Design of aircraft structures against threat of bird strikes

Jun LIU,Yulong LI,*,Xinheng YU,b,Xiosheng GAO,Zongxing LIU

aSchool of Aeronautics,Northwestern Polytechnical University,Xi’an 710072,China

bDyson School of Design Engineering,College of Engineering,Imperial College,London sw71na,UK

cDepartment of Mechanical Engineering,the University of Akron,Akron,OH 44325,USA

KEYWORDS Bird strike;Cover sheet;Design of aircraft structures;Horizontal tail;Vertical tail

Abstract In this paper,a method to design bird-strike-resistant aircraft structures is presented and illustrated through examples.The focus is on bird strike experiments and simulations.The explicitfinite element software PAM-CRASH is employed to conduct bird strike simulations,and a coupled Smooth Particles Hydrodynamic(SPH)and Finite Element(FE)method is used to simulate the interaction between a bird and a target structure.The SPH method is explained,and an SPH bird model is established.Constitutive models for various structural materials,such as aluminum alloys,composite materials,honeycomb,and foam materials that are used in aircraft structures,are presented,and model parameters are identified by conducting various material tests.Good agreements between simulation results and experimental data suggest that the numerical model is capable of predicting the dynamic responses of various aircraft structures under a bird strike,and numerical simulation can be used as a tool to design bird-strike-resistant aircraft structures.?2018 Chinese Society of Aeronautics and Astronautics.Production and hosting by Elsevier Ltd.Thisis anopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1.Introduction

Bird strikes have long been a significant threat to aviation safety.The first bird strike dated back to September 7,1905,recorded by the Wright brothers.1The United States Air Force reported 13427 bird/wildlife strikes to aircraft worldwide between 1989 and 1993,and estimated the damage to civilian and military aircraft to cost hundreds of millions of dollars every year.Most bird strikes occurred when an aircraft was at a lo2w altitude during the take-off and landing phases of a flight.The most vulnerable components to bird strikes are aircraft engines,nose,and wings/empennages,3,4as shown in Fig.1.To ensure flight safety,aviation regulations require a certain level of bird strike resistance for critical components.As specified in Part 25 of Federal Aviation Regulations,an airplane must be capable of successfully completing a flight during which likely a structural damage occurs as a result of impact with a 4 lb(1 lb=0.4536 kg)bird(8 lb for an empennage structure)when the velocity of the airplane relative to the bird along the airplane’s flight path is equal toVc(design cruise speed)at the sea level or 0.85Vcat 8000 ft(1 ft=0.3048 m),whichever is more critical.Similar requirements have also been included in the China Civil Aviation Regulations(CCAR)and the certification specifications CS-25(Large Airplanes)by the European Aviation Safety Agency.5–7This calls for continuous efforts to design bird-strike-resistant aircraftcomponentsthrough acombinationoftestsand simulations.

Fig.1 Illustration of aircraft components vulnerable to bird strikes4.

Most early researches on bird strikes were experimental studies.Barber et al.8were among the pioneers to conduct bird strike experiments.They performed bird impact tests on a rigid plate,and found that the peak pressure generated during the impact was proportional to the square of the impact velocity,independent of the bird geometry.While a bird impact experiment provides a direct method to examine a component’s bird-strike resistance,with rapid development of the computer technology and thefinite element method,numerical simulation is proven as a viable alternative method to design and certify a bird-strike-resistant component more economically.Airoldi and Cacchione9investigated the performance of Lagrangian bird models,considering different bird material characteristics and focusing on numerical modeling of the pressure distributions on a target.Vijay et al.10,11developed various bird models,and compared Lagrangian,arbitrary Lagrangian Eulerian(ALE),and Smooth Particle Hydrodynamic(SPH)methods to simulate a bird-strike event.Lakshmi and Walter12studied the effect of the equation of state models of a bird material on the predicted impact load.Lavoie et al.13performed bird impact tests using gelatin as a bird substitute to validate numerical models,and promoted the use of numerical tools in the aircraft design and certification process.Hanssen et al.14performed bird-strike tests on a double-sandwich panel,and developed a numerical model to simulate the test process.Their result showed that the Finite Element(FE)method was able to represent failure of both aluminum cover plates and aluminum foam cores.McCarthy et al.3,15studied bird strikes on an aircraft wing’s leading edge made from two kinds of fiber metal laminates with different lay-ups,using SPH bird and various material models.More recently,Georgiadis et al.16established a simulation methodology to support the birdstrike certification of a moveable trailing edge of the Boeing 787 Dreamliner,made of a carbon fiber epoxy composite.In their work,the explicitfinite element software PAM-CRASH was selected to perform simulations where the bird was modeled with the SPH method and the joints were represented by different advanced fastener elements.

The current research on bird-strike simulation is mainly focused on the following three aspects:

(1)Constitutive models for a bird.The bird body is not a homogeneous medium.It includes bones,meat,feathers,etc.,and thus an equivalent model must be established.Moreover,the bird shows different mechanical behaviors under different impact velocities:under low-speed impact,the bird can be considered as an elastic–plastic body,while under high-speed impact,its behavior is similar to that of afluid.17

(2)Mechanical behaviors of aeronautical materials under high strain rates.A bird strike is a rapid process,which often occurs in a few mini seconds,and as a result,the material under a bird strike may experience intermediate to high strain rates.For many materials,the strain rate can significantly affect their mechanical behaviors,which in turn will influence the dynamic response of the structure.

(3)Numerical techniques to simulate the bird strike process.Three methods have been used to simulate the dynamic response of aircraft structures under a bird strike:the uncoupled method,the coupled method,and the SPH method.The uncoupled method disregards the bird and treats the impact load as a known quantity to analyze the structural response.The coupled method models both the bird and the structure,and simulates the birdstructure interaction by applying a contact-impact coupling algorithm or afluid–structure coupling algorithm.The SPH method is a meshless method and capable of solving large-deformation and nonlinear dynamic problems.Technically speaking,the SPH method is a special case of the coupled method.18

Three basic design principles can be employed to protect aircraft structures against bird strikes:

(1)To improve the energy absorption ability of a structure.When the target structure is almost perpendicular to the bird strike direction,such as the aircraft nose or engine,all the impact energy would be applied to the structure.In such cases,the energy absorption ability of the structure determines its anti-bird strike performance.Energy absorbing materials,such as honey comb and foam,are usually adopted to improve the structure’s energy absorption ability.

(2)To improve the energy dissipation ability of a structure.When the target structure is thin or not perpendicular to the bird strike direction,such as the horizontal tail and the windshield,much of the impact energy can be dissipated if the impact direction can be changed to some extent.This requires novel structural design and matching structural stiffness.

(3)To adopt the multi-layer protection design.Some components,such as the nose and the vertical tail,contain important pipelines and flight control systems,so birdstrike damage can lead to extremely dangerous consequences.These components are usually designed as multi-layer structures to improve the bird-strike resistance.

In this paper,a method to design bird-strike-resistant aircraft structures is presented and illustrated through examples.The structure of the paper is as follows.In Section 2,methods to conduct bird strike simulations and experiments are presented.Details of the SPH bird model and constitutive models for various structural materials,such as aluminum alloys,composite materials,and honeycomb and foam materials,are described.Sections 3–5 show three examples,where designs of a composite vertical tail structure,a cover sheet behind the radome of an aircraft,and a horizontal tail structure are modified to improve their anti-bird strike performances.Finally,some concluding remarks are made in Section 6.

2.Design method of structure impacted by bird

Fig.2 shows a flowchart for designing an aircraft component subject to the threat of a bird strike.19In the first step,an initial design is made by considering regular static and dynamic loads.Next,numerical simulations of a bird strike are conducted to investigate the structure’s resistance to the bird strike.During this process,constitutive models of the bird and structural materials need to be developed,and model parameters need to be calibrated and verified by tests.Based on the damage prediction from numerical simulations,improvements to the design are made,and the process iterates until the predicted damage is within the allowed tolerance.Last but not least,physical tests of the component under various conditions,including a bird strike,need to be conducted to validate the design.If the design does not pass these validation tests,it will have to be revised,and the above design process will have to start over.This paper focuses on the second step–bird strike simulation and damage prediction.

2.1.Numerical model for bird

2.1.1.SPH bird model

We limit our attention to a high-speed bird strike.During the high-speed impact process,the bird is observed to be highly distorted and crushed into pieces,behaving likefluids.20This kind of behavior is difficult to simulate using the conventional FE model with a Lagrangian mesh as large mesh distortion leads to severe stability problems.21The SPH method is a meshless method,which overcomes this difficulty,and is thus adopted to simulate the bird in this paper.

Fig. 2 Flowchart for designing bird-strike-resistant component.19

SPH is a Lagrange particle method introduced by Lucy in the 1970s22in order to solve hydrodynamic problems in astrophysical contexts.It has been extensively applied in the study of accretion disc,galaxy dynamics,and star collisions among other problems.This method has been implemented in the commercialfinite element software PAM-CRASH.23A smooth particle is input like a 3-DOF(degree-of-freedom)solid element and defined by its center of mass,volume,part number,and domain of influence.It can be used with great advantages to model bulk materials with no cohesion(sand,liquid,gas)or in situations where perforation or mixing is expected.Note,however,that it can be much more expensive to use than a classical solid element.A smooth particle,similar to afinite element,has its own shape functions,reconstructed at each cycle from its dynamic connectivity.Localization and information transmission from one particle to another are achieved through the notion of an interpolation distance called the smoothing length.ParticleJis said to have a contributing neighbor ParticleI,when ParticleIlies within the sphere of influence of ParticleJ.In such a case,ParticleIis said to be connected to the central ParticleJ.The sphere of influence of a particle is a multiple to its smoothing length.The multiplication factor depends on the type of kernel used to construct the smooth particle’s shape functions.Details of the kernel functions are given in Ref.24ParticlesIandJas well as the SPH bird model are shown in Fig.3.

Structural nodes and smooth particles must have distinct numbers.Smooth particles can be subjected to constraints or loads as if they are nodes.In particular,finite element-tosmooth particle coupling is achieved by the use of penalty contacts,where the particles are slave nodes.25The interaction between the particles and thefinite elements may be modeled by the existing sliding interface algorithms available within PAM-CRASH.The most frequently used sliding interface types to relate smooth particles tofinite elements are type 34 in case that the materials may be considered as sliding with dynamic contact or as a tied interface when the particles are assumed to stick to thefinite element surface.For afinite element mesh,the contact thickness indicates the distance away from a contact face where a physical contact is established.For particles interacting withfinite elements,the contact thickness should be representative of the particle radius,possibly augmented with the half-thickness of the shell structure.Therefore,the node to the surface contact is used to model the interaction between the SPH particles and the Lagrange elements.

2.1.2.Constitutive model of bird

Fig.3 Particles I and J in SPH and SPH bird model.

The bird geometry is approximated as a hemispherical ended cylinder in the present paper.Barber et al.26measured the den-sities of birds of various sizes,and obtained an average value of about 950 kg/m3.This value has been adopted as the bird density in many published studies15and is also adopted here.

It has been shown that at high speed,the bird can be considered as a homogeneous jet offluid impinging a structure.Therefore,it is adequate to use an Equation of State(EOS)to model the bird material.The SPH method and the Murnaghan equation of state implemented in PAM-CRASH provide a tool to simulate the bid material at high-speed impact.The EOS is given by

wherep1is a pressure,p0is a reference pressure,Band γ are constants to be determined,15and ρ/ρ0is the ratio of the current mass density to the initial mass density.

The material constants in the EOS are difficult to obtain as they cannot be measured directly.McCarthy et al.15performed bird strike tests on flat plates to identify the most suitable material parameters by matching the responses of a number of pressure sensors using the program PAM-OPT.The parameters determined by McCarthy et al.15are used in this study,i.e.,B=128 MPa and γ=7.98.

2.2.Constitutive models for structural materials

Commonly used materials for aircraft structures are aluminum alloys,composite materials,and honeycomb materials.The maximum strain rate in the process of a bird strike at the designed velocity(180 m/s)can reach as high as 300 s-1,and thus the strain rate effect should be taken into account.In the following,material models for aluminum alloys and composite materials are described in detail,and model parameters are identified through various material tests.And the model parameters for honeycomb materials are referenced and listed.

2.2.1.Aluminum alloys

An elastic–plastic material with isotropic damage for thin shell elements,corresponding to material model 105 in PAMCRASH,is adopted to describe four different aluminum alloys(Al6061-T4,Al7050-T7451,Al7075-T6,and Al2024-T3)considered in this study.In material model 105,the plastic response is described by the strain rate dependent Cowper-Symonds law as

wherea+b(εp)nrepresents the static yield stress including strain hardening,in whichais the yield stress,εpis the effective plastic strain,andbandnare materials constants.The strain rate effect is included by the factor 1+（˙ε/D）1/p,which depends on the strain rate˙ε and the control parametersDandp.27

In order to obtain the material parameters of these four different aluminum alloys,uniaxial compression tests are performed,including static compression tests and dynamic Hopkinson bar tests.The material constants have been obtained by curve-fitting test data.20,21A failure strain criterion is used to model the damage of the structure.Once the equivalent tensile strain in an element reaches the critical failure strain(εfail)of the material,the element will be deleted.All the necessary parameters needed to define the material models of the four aluminum alloys are listed in Table 1.

2.2.2.Composite materials

The strain rate effect on the mechanical behavior of a carbon fiber is shown to be significant,which is also considered in the simulation.The failure and damage modeling of a composite material is achieved by multilayered shell elements in PAMCRASH.A single composite ply is modeled with the unidirectional composite global ply model(the modified Ladeve`ze composite model,27whose properties are degraded as loading continues due to progressive damage processed prior to ultimate failure.Material parameters are defined with respect to the coordinate system shown in Fig.4(a).The basic constitutive relationship of the modified Ladeve`ze model can be expressed as follows:

(1)In the fiber direction(1-direction),if ε11≥ 0,i.e.,the composite ply is under tension,then

and if ε11＜ 0,i.e.,the composite ply is under compression,then

(2)In the transverse direction(2-direction),if ε22≥ 0,i.e.,the composite ply is under tension,then

Table 1 Material parameters for four aluminum alloys.

Fig.4 Frame definitions of composite material and honeycomb material.

and if ε22＜ 0,i.e.,the composite ply is under compression,then

(3)The degradation of the shear modulus is described by

wheredftanddfcare damage parameters for fibers,anddandd′are damage parameters for the matrix.Detailed descriptions of the modified Ladeve`ze model,model parameters,and damage evolution equations can be found in Ref.27

CCF300 carbon fibers are used for the composite materials in this study.Material data is obtained by carrying out various compressive and tensile static tests as well as dynamic tests.Different strain rates are selected for the dynamic tests so that the strain rate effect can be identified.The material parameters of unidirectional CCF300 carbon fibers are listed in Table 2.

2.2.3.Honeycomb and foam materials

To simplify the simulation,honeycomb materials are modeled ascontinuum solid elementswith orthotropicmaterial properties.Material type 42 in PAM-CRASH is employed,representing an improved material model for impact.The orthotropic material frame definition is shown in Fig.4(b).A detailed description of material type 42 can be found in Ref.27In order to model the Nomex honeycomb material,quasi-static compressive tests in three directions,L,T,andW,as well as shear tests in three planes,L-T,L-W,andT-W,havebeen found in Ref.28.Fig.528showsthe compressive and shearstress–strain curvesin different directions and planes.The material parameters for this material are listed in Table 3.

The aluminum honeycomb material used in this study is the Divinycell F50 honeycomb,with a density of 50 kg/m3.According to the parameters provided by the manufacturer,the material constants for this material are given in Table 4.

The aluminum foam used in this paper is manufactured by blowing gas into a molten aluminum alloy.Its density is 275 kg/m3,and porosity is 90%.Material type 35 in PAMCRASH is employed to model this material.The material properties are listed in Table 5 according to the parameters provided by the manufacturer.

2.2.4.Models for joints

Fastened joints,such as rivets and bolts,are represented by the PLINK elements in PAM-CRASH.It has been found that the PLINK model is a convenient method with an adequate accuracy for modeling fastened composite joints in crash and impact simulations.30The PLINK is a very beneficial element,which is mesh-independent and capable of joining up to five layers per element.To ensure a stable response,the penalty stiffness of a joint is internally calculated,but can be modified by user input if necessary.16The rupture model of the PLINK element used in this task is shown as follows:

where the normal forceNand shear forceTare calculated in the simulation process.The maximum normal forceNmaxand shear forceTmax,as well asn,m,and φ,are user input parameters.The following parameters are taken from Ref.15:Nmax=5100 N,Tmax=3200 N,n1=1.5,m=2.1,and φ=1.

Bonded and so-cured joints can be modeled by using TIED elements,coupled with a cohesive zone fracture model.16The TIED model provides a satisfactory and relatively-simple way to model such fasteners.

2.3.Bird-strike test

The simulation result must be validated by bird strike tests,which should be conducted according to certain standards and aviation regulations,which specify the impact direction,velocity,and the weight of the bird.

To maintain the impact direction for multiple tests,a specimen is usually mounted on a steelfixture designed to withstand loads expected from an impact.The steelfixture is connected to a heavy support attached to a concrete ground.Different bolts are used to connect the specimen to the steelfixture and tofix the steelfixture to the ground.

Soft bodies such as fowls and gelatin are used by the aircraft industry as substitute birds,and are observed to flow over a structure during impact, spreading the load over a significant surface area which limits local impact damage.30In this study,a fowl,which has physical properties closest to those of living birds,is used as the substitute bird impactor.Depending on the different types of target,an 8 lb or 4 lb impactor is used.To prepare the impactor,a fowl weighing more than 8 lb or 4 lb is killed just before a test.Some meat and bones in the legsare removed in order to ensure the mass of the projectile being 8 lb or 4 lb.The bird is then packeted using a nylon bag to make the impactor,as shown in Fig.6.A pneumatic gun,shown in Fig.7,launches the bird held inside a sabot,and a sabot stopper traps the sabot when it reaches the end of the barrel so that only the bird is launched to impact the target at the designed velocity.

Table 2 Material parameters for unidirectional CCF300 carbon fibers.

The effective bird impact velocity is measured by laser sensors.A high-speed video camera is also set up as an auxiliary speed trap to monitor the velocity of the projectile.Another high-speed video camera is aimed at the impact region to capture the impact process.

The entire bird strike process is very short,lasting only 3–5 ms.Therefore,a high-speed data acquisition system must be used to collect data for strain,displacement,and force responses.The high-speed data acquisition system usually includes strain sensors and super dynamic strain gauges,displacement sensors and amplifier,load cells,an impedance variation device,and a transient recorder.The logical connections among these devices are shown in Fig.8.Displacement and strainsignalsarecollected withasamplingfrequencyof8 MHz.

3.Test 1:Design of a composite vertical tail

3.1.Test

Fig.5 Compressive and shear stress–strain curves of Nomex honeycomb28.

Table 3 Material parameters for Nomex honeycomb.

Table 4 Material constants for aluminum alloy honeycomb.

Table 5 Mechanical properties of aluminum alloy foam.

The component considered in this example is a simplified vertical tail structure.Fig.9 shows the detailed configuration of this representative vertical tail structure.With about 1000 mm in length,it consists of a metal skin,a leading edge skin,an auxiliary beam,a front beam,two ribs,attaching frames,and upper and lower skins.These structural items are made from different materials:the metal skin is made of the 6061-T4 aluminum alloy,the auxiliary beam and ribs are made of the 7050-T7451 aluminum alloy,and the attaching frames thatfix the test structure to the steel testfixture are made of the 7075-T6 aluminum alloy.The thicknesses of these components vary between 2.5 mm and 3 mm.For the sake of lightweight design,the front beam is made of CCF300 carbon fibers,with a thickness of 4.135 mm.The leading edge skin and the upper and lower skins are sandwiched structures,consisting of CCF300 carbon fibers face-layers and a core made of the Nomex honeycomb,as shown in Fig.10.The thicknesses of the outer and inner layers are 1.2 mm and 1.68 mm,respectively.The outer layer has 10 plies with a[(±45)2/(0,90)/(±45)/(0,90)]lay-out,and the inner layer has 14 plies with a[(0,90)/(±45)/(0,90)/(±45)/(0,90)/(±45)2]layout.The thickness of the honeycomb core is 15 mm.

The test structure is mounted horizontally on a steelfixture,and the steelfixture is connected to a heavy support attached to a concrete ground,as shown in Fig.11(a).The impact point,shown in Fig.11(b),is approximately 300 mm away from the lower rib side,while the angle between the leading edge and the strike direction is about 49.6°.For the vertical tail structure,the specified bird strike velocity is 180 m/s.

Fig.12 shows the locations where the displacement(D)and strains(S1–S6)are measured.Dis at the mid-point of the vertical axis,which is 50 mm left from the right edge of the auxiliary beam.S2 is at the mid-point of the metal skin.S1 is 250 mm right to S2 on the horizontal axis,and S3 is 250 mm left to S2 in the opposite direction.S4 and S5 are placed on the auxiliary beam with a gap of 200 mm to the right edge of the auxiliary beam.S6 is bonded to the mid-point of the outer layer on the upper skin with a distance of 300 mm from the left boundary of the test structure.

3.2.Simulation results vs test data

3.2.1.Finite element model

Fig.6 Production process of bird projectile.

Fig.7 Pneumatic gun for launching bird projectile.

Fig.8 Logical connections of data acquisition equipment.

Fig.9 configuration of vertical tail.

Fig.10 Sandwiched structures of leading edge skin and lower skin.

Fig.11 Composite vertical tail structure and schematic diagram of bird impact test.

Fig.12 Locations where displacement and strains are measured.

Thefinite element model includes all the components,as shown in Fig.13,and consists of 9622 solid elements and 39262 shell elements with refinement in the highly loaded areas.In simulation,the structure is mounted horizontally on a steelfixture,and the steelfixture is connected to a heavy support attached to a concrete ground.The steelfixture is meshed by beam elements in order to be coincident with the real boundary condition.It has long been accepted that increasing the mesh density in such areas can increase deformability and allow for a reduction of the contact interface stiffness.21The highly loaded areas include the metal skin,the leading edge skin,and the auxiliary beam.The average element size of the conventional shell element is 10 mm,while the average element size in the refined area is set to be 5 mm.

Fig.13 FE model of test structure and SPH bird model.

Fig.14 Comparison between strain vs time responses obtained from three differentfinite element meshes.

Fig.15 Comparison of simulated bird strike process with pictures taken with a high-speed camera.

The bird used in the test weighs 3.624 kg(7.99 lb).The actual impact velocity of the bird is measured as 184 m/s,which is very close to the designed velocity.Thus this value is defined as the bird strike velocity in the numerical simulation.

A convergence study is conducted by developing models with differentelementsizesand comparingthemodel responses.Three different element sizes(5,10,and 12 mm)for the highly loaded and heavily deformed areas are chosen.Fig.14 shows the strain vs time responses at the same location,which is on the auxiliary beam with a gap of 200 mm to the right edge,obtained using the three meshes.Three curves coincide,indicating that a converged solution has been obtained with the current mesh refinement.

3.2.2.Deformation of leading edge and damage of auxiliary beam

Fig.15 compares the deformation of the leading edge during the bird-strike process resulted from the numerical simulation with what has been captured by the side-on high-speed camera during the test.The deformation behavior of the SPH bird appears to be in good agreement with the deformation of the real bird in the test.Fig.16 compares the final structural damage between the test and the simulation.To show the internal damage,the leading edge of the FE model has been removed from the figure.The shape and size of the penetration hole in the simulation are very close to those in the test.These results show that the deformation and damage predicted by the simulation are in good agreement with the test results.

Fig.16 Comparison of final structural damage between test and simulation.

Fig.17 Comparison of displacement vs time response between test and simulation.

3.2.3.Displacement and strains

Fig.17 shows the comparison of the displacement vs time response between the simulation and the test.The simulation result is in good agreement with the test data before 2.7 ms.However,after 2.7 ms,the two curves appear to show different trends.This difference is due to the horizontal displacement of the measurement point.Since the displacement is obtained by non-contact measurement,when the structure deforms and causes a horizontal displacement,the measurement point is no longer the originally designed point.Another possible factor that contributes to this difference is the effect of the damping of the structure,which is difficult to define in the simulation.A comparison of the strain vs.time response between the simulation result and the test record is shown in Fig.18.The strain data at S1 is not available because this location is close to the impact point,and the strain gauge was damaged right after the bird touched the leading edge.For the strain at S2,the test data is only valid during the first 1.4 ms,after which the strain gauge was disconnected by the shock wave caused by the high-speed impact.For most of the strain vs time responses,there is a good overall agreement between the simulation results and the test records.

The good agreement between the simulation results and the test data indicates that the numerical model established here is capable of predicting the dynamic response of the vertical tail structure under a bird strike.Therefore,it can be used as a tool to design the vertical tail structure.

3.3.Improvement of design

Structural damage is inevitable during a bird strike.However,a good design should ensure that an airplane still has the capability of further safe flight and landing,and keep important pipelines and the control system intact after an accidental impact.As for the vertical tail,important components and operating systems are placed behind the front beam,in front of which the leading edge and the auxiliary beam are located.As a result of this layout,the front beam should not be penetrated,which requires the auxiliary beam to withstand the impact load.

Under the above prerequisite,the anti-bird strike performance of the vertical tail depends mainly on the leading edge and auxiliary beam structures,whose performances are determined by their configurations and materials.This section focuses on investigating the configuration of the auxiliary beam on the anti-bird strike performance of the vertical tail to develop better auxiliary beams with different configurations.While the configuration of the auxiliary beam changes,other components of the vertical tail structure remain the same as in the original structure.

The auxiliary beam of the current test structure is made of the Al7050-T7451 aluminum alloy,and its configuration is shown in Fig.19(a).On the back of the auxiliary beam,there are five stringers,of which the outer two stringers are designed to be riveted with ribs.These stringers are employed to strengthen the structure.However,they cannot withstand bird-strike load very well.Fig.19(b)gives the equivalent stress distribution of the original auxiliary beam before it was penetrated.The maximum equivalent stress occurs in the flat area,where penetration also starts.For purposes of reducing the maximum stress and avoiding penetration,reinforcement in the flat area should be considered.

Fig.18 Comparison of strain vs time responses between tests and simulations.

Fig.19 configuration of auxiliary beam and equivalent stress(view from rear).

Fig.20 Improved auxiliary beam structures.

Fig.21 Structural deformations of reinforced auxiliary beam and arc-shaped auxiliary beam.

Fig.22 Final deformations(view from rear)and equivalent stress distributions of three auxiliary beams.

Fig.20(a)shows a novel reinforced auxiliary beam structure with strengthening stiffeners while the weight is kept the same as that of the test structure by reducing the thickness of the flat area.This new auxiliary beam is made with the same material(Al7050-T7451)and connecting pattern as those of the current test structure.

Fig.20(b)shows another novel structure,which has a different configuration and connecting pattern from those of the original structure.This auxiliary beam is developed based on the consideration of energy absorption and deformation capacity.It is a split structure,which consists of two connecting girders and an arc-shaped board,both made of Al7050-T7451.The thicknesses of these two components are determined to keep the weight of this novel structure the same as that of the auxiliary beam used in the original structure.Although there are no stringers or strengthening stiffeners on the back of the arc-shaped board,the shape of the board is responsible for the large capacity of impact energy absorption as the arc shape allows for much larger deformation than that of the flat shape.

3.4.Numerical predictions

The bird-strike resistances of the two improved structures are studied using the numerical model developed in the previous section.Results of structural deformation and damage are compared with those of the original design in this section.

Fig.21 shows the structural deformations of the two improved structures.To show the deformation of the auxiliary beam,the leading edge of the FE model has been removed from the figure.They both display an exceptional capacity of large deformation without damage.There is no penetration on the auxiliary beam.The simulation results confirm that these two novel auxiliary beams have good anti-bird strike performances.

Fig.23 Comparison of the maximum equivalent stress vs time responses between three configurations.

Fig.22 compares the final deformations of the three auxiliary beams,and the equivalent stress distributions att=3.0 ms,the moment before the original auxiliary beam was penetrated.To compare the stress levels in the three configurations,a uniform spectrum range is applied to the three equivalent stress contours.The contours illustrate that the overall equivalent stresses on the improved auxiliary beams are considerably reduced compared with that on the original structure.

Fig.23 gives a comparison of the maximum equivalent stress vs time responses between the three auxiliary beams.For the original auxiliary beam,the maximum equivalent stress point lies in the penetration holes region.For the reinforced and arc-shaped auxiliary beams,the point is in the middle area of the auxiliary beam.It is clear that att=3.1 ms,the maximum equivalent stress on the original auxiliary beam reaches the tensile strength(660 MPa)of the Al7050-T7451 aluminum alloy.However,the maximum equivalent stresses on the improved auxiliary beams never exceed 500 MPa during the bird strike process.

Fig.24 shows the equivalent strain distributions in the three auxiliary beams att=3.2 ms,when the penetration hole of the original auxiliary beam is enlarged,and att=8.0 ms,when the impact process is over.For the original auxiliary beam,att=3.2 ms,the strains of the shell elements on the vertexes of the penetration hole have reached the failure strain(0.14)of the Al7050-T7451 aluminum alloy.For the two improved structures,att=3.2 ms,strains on the auxiliary beams are well below the failure strain.

Fig.24 Equivalent strain distributions(view from front)at t=3.2,8.0 ms of three auxiliary beams.

Att=8.0 ms,for the original auxiliary beam,the strains of the shell elements on the edge of the penetration hole have all reached the failure strain.The failed elements are deleted and not shown in the figure.For the reinforced auxiliary beam,the overall strain level is very low,except some shell elements on the stringers.Even so,strains in these elements have never reached the failure strain.The arc-shaped auxiliary beam experiences the lowest strain among the three designs with the highest strain never exceeding 0.1.The strain contours clearly show that the strains in the two improved auxiliary beams are considerably reduced compared with that in the original design.

The improved performances are owing to different reasons.The strengthening stiffeners of the reinforced auxiliary beam provide local support and increase the anti-tension stiffness to withstand the impact load and to reduce the tearing damage.The arc-shaped auxiliary beam performs in a different way.As the arc-shaped board has an outward arc shape,which has a superior ability to support load,the board has more deformation space before the failure limit of the aluminum alloy is reached,which enables much more energy absorption.

4.Test 2:Design of cover sheet behind radome

4.1.Test

The specimen tested in this example is the cover sheet behind the radome of an aircraft.Fig.25 shows the detailed configuration of this cover sheet structure.This cover sheet is connected to the ribs of the aircraft nose.It is a doublehoneycomb sandwich panel,which consists of two aluminum alloy honeycomb cores and three aluminum alloy cover plates.These structural items are made from different materials:the honeycomb core is made of the aluminum alloy,the cover plates are made of the 7075-T6 aluminum alloy,the ribs are made of the 7050-T7451 aluminum alloy,and the vertical plates are made of the 2024-T3 aluminum alloy.The thickness of each honeycomb core is 12.7 mm,and the thicknesses of other components vary between 0.5 mm and 3 mm.

The cover sheet structure is mounted on a steelfixture,and the steelfixture is connected to a heavy support attached to a concrete ground,as shown in Fig.26.The impact point,shown in Fig.26,is at the mid-point of the cover sheet.In this test,the specified bird strike velocity is 150 m/s.

4.2.Simulation results vs test data

4.2.1.Finite element model

Thefinite element model includes all the components and somefixtures of the test structure.As shown in Fig.27,the model consists of 13816 solid elements and 19365 shell elements.

The average element size is 10 mm.

The bird used in this test weighs 1.81 kg(3.99 lb).The actual impact velocity of the bird is measured as 151.9 m/s,which is very close to the designed velocity.Thus this value is defined as the bird strike velocity in the numerical simulation.

Fig.25 Detailed configuration of cover sheet behind radome.

Fig.26 Cover sheet structure and steelfixture.

Fig.27 Finite element model of test structure and SPH bird model.

Similar to the numerical simulations presented in Test 1,a convergence study is conducted by comparing results of models with different element sizes,and the mesh adopted here is shown to provide a converged solution.

4.2.2.Deformation of cover sheet

Fig.28 compares the deformation of the cover sheet during the bird-strike process resulted from the numerical simulation with what has been captured by the side-on high-speed camera during the test.The good agreement between the simulation results and the test data indicates that the numerical model is capable of predicting the dynamic response of the cover sheet structure under a bird strike.Therefore,it can be used as a tool to design the cover sheet structure.

Fig.28 Comparison of simulated bird strike process with pictures taken with high-speed camera.

4.3.Improvement of design

4.3.1.Improved cover sheet structure

Fig.29 illustrates the original cover sheet structure and an improved design.In the original design,Fig.29(a),there is no supporting structure behind the double-honeycomb sandwich panel,resulting in the sandwich panel experiencing large deformation and being penetrated.In order to provide local support for the sandwich panel,aluminum foams are placed and fixed between the sandwich panel and the back plate,as shown in Fig.29(b).The density of the aluminum foams is 275 kg/m3and the total increased weight is 2.38 kg.

Fig.29 configurations of two cover sheet structures.

Fig.30 Simulated bird strike process of improved cover sheet structure.

4.3.2.Bird-strike simulation of improved design and test validation

Fig.30 shows the simulated bird-strike process of the improved cover sheet structure.The anti-bird strike performance is dramatically enhanced due to the support of the aluminum foams.As a result,there is no penetration on the cover sheet.

Fig.31 Improved cover sheet structure and result of bird-strike test.

Fig.32 configuration of representative horizontal tail structure.

Fig.33 Horizontal tail structure and steelfixture.

Fig.34 Finite element model of test structure and SPH bird model.

To validate the simulation result,a bird-strike test is conducted on the improved cover sheet structure.The test result,shown in Fig.31,is in good agreement with the result of numerical simulation.The sandwich panel is deformed but not penetrated due to the addition of the aluminum foams.The result suggests that the improved structure performs better than the original design and can successfully withstand the impact load due to a bird strike without catastrophic failure.

5.Test 3:Design of horizontal tail

5.1.Test

The specimen tested in this example is a representative horizontal tail structure.Fig.32 shows the detailed configuration of this structure.To show the internal layout,the upper skin has been removed from the figure.With about 1500 mm in length,it consists of a leading edge skin,two ribs,a front beam,a back beam,and upper and lower skins.The materials of these structural items are different:the leading edge and the skins are made of the 2024-T3 aluminum alloy,and the ribs and beams are made of the 7075-T6 aluminum alloy.The thicknesses of these components vary between 1.6 mm and 3 mm.

The horizontal tail structure is mounted on a steelfixture,and the steelfixture is then connected to a heavy support attached to a concrete ground,as shown in Fig.33.The impact point is at the mid-point of the leading edge skin,and the specified bird impact velocity is 180 m/s.

Fig.35 Simulated bird strike process of horizontal tail structure.

Fig.36 Comparison of final structural damage between numerical simulation and test.

5.2.Simulation results vs test data

5.2.1.Finite element model

The numerical model includes all the components of the test structure.In a test,a laser displacement sensor was used to monitor the deformation of the steelfixture.The results indicate that the steelfixture is very rigid,so in simulation,the steelfixture is not modeled,and the structure is just fixed at the component which is connected with the steelfixture.As shown in Fig.34,the model consists of 19992 shell elements.The average element size of the shell elements is 10 mm.Convergence studies suggest that the mesh adopted here has sufficient refinement to provide a converged solution.

The bird used in this test weighs 3.618 kg(7.98 lb).The actual impact velocity of the bird is measured as 178 m/s,which is very close to the designed velocity.Thus this value is defined as the bird strike velocity in the numerical simulation.

5.2.2.Deformation of horizontal tail structure

Fig.35 shows the simulated bird strike process.The leading edge skin fails to withstand the bird strike load and is penetrated.Fig.36 compares thefinial structural damage between the numerical simulation and the test.The simulation results of the deformation type and damage mode are very close to the test results.The good agreement between the simulation results and the test data indicates that the numerical model is capable of predicting the dynamic response of the horizontal tail structure under a bird strike,and the numerical simulation can be used as a tool to design the horizontal tail structure.

Fig.37 Original horizontal tail and improved structure.

Fig.38 Simulated process of bird strike on structure reinforced by single plate.

5.3.Improvement of design

5.3.1.Improved horizontal tail structure

Fig.37(a)shows the deformed original horizontal tail structure under the impact of a bird strike.Based on this result,two types of improved designs are proposed.For the first design,a single reinforcement plate is used to change the motion direction of the bird after it strikes the structure,as shown in Fig.37(b).The thickness of the reinforcement plate is 5.08 mm,and it is made of the 2024-T3 aluminum alloy.The total increased weight is 7.0 kg.For the second design,a triangular reinforcement is fixed on the front beam to cut the bird and change its direction,as shown in Fig.37(c),so that more impact energy can be dissipated.To keep the same weight as that of the single reinforcement plate,2.54 mm thick plates are used in the triangular reinforcement.To maintain the total weight of the horizontal tail structure the same as that of the original design,the thickness of the front beam is reduced in the two improved designs.

5.3.2.Numerical simulations and test validations

Fig.39 Support angle of single reinforcement plate.

Fig.40 Simulation results of structure reinforced by single plate with different support angles.

Fig.38 shows the simulated process of a bird strike on the structure reinforced by a single plate.After impacting on the structure,the bird is cut by the leading edge skin and the reinforcement plate.Before 2 ms,the leading edge structure can support the bird strike load effectively.However,after 2 ms,the reinforcement plate buckles,and finally the leading edge collapses and some bird debris are held by the pockets created on the deformed leading edge skin.

To further study the effect of the single reinforcement plate,it is placed at different angles,as shown in Fig.39.As the support angle has been changed,the plate thickness has also been changed to keep the weight the same.As shown in Fig.40,the leading edge structure suffers more deformation and damage as the support angle of the reinforcement plate decreases.

Fig.41 shows the simulated bird-strike process of the horizontal tail structure with a triangular reinforcement.The bird-strike resistance is dramatically increased due to the energy dissipation ability of the triangle reinforcement.The bird is cut into two parts so that the impact direction is changed.As a result,there is no significant damage on the leading edge skin.

To further study the effect of the triangular reinforcement,the distancehfrom the tip of the reinforcement to the leading edge is changed as demonstrated in Fig.42.As the distance has been changed,the plate thickness has also been changed to keep the weight the same.As shown in Fig.43,ashbecomes nonzero,the benefit of the reinforcement quickly diminishes.

Fig.41 Simulated process of bird strike on structure with triangular reinforcement.

Fig.42 Distance h between tip of triangular reinforcement and leading edge.

The horizontal tail structure having a triangular reinforcement withh=0 mm is shown by numerical simulation to have a significantly increased bird-strike resistance.To validate this result,a bird-strike test is conducted on this improved structure.The experimental result shown in Fig.44 confirms the simulation result.The improved structure shows a good anti-bird strike performance,where the deformation of the leading edge is very small and no damage is observed on the structure.

Fig.43 Simulation results of structure having triangular reinforcement with different distance h.

6.Concluding remarks

In this paper,a method to design bird-strike-resistant aircraft structures is presented and illustrated through three examples.The focus is on bird strike experiments and simulations.The explicitfinite element software PAM-CRASH is employed to conduct bird strike simulations,and a coupled SPH and FE method is used to simulate the interaction between a bird and a target structure.The SPH method is explained,and an SPH bird model is established.Constitutive models for various structural materials,such as aluminum alloys,composite materials,honeycomb,and foam materials that are used in aircraft structures,are presented,and model parameters are identified by conducting various material tests.

In the three examples,bird strike tests are carried out on the original structures,and simulation results are compared with experimental results to verify the numerical model.Next,based on the simulation results,the original designs are modified according to three design principles:(A)to improve the energy absorption ability of the structure,(B)to improve the energy dissipation ability of the structure,and(C)to adopt the multi-layer protection design.Finally,bird-strike simulations of the improved designs are conducted,and simulation results are validated by experiments.

Good agreements between the simulation results and the experimental data obtained in this study suggest that the numericalmodeliscapableofpredictingthedynamic responses of various aircraft structures under a bird strike,and numerical simulation can be used as a tool to design bird-strike-resistant aircraft structures.

Acknowledgement

This study was supported by Natural Science Foundation of China(No.11472225).