Composite Structures 202 (2018) 590–602

Contents lists available at ScienceDirect

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

Numerical methodologies for simulating bird-strike on composite wings T

A. Riccio, R. Cristiano, S. Saputo, A. Sellitto
Department of Engineering, Università Degli Studi della Campania “Luigi Vanvitelli”, Aversa, CE, Italy

A R T I C LE I N FO A B S T R A C T

Keywords: During its service life, an airplane can experience impact events with different foreign objects. In case of impact
Bird strike with a flying bird, the term “Bird Strike” is commonly adopted. Bird strike can be catastrophic, especially when
Bird modeling small general aviation airplanes are involved. Indeed, the Federal Aviation Administration (FAA), to prevent
Lagrangian method catastrophic failures with casualties, obliges airplanes to be able to complete the flight after an impact with a
ALE method
medium dimensions bird. Hence, it is clear that, when designing aerospace components, the potential effects of
SPH method
CEL method
the bird strike events must be taken into account to guarantee the structural integrity and, as a consequence, the
Composite structures passengers and the pilot safety. In order to optimize the costs, the experimental activity for certification purpose
should be suitably prepared and driven by an extensive campaign of numerical simulations. Therefore, the
accuracy and the effectiveness of the numerical models able to simulate the bird strike event and its con-
sequences on the structural integrity become of major concern when designing aerospace components. In this
paper, the numerical methodologies, commonly adopted for the simulation of the bird strike event, are presented
and assessed focusing on their capability to predict the induced damage and the composite components’ residual
strength.

1. Introduction
Impact with flying birds also called “bird strikes” can be critical for
flight safety especially when small general aviation airplanes are in-
volved. Nowadays, collisions between airplanes and birds are becoming
much more frequent because of the increasing air traffic. The Federal
Aviation Administration airworthiness rules obliges an airplane to be
able to complete the flight after an impact with a bird of medium di-
mension. Indeed, during the certification process, an airplane must
prove its capability to land safely after a representative impact event
[1]. The bird strike event is a short duration (milli-seconds range)/high
intensity loads event characterized by a strong interaction among the
impact energy, the bird deformation and the airplane component
structural behavior. During the bird strike, the target (airplane struc-
ture) and the projectile (bird) are both subjected to large deformations
and inelastic strain rates.
To obtain an accurate prediction of the bird strike event induced
damage, it is important to model carefully the bird shape. Actually,
several bird models have been developed, based on test data during the
last 30 years [2]. One of the first researchers investigating the bird
mechanical behavior during an impact was Wilbeck [2–4]; he defined
the bird shape (Fig. 1) and the mechanical properties to be used during
the numerical simulations.
During the tests, Wilbeck impacted the projectile on a rigid plate
within a 100–300 m/s velocity range. The peak pressure, changing with
projectile velocity, was monitored at different position on the plate by
transducers. The normalized results, in terms of time vs. stagnation
pressure, are shown in Fig. 1. The results uncertainty is due to the
piezoelectric transducers (mounted on the rigid plate) limited fre-
quency response, which did not allow to measure the pressure for
period exceeding 5 µs. Fig. 1 shows that the maximum pressure value is
next to the impact point and then it goes through a gradual stabilization
around the stagnation pressure value. Moreover, the normalized Hu-
goniot pressure value decreases with the increasing of the impact ve-
locity value because the Hugoniot pressure value is punctual and the
impact phenomenon duration is very short. Taking into account the
above considerations, the bird strike event can be characterized by two
stages: the initial shock and the steady flow. The initial shock pressure
(Hugoniot pressure) is evaluated by means of Eq. (1) whereas the
steady flow pressure is represented by Eq. (2) formulated according to
the Bernoulli’s theory:
Pshock = ρ0 νshock νimpact (1)
1 2
Pstagnation = ρ νimpact
2 0 (2)
From Eqs. (1) and (2), it can be stated that pressure values do not
depend from the projectile mass. Furthermore, the Hugoniot pressure
can be considered as the maximum pressure value for an impact event
while the steady flow pressure is representative of the final pressure
value once the flow stabilizes.
Among the several materials tested during the last decades, the
ballistics gelatin with 10% porosity and 950 kg/m3 mean density has
been found the most suitable to mimic the bird impact behavior.
Indeed, the ballistic gelatin is characterized, as for real birds, by a mean

https://doi.org/10.1016/j.compstruct.2018.03.018
Received 26 January 2018; Received in revised form 28 February 2018; Accepted 7 March 2018
Available online 09 March 2018
0263-8223/ © 2018 Elsevier Ltd. All rights reserved.
A. Riccio et al.
Fig. 1. Geometrical characteristics of the bird model and simulations results by Wilbeck [5].
density which is lower than the water’s one. To represent the pressure-
density relationship in a medium sizes bird, the equation of state (EOS)
can be used. The most frequently adopted equation for the re-
presentation of a bird strike event is the third-degree order equation,
defined in Eq. (3) [2]:
ρ body which behaves as a fluid and spreads the impact force on a great
P = C0 + C1 μ + C2 μ2 + C3 μ3; μ = −1(3)
ρ0
where ρ is the current density and ρ0 is the reference state density. As
defined in [2], the constant Ci in Eq. (3) are function of the initial bird
density, of the sound speed in water, and of an experimental constant k.
Concerning the bird shape, circular cylinder, sphere, hemi-spherical
cylinder, and ellipsoid [6–9] shapes can be adopted. However, in lit-
erature, detailed bird shapes can be found such as in [10], where a
detailed bird shape with wings, feathers, and beak is introduced.
Budgey [11] and Stoll [12], after using truncated cylinders and ellip-
soid bird shapes, realized that the most suitable shape was a hemi-
spherical cylinder with 2:1 length-diameter ratio. Other authors per-
formed studies considering variations in bird mass, speed and impact
angles. As a matter of facts, Barber at al. [13,14] conducted tests
varying the bird mass (from 60 g to 4 kg) and the impact velocity (from
50 m/s to 300 m/s), considering the impact angles from direct normal
to oblique at 25 degrees. From these tests, non-dimensional impulse-
impact velocity curves, non-dimensional impulse-impact duration
curves, and rise time-impact velocity information have been plotted
with a considerable improvement in knowledge about the impact re-
sponse. Edge and Degrieck [15] focused on the relations between the
bird mass, bird density, and under/over-wing diameters determination.
Budgey [11] investigated the influence on the main bird mass and
geometrical parameters on the impact response providing a compre-
hensive insight on the advantages of using detailed artificial birds for
impact events experiments.
As recent developments, some authors pointed out that, during a
bird strike event, the bird behaves as a high deformable projectile
characterized by a yield strength lower than the sustained stress. Hence,
the impact phenomenon can be seen as a hydrodynamic impact
[4,5,11–13,16–24] and the bird can be modelled as a mass of water,
being the hydrodynamic impact characterized by three bird loading
stages: shock pressure, shock pressure decay, and the steady state
pressure.
In the following sections, several literature numerical approaches
are explored both for the modelling of the bird and the composite wing.
A specific section is dedicated to the damage mechanisms to be taken
into account when simulating the impact event on composite structures.
Finally a numerical application is presented which allows to compare
the effects of the bird modelling approaches on numerical results in
terms of leading edge deformation and damage distribution in the
Composite Structures 202 (2018) 590–602
composite wing-box.
2. Numerical approaches for modelling the bird and the structure
During the design phase, the projectile can be modelled as a soft
area of the target. Even if, accurate results prediction depends on bird
(3)
model, different numerical approaches for modelling the bird are
available in literature which lead to a relevant variation in numerical
predictions: Lagrangian Model, Arbitrary Lagrangian Eulerian Model
(ALE), and Smooth Particle Hydrodynamics (SPH) [25–27].
The Lagrangian method has been widely used for modelling the bird
and the structure involved in a bird strike event, even if this method has
been proved to be unsatisfactory for simulating the bird behavior.
Indeed, according to the Lagrangian method, which considers the same
element formulation both for the bird and the structure, the bird mesh
distorts (due to great deformation after impact) together with the
structure affecting the results accuracy. This is the main drawback of
using the Lagrangian method, commonly adopted in solids mechanics,
to represent the bird which is characterized by a fluid dynamics be-
havior, best represented by a Eulerian formulation.
More realistic prediction of the bird strike, which can be regarded as
a fluid-structure interaction problem, can be obtained by means of the
ALE and SPH methods. According to these approaches, bird and
structure are individually modelled by using the most suitable tech-
nique and then they are coupled in the same simulation. The ALE
method models the fluid-structure interaction problems by means of a
Lagrangian structure and a Eulerian bird. On the other hand, the SPH
method models the bird as an agglomerate of particles and the structure
with a Lagrangian mesh.
In recent years, a great number of studies on these methods have
been carried out by numerous researchers. Hormann et al. [28] in-
vestigated the bird strike phenomenon on a horizontal tail leading edge
using the non-linear finite element code LS-Dyna, modelling the bird
with the Lagrangian and the Arbitrary Lagrangian Eulerian methods.
The plastic strain was compared with the experimental data.
Guida [29] compared the computational time needed to simulate a
bird impact event by adopting Lagrangian, Eulerian, and SPH methods.
The bird shape and the composite material lay-up influence have been
discussed. Moreover, the material energy absorption, at different im-
pact phenomenon stages, has been examined by using the MSC Dytran
software.
Blair [30] numerically validated the cylindrical bird shape with
hemi-spherical edge by impacting the bird with a rigid plate, con-
sidering different mesh densities. The contact elements between the
rigid plate and the bird have been changed and noted during the ana-
lysis. Blair tabulated also the bird’s dimensions, the material properties,
591
A. Riccio et al.
Fig. 2. Lagrangian elements description.
and the adopted equation of state.
Ubels et al. [31] simulated the bird strike event on a leading edge
using low velocities, and suggested the use of SPH method for studying
the impact phenomenon.
Lastly, Ortecho [32] studied the bird strike event by using the La-
grangian, Arbitrary Lagrangian Eulerian, and Smooth Particle Hydro-
dynamics numerical models. The analyses were carried out at angled
and normal directions comparing the resulting peak force. The Smooth
Particle Hydrodynamics method has been recommended for modelling
and simulating the impact phenomenon.
A brief description of the introduced numerical approaches for the
bird modelling is giver hereafter in the next subsections.
2.1. Lagrangian formulation
The Lagrangian technique is generally used for solving problems
connected to solid mechanics. The mesh nodes are coincident with the
material points, which identify the location of the material during the
simulation, so the mesh distorts together with the structure during the
analysis, as shown in Fig. 2.
Since the material interfaces and the free surfaces move together
with the mesh, the Lagrangian solvers offer easier behavior for the
boundaries. The material properties at each point can be easily mon-
itored because the mesh is embedded into the structure. For such a
reason, the distortions and the deformations are reflected on the mesh
with the risk to introduce great inaccuracies in the numerical model.
To define the spatial material motion for any domain, the mass
conservation, linear momentum conservation, and energy conservation
equations have to be verified.
Mass Conservation:
dρ
+ ρ∇x · v = 0
dt (4)
Linear momentum conservation:
dv
ρ −∇x · σ −ρb = 0
dt (5)
Energy conservation:
Fig. 3. Eulerian elements description.
592
Composite Structures 202 (2018) 590–602
de int
ρ − D : σ + ∇x · q−ρs = 0
dt (6)
where ρ represents the density, v is the velocity of the material point, σ
is the second order Cauchy stress tensor, b denotes the force per mass
unit, eint is the internal energy, D denotes the strain rate tensor, q re-
presents the heat flux applied, and, finally, s is the applied heat source.
Each interaction time step during an explicit simulation depends on
the smallest element size of the whole computational domain; there-
fore, the time step becomes extremely small in case of great distortions
and the simulation becomes very slow. For solving this issue, the
warped elements can be deleted from the mesh by imposing a plastic
strain limit. The deletion eliminates mass and energy, violating the
principles of mass and energy conservation and introducing other kinds
of inaccuracies [33].
Hence, the Lagrangian method would need the remeshing of the
model due to excessive mesh deformations [31], but this additional
remeshing step would increase the solution time. Moreover, great de-
formations would, in any case, lead to mesh quality decay and would
cause inaccurate results.
Some authors [33] introduced the “small-strain” option as a coun-
termeasure for the extremely small time steps associated to the La-
grangian formulation. The “small-strain” option allows to not update
the Jacobian matrix at every time steps; hence, wide structural distor-
tions are neglected. Unluckily, this option is not appropriate for solving
problems with very great deformations/distortions such as bird strike
problems.
2.2. Eulerian formulation
The Eulerian technique is generally used for solving problems
connected to fluid dynamics [34,35]. The mesh is considered as a
control volume, the computational grid is settled in the space, and the
material gets through the mesh as shown in Fig. 3.
The computational domain includes both the region where the
material exists and the region where the material could exist later,
therefore the Eulerian method is very good for investigating problems
involved in wide distortions. Moreover, more than one material may be
A. Riccio et al. Composite Structures 202 (2018) 590–602

contained in a Eulerian mesh cell and the solver is completely able to
manage multiple materials in one cell.
Studying objects in motion, involved in the impact analyses, is very
difficult. Free surfaces and material interfaces move by the mesh, dif-
fusing everywhere in the computational grid. To avoid this issue, pre-
ferential transport of materials [36,37], based on separated calculations
for each interface material, is needed.
The material motion from the initial domain (t0) to the final domain
(tf) can be described by a ϕ function. The latter associates each
Lagrangian material point coordinate X to an Eulerian position x in the
space. It is defined by Eq. (7):
x = ϕ ( X ,t ) (7)
simplify the governing equations by separating them into two sets. In
the first phase the material moves with the mesh (Lagrangian step), in
the second phase the solution is mapped from the Lagrangian to the
reference domain. The Lagrangian and the Eulerian techniques are
performed individually and independently to each other, but the solu-
tion could not be robust as the fully coupled equations.
To evaluate the ALE approach, the equation of mass conservation,
the linear momentum conservation and the energy conservation are
reported below in terms of Eulerian coordinates.
Mass conservation:
∂ρ
+ ( v− v ALE )·∇x ρ + ρ∇ x · v = 0
∂t (9)

The Lagrangian coordinates are coupled to the material and in- Linear momentum conservation:
variant during time for the local configuration attached to the final
∂v
domain, whereas the Eulerian coordinates are invariant in time and ρ + ( v− v ALE )·∇x v−∇x · σ −ρb = 0
∂t (10)
position.
( )
d
The material derivative dt is joint to the partial derivative ex- Energy conservation:
pressed in the Eulerian framework as reported below:
∂e int
ρ + ( v− v ALE )·∇x e int− D : σ + ∇x · q−ρs = 0
d ∂ ∂t (11)
= + v·∇x
dt ∂t (8)
where vALE is the ALE domain velocity.
The above mentioned computational problems, and some others
related to the Eulerian technique, increase the computational time with
2.4. Smooth Particle Hydrodynamics (SPH)
respect to Lagrangian technique applied to structural problems. For
these reasons, the pure Eulerian technique it is not suggested for impact
The Smoothed Particle Hydrodynamics (SPH) technique is a mesh-
phenomena investigations also because, generally, the Eulerian
less method based on Lagrangian formulation. It has been used in as-
methods needs a finer mesh than the Lagrangian method to have the
trophysics for managing the fluid masses in 3D space [27,48,49]. The
same accuracy level [38].
SPH technique can be used for solving problems involved in severe
deformations and for high projectile distortions after impact. First at-
2.3. Arbitrary Lagrangian Eulerian (ALE) formulation tempts to use this approach have been made by Lucy [50], Gingold and
Monaghan [51] for astrophysical problems. Lately, it has been used also
The Arbitrary Lagrangian Eulerian (ALE) formulation (also known in high velocity impact phenomena [52–57].
as mixed Lagrangian Eulerian method) is often used for studying en- In the SPH formulation, the continuum is considered as a set of
gineering problems involved in fluid-structure interaction. The spatial discrete independent particles interacting each other. Since the SPH is a
and material domains connected to the ALE method are, respectively, Lagrangian technique, it can prevent the issues associated to the
the same of the Eulerian and Lagrangian methods. In addition, the ALE Eulerian formulations [56] and it can be adopted in conjunction with a
method introduces a reference domain to deal with the computational standard finite element method. The mass, momentum and energy
mesh. The model, shown in Fig. 4, moves with a velocity w, with respect conservation equations are changed from partial differential equations
to the reference domain independently from the spatial domain and the to integral equations by using the smoothing kernel function.
material particles, as for the Eulerian technique. Mass conservation:
To use the ALE technique, the differential equations need to be
modified including also the referential domain arbitrary velocity effect. dρi mj
+ ρi ∑ (vj−vi ) Aij = 0
The ALE domain becomes a Eulerian domain when w = 0, whereas the dt j
ρj (12)
ALE domain becomes a Lagrangian domain when w is the material
particles velocity. In other cases, the reference domain is an ALE do- Linear momentum conservation:
main.
There are two different ways for solving the ALE equations [39–47]. dvi ⎛ σj σ ⎞
+ ∑ mj ⎜ 2 Aji − i2 Aij ⎟ = 0
The first way considers the fully coupled ALE equations resolution, dt ρj ρi (13)
j ⎝ ⎠
managing a single material into an element. The second way (called
operator split approach) divides each solution step in two phases and Energy conservation:

Fig. 4. Arbitrary Lagrangian Eulerian (ALE) elements description.

593
A. Riccio et al.
Fig. 5. Particle spacing and Kernel function.
dei P equations of conservation.
+ i2 ∑ mj (vi−vj ) Aij = 0
dt ρi j (14)
where Pi, σi, and mi represent the pressure term, the Cauchy stress, and
the particle mass, respectively. Aij is the kernel function gradient and is
expressed by the following equation:
∂ The Coupled Eulerian Lagrangian (CEL) method eliminates the ex-
Aij = W (x −x j,h)
∂x i
W is the approximate Dirac Function based on the smoothing length
h, whereas α is a linear coefficient.
1 ⎛d⎞
W (d,h) = θ by means of which the Eulerian material goes and interacts with the
hα ⎝ h ⎠
The particles in the support domain of the kernel is described in
Fig. 5.
This latter supplies the kernel estimation of the relevant variable
field in a specific location in the space. The smoothing kernel is pro-
portional to the smoothing length h. As soon as h → 0, the smoothing
kernel moves closer the delta function. The kernel function describes
the datum point influence range in the continuum, and the smoothing
length defines the number of neighboring particles included in the
above-mentioned range (a particle can be seen as a neighbor if it falls
within another particle smoothing length). Fig. 6 shows the k-particle
neighborhood, defined as a circle with a 2hk radius. All the particles
which fall within the circle will interact with the k-particle.
The integration by parts allows the spatial derivatives of the in-
definite field variables to be substituted by the well-known derivatives
of the smoothing kernel.
Summing the discrete points into the space, evaluated as the cen-
troids of the “smoothed particles”, it is possible to substitute the re-
sultant integral equations for numerical estimation. Each particle has an
own mass, velocity and stress state, which change according to the
Fig. 6. SPH k-particle neighborhood.
Composite Structures 202 (2018) 590–602
The SPH method is becoming more common for bird strike nu-
merical simulations because the mesh distortions do not exist anymore.
2.5. Coupled Eulerian Lagrangian (CEL) method
(15) treme numerical bird mesh distortions related to Lagrangian formula-
tion.
The Lagrangian formulation is used to model the structure. The
Eulerian-Lagrangian contacts [58] are considered for moving the im-
pacting loads to the Lagrangian structure. The mesh is a stationary cube
(16)
Lagrangian structure. The flowing material through the mesh is tracked
by the Eulerian Volume Fractions (EVF) which represents the ratio
whereby the material is filled with the Eulerian elements. EVF = 1
means that the element fills with material; EVF = 0 means that the
element is completely void.
With Coupled Eulerian Lagrangian method, the advantages of
Eulerian and Lagrangian methods have been implemented. The contacts
between the two domains have been discretized using the general
contact algorithm based on the penalty contact method. Moreover, the
contact force fp applied between the two domains is proportional to
their penetration distance dp.
f p = k p dp (17)
where kp is the penalty stiffness and depends on the Eulerian and
Lagrangian material properties.
In the frame of bird-strike investigations, the grid dimensions con-
taining the Eulerian elements need to be adequately large. It is neces-
sary to avoid a bird material loss after an impact because it is associated
with a loss of kinetic energy, and, hence, a loss of accuracy level of the
obtained results.
3. Numerical approaches for modelling the damage evolution in
the composite wing
Carbon Fibers Reinforced Plastics composites have been increas-
ingly used for aerospace structural applications due to their high spe-
cific strength and stiffness compared to conventional metallic alloys.
Nonetheless, structures made of composite material have been proved
to be very sensitive to damage rising during the manufacturing phase or
the impact with foreign objects. The damage may quickly grow leading
to the whole structure failure, notably under a compressive load. The
laminated composite structures have been demonstrated to be involved
by inter-laminar fractures, such as delamination or skin-stringer de-
bonding, and intra-laminar damage mechanisms, like matrix and fibers
failure. Therefore, both these types of damage have to be taken into
account when simulating a bird strike event on a composite structures
594
A. Riccio et al. Composite Structures 202 (2018) 590–602

where Gj is the Energy Release Rate associated to the fracture mode j,

Gjc is the critical Energy Release Rate associated to the fracture mode j,
and α is an experimental coefficient included in the range 1 ÷ 1.6.
According to Eq. (19), the energy Gc dissipated due to the damage can
be evaluated as:
1
Gc =
1 α 1 α α 1/ α
(( GI
) +(
GI + GII + GIII GIc
GII
GI + GII + GIII GIIc ) +( GIII 1
))
GI + GII + GIII GIIIc

Fig. 7. Traction-separation law for cohesive material.
in order to obtain accurate results.
Many authors have investigated the composite structure mechanical
behavior in case of inter-laminar and intra-laminar damages. In
[59–66] experimental studies which have improved the inter-laminar
defects knowledge are reported. In [67–69] the buckling behavior of
two-dimensional structures with trough-the-width delaminations have
been investigated, while in [70–72] three-dimensional approaches have
been examined.
To study the inter-laminar damage onset and evolution, a Cohesive
Zone Model (CZM) has been adopted in the frame of the present paper.
Cohesive elements have been used to reproduce the interface behavior
between different layers by evaluating the interface strength, the frac-
ture toughness and the stiffness. These elements follow the traction-
separation law based on the energy release rate, as highlighted in Fig. 7.
As shown in Fig. 7, the first segment OA identifies the damage onset
governed by the quadratic nominal stress criterion of Eq. (18)
(QUADS), which has been used to evaluate the point A of Fig. 7, re-
presentative of the damage onset:
2 2 2 failure, matrix tensile failure and matrix compressive failure. The
⎛ 〈σn 〉 ⎞ + ⎛ σt ⎞ + ⎛ σs ⎞ = 1
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎝ NMAX ⎠ ⎝ TMAX ⎠ ⎝ SMAX ⎠
(20)
The damage coefficient D, used to compute the degradation of the
cohesive stiffness (Point B in Fig. 7) is evaluated as:
δf (δmax−δ0)
D=
δmax (δf −δ0) (21)
where δf = 2Gc/Teff and δmax is the maximum effective displacement
attained during the analysis.
The intra-laminar damages are usually numerically simulated by
using two main approaches [73]: Progressive Ply Damage based
methodologies (PDA) and Continuum Damage Mechanics based meth-
odologies. (CDM).
Progressive ply damage based methodologies use failure criteria for
the failure identification [74–83] and ply-discount methods [84–92] for
the selected elastic ply properties instantaneous degradation in the
frame of Instantaneous Degradation Models (IDM).
The Continuum Damage Mechanics methodologies consider the
microstructure of the composite material using equations related to the
damage onset and evolution. These techniques are suitable for Gradual
Degradation models (GDM).
To evaluate the damage onset, Hashin’s criteria have been im-
plemented here. These criteria allow to characterize each different
intra-laminar failure mode, such as fiber tensile and compressive
adopted Hashin’s formulation follows Eqs. (22)–(25):
(18)
2 2
σ̂ σ̂
Fiber tension (σ11̂ ⩾ 0) Fft = ⎛ 11T ⎞ + α ⎛ 12L ⎞ = 1
⎜ ⎟ ⎜ ⎟

In Eq. (18) σn, σt, and σs represent the peak values of the nominal ⎝X ⎠ ⎝S ⎠ (22)
stress considering a deformation purely normal to the interface or
2
purely in the first or the second shear direction respectively, and NMAX, σ̂
Fiber compression (σ11̂ < 0) Ffc = ⎛ 11C ⎞ = 1⎜ ⎟

TMAX, and SMAX are the nominal stresses in the pure normal mode, first ⎝X ⎠ (23)
and second shear directions respectively. The Macaulay brackets
2 2
(< >) indicate that the damage is not initiated by a pure compressive σ̂ σ̂
Matrix Tension (σ22̂ ⩾ 0) Fmt = ⎛ 22T ⎞ + ⎛ 12L ⎞ = 1
⎜ ⎟ ⎜ ⎟

stress state. ⎝Y ⎠ ⎝S ⎠ (24)

The evolution phase (segment OC in Fig. 7) identifies the damage
2 2
growth until the complete failure of the cohesive elements. The evo- σ̂ C
⎛Y ⎞ ⎤ σ22̂
Matrix Compression (σ22̂ < 0) Fmc = ⎛ 22T ⎞ + ⎡
⎢ 2S T −1⎥ Y C
⎜ ⎟ ⎜ ⎟
lution law adopted in the present work is the Power law based on the ⎝ 2S ⎠ ⎣⎝ ⎠ ⎦
energy release rate, reported in the following equation: 2
σ ̂
+ ⎛ 12L ⎞ = 1
⎜ ⎟

α α α
⎛ GI ⎞ + ⎛ GII ⎞ + ⎛ GIII ⎞ = 1 ⎝S ⎠ (25)
⎜ ⎟ ⎜ ⎟ ⎜ ⎟

⎝ GIc ⎠ ⎝ GIIc ⎠ ⎝ GIIIc ⎠ (19) where the coefficient α (set = 1 in this paper) allows to take into

Fig. 8. Constitutive relation adopted for each intra-laminar failure mode.

595
A. Riccio et al. Composite Structures 202 (2018) 590–602

Fig. 9. Global-Local interface.

Table 1 material is completely damaged (dI = 1).

Section Wing dimensions. In order to evaluate both the inter-laminar and intra-laminar da-
mages, a finite element model with a discretization ply by ply is man-
Section wing length (L) 2.71 × 103 mm
Air foil profile chord (c) 3.10 × 103 mm datory. In order to reduce the computational time linked to the very
Wing ribs interaxle spacing (d) 5.92 × 102 mm fine mesh, a global-local approach can be introduced. The studied local
Wing rib height (h) 4.23 × 102 mm domain can be connected to the global domain by means of transition
elements, as developed by Davila and Johnson [93,94]. Each transition
element can be employed to link shell and brick elements for calcu-
account the shear stress contribution to the fiber tensile failure mode. lating the three-dimensional response. These kinds of elements have
The constitutive relation adopted for each intra-laminar failure mode is been adopted also by Cho and Kim [95] for investigating the buckling
graphically showed in Fig. 8. phenomenon of a delaminated composite structure by connecting the
In Fig. 8, two different phases can be distinguished: the undamaged global domain to the local delaminated area. In [96] the post-buckling
phase from the beginning to the point A which represents the damage behavior of a delaminated composite structure under compressive load
onset, and a propagation phase. The damage effects for each failure has been studied. The adopted transition elements have been demon-
modes have been taken into account thanks to a stiffness coefficients strated to be efficient for this kind of investigation. However, the
reduction given by Eq. (26): transition elements play an important role during the meshing phase
δIt,eq (δI ,eq−δI0,eq) because the distorted elements could nullify the results in the transition
dI = ; (δI0,eq ⩽ δI ,eq ⩽ δIt,eq); I ∈ (fc ,ft ,mc ,mt ) regions. In order to study the local phenomena, different FE analytical
δI ,eq (δIt,eq−δI0,eq) (26) tools which combine finite element models with different mesh den-
where δ0I,eq is the equivalent displacement at which the specific Hashin’s sities, and/or different element types are available. The Multipoint
criterion is satisfied, δtI,eq is the equivalent displacement at which the Constraint (MPC) [58] option in ABAQUS® code has been adopted by

Fig. 10. Wing Section Structure (I-II-III).

596
A. Riccio et al. Composite Structures 202 (2018) 590–602

Table 2 ui = NA uA + NB uB + NC uC
Material mechanical properties and stacking sequence. vi = NA vA + NB vB + NC vC
wi = NA wA + NB wB + NC wC
Properties Value
ϕxi = NA ϕxA + NB ϕxB + NC ϕxC
Density 1400 kg/m3 ϕyi = NA ϕyA + NB ϕyB + NC ϕyC (28)
Orthotropic E11 = 130050 MPa; E22 = E33 = 11550 MPa
properties G12 = G13 = G23 = 6000 MPa; The kinematic relations between the node P of the solid model and
ν12 = ν13 = ν23 = 0.312 the point i can be formulated by applying Kirchhoff's theory. The dis-
Strength XT = 1022.7 MPa; XC = 613.5 MPa; YT = 54 MPa;
YC = 170 MPa
placements of the node P are expressed as a function of the displace-
S12 = S13 = 63 MPa; S23v= 28 MPa ment of the point i:
Fracture energies GT1c = 11.48 kJ/m2; GT2c = 4.13 kJ/m2
GC1c = 0.35 kJ/m2; GC2c = 3.23 kJ/m2
up = ui + hϕyi

Stacking Sequence
vp = vi−hϕxi
Stringer [0/45/−45/90/90/−45/45/0/0/90/0/0/90/0/0/45/ wp = wi (29)
−45/90/90/−45/45/0]
Upper/Lower [0/45/−45/90/0/45/−45/0/0/−45/45/0/90/−45/45/ By making suitable replacements, it is possible to express the node P
0/0/0] degrees of freedom in relation to the A-C nodes degrees of freedom:
Ribs [0/45/−45/90/90/−45/45/0/0/90/0/0/90/0/0/45/
−45/90/90/−45] up uA uB uC
⎧ ⎫
Omega Stringer [0/45/−45/90/90/−45/45/0/0/45/−45/90/0/0/0/90/ vp = NA·⎧ vA ⎫ + NB ·⎧ vB ⎫ + NC ·⎧ vC ⎫ + NA·h
−45/45/0/0/45/−45/90/90/−45/45/0] ⎨w ⎬ ⎨ wA ⎬ ⎨ wB ⎬ ⎨ wC ⎬
⎩ p⎭ ⎩ ⎭ ⎩ ⎭ ⎩ ⎭
Spar [0/45/−45/90/0/−45/45/0/0/90/0/0/90/0/0/45/
−45/0/90/−45/45/0] ⎧ ϕ yA ⎫ ⎧ ϕyB ⎫ ⎧ ϕ yC ⎫
· −ϕxA + NB ·h· −ϕxB + NC ·h· −ϕxC
⎨ ⎬ ⎨ ⎬ ⎨ ⎬
⎩ 0 ⎭ ⎩ 0 ⎭ ⎩ 0 ⎭ (30)
Alesi et al. [97] to investigate the skin/stringer interfaces in stiffened
panels. The shell to solid coupling option ABAQUS® code has been used The latter has been written on the assumption that global and local
by Krueger and Brien [98] and Krueger and Minguet [99] to study the reference systems are coincident. When the two systems do not coin-
propagation of the delamination and the skin-stiffener debonding in cide, it is possible to reformulate the previous equation in the global
composite laminates, respectively. Furthermore, the global-local tools reference system by adopting a rotation matrix R∗∗:
available in the most common FE based codes have been discussed in
0
[100]. ABAQUS® code and ANSYS® code have been compared through ⎧ up ⎫ u0 u0 ⎧ u0 ϕ0
⎪ 0⎪ ⎧
⎪ A⎫ ⎪ ⎧
⎪ B⎫ ⎪ ⎪ C⎫ ⎪ ⎧
⎪ yA ⎫
⎪
a common numerical test case on the non-linear buckling analysis of a vp = NA· vA0 + NB · vB0 + NC · vC0 + NA·h·R∗∗· −ϕ0
⎨ ⎬ ⎨ 0⎬ ⎨ 0⎬ ⎨ 0⎬ ⎨ xA ⎬
square plate. The obtained results have been compared to reference ⎪ wp0 ⎪ ⎪ wA ⎪ ⎪ wB ⎪ ⎪ wC ⎪ ⎪ 0 ⎪
⎩ ⎭ ⎩ ⎭ ⎩ ⎭ ⎩ ⎭ ⎩ ⎭
results, obtained by adopting a uniform discretization with a detailed
mesh. ⎧ ϕ0 ϕ0
⎪ yB ⎫ ⎪ ⎧
⎪ yC ⎫
⎪
A multi-point constraint (MPC) technique has been used to join + NB ·h·R∗∗· 0 ∗∗
+ NC ·h·R · −ϕ0
⎨−ϕxB ⎬ ⎨ xC ⎬
zones with different discretization and/or different elements type. In ⎪ 0 ⎪ ⎪ 0 ⎪
⎩ ⎭ ⎩ ⎭ (31)
Fig. 9, a global 2-D elements domain and a local 3-D elements domain
connected by MPC have been showed. By solving this equation, it is possible to obtain the node P degrees
In Fig. 9, the A, B, and C nodes belonging to the two-dimensional of freedom of a solid domain depending on the degrees of freedom of
domain, are located on the border between the 2D model (colored) and nodes A, B, and C of a two-dimensional domain.
the discretized region with solid elements; the i-point is located on the
intersection between the A-C line and the orthogonal line to the A-C 4. Numerical model and finite element implementation
line containing the node P of the solid model. Considering the shape
functions of the 2D element: In order to give a realistic representation of the above mentioned
methodologies applied to a common test case, in this section a nu-
NA = −0.25(1 + ξi ξA)(1 + ηi ηA)(1−ξi ξA−ηi ηA) merical application is presented. A bird impact on a composite wing-
NB = +0.50(1 + ξi ξB )(1−ηi2) box is simulated by modelling the bird as a Rigid body, as an elastic
NC = −0.25(1 + ξi ξC )(1 + ηi ηC )(1−ξi ξC−ηi ηC ) continuum with a Lagrangian formulation, as a group of particles with
(27)
the SPH approach and as a fluid with an Eulerian approach coupled
with a Lagrangian approach for the composite wing. Comparing the
It is possible to formulate the following kinematic relations, which
results obtained with these formulations in term of deformation of the
link the A-C nodes and point i:
structure, maximum displacement, and damage onset and propagation,

Fig. 11. Wing Section Boundary condition (I). Mesh dimension and shape (II).

597
A. Riccio et al. Composite Structures 202 (2018) 590–602

considering one element through the thickness. All degrees of freedom

on both external edges have been constrained along y-direction during
the simulations, according to Fig. 10.
The mechanical material properties and the different sub-
components stacking sequences, highlighted in Fig. 10-I, have been
reported in Table 2. Surface-to-Surface tie constraints have been adopted
to connect different mesh discretization related to the different wing
section components and to obtain a finer mesh in the impacted zone, as
shown in Fig. 11-II.

4.2. Bird models

Fig. 12. Bird Structure. An impact velocity of 150 m/s has been considered. Different stu-
dies found that the most appropriate material, for defining a bird during
a high velocity impact event, is the gelatin with 10% porosity and
950 kg/m3 density. Moreover, one of the best geometrical bird shape is
a hemispherical cylinder with Lb/ϕ = 2 because it shows the real bird
pressure time history during the impact event, as declared in [12].
Anyway, the bird material density used in the present work has been
reduced to be 938 kg/m3, as suggested in [101]. The bird numerical
model and its dimensions are shown in Fig. 12.

• L = 228 mm, bird length;

b

Fig. 13. Rigid and Lagrangian model.

• Φ = 114 mm, bird diameter.
The bird model has been considered as an equivalent mass of water
because real birds are made of air and water in their lungs and bones
[102–104]. The hydrodynamic response of the bird has been studied
adopting the equation of state (EOS) materials. The peak pressure value
has been evaluated considering the following theoretical value (Hugo-
niot pressure):

PH = ρ0 US (U0) U0 (32)

where ρ0 is the initial material density, U0 is the impact velocity, and US

is the shock velocity, which is in turn a function of U0. The maximum
Fig. 14. SPH model. pressure value is reached at the beginning of the impact event. Gra-
dually, it is followed by a constant and stable pressure flow calculated
by Eq. (33).

1
P= ρ (U0)2
2 0 (33)

As already mentioned, the bird has been modelled by considering

Fig. 15. CEL approach.
it will be possible to have a realistic insight on the effect of the different
bird modelling formulations on the response of the composite structure
to the impact event.
4.1. Wing FEM model
The investigated composite wing section is composed by laminates
stacked with 0.125 mm thick plies, whose nominal geometry data are
reported in Table 1 and Fig. 10. The model has been discretized using
112,636 nodes and 56,724 continuum shell elements with reduced in-
tegration points (SC8R), available in the ABAQUS FE code library [58],
four different formulations: rigid body with lagrangian formulation,
Elastic body with Lagrangian formulation, Smooth Particle
Hydrodynamics (SPH) formulation and Coupled Lagrangian Eulerian
(CEL) formulation. The model has been validated by comparing
Hugoniot and stagnation pressures by impacting a bird on a rigid plate
at 116 m/s velocity, perpendicularly to the target [103]. The rigid and
the Lagrangian models, shown in Fig. 13, have been modelled adopting
the standard eight node elements with reduced integration points
(C3D8R) to reduce the computational costs.
The bird SPH model is shown in Fig. 14. According to this for-
mulation, the bird has been modelled by considering discrete particles,
with a distance to each other called “smoothing length”. The single
particle physical properties have been obtained by summing the prop-
erties of all particles.
Concerning the coupled Eulerian Lagrangian (CEL) formulation, a
Eulerian control volume have been introduced and discretized by
8000000 EC3D8R elements. The bird region has been modelled within
the above-mentioned control region, as shown in Fig. 15.
The bird region is characterized by the previously described EOS
property and the control region is not characterized by any property.
Since the bird can move around the control region, a part of the control
volume has been overlaid to the wing structure.

598
A. Riccio et al. Composite Structures 202 (2018) 590–602

Fig. 16. Maximum impact direction displacements contour plots a) Rigid Body displacements; b) Lagrangian model displacements; c) SPH model displacements; d) CEL model dis-
placement.

Fig. 17. Matrix traction cracking: a) Rigid Body; b) Lagrangian model; c) SPH model; d) CEL model.

4.3. Numerical results
The numerical results obtained by adopting the four mentioned
formulations for modelling the bird have been compared in terms of
wing maximum deformation, matrix traction cracking and bird de-
formations.
In Fig. 16 a comparison among the maximum displacements contour
plots along the impact direction obtained by means of the adopted
formulation for modelling the bird is introduced. According to the re-
sults shown in Fig. 16, the maximum deformation value, in bird impact
direction, has been found for the bird rigid body approach (3.069·102
mm). On the other hand, SPH, elastic Lagrangian and CEL approaches
provide a similar maximum deformation values (2.35·102 mm for the
SPH approach, 2.38·102 mm for the elastic Lagrangian approach and
2.457·102 mm for the CEL approach). As expected, the SPH formula-
tion provides the smaller displacement if compared to the other for-
mulations. Indeed, the modelled weak particles interaction causes a
larger impact area characterized by a lower maximum deflection.
As a consequence of a bird strike event on a composite laminate, the
matrix traction damage is the most extended damage mechanism and
can lead to a significant stiffness reduction.
In Fig. 17, the matrix traction cracking distributions obtained
thanks to the four adopted bird modelling formulations are compared.
The shape and the dimension of the damaged area are again

599
A. Riccio et al.
Fig. 18. Bird deformation and maximum displacement: a) Rigid Body; b) Lagrangian; c) SPH; d) CEL.
comparable for SPH, Lagrangian and CEL methods. The damaged area
obtained in the case of bird Rigid Body modelling (Fig. 17.a) is, as
expected, more extended. Indeed, all the impact energy is completely
transferred to the structure differently from the other formulations.
Remarkable differences can be appreciated in Fig. 18 where the bird
deformation obtained for the four adopted formulations at maximum
impact displacement are introduced.
From Fig. 18, the SPH and the CEL formulations seem to provide the
most realistic results in terms of bird deformation. Indeed, in the frame
of the Rigid Body formulation the bird does not undergo shape and
dimension deformations during the impact event which is, evidently,
not realistic as well as the elastic Lagrangian formulation which un-
derestimates the bird deformation without taking into account the mass
separation of the bird .
As a separated note, the CEL approach has been found to be the less
effective method in terms of computational costs, while the Rigid Bird
Model has been revealed to be the most effective one.
5. Conclusions
The numerical approaches commonly adopted for the simulation of
the bird strike phenomenon on a composite wing section have been
introduced. The state of the art formulations for the bird and structure
modelling have been analysed in detail providing discussion on the
added value and the weak points. An additional insight on the meth-
odologies adopted to simulate the damage onset and evolution in
composite structure has been given. Four different numerical ap-
proaches (Rigid Bird Model, Lagrangian Model, SPH Model and CEL
Model) to model the bird have been presented and compared for the
same bird mass, velocity, and boundary conditions. As expected, the
results show variable outputs in terms of wing indentation, matrix
cracking and bird deformation. The CEL approach has been found to be
the more accurate but at the same time, as expected, the less compu-
tationally effective one, whereas the Rigid Bird Model, providing un-
realistic results has revealed to be the computationally cheapest one.
References
[1] Policy Statement Number ANE-2001-35.13-R0, Policy for Bird Strike, Federal
Aviation Administration, U.S. Department of Transportation; April 2002.
[2] J.S. Wilbeck. Impact behavior of low strength projectiles, Air Force Materials
Laboratory, Technical Report AFML-TR-77-134; 1977.
600
Composite Structures 202 (2018) 590–602
[3] Wilbeck JS, Rand JL. The development of a substitute bird model. J Eng Gas
Turbines Power 1981;103(4):725–30.
[4] J.S. Wilbeck, J.P. Barber. Bird impact loading. The shock and vibration bulletin;
1978, p. 48.
[5] Lavoie M-A, Gakwaya A, Nejad Ensan M, Zimcik DG. Validation of available ap-
proaches for numerical bird strike modeling tools. Int Rev Mech Eng (IREME)
2007;1(4):225–31.
[6] Anghileri M, Castelletti LML, Mazza V. Birdstrike: approaches to the analysis of
impacts with penetration. In: Alves M, Jones N, editors. Impact loading of light-
weight structures. Southampton: WIT Press; 2005. p. 63–74.
[7] J. Frischbier, A. Kraus. Multiple Stage turbofan bird ingestion analysis with ALE
and SPH method. MTU Aero Engines GmbH ISABE-2005, D-80976 Muenchen,
Germany; 2005.
[8] McCarthy MA, Xiao JR, McCarthy CT, Kamoulakos A, Ramos J, Gallard JP, et al.
Modeling of bird strike on an aircraft wing leading edge made from fibre metal
laminates – part 2: modeling of impact with SPH bird model. Appl Compos Mater
2004;11(5):317–40.
[9] Rueda F, Beltrán F, Maderuelo C, Climent H. Birdstrike analysis of the wing slats of
EF-2000. Struct Mater 2002;11:189–98.
[10] S.C. McCallum, C. Constantinou. The influence of bird-shape in bird strike analysis.
In: 5th European LS-DYNA Users Conference, Birmingham; 2005.
[11] R. Budgey. The development of a substitute artificial bird by the international Bird
strike Research Group for use in aircraft component testing. In: International Bird
Strike Committee ISBC25/WP-IE3, Amsterdam; 17–21 April 2000.
[12] F. Stoll, R.A. Brockman. Finite element simulation of high-speed soft-body impacts.
In: 38th AIAA/ASME/ASCE/ AHS/ ASC Structures, Structural Dynamics, and
Material Conference; 1997, p. 334–344.
[13] J.P. Barber, H.R. Taylor, J.S. Wilbeck. Bird impact forces and pressures on rigid
and compliant targets. In: Air Force Flight Dynamics Laboratory Technical Report
AFFDL-TR-77-60; 1978.
[14] J.P. Barber, H.R. Taylor, J.S. Wilbeck. Characterization of bird impacts on a rigid
plate: part 1. Air Force Flight Dynamics Laboratory Technical Report AFFDL-TR-
75-5; 1975.
[15] C.H. Edge, J. Degrieck. Derivation of a dummy bird for analysis and test of air-
frame structures. In: 1999 Bird Strike Committee-USA/Canada, First Joint Annual
Meeting, Vancouver, BC. Paper 14.
[16] R. L. Peterson, J.P. Barber. Bird impact forces in aircraft windshield design. Air
Force Flight Dynamics Laboratory Technical Report AFFDL-TR-75-150; 1976.
[17] Airoldi A, Cacchione B. Modelling of impact forces and pressure in Lagrangian bird
strike analysis. Int J Impact Eng 2006;32(10):1651–77.
[18] Castelletti LML. Fluid-structure interaction for airworthiness problems using ex-
plicit finite element codes (PhD. thesis). Politecnico di Milano: Dipartimento di
Ingegneria Aerospaziale; 2004.
[19] M. Anghileri, G. Sala. Theoretical assessment, numerical simulation and compar-
ison with tests of birdstrike on deformable structures. In: 20th Congress of the
International Council of the Aeronautical Sciences, ICAS 96; 1996. Sorrento, Italy.
[20] Langrand B, Bayart A-S, Chaveau Y, Deletombe E. Assessment of multi-physics fe
methods for bird strike modeling – application to a metallic riveted airframe. Int J
Crashworthiness 2002;7(4):415–28.
[21] Hanssen AG, Girard Y, Olovsson L, Berstad T, Langseth M. A numerical model for
bird strike of aluminium foam-based sandwich panels. Int J Impact Eng
2006;32(7):1127–44.
[22] Johnson AF, Holzapfel M. Modelling soft body impact on composite structures.
Compos Struct 2003;61(1–2):103–13.
A. Riccio et al.
[23] V.K. Goyal, C.A. Huertas, T.R. Leutwiler, J.R. Borrero. Robust bird-strike modeling
based on SPH formulation using LS-DYNA. In: 47th AIAA/ASME/ASCE/AHS/ASC
Structures, Structural Dynamics, and Materials Conference, Newport; 2006.
[24] A.J. Tudor. Bird ingestion research at Rolls-Royce. In: Symposium on the me-
chanical reliability of turbo-Machinery Blading, Derby and District College of
Technology; April 1968.
[25] Nizampatnam LS. Models and methods for bird strike load predictions (PhD thesis)
Wichita State University; 2007.
[26] McCarthy MA, Xiao JR, McCarthy CT, Kamoulakos A, Ramos J, Gallard JP, et al.
Modelling bird impacts on an aircraft wing – part 2: modelling the impact with an
SPH bird model. Int J Crashworthiness 2005;10(1):51–9.
[27] Monaghan JJ. Smoothed particle hydrodynamics. Ann Rev Astron Astrophys
1992;30(1):543–74.
[28] M. Hörmann, U. Stelzmann, M.A. McCarthy, J.R. Xiao. Horizontal tail plane sub-
jected to impact loading. In: 8th International LS-DYNA Users Conference.
[29] Guida M. Study, design and testing of structural configurations for the bird strike
compliance of aeronautical components (PhD. thesis). University of Naples; 2008.
[30] Blair A. Aeroengine fan blade design accounting for bird strike. University of
Toronto; 2008.
[31] R. Ubels, A.F. Johnson, J.P. Gallard, M. Sunaric, Design and Testing of a Composite
Birdstrike resistant leading edge. SAMPE Europe Conference & Exhibition, Paris,
France; April 2003.
[32] Huertas-Ortecho CA. Robust birdstrike modeling using Ls Dyna. University of
Puerto Rico; 2006.
[33] R.A. Brockman, T.W. Held. Explicit finite element method for transparency impact
analysis. Wright Laboratory Technical Report WL-TR-91-3006; 1991.
[34] Benson DJ. Eulerian finite element methods for the micromechanics of hetero-
geneous materials: dynamic prioritization of material interfaces. Comput Methods
Appl Mech Eng 1998;151(3–4):343–60.
[35] Nakayama T, Mori M. An Eulerian Finite Element method for time-dependent free
surface problems in hydrodynamics. Int J Numer Meth Fluids 1996;22(3):175–94.
[36] Zukas JA, Nocholas T, Swift HF, Greszczuk LB, Curran DR. Impact dynamics. John
Wiley & Sons Inc; 1982.
[37] A.K. Sareen, M.R. Smith, B.R. Mullins. Applications of a nonlinear dynamics tool to
rotorcraft design problems at Bell Helicopter Textron Inc. In: 27th European
Rotorcraft Conference, Moscow; 2001.
[38] Birnbaum NK, Francis NJ, Gerber BI. Coupled techniques for the simulation of
fluid-structure and impact problems. In: Stein E, Borst R, Hughes TJR, editors.
Computer assisted mechanics. Fundamentals, vol. 1. John Wiley & Sons Ltd; 2004.
[39] Donea J, Huerta A, Ponthot J-Ph, Rodriguez-Ferran A. Arbitrary Lagrangian-
Eulerian methods. In: Stein E, Borst R, Hughes TJR, editors. Chapter 14,
Encyclopedia of computational mechanics. Fundamentals, vol. 1. John Wiley &
Sons Ltd; 2004.
[40] A.A. Amsden, C.W. Hirt. YAQUI: an arbitrary Lagrangian-Eulerian computer pro-
gram for fluid flow at all speeds. Los Alamos Scientific Laboratory Techical Report
LA-5100;1973.
[41] Benson DJ. Computational methods in Lagrangian and Eulerian hydrocodes.
Comput Methods Appl Mech Eng 1992;99(2–3):235–394.
[42] Souli M, Zolesio JP. Arbitrary Lagrangian-Eulerian and free surface methods in
fluid mechanics. Comput Methods Appl Mech Eng 2001;191(3–5):451–66.
[43] Souli M, Ouahsine A, Lewin L. ALE formulation for fluid-structure interaction
problems. Comput Methods Appl Mech Eng 2000;190(5–7):659–75.
[44] Liu WK, Belytschko T, Chang H. An arbitrary Lagrangian-Eulerian finite element
method for path-dependent materials. Comput Methods Appl Mech Eng
1986;58(2):227–45.
[45] Aquelet N, Souli M, Gabrys J, Olovson L. A new ALE formulation for sloshing
analysis. Struct Eng Mech 2003;16(4):423–40.
[46] Donea J, Giuliani S, Halleux JP. An arbitrary Lagrangian-Eulerian finite element
method for transient dynamic fluid-structure interactions. Comput Methods Appl
Mech Eng 1982;33(1–3):689–723.
[47] Benson DJ. An efficient, accurate, simple ALE method for nonlinear finite element
programs. Comput Methods Appl Mech Eng 1989;72(3):305–50.
[48] Monaghan JJ. Particle methods for hydrodynamics. Comput Phys Rep
1985;3(2):71–124.
[49] Monaghan JJ. An introduction to SPH. Comput Phys Commun 1988;48(1):89–96.
[50] Lucy LB. A numerical approach to the testing of fusion process. Astron J
1977;88:1013–24.
[51] Gingold RA, Monaghan JJ. Smoothed particle hydrodynamics: theory and appli-
cation to non-spherical stars. MNRAS 1977;181(3):375–89.
[52] Groenenboom PHL. Numerical simulation of 2D and 3D hypervelocity impact
using the SPH Option in PAM-SHOCK™. Int J Impact Eng 1997;20(1–5):309–23.
[53] Faraud M, Destefanis R, Palmieri D, Marchetti M. SPH simulations of debris im-
pacts using two different computer codes. Int J Impact Eng 1999;23(1 Part
I):249–60.
[54] Sulsky D, Chen Z, Schreyer HL. A particle method for history-dependent materials.
Comput Methods Appl Mech Eng 1994;118(1–2):179–96.
[55] Zhang X, Sze KY, Ma S. An explicit material point finite element method for hyper-
velocity impact. Int J Numer Meth Eng 2006;66(4):689–706.
[56] Johnson GR, Stryk RA, Beissel SR. SPH for high velocity impact computations.
Comput Methods Appl Mech Eng 1996;139(1–4):347–73.
[57] A. Kamoulakos, P.H.L. Groenenboom. Moving form FE to SPH for space debris
impact simulations – experience with PAMSHOCK. In: Proceedings of European
conference on spacecraft structures: materials and mechanical testing,
Braunschweig, Germany; 1998.
[58] Abaqus 6.14: Analysis User's Manual.
[59] Citarella R, Cricrì G. Three-dimensional BEM and FEM submodelling in a cracked
601
Composite Structures 202 (2018) 590–602
FML full scale aeronautic panel. Appl Compos Mater 2014;21(3):557–77.
[60] R.L. Ramkumar. Performance of a quantitative study of instability-related dela-
mination growth. NASA CR-166046; 1983.
[61] Armentani E, Citarella R, Sepe R. FML full scale aeronautic panel under multiaxial
fatigue: experimental test and DBEM simulation. Eng Fract Mech
2011;78(8):1717–28.
[62] Chai H, Knauss WG, Babcock CD. Observation of damage growth in compressively
loaded laminates – the phenomenological aspects of composite-panel-failure under
compressive in-plane loading and low-velocity transverse impact are studied ex-
perimentally via real-time recording of the failure-propagation event. Exp Mech
1983;23(3):329–37.
[63] Rhodes MD, Williams JG, Starnes Jr. JH. Low-velocity impact damage in graphite-
fiber reinforced epoxy laminates. Polym Compos 1981;2(1):36–44.
[64] Rhodes MD. Damage tolerance research on composite compression panels. NASA
CP 1980;2142:107–42.
[65] Takeda N, Sierakowski RL, Ross CA, Malvern LE. Delamination-crack propagation
in ballistically impacted glass/epoxy composite laminates – velocities of delami-
nation-crack propagation and of development of the generator strip which initiates
delamination were measured in ballistically impacted composite laminates. Exp
Mech 1982;22(1):19–25.
[66] Chai H. The growth of impact damage in compressively loaded laminates (PhD
thesis) Cal. Institute of Tech; 1982.
[67] Whitcomb JD. Approximate analysis of postbuckled through-width delaminations.
Composites Technol Rev 1982;V4(N3):71–7.
[68] Whitcomb JD. Parametric analytical study of instability-related delamination
growth. Compos Sci Technol 1986;25(1):19–48.
[69] Chai H, Babcock CD, Knauss WG. One dimensional modelling of failure in lami-
nated plates by delamination buckling. Int J Solids Struct 1981;17(11):1069–83.
[70] Kim H-J, Hong C-S. Buckling and postbuckling behavior of composite laminates
with a delamination. Compos Sci Technol 1997;57(5):557–64.
[71] Whitcomb JD. Analysis of a laminate with a postbuckled embedded delamination,
including contact effects. J Compos Mater 1992;26(10):1523–35.
[72] Whitcomb JD. Three-dimensional analysis of a postbuckled embedded delamina-
tion. J Compos Mater 1989;23(9):862–89.
[73] Orifici AC, Herszberg I, Thomson RS. Review of methodologies for composite
material modelling incorporating failure. Compos Struct 2008;86(1–3):194–210.
[74] Hashin Z. Failure criteria for unidirectional fiber composites. J Appl Mech Trans
ASME 1980;47(2):329–34.
[75] C.T. Sun. Strength analysis of unidirectional composites and laminates.
In:Comprehensive composite materials; 2000, p. 641–666.
[76] Tsai SW. Theory of composite design. Dayton: Think Composites; 1992.
[77] Davies GAO, Zhang X. Impact damage prediction in carbon composite structures.
Int J Impact Eng 1995;16(1):149–70.
[78] Abrate S. Modeling of impacts on composite structures. Compos Struct
2001;51(2):129–38.
[79] Garnich MR, Akula VMK. Review of degradation models for progressive failure
analysis of fiber reinforced polymer composites. Appl Mech Rev 2009;62(1):1–33.
[80] Matzenmiller A, Lubliner J, Taylor RL. A constitutive model for anisotropic da-
mage in fiber-composites. Mech Mater 1995;20(2):125–52.
[81] Reddy YSN, Moorthy CMD, Reddy JN. Non-linear progressive failure analysis of
laminated composite plates. Int J Non Linear Mech 1995;30(5):629–49.
[82] Olsson R, Donadon MV, Falzon BG. Delamination threshold load for dynamic
impact on plates. Int J Solids Struct 2006;43(10):3124–41.
[83] Lapczyk I, Hurtado JA. Progressive damage modeling in fiber-reinforced materials.
Compos A Appl Sci Manuf 2007;38(11):2333–41.
[84] Ochoa OO, Reddy JN. Finite element analysis of composite laminates. Kluwer
Academic Publishers; 1992.
[85] D.W. Sleight. Progressive failure analysis methodology for laminated composite
structures. NASA/TP-1999-209107; 1999.
[86] Chang F-K, Lessard LB. Damage tolerance of laminated composites containing an
open hole and subjected to compressive loadings: part I—analysis. J Compos Mater
1991;25(1):2–43.
[87] Lin W-P, Hu H-T. Nonlinear analysis of fiber-reinforced composite laminates
subjected to uniaxial tensile load. J Compos Mater 2002;36(12):1429–50.
[88] Chang F-K, Chang K-Y. A progressive damage model for laminated composites
containing stress concentrations. J Compos Mater 1987;21(9):834–55.
[89] Tan SC. A progressive failure model for composite laminates containing openings.
J Compos Mater 1991;25(5):556–77.
[90] Engelstad SP, Reddy JN, Knight NF. Postbuckling response and failure prediction
of graphite-epoxy plates loaded in compression. AIAA J 1992;30(8):2106–13.
[91] Chou S-C, Orringer O, Rainey JH. Post-failure behavior of laminates I — no stress
concentration. J Compos Mater 1976;10(4):371–81.
[92] Chou S-C, Orringer O, Rainey JH. Post-failure behavior of laminates: II — stress
concentration. J Compos Mater 1977;11(1):71–8.
[93] Dávila CG, Johnson ER. Analysis of delamination initiation in postbuckled
dropped-ply laminates. AIAA J 1993;31(4):721–7.
[94] Dávila CG. Solid-to-shell transition elements for the computation of interlaminar
stresses. Comput Syst Eng 1994;5(2):193–202.
[95] Cho M, Kim J-S. Bifurcation buckling analysis of delaminated composites using
global-local approach. AIAA J 1997;35(10):1673–6.
[96] Kim J-S, Cho M. Postbuckling of delaminated composites under compressive loads
using global-local approach. AIAA J 1999;37(6):774–8.
[97] Alesi H, Nguyen VM, Mileshkin N, Jones R. Global/local postbuckling failure
analysis of composite stringer/skin panels. AIAA J 1998;36(9):1699–705.
[98] Krueger R, O’Brien TK. Shell/3D modeling technique for the analysis of delami-
nated composite laminates. Compos A Appl Sci Manuf 2001;32(1):25–44.
A. Riccio et al.
[99] R. Krueger, P. Minguet. Skin-stiffener debond prediction based on computational
fracture analysis, NASA/CR-2005-213915.
[100] Sellitto A, Borrelli R, Caputo F, Riccio A, Scaramuzzino F. Methodological ap-
proaches for kinematic coupling of nonmatching finite element meshes. Procedia
Eng 2011;10:421–6.
[101] Liu J, Li Y-L, Xu F. The numerical simulation of a bird-impact on an aircraft
windshield by using the SPH method. Adv Mater Res 2008;33–37(Part 2):851–6.
[102] Georgiadis S, Gunnion AJ, Thomson RS, Cartwright BK. Bird-strike simulation for
602
Composite Structures 202 (2018) 590–602
certification of the Boeing 787 composite moveable trailing edge. Compos Struct
2008;86(1–3):258–68.
[103] Smojver I, Ivančević D. Numerical simulation of bird strike damage prediction in
airplane flap structure. Compos Struct 2010;92(9):2016–26.
[104] M.A. Lavoie, A. Gakwaya, M.N. Ensan, D.G. Zimcik. Review of existing numerical
methods and validation procedure available for bird strike modelling. Miami, FL,
USA; 2007, p. 111–118.