Aerospace Science and Technology 28 (2013) 9–20

Contents lists available at SciVerse ScienceDirect

Aerospace Science and Technology

www.elsevier.com/locate/aescte

A new bird model and the effect of bird geometry in impacts from
various orientations
Reza Hedayati ∗ , Saeed Ziaei-Rad
Mechanical Engineering Department, Isfahan University of Technology, Isfahan 84156-83111, Iran

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

Article history: In this paper, a bird model with geometry similar to a real bird (Bufflehead Duck) is introduced
Received 14 July 2011 and compared to traditional bird models and also to experimental data. A bird can impact a part of
Received in revised form 1 August 2012 an airplane from its head, tail, bottom or wings. Any of these orientations can have a different effect on
Accepted 18 September 2012
the response of an airplane part. Since all birds do not have the same body shapes and sizes, and, since,
Available online 5 October 2012
it is not always convenient to model a complex bird body, four substitute bird models are introduced.
Keywords: The models results are then compared in order to determine the best substitute bird which can properly
Bird strike model the pressure and force exerted by real birds when impacting from different orientations. It was
SPH method concluded that the impact from bird bottom side is the most damaging scenario, while the tail side
Bird orientation impact is the less dangerous one. It was also found that for the tail side impact scenario, a hemispherical-
Realistic bird model ended cylinder shows the best results, while for the bottom side impact scenario, an ellipsoid can be the
Bufflehead duck best candidate for the bird substitute model.
Ellipsoid shaped bird model
© 2012 Elsevier Masson SAS. All rights reserved.

1. Introduction civil or military bird strike design requirements. However, analy-

sis techniques must be validated before they can be employed to
Bird strike is a common hazard for flight safety that causes accurately guide the design process.
remarkable losses every year. Statistics indicate that although all In the absence of high-tech computers, Barber et al. [2] per-
aircraft designs have to be certified for a level of bird impact formed numerous bird impact tests on a rigid target plate. He used
resistance, the number of bird strikes has been increasing every birds with different body shapes, weights, etc. He found out that
year [5]. This can be attributed to globalization which has resulted the peak pressures generated on the center of the target plate
in the steady increase in air traffic density levels as well as the due to bird impact is independent of the bird size and is propor-
dramatic expansion of wild bird populations [5]. Based on the Fed- tional to the square of the impact velocity. Niering used Lagrangian
eral Aviation Administration National Wildlife Strike Database [25], method to model the bird impact phenomenon [21]. He used
during the years 1912 to 2004, there were 47 fatal bird strike different finite element methods in his work, but finally stated
accidents that have killed 242 people. The total number of air- that refinements in the models are necessary due to high dis-
craft destroyed due to the bird strike in that period was 90. Bird tortion of bird elements during the impact. Stoll and Brockman
strike costs the U.S. civil aviation industry more than $495 million tried to overcome the problem by eliminating highly distorted el-
yearly [5]. ements [24]. This, however, leads to the bird mass loss during the
An aircraft must show compliance with “continued safe flight impact. Due to Lagrangian method disadvantages, some authors
and landing” requirements following specified types of high-energy used other approaches like Arbitrary Lagrangian–Eulerian (ALE) for
bird impact. Experimental tests for bird strike certification are very modeling bird behavior during the impact. Moffat and Cleghorn
expensive and time consuming. In order to decrease the number implemented the MSC/DYTRAN code to develop a bird model for
of costly sample tests, a trustworthy analysis tool is necessary to an ALE description [19]. They reproduced the impacts of the bird in
reliably predict the structural responses of bird and target. The rigid and flexible targets. Hanssen et al. also used the ALE approach
use of Finite Element (FE) simulation provides the opportunity to to study the bird impact against aluminum foam-based sandwich
effectively evaluate numerous structural design approaches that panels [8].
minimize structural weight and reduce the risk of not meeting Another approach to study the bird strike simulation is a mesh-
less method called Smoothed Particles Hydrodynamics (SPH). Hutt
et al. introduced the SPH method, showed its utilization and pro-
vided comprehensive data of its computational parameters [9].
* Corresponding author. Tel.: +98 311 3915244; fax: +98 311 3912628.
Martin studied a transient, material, and geometric nonlinear fi-
E-mail addresses: [email protected] (R. Hedayati), [email protected]
(S. Ziaei-Rad). nite element based impact analysis using PW/WHAM [16]. He used

1270-9638/$ – see front matter © 2012 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ast.2012.09.002
10 R. Hedayati, S. Ziaei-Rad / Aerospace Science and Technology 28 (2013) 9–20

formations and inelastic strains. The constraint derived from the

contacts and the elastic–plastic behavior considered for the mate-
rial can be used to study the interaction between the bird and the
target.
Since late eighties, FE codes adopted various finite element ap-
proaches to model the impact phenomena. These include the La-
grangian approach, Arbitrary Lagrangian–Eulerian (ALE) approach,
and recent solvers based on Smoothed Particle Hydrodynamics
(SPH).
Unlike the implicit technique, the explicit integration technique
used in the bird strike analysis is conditionally stable, requiring
the critical time step to meet the Courant criterion, which is ulti-
mately determined by the smallest element size in the FE model.
The advantage of Lagrangian method is its ease of use and also
the ability to track history dependant materials. However, by us-
ing the Lagrangian approach for bird modeling, the element size
tends to become very small and distorted. This ultimately results
Fig. 1. A flock of birds striking an airplane wing (courtesy of Juergen Kienast/Bild). in prohibitive computational time and often produces an unsta-
ble numerical solution. In fact, in an explicit FE analysis, the time
a formulation very similar to the concept of SPH technique. Ubels step t is determined by the characteristic length of the smallest
et al. also used the SPH method to simulate the bird strike against element L:
airplane wing leading edge structure [26]. They employed the L
PAM-CRASH code to model the bird using SPH elements and the t = (1)
c
impacted structure using Lagrangian elements.
Researchers have used three simple configurations to model where c is the wave propagation velocity in the material. Since
the bird in a bird strike event. By reviewing the literature, it was severe mesh distortion decreases the smallest side size of element,
found out that the effect of implementing a realistic bird model it causes the time step to decrease to an excessively low value for
has not been investigated so far. As a result in this paper, a bird the calculations to be continued.
model with geometry similar to a real bird (Bufflehead Duck) is The Arbitrary Lagrangian–Eulerian (ALE) formulation is a com-
introduced and compared to traditional bird models and also to bination of the Lagrange and Eulerian formulations in which the
the experimental data. There are different bird types with differ- reference is set arbitrarily by the user in order to capture the ad-
ent body shapes; but most have an ellipsoid shaped torso with vantages of the methods while minimizing the disadvantages. The
a round head much smaller than their torso. Although there are user must set the mesh motion that best suites the problem in or-
many different types of birds, the obtained results can be extended der to minimize the mesh distortions and obtain the best results.
to other bird species in a qualitative way. Modeling a bird with The main disadvantage of this formulation is that the user must
a realistic body shape also makes possible to investigate the ef- be experienced in order to select the best method, and interpret
fect of bird orientations in the bird strike on structures. When not the results obtained. A very comprehensive description of the ALE
many investigations have been carried out on the effect of ori- method has been presented by Stoker [23].
entation a bird impacts a surface, it can be an important factor. SPH is a meshless method first introduced by Lucy [15]. It is
It is because in the real bird strike events the birds do not always a Lagrangian numerical technique used to solve the fluid equations
impact a surface from the same orientation. Fig. 1 shows a real of motion. Specifically, it uses an interpolation kernel of compact
situation in which a flock of birds with different orientations are support to represent any field quantity in terms of its values as
going to strike an airplane. A bird can impact a part of an airplane a set of disordered particles. The material is discretized, and prop-
from its head, tail, bottom or wings. Any of these orientations can erties of these nodes are associated with the center of the parti-
have a different effect on the integrity of airplane parts. The re- cles. A chosen interpolation kernel determines the amount of effect
sults are then compared to the traditional bird strike simulations that a known point value contributes to the point of interest. The
as well as experimental results published by Wilbeck [27]. Since advantages of the SPH method is that first, it does not require el-
all birds do not have the same body shapes and sizes, and since, ement meshing and thus avoids mesh tangling; second, unlike the
it is not always convenient to model a complex bird body, four ALE method it does not require considering and setting many pa-
substitute bird models are introduced. The model results are then rameters and is less complex and; third, since in this paper the
compared in order to determine the best substitute bird which can effect of orientation at which a bird impacts a surface is investi-
best model the pressure and force exerted by real birds when im- gated, the different mesh size at different parts of a bird can affect
pacting from different orientations. In the present study, all the the results. In the SPH method, the particles are described by:
specifications for simulations are similar to traditional bird models  
xi (t ), mi (t ) i ∈ P (2)
used by other researchers, except for the geometry of the bird.
where P is the set of moving particles and xi (t ) and mi (t ) are
2. Numerical methods the location and the mass of each moving particle, respectively.
The movement of each particle is related to the mass change
Since the experimental verifications are time consuming, ex- by [10]:
pensive and difficult to perform, numerical simulations can provide
dmi
significant help in designing high-efficiency bird-proof structures. = ∇. V (xi , t )mi (3)
Explicit nonlinear finite element (FE) codes, which are available in dt
several high-end commercial FE solvers, have been used to treat where V is the velocity of the particle i. Smoothing kernel
different problems of this class. is a valuable concept in SPH technique. Before presenting the
A bird strike is characterized by loads with high intensity and smoothing kernel, auxiliary cubic B-spline function must be in-
short duration. The materials undergo high strain rates, large de- troduced:
 R. Hedayati, S. Ziaei-Rad / Aerospace Science and Technology 28 (2013) 9–20 11

⎧ 3 3
⎨ 1 − 2 y2 + 4 y3 y  1
θ( y ) = α1 × 1
⎩ 4 (2 − y ) 1 y2
3 (4)
0 y2
where α1 is a constant depending on the dimension and the slope
of kernel function. In two dimensions: α1 = 710
π.
Lacome suggested the smoothing kernel as [10]:
 
1 xi − x j
W (xi − x j , h̄) = θ (5)
h̄ h̄
where h̄ is the smoothing length. Initially, while using the SPH
method, the smoothing length was considered constant during the
entire simulation. However it has been shown that it is more
appropriate for each particle to have its own smoothing particle
depending on the number of neighbor particles. Therefore, in the
Fig. 2. Pressure profile at the center of impact from Wilbeck experimental test
above equations, h̄ can be replaced by h(xi ).
(1977).
In 1989, Monaghan defined the mass density by [20]:
N 3. Theoretical methods
ρ ( xi ) = m j W (xi − x j , h̄) (6)
j =1 Wilbeck was one of the first researchers who investigated the
The equation of conservation of mass in Lagrangian form is [20]: experimental behavior of a bird under impact [27]. His conclusions
and results have kept their importance till now since most numer-
dρ ical case studies carried out by other researchers used the same
(xi ) = −ρ ∇ V (7)
dt shape and characteristics as used in Wilbeck’s paper. Also, results
The SPH approximation for the conservation of mass can be writ- from his experimental data were used as a reference for compari-
ten as [20]: son with numerical results. Among Wilbeck’s numerous tests, the
impact of a real bird having an initial velocity of 116 m s−1 was
N
dρ   selected for this study (see Fig. 2).
( xi ) = m j v ( x j ) − v ( xi ) ∇ W i j (8)
When a projectile of any material impacts a target plate, the
dt
j =1 particles on the front surface of the projectile are instantaneously
The neighbor search is a very important step in the SPH computa- brought to rest relative to the target face and a shock is formed.
tion. It is important to know at any time of the calculation which The purpose of this shock wave is to bring each succeeding layer
particle is going to interact with which other. The sphere of influ- of particles to rest. The pressure in the shock compressed region is
ence of each particle is a finite domain of a radius of 2 h. The goal very high initially and is constant throughout the region at early
of the neighbor search is to list the particle inside that domain times. As the shock propagates up the projectile, the particles
at each time step. In a direct search for a set of N particles, the along the projectile’s edge are subjected to a very high pressure
number of distance comparisons required is N − 1. Since this com- gradient due to the shock loading on one side and the free sur-
parison needs to be made for each particle, the total number of face on the other. This pressure gradient causes the particles to be
comparisons is N ( N − 1) which can dominate the total CPU time accelerated radially outward and a release wave is formed.
for large models. The idea of the search for neighbors is to use If, at any time, the state of stress is such that the strength of
the same algorithm than the one used for the contact search: the the material is exceeded, the material will “flow”. For soft body
bucket sort. The domain covered by the particles is split in several impact, it will be assumed that the stresses throughout the impact
boxes of a given size. Then for each particle, we search for neigh- event greatly exceed the material strength, so that the flow will
bors inside the main box and also the neighbor boxes contained continue indefinitely. After several reflections of the release waves,
in the domain of influence of the given particle. Then, when the a condition of steady flow is established. A constant pressure and
list of hypothetic neighbors is known, we compute the distance velocity field is set up in the projectile, and the particles flow along
between each couple of particles and check if it is smaller than paths which are fixed in space, called streamlines [27].
twice the smoothing length. The total cost of the grouping opera- Using a hydrodynamic impact theory and considering a ho-
tions and comparisons, needed to form the buckets, can be nearly mogeneous, right-circular cylinder impacting normally on a rigid
linear with the number of particles N [11]. plate, the flow across a shock can be considered one-dimensional
Many researchers [6,12,17,1] have simplified the bird torso as and adiabatic. A short time after the initial shock wave propagates,
a hemispherical-ended cylinder. The ellipsoid geometry is also a the equations of conservation of mass (continuity) and momentum
well-accepted choice, which has been suggested by the Interna- for materials behind and after the shock wave can be written as:
tional Bird Strike Research Group [22], and has been used by
Guan et al. [7]. Besides these two configurations, the straight- ρ1 V sh = ρ2 ( V sh − V p ) (9)
ended cylinder has also been adopted by Brockman and Held [3], 2 2
P1 + ρ 1 V sh = P 2 + ρ2 ( V sh − V p ) (10)
but its application remains somewhat infrequent. Recently, Meguid
et al. focused on the three most-frequently used configurations where ρ1 , P 1 , ρ2 and P 2 are the density and pressure of mate-
mentioned above: namely, hemispherical-ended cylinder, straight- rial before and behind the shock wave. V sh is the velocity of the
ended cylinder, and ellipsoid, at various length-to-diameter aspect shock propagating into the fluid and V p is the velocity of the parti-
ratios [18]. The results of their study showed that the initial con- cles behind the shock. For a circular cylinder it can be proved that
tact area between the bird and target in the early phase of the V p is equal to the initial velocity of bird V i . Combining Eqs. (9)
impact event would have a significant effect on the peak impact and (10) and by defining the Huguenot pressure as P H = P 2 − P 1 ,
force. The aspect ratio of the bird striking both rigid panel and the pressure behind the shock is found to be:
flexible fan blade was found to have little influence on the nor-
malized impact force and impulse. P H = ρ1 V sh V i (11)
12 R. Hedayati, S. Ziaei-Rad / Aerospace Science and Technology 28 (2013) 9–20

Fig. 4. A typical flying Bufflehead. (For interpretation of the references to color in

this figure legend, the reader is referred to the web version of this article.)

Table 1
Fig. 3. Variation of shock velocity versus initial velocity for a right-ended cylinder Bird geometrical specifications.
with porosity of 10%. Bird head-to-tail length 18.7 cm
Bird torso radius 5.4 cm
The proportion of densities before and after impact can be derived Head radius 2.7 cm
Wing span 31 cm
from:

 − c1
ρ1 PH 2
= (1 − α ) +1 (12)
ρ2 c1

with constants:

ρ1 c02
c1 = (13)
4k − 1
c 2 = 4k − 1 (14)

where k is an experimental constant, c 0 is the speed of sound in

material and α is the porosity of material. Wilbeck found out that
a bird can best be described as a low strength material with an av-
erage density of 950 kg m−3 and 10% porosity, i.e. α = 0.1. By solv-
ing Eqs. (9) and (12) simultaneously and replacing P H from (11)
into (12), the shock velocity can be obtained by the nonlinear
equation:

 − 4k1−1
V sh V sh V i (4k − 1)
= (1 − α ) (15)
V sh − V i 2
c0

As it can be seen from Eq. (15), the shock velocity is a func- Fig. 5. A real Bufflehead compared to its SPH model.

tion of initial velocity. By solving Eq. (15) for V sh , Fig. 3 for shock
velocity as a function of initial velocity is obtained. 4. A realistic bird model
In the steady-state condition, since the bird materials flow in
streamlines, using Bernoulli’s equation, the stagnation pressure at 4.1. Bird model geometry
the center of impact is found by:
In Fig. 4, a typical flying Bufflehead is shown. Two red contours
1 show how the body of a Bufflehead can be simplified. As already
P stag = ρ V i2 (16) mentioned, for the realistic bird model and all other models intro-
2
duced in this paper, only the SPH method is used.
For the test considered, having the initial velocity of V i = The model bird is 0.3 kg, and has a length of 0.187 m and
116 m s−1 and initial density of ρ1 = 950 kg m−3 , using Eqs. (11) a wing span of 0.31 m. The geometrical dimensions considered for
and (16), Huguenot pressure and Stagnation pressure are theoreti- the bird model are more specified in Table 1. It is assumed that
cally found to be 88 MPa and 6 MPa, respectively. These values for at the time of impact, the wings are completely spread. In Fig. 5
the experimental test (Fig. 2) are respectively 23 MPa and 5 MPa the real Bufflehead is compared to its SPH model from two views.
while it is 37 and 6 MPa for the new FE bird model used (the The SPH bird model consists of 16 284 SPH elements. The distance
simulation results are given under Subsection 5.2). The inconsis- between two adjacent SPH elements is chosen to be 3 mm.
tency in theoretical and experimental values can be attributed to The bird is hit to the target from 4 different orientations: head,
the limitations in the response of pressure gages. In fact, while tail, bottom and wing. The position of sensors with respect to the
the Huguenot pressure peak takes place in a very short time, SPH bird model is shown in Fig. 6. For impact from head, tail and
the 100 kHz frequency response of the transducers prevents them bottom side, four sensors were attached on the target. For the wing
from measurement of rise times of less than 5 μs. In FE simula- side, seven sensors were mounted on the target in order to capture
tions, 100 kHz frequency has also been chosen for gauges response. the pressure profile in both x and y directions.
 R. Hedayati, S. Ziaei-Rad / Aerospace Science and Technology 28 (2013) 9–20 13

Fig. 7. Setup of bird and target with the positions of four sensors.

Fig. 6. Position of sensors with respect to the SPH bird model in impacts with dif-
ferent orientations.

4.2. Bird material model

At high impact velocities, in which most fatal bird strikes take

place, the bird behaves like a fluid and therefore a hydrodynamic
material model can be used for the bird with a reasonable ap-
proximation. Thus, a null material model with an Equation of State
(EOS) was employed for the SPH bird constitutive modeling. It is
noteworthy that the null material model allows only the isotropic
components of the stress tensor.
Various equations of states have been proposed for the bird
so far: Polynomial, Tabulated and Gruneisen. However, in this
study a parametric study was conducted and it was found that
a Gruneisen equation of state can best predict the behavior of bird
in bird strike impacts. Gruneisen equation (also called v s − v p
equation of state) describes a linear relationship between the
shock and particle velocities. In this study, for the bird a null ma-
terial model with Gruneisen EOS was used. Gruneisen EOS with
cubic shock velocity defines pressure for compressed materials
as [14]:

ρ0 C 2 μ 1 + (1 − γ20 )μ − 2a μ2
P= 2
+ (γ0 + aμ) E (9)
μ2 μ3
1 − ( S 1 − 1)μ − S 2 μ+1 − S 3 (μ+1)2
Fig. 8. Deformation of SPH bird model at different times for the birds impacting the
and for expanded material as: target from bottom side (first row) and tail side (second row).

P = ρ0 C 2 μ + (γ0 + aμ) E (10) target plate is composed of 14 640 solid elements. In addition to
where C is the intercept of the v s − v p curve; S1, S2 and S3 that, one 1 cm × 1 cm shell element at the center of target and
are the coefficients of the slope of the v s − v p curve; γ0 is the three 0.25 cm × 0.25 cm shell elements at upper positions are used
Gruneisen gamma; and μ = ρ /ρi − 1 where ρ is the material den- in order to capture the pressure profile imposed by the bird model
sity. onto the target disc. Both the target solid elements and sensors
For the water Gruneisen equation of state, parameters C = 1480 shell elements were given the linear isotropic material model of
and S 1 = 1.92 were used and other parameters were set to zero steel with density of 7800 kg m−3 , Young Modulus of 207 GPa and
as defined in [4]. In previous numerical simulations, authors have Poisson’s ratio of 0.3. The target and the positions of four sensors
used densities ranged from 900 kg m−3 to 950 kg m−3 for bird are shown in Fig. 7.
material. Here, a density of 938 kg m−3 is used for the realistic In all calculations, a node to surface contact algorithm was used
and other substitute bird models as it has been used by most au- in order to model the contact between the bird and the rigid tar-
thors previously. In a real bird, the wings structures and materials get. All the nodes placed on the SPH particles were used as the
are different from the rest of the body. However, in this study the slave set and all the external segments of target plate elements
wings are modeled as the rest of the body – i.e. as a “water” bullet. were used as the master set.

4.3. Target specification 5. Results and discussion

A steel plate, 60 cm in diameter and 6 cm in thickness, is used Deformation of SPH bird model at different time instants for the
as target for bird impact simulations. Large thickness of the target birds impacting the target from tail and bottom sides are shown in
plate allows it to be considered appropriately as a rigid plate. The Fig. 8.
14 R. Hedayati, S. Ziaei-Rad / Aerospace Science and Technology 28 (2013) 9–20

from its tail side are compared with the Wilbeck experimental
data.
Pressure profile at the center of impact is the most crucial
criterion for proving the accuracy of a bird model. The pres-
sure profiles at the center of impact for the impact from tail
side and the hemispherical-ended bird model have been com-
pared to the Wilbeck experimental results in Fig. 11. For the
sake of brevity, specifications of hemispherical-ended bird model
are not presented here, but will be explained in Section 6. As it
can be seen, the hemispherical-ended cylinder model, the realis-
tic bird model, and Wilbeck test have the pressure peak of 79,
37 and 23 MPa, respectively. When the bird impacts the tar-
get from the tail side, the pressure profile correlates well with
Wilbeck test results, while the hemispherical-ended cylinder bird
model predicts much higher pressure peak. Therefore, one can con-
clude that the new bird model can better predict the behavior of
bird.
Initial contact area has a significant effect on the Huguenot
pressure in each impact scenario. The larger the initial contact area
is the higher the Huguenot pressure will be. It is because when the
Fig. 9. Deformation of different bird models in the perpendicular impact. initial contact area is larger, the material at the initial contact area
is more limited by the neighbor particles. These neighbor particles
5.1. Comparison between the Lagrangian, ALE and SPH formulations prevent the materials at the initial contact area to move outward
which then causes the Huguenot pressure increase. As it can be
seen from Fig. 6, the initial contact area of the head side impact is
Before selecting the appropriate formulation for the bird strike
larger than that in the tail-side impact. That is why the head-side
simulations, a comprehensive set of simulations were carried out.
peak pressure is higher. The new bird model tail side has smaller
Three types of bird formulations, i.e. Lagrangian, ALE and SPH were
initial contact area than that of the hemispherical-ended cylinder.
considered. The hemispherical-ended cylinder bird models were
That is why the peak pressure of the new bird model is lower than
created and then impacted to perpendicular rigid target plates.
that of the hemispherical-ended cylinder.
The pressure profiles were compared to Wilbeck experimental test.
The other criterion which is commonly used to validate the nu-
A steel plate, previously described in Subsection 4.3 is used as the
merical bird model is the distribution of stagnation pressure on
target for all bird impact simulations. Large thickness of the tar-
the target (Fig. 12). The stagnation pressure value at the center of
get plate allows it to be considered appropriately as a rigid plate.
impact for the realistic bird model, Wilbeck experiment and theory
Since the target is assumed to be rigid and not allowed to have
(Eq. (8)) are 6 MPa, 5 MPa and 6 MPa, respectively. These values
large deformations, dimensions and specifications used for the tar-
are in a good agreement with each other.
get are not crucial. In fact, the pressures captured by the sensors
are of interest and not the target deformation itself. In this sec-
5.3. Effect of bird orientation on pressure distribution
tion the bird was modeled as a water projectile with the shape
of a hemispherical-ended cylinder with a length to diameter ratio
The pressure plots corresponding to four sensors for impacts
of 2. The impact velocity was set to 116 m/s.
with different orientations are shown in Fig. 13(a)–(d). For impact
The deformation of the bird model obtained from different for-
from the bottom side, three sensors reach pressures more than
mulations with respect to time can be seen in Fig. 9. The pressure
120 MPa while the sensor which is placed far from the center of
profile at the center of the impact for three formulations and for
impact reaches the peak pressure of 42 MPa (see Fig. 13(a)). After
different elements size are depicted in Fig. 10(a)–(c).
0.7 ms, pressures captured by all sensors diminish. For all sen-
As it can be seen from the figures, all the three methods pre-
sors after the initial peak pressure, a phase of substantially steady
dict relatively close results for fine mesh in comparison to each
flow is developed. This is because after the initial shock waves de-
other as well as to the experimental tests. Anyway for this study
veloped by the first touch between the bird and target, the shock
the SPH method was selected due to the following reasons: The constantly gets weakened. For a low velocity impact (subsonic), the
SPH elements can be defined on the vertexes of same dimensioned shock wave will be weakened to the point that it will disappear.
adjacent cubes, thus, the mesh is uniform and homogeneous in For impact from the head side, the central sensor reaches the
all directions, i.e. the distance between two neighbor nodes is the peak pressure of 104 MPa when other sensors do not reach pres-
same at any position in a simple or complex SPH bird model, while sures more than 25 MPa (Fig. 13(b)). The duration of impact for the
for a complex Lagrangian or ALE bird model the size of elements impact from the head side is much longer than the impact from
at various parts is essentially different, especially for a complex the bottom side. This is because the dimension of a bird is much
geometry. By adopting the SPH bird model, the mesh does not af- longer from head to tail than from the bottom to the top. The cen-
fect the results when the bird impacts the target from its different tral sensor captures an initial pressure peak and then a phase of
parts. Therefore, hereafter in this paper, all the simulations were steady flow. However, other sensors do not capture an initial pres-
conducted using the SPH method. sure peak and in turn, show a steady flow pressure during the
impact process.
5.2. Validation of the realistic bird model For the impact from the tail side, the pressure profile is similar
to that of the impact from the head side (Fig. 13(c)). One of the
In Wilbeck experimental tests, all birds were launched tail lead- differences is that the central sensor captures much lower pressure
ing and impacted the plate in that orientation for increasing the than that of the head side (38 MPa versus 104 MPa) when other
stability of bird after being shot. In order to validate the new bird sensors capture higher pressure peaks. It can be concluded that in
model, the results from the realistic bird impacting the target plate impact from the tail side, the slope of variation of pressure peak
 R. Hedayati, S. Ziaei-Rad / Aerospace Science and Technology 28 (2013) 9–20 15

Fig. 10. Pressure profile at the center of impact for (a) Lagrangian formulation, (b) SPH formulation, and (c) ALE formulation.

Fig. 11. Pressure profile at the center of impact for old and new bird models com- Fig. 12. Distribution of stagnation pressure on the target.
pared to experimental result.

on the plate is much lower than that of the impact from the head The pressure profile for the wing side shows that the pressure
side. It can also be seen that in the impact from tail side, all the is higher for the tail-to-head direction as compared with the torso-
sensors show a phase of substantially steady flow after an initial to-wing direction (Fig. 13(d)). It can also be derived from Fig. 16
peak of pressure. that pressure profiles for all the sensors reach their highest point
16 R. Hedayati, S. Ziaei-Rad / Aerospace Science and Technology 28 (2013) 9–20

Fig. 13. Pressure profiles captured by sensors 1 to 4 for the impact from (a) bottom side, (b) head side, (c) tail side, and (d) wing side.

when the torso of bird starts to impact the target. Sensors 3, 2

and 4 start propagating earlier and their signals diminish later than
other sensors. This is because they are closer to the tip of the tail.

5.4. Pressure peak distribution for impacts with different bird

orientations

In Fig. 14, the peak pressure imposed on the target plate is

plotted versus distance from the center of the impact. The trend
for variations in pressure profile is similar for all the impacts, i.e.
for all the impacts, the pressure peak value is maximum at the
center of target and decreases when the distance from the cen-
ter increases. It can also be seen that the peak pressure profile for
the bird impacting from the tail correlates relatively well with that
of the hemispherical-ended cylinder bird model. It is obvious that
during the impact, in which the bird impacts the target from bot-
tom side, three sensors reach peak pressures of 120 MPa. By com-
paring it with the impacts from other sides, it can be concluded Fig. 14. Peak pressures of sensors in impacts with different orientations and the
that the highest pressure is formed on the target when the bird hemispherical-ended cylinder bird strike modeling.
impacted it from its bottom side. In other words, for other impact
cases, the pressure peak is lower for sensors located far from the impacts the target from the tail side, it imposes the lowest pres-
center of the target. It is also notable that the bottom side, head sure peak than in other impact cases. It can, therefore, be careless
side, hemispherical-ended cylinder model, wing side and tail side to consider the experimental/numerical results from the tail side
impose the higher pressure on the target, respectively. This is an impact as a criterion for designing aircraft components resistant
important point because it can be concluded that when the bird enough against bird strike.
 R. Hedayati, S. Ziaei-Rad / Aerospace Science and Technology 28 (2013) 9–20 17

Based on the FE simulations, it can be concluded that the im-

pact from the bird bottom side is the most damaging scenario,
while the tail side impact is the less dangerous one. The peak pres-
sure for the bottom side is 128 MPa while for the tail side it is
38 MPa. For bottom scenario, three sensors reach pressures more
than 120 MPa, while for the tail scenario three sensors have the
peak pressure of lower than 25 MPa. Therefore, changes in tradi-
tional bird models used for the bird strike analysis seem necessary
in order to better simulate realistic bird strike events.
Although the peak pressure is much lower in tail-side impact
in comparison with the bottom side impact, but the impact dura-
tion in tail-side impact is longer which, on the other hand, can
increase the damage level as well. The shorter the impact du-
ration is, the more vulnerable the impacted structure is to local
penetration. As the impact duration gets longer, the global defor-
mation of the structure becomes more significant. Therefore for
the impact of bird to the structure of an aircraft, the simulations
or tests must be carried out to find out which parameter (pres-
sure peak or impact duration) has more influence on the integrity
of the impacted structure and as a result the worst impact sce- Fig. 15. Various SPH bird models – number of SPH elements.
nario will be found. Since in most bird strike events, the local
penetration is the main reason for damage, therefore it can still
be concluded that the bottom side impact is the most damaging
scenario.
In the next section, four substitute bird models are introduced
and their results are studied. The aim is to introduce a proper
substitute which is simple, accurate enough and needs less com-
putational efforts in comparison to a realistic bird model.

6. Substitute bird model

From Section 5, it was found that the impact from bird bot-
tom side is the most damaging scenario, while the tail side im-
pact is the less dangerous one. In bird strike simulations in which
a bird impacts front facing aircraft components like windshield,
wing leading edge and compressor blades, it is time consuming to
always model a bird with geometry similar to a real bird which
has the organs head, neck and wings. In addition to that, not all
birds have the same body shapes or the same dimensions. There-
fore, a substitute bird model which can be used as a reference bird
model in different situations seems to be necessary. The substi-
tute bird model should be able to predict the pressure distribu-
tion on the targets appropriately when impacting from the tail or
bottom side. Until now, three bird configurations have been pro-
posed as a substitute for real birds and their results have been
compared to Wilbeck’s experimental results [27]. The problem is
that the three substitute results have been compared to exper-
imental results from the tail side impact. In the current study,
four different configurations, namely: sphere, straight-ended cylin-
der, hemispherical-ended cylinder and ellipsoid were selected and Fig. 16. Deformation of different bird configurations at different time intervals for
their results are compared to each other and also to Wilbeck ex- the impact from tail side.
perimental results when impacting from the tail side. The four
configuration results are also compared to the more realistic bird ements. Similar to previous simulations, the null material model
model introduced in Section 5 when impacting from the bottom with Gruneisen equation of state is employed for all models. More-
side and tail side. over, the target and sensor elements have the same dimensions
and the material parameters are the same as the previous simula-
6.1. Bird and target discretization tions.

For all the configurations except for the sphere, a length-to- 6.2. Tail side impact results
diameter ratio of two was selected, as suggested by previous
works [13]. For the considered mass of 0.3 kg, the diameters Since the tail side impact for different bird models has been
0.0647, 0.0626, 0.0588 and 0.0848 m are obtained for ellipsoid, previously discussed by other authors [18,13], the results of tail
hemispherical-ended cylinder, straight-ended cylinder, and sphere, side impact for various substitute bird models will be presented
respectively. Various SPH bird model are shown in Fig. 15. The briefly. Deformation of different bird configurations at different
number next to each bird model shows the number of its SPH el- time intervals is shown in Fig. 16. The deformation of different
18 R. Hedayati, S. Ziaei-Rad / Aerospace Science and Technology 28 (2013) 9–20

Fig. 17. Pressure at the center of impact for Wilbeck experimental result, the realis-
tic bird model and substitute bird configurations.

Fig. 18. The peak of pressure imposed on the target plate versus distance from the Fig. 19. Deformation of different bird configurations at different time intervals for
center of impact for the tail side impact. the impact from bottom side.

bird configurations is different at the initial moments of impact, ended cylinder is 80 MPa, while for the realistic bird model and
but then after the middle of impact, they deform in a similar way. Wilbeck experiment it is 38 MPa and 24 MPa, respectively.
In Fig. 17, pressure at the center of impact is compared for
the Wilbeck experimental result, the realistic bird model and four 6.3. Bottom side impact results
substitute bird configurations. As it can be seen, straight-ended
cylinder and sphere predict pressure profiles much higher than As concluded earlier in the paper, the bottom side impact is
Wilbeck experimental result. Ellipsoid, also, predicts a high pres- the most damaging scenario in a bird strike event. Deformation of
sure peak, although its peak pressure is more accurate than that different bird configurations at different time intervals is shown in
of sphere and straight-ended cylinder. Among the four bird con- Fig. 19.
figurations, the hemispherical-ended cylinder predicts a pressure Pressure–time curves captured by four sensors for different sub-
peak closer to the experimental result. It is notable that the re- stitute bird configurations are plotted in Figs. 20(a)–(c). Pressure
alistic bird model, predicts the best pressure peak in comparison profile distribution for the ellipsoid is shown in Fig. 20(a). In the
with other numerical models. It is because its initial contact area impact of ellipsoid bird model to the rigid target, the three cen-
is more similar to a real bird than that of other geometries. tral sensors reach their higher peaks at the same time while the
In Fig. 18, the peak of pressure imposed on the target plate is fourth sensor reaches its highest value after a short delay. It is also
plotted versus distance from the center of impact. The trend for notable that the central sensor has the pressure peak of 155 MPa
variations in pressure peak is similar for all the impacts, i.e. for all while the three other sensors show a maximum pressure of ap-
the impacts, the pressure peak is maximum at the center of tar- proximately 120 MPa. All the sensors show a phase of steady state
get and it decreases when the distance from the center increases. show after the initial shock phase.
Again, the straight-ended cylinder and the sphere predict pressure In Fig. 20(b), pressure profiles recorded by sensors 1 to 4 are
distribution much higher than the realistic bird model. Ellipsoid shown for hemispherical-ended cylinder. As it can be seen, all
and hemispherical-ended cylinder correlate better with Wilbeck sensors reach their highest values at the same time. This can be at-
experiment, while the hemispherical-ended cylinder seems to be tributed to the shape of hemispherical-ended cylinder which is flat
the most accurate model among presented substitute bird models. at sides and as a result starts to touch all the sensors at the same
The peak pressure at the center of impact for the hemispherical- time. Sensors 1 to 3 have the maximum pressures of more than
 R. Hedayati, S. Ziaei-Rad / Aerospace Science and Technology 28 (2013) 9–20 19

Fig. 20. Pressure profiles captured by sensors 1 to 4 for the impact of (a) ellipsoid, (b) hemispherical-ended, (c) straight-ended bird model.

140 MPa, while the 4th sensor has the peak pressure of 95 MPa
only.
In Fig. 20(c), pressure profiles recorded by sensors 1 to 4 are
shown for the case of straight-ended cylinder. In this impact, like
the hemispherical-ended cylinder, all the sensors reach their high-
est values at the same time. This is because straight-ended cylin-
der, like the hemispherical-ended cylinder, has a flat surface in the
longitudinal side. Again, sensors 1 to 3 have the maximum pres-
sures of more than 120 MPa, while the 4th sensor has a pressure
peak of 70 MPa.
In Fig. 21, distribution of maximum pressures for different bird
substitutes are shown and compared to each other, as well as to
that of the realistic bird impacting from its bottom side. The sphere
configuration pressure distribution is different from other config-
urations. It also shows a significant deviation from the realistic
bird model. Therefore, it can be concluded that it is an unsuit-
able substitute for real birds. Hemispherical-ended cylinder and
straight-ended cylinder both predict peak pressures higher than Fig. 21. Distribution of maximum pressures for different bird substitutes and the
the realistic bird model. On the other hand, the ellipsoid bird realistic bird impacting from bottom side.

model maximum pressure distribution shows a good correlation

with that of the realistic bird model. One may conclude that the pacts from both the tail and bottom side. Comparing the two re-
ellipsoid bird model can more precisely model the pressure distri- maining configurations, the hemispherical-ended cylinder creates
bution imposed on the target. more accurate result when impacting from the tail side, while
According to the results from impacts from tail and bottom it shows inferior results in the impact from the bottom side.
side by different bird configurations, it can be concluded that the On the other hand, the ellipsoid correlates well with the experi-
sphere and straight-ended cylinder have imprecise results in im- mental result, when impacting from the bottom side, but does not
20 R. Hedayati, S. Ziaei-Rad / Aerospace Science and Technology 28 (2013) 9–20

indicate a comparatively accurate result when impacting from the [3] R.A. Brockman, T.W. Held, Explicit finite element method for transparency im-
tail side. pact analysis, University of Dayton Research Institute Report WL-TR-91-3006,
1991.
[4] M. Chizari, L.M. Barrett, S.T.S. Al-Hassani, An explicit numerical modeling of the
7. Conclusions water jet tube forming, Comput. Mater. Sci. 45 (2009) 378–384.
[5] E. Cleary, Wildlife strikes to civil aircraft in the United States 1990–2004, Fed-
In experimental tests of bird impact, birds are usually launched eral Aviation Administration National Wildlife Strike Database, No. 11, 2005.
[6] J. Frischbier, Bird strike capability of a transonic fan blisk, in: Proceedings of
tail leading and are impacted to the target in that orientation for
the ASME Turboexpo, Orlando, FL, June 1997, pp. 2–5.
increasing the stability of shooting. On the other hand, in the real [7] Y.P. Guan, W. Chen, Z.Y. Huang, Sliced model for bird impacting blades, J. Nan-
bird strike events the birds do not always impact a surface from jing Univ. Aeronaut. Astronaut. 36 (6) (2004) 784–786.
the tail side orientation. In this study, using a SPH realistic bird [8] A.G. Hanssen, Y. Girard, L. Olovsson, T. Berstad, M. Langseth, A numerical model
model, the effect of orientation in bird strike events is investi- for bird strike of aluminium foam-based sandwich panels, Int. J. Impact Eng. 32
(2006) 1127–1144.
gated. Based on the FE simulation, it can be concluded that the [9] P. Hut, L. Hernquist, G. Lake, J. Makino, S. McMillan, T. Sterling, Smooth par-
impact from bird bottom side is the most damaging scenario while ticle hydrodynamics: Models, applications, and enabling technologies, in: Pro-
the tail side impact is a less dangerous one. The pressure peak for ceedings form the Workshop Presented by the Institute for Advance Study at
the bottom side impact is about three times of that in the impact Princeton, 1997.
[10] J. Lacome, Smooth Particle Hydrodynamics (SPH): A new feature in LSDYNA,
from tail side. Generally it was found that the bottom side, head
in: Proceedings of the 6th International LS-DYNA Users Conference, Session 7-
side, hemispherical-ended cylinder model, wing side and tail side 3, 2000.
impose the higher pressure on the target, respectively. This is an [11] J.L. Lacome, Smoothed particle hydrodynamics – Part II, in: FEA Information In-
important point because it can be concluded that when the bird ternational News for the World-Wide Engineering Community, 2001, November
impacts the target from the tail side, it imposes the lowest pres- 27, pp. 6–11.
[12] B. Langrand, A.S. Bayart, Y. Chauveau, E. Deletombe, Assessment of multi-
sure peak than in other impact cases. It can, therefore, be careless physics FE methods for bird strike modelling—Application to a metallic riveted
to consider the experimental/numerical results from the tail side airframe, Int. J. Crashworth. 7 (4) (2002) 415–428.
impact as a criterion for designing aircraft components resistant [13] M.A. Lavoie, A. Gakwaya, M. NejadEnsan, D.G. Zimcik, Validation of available
enough against bird strike. Therefore, changes in traditional bird approaches for numerical bird strike modeling tools, Int. Rev. Mec. Eng. 1
(2007) 380–389.
models used for bird strike analysis are necessary in order to bet-
[14] Livermore software Inc., LS-DYNA Keyword User’s Manual, ver. 9.71, Livermore,
ter simulate real bird strike events. CA, 2006.
In bird strike simulations in which a bird impacts aircraft com- [15] L.B. Lucy, A numerical approach to the testing of the fission hypothesis, Astron.
ponents such as windshield, wing leading edge or compressor J. 82 (1977) 1013–1020.
blades, it is time consuming to always model a bird with ge- [16] N.F. Martin, Nonlinear Finite Element Analysis to Predict Fan Blade Impact
Damage, Pratt & Whitney, 2004.
ometry similar to a real bird which has the organs head, neck [17] M.A. McCarthy, J.R. Xiao, C.T. McCarthy, A. Kamoulakos, J. Ramos, J.P. Gallard,
and wings. In addition to that, not all the birds have the same Modelling of bird strike on an aircraft wing leading edge made from fiber
body shapes nor the same dimensions. Therefore, a substitute bird metal laminates—Part 2: Modeling of impact with SPH bird model, Appl. Com-
model which can be used as a reference bird model for differ- pos. Mater. 11 (5) (2004) 317–340.
[18] S.A. Meguid, R.H. Mao, T.Y. Ng, FE analysis of geometry effects of an artificial
ent problems is necessary. Four substitute bird models, namely,
bird striking an aero engine fan blade, Int. J. Impact Eng. 35 (2007) 487–498.
sphere, straight-ended cylinder, hemispherical-ended cylinder and [19] T.J. Moffat, W.L. Cleghorn, Prediction of bird impact pressures and damage us-
ellipsoid were presented and their results were analyzed. It was ing MSC/DYTRAN, in: Proceedings of ASME TURBOEXPO 2001, Louisiana.
found that for the tail side impact scenario, the hemispherical- [20] J. Monaghan, On the problem of penetration in particle methods, J. Comput.
Phys. 82 (1) (May 1989) 1–15.
ended cylinder shows the best results, while for the bottom side
[21] E. Niering, Simulation of Bird Strikes on Turbine Engines, Turbinen-Union Mu-
impact scenario, the ellipsoid can be the best candidate for the nich, Germany, 1988.
bird substitute model. If it is necessary to choose only one of the [22] B. Richard, The development of a substitute artificial bird by the International
configurations to model impacts from different orientations, it is Bird strike Research Group for use in aircraft component testing, International
suggested to select the ellipsoid. It is because of two reasons: first, Bird Strike Committee ISBC25/ WP-IE3 (2000), Amsterdam.
[23] C. Stoker, Developments of the arbitrary Lagrangian–Eulerian method in non-
in the tail side impact of ellipsoid, the pressure peak is inaccurate
linear solid mechanics, Enschede, The Netherlands, 1999.
only for the central sensor, while other sensors show good correla- [24] F. Stoll, R.A. Brockman, Finite element simulation of high speed soft-body im-
tion with experimental data. Second, most birds have an ellipsoid pacts, in: Proceedings of the 1997 38th AIAA/ASME/ASCE/AHS/ASC Structure,
shaped body rather than hemispherical-ended cylinder. Structural Dynamics, and Materials Conference, Part 1 (of 4), Kissimmee, FL,
USA, 1997, pp. 334–344.
[25] J. Thorpe, Fatalities and destroyed civil aircraft due to bird strikes, 2002 to
References 2004, International Bird Strike Committee, Athens, May 2005, pp. 23–27.
[26] L.C. Ubels, A.F. Johnson, J.P. Gallard, M. Sunaric, Design and testing of a compos-
[1] A. Airoldi, B. Cacchione, Modelling of impact forces and pressures in Lagrangian ite bird strike resistant leading edge, in: SAMPE Europe Conference Exhibition,
bird strike analyses, Int. J. Impact Eng. 32 (2006) 1651–1677. National Aerospace Laboratory, 2003.
[2] J.P. Barber, H.R. Taylor, J.S. Wilbeck, Characterization of bird impacts on a rigid [27] J. Wilbeck, Impact behavior of low strength projectiles, Report No. AFML-TR-
plate: Part I, University of Dayton Research Institute, Technical Report AFFDL- 77–134, Air Force Materials Lab., Air Force Wright Aeronautical Lab’s, Wright-
TR-75-5, 1975. Patterson Air Force base, OH, 1977.