Composite Structures 184 (2018) 554–567

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Spatial vulnerability analysis for the first ply failure strength of composite MARK
laminates including effect of delamination
⁎
P.K. Karsha, T. Mukhopadhyayb, , S. Deya
a
Mechanical Engineering Department, National Institute of Technology Silchar, India
b
Department of Engineering Science, University of Oxford, Oxford, United Kingdom

A R T I C L E I N F O A B S T R A C T

Keywords: The present investigation deals with the first ply failure strength of laminated composite plates for spatial
First ply failure variation of loading position, which can render a clear idea about the locational sensitivity of the loading po-
Laminated composites sitions in a two dimensional space. In this context, the effect of delamination is investigated on the failure
Stacking sequence strengths considering angle ply and cross ply laminates. A finite element model is developed based on different
Failure criteria
failure criteria of composites, such as maximum stain, maximum stress, Tsai-Hill, Tsai-Wu-Hahn and Tsai-Hill-
Spatially varying loading position
Hoffman. An eight noded isoparametric quadratic element is considered in the present finite element for-
Stochastic analysis
Locational sensitivity mulation incorporating transverse shear and rotary inertia. Results are presented in deterministic as well as
stochastic regime. For obtaining the probabilistic descriptions of failure strengths following different failure
criteria, Monte Carlo simulation is carried out in conjunction with the finite element model following a non-
intrusive approach. The variation of failure strength is portrayed considering the effect of stacking sequence, ply
orientation, number of layers, degree of orthotropy and ply thickness. In this article, consideration of stochastic
material and structural attributes along with critical service-life characteristics such as delamination for spatially
varying loading positions provides a comprehensive understanding about the failure strength of composite la-
minates for practical applications.

1. Introduction interactive effects of elastic-brittle rupture of fibers and yielding of the

matrix. The assessment of the failure process can be predicted by the
Modern competitive industries exhaustively use the laminated maximum amount of load withstood prior to first-ply failure analysis
composite materials. From under water to ground even space vehicles, based on some failure criteria. In spite of the practical importance of the
applications of composites are found as cost-effective and efficient, aspect of failure in composite structures, the number of technical papers
while being more sound and application specific in terms of structural and reports dealing with this crucial subject is limited due to the
attributes. The extended applications range from medical prosthetic complexity involved. A concise review on this aspect is provided in the
devices, sports equipment, automotive parts to high-performance de- next paragraph.
fence structures. The use of composite materials in structural applica- A review of all the failure theories of composite materials was
tions is dictated by the outstanding strength and stiffness, low specific presented by Talreja [1]. Dvorak, Laws, Reddy and Pandey [2–5] were
gravity, low maintenance costs, and the flexibility in tailoring the the first pioneers to analyze the first ply failure of the composite la-
stiffness and strength in the preliminary design stage of complex minates including cracks induced failure. Bruno and Zinno [6] extended
structures. The use of composite structures in advanced underwater, their work on the composite plate using the Reissner-Mindlin plate
ground and space vehicles can result in significant increase in payload, theory that accounts for moderate rotation. Almost half a decade later
weight reduction, range and speed, maneuverability, fuel efficiency and [7–9], various works on first ply failure using the non-linear and pro-
safety. The functional requirements and economic considerations of gressive analysis methods were adopted. Different theories were ap-
design require designers to use reliable and accurate but economical plied as to analyze the crack development which could lead to the
methods of determining stresses and identifying failure mechanisms. failure of the laminate. Reliability analysis using the finite element
The mechanism of failure of laminated composites is complex in nature method was performed during the mid-90s by Kam and Chao [10] and
predominantly governed by dissimilar phases of independent and Kam and Chang [11] in this context. Later the above methods were

⁎
Corresponding author.
E-mail address: [email protected] (T. Mukhopadhyay).
URL: https://www.tmukhopadhyay.com (T. Mukhopadhyay).

http://dx.doi.org/10.1016/j.compstruct.2017.09.078
Received 16 May 2017; Received in revised form 28 August 2017; Accepted 26 September 2017
Available online 06 October 2017
0263-8223/ © 2017 Elsevier Ltd. All rights reserved.
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

applied to the composite shells [12] and other physical characteristics Moreover, stochasticity in structural and material attributes of com-
and responses of composite laminates such as buckling and bending posites may lead to significant variability in the predicted failure
[13–18], wherein the composite structures were studied along with the strengths. Even though recent studies have focused on different aspects
progressive failure. Of late, Kumar and Srivastava [21] analyzed the of probabilistic [56–67] and non-probabilistic [68] variabilities in
failure of stiffened plate, while Chang [19] studied the experimental material and structural properties of composite structures to quantify
and theoretical analysis of the first ply failure on composite laminate. their effect on dynamics and stability characteristics, the aspect of
Pal and Ray [20] investigated the progressive failure analysis of lami- failure is yet to be addressed in this context. Aim of this article is to
nated composites using the finite element method. Rattana- comprehensively analyze the effect of spatially varying loading position
wangcharoen [22] studied on the first ply failure analysis on the la- on the failure strength of composites considering service-life char-
minated composite cylindrical panels employing the 3D layerwise acteristics like delamination and the effect of stochasticity in material
mixed stress-displacement finite element model. The other subsequent and structural attributes to closely simulate the practical field situation.
works in this area focused on the failure of composite laminates with This paper, hereafter, is organized as follows: Section 2 describes the
various physical characteristics and responses including the presence of mathematical formulation for the failure of composite laminates along
different external influences such as static and impulsive water pres- with the effect of delamination and stochasticity; Section 3 presents
sure, static and fatigue loads etc. [23–28]. A Statistical model of the deterministic as well as stochastic results concerning the failure
transverse ply cracking in cross-ply laminates by strength and fracture strength of composite laminates considering different failure criteria;
toughness based failure criteria was developed by Andersons and finally Section 4 renders a perspective of this article along with con-
Spārniņš [29], who studied the elucidation of the relative importance of clusion and the prospective path of future research.
the crack initiation and propagation phenomena in the fragmentation
process at different transverse and longitudinal ply thickness ratios. 2. Theoretical formulation
Meiche et. al. [30] studied the analysis of hybrid cross-ply composite
laminates using the modified shear-lag analysis to evaluate the effect of The governing equation of a laminated composite plate (refer to
transverse cracks on the stiffness reductions. Adali and Cagdas [31] Fig.1) is derived based on the principle of minimum total potential
studied the failure analysis of curved composite panels based on first- energy. The total potential energy ‘Π’ is expressed as sum of strain
ply and buckling failures. Gadade et al. [32] worked on the stochastic energy ‘U’ and work done due to external load ‘V’ as,
buckling and first ply failure analysis of laminated composite plates (1)
Π=U+V
using the higher order shear deformation theory in conjunction with
Tsai-Wu failure theory. Gohariet al. [33] and Ghosh and Chakravorty Strain energy can be expressed as a volume integral,
[34] modelled the progressive failure behavior for unsymmetrical 1
composite laminates using the prediction theory. Moreno et al. [35]
U=
2
∫ϕ {ε}T {σ } dϕ (2)
determined the difference in behavior of laminated composite under
Work done by external load is the area integral,
tensile and compressive loading in context to the failure analysis of
composites. Patel and Guedes Soares [36] investigated on system V= ∫ ∫A {u}T {q} dA (3)
probability of failure and sensitivity under low velocity impact while
Romanowicz [37] used computational micromechanics to find the ef- External load on plate can be expressed as {q} = { 0 0 qz 0 0 }T , where
fect of first ply failure when subjected to uniaxial tensile load. Rehan {qz } represents transverse load intensity on the plate. The constitutive
et al. [38] investigated the effect of ply orientation angle considering equation of the plate is given by,
carbon/epoxy composite and found that toughness decrease with both {F } = [D]{ε } (4)
sub-adjacent and adjacent ply angle at the crack initiation. Dimitri et al.
[39] and Abdullah et al. [40] presented the effect of crack and dela- where the stress resultant vector is {F}, the strain vector is {ε} and the
mination in composites using XFEM and SFEM. Kumar et al. [41,42] laminate elasticity matrix is [D]. Assuming linearly elasticity and ne-
investigated on the failure of sandwich panels considering deterministic glecting the normal stress perpendicular to the plate of the lamina, the
as well as probabilistic input parameters. Of late, Naskar et al. [43] constitutive relations, in the principal material direction 1, 2 and 3 are
have studied the effect of matrix cracking on the dynamic response of given by
composite circular beams in a probabilistic framework, while Li et al.
σ ⎡Q11 Q12 0 0 0 ⎤ ε1
[44] have carried out analytical and numerical investigation on the ⎧ σ1 ⎫ ⎢Q Q 0 0 0 ⎥ ⎧ ε2 ⎫
stiffness matrix for edge-cracked circular shafts. Yao et al. [45] used ⎪ 2 ⎪ ⎢ 12 22 ⎪ ⎪
τ12 = 0 0 Q66 0 0 ⎥ ε12
Paris relation and Hartman-Schijve equation for determining the fa- ⎨ τ23 ⎬ ⎢ ⎥⎨ε ⎬
⎪τ ⎪ ⎢ 0 0 0 Q44 0 ⎥ ⎪ 23 ⎪
tigue delamination of composite delaminates. Jang and Kim [46] uti- ⎩ 13 ⎭ ⎢ 0 0 0 0 Q55 ⎥ ⎩ ε13 ⎭ (5)
lized high speed Fiber Bragg Grating (FBG) sensing system for the ⎣ ⎦
analysis of delamination of composite when subjected to impact where,
loading, while Yelve et al. [47] used Lamb wave based nonlinear
method to determine the delamination of composite and later on, a
generalized probabilistic approach (GPA) is employed on failure as-
sessment by Carpinteri et al. [48]. Progressive failures of composite
with different loading conditions are analyzed [49–53]. The floating
node method is employed to modelling tensile failure of composite by
Chen et al. [54] while a hybrid failure criterion is applied to determine
the matrix failure of composite by Chowdhury et al. [55].
The literature review presented above shows that some of the cru-
cial aspects in context to failure of composites have not yet received
proper attention, such as the locational sensitivity of applied load and
the effect of service-life characteristics like delamination. The point of
application of load on the composite panel has considerable effect in Fig. 1. Laminated cantilever composite plate.
the failure strength and it is also significantly affected by delamination,
which is one of the most common modes of damage in composites.

555
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

E1 E2 ν12 E2 1, 2 and 3 are freely allowed to deform prior to imposition of the

Q11 = Q
1 − ν12 ν 21 22
= 1 − ν12 ν 21
and Q12 = 1 − ν12 ν 21
constraints conditions. The nodal displacements of elements 2 and 3 at
Q66 = G12 Q44 = G23 and Q55 = G13 (6)
the crack tip are expressed as [69]
The constitutive relations of the lamina, with respect to the x, y and z
u j = u jo−(z−zoj ) θxj
axes, are expressed as
νj = vjo−(z−zoj ) θyj
σ ⎡Q11 Q12 Q16 0 0 ⎤ εxx wj = wjo (where, j= 2,3) (12)
⎧ σxx ⎫ ⎢ ⎧ ⎫
⎪ yy ⎪ ⎢Q12 Q22 Q26 0 0 ⎥ ⎪ εyy ⎪
τxy = Q16 Q26 Q66 0 ⎥ where uoj,
and voj woj
are the mid-plane displacements and is the z- zoj
0 ⎥ εxy
⎨ τyz ⎬ ⎢ ⎨ ⎬ coordinate of mid-plane of element j and θx, θy are the rotations about x
⎪ ⎪ ⎢ 0 0 0 Q44 Q45 ⎥ ⎪ εyz ⎪
τxz ⎢ ⎥ εxz and y axes, respectively. The above equation also holds good for ele-
⎩ ⎭ 0 Q45 Q55 ⎦ ⎩ ⎭
⎣ 0 0 (7)
ment 1 and zo1 equal to zero. The transverse displacements and rotations
where, at a common node have values expressed as,

Q11 = Q11Cos4θ + 2(Q12 + 2Q66)Sin2θCos2θ + Q22Sin4θ, w1 = w 2 = w 3 = w

θx1 = θx2 = θx3 = θx
Q12 = (Q11Q22−4Q66)Sin2θCos2θ + Q12 (Sin4θ + Cos4θ),
θy1 = θy2 = θy3 = θy (13)
Q22 = Q11Sin4θ + 2(Q12 + 2Q66)Sin2θCos2θ + Q 22Cos4θ,
Q16 = (Q11−Q12−2Q66)SinθCos3θ + (Q12−Q22 + 2Q66)Sin3θCosθ, In-plane displacements of all the three elements at a crack tip are equal
Q26 = (Q11−Q12−2Q66)Sin3θCosθ + (Q12−Q22 + 2Q66)SinθCos3θ, and they are related as
Q66 = (Q11 + Q22−2Q12−2Q66)Sin2θCos2θ + Q66 (Sin4θ + Cos4θ), u2o = u1o−z2o θx
Q44 = Q44Cos2θ + Q55Sin2θ, ν2o = ν1o−z2o θy
Q45 = (Q55−Q44)SinθCosθ,
u3o = u1o−z3o θx
Q55 = Q55Cos2θ + Q44Sin2θ (8) ν3o = ν1o−z3o θy (14)
The in-plane displacements of a plate u and v of any point at a distance
uo1
where is the mid-plane displacement of element 1. Eqs. (13) and (14)
z from the mid-surface are given by
relating the nodal displacements and rotations of elements 1, 2 and 3 at
u (x ,y,z ) = u0 (x ,y )−zθx (x ,y ) the delamination crack tip, are the multipoint constraint equations used
v (x ,y,z ) = v 0 (x ,y )−zθy (x ,y ) in finite element formulation to satisfy the compatibility of displace-
w (x ,y,z ) = w 0 (x ,y ) = w (x ,y ), ments and rotations. Mid-plane strains between elements 2 and 3 are
(9)
related as,
In general, the force and moment resultants of a single lamina are ob-
{ε o} j = {ε o}1 + zoj {k } (where j = 2,3) (15)
tained from stresses as
o
{F } = {Nx Ny Nxy Mx My Mxy Qx Q y}T where {ε } represents the normal strain vector at mid-plane and {k} is
h /2
the curvature vector being identical at the crack tip for elements 1, 2
= ∫ −h /2
{σx σy τxy σx z σy z τxy z τxz τyz }T dz (10) and 3. This equation can be considered as a special case for element 1
when z′1 is equal to zero. In-plane stress-resultants, {N} and moment
In matrix form, the in-plane stress resultant {N}, the moment re-
resultants, {M} of elements 2 and 3 can be expressed as
sultant {M}, and the transverse shear resultants {Q} can be expressed as
{N } j = [A]j {ε o}1 + (zoj [A]j + [B]j ){k } (where j= 2,3)
{N } = [A]{ε 0} + [B]{k }
{M } j = [B]j {ε o}1 + (zoj [B]j + [D]j ){k } (where j= 2,3)
{M } = [B]{ε 0} + [D]{k }
{Q} = [A∗ ]{γ } {Q}j = [A∗ ]j {γ} (where j = 2,3) (16)
(11)
In other form,
2.1. Multipoint constraint – delamination model 0
⎧ Nx ⎫ n=8 ⎧ ⎧ εx ⎫ κ
⎪ zk ⎪ 0⎪ zk ⎧ x⎫
The cross-sectional view of a typical delamination crack tip is shown
Ny =
⎨ ⎬
∑ [Qij]
⎨
∫zk−1
ε y dz +
⎨ 0⎬
∫zk−1
κ y zdz
⎨ κ xy ⎬
N k=1
in Fig. 2 where the nodes of three plate elements meet together to form ⎩ xy ⎭ ⎪ ⎪ γxy ⎪ ⎩ ⎭
⎩ ⎩ ⎭
a common node. The undelaminated region is modelled by plate ele-
ment 1 of thickness h, and the delaminated region is modelled by plate κ ⎫
zk ⎧ x⎫ ⎪
elements 2 and 3 whose interface contains the delamination (h2 and h3 + zj0 ∫z k−1
κ y dz , ...i,j = 1,2,6,
⎨ κ xy ⎬ ⎬
are the thicknesses of the elements 2 and 3, respectively). The elements ⎩ ⎭ ⎪
⎭
0
⎧ Mx ⎫ n=8 ⎧ ⎧ εx ⎫ κ
⎪ zk ⎪ 0⎪ zk ⎧ x⎫ 2
⎨
My =
⎬
∑ [Qij]
⎨
∫z k−1
ε y zdz +
⎨ 0⎬
∫z k−1
κ y z dz
⎨ κ xy ⎬
M k=1
⎩ xy ⎭ ⎪ ⎪ γxy ⎪ ⎩ ⎭
⎩ ⎩ ⎭

κ ⎫
zk ⎧ x⎫ ⎪
+ z j0 ∫z k−1
κ y zdz , ...i,j = 1,2,6,
⎨ κ xy ⎬ ⎬
⎩ ⎭ ⎪
⎭ (17)

n=8 0
⎧Q x ⎫ = ⎧ zk ⎧ γxz ⎫ ⎫
∑ [Qij] ∫ dz , ...i,j = 4,5.
⎩Qy ⎬
⎨ ⎨ z k−1
0⎬
⎨ γyz ⎬
⎭ k=1
⎩ ⎩ ⎭ ⎭
Fig. 2. Plate elements at a delamination crack tip.
Thus in matrix form

556
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

⎡ A11 A12 A16 B11 + z j0 A11 B12 + z j0 A12 B16 + z j0 A16 0 0 ⎤ ε0 2.2.1. Maximum strain theory
⎧ x ⎫
⎧ Nx ⎫ ⎢ ⎥ In the maximum strain criterion, failure of material is assumed to
⎪ Ny ⎪ ⎢ A12 A22 A26 B12 + z j0 A12 B22 + z j0 A22 B26 + z j0 A26 0 0 ⎥ ⎪ ε y0 ⎪
⎪ ⎪ occur if one of the following conditions is satisfied:
⎪N ⎪ ⎢ A A26 A66 B16 + z j0 A16 B26 + z j0 A26 B66 + z j0 A66 0 0 ⎥⎪γ0 ⎪
⎪ xy ⎪ ⎢ 16 ⎥ ⎪ xy ⎪
⎪ Mx ⎪ ⎢ B11 B12 B16 D11 + z j0 B11 D12 + z j0 B12 D16 + z j0 B16 0 0 ⎥ ⎪ κx ⎪ ε1 > XεT , ε2 > YεT , ε3 > ZεT ε4 > Rε , ε5 > Sε, ε6 > Tε (23)
=
⎨ My ⎬ ⎢ ⎥ ⎨κy ⎬
⎪M ⎪ ⎢ B12 B22 B26 D12 + z j0 B12 D22 + z j0 B22 D26 + z j0 B26 0 0 ⎥⎪ ⎪ where ε1, ε2, and ε3 are the normal strain components; ε4, ε5, and ε6 are
⎪ xy ⎪ ⎢ ⎥ κ xy
⎢ B B26 B66 D16 + z j0 B16 D26 + z j0 B26 D66 + z j0 B66 0 0 ⎥⎪γ0 ⎪ the shear strain components; XεT, YεT and ZεT are the lamina normal
⎪Qx ⎪ ⎢ 16 ⎪ xz ⎪
⎪Q y ⎪ ⎢ 0 0 0 0 0 0 S44 S45 ⎥
⎥⎪ 0 ⎪
strain strengths along the 1, 2 and 3 directions, respectively and Rε, Sε
⎩ ⎭ ⎪ γyz ⎪
⎣ 0
⎢ 0 0 0 0 0 S45 S55 ⎥
⎦⎩ ⎭ and Tε are the shear strain strengths in the 2-3, 1-3 and 1-2 planes,
(18) respectively. Normal strain strengths in compression are compared with
ε1, ε2, and ε3 when they are of compressive nature.
where [A], [B] and [D] are the extension, bending-extension coupling
and bending stiffness coefficients of the composite laminate, respec-
2.2.2. Maximum stress theory
tively and for element 1 can be written as
Failure of material is assumed to occur if any of the following
zA h /2 conditions is satisfied:
([A],[B],[D]) = ∫−h/2 [Q ](1,z,z2) dz + ∫z A
[Q ](1,z ,z 2) dz
(19)
σ1 > XT , σ2 > YT , σ3 > ZT σ4 > R, σ5 > S, σ6 > T (24)
where [Q ] is the transformed reduced stiffness as defined in [70] while
where σ1, σ2, σ3 are the normal stress components, σ4,σ5,σ6 are the shear
zto is the z-co-ordinate of mid-plane of t-th sublaminate. Thus the for-
stress components; XT, YT, ZT are the lamina normal strengths along the
mulation based on the multi-point constraint conditions leads to un-
1, 2 and 3 directions, respectively and R, S and T are the shear strain
symmetric stiffness matrix. The resultant forces and moments at the
strengths in the 2-3, 1-3 and l-2 planes, respectively. When σ1, σ2, σ3 are
delamination front for the elements 1, 2 and 3 satisfy the following
of compressive nature, they should be compared with XC, YC, and ZC,
equilibrium conditions,
which are normal strengths in compression along the 1, 2 and 3direc-
{N } = {N }1 = {N } 2 + {N }3 tions, respectively. This is regarded as an independent mode criterion.
{M } = {M }1 = {M }2 + {M }3 + z2′ {N }2 + z3′ {N }3 The polynomial type maximum stress criterion is expressed as,
{Q} = {Q}1 = {Q} 2 + {Q}3 (20) (σ1−XT )(σ1 + XT )(σ2−YT )(σ2 + YT )(σ3−ZT )(σ3 + ZT )
where {Q} denotes the transverse shear resultants. (σ4−R)(σ4 + R)(σ5−S )(σ5 + S )(σ6−T )(σ6 + T ) = 0
(25)
2.2. Governing failure theories
Comparing Eq. (4) with Eq. (1) and ignoring higher-order terms in
both the equations, it can be expressed as,
In the present study, the cantilever composite plate is considered
with typically thin layers. The material of each lamina consists of
parallel, continuous fibers embedded in a matrix material. Different
F1 = ( 1
−
XT XC
1
), F2 = ( 1
−
YT YC
1
), F3 = ( 1
−
ZT ZC
1
),
1 1 1
quadratic polynomial criteria differ in the way polynomial constants F11 = X X , F22 = Y Y , F33 = Z Z ,
T C T C T C
are determined. The most general criterion for composite materials is 1 1 1
F44 = 2 , F55 = 2 , F66 = 2 ,
the tensor polynomial criterion proposed by Tsai and Tsai and Wu [71]. R S T
F F F F F F
The failure criteria used are macroscopic in nature and are incorporated F12 = − 12 2 , F13 = − 12 3 , F23 = − 22 3 (26)
into the finite element computational procedure. All the failure theories
are considered for the tensor polynomial failure criterion [72]. The remaining strength constants are zero.

F1 σ1 + F2 σ2 + F3 σ3 + 2F12 σ1 σ2 + 2F13 σ1 σ3 + 2F23 σ2 σ3 + F11 σ12 + F22 σ22 2.2.3. Energy based interaction (Tsai-Hill) theory
+ F33 σ32 + F44 σ42 + F55 σ52 + F66 σ62 + ⋯ ⩾ 1 (21) Tsai ill criterion can be expressed as

The failure criteria require the stresses and strains computed with re- σ 2 σ 2 σ 2 1 1 1 1 1 1
⎛ 1 ⎞ + ⎛ 2 ⎞ + ⎛ 3 ⎞ −⎛ + − ⎞ σ1 σ2−⎛ 2 + 2 − 2 ⎞ σ2 σ3
spect to the material axes of each lamina. In general, the laminate axes ⎝X⎠ ⎝Y ⎠ ⎝ Z ⎠ ⎝ X2 Y 2 Z2 ⎠ ⎝Y Z X ⎠
do not coincide with the material principal axes of the individual plies. 1 1 1 σ 2 σ 2 σ 2
−⎛ 2 + 2 − 2 ⎞ σ1 σ3 + ⎛ 4⎞ + ⎛ 5⎞ + ⎛ 6⎞ ⩾ 1
Therefore, a transformation of the stresses and strains from laminate ⎝Z X Y ⎠ ⎝R⎠ ⎝S⎠ ⎝T ⎠ (27)
coordinates to the lamina material coordinates is performed. If θm de-
The F1, F2 and F3 terms are zero as the stress terms do not appear as
notes the lamination angle of the m-th layer, then the stresses σi(m) in its
linear terms. The values of X, Y and Z are taken as either XT, YT, ZT or as
material coordinates can be obtained from the stresses σxx, σyy, τxy etc.
XC, YC, ZC depending upon the sign of σ1, σ2 and σ3, respectively.
in laminate coordinates using the following transformation:

σ1(m) = σxx cos2 θm + σyy sin2 θm + 2τxy sinθm cosθm 2.2.4. Interaction tensor polynomial (Tsai-Wu) theory
σ2(m) = σxx sin2 θm + σyy cos2 θm−2τxy sinθmcosθm Based on the generalization of the von Mises criterion, Tsai-Wu
criterion is expressed as
σ3(m) = σzz
Fi σi + Fij σi σj ⩾ 1 (28)
σ4(m) = −τxz sinθm + τyz cosθm
σ5(m) = τxz cosθm + τyz sinθm where

σ6(m) = −σxx sinθmcosθm + σyy sinθmcosθm + τxy (cos2 θm−sin2 θm ) (22) F1 =

1 1 1 1 1
− , F1 = Y − Y , F1 = Z − Z
1
XT XC T C T C
1 1 1
Similarly, εi(m) in its material coordinates can be obtained from εxx, F11 = X X ,F22 = Y Y , F33 = Z Z ,
T C T C T C
εyy, εzz etc., for the mth layer. The lamina (i.e., material coordinate) 1 1 1
F44 = 2 , F55 = 2 , F66 = 2 ,
stresses or strains are then used in a chosen failure criterion to de- R S T
1 1 1
termine failure of each finite element and each lamina. The load that F12 = − 2 X X Y Y , F13 = − 2 X X Z Z , F23 = − 2 Y Y Z Z
T C T C T C T C T C T C (29)
causes ply failure is calculated based on commonly used strength-of-
material type failure criteria as described below. All other strength tensor components are zero.

557
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

2.2.5. Hoffman’s criterion

Hoffman’s criterion is expressed as

1⎛ 1 1 1 ⎞ 1 1 1 1 ⎞
⎜ + − (σ2−σ3)2 + ⎛ ⎟+ − (σ3−σ1)2
⎜ ⎟

2 ⎝ YT YC ZT ZC XT XC ⎠ 2 ⎝ ZT ZC XT XC YT YC ⎠
1 1 1 1 ⎞ 1 1 1 1
+ ⎛ +
⎜ − (σ1−σ2)2 + ⎛ − ⎞ σ1 + ⎛ − ⎞ σ2
⎟ ⎜ ⎟ ⎜ ⎟

2 ⎝ XT XC YT YC ZT ZC ⎠ ⎝ XT XC ⎠ ⎝ YT YC ⎠
1 1 σ 2 σ 2 σ 2
+ ⎛ − ⎞ σ3 + ⎛ 4 ⎞ + ⎛ 5 ⎞ + ⎛ 6 ⎞ ⩾ 1
⎜ ⎟

⎝ ZT ZC ⎠ ⎝R⎠ ⎝S⎠ ⎝T ⎠ (30)

In the present investigation, results are presented based on the
above mentioned five failure criteria and a comparative assessment is
carried out in the context to failure strength of composite laminates.

2.3. Finite element formulation

The present finite element formulation is based on a composite plate

element. An eight noded isoparametric quadratic plate bending element Fig. 3. An eight nodded isoparametric quadratic plate bending element with numbering
with five degrees of freedom at each node (three translation and two scheme of the nodes and its mapping in XY space.

rotations) is considered. The shape functions or the interpolation

functions are polynomials of the natural co-ordinates (ξ ,η,ζ ), which
relate the generalized displacements at any point within an element to x ,ξ y,ξ
where [J ] = ⎡ x y ⎤ is the Jacobian matrix.
the nodal values of the displacements. These are derived from an in- ⎣ ,η ,η ⎥
⎢ ⎦
terpolation polynomial in terms of the natural co-ordinates so that the The generalized inertia matrix per unit area includes the translatory
displacement fields are satisfactorily represented. For the analysis and rotatory inertia terms. The mass and moment of inertia are re-
where the final element is assumed to have mid-surface nodes only, the spectively the measure of translatory and rotatory inertial resistances
interpolation polynomial is a function of ξ and η and has the following and are given by the following equations respectively. Mass per unit
form area is denoted by P and is given by
n=8
u (ξ , η) = C0 + C1 ξ + C2 η + C3 ξ 2 + C4 ξη + C5 η2 + C6 ξ 2η + C7 ξη2 zk

(31)
P= ∑ ∫zk−1 ρdz
k=1 (37)
where, C0,C1,C2…C7 are the generalized degree of freedom.
where ρ is the mass density of the laminate. Moment of inertia per unit
The shape functions derived from interpolation polynomial are as
area is denoted by I1 and is given by
[73]
n=8
zk
1
Ni = 4 (1 + ξξi )(1 + ηηi )(ξξi + ηηi −1) ⋯i = 1,2,3,4 I1 = ∑ ∫zk−1 zρdz.
1 k=1 (38)
Ni = 2
(1 + ηηi )(1−ξ 2) ⋯i = 5,7.
1 Incorporating both the translatory and rotatory inertia terms, the
Ni = 2
(1 + ξξi )(1−η2) ⋯i = 6,8. (32) element mass matrix takes the following form
where ξη are the local natural coordinates of the element. Here ξi = +1 [Me] = ∫Φ [N ]T [P][N ] dΦ
for nodes 2, 3, 6, ξi = −1 for nodes 1, 4, 8 and ηi = +1 for nodes 3, 4, 7 (39)
and ηi = −1 for nodes 1, 2, 5 (Fig. 3). The correctness of the shape where,
functions is checked from the relations
⎡ {Ni} ⎤
8 8
∂Ni
8
∂Ni ⎢ {Ni} ⎥ ⎡P 0 0 0 0⎤
∑ Ni = 1 , ∑ ∂ξ
= 0 and ∑ ∂η
=0 ⎢ ⎥ ⎢0 P 0 0 0⎥
i=1 i=1 i=1 (33) [N ] = ⎢ {Ni} ⎥ and [P ] = ⎢ 0 0 P 0 0 ⎥.
⎢ {Ni} ⎥ ⎢0 0 0 I 0⎥
In an isoparametric plate bending formulation, the generalized ⎢ ⎥ ⎢0 0 0 0 I⎥
{ N } ⎣ ⎦ (40)
displacements and co-ordinates are interpolated from their nodal values ⎣ i ⎦

by the same set of shape functions. Hence, the co-ordinates (x, y) of any
point within an eight-noded element are obtained as
2.4. Stochastic failure analysis of composite laminates
8 8
x= ∑ Ni x i and y = ∑ Ni yi The flowchart for carrying out stocahstic analysis of failure loads
i=1 i=1 (34)
considering stochasticity in structural and material attributes (such as
The relations between the displacement at any point with respect to longitudinal elastic modulus, transverse elastic modulus, longitudinal
the coordinates (ξ, η) and nodal degrees of freedom are expressed as shear modulus, transverse shear modulus, Poisson’s ratio, mass density,
8 8 8 8 ply orientation angle) is elucidated in Fig. 4. Monte Carlo simulation
u= ∑ Ni ui, v = ∑ Ni vi, w = ∑ Ni wi, θx = ∑ Ni θxi, θy technique [56] is employed in conjunction with the finite element
i=1 i=1 i=1 i=1 model for failure strength analysis of composite laminates. The com-
8
pound effect { g (ω ) } of the variation of all the aforementioned sto-
= ∑ Ni θyi
chastic paramenters are considered in the analysis as follows:
i=1 (35)
g {θ (ω ),ρ (ω ),G12 (ω ),G23 (ω ),E1 (ω )} = {Φ1 (θ1 ⋯θl ),Φ2 (E1(1) ⋯E1(l) ),Φ3 (E2(1) ⋯E2(l) )
and
Φ4 (G12(1) ⋯G12(l) ),Φ5 (G23(1) ⋯G23(l) ),Φ6 (μ1 ⋯μl ),Φ7 (ρ1. .ρl )}

⎡ Ni,x ⎤ = [J ]−1 ⎡ Ni,ξ ⎤ (41)

⎢ Ni,y ⎥ ⎢ Ni,η ⎥ (36)
⎣ ⎦ ⎣ ⎦ where θi, E1(i), E2(i), G12(i), G23(i), μ(i) and ρ(i) are the ply orientation

558
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

Fig. 4. Flow diagram depicting the Monte Carlo simulation

based stochastic analysis of failure load for laminated
composite plate.

angle, elastic modulus along longitudinal and transverse direction, in Table 1, wherein the results for first-ply failure load of three layered
shear modulus along longitudinal direction, shear modulus along angle-ply composite plates with mesh sizes (2 × 2), (4 × 4) and (8 × 8)
transverse direction, Poisson’s ratio and mass density, respectively and are presented and the results are compared with Reddy and Pandey [3].
‘l’ denotes the number of layer in the laminate. ω represents the sto- Further results for validation of the finite element code are provided in
chastic character of the input parameters. Tables 2 and 3 with Kam et al. [74]. The failure loads obtained using
the present finite element code with 8 × 8 mesh and the results pre-
sented by Reddy and Pandey [3] and Kam et al. [74] are found to be in
3. Results and discussion
good agreement corroborating the validity of the finite element model.

3.1. Description of laminate configuration and validation

3.2. Parametric study following deterministic approach
In this section, the results obtained for the failure load of delami-
nated and undelaminated composites following different failure criteria Parameteric studies are conducted based on different influential
are presented in deterministic as well as stochastic regime. The five material and geometric factors of composite laminates. Fig. 5 shows the
different failure criteria considered in this study are: maximum stain, effect of ply thickness on failure load considering five different failure
maximum stress, Tsai-Hill, Tsai-Wu-Hahn and Tsai-Hill-Hoffman. criteria. It can be observed from the figure that as thickness increases,
Parametric studies are carried out to investigate the effect on first-ply the load for the failure also increases. According to the maximum strain
failure loads with respect to ply orientation angle, number of ply, de- theory, the slope of the curve is comparatively steeper than the other
gree of orthotropy and thickness. Three layered graphite-epoxy angle- theories, followed by maximum stress criterion. Tsai–Hill theory,
ply ([45°/−45°/45°]) laminated composite cantilever plate is con- Tsai–Wu–Hahn theory and Tsai–Hill–Hoffman theory obtain almost
sidered to present the results, unless otherwise mentioned. The length, similar variation. Fig. 6 shows the variation of failure load with respect
width and thickness of the composite laminate considered in the pre- to degree of orthotropy, wherein it can be observed that when degree of
sent analysis are 1 m, 1 m and 5 mm, respectively. Deterministic mean orthotropy increases, failure load also increases. In this result also, in
value of material properties for graphite–epoxy composite are con- maximum strain theory for the same degree of orthotropy, the failure
sidered as E1 = 138.0 GPa, E2 = 8.96 GPa, G12 = 7.1 GPa, load is more. The maximum strain theory obtains the failure loads with
G13 = 7.1 GPa, G23 = 2.84 GPa, μ = 0.3, ρ = 3202 kg/m3 [58]. A dis- higher values, compared to the other failure critaria, followed by
cretization of (8 × 8) mesh on plan area with 64 elements and 225 maximum stress theory and Tsai-Hill theory. The failure load obtained
nodes with natural coordinates of an isoparametric quadratic plate by Tsai–Wu–Hahn theory and Tsai–Hill–hoffman theory are same and
bending element is adopted for the present finite element model. The they are the lower than the other failure criteria. However, no noticable
finite element mesh size is finalized using convegence studies as shown difference in the slope of the curves corresponding to different failure

559
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

Table 1
Validation of results for first-ply failure load for in-plane loading according to different failure criteria considering a [45°/−45°/45°] laminate.

Failure theory Failure load (N)

(2× 2) (4× 4) (8× 8)

Reddy and Pandey [3] Present FEM Reddy and Pandey [3] Present FEM Reddy and Pandey [3] Present FEM

Max. stress 2854.4 3408.65 2164.32 2486.54 1908.16 1962.24

Max. strain 2947.68 3273.19 2268.64 2421.34 1940.48 1994.73
Tsai-Hill 2788.8 3091.37 1803.84 1897.86 1530.4 1563.62
Tsai-Wu 2886.72 3337.68 2218.88 2342.64 1917.76 1957.43
Hoffman 2850.24 3224.46 2156.8 2269.47 1905.76 1961.94

Table 2
First-ply failure loads for a partially clamped (0°/90°)s plate subjected to a concentrated
transverse load at the center.

Failure theory Failure load (N)

Kam et al. [74] Present FEM

Max stress 64.94 64.694

Max strain 76.04 75.071
Tsai Hill 64.03 64.694
Tsai-Wu 68.30 66.029
Hoffman 63.60 64.574

Table 3
First-ply failure loads for a partially clamped (0°2/90°)s plate subjected to a concentrated
transverse load at the center.

Failure theory Failure load (N)

Kam et al. [74] Present FEM

Max stress 108.26 109.095 Fig. 5. Effect of ply thickness on failure load.
Max strain 122.86 120.642
Tsai Hill 107.06 109.095
Tsai-Wu 112.77 110.639
Hoffman 106.45 108.899

criteria is observed in case of variation in degree of orthotropy.

Fig. 7 presents the effect of variation in number of plies on failure
load for angle ply and cross ply laminates. For angle ply and cross ply
laminates it is observed that maximum strain theory obtains highest
failure load, while Tai–Hill–Hoffman theory obtains the least value of
failure load. However, the difference in the failure loads predicted by
maximum stress theory and maximum strain theory is found to be
minimal in case of cross ply laminates, the prediction of other failure
criteria being lesser than the aforementioned two criteria. Fig. 8 shows
the effect of variation in ply orientation angle on failure load con-
sidering a [θ°/−θ°/θ°] family of composite laminates, from which it is
observed that the variation in failure load is more in maximum strain
theory. Variations in failure load obtained following Tsai–Wu–Hahn Fig. 6. Effect of degree of orthotropy on failure load.
and Tai–Hill–Hoffman criteria are minimum. From the figures a critical
value of ply orientation angle can be identified (between 45° and 60°), presented in Fig. 9, while the effect of delamination is portrayed in
for which the predicted failure load is found to be maximum. Such Fig. 10 by comparing results for both delaminated and undelaminated
observations along with the parametric results presented in the pre- angle ply composite laminates. In the present analysis, a laminated
ceeding paragraphs cater to a clear understanding about the failure load composite plate having a square plan-form (L/b = 1) with 25% dela-
of composite laminates from a designer’s perspective. mination at mid-plane corresponding to relative location of delamina-
tion distance, d/L = 0.33, 0.5 and 0.66. A clear distinction in the pat-
3.3. Locational vulnerability analysis of failure load tern of failure loads can be identified for angle ply and cross ply
laminates. The spatially varying values of failure load are found to be
Based on the five failure criteria, the effect of spatial variation in significantly affected by the effect of delamination, as shown in Fig. 10.
loading position is investigated to analyze the relative vulnerability in For a cantilever plate, as considered in this investigation, the central
failure strength of composites. Figs. 9 and 10 show the effect of spatial strip perpendicular to the fixed end and the edge parallel to the fixed
variation of loading position on the failure loads considering different end show least value of failure load (i.e. more susceptible to failure),
failure criteria. A comparative assessment of the locational sensitivity of while the zones near the fixed end show higher value of failure load.
loading positions between angle ply and cross ply laminates is Such plots provide a clear perspective about the vulnerability of the

560
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

Fig. 7. Effect of number of plies on failure load for (a)

angle-ply and (b) cross-ply laminate (n – number of layers,
FC – 1: Maximum stain, FC – 2: Maximum stress, FC – 3:
Tsai-Hill, FC – 4: Tsai-Wu-Hahn, FC – 5: Tsai-Hill-Hoffman).

criteria. The stochastic bounds shown in Fig. 11(a)–(e) for different

loading positions indicate that the effect of stochasticity has significant
influence on the failure loads and subsequently, such effects should be
accounted in the analysis and design of composite laminates. To char-
acterize the relative sensitivity of stochastic behavior for the three
loading positions, the respective coefficient of variations are plotted in
Fig. 11(g), wherein it can be discerned that Point-1 is least sensitive to
the effect of stochasticity, followed by Point-3 and Point-2 respectively.
In turn, the Fig. 11(g) also affirms that the sensitivity of stochasticity for
the failure loads at different points varies spatially and this aspect needs
further attention in the process of analyzing composite laminates.
To access the relative effect of stochasticity in the individual ma-
terial and geometric parameters, a separate study is carried out for the
loading position 2. Monte Carlo simulations are performed for the in-
dividual stochasticity of different input parameters. The coefficient of
Fig. 8. Effect of ply orientation angle on failure load on a [θ°/−θ°/θ°] family of com- variations (COV) in the failure loads following different failure criteria
posite laminates. are presented in Fig. 12, which provides a measure of sensitivity for
stochasticity in the input parameters. From the figure, it can be ob-
location of applied load in the analysis of composite laminates. served that the structural attributes such as thickness and ply orienta-
tion angle are most sensitive to the failure loads, followed by the ma-
terial properties E1, E2, G12 and G13. Sensitivity of G23, Poisson’s ratio
3.4. Stochastic failure analysis
and mass density are found to be negligible. The probabilistic de-
scriptions of failure loads following different failure criteria for in-
In the present study, the spatial location of point 1 (nearer to fixed
dividual stochasticity in the most sensitive parameters (as per Fig. 12)
end of cantilever plate), point 2 (mid-point) and point 3 (towards free
are presented in Fig. 13 (considering loading position 2). Such analysis
end of cantilever plate) are considered as shown in Fig. 11(f) to eluci-
considering individual effect of stochasticity can provide a compre-
date the effect of stochasticity in failure load. The effect of inevitable
hensive understanding about the extent of relative control required for
random variability in the structural and material attributes of compo-
different parameters in practical situations.
site laminates is accounted by considering 10% stochastic bound (in-
Future research can be extended to address the aspect of computa-
dustry standard) in the input parameters with respect to their respective
tional efficiency for the stochastic analysis of the failure strength of
mean values. The results are presented in Fig. 11 considering different
composites. In general for complex structural forms like laminated
failure criteria. The probabilistic description of failure loads are ob-
composites, the performance functions for failure strengths are gen-
tained for three different representative loading positions. The prob-
erally not available as an explicit analytical model of the input para-
ability density function plots vary considerably for the three different
meters. If explicit analytical expressions are available for a stochastic
loading positions and they also differ based on the adopted failure

561
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

Fig. 9. Locational sensitivity of applied load for

angle ply ([45°/−45°/45°]) and cross ply ([0°/
−90°/0°]) composite laminates considering dif-
ferent failure criteria (FC – 1: Maximum stain, FC
– 2: Maximum stress, FC – 3: Tsai-Hill, FC – 4:
Tsai-Wu-Hahn, FC – 5: Tsai-Hill-Hoffman).

system [75–80], carrying out a Monte Carlo simulation for the sto- which is a time-consuming and computationally expensive approach. A
chastic responses could be computationally viable. However, the pre- direct Monte Carlo simulation based uncertainty quantification ap-
sent analysis of stochastic failure strength for composite laminates can proach, as followed in this article, requires thousands of expensive fi-
only be performed numerically such as the finite element method, nite element analyses, which becomes a major setback in carrying out

562
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

Fig. 10. Locational sensitivity of applied load for undelaminated and delaminated composite laminates with angle ply ([45°/−45°/45°]) configuration considering different failure
criteria (FC – 1: Maximum stain, FC – 2: Maximum stress, FC – 3: Tsai-Hill, FC – 4: Tsai-Wu-Hahn, FC – 5: Tsai-Hill-Hoffman).

563
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

Fig. 11. (a–e) Probabilistic description of failure loads following different failure criteria considering an angle ply laminate (45°/−45°/45°) for three different loading positions (as shown
in Fig. 10(f)). The corresponding deterministic values are shown in subset. (g) Coefficient of variation for three different loading positions corresponding to the five considered failure
criteria (FC – 1: Maximum stain, FC – 2: Maximum stress, FC – 3: Tsai-Hill, FC – 4: Tsai-Wu-Hahn, FC – 5: Tsai-Hill-Hoffman).

564
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

Fig. 12. COV (Coefficient of variation) of the failure loads

for the stochasticity in individual parameters (computed for
different failure criteria considering an angle ply laminate
(45°/−45°/45°) with the loading position 2, as shown in
Fig. 11(f)).

Fig. 13. (a) Probabilistic description of failure loads for individual stochasticity in only E1 following different failure criteria considering an angle ply laminate (45°/−45°/45°) for
loading position 2 (refer to Fig. 11(f)) (b) Probabilistic description of failure loads for individual stochasticity in only E2 following different failure criteria considering an angle ply
laminate (45°/−45°/45°) for loading position 2 (refer to Fig. 11(f)) (c) Probabilistic description of failure loads for individual stochasticity in only G12 following different failure criteria
considering an angle ply laminate (45°/−45°/45°) for loading position 2 (refer to Fig. 11(f)). (d) Probabilistic description of failure loads for individual stochasticity in only ply
orientation angle following different failure criteria considering an angle ply laminate (45°/−45°/45°) for loading position 2 (refer to Fig. 11(f)). (e) Probabilistic description of failure
loads for individual stochasticity in only thickness following different failure criteria considering an angle ply laminate (45°/−45°/45°) for loading position 2 (refer to Fig. 11(f)).

detail stochastic analyses for the failure strength of composites any structures, the present study addressing the practical aspects of loca-
further. This lacuna can be effectively mitigated by following a surro- tional sensitivity of failure loads, delamination and stochasticity will
gate based approach [81–88,60,89], where the objective is to carry out bring about a comprehensive perspective in the failure analysis of such
the Monte Carlo simulation by conducting minimum number of finite structures.
element simulations. Some crucial definitive inferences can be drawn from the numerical
results, which are clearly indicated in this paper. Maximum strain cri-
4. Conclusion terion is found to overestimate the failure loads in most of the cases.
The plots presenting deterministic failure strengths for spatially varying
The effect of spatially varying loading position on the failure loading positions can be useful to identify the critical contours for ap-
strength of laminated composites is investigated following determi- plication of point load. This will render an idea about the relative
nistic as well as stochastic analyses. The crucial aspect of delamination vulnerability of failure based on the locational sensitivity of the applied
in composites is accounted in the investigation to quantify its effect on load. The stochastic analysis further quantifies the variability in failure
the failure strengths following different criteria (such as maximum strengths due to inherent uncertainty in the system parameters.
stain, maximum stress, Tsai-Hill, Tsai-Wu-Hahn and Tsai-Hill- Consideration of defects in composites (such as delamination), which
Hoffman). Novelty of the present study includes incorporation of the commonly occurs during the service period, simulates the practical
aspect of spatially varying loading position as well as the effect of conditions quite closely in the present failure analysis. The probabilistic
stochasticity in structural and material attributes in context to dela- plots and statistical results presented on the basis of stochastic analysis
minated composite plates. The results obtained are validated with indicate that the probabilistic bound of failure strength varies with
previous benchmarking literature. Both angle ply and cross ply lami- different loading position and the considered failure criteria. This ob-
nated composite plates are considered to investigate the failure strength servation leads to the realization for the need of further detailed ana-
of composite laminates along with the effect of variation in stacking lysis to quantify the probabilistic variability for spatially varying
sequence, ply orientation, number of layers, degree of orthotropy and loading positions and subsequent quantification of their sensitivities.
ply thickness. Individual as well as compound effect of stochasticity in The present article will serve as a valuable reference for such future
the material and structural attributes are analyzed for the failure loads investigations.
considering different failure criteria. As composite materials are being
increasingly used in various fields of Mechanical, Civil and Aerospace

565
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

Acknowledgement surface-bounded sensors and internally pressurized. Compos Part B: Eng

2015;77:502–18.
[34] Ghosh A, Chakravorty D. Prediction of progressive failure behaviour of composite
PKK would like to acknowledge the financial support received from skewed Hypar shells using finite element method. J Struct 2014;2014(2014):8.
MHRD, India during the period of this research work. [35] Moreno MCS, Gutierrez AR, Vicente JLM. First flexural and interlaminar shear
failure in symmetric cross-ply carbon-fibre laminates with different response under
tension and compression. Compos Struct 2016;146:62–8.
References [36] Patel S, Guedes Soares C. System probability of failure and sensitivity analyses of
composite plates under low velocity impact. Compos Struct 2017. http://dx.doi.
org/10.1016/j.compstruct.2017.08.054.
[1] Talreja R. Assessment of the fundamentals of failure theories for composite mate-
[37] Romanowicz M. Determination of the first ply failure load for a cross ply laminate
rials. Compos Sci Technol 2014;105:190–201.
subjected to uniaxial tension through computational micromechanics. Int J Solids
[2] Dvorak George J, Norman Laws. Analysis of first ply failure in composite laminates.
Struct 2014;51:2549–56.
Eng Fract Mech 1986;25(5–6):763–70.
[38] Rehan MSBM, Rousseau J, Fontaine S, Gong XJ. Experimental study of the influence
[3] Reddy JN, Pandey AK. A first-ply failure analysis of composite laminates,
of ply orientation on DCB mode-I delamination behavior by using multidirectional
Department of Engineering Science and Mechanics, Virginia Polytechnic Institute
fully isotropic carbon/ epoxy laminates. Compos Struct 2017;161:1–7.
and State University, Blacksburg, VA 24061, U.S.A. 1987; 25(3): 371–393.
[39] Dimitri R, Fantuzzi N, Li Y, Tornabene F. Numerical computation of the crack de-
[4] Dvorak George J, Norman Laws. Analysis of progressive matrix cracking in com-
velopment and SIF in composite materials with XFEM and SFEM. Compos Struct
posite laminates II. First ply failure. J Compos Mater 1987;21(4):309–29.
2017;160:468–90.
[5] Pandey A. A post first-ply failure analysis of composite laminates, American
[40] Abdullah NA, Curiel-Sosa JL, Taylor ZA, Tafazzolimoghaddam B, Vicente JLM,
Institute of Aeronautics and astronautics; 1987.
Zhang C. Transversal crack and delamination of laminates using XFEM. Compos
[6] Bruno D, Spadea G, Zinno R. First-ply failure of laminated composite plates. Theoret
Struct 2017;173:78–85.
Appl Fract Mech 1993;19(1):29–48.
[41] Kumar A, Chakrabarti A, Bhargava P. Efficient failure analysis of laminated com-
[7] Reddy YSN, DakshinaMoorthy CM, Reddy JN. Non-linear progressive failure ana-
posites and sandwich cylindrical shells based on higher order zigzag theory. J
lysis of laminated composite plates. Int J Non-Linear Mech 1995;30(5):629–49.
Aerosp Eng, http://dx.doi.org/10.1061/(ASCE)AS.1943-5525.0000433.
[8] Kam TY, Sher HF. Nonlinear and first-ply failure analyses of laminated composite
[42] Kumar A, Chakrabarti A, Bhargava P, Chaudhary R. Probabilistic failure analysis of
cross-ply plates. J Compos Mater 1995;29(4):463–82.
laminated sandwich shells based on higher order zigzag theory. J Sandwich Struct
[9] Kam TY, Jan TB. First-ply failure analysis of laminated composite plates based on
Mater 2015;17(5):546–61.
the layerwise linear displacement theory. Compos Struct 1995;32(1–4):583–91.
[43] Naskar S, Mukhopadhyay T, Sriramula S, Adhikari S. Stochastic natural frequency
[10] Kam TY, Sher HF, Chao TN, Chang RR. Predictions of deflection and first-ply fai-
analysis of damaged thin-walled laminated composite beams with uncertainty in
lure load of thin laminated composite plates via the finite element approach. Int J
micromechanical properties. Compos Struct 2017;160:312–34.
Solids Struct 1996;33(3):375–98.
[44] Li Y, Cao S, Dimitri R, Fantuzzi N, Tornabene F. Analytical and numerical in-
[11] Kam TY, Chang ES. Reliability formulation for composite laminates subjected to
vestigation of the stiffness matrix for edge-cracked circular shafts. Fatigue Fract Eng
first-ply failure. Compos Struct 1997;38(1–4):447–52.
Mater Struct. http://dx.doi.org/10.1111/ffe.12503.
[12] Kam TY, Liu YW, Lee FT. First-ply failure strength of laminated composite pressure
[45] Yao Liaojun, Sun Yi, Guo Licheng, Zhao Meiying, Jia Liyong, Alderliesten RC, et al.
vessels. Compos Struct 1997;38(1–4):65–70.
A modified Paris relation for fatigue delamination with fibre bridging in composite
[13] Sleight David W, Knight, Jr. Norman F, Wang John T. Evaluation of a progressive
laminates. Compos Struct 2017;176:556–64.
failure analysis methodology for laminated composite structures. In: 38th struc-
[46] Jang Byeo-Wook, Kim Ch-Gon. Real-time detection of low-velocity impact-induced
tures, structural dynamics, and materials conference, NASA Langley Research
delamination onset in composite laminates for efficient management of structural
Center; 1997.
health. Compos B 2017;123:124–35.
[14] Lin SC, Kam TY, Chu KH. Evaluation of buckling and first-ply failure probabilities
[47] Yelve Nitesh P, Mitra Mira, Mujumdar PM. Detection of delamination in composite
of composite laminates. Int J Solids Struct 1998;35(13):1395–410.
laminates using Lamb wave based nonlinear method. Compos Struct
[15] Liu K-Shih, Tsai Stephen W. A progressive quadratic failure criterion for a laminate.
2017;159:257–66.
Compos Sci Technol 1998;58(7):1023–32.
[48] Carpinteri Andrea, Fernández-Canteli Alfonso, Fortese Giovanni, Muñiz-Calvente
[16] Padhi GS, Shenoi RA, Moy SSJ, Hawkins GL. Progressive failure and ultimate col-
Miguel, Ronchei Camilla, Scorza Daniela, et al. Probabilistic failure assessment of
lapse of laminated composite plates in bending. Compos Struct
Fibreglass composites. Compos Struct 2017;160:1163–70.
1998;40(3–4):277–91.
[49] Chen Ji-Fen, Morozov Evgeny V, Shankar Krishnakumar. Simulating progressive
[17] Kam TY, Lai FM. Experimental and theoretical predictions of first-ply failure
failure of composite laminates including in-ply and delamination damage effects.
strength of laminated composite plates. Int J Solids Struct 1999;36(16):2379–95.
Compos: Part A 2014;61:185–200.
[18] Lin S, Kam T. Probabilistic failure analysis of transversely loaded laminated com-
[50] Rivera Juan A, Aguilar Enrique, Cárdenas Diego, Elizalde Hugo, Probst Oliver.
posite plates using first-order second moment method. J Eng Mech
Progressive failure analysis for thin-walled composite beams under fatigue loads.
2000;126(812):812–20.
Compos Struct 2016;154:79–91.
[19] Chang RR. Experimental and theoretical analyses of first-ply failure of lamina-
[51] Luo Haibo, Yan Ying, Zhang Taihua, He Zeqing, Wang Sheng. Progressive failure
ted composite pressure vessels. Compos Struct 2000;49(2):237–43.
numerical simulation and experimental of carbon-fiber composite corrugated
[20] Pal P, Ray C. Progressive failure analysis of laminated composite plates by finite
beams under dynamic impact. Polym Test 2017;63:12–24.
element method. J Reinf Plast Compos 2002;21(16):1505–13.
[52] Pan Zhongxiang, Bohong Gu, Sun Baozhong, Xiong Jie. Progressive failure of 3-D
[21] Satish Kumar YV, Anand Srivastava. First ply failure analysis of laminated stiffen-
textile composites under impact loadings. Compos Struct 2017;168:710–24.
ed plates. Compos Struct 2003;60(3):307–15.
[53] Zhao Chao, Huang Yuanchen, Chen Zhong, Kyu Ha Sung. Progressive failure pre-
[22] Rattanawangcharoen N. First-ply failure analysis of laminated composite cylindrical
diction of a landing gear structure of braided composites. Compos Struct
panels. J Reinf Plast Compos 2005;24(14):1521–37.
2017;161:407–18.
[23] Cheng-Hsien Chung, LeeYa-Jung. Progressive failure of marine GFRP laminated
[54] Chen BY, Tay TE, Pinho ST, Tan VBC. Modelling the tensile failure of composites
plates under static water pressure. J Compos Mater 2005;39(12):1081–102.
with the floating node method. Comput Methods Appl Mech Eng 2016;308:414–42.
[24] Hochard Ch, Payan J, Bordreui C. A progressive first ply failure model for
[55] Tawqir Chowdhury Nayeem, Wang John, Chiu Wing Kong, Yan Wenyi. Predicting
woven ply CFRP laminates under static and fatigue loads. Int J Fatigue
matrix failure in composite structures using a hybrid failure criterion. Compos
2006;28(10):1270–6.
Struct 2016;137:148–58.
[25] Daxu Z, Jianqiao Y, Dennis L. Ply cracking and stiffness degradation in cross-ply -
[56] Dey S, Mukhopadhyay T, Spickenheuer A, Adhikari S, Heinrich G. Bottom up sur-
laminates under biaxial extension, bending and thermal loading. Compos Struct
rogate based approach for stochastic frequency response analysis of laminated
2006;75(1–4):121–31.
composite plates. Compos Struct 2016;140:712–27.
[26] Hisashi K, Hirotaka H, Ippei S, Takao U. Damage and permeability evolution in
[57] Dey S, Mukhopadhyay T, Naskar S, Dey TK, Chalak HD, Adhikari S. Probabilistic
CFRP cross-ply laminates. Compos Struct 2006;76(1–2):73–81.
characterization for dynamics and stability of laminated soft core sandwich plates. J
[27] Lopes CS, Gürdal Z, Camanho PP. Variable-stiffness composite panels: buckling
Sandw Struct Mater. http://dx.doi.org/10.1177/1099636217694229.
and first-ply failure improvements over straight-fibre laminates. Comput Struct
[58] Mukhopadhyay T, Naskar S, Dey S, Adhikari S. On quantifying the effect of noise in
2008;86(9):897–907.
surrogate based stochastic free vibration analysis of laminated composite shallow
[28] Cheng-Hsien C, Ya-Jung L. Progressive failure of marine GFRP laminated plates
shells. Compos Struct 2016;140:798–805.
under impulsive water pressure. J Compos Mater 2008;42(4):335–65.
[59] Dey S, Mukhopadhyay T, Sahu SK, Adhikari S. Stochastic dynamic stability analysis
[29] Andersons J, Joffe R, Spārniņš E. Statistical model of the transverse ply cracking in
of composite curved panels subjected to non-uniform partial edge loading. Eur J
cross-ply laminates by strength and fracture toughness based failure criteria. Eng
Mech/A Solids http://dx.doi.org/10.1016/j.euromechsol.2017.09.005.
Fract Mech 2008;75(9):2651–65.
[60] Dey S, Mukhopadhyay T, Adhikari S. Metamodel based high-fidelity stochastic
[30] El Meiche N, Tounsi A, Adda-Bedia EA, Megueni A. Analysis of the transverse
analysis of composite laminates: a concise review with critical comparative as-
cracking in hybrid cross-ply composite laminates. Comput Mater Sci
sessment. Compos Struct 2017;171:227–50.
2009;46(4):1102–8.
[61] Dey S, Mukhopadhyay T, Sahu SK, Adhikari S. Effect of cutout on stochastic natural
[31] Adali S, Cagdas Izzet U. Failure analysis of curved composite panels based on first-
frequency of composite curved panels. Compos B Eng 2016;105:188–202.
ply and buckling failures. Proc Eng 2011;10:1591–6.
[62] Dey S, Mukhopadhyay T, Spickenheuer A, Gohs U, Adhikari S. Uncertainty quan-
[32] Gadade AM, Lal A, Singh BN. Stochastic buckling and first ply failure analysis of
tification in natural frequency of composite plates – an Artificial neural network
laminated composite plate. Adv Struct Eng 2014;125–135.
based approach. Adv Compos Lett 2016;25(2):43–8.
[33] Gohari S, Sharifi S, Vrcelj Z, Yahya MY. First-ply failure prediction of an un-
[63] Dey S, Mukhopadhyay T, Khodaparast HH, Kerfriden P, Adhikari S. Rotational and
symmetrical laminated ellipsoidal woven GFRP composite shell with incorporated

566
P.K. Karsh et al. Composite Structures 184 (2018) 554–567

ply-level uncertainty in response of composite shallow conical shells. Compos Struct [78] Mukhopadhyay T, Adhikari S. Stochastic mechanics of metamaterials. Compos
2015;131:594–660. Struct 2017;162:85–97.
[64] Dey S, Mukhopadhyay T, Sahu SK, Li G, Rabitz H, Adhikari S. Thermal uncertainty [79] Mukhopadhyay T, Adhikari S. Effective in-plane elastic moduli of quasi-random
quantification in frequency responses of laminated composite plates. Compos B Eng spatially irregular hexagonal lattices. Int J Eng Sci 2017;119:142–79.
2015;80:186–97. [80] Mukhopadhyay T, Adhikari S, Batou A. Frequency domain homogenization for the
[65] Dey S, Naskar S, Mukhopadhyay T, Gohs U, Spickenheuer A, Bittrich L, et al. viscoelastic properties of spatially correlated quasi-periodic lattices. Int J Mech Sci
Uncertain natural frequency analysis of composite plates including effect of noise – http://dx.doi.org/10.1016/j.ijmecsci.2017.09.004.
a polynomial neural network approach. Compos Struct 2016;143:130–42. [81] Mukhopadhyay T, Dey TK, Dey S, Chakrabarti A. Optimization of fiber reinforced
[66] Dey S, Mukhopadhyay T, Khodaparast HH, Adhikari S. A response surface model- polymer web core bridge deck – a hybrid approach. Struct Eng Int
ling approach for resonance driven reliability based optimization of composite 2015;25(2):173–83.
shells. Period Polytech – Civ Eng 2016;60(1):103–11. [82] Dey TK, Mukhopadhyay T, Chakrabarti A, Sharma UK. Efficient lightweight design
[67] Dey S, Mukhopadhyay T, Khodaparast HH, Adhikari S. Stochastic natural frequency of FRP bridge deck. Proc Inst Civ Eng – Struct Build 2015;168(10):697–707.
of composite conical shells. Acta Mech 2015;226(8):2537–53. [83] Mukhopadhyay T, Dey TK, Chowdhury R, Chakrabarti A, Adhikari S. Optimum
[68] Dey S, Mukhopadhyay T, Khodaparast HH, Adhikari S. Fuzzy uncertainty propa- design of FRP bridge deck: an efficient RS-HDMR based approach. Struct
gation in composites using Gram-Schmidt polynomial chaos expansion. Appl Math Multidiscip Optim 2015;52(3):459–77.
Model 2016;40(7–8):4412–28. [84] Mahata A, Mukhopadhyay T, Adhikari S. A polynomial chaos expansion based
[69] Gim CK. Plate finite element modeling of laminated plates. Comput Struct molecular dynamics study for probabilistic strength analysis of nano-twinned
1994;52(1):157–68. copper. Mater Res Express 2016;3:036501.
[70] Jones RM. Mechanics of composite materials, Scripta Book Co.; 1975. [85] Mukhopadhyay T, Chowdhury R, Chakrabarti A. Structural damage identification: a
[71] Reddy YSN, Reddy JN. Linear and nonlinear failure analysis of composite laminates random sampling-high dimensional model representation approach. Adv Struct Eng
with transverse shear. Compos Sci Technol 1992;44:227–55. 2016;19(6):908–27.
[72] Daniel IM, Ishai O. Engineering mechanics of composite materials, 2nd ed. [86] Mukhopadhyay T. A multivariate adaptive regression splines based damage iden-
University Press; 2006. tification methodology for web core composite bridges including the effect of noise.
[73] Bathe KJ. Finite element procedures in engineering analysis. New Delhi: Prentice J Sandw Struct Mater http://dx.doi.org/10.1177/1099636216682533.
Hall Inc.; 1990. [87] Mukhopadhyay T, Mahata A, Dey S, Adhikari S. Probabilistic analysis and design of
[74] Kam TY, Sher HF, Chao TN, Chang RR. Predictions of deflection and first-ply failure HCP nanowires: an efficient surrogate based molecular dynamics simulation ap-
load of thin laminated composite plates via the finite element approach. Int J Solids proach. J Mater Sci Technol 2016;32(12):1345–51.
Struct 1996;33(3):375–98. [88] Metya S, Mukhopadhyay T, Adhikari S, Bhattacharya G. System reliability analysis
[75] Mukhopadhyay T, Adhikari S. Equivalent in-plane elastic properties of irregular of soil slopes with general slip surfaces using multivariate adaptive regression
honeycombs: an analytical approach. Int J Solids Struct 2016;91:169–84. splines. Comput Geotech 2017;87:212–28.
[76] Mukhopadhyay T, Adhikari S. Free vibration analysis of sandwich panels with [89] Mukhopadhyay T, Chakraborty S, Dey S, Adhikari S, Chowdhury R. A critical as-
randomly irregular honeycomb core. J Eng Mech 2016;142(11):06016008. sessment of Kriging model variants for high-fidelity uncertainty quantification in
[77] Mukhopadhyay T, Adhikari S. Effective in-plane elastic properties of auxetic hon- dynamics of composite shells. Arch Comput Methods Eng 2017;24(3):495–518.
eycombs with spatial irregularity. Mech Mater 2016;95:204–22.

567