Composite Structures 94 (2011) 15–23

Contents lists available at SciVerse ScienceDirect

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

Hybrid approach in bird strike damage prediction on aeronautical

composite structures
D. Ivančević, I. Smojver ⇑
Faculty of Mechanical Engineering and Naval Architecture, University of Zagreb, I. Lučića 5, HR-10000 Zagreb, Croatia

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

Article history: This paper deals with the problem of numerical prediction of bird strike induced damage on aeronautical
Available online 10 August 2011 structures. The problem of soft body impacts has been tackled by applying a hybrid Eulerian Lagrangian
technique, thereby avoiding numerical difficulties associated with extensive mesh distortion. Eulerian
Keywords: modeling of the bird impactor resulted in a more realistic behavior of bird material during impact, which
Composite structures has lead to an enhanced response of the impacted structure. The work presented in this paper is focused
Impact damage on damage modeling in composite items of aeronautical structures. The bird impactor model and damage
Bird strike
modeling approaches have been validated by comparison with experimental gas gun results available in
Aeronautical structures
the open literature, while the complete damage prediction procedure has been demonstrated on a
complex airplane flap structure finite element model.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction CEL application is the improved realistic behavior of the bird

The work presented in this paper presents improvements in the
constitutive and damage modeling of the bird strike simulation
presented in [1–3]. Lagrangian impactor modeling presents an effi-
cient technique for numerical modeling of bird deformation behav-
ior as demonstrated in [1]. Excessive distortion of Lagrangian finite
elements can in some impact conditions lead to numerical difficul-
ties, as stated in [2], where the Lagrangian bird model has been re-
placed by a hybrid Lagrangian Eulerian technique. The application
of hybrid methods in bird strike analyzes allows the bird to be
modeled as Eulerian material, while the impacted structure is rep-
resented by traditional Lagrangian finite elements. Eulerian mate-
rial motion is independent of finite element motion, thereby
avoiding numerical difficulties associated with extreme mesh dis-
tortion. Application of hybrid techniques is particularly suited to
bird strike modeling as very large deformation of the impactor is
expected.
The explicit dynamic finite element analyzes illustrated in this
work have been performed using Abaqus/Explicit and the hybrid
Coupled Eulerian Lagrangian (CEL) method. The Eulerian model
in CEL analyzes is usually represented by a stationary cube con-
taining Eulerian finite elements. Abaqus provides multi-material
EC3D8R volume elements to model Eulerian problems, which
may be completely or partially occupied by the Eulerian material
[4]. In addition to numerical stability, an important outcome of
⇑ Corresponding author.
E-mail address: ismojver@fsb.hr (I. Smojver).
URL: http://www.aerodamagelab.fsb.hr
impactor upon impact, since large deformation and even disinte-
gration of the material is possible without causing numerical diffi-
culties, as stated in [2].
The Eulerian material of the bird model is able to interact with
the Lagrangian finite element model, which in this work presents a
large airliner flap structure. The bird material boundary does not
have to match element geometry at any time during the analysis
and has to be recomputed in each time increment as the material
flows through the mesh. Coupling between Eulerian and Lagrang-
ian meshes is introduced by an extension of the general contact
algorithm. The contact is created between Lagrangian mesh sur-
faces and Eulerian material surfaces, which are automatically com-
puted and tracked during the analysis. Abaqus, like most of the
commercial FE codes, uses penalty contact algorithms to introduce
coupling between Eulerian and Lagrangian instances, as this ap-
proach uses the simplest computational level and increases robust-
ness, as described in [5].
The size of the volume enclosing Eulerian elements must be suf-
ficiently large to prevent loss of material during the analysis. The
loss of material leads to a loss of kinetic energy and could under
some conditions lead to numerical instabilities. An important
restriction is placed on the mesh size of the Eulerian finite element
model. A very fine mesh of the Eulerian grid is necessary to effi-
ciently capture the contact between Eulerian material surfaces
and Lagrangian elements in order to prevent physically unaccept-
able penetration of the bird impactor through the Lagrangian finite
element mesh.
The capability of the described damage prediction procedure
introduced in [1–3] has been extended by employing new material

0263-8223/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruct.2011.07.028
16 D. Ivančević, I. Smojver / Composite Structures 94 (2011) 15–23
and damage models using the Abaqus user material subroutine
VUMAT. The application of this subroutine enabled implementa-
tion of strain rate effects for Aluminum alloys and Puck’s phenom-
enological damage model for composite materials. Additionally,
the Mie–Grüneisen equation of state has been replaced by the
polynomial equation of state consequently enabling inclusion of
porosity effects for the bird impactor.
2. Bird modeling
Theoretical analysis of the bird strike introduces some assump-
tions on the bird material constitutive behavior which allow
hydrodynamic theory to be applied for the bird impact [6]. As
the strength and viscosity in this approach are neglected, a simple
pressure vs. density equation of state can be used to describe the
constitutive behavior of the bird material in numerical analyzes.
Theoretical and numerical validations of bird impactor geometry
suggest that a hemispherical cylinder with a length to diameter ra-
tio equal to two, enclosing a material having the density of 950 kg/
m3, best resembles a real bird during impact [6,7].
Numerical bird material models have the properties of a water
and air mixture, as real birds mostly consist of water and trapped
air inside internal cavities. The constitutive behavior of fluid-like
materials is governed by hydrodynamic material models, whose
volumetric strength is calculated using equations of state (EOS).
An incomplete equation of state, which does not cover heat con-
duction effects, determines values of the hydrostatic pressure
depending on a combination of two internal variables: density
(q) and specific internal energy (e); volume (V) and temperature
(T); or volume and specific internal energy, after [8]
p ¼ pðq; eÞ _ pðV; TÞ _ pðV; eÞ:
Experimental study of soft body impactors [6] illustrates that
the time-dependence of the pressure values at the impacted plate
shows a recognizable pattern with three distinct regions. Immedi-
ately after the initial contact very high values of pressure arise. The
peak pressure value (also called Hugoniot pressure) has the theo-
retical value [6]
pH ¼ q0 U S ðU 0 ÞU 0 :
The second stage is characterized by release waves which de-
crease pressure values [6]. After several reflections of release
waves, a region of stable and constant pressure is established.
The steady flow stage is characterized by the stagnation pressure
1
pstag ¼ q U2 :
2 0 0
In order to improve the time dependent pressure response at
the impact, porosity effects of water to air mixtures have been ta-
ken into account by programming the polynomial equation of state
using Abaqus/Explicit user material subroutine VUMAT. According
to [6], porosity has a significant effect on the shock velocity and
compressibility of soft body impactors and, consequently, needs
to be considered in order to realistically replicate the forces gener-
ated at an impact of real bird. Porosity decreases the shock velocity
in the material, resulting in the lower Hugoniot and stagnation
pressures, although the effect on stagnation pressures is not as pro-
nounced. According to [6] the effect of porosity (a) on the pressure
to density relation is accounted for by the relation
p 1B  1c
q0 p
¼ ð1  aÞ þ 1 þa
q A p0
where q0 and p0 are the initial density and pressure, respectively,
while c is the ratio of specific heats of the air. The empirical con-
stants A and B are defined by Eqs. (5) and (6), where s is the
coefficient which relates impacting and shock velocities, while c0
is the speed of sound in the material
q0 c20
A¼ ; ð5Þ
ð4s  1Þ
B ¼ 4s  1: ð6Þ
The effect of porosity on the pressure vs. relative specific vol-
ume relation (Hugoniot curves), after Eq. (4), is shown by Fig. 1.
The polynomial equation of state which has been programmed in
VUMAT has the form
p ¼ C 0 þ C 1 l þ C 2 l2 þ C 3 l3 ; ð7Þ
where l is a dimensionless parameter which is defined in terms of
the ratio of initial to current density
q
l¼  1: ð8Þ
q0
The polynomial EOS has been used as bird material in [7,9–11].
As in [7], the coefficients C0–C3 in the polynomial EOS (Eq. (7)) have
been varied in order to achieve a reasonable fit of the Hugoniot
curves of the homogenized materials with 10% porosity. The
Hugoniot curves of the approximated material properties (after
Eq. (7) for 10% porosity) along with the ones for water with various
porosity, after Eq. (4), have been plotted in Fig. 1.
The porous EOS material has been validated in an impact on a
rigid plate as to compare the impact pressures with experimental
values [6]. An Eulerian model containing 500,000 finite elements
has been used to discretize bird material motion in this analysis.
The results of the bird material validation are shown on Fig. 2 in
which the pressure values have been normalized by the stagnation
pressure in order to compare the values with experimental results.
ð1Þ
It can be concluded that the pressure temporal response follows
the general trend observed in experimental results with distinct
Hugoniot, pressure release and stagnation pressure stages. The
theoretical Hugoniot pressure value for an impact at 116 m/s is
93.6 MPa, after [12], which has given normalized value of 14.9.
Experimental results for real birds show significantly lower peak
pressure values, reaching normalized pressure value of 3.5. Fig. 2
shows that the EOS with approximated material properties shows
ð2Þ
significantly higher Hugoniot pressure values than the experimen-
tal values reported in [6], reaching normalized pressure value of
ð3Þ
ð4Þ
Fig. 1. Hugoniot curves of porous water and approximated homogenized bird
material with 10% porosity.
 D. Ivančević, I. Smojver / Composite Structures 94 (2011) 15–23 17

Fig. 2. Results of EOS material validation (left-hand side image) and deformation of the Eulerian impactor (right-hand side image).

9.1. The same conclusion has been reported in [9,10]. The Mie– where rn is the dynamic yield stress, ry is the static yield strength,
Grüneisen EOS of the non-porous material (used in [1]) predicts and e_ is the equivalent strain rate. The parameters D and p of the
the theoretical pressure value very accurately, but overestimates Cowper–Symonds law for the Al 2024 are taken from [13]. Combin-
even more the experimental values, with normalized peak pres- ing Eqs. (10) and (11) results in the final form of hardening rule
sure values reaching 13.81. The discrepancy between numerical "  1=p #
e_
and experimentally obtained results could be explained by the rðe; e_ Þ ¼ ½a þ bðep Þn  1 þ : ð12Þ
inability of the measuring equipment which has been used in the D
experiments to capture the instantaneous peak pressure [12].
All necessary parameters needed to define the constitutive behavior
The stagnation pressure values for the 10% porous EOS oscillate
of the Al 2024 alloy are summarized in Table 1.
around the theoretical and experimental value (normalized to
Fig. 3 shows the effect of different strain rates on the hardening
1.00). The frequency and amplitude of stagnation pressure oscilla-
behavior of the Al 2024 alloy as calculated after Eq. (12), with the
tion of the porous material are much lower compared to the Mie–
parameters listed in Table 1. Constitutive and damage modeling of
Grüneisen EOS (reported in [1]). The polynomial EOS with proper-
other metallic materials (material properties are taken from [17])
ties approximating 10% porosity will be used as bird replacement
and sandwich structures is explained in detail in [1].
material throughout this work, since Hugoniot and stagnation
Strain rate effects have not been included into the constitutive
pressure values are better approximated by the porous material.
behavior of the CFRP materials since strain rate effects do not cause
The deformation of the Eulerian impactor is shown in Fig. 2,
significant effect on carbon fiber properties, after [18,19]. Further-
right-hand side image. The impactor deformation closely resem-
bles gas-gun experiments published in [10,13,14] and numerical
results for bird replacement materials published in [10,14,15].

3. Damage modeling

Damage modeling of metallic structural components has been

improved by inclusion of strain rate effects. The deformation rates
in crashworthiness problems usually reach very high values, at
which strain rate effects on material behavior are significant and
cannot be neglected. The strain rate dependency has been included
for the Al 2024 alloy, which is used for the ribs in the analyzed flap
model. Strain rate effects for other metallic materials have been
neglected, since the impact location in this work is distant from
the other structural components and, consequently, no high strain-
ing rates are expected. Generally, the yield stress depends on
strain, strain rate and temperature
r ¼ f ðe; e_ ; TÞ: ð9Þ
The implemented VUMAT constitutive model for the Al 2024 al-
loy includes a Von Mises yield criterion and an algorithm that takes
into account strain rate sensitivity by enforcing the Cowper–Sy-
monds law. The elastic–plastic behavior has been defined as a
Fig. 3. Effect of strain rate on Al 2024 alloy after Eq. (12) with properties listed in
power law, after [16]
Table 1.
n
rðeÞ ¼ a þ bðep Þ : ð10Þ
The parameters a, b and n for the Al 2024 alloy are taken from
[16]. The mathematical description of the assigned hardening law Table 1
Al 2024 properties, after [13,16].
has the form
 1=p E m a b n D (s1) p efail q (kg/
rn e_ (GPa) (MPa) (MPa) m3)
¼1þ ; ð11Þ
ry D 72.4 0.33 277 485 0.55 1.28  10 5
4.0 0.18 2780
18 D. Ivančević, I. Smojver / Composite Structures 94 (2011) 15–23

more, epoxy matrices are also not prone to strain rate effects [18]. and
The properties of CFRP are taken from [20] and listed in Table 2. In  
r12  jsc j
order to extend the capability of the damage prediction procedure, 0 6   6 A ;
Puck’s failure and degradation model has been implemented in the
r22 R23
analysis. Puck’s criterion has been selected as it has achieved very with
good results in the World Wide Failure Exercise [21] and therefore qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pc12 pc23
presents one of the most potent failure criteria. The criterion be- sc ¼ R12 1 þ 2pc23 ; ¼ ;
longs to the group of phenomenological models as it distinguishes R12 RA23
between fiber failure (FF) and matrix failure or interfiber failure and
(IFF). It is restricted to continuous fiber reinforced composites 0sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
and transversely isotropic materials [22]. R12 @ Rc
Fiber failure is simply defined as the ratio between laminate RA23 ¼ c 1 þ 2pc12 22  1A:
2p12 R12
stress in the fiber direction and laminate strength in the same
direction [22]. This relation can be expressed as Puck’s IFF criterion distinguishes between three distinct modes
 
r  of matrix failure – Mode A calculated after Eq. (16), Mode B after
 11 
 t;c  6 1; ð13Þ Eq. (17) and Mode C after Eq. (18). Selection of the critical interfi-
 R11 
ber failure mode is done based on the stress components in the ply
where R11 t;c
represents tensile or compressive laminate strengths in coordinate system. The R parameters in Eqs. (16)–(18) are ply frac-
fiber direction (Sþ  ture resistances (defined in [22]), while p parameters are explained
1 or S1 in Table 2). Puck’s interfiber criterion is
based on the hypothesis that failure under a combined stress state in [22] and listed in Table 3, after [24].
is initiated in the plane which offers the lowest failure resistance to- At high longitudinal stresses an additional weakening factor (fw)
wards the acting stresses, thus having the highest probability of is introduced which scales the failure envelope (calculated with
failure. Therefore, interfiber failure is calculated in the fracture Eqs. (16)–(18)) thus accounting for microdamage at these stress
plane, which is parallel to the fiber direction and inclined to the states [22]
laminate plane by an angle hfp [22]. The failure criterion for the frac- vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  !2
u r 
ture plane distinguishes between tensile and compressive normal
u 1  11 
t
fw ¼ 1     0:7 for jr11 j P 0:7Rt;c ð19Þ
11 :
stress of the fracture plane. 0:12  Rt;c 11

!2 !2 !2
rn snt sn1 The damage evolution law presented in [22] has been applied in
þ þ ¼ 1 for rn P 0; ð14Þ
this work as well. According to this model, degradation of elastic
RðþÞA
? RA?? RA?k
!2 !2 properties due to fiber failure is not considered, as fiber failure
snt sn1 leads to ultimate ply failure. Matrix failure initiates damage which
ðÞ
þ ðÞ
¼ 1 for rn < 0: ð15Þ
is enforced by the degradation function g
RA??  p?? rn RA?k  p?k
1  gr
Eqs. (14) and (15) contain stress components which correspond g¼ : ð20Þ
to fracture plane coordinate system in which the 1 direction
1 þ cðfE  1Þf
coincides with fiber direction, while n and t designate fracture The parameter fE in Eq. (20) is called the factor of effort and cor-
plane normal and tangent, respectively. Using the well known responds to the left-sides of the Eqs. (16)–(18). The parameters c,
transformation rules for stress components which relate the frac- gr, and f are material parameters which are determined from
ture plane stresses to the laminate coordinate system, and plane experiments. As no experimental data are available for CFRP in
stress assumption, leads to a more commonly used form of Puck’s these analyzes, an approximation proposed in [22] is used
IFF criterion [23]:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
u !  2
g¼ : ð21Þ
u 1 pt 2 fE
t r12 pt
t  12
r 2
22 þ þ 12 r22 ¼ 1; hfp ¼ 0 for r22 P 0;
R22 R 12 R 12 R 12 Degradation of elastic properties is computed after the follow-
ing rules:
ð16Þ
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Ed2 ¼ gEinit
2 ; Gd12 ¼ gGinit
12 r22 > 0
for ð22Þ
 c 2  2
p12 r pc Ed2 ¼ Einit
2 ; Gd12 ¼g c
Ginit for r22 < 0; ð23Þ
r222 þ 12 þ 12 r22 ¼ 1; 12
 
R12 R12 R12   2
  where g ¼ g cos ð1Þ þ sin ð1Þ and 1 ¼ arctanrr22
c 2
. The Poison’s coef-
r22  RA ficients keep their initial values.
12

hfp ¼ 0 for r22 < 0 and 0 6   6 23 ; ð17Þ

r12 jsc j Puck’s failure initiation criterion and degradation model has
" been compared with Abaqus failure and degradation model for
2  2 #
r12 r22 Rc22 composite materials which is based on Hashin’s failure criteria
þ ¼ 1; [4]. The analyzes in this comparison try to replicate experimentally
2ð1 þ pc23 ÞR12 Rc22 r22
sffiffiffiffiffiffiffiffiffiffiffi observed substitute bird high velocity impact damage on compos-
RA23 ite plates [25]. The numerical model, consisting of 860 conven-
cos hfp ¼ for r22 < 0; ð18Þ tional shell elements for the target plate and 487,920 Eulerian
r22

Table 2
CFRP properties, after [20]. Table 3
CFRP properties for Puck’s IFF criterion, after [24].
q (kg/ E1 E2 G12 m12 Sþ
1¼ S
1 Sþ
2
S
2 S12
m3) (GPa) (GPa) (GPa) (MPa) (MPa) (MPa) (MPa) pt12 pc12 pt23 pc23

1600 181 10.3 7.17 0.28 1500 40 246 68 0.35 0.3 0.27 0.27
 D. Ivančević, I. Smojver / Composite Structures 94 (2011) 15–23 19

elements, is explained in detail in [2]. The experimental work in

[25] investigates the impact behavior of composite plates impacted
by soft body projectiles in the velocity range from 200–280 m/s.
The impacted plate is 3 mm thick and consists of 21 CFRP T300/
914 layers with a ½ð0=90Þ5 =0  layup. Dimensions of the composite
S
plate are 216  102 mm, with one end clamped on the shorter
edge with a 25 mm wide clamp. The initial velocity vector is de-
flected by 40° with respect to the plate plane. The diameter of
the gelatin projectile is 25 mm and its mass is 10 g. Fig. 4 shows
the numerical model and initial conditions. Comparison of Hashin’s
damage initiation and degradation model with visually observed
damage state of the impacted plates given in [25] resulted in
acceptable resemblance between numerical and experimental re-
sults as illustrated in [2]. Therefore the implemented Puck’s model
will be compared only with Hashin’s damage model.
The results of the comparison at an impact velocity of 200 m/s
are shown on Fig. 5. Both failure and degradation models predict
that the structural integrity of the impacted plate is intact as none
of the plate finite elements has reached the criterion for element
deletion according to both investigated criteria. This observation
Fig. 5. Failure initiation criteria at 200 m/s. Contours of maximal through thickness
is in accordance with experimental results [25], in which damage values are shown. Clamped end of the plate is on the left side.
is visually detected only at the clamped end of the plate. The last
observation is in correspondence with Hashin’s and Puck’s failure
criteria, which both predict fiber damage initiation at this location.
Puck’s IFF Mode A agrees very well with Hashin’s matrix tensile
failure criterion. The values for Puck’s IFF Mode C do not reach
the critical value even for the 280 m/s impact and are therefore
not shown on Figs. 5 and 6.
The results of the impact at 280 m/s are shown in Fig. 6. Exper-
imental testing revealed these at this initial impact conditions, the
CFRP plate has completely failed and has broken into two halves.
Numerically predicted responses of Hashin’s and Puck’s damage
models both underestimate the damaged state of the plate as only
one element of the mesh reached the element deletion criterion
(indicating complete failure of fiber damage modes through all
material points of the element). Complete compliance with exper-
imental results is difficult to obtain as numerical prediction of
complex phenomena like perforation of composite plates is an ex-
tremely difficult task. The results of failure criteria for fiber and
matrix failure modes reach maximal values in the area, where
the plate broke into two halves in the experiment, indicating that
numerically predicted damage by Puck’s and Hashin’s failure initi-
ation criteria are very similar. Elements with maximum damage
values have only limited load carrying capability and are likely to
fail with load increase. The applied damage models are last ply fail- Fig. 6. Failure initiation criteria at 280 m/s. Contours of maximal through thickness
ure criteria, as finite elements are removed from the analysis when values are shown. Clamped end of the plate is on the left side.

all material points in the element reach the critical degradation va-
lue. There are several possible sources of discrepancy between
numerical and experimental results. The most obvious of them
has been modeling of the impactor, as in the gas-gun experiments
gelatin substitute birds are being used, while in the numerical sim-
ulations equation of state materials having properties of a mixture
of water and air were modeled. As reported in [18] the damage in-
duced by gelatin bird substitutes is more severe compared to im-
pact involving real birds.

4. LESAD

Manual creation of new input files for CEL analyzes is a very

Fig. 4. Numerical model for damage model validation.
time-consuming task. In order to reduce the effort needed to create
new models for CEL analyzes, the substructure generation proce-
dure, introduced in [1], has been extended in order to enable inclu-
sion of Eulerian impactors. The new programme is called Lagrange
Euler Submodeling of Aeronautical Damage (LESAD). The auto-
matic procedure places the bird model in the desired initial posi-
tion and orientation based on the selected impact location and
flight parameters at the time of impact. LESAD enables damage
20 D. Ivančević, I. Smojver / Composite Structures 94 (2011) 15–23
prediction analyzes on smaller target structure models, and as an
additional feature is the ability of reducing computational time
by enabling analyzes on smaller finite element models of the im-
pacted structure. This capability allows short analyzes on complex
models in the preliminary stage of a damage prediction in which
various parameters have to be defined. Reduction of the finite ele-
ment model has been achieved by extracting smaller parts of the
complete model, based on the impact location and desired size of
the smaller model.
Lagrange and Eulerian bird model are offered as impactors for
the impact simulation. The Lagrange bird model enables shorter
calculations, but the results of such analyzes have to be controlled
carefully, as Lagrangian bird impactors are prone to cause numer-
ical errors, as discussed in [2]. LESAD offers option of two alterna-
tive sets of input flight parameters which are used to calculate
impactor orientation and velocity vector components. The first
set of parameters requires definition of aircraft velocity magnitude,
angle of attack and sideslip angle, while the second option requires
definition of aircraft velocity magnitude, angle of climb, heading
angle and orientation in space.
Fig. 7 shows the flowchart of the programme. The complete
algorithm of LESAD has been programmed within Visual Studio
2008 Express package by using the C# programming language.
Additionally, several Python scripts were written which run Aba-
qus/CAE in order to employ its volume fraction tool used to calcu-
late the Eulerian volume fractions for the Eulerian finite element
model. LESAD is fully object-oriented and all important keywords
within the input file are represented as objects with various prop-
erties. Consequently, the input file is not just read into memory
and then combined together with other input files, but is logically
divided into collection of modeling primitives (nodes, elements,
surfaces, etc. . .) which then enables faster submodel generation
in the next step. Error checking while reading the input file is per-
formed in such a way that unrecognized keywords, and subsequent
data cards until the next known keyword is recognized, are written
into a separate log file that serves as a necessary diagnostics tool.
Fig. 7. LESAD flowchart.
5. Analysis
The structure subjected to bird strike analyzed in this work is an
inboard flap of a typical large transport aircraft. The flap finite ele-
ment model is the same as the model analyzed in [1,2] and thus
only basic pieces of information about the modeling approach
are given in this work. Dimensions of the modeled structure are
span of 4.38 m and chord of 1.05 m. The flap structural layout con-
sists of front, rear and auxiliary spars and 14 ribs. These structural
items have been manufactured out of aluminum alloys. The flap
skins are made of CFRP with variable number of unidirectional lay-
ers resulting in gradually variable skin thickness. The flap skins are
additionally strengthened with 10 CFRP stringers, of which six are
positioned along the upper skin and four are positioned on the
lower flap skin. Flap leading and trailing edges have been designed
as sandwich structures consisting of CFRP face layers for the lead-
ing edge and aluminum alloy for the trailing edge face layers. No-
mex honeycomb cores have been used in both sandwich
structures.
The flap finite element model consists of a total number of
109,585 elements, of which 69,696 are conventional shell, 16,020
continuum shell, and 23,665 solid elements. Additionally, front
spar reinforcements are modeled with 204 beam elements. Most
of the flap geometry is suitable for two-dimensional discretisation,
whereas three-dimensional finite elements have been employed to
model sandwich structures. Tie surface kinematic constraints had
to be introduced at the interface of different elements as to effi-
ciently connect finite elements having different nodal degrees of
freedom. The gradually variable composite skin in the impact area
is 2.5 mm thick, having a composite layup [45/0/45/90/45/0/
±45/90/45/02/452/0/45/90/45/0/45]. The 0° oriented layer cor-
responds to the flap span direction and the first layer in the lami-
nate stacking sequence is located on the internal composite skin.
The numerical impact model is shown in Fig. 8. Puck’s damage
model has been included only for lower flap CFRP skin and string-
ers as no damage is expected on the upper skin.
 D. Ivančević, I. Smojver / Composite Structures 94 (2011) 15–23 21

6. Results

Results of the analysis are presented in Figs. 9–12. The bird

Fig. 8. Numerical model of impact analyzed in this work.
As high lift devices are extended only during the take-off and
landing phases of the flight, flap structures are exposed to bird
strikes at relatively low velocities. Taking this into account, this
work presents results of a bird strike simulation which included
a bird mass used for certification after FAR 25.571 at 100 m/s.
Taking into account the approximate landing glide slope, wing
incidence angle, aircraft pitch angle and maximal flap deflection
position (40°), the resulting total angle of attack of the bird is 55°
with regard to the flap reference plane. The velocity component
in the flap span direction is zero, as the inboard flap is perpendic-
ular to the airplane’s plane of symmetry. The Euelrian model in the
presented analysis consists of 600,000 elements enclosing a
volume having dimensions 1  1  0.6 m.
impactor in CEL analyzes is modeled as Eulerian material which
flows through the finite element mesh. The material motion is
tracked as it flows through the mesh by means of Eulerian Volume
Fractions (EVF) which represent the ratio by which each Eulerian
element is filled with material, with EVF equal to one representing
element completely filled with material, contrary to the com-
pletely void elements with EVF equal to zero. Interaction of the flap
structure with the bird material and evolution of EVF values during
the impact are shown by Fig. 9.
Bird material deformation with Von Mises stress contours on the
flap lower skin is shown by Fig. 10. Eulerian formulation of the bird
impactor enables extreme material deformation, as the material can
move through the Eulerian mesh without limitations existing in the
Lagrangian model. The advantages compared to the Lagrangian bird
model (presented in [1]) are obvious, as even material disintegration
is modeled without causing numerical instabilities. To the contrary,
Lagrangian bird models are always integrated, which is not in accor-
dance with the behavior of real birds during a bird strike.
The kinetic energy of the bird in this analysis has not been suf-
ficient to completely penetrate through the flap skin, although
some elements reached the element removal criterion. According

Fig. 9. Eulerian volume fractions in the cross section at the impact.

Fig. 10. Deformation of the bird impactor. Contours of Von Mises stress are shown.
22 D. Ivančević, I. Smojver / Composite Structures 94 (2011) 15–23

Fig. 11. Contours of Puck’s damage initiation criterion (upper skin and stringers are removed).

Fig. 12. Degraded values of E2.

to the implemented damage model, an element is removed from
the mesh when all material points in an element reach Puck’s fiber
(tensile or compressive) failure initiation criteria. Fig. 11 shows
maximal through thickness values of Puck’s failure initiation crite-
ria calculated during the analysis. The fiber failure (FF) tensile cri-
terion reaches critical values, as expected, in the vicinity of the
impact location. The compressive fiber criterion reaches the critical
value only in the skin stringers. Contrary to FF modes, which reach
critical values locally in the area of initial impact, interfiber failure
(IFF) modes reach very high values on a large segment of the skin
and stringers. IFF Mode A is the most critical matrix failure crite-
rion, contrary to Mode C which does not reach the critical value
and is therefore not shown on Fig. 11. An example of the degrada-
tion effect on the E2 ply properties for the outer most ply of the im-
pacted skin and stringers (oriented by 45° with regard to the flap
span direction) is shown in Fig. 12. The figure shows contours of
Young’s modulus in the two direction which has been reduced
after Eqs. (22) and (23) by more than two orders of magnitude
compared to the initial value in some material points.
7. Conclusions
The work presented in this paper shows improvements in the
damage prediction procedure introduced in [1]. The procedure is
based on nonlinear explicit finite element methods in combination
with complex structural finite element models, various failure and
degradation modes and an Eulerian impactor model. Compared to
the previous work, improvements have been made in the field of
damage modeling of composite structures by implementation of
Puck’s failure and degradation model. Puck’s failure criterion has
achieved remarkable results in the World Wide Failure Exercise
[21] and presents one of the most potent failure criteria for further
development of damage models [23]. The model described in this
work has been compared to the Hashin’s damage model and dem-
onstrated acceptable similarity. A further important improvement
has been made in the field of bird modeling, as the polynomial
equation of state has been programmed in order to replicate the
constitutive behavior of a water and air mixture with 10% porosity.
An equation of state with approximated polynomial coefficients
has been validated by comparison with experimental results pub-
lished in the literature. The obtained results show that application
of the new EOS properly predicts the stagnation pressure. On the
other hand, the peak pressure value is still overestimated, although
to a much lower extent then the Mie–Grüneisen EOS which has
been used in the previous work. The ability of presented method-
ology to simulate the bird strike on a complex aeronautical struc-
ture is demonstrated in an impact simulation on a typical large
airliner flap structure. The flap structure is able to withstand the
impact of the bird mass used for certification requirements with-
out complete penetration or loss of load carrying ability.
 D. Ivančević, I. Smojver / Composite Structures 94 (2011) 15–23 23

References [13] Guida M, Marulo F, Meo M, Riccio M. Analysis of bird impact on a composite
tailplane leading edge. Appl Compos Mater 2008;15:241–57.
[14] Lavoie MA, Gakwaya A, Nejad Ensan M, Zimcik DG, Nandlall D. Bird’s
[1] Smojver I, Ivancevic D. Numerical simulation of bird strike damage prediction
substitude test results and evaluation of available numerical methods. Int J
in airplane flap structure. Compos Struct 2010;92:2016–26.
Imp Eng 2009;36:1276–87.
[2] Smojver I, Ivancevic D. Bird strike damage analysis in aircraft structures using
[15] McCarthy MA, Xiao JR, McCarthy CT, Kamoulakos A, Ramos J, Gallard JP, et al.
abaqus/explicit and coupled Eulerian Lagrangian approach. Compos Sci
Modelling of bird strike on an aircraft wing leading edge made from fibre
Technol 2011;71:489–98.
metal laminates – Part 2: Modelling of impact with SPH bird model. Appl
[3] Smojver I, Ivancevic D, Mihaljević D. An explicit numerical modeling of soft
Compos Mater 2004;11:317–40.
body impact damage in metallic airplane structures. CMES: Comput Model Eng
[16] McCarthy MA, Xiao JR, Petrinic N, Kamoulakos A, Melito V. Modelling of bird
Sci 2010;70(2):191–216.
strike on an aircraft wing leading edge made from fibre metal laminates – Part
[4] Abaqus Analysis User’s Manual. Version 6.8. Dassault Systémes; 2008.
1: Material modelling. Appl Compos Mater 2004;11:295–315.
[5] Benson DJ, Okazava S. Contact in a multi-material Eulerian finite element
[17] Metallic Materials and Elements for Aerospace Vehicle Structures MIL-HDBK-
formulation. Comput Methods Appl Mech Eng 2004;193:4277–98.
5J. Department of Defence Handbook, Washington (DC); 2003.
[6] Wilbeck JS. Impact behavior of low strength projectiles. Air Force Materials
[18] Cantwell WJ, Morton J. Comparison of the low and high velocity impact
Laboratory, Technical Report AFML-TR-77-134; 1977.
response CFRP. Composites 1989;20:545–51.
[7] Airoldi A, Cacchione B. Modelling of impact forces and pressures in lagrangian
[19] Will MA, Franz T, Nurick GN. The effect of laminate stacking sequence of CFRP
bird strike analyses. Int J Imp Eng 2006;32:1651–77.
filament wound tubes subjected to projectile impact. Compos Struct
[8] Hiermaier S. Structures under crash and impact. New York: Springer
2002;58:259–70.
Science+Business Media; 2008.
[20] Springer GS. Mechanics of composite structures. Cambridge: Cambridge
[9] Tho CH, Smith MR. Accurate bird strike simulation methodology for BA609
University Press; 2003.
tiltrotor. In: Presented at the American helicopter society 64th annual forum;
[21] Hinton MJ, Kaddour AS, Soden PD. A further assessment of the predictive
2008.
capabilities of current failure theories for composite laminates: comparison
[10] Johnson AF, Holzapfel M. Modelling soft body impact on composite structures.
with experimental evidence. Compos Sci Technol 2004;64:549–88.
Compos Struct 2003;63:103–13.
[22] Puck A, Schürmann H. Failure analysis of FRP laminates by means of physically
[11] Hanssen AG, Girard Y, Olovsson L, Berstad T, Langseth M. A numerical model
based phenomenological models. Compos Sci Technol 1998;58:1045–67.
for bird strike of aluminium foam-based sandwich panels. Int J Imp Eng
[23] Schücker C. Mechanism based modeling of damage and failure in fiber
2006;32:1127–44.
reinforced polymer laminates. Düsseldorf: VDI Verlag; 2006.
[12] Lavoie MA, Gakwaya A, Nejad Ensan M, Zimcik DG. Review of existing
[24] Puck A, Kopp J, Knops M. Guidelines for the determination of the parameters in
numerical methods and validation procedure available for bird strike
Puck’s action plane strength criterion. Compos Sci Technol 2002;62:371–8.
modeling. In: International conference on computational & experimental
[25] Hou JP, Ruiz C. Soft body impact on laminated composite materials. Compos:
engineering and sciences; 2007.
Part A 2007;38:505–15.