Composite Structures 96 (2013) 616–630

Contents lists available at SciVerse ScienceDirect

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

Parametric study of a SPH high velocity impact analysis – A birdstrike

windshield application
A. Grimaldi a,⇑, A. Sollo a, M. Guida b, F. Marulo b
a
Piaggio Aero Industries, Structure Department Office, Pozzuoli (NA), Italy
b
University of Naples ‘‘Federico II’’, Department of Aerospace Engineering, Napoli, Italy

a r t i c l e i n f o a b s t r a c t

Article history: During its life cycle an aircraft flies on the risk of impacting foreign objects. According to the aeronautical
Available online 27 October 2012 specifications, the term ‘‘birdstrike’’ indicates the collision between a bird and an aircraft front facing
component, which includes windshield, nacelles, compressor blade, wing leading edge or even flaps,
Keywords: when extended.
SPH technique This paper is part of a research focused on the study, with the help of finite element analysis, of an air-
High-velocity impact craft windshield-surround structure with an innovative configuration that satisfies the bird-strike
Birdstrike
requirement according to the European and US aviation regulations 25.631 on the ‘‘Bird-strike Damage’’
Glass composite material
Windshield component
[1,2].
Firstly the paper provides a numerical analysis of a simplified, but dimensionally realistic, square flat
windshield model subjected to impact by a 1.8 kg bird model at 155 m/s with an impact angle of 90°. The
FE-SPH coupled approach is used to simulate the birdstrike by using the explicit finite element solver
code LS-Dyna.
The second step involves a parametric analysis on the square model to estimate the influence of the
target geometry, the impact angle, and the plate curvature on the impact response of the windshield
structure. The goal of these numerical simulations is the evaluation of the windshield capability to absorb
the impact energy of a birdstrike event, in a safe and efficient way which is compliant with the airwor-
thiness rules.
The aim of this work is to define a scientific and methodological approach to the study of the birdstrike
problem. The collection of results and experiences achieved by the previous simplified realistic model can
be applied to perform a certification test simulation and define possible guidelines for the design of a
bird-proof airplane windshield.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction investigated the bird impact on an aircraft wing leading edge struc-
ture using PAM-CRASH software. They showed that the SPH meth-
The high probability that a bird impacts on an aircraft front fac- odology was able to capture the breakup of the bird into debris
ing component (e.g. windshield, nacelles, wing components and particle after the collision. Guida et al. [5], found that the Lagrang-
compressor blade) during its operative life, poses important prob- ian–SPH combination provided the best results in terms of impact
lems in aircraft design. visualisation and a good prediction of the deceleration of the pro-
The probability of an accident is higher in the airport area dur- jectile, compared to the test results.
ing the take-off and landing phases, and especially in the early The birdstrike in addition to involve several front facing compo-
morning and late afternoon. In recent years the severity and nents of the aircraft, will concern different kind of materials too. In
importance of the birdstrike has grown because of the remarkable the recent years many authors analysed and simulated the bird-
increase of the air traffic and airplane performances in term of strike event against a specific part of the airplane. Georgiadis
velocity, followed by an increment of energy density and impulsive et al. [6], provided a validated simulation methodology to support
loads during the impact. the birdstrike certification of the carbon fibre epoxy composite
Other authors used SPH approach to model the birdstrike phe- moveable trailing edge of the Boeing 787 Dreamliner. Anghileri
nomenon, in particular Ubels et al. [3], and McCarthy et al. [4], et al. [7], performed numerical simulation of the bird impact
against the intake of turbofan engine and discussed the advantages
⇑ Corresponding author. and disadvantages about using of different bird modelling
E-mail address: [email protected] (A. Grimaldi). techniques. Hanssen et al. [8], investigated bird impact against

0263-8223/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.compstruct.2012.09.037
 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630 617

Table 1 from flying windshield fragments due to bird impact’’, the regula-
FAR – birdstrike test requirements. tions state.
Aircraft component Bird weight (lb) FAR section In the case of the newest aircraft family being under develop-
Windshields and frames 4 25.775 (b), 25.775 (c) ment at Piaggio Aero, the route dedicated to a study of a new con-
Wing leading edges 4 25.571 (e) (l) figuration keeping a single piece laminated glass windshield is
Empennage leading edges 8 25.631, 25.571 (e) (l) maintained, so that the designers are focusing in prevent breaking
Engine – inlet lip 4 25.571 (e) (l) of glass from entering the flight deck during a birdstrike event. In
Engine – fan integrity 4 33.77, 25.571 (e) (l)
addition the design of the cockpit is aimed to prevent loss and/of
limited visibility for both pilots in case of birdstrike. From the early
aluminium foam-based sandwich panel using the ALE approach. stage of the design development, big attention has been paid to the
Guida et al. [9], developed a finite element model for simulating nose design in order to maintain the same Piaggio Aero P180 brand
the birdstrike test on the tailplane leading edge structure. They for the unique Front Fuselage Shape and the single windshield has
studied the impact on a composite component made of aluminium been identified as one of the key product differentiators compared
skin and flexcore as interior sandwich structure, finding good cor- to other business aircraft competitors.
relation between the numerical and experimental results. Both Federal Aviation Administration (FAA) and European Avia-
The windshield structure plays a key role relating to the bird- tion Safety Agency (EASA) list regulations for the aircraft certifica-
strike problem, because of its exposure in front of the airplane. tion process to ensure that each front facing aircraft components
So many authors developed study in order to design bird-proof should be capable of withstanding birdstrike at critical flight speed
windshield for protecting the safety of aircraft. Yang et al. [10], to a certain degree (Table 1).
elaborated an experimental and FEM of windshield subjected to For the final certification both FAA and EASA require full-scale
high-speed bird impact. Liu et al. [11], focused on the analysis of tests to demonstrate the efficiency of every bird-proof structures,
an effective numerical method to simulate bird impact aircraft like a windshield. Because of the excessive costs necessary to the
windshield events, using the SPH approach and the explicit finite execution of the test, the manufacturers perform many numerical
element program PAM-CRASH. Recently Salehi et al. [12], investi- analysis of the birdstrike event with explicit non-linear finite ele-
gated the effect of the birdstrike on different aircraft windows both ment (FE) codes in order to identify the most critical impacting
numerically and experimentally. They studied structures made up locations, to decrease the number of destructive tests required
of different geometries and materials by using various modelling and to avoid any delay during the certification.
approach (ALE and SPH).
The birdstrike is not only relative to the flight safety, but also to 2. Modelling technique
not negligible maintenance costs, which the companies must meet
to repair possible damages in case of an accident. In order to better Explicit FE analysis is a numerical technique used in case of
understand the nature of birdstrike and also prevent the hazard of highly non-linear behaviour of materials with inelastic strains,
an accident, they have been formed international committees, such high strain rates and large deformations, such as it occurs during
as the Birdstrike Committee, USA. Only in the United States more a birdstrike. For a birdstrike phenomenon, to obtain a good predic-
than 60,000 wildlife strikes to civil aircraft have been reported be- tion of the impact loads and damage of an aircraft structure under
tween 1990 and 2005 Clearly et al. [13], and about 4500 birdstrikes impact loading, it is essential to adopt a realistic material model for
were reported by the U.S. Air Force in 2011. a bird and its associated material and geometrical parameters. To
Therefore more and more companies and government authori- achieve an accurate bird model an explicit code, like LSTC/LS-Dyna,
ties have initiated advanced research and development programs offers different approaches: Lagrangian, Arbitrary Lagrangian Eule-
to ensure that every structural part of an aircraft is able to with- rian (ALE) and Smoothed Particle Hydrodynamics (SPHs) approach.
stand the loads due to a high velocity impact and at least guarantee The first and the third method are based on the Lagrangian fi-
the safe landing of the airplane, in according to the International nite element formulation, with the difference that in case of SPH
Certification Standards. approach the bird is modelled with a meshless technique, in which
For the windshield, manufacturers must either show by analysis the elements are a set of discrete and mutually interacting nodes.
or tests that the likelihood of the glass fragmenting as a result of Instead in the ALE method the bird material flows relative to an
the impact and hitting the pilots during the event is ‘‘of low order’’ Eulerian mesh and the impacting loads are transferred to the
or they must ‘‘have a means to minimise the danger to the pilots Lagrangian mesh (the impacted structure) through an ALE coupling

Fig. 1. Discretization of volume: (a) FEM and (b) SPH.

618 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630

Fig. 5. Laminated glass.

Fig. 2. Domain of the sphere in the SPH method.

Fig. 6. Windshield lay-up configuration.

useful to couple different numerical approach in order to treat each

domain of the problem more appropriately.
For this study it has been used the SPH modelling approach be-
cause of its suitability to model the flow of fluid, such as the bird.
The Smooth Particle Hydrodynamics was developed by Mona-
ghan in the late 1970s for astrophysics problems with application
to hypervelocity impacts (10 km/s) Monaghan [15]. It is both
effective and accurate at modelling material deformation as well
as adaptable in terms of specific material models and besides to
solve computational fluid dynamic problems, it can be also applied
for continuum mechanics problems with large deformations, as
crash simulations.
The main difference between the classical FE method and SPH is
the methodology of discretization of the model, as shown in Fig. 1.
In the SPH formulation the fluid is represented as a set of mov-
Fig. 3. Flat windshield model. ing particles, each one representing an interpolation point, where
all the fluid properties are known. The estimation of the field vari-
ables at any point is allowed by an interpolation formula, called
kernel function.
In the definition of the methodology a very important role is
played from the neighbour search procedure, i.e. the evaluation
step-by-step of which particle will interact with the others. This
influence of each particle is established inside of a sphere of radius
of 2 h, called support domain Xh, where h is the smoothing length,
as shown in Fig. 2. The smoothing length of every particle changes
with the time. When particles separate the smoothing length in-
creases, while when they come close to each other, the smoothing
length decreases accordingly. It is necessary to keep enough parti-
cles in the neighbourhood to validate the approximation of contin-
uum variables.
Because of the gridless nature of the methodology, the SPH does
not suffer from the usual disadvantage relative to mesh tangling in
large deformation problems, like a pure Lagrangian formulation,
and uses fewer elements than the ALE method, avoids the material
Fig. 4. Cylindrical bird model. interface problems associated with it. Again it allows the modelling
of fragmentation and fracture phenomena, and, as well as the
Lagrangian formulation, the SPH method allows a good tracking
algorithm. Each method presents both advantages and disadvan- of the material deformation and history. Nevertheless it remains
tages, so the choice of the more appropriate approach for the mod- some disadvantages about the stability and consistency of the
elling is often connected to the experience of the user, as well as methodology. One disadvantage is the computationally demanding
the nature of crash phenomenon. of the method, both in memory and in CPU time. This can be over-
There is no a unique ‘‘best’’ numerical method to analyse the come using a parallel analysis with more than one CPU. Further-
impact phenomenon, and in particular the fluid–structure interac- more another disadvantage can be the difficulty of establishing
tion problems. During a single impact analysis it could be often the boundary conditions with a loss of smoothness and accuracy.
 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630 619

Table 2
Glass properties.

Density Poisson ratio Young modulus Failure strain

3 10 2
q = 2500 kg/m m = 0.22 E = 6.895e N/m ef = 0.35%

Fig. 9. Windshield-surround installation – Geometry.

penetration, but also avoid a complete fragmentation of all trans-

parencies, so that to ensure a sufficient vision through at least
one panel to permit continued safe flight and landing. This specifi-
cation could be critical in the case of the windshield composed by
only two panes because an impact on a transparent could cause a
damage on the other panel too, or again an impact on the centre
beam, which divides the two panels, could create a fragmentation
of both transparencies at the same time with following loss of
Fig. 7. Glass r–e curve.
visibility.
Before studying a full-scale model of the real windshield struc-
Table 3 ture, it has been carried out a preliminary parametric analysis of
PVB-interlayer properties. the birdstrike phenomenon through a series of numerical simula-
Density q = 2500 kg/m3
tions on a simplified but representative structure, shown in
Short-time shear modulus G0 = 0.33e9 N/m2 Fig. 3. This simplified model is a square flat plate made up of the
Long-time shear modulus G1 = 0.33e9 N/m2 same materials and lay-up configuration of the full-scale structure.
Bulk modulus K = 2.0e6 N/m2 This parametric analysis was useful to identify the most impor-
Decay coefficient b = 12.6 s1
tant parameters that affect the impact response of the target struc-
Failure strain 175%
ture in case of bird impact and to give the chance for the definition
of some best guidelines for the design and optimisation process of
bird-proof windshield structure.
About the projectile configuration, in according with the Certi-
fication Requirements, the weight of the bird is W = 1.8 kg, the
density is q = 950 kg/m3, and the geometric model is approximated
as a right circular cylinder, as shown in Fig. 4:
where
sffiffiffiffiffiffiffiffi
3 2W
D¼ ¼ 0:106 m ð1Þ
pq
and
L ¼ 2D ¼ 0:212 m ð2Þ

3.2. Mechanical behaviour of materials

Fig. 8. PVB shear modulus function. In order to comply with fail-safe requirements, an aircraft wind-
shield is built by laminated glass. Laminated glass is a type of
safety glass used in different fields of the industrial applications,
3. Bird and target description e.g. in automotive structures, as well as aerospace, and permits
to avoid serious injuries of the passengers in case of an impact
3.1. Geometry event.
The basic construction of a laminated glass involves two panels
This section presents a geometry description of the target struc- of glass joined to a polyvinyl-butyral (PVB) interlayer, as shown in
ture onto which the projectile (bird) impacts. The application stud- Fig. 5. In the case of an impact, such as the birdstrike, the splinters,
ied in this work is part of study relative to a birdstrike event caused by the glass failure, and remain connected to the PVB-
against a windshield structure of a business jet airplane. interlayer.
The International Certification Standards prescribe a windshield In a short-time dynamic event the glass and PVB layers work in a
structure must, not only, withstand to the birdstrike event without different way. For small deformations the glass determines the elastic
620 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630

Fig. 10. Square plate FE model.

Table 4
Thickness distribution of the glass laminated model. material. Fig. 7 shows the stress strain curve, the ultimate tensile
Outer glass Outer PVB Middle glass Inner PVB Inner glass
stress is equal to rupture stress and yield stress at the same time
because of the brittleness of the material.
4.0 mm 4.0 mm 4.0 mm 4.0 mm 4.0 mm
The plastic material is the polyvinyl-butyral (PVB) and it is
interposed between two layers of glass. It has typical characteris-
tics of a viscoelastic interlayer. This type of material shows good
characteristics of strength and transparency, besides allowing a
high deformation before the failure and a good tearing strength.
Table 3 shows standard literature characteristics of the PVB
material.
The interlayer gives special impact strength to the glass, which
can absorb a part of the impact energy thanks to its deformation.
Furthermore it avoids the fragmentation of glass by its adhesive
property, which could be very dangerous for the occupants of the
aircraft. The interlayer is modelled as a linear viscoelastic material
and the shear law is described in the following equation:
GðtÞ ¼ G1 þ ðG0 þ G1 Þebt ð3Þ
The plot of this function is shown in Fig. 8.
Fig. 11. Bird SPH model.
3.3. Boundary Conditions
behaviour of the composite, while for large deformations it is not able
to withstand load anymore; in this case the PVB-interlayer plays a Fig. 9 shows a section of the windshield-surround installation. A
key role, because still has a load-carrying capacity left. key role is played by the joint technique used to stick together the
The glass laminate studied in this work consists of three plies of layers of the glass laminate. In fact the choice of a method, rather
glass and two plies of a polyvinyl-butyral (PVB) interlayer. than the bolts (i.e. mechanical clamp), can appreciably affect the
In this research different thickness lay-up configurations have performances of the structure in case of both static and dynamic
been investigated in order to obtain some design guidelines for load application. The plug installation defines this method of clamp
the optimisation of the thickness structure lay-up. For instance the windshield with a surround without bolts and joints.
Fig. 6 shows a configuration for which the thickness of each glass
layer is equal to PVB-interlayer one. 4. Numerical strategies
The glass is an amorphous, transparent and isotropic material,
with characteristics of high brittleness if subjected to both dy- 4.1. Windshield FE modelling
namic and static loads. The most prevalent type of glass is soda-
lime glass, mainly made up of silica and several minor components. In this work the commercial explicit FE software LS-Dyna was
Table 2 shows the glass mechanical properties. used for all numerical simulations Hallquist [14]. It was carried
The glass is considered as an elastic–plastic material with an out a preliminary parametric numerical analysis of a simplified,
infinitely small plastic part of the r–e curve typically of a brittle but realistic, structure subjected to the birdstrike phenomenon.

Table 5
Tabulated equation of state.

1 2 3 4 5 6 7 8 9 10
eV 0.000 0.105 0.118 0.128 0.137 0.154 0.169 0.183 0.195 0.217
C (Pa) 0.000 2.37e8 4.25e8 5.86e8 7.27e8 9.72e8 1.18e9 1.37e9 1.54e9 1.84e9
A. Grimaldi et al. / Composite Structures 96 (2013) 616–630 621

Fig. 12. Windshield-surround installation – FEM.

Fig. 13. Birdstrike vs simplified square panel.

Fig. 14. Impact scenario at t = 0.001 s.

622 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630

Fig. 15. Impact scenario at t = 0.004 s.

t = 0.001 sec t = 0.004 sec

Fig. 16. Failure propagation on the outer glass layer.

t = 0.001 sec t = 0.004 sec

Fig. 17. Failure propagation on the middle glass layer.

t = 0.001 sec t = 0.004 sec

Fig. 18. Failure propagation on the inner glass layer.

This simplified model is a 1 m  1 m square flat plate made up of KINEMATIC Card. The PVB-interlayer ply was modelled by a
the same materials and lay-up configuration of a real windshield. viscoelastic material law using the LS-Dyna MAT-VISCOELASTIC
The square plate windshield is a glass laminate structure, com- Card.
posed of three layers of glass and two layers of PVB-interlayer. In
Fig. 10 is shown an exploded view of the FE target square model. 4.2. Bird SPH modelling
The numerical model consists of five plies and each of these was
modelled by default 3D eight-nodes brick elements. The element The SPH approach was used to model the bird. The numerical
size was approximately 5 mm and the mesh is uniform throughout model consists of 28,620 SPH nodes with an average distance be-
the thickness. Table 4 reports the thickness of each layer of the tween two nodes of 5 mm, as shown in Fig. 11.
glass laminate. In according to the International Certification Standards every
The glass ply was modelled by a bilinear elastic–plastic material simulation is performed with a bird impact velocity equal to
law with negligible plastic region using the LS-Dyna MAT-PLASTIC- 155 m/s.
 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630 623

Fig. 19. Centre node displacement.

Fig. 20. Total internal energy for each layer.

Fig. 21. Total hourglass energy.

The appropriate substitute bird has a density of 950 kg/m3 and a P ¼ CðeV Þ ð4Þ
porosity of 10%, i.e. it is composed of 90% of water and 10% of air.
where eV is the volumetric strain given by the natural logarithm of
Regardless of the modelling method chosen, a fluid dynamic
the relative volume V. The values of the equation of state parame-
material model was used for the bird modelling. In particular it
ters are shown in Table 5 as reported in paper [5].
was defined a material constitutive model to relate stress to strain
and an equation of state to relate pressure to volume. 4.3. Surround FE modelling and boundary conditions
The MAT-NULL Card calculates the pressure p from a specified
tabulated equation of state (Eq. (4)), defined by the LS-Dyna EOS- An important part of this analysis is the modelling of the sur-
TABULATED Card round structures and the definition of its interaction with the
624 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630

Fig. 22. Energy balance check.

Fig. 23. Curvature changes.

Fig. 24. Impact angle changes.

windshield. Fig. 12a and b shows the FE model of the C-beam that in the previous sections, a cylinder with length-to-diameter ratio
surrounding the square flat plate model and its section, whereas of 2 was predefined for the current 1.8 kg (4 lb) bird model, with
Fig. 12c shows the section of a typical windshield-surround a diameter of 0.106 m. This soft projectile travelling with an inci-
installation. dent velocity of 155 m/s striking the 1 m  1 m flat target with
As well as for the laminate layers the surround structure was and impact angle of 90° respect to the impact axis.
modelled by eight-nodes brick elements. For the interaction be- Every analysis was performed on a 8-processor HP Workstation
tween surround and laminate, it has not been considered the pres- machine with a WIN64 version of LS-Dyna3D (V971), taking
ence of the bolts, but a simple plug approach, so the boundary approximately 12 h of wall clock time to complete the numerical
conditions are defined fixing every translational and rotational de- analysis. It was set the output interval at 50 ls for a total event
gree of freedom of the surround contour. time of 0.005 s, with producing of an output database of more than
4 GB size.
5. Results and discussion Fig. 14 shows a plots about the deformation behaviour of the
impacted panels at t = 0.001 s from the beginning of the impact
5.1. Birdstrike vs square windshield model phenomenon. During the first 2 ms it is happens the transfer of
the almost 90% of the bird load to the plate, with a consistent
This section presents the numerical results of the bird impact deformation of the centre of the model, correspondent to the bird
on the square plate windshield, as shown in Fig. 13. As mentioned impacted zone.
 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630 625

Fig. 25. Case studied as changes of curvature and impact angle.

Instead during the following 3 ms it can see an expansion of the nal energy, Ehg the hourglass energy, Esi the sliding interface energy,
plate deformation from the centre zone to the plate edges, because Erw the rigid wall energy, and Edamp is the damping energy.
of the presence of the PVB-interlayer, as shown in Fig. 15. Further- In this birdstrike analysis there is not rigid wall and damping
more it is clearly visible the deformation of the bird during the im- energy, the sliding energy is negligible and the initial energy is only
pact, and its squashing into the windshield is adequately simulated represented by kinetic energy of the bird, given by
by the SPH modelling approach.
1
Figs. 16–18 show the failure propagations of the each layer of Ekino ¼ mv2 ¼ 21622:5 J ð6Þ
the glass laminate. About the glass plies it can already see a crack 2
initiation after just 1 ms, and a propagation of this from the centre The time-history of the internal energy for each layers of the
to the angles of the plate. laminated glass is reported in Fig. 20. The outer glass, on which
The failure of the glass layers is more noticeable for the inner the bird impacts directly, is the ply that absorbs the bigger amount
glass layer, because of more intensive bending. Both the outer of the bird impact energy, but contrarily to what you might think,
and inner PVB-interlayer ply do not present any failure. after that the two PVB layers turn out to be very efficient about the
In Fig. 19 the resultant displacement for the central element energy absorbing, because of their plastic behaviour.
(the element of the impact) of the plate is reported. At the end of The time-history of the hourglass energy in function of time is
the simulation it presents a maximum value of ’67 mm. reported in Fig. 21, it is equal to 10% of peak of the total internal
In order to perform a check on the analysis, it can be useful to energy and therefore it may be considered acceptable for the
carry out a balance of the energy data. The following Eq. (5) should analysis.
hold at all times during an analysis: The check on the total energy, shown in Fig. 22, ensures the
accuracy of the results obtained by the birdstrike numerical analy-
Ekin þ Eint þ Ehg þ Esi þ Eraw þ Edamp ¼ Ekin0 þ Eint0 þ Eext ð5Þ
sis performed. It has to be taken into account that the internal en-
where Ekin0 is the initial kinetic energy, Eint0 the initial internal en- ergy shown in the previous plot is associated to the whole glass
ergy, Wext the external energy, Ekin the kinetic energy, Eint the inter- laminate, sum of the energy associated to each layer, to the sur-
626 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630

was identified between a number of 27 different impact scenarios

in order to define the behaviour of the windshield in term of en-
ergy absorbing and failure.

5.2.1. Effect of the curvature

This section describes the effect of the curvature on the impact
energy absorbed from windshield. Fig. 27 shows the maximum val-
ues of the energy transferred from bird to windshield during the
impact as curvature changes for an impact angle of 90°, and for
each thickness lay-up configuration.
It can see that the amount of the energy of impact remains al-
most constant going from a flat plate (blue1 bar) through a curva-
ture radius of r = 1.273 (red bar), while it decreases by 10% going
from a curvature radius of r = 1.273 (red bar) through a curvature ra-
dius of r = 0.636 (green bar), in addition the absorbing energy de-
crease with decreasing of the PVB thickness.
It can also see similar trends of the energy as curvature changes
in the case of impact with an impact angle of 60° (Fig. 28) and 30°
(Fig. 29), with the difference that only for the impact angle of 30°
there is no failure of any glass layer.

5.2.2. Effect of the impact angle

In this section is presented the variation of the absorbing energy
as the impact angle changes. Fig. 30 shows the maximum values of
the energy for a flat plate and for each thickness lay-up configura-
tion. It can see that the amount of the energy of impact decreases
by 40–50% going from an impact angle of 90° (blue bar) through
60° (red bar), and again decreases even by 70–80% going from an
Fig. 26. Thickness lay-up configurations. impact angle of 60° (red bar) through 30° (green bar).
Figs. 31 and 32 report similar trends of the energy as curvature
round deformation, and to the ‘‘deformation’’ of the bird, or to be changes in the case of impact against a windshield with a curva-
more exact associated with its squashing onto the panel. ture r = 1.273 and r = 0.636, the only difference is a light decrease
of the absolute values with increase of the curvature. Again in this
5.2. Parametric analysis case only the impacts with an angle of 30° do not present any fail-
ure in the glass layers.
A parametric analysis was carried out to investigate the influ-
ence of various geometric parameters on the impact response of 5.2.3. Effect of the thickness lay-up configuration
the windshield, and in particular respect to the capability to absorb In this section is presented the variation of the energy as the
the impact energy in a safe way without any failure. glass-PVB thickness ratio changes. Fig. 33 shows the maximum val-
For the parametric study, the influence of the following three ues of the energy for a flat plate and for each impact angle. It can
parameters: see that the amount of the energy of impact decreases by 20–
(1) Windshield Curvature, (2) Impact Angle and (3) Glass-PVB 30% going from ratio of 1 to 1 (blue bar) through to a ratio of 1
Thickness Ratio. to 2, while it remain constant going from ratio of 1 to 2 (red bar)
The first parameter investigated was the curvature of the wind- through to a ratio of 1 to 3 (green bar).
shield. In addition to the case of a square flat plate, analysed in pre- Figs. 34 and 35 report similar trends of the energy as curvature
vious section, the birdstrike analysis was also performed onto two changes in the case of impact against a windshield with a curva-
further types of windshield with a radius of curvature of 1.273 m ture r = 1.273 and r = 0.636. Again in this case only the impacts
and 0.636 m, as shown in Fig. 23. with an angle of 30° do not present any failure in the glass layers.
The second parameter analysed was the impact angle of the
bird respect to the horizontal x-axis. The simulation was performed 6. Conclusions
for three impact angles: a = 90°, a = 60°, and a = 30°, as it was re-
ported in Fig. 24. The goal of this work was the development a scientific and meth-
Combining this two parameters it was obtained the matrix of odological approach to the study of the birdstrike problem for the
nine analysis presented in Fig. 25, i.e. for each curvature they were design, verification, and optimisation of a bird-proof windshield of
performed three simulations correspondent to the three impact an airplane. The SPH approach results to be the most suitable and
angles. feasible methodology to simulate the dynamics of an high speed im-
The third parameter analysed was the thickness ratio between pact phenomenon, like the birdstrike against an aircraft component.
the glass and PVB layers, with a constant total thickness equal to Both bird SPH and target FE model were prepared by the LS-Pre-
20 mm. In detail they were studied three type of lay-up configura- Post pre-processor software, while every numerical simulation was
tions for which: (a) glass and PVB layers have the same thickness, performed by using LSTC/LS-Dyna explicit solver.
(b) the thickness of the glass is twice as much as the PVB one, and A preliminary validation of the birdstrike methodology was
(c) the thickness of the glass is three times the PVB one, as shown achieved through a simulation on a simplified, but representative,
in Fig. 26.
Each of these three thickness lay-up configurations was applied
to the nine previous cases obtained from the combination of 1
For interpretation of colour in Figs. 27, 30 and 33, the reader is referred to the web
curvature and impact angle, and finally the best case scenario version of this article.
 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630 627

Fig. 27. Maximum energy as curvature changes for a = 90°.

Fig. 28. Maximum energy as curvature changes for a = 60°.

Fig. 29. Maximum energy as curvature changes for a = 30°.

windshield structure. It was studied a simplified flat square wind- Secondly a parametric analysis was executed on the previous
shield model impacted by a 1.8 kg bird model at 155 m/s with an square model to evaluate the effects on its structural response
impact angle of 90°. Numerical results for this pane showed that of: (1) the target geometry, (2) the impact angle, and (3) the plate
it would not withstand the bird impact under the conditions stated curvature. The aim of this analysis was to evaluate the capability of
in the CS 25 Standards. windshield to absorb the impact energy, involving during a bird-
628 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630

Fig. 30. Maximum energy as impact angle changes for r = 1.

Fig. 31. Maximum energy as impact angle changes for r = 1.273.

Fig. 32. Maximum energy as impact angle changes for r = 0.636.

strike event, in a safe and efficient way without any catastrophic strike, it is preferable to have a windshield structure with an im-
failure of the windshield or total loss of visibility by pilots. pact angle smaller than 30°.
The results of the parametric analysis reveal that: (1) the energy It is also clear that the choice of this angle cannot to be only
transferred to the windshield during impact is strongly dependent function of birdstrike requirements, because further factors, like
of the impact angle, and (2) in order to design a structure capable for example optical problems, come into play in the design of a
to absorb safely the energy of impact involved during the bird- windshield structure.
 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630 629

Fig. 33. Maximum energy as thickness lay-up changes for r = 1.

Fig. 34. Maximum energy as thickness lay-up changes for r = 1.273.

Fig. 35. Maximum energy as thickness lay-up changes for r = 0.636.

The main achievement of this work was the collection of results and [2] Federal Aviation Administration. Dept. of transportation, ‘‘Bird Strike Damage’’.
Part 25 Airworthiness Standards: Transport Category Airplanes, Sec. 25.631,
experiences to define a ‘‘rule of thumb’’ assessment for the design of an
Washington, DC; 2003.
airplane windshield structure able to withstand to a birdstrike event in [3] Ubels L, Johnson A, Gallard J, Sunaric M. Design and testing of a composite
according with the conditions stated in the CS 25 standard requirements. birdstrike resistant leading edge. Tech. rep., National Aerospace Laboratory
NLR, Amsterdam, The Netherlands; 2003.
[4] McCarthy MA, Xiao JR, McCarthy CT, Kamoulakos A, Ramos J, Gallard JP, et al.
References Modelling of birdstrike on an aircraft wing leading edge made from fibre metal
laminates – Part 2: modelling of impact with SPH bird model. Appl Compos
[1] European Aviation Safety Agency. CS-25.631 Bird strike damage – certification Mater 2004;11:317–40.
specifications and acceptable means of compliance for large aeroplanes. Annex [5] Guida M, Marulo F, Meo M, Grimaldi A, Olivares G. SPH–Lagrangian study of
to ED decision 2011/004/R. bird impact on leading edge wing. Compos Struct 2011;93:1060–71.
630 A. Grimaldi et al. / Composite Structures 96 (2013) 616–630

[6] Georgiadis S, Gunnion AJ, Thomson RS, Cartwright BK. Birdstrike simulation for [11] Liu J, Li Y, Xu F. The numerical simulation of a bird-impact on an
certification of the boeing 787 composite movable trailing edge. Compos Struct aircraft windshield by using the SPH method. Adv Mater Res 2008;33:
2008;86:258–68. 851–6.
[7] Anghileri M, Castelletti LML, Mazza V. Birdstrike: aapproaches to the analysis [12] Salehi H, Ziaei-Rad S, Vaziri-Zanjani M-A. Bird impact effects on different types
of impacts with penetration. In: International conference on impact loading of of aircraft bubble windows using numerical and experimental methods. Int J
lightweight structures; 2004. Crashworthiness 2010;15:93–106.
[8] Hanssen A, Girard Y, Olovsson L, Berstad T, Langseth M. A numerical model for [13] Clearly EC, Dolbeer RA, Wright SE. Wildlife strike to civil aircraft in the United
birdstrike of aluminium foam-based sandwich panels. Int J Impact Eng States 1990–2005. Tech. report – FAA, Office of Airport Safety and Standards;
2006;32:1127–44. 1994.
[9] Guida M, Marulo F, Polito T, Meo M, Riccio M. Design and testing of a fiber [14] Hallquist JC. LS-Dyna Theory Manual v1.0. Livermore Software Technology
metal laminate birdstrike resistant leading edge. J Aircraft 2009;46:2121–9. Corporation; 2006.
[10] Yang J, Cai X, Wu C. Experimental and fem study of windshield subjected to [15] Monaghan JJ. Smoothed particle hydrodynamics. Ann Rev Astron Astrophys
high speed bird impact. Acta Mech Sin 2003;19:543–50. 1992;30:543–74.