions, the shear forces Qx, Qy are con- 4 x5 ¼ 4 C21 C22 0 5 4 "y 5,
tinuous across the delamination front. xy 0 0 C66 xy
 ðkÞ " #ðkÞ  
xz C55 0 xz
Calculation of the j-integral yz
¼
yz
ð53Þ
0 C44
50,51
In the general 3D case, the J-integral is defined as:
Z Z
Jk ¼ ðWnk  ij ui, k nj Þds  ði3 ui,k Þ,3 dA, k¼1, 2, 3 where {Cij } is the stiffness matrix of the kth laminate.
C A Taking the stresses and strains calculated by FSDT
ð50Þ back into J1 in equation (52), we obtain an expression

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

Szekrényes 9

  

including the product of stress resultants and strain 1 ð0Þ  ð0Þ 

JIII ¼ 2  Nxy1 ^xy1  Nxy2 ^xy2 
components in the form of 2 x¼þ0 x¼0

1  

  ð1Þ  ð1Þ 
1  

 Mxy1 ^xy1  Mxy2 ^xy2  ð58Þ
J1 ¼ Nx1 "ð0Þ
x1  Nx2 "ð0Þ
x2 
2 x¼þ0 x¼0
2 x¼þ0 x¼0
1  
which is valid for both theories. The production by 2 is
ð0Þ  ð0Þ 
 Ny1 "y1  Ny2 "y2 
2 x¼þ0 x¼0 necessary because we have top and bottom plates and
1  
only the top plate was analyzed. For the CLPT, the
 ð1Þ 
þ Mx1 "ð1Þ x1  M x2 " x2  stress resultants are calculated by the first of equation
2 x¼þ0 x¼0
1  
(2), the strains are calculated by equations (6), (25) and
 ð1Þ 
 My1 "ð1Þ y1  M y2 " y2  from the second of (17). In the case of FSDT, the stress
2 x¼þ0 x¼0
1  
resultants are obtained from equation (2), the strains
ð0Þ  ð0Þ 
 Nxy1 ^xy1  Nxy2 ^xy2  are determined from equations (6) and (17). Also, the
2 x¼þ0 x¼0
1  
present formulation does not incorporate any physic-
ð1Þ  ð1Þ 
 Mxy1 ^xy1  Mxy2 ^xy2  ally inconsistent parameters (e.g. shear compliances, see
2  x¼þ0
 x¼0
1 ð0Þ @w1  e.g. [8,41,45,46]), it is based on an entirely exact formu-
þ Qx1 xz1  lation including the material law of orthotropic solids.
2 @x x¼þ0
 
1 ð0Þ @w2 
 Qx2 xz2  Results and discussion
2 @x x¼0
1  
The properties of the analyzed simply-supported plate
ð0Þ  ð0Þ 
þ Qy1 yz1  Qy2 yz2  ð54Þ were (refer to Figure 3): a ¼ 105 mm (crack length),
2 x¼þ0 x¼0
c ¼ 45 mm (uncracked length), b ¼ 100 mm (plate
where e.g. Nx1 is the in-plane normal force for the dela- width), t ¼ 2 mm (half plate thickness), Q0 ¼ 1000 N
minated portion, "x 1ð0Þ is the constant part of the strain (point force magnitude), xQ ¼ 31 mm, yQ ¼ 50 mm
in the x direction as it is defined in equation (6) etc. (point of action coordinates of Q0), d0 ¼ 0.1 mm and
Moreover, the shear strains with the hat are q0 ¼ Q0/2d0. The plate is made of a carbon/epoxy
material, the lay-up of the plate was [  45f2 ;
ð0Þ @u0 @v0 ð1Þ @x @y 012;  45f2 ] for the delaminated and [  45f2 ; 012;  45f2 ]s
^xy ¼  , ^xy ¼  ð55Þ
@y @x @y @x for the uncracked part. The superscript ‘‘f’’ refers to the
fact that the cross-ply laminate is a woven fabric panel.
for the FSDT theory and The properties of the individual laminae are given in
Table 1.1 The computation was performed in the code
@u0 @v0 ð1Þ
ð0Þ
^xy ¼  , ^xy ¼ 0 ð56Þ MAPLE 1252 in accordance with the following points.
@y @x The stiffness matrices of each single layer, the stiffness
for the CLPT. The latter is not surprising after substi- matrices ([A] and [D]) of the plate, were determined
tuting equation (24) into equation (55). Considering the based on the elastic properties of the laminates as
fact that Qx, Qy and the corresponding strains are con- given in Table 1. The problem in Figure 2 was solved
tinuous across portions ‘‘1’’ and ‘‘2’’, the relevant part varying the number of Fourier series terms (N) by
of the J-integral vanishes. In the case of Kirchhoff plate creating a for-do cycle. Based on the displacement
theory, the corresponding terms are apparently zero. parameters, the stress resultants and the stresses were
The remaining part can be separated based on the calculated, while the ERRs were computed using the
direct method (refer to equation (52)) or by simply J-integral. The convergence of the results was analyzed
separating the terms including the sin (mode-II) and and it was found that after the 13th Fourier term there
cos (mode-III) functions, leading to was no change in the displacement field, stresses, stress
 resultants and ERRs.
1  

 ð0Þ 
JII ¼ 2 Nx1 "ð0Þ
x1  N "
x2 x2 
2 x¼þ0 x¼0
FE model
1  

 ð0Þ 
 Ny1 "ð0Þ
y1  N y2 " y2  In order to verify the analytical results, FE analysis was
2 x¼þ0 x¼0
1  
carried out. The 3D FE model of the plate was created
ð1Þ  ð1Þ 
þ Mx1 "x1  Mx2 "x2  in the code ANSYS 12 using 8-node solid elements.
2 x¼þ0 x¼0
1  
 Similar 3D models are documented by de Morais and
ð1Þ  ð1Þ 
 My1 "y1  My2 "y2  ð57Þ Pereira22 and Pereira and de Morais,30 therefore the
2 x¼þ0 x¼0
model is not shown here; 50, 78 and 10 elements were

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

10 Journal of Composite Materials 0(0)

applied along the plate width (y), length (x) and thick- agrees excellently with the FE results. This can be explained
ness (z), respectively. The global element size was by the fact that CLPT does not consider the transverse (or
2 mm  2 mm  0.4 mm. In the vicinity of the crack interlaminar) shear effect. Only a small amount of nonli-
tip, a refined mesh was constructed including trapezoid nearity can be observed in the numerical results.
shape elements.12 The z displacements of the contact The distribution of the normal stresses,  x and  y
nodes over the delaminated surface were imposed to along the same lines as in Figure 5, are demonstrated
be the same. The mode-II and mode-III ERRs were in Figure 6. The distributions were determined layer-
calculated by the VCCT; the size of the crack tip elem- wise using the stress–strain relations given by equation
ents were x ¼ 0.2 mm, y ¼ 0.2 mm and z ¼ 2 mm. (53). In the case of  x, there are significant differences
For the determination of GII and GIII along the delam- between the analytical and FE results. The reasons for
ination front a so-called MACRO was written in the these discrepancies are the followings. First, the FE
ANSYS Design and Parametric Language (ADPL). solution is based on the nodal stresses, which are cal-
The MACRO gets the nodal forces and displacements culated by taking the average of the stresses from the
at the crack tip at each pair of nodes, respectively, eight neighboring elements in the mutual node. That is
then by defining the size of crack tip elements it deter- why it is not possible to obtain piecewise continuous
mines and plots the ERRs at each node along the crack distributions by the FEM. Second, it is seen that
front. although the FEM provides average stresses, the
linear approximation is unable to capture accurately
the strain and stress states in the vicinity of the delam-
Analytical and numerical results
ination front. In a recent paper,53 it has been found that
Figure 5 shows the in-plane displacements in the vicinity of second-order plate theory gives much better predictions
two points of the plate: u(0,b/2,z) and v(0,b,z), i.e. each for the stress state. In the case of  y, the agreement
point lies in the crack front. It is seen that the CLPT under- between analytical and numerical results is better.
predicts the FE solution; on the other hand, the FSDT Finally, the difference between CLPT and FSDT is

Table 1. Elastic properties of single carbon/epoxy laminates.

Ex (GPa] Ey (GPa) Ez (GPa) Gyz (GPa) Gxz (GPa) Gxy (GPa)

yz
xz
xy

45 f 16.39 16.39 16.4 5.46 5.46 16.4 0.5 0.5 0.3
0 148 9.65 9.65 4.91 4.66 3.71 0.27 0.25 0.3

Z, z
2 2
±45°f FSDT (Mindlin) FSDT (Mindlin)
z(t) Kirchhoff Kirchhoff
4
z(t) FEM FEM
3
0° 1 1
x,y
z(t)
2
z(t)
1
±45°f
Z [mm]

Z [mm]

0 0
±45°f X,Y
z(b)
4
z(b)
3
0° −1 −1
x,y
z(b)
2
z(b)
1
±45°f
−2 −2
−0.10 −0.05 0.0 0.05 0.10 −0.15 −0.10 −0.05 0.0 0.05 0.10 0.15
u(0,b/2,z) [mm] v(0,b,z) [mm]

Figure 5. Distribution of the in-plane displacements over the plate thickness.

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

Szekrényes 11

not significant. It has to be mentioned that in the FE FSDT (piecewise constant, purple line) as well as the
model, the displacement and stress continuity is solution calculated by the 3D equilibrium equations are
ensured, but in the analytical models only the continu- equally presented. The latter ones were calculated by
ity of displacement parameters and shear forces Qx, Qy the r ¼ 0 equation2,49 including both theories. The
(FSDT) can be realized. mismatch between the FE and analytical solutions is
Consequently, the stresses are not continuous in significant, especially in the interface plane of the
accordance with either the CLPT or FSDT, however, plate, where the FE model predicts a sharp peak.
the discrepancies in stresses are not so significant in the From another point of view, the FSDT and 3D equi-
transition between the delaminated and undelaminated librium solutions agree well, while the CLPT predicts
plate portions. negligible shear stresses. The disagreement between the
The distributions of transverse shear stresses are numerical and analytical models is attributed to the fact
plotted in Figure 7. The FE solution, the solution by that the assumed linear displacement field in terms of z

Z, z Kirchhoff Mindlin FE solution

2 2
±45°f
z(t)
4
z(t)
3
0° 1 1
x,y
z(t)
2
z(t)
1
±45°f
Z [mm]

Z [mm]
0 0
±45°f X,Y
z(b)
4
z(b)
3
0° −1 −1
x,y
z(b)
2
z(b)
1
±45°f
−2 −2
−150 −100 −50 0 50 100 150 −100 −50 0 50 100
σx (0,b/2,z) [MPa] σy (0,b/2,z) [MPa]

Figure 6. Distribution of the normal stresses over the plate thickness.

Z, z
2 2
±45°f
z(t) FSDT (Mindlin) FSDT (Mindlin)
4
FSDT-3D eq. FSDT-3D eq.
z(t)
3 Kirchhoff-3D eq. Kirchhoff-3D eq.
0° 1 FE solution 1 FE solution
x,y
z(t)
2
z(t)
1
±45°f
Z [mm]

Z [mm]

0 0
±45°ff X,Y
z(b)
4
z(b)
3
0° −1 −1
x,y
z(b)
2
z(b)
1
±45°f
−2 −2
−30 −20 −10 0 10 −60 −50 −40 −30 −20 −10 0
τxz (0,b/2,z) [MPa] τyz (0,0,z) [MPa]

Figure 7. Distributions of shear stresses over the plate thickness.

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

12 Journal of Composite Materials 0(0)

is unable to provide better results. The distributions of effect is not considered by CLPT, and the absence of
the in-plane normal force Nx(x,y) and the shear force the independent rotational parameters results in the
Nxy(x,y) are plotted in Figure 8. discrepancies between the stress resultants presented
Although these stress resultants could be calculated in Figure 8(a)–(d). Also, the different displacement
by the FE model too (by integrating the normal and in- fields of CLPT and FSDT lead to different equilibrium
plane shear stresses in the through-thickness direction), equations (compare equation (18)–(20) and (29)) and so
this would be a very long process, therefore, in this case different stress resultants.
only analytical results are presented. It is seen that near The interlaminar shear stress (¼transverse shear in
the delamination front (x ¼ 0) these stress resultants accordance with the duality of shear stresses) distribu-
change suddenly and significantly. Moreover, Nx tions along the global midplane of the undelaminated
involves the sin, while Nxy involves the cos function plate portion (top plate, refer to Figure 1) can be cal-
(refer to equation (43)) leading to the completely differ- culated based on the FSDT as:
ent nature of these stress resultants. One of the advan-  
tages of plate theory over the FE model is that these 45 45 @w2
xz z¼2 ¼ G13 xz ¼ G13 x2 þ
j t ,
plots can be obtained very simply in the analytical way. @x
Considering the values, the difference between the   ð59Þ
 @w2
CLPT (Figure 8(a) and (b)) and FSDT (Figure 8(c) yz z¼ t ¼ G45 45
23 yz ¼ G23 y2 þ
2 @y
and (d)) theories is significant. The FSDT theory indi-
cates that the stress resultants change more suddenly at where the shear moduli are defined in Table 1. The
the crack front (x ¼ 0) compared to the CLPT. The function plots are given by Figure 9. Both stresses
reason for that is again the fact that transverse shear change significantly near the delamination front and

(a) (b)
Nx (x,y) [N/mm] Nxy (x,y) [N/mm]

40 24
30
4
20
-16
10

0 −44 −36 −44

0 −34 0 −34
25 −24 25 −24
X [mm] 50 −14 X [mm] 50 −14
75 −4 75 −4
100 Kirchhoff 100
Y [mm] Y [mm]

(c) (d)

Nx (x,y) [N/mm] Nxy (x,y) [N/mm]

36
27
16
17
−4
7 −24

−3 −44 −44 −44

0 −34 0 −34
25 −24 25 −24
X [mm] 50 −14 X [mm] 50 −14
75 −4 75 −4
100 Mindlin 100
Y [mm] Y [mm]

Figure 8. Distribution of the in-plane normal and shear forces over the uncracked (top) plate portions.

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

Szekrényes 13

decay suddenly by getting far from it. It is important to curvature are proportional to this derivative.
note that  xz does not vanish at the x ¼c boundary, on Consequently, in the case of CLPT GIII comes from
the other hand,  yz vanishes entirely at the same the in-plane shear force Nxy only. Seeing the results
location. by FSDT, a better agreement with the numerical results
The mode-II, mode-III ERRs and the mode ratio is obtained. Although the mode-II ERR compared to
along the delamination front are plotted in Figure 10. the VCCT result is significantly underpredicted, the
The symbols show the results of the VCCT, while the overall agreement is acceptable. These results show
curves represent the results by CLPT and FSDT. that a displacement field with at least five parameters
According to Figure 10(a), it can be seen that the (three parameters with interface constraints) is neces-
CLPT produces some overprediction for GII compared sary to reach an acceptable agreement with the VCCT
to the VCCT but underpredicts significantly GIII. We results. The mode ratios are plotted in Figure 10(b).
note that in accordance with equations (58) and (56), The main conclusion is that the FSDT captures this
the ERR by Mxy vanishes in the case of CLPT. The ratio even better than CLPT. It is important to mention
reason for that is the continuity of the derivative of that the VCCT method is mesh-sensitive to a certain
the deflections in equation (45) in the transition zone degree and the investigation of the effect of mesh refine-
and that both the twisting moment and twisting ment was outside the scope of this paper.

(a) (b)
τxz (x,y) [MPa] τyz (x,y) [MPa]

10
−0,6
5
−1,6
0

−2,6 −5

−3,6 −44 −10 −44

0 −34 0 −34
25 −24 25 −24
X [mm] 50 −14 X [mm] 50 −14
75 −4 75 −4
100 100
Y [mm] Y [mm]

Figure 9. Distribution of the interlaminar shear stresses over the uncracked (top) plate portion.

(a) (b)
G [J/m2] GII/GIII

250 4
GII, FSDT(Mindlin)
GIII, FSDT(Mindlin) FSDT (Mindlin)
GII, Kirchhoff Kirchhoff
200 VCCT
GIII, Kirchhoff 3
GII, VCCT
GIII, VCCT
150
2
100

1
50

0 0
0 20 40 60 80 100 y [mm] 0 20 40 60 80 100 y [mm]

Figure 10. Distribution of the ERRs and the mode ratio by CLPT, FSDT and VCCT. ERR: energy release rate; CLPT: classical
laminated plate theory; FSDT: first-order shear deformable plate theory; VCCT: virtual-crack closure technique.

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

14 Journal of Composite Materials 0(0)

(a) (b)
G [J/m2] GII/GIII

140 GII, FSDT GII, Kirchhoff

4
FSDT (Mindlin)
GIII, FSDT GIII, Kirchhoff
Kirchhoff
120 GII, VCCT
VCCT
GIII, VCCT
3
100

80
2
60

40 1
20

0 0
0 20 40 60 80 100 120 140 160 y [mm] 0 20 40 60 80 100 120 140 160 y [mm]

Figure 11. Distribution of the ERRs and the mode ratio by CLPT, FSDT and VCCT (b ¼ 160 mm). ERRs: energy release rate; CLPT:
classical laminated plate theory; FSDT: first-order shear deformable plate theory; VCCT: virtual-crack closure technique.

For the sake of completeness, the ERRs and mode The present model eliminates the physically incon-
ratios were also calculated for the case when the plate sistent shear compliance of the flexible joint models and
width was b ¼ 160 mm. The results are presented in includes the effect of interface deformation based on the
Figure 11. Although the CLPT captures better the equations of linear elasticity and the material law of
mode-II ERR in the middle part of the plate, the orthotropic solids. It was shown that although the dis-
mode-III ERR differs significantly in comparison with placement components are continuous across the
the FE results. The FSDT provides better results, espe- delamination, there are stress resultants, which remain
cially in the case of GIII. Independent of the fact discontinuous. Moreover, the shear forces vanish in the
whether the CLPT or FSDT theory is utilized, the ana- J-integral, and the mode-II and mode-III ERRs are
lytical models do not capture the edge effect. More defined in a relatively simple way. However, it must
clearly, the FE solution indicates some oscillatory be mentioned that only mixed-mode II/III fracture
change at the plate edges; on the contrary, the analyt- problems were considered in this study and the trans-
ical solutions remain smooth. The final conclusion is verse deflection of the top and bottom plates was con-
that the transverse (or interlaminar) shear effect plays sidered to be the same. It has been shown that the
an important role in the development of an accurate difference between the FE and plate theory solution
plate model to analyze delaminated composite plates becomes the largest at the edges of the plate.
even in that case when the plate is relatively thin. Considering the available methods for the calcula-
tion of the ERR in plates, the first alternative is in
general the VCCT. However, for the 3D FE model,
Conclusions the computation could be lengthy, especially if the
The classical laminated and first-order shear deform- model size is relatively large. Furthermore, in the
able plate theories are utilized in this work to develop crack tip a refined mesh should be constructed to
a double-plate system for delaminated orthotropic obtain accurate ERR values. Finally, in most of the
composite plates. The models are based on the continu- commercial FE packages, the VCCT has not yet been
ity of the displacement field across the delamination implemented. The present work provides another pos-
front by imposing the kinematic constraints along the sibility for the calculation of the ERR in plates sub-
interface. A simply-supported delaminated plate sub- jected to bending. The possible application field of the
jected to a point force was analyzed using Lévy plate presented method is the fracture mechanics of compos-
formulation, the stresses and the ERRs were calculated. ite materials. In the last few years, fracture test methods
The results were compared to those of a 3D FE model including plate-like specimens have been developed to
and a comparison was made among the results characterize the mode-III, mixed-mode II/III and
obtained. The final conclusion is that transverse shear mixed-mode I/III fracture behavior of laminated mater-
effect and the coupling between the stress resultants ials.18,22,30 By preparing a detailed user-friendly work-
play important role in the development of an accurate sheet in MAPLE, it is possible to provide a data
plate model for delaminated composite plates. reduction scheme for the experimentalists. Also, the

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

Szekrényes 15

application to asymmetrically delaminated orthotropic 10. Plain KP and Tong L. An experimental study on mode I
and angle-ply laminated plates as well as sandwich and II fracture toughness of laminates stitched with a
panels needs to be investigated. The present model one-sided stitching technique. Compos Part A Appl Sci
can be implemented in the vibration analysis of Manufact 2011; 42: 203–210.
simply-supported composite plates too. Finally, the 11. Reeder JR and Crews JR. Mixed-mode bending method
for delamination testing. AIAA J 1990; 28: 1270–1276.
analysis developed in this paper may be utilized as a
12. Davidson BD and Sundararaman V. A single leg bending
means for benchmarking fracture mechanical tools54
test for interfacial fracture toughness determination. Int J
that are available in commercial FE packages. These Fract 1996; 78: 193–210.
tasks will be carried out in the near future. 13. Lee SM. An edge crack torsion method for mode III
delamination fracture testing. J Compos Technol Res
Funding 1993; 15: 193–201.
This research received no specific grant from any funding 14. Li J and O’Brien TK. Simplified data reduction
agency in the public, commercial, or not-for-profit sectors. methods for the ECT test for mode III interlaminar
fracture toughness. J Compos Technol Res 1996; 18:
Conflict of Interest 96–101.
15. Szekrényes A. Improved analysis of the modified split-
None declared. cantilever beam for mode III fracture. Int J Mech Sci
2009; 51: 682–693.
Acknowledgments 16. Browning G, Carlsson LA and Ratcliffe JG.
This work was supported by the János Bolyai Research Modification of the edge crack torsion specimen for
Scholarship of the Hungarian Academy of Sciences. This mode III delamination testing. Part II-experimental
work is connected to the scientific program of the study. J Compos Mater 2011; 45: 2633–2640.
‘‘Development of quality-oriented and harmonized RþDþI 17. Browning G, Carlsson LA and Ratcliffe JG. Redesign of
strategy and functional model at BME’’ project. This project the ECT test for mode III delamination testing. Part I:
is supported by the New Hungary Development Plan (Project Finite element analysis. J Compos Mater 2010; 44:
ID: TÁMOP-4.2.1/B-09/1/KMR-2010-0002). 1867–1881.
18. de Morais AB and Pereira AB. Mode III interlaminar
References fracture of carbon/epoxy laminates using a four-point
bending plate test. Compos Part A Appl Sci Manufact
1. Kollár LP and Springer GS. Mechanics of composite struc-
2009; 40: 1741–1746.
tures. Cambridge, New York, Melbourne, Madrid, Cape
19. de Moura MFSF, Fernandez MVC, de Morais AB, et al.
Town, Singapore, São Paolo, PA: Cambridge University
Numerical analysis of the edge crack torsion test for
Press, 2003.
mode III interlaminar fracture of composite laminates.
2. Reddy JN. Mechanics of laminated composite plates and
shells – Theory and analysis. Boca Raton, London, New Eng Fract Mech 2009; 76: 469–478.
York, Washington DC, PA: CRC Press, 2004. 20. Mehrabadi FA and Khosravan M. Mode III interlaminar
3. Rudolph Sz. Theories and applications of plate analysis. fracture and damage characterization in woven fabric-
Hoboken, New Jersey, PA: John Wiley & Sons, 2004. reinforced glass/epoxy composite laminates. J Compos
4. Carlsson LA and Kardomateas GA. Structural and failure Mater 2012. DOI: 10.1177/0021998312449770.
mechanics of sandwich composites. Dordrecht, Heidelberg, 21. Szekrényes A. Delamination fracture analysis in the
London, New York. PA: Springer, 2004. GII-GIII plane using prestressed composite beams. Int J
5. Anderson TL. Fracture mechanics – fundamentals and Solid Struct 2007; 44: 3359–3378.
applications, 3rd edn. Boca Raton, London, New York, 22. de Morais AB and Pereira AB. Mixed mode II þ III
Singapore, PA: CRC Press, Taylor & Francis Group, interlaminar fracture of carbon/epoxy laminates.
2005. Compos Sci Technol 2008; 68: 2022–2027.
6. Jumel J, Budzik MK and Shanahan MER. Beam on elastic 23. Kondo A, Sato Y, Suemasu H, et al. Characterization of
foundation with anticlastic curvature: Application to ana- fracture resistance of carbon/epoxy composite laminates
lysis of mode I fracture tests. Eng Fract Mech 2011; 78: during mixed-mode II and III stable damage propaga-
3253–3269. tion. J Jpn Soc Compos Mater 2010; 36(5): 179–188.
7. Peng L, Zhang J, Zhao L, et al. Mode I delamination 24. Suemasu H, Kondo A, Gozu K, et al. Novel test method
growth of multidirectional composite laminates under fati- for mixed mode II and III interlaminar fracture tough-
gue loading. J Compos Mater 2011; 45: 1077–1090. ness. Adv Compos Mater 2010; 19(4): 349–361.
8. Wang J and Qiao P. Novel beam analysis of the end 25. Suemasu H, Kondo A, Gozu K, et al. Double notched
notched flexure specimen for mode-II fracture. Eng Fract split cantilever test method to measure the mixed mode II
Mech 2004; 71: 219–231. and III interlaminar toughness. In: ICCM-17 17th
9. Argüelles A, Viña J, Canteli AF, et al. Influence of resin International Conference on Composite Materials. Paper
type on the delamination behavior of carbon fiber rein- ID. F9.6. 27 Jul 2009–31 Jul 2009, pp. 1–10, Edinburgh
forced composites under mode-II loading. Int J of International Convention Centre (EICC), Edinburgh,
Damage Mech 2011; 20(7): 963–978. UK.

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

16 Journal of Composite Materials 0(0)

26. Kondo A, Sato Y, Suemasu H, et al. Fracture resist- 43. Qiao P and Chen F. On the compliance and energy
ance of carbon/epoxy composite laminates under release rate of generically-unified beam-type fracture spe-
mixed-mode II and III failure and its dependence on frac- cimens. J Compos Mater 2011; 45(1): 65–101.
ture morphology. Adv Compos Mater 2011; 20(5): 44. Chen F. Interface mechanics of layered composite beam-
405–418. type structures. PhD Thesis. Department of Civil and
27. Szekrényes A. Interlaminar fracture analysis in the GII- Environmental Engineering, Washington State
GIII plane using prestressed composite beams. Compos University, 2011.
Part A Appl Sci Manufact 2012; 43: 95–103. 45. Szekrényes A. Interlaminar stresses and energy release
28. Miura M, Shindo Y, Takeda T, et al. Interlaminar frac- rates in delaminated orthotropic composite plates. Int J
ture characterization of woven glass/epoxy composites Solid Struct 2012; 49: 2460–2470.
under mixed-mode II/III loading conditions at cryogenic 46. Szekrényes A. Interface crack between isotropic
temperatures. Eng Fract Mech 2012; 96: 615–625. Kirchhoff plates. Meccanica 2013; 48(3): 507–526.
29. Mehrabadi FA. Analysis of pure mode III and mixed 47. Rice JR. A path independent integral and the approxi-
mode (IIIþII) interlaminar crack growth in polymeric mate analysis of strain concentration by notches and
woven fabrics. Mater Design 2013; 44: 429–437. cracks. J Appl Mech 1968; 35: 379–386.
30. Pereira AB and de Morais AB. Mixed-mode I þ III inter- 48. Cherepanov GP. Methods of fracture mechanics: Solid
laminar fracture of carbon/epoxy laminates. Compos Part matter physics. Dordrecht, Boston, London, PA:
A Appl Sci Manufact 2009; 40: 518–523. Kluwer Academic Publishers, 1997.
31. Szekrényes A. Interlaminar fracture analysis in the 49. Chou PC and Pagano NJ. Elasticity - tensor, dyadic, and
GI-GIII plane using prestressed composite beams. engineering approaches. Princeton, New Jersey, Toronto,
Compos Part A Appl Sci Manufact 2009; 40: 1621–1631. London, PA: D. Van Nostrand Company Inc, 1967.
32. Davidson BD, Sediles FO and Humphreys KD. A shear- 50. Rigby RH and Aliabadi MH. Decomposition of the
torsion-bending test for mixed-mode I-II-III delamin- mixed-mode J-integral - revisited. Int J Solid Struct
ation toughness determination. In: 25th Technical 1998; 35: 2073–2099.
51. Shivakumar KN and Raju IS. An equivalent domain
Conference of the American Society for Composites and
integral method for three- dimensional mixed-mode frac-
14th US-Japan Conference on Composite Materials (ed JB
ture problems. Eng Fract Mech 1992; 42: 935–959.
Lantz), 2010, 2, pp.1001–1020. Curran Associates, Inc.
52. Garvan F. The Maple Book. Boca Raton, London, New
33. Szekrényes A. Interlaminar fracture analysis in the GI-
York, Washington DC, PA: Chapman & Hall/CRC,
GII-GIII space using prestressed composite beams. J
2002.
Reinf Plast Compos 2011; 30: 1655–1669.
53. Szekrényes A. Interface fracture in orthotropic composite
34. Davidson BD and Sediles FO. Mixed-mode I-II-III
plates using second-order shear deformation theory. Int J
delamination toughness determination via a shear-tor-
Damage Mech 2013. DOI: 10.1177/1056789513478957.
sion-bending test. Compos Part A Appl Sci Manufact
54. Orifici A and Krueger R. Benchmark assessment of auto-
2011; 42: 589–603.
mated delamination propagation capabilities in finite
35. Davidson BD, Yu L and Hu H. Determination of energy
element codes for static loading. Finite Element Anal
release rate and mode mix in three-dimensional layered Design 2012; 54: 28–36.
structures using plate theory. Int J Fract 2000; 105:
81–104.
36. Sankar BV and Sonik V. Pointwise energy release rate in
delaminated plates. AIAA J 1995; 33: 1312–1318. Appendix 1
37. Park O and Sankar BV. Crack-tip force method for com-
puting energy release rate in delaminated plates. Compos Coefficients for equations (21)–(23) and (38)
Struct 2002; 55: 429–434.
38. Pao YC. Simple bending analysis of laminated plates by a^ 1 ¼ A11 t2 =4 þ D11 , a^ 2 ¼ A66 t2 =4 þ D66 , a^ 3 ¼ a^5
large-deflection theory. J Compos Mater 1970; 4:
380–389.
¼ kA55 , a^ 4 ¼ ðA12 þ A66 Þt2 =4 þ D12 þ D66
39. Whitney JM. The effect of transverse shear deformation b^1 ¼ a^4 , b^2 ¼ a^2 , b^3 ¼ A22 t2 =4 þ D22 , b^4 ¼ b^5 ¼ kA44
ont he bending of laminated plates. J Compos Mater
c^1 ¼ c^3 ¼ kA55 , c^2 ¼ c^4 ¼ kA44 , c^5 ¼ 1
1969; 3: 534–547.
40. Shi G and Bezine G. A general boundary integral formu- d^1 ¼ ð2 a^ 2  a^ 3 Þ=a^ 1 , d^2 ¼ a^ 4 =a^1 , d^3 ¼ a^ 5 =a^1
lation for the anisotropic plate bending problems. J
Compos Mater 1988; 22: 694–716. e^1 ¼ b^1 =b^2 , e^2 ¼ ð2 b^3  b^4 Þ=b^2 , e^3 ¼ b^5 =b^2
41. Qiao P and Wang J. Mechanics and fracture of crack-tip f^1 ¼ c^1 =c^3 , f^2 ¼ c^2 =c^3 , f^3 ¼ 2 c^4 =c^3 , f^4 ¼ c^5 =c^3
deformable bimaterial interface. Int J Solid Struct 2004;
41: 7423–7444.
ð60Þ
42. Wang J and Qiao P. Interface crack between two shear
deformable elastic layers. J Mech Phys Solid 2004; 52:
891–905.

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

XML Template (2013) [16.5.2013–10:28am] [1–17]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JCMJ/Vol00000/130360/APPFile/SG-JCMJ130360.3d (JCM) [PREPRINTER stage]

Szekrényes 17

Appendix 2
Coefficients for equation (34)

a1 ¼ A11 , a2 ¼ A66 , a3 ¼ A12 þ A66 , ð61Þ

b1 ¼ a3 , b2 ¼ a2 , b3 ¼ A22
c1 ¼ D11 , c2 ¼ D66 , c3 ¼ c5 ¼ kA55 , c4 ¼ D12 þ D66
d1 ¼ c4 , d2 ¼ c2 , d3 ¼ D22 , d4 ¼ d5 ¼ kA44
e1 ¼ e2 ¼ kA55 , e2 ¼ e4 ¼ kA44 , e5 ¼ 1
f1 ¼ 2 a2 =a1 , f2 ¼ a3 =a1
g1 ¼ b1 =b2 , g2 ¼ 2 b3 =b2
h1 ¼ ð2 c2  c3 Þ=c1 , h2 ¼ c4 =c1 , h3 ¼ c5 =c1
j1 ¼ d1 =d2 , j2 ¼ ð2 d3  d4 Þ=d2 , j3 ¼ d5 =d2
k1 ¼ e1 =e3 , k2 ¼ e2 =e3 , k3 ¼ 2 e4 =e3 , k4 ¼ e5 =e3
ð62Þ

Downloaded from jcm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016