Liu Fu 1,2, Zhang Jiazhen 1, Hu Zhongmin 1, and Zhang Mingyi 1 1 Beijing Aeronautical Science & Technology Research Institute of COMAC, Beijing 102211, PR China 2 Nanjing University of Aeronautics and Astronautics, Nanjing 210016, PR China Summary GLARE is a Glass Fibre Reinforced Epoxy/Aluminum FMLs (Fibre Metal Laminates) and has been reported for its excellent impact resistance. In this paper, a numerical model for bird impact on an aluminium alloy plate was established, using the explicit finite element code PAM-CRASH first. A corresponding bird impact experiment was carried out and the simulations were compared with the experimental results. The good agreement between them confirms the accuracy of the numerical analysis model, including the Smoothed Particle Hydrodynamics (SPH) bird model and the aluminium alloy constitutive model. In the second part of the paper,, another simulation on a thinner 2024-T3 aluminum alloy plate was performed, which was used to determine the critical bird impact speed for the plate. By combining the Johnson-Cook model for aluminum alloy and Continuum Damage Mechanics (CDM) theory for the glass fibre reinforced composites, a constitutive model for GLARE was derived. Impact simulations on the GLARE plate with the same thickness and impact speed as the 2024-T3 plate shows no penetration on the GLARE plate and indicates that GLARE can effectively improve the impact resistance of a structure. Keywords: Bird impact, GLARE, FMLs, Experiment, Smoothed Particle Hydrodynamics 1. Introduction Bird strike is a serious and damaging event for aircraft during take off and landing 1, and hence specific design criteria with respect to bird strike events have been issued by civil aviation authorities. The problem of numerically analyzing the damage of aircraft structure caused by bird strike has been studied by many researchers for many years. The explicit finite element code PAM-CRASH was selected to perform the simulations of the bird-strike certification of the carbon fibre epoxy composite, moveable trailing edge of the Boeing 787 Dreamliner by Georiadis, which had improved design efficiency and safety and reduced certification costs 2. Smojver used ABAQUS to analyze bird impact damage in complex aircraft flap structures. Parametric analyses had Smithers Rapra Technology, 2013 been performed using different bird sizes, impact locations and velocity 3. Weight reducing is an important object in aircraft design. Due to their high specific strength, specific stiffness and other advantages, composites materials are increasingly used in large civil aircraft. With the development of composites, a new kind of material Fibre Metal Laminates (FMLs) has been successfully developed at the Delft University of Technology. FMLs are hybrid composite structures based on thin sheets of metal alloys and plies of fibre reinforced polymeric materials, which combine the advantages of monolithic aluminum alloys and conventional fibre-reinforced laminates. One of those FMLs, known as GLARE Glass Fibre Reinforced Epoxy/Aluminum FMLs, has been successfully used in the main fuselage skin, the leading edges of the horizontal and vertical stabilizers of the Airbus A380. GLARE has also been selected for the Boeing 777 impact resistant bulk cargo door. Glare laminates consist of alternating layers of unidirectional glass fibre reinforced prepregs with a thickness of 0.25~0.5 mm, and high strength aluminum alloy sheets with a thickness of 0.3~0.5 mm. GLARE has been reported for its excellent fatigue and impact resistant properties. McCarthy designed and built two different leading edge structures, one made of FMLs and one made of aluminum, and numerically simulated bird impact experiments. The simulation and experimental results show that the use of FMLs considerably improves the bird strike performance of the structure 4. Guida used Ls-dyna and MSC-Dytran to simulate bird impact on a leading edge bay where the inboard ply of the layup was made of aluminum alloy 2024-T3, the outboard ply was GLARE and the core section of the layup was Polymers & Polymer Composites, Vol. 21, No. 9, 2013 587
Liu Fu, Zhang Jiazhen, Hu Zhongmin, and Zhang Mingyi presented by the honeycomb. The structural damage due to the impact and the residual structural strength of the component were investigated 5. In this paper, based on the explicit finite element code PAM-CRASH, the numerical model of bird impact on an aluminum alloy LY12 plate was established. A corresponding bird impact experiment was conducted. The calculation results were compared with the experimental results. The good agreement between them has confirmed the numerical method is reasonable and reliable. Then, the comparison of anti-bird strike performance on a thinner aluminum alloy 2024-T3 plate and a GLARE plate with the same thickness and impact velocity were performed by the numerical calculation. 2. The bird impact analysis and its validation 2.1 Constitutive Model of the Aluminum Alloy A bird impact event is very sudden and happens just within a few milliseconds. Hence rate dependent material parameter need to be used to describe the material behavior. Influence of strain rate for material must be considered. The constitutive model for the aluminum alloy LY12 was described by the Johnson-Cook equation: ( ) σ = ( A + Bε n )( 1+ C ln ε * ) 1 ( T * ) m (1) Where, s is the stress, e is the effective plastic strain, ε * = ε / ε 0 is the dimensionless plastic strain rate. In the paper, we chose ε 0 as 1 s -1, T * = (T T r )/(T m T r ), where T r is the room temperature, T m is the melting point of material. Hardening effects, strain rate effects and temperature softening effects are respectively described by (A + Be n ), 1+ C ln ε * ( ) and (1 (T * ) m ). By using a Split Hopkinson Pressure Bar (SHPB) to perform the material dynamic mechanics experiments at different strain rates, values of five parameters in Johnson-Cook equation were determined, which are shown in Table 1. 2.2 The SPH Model of the Bird In the analysis, the bird was represented by a cylinder caped by hemispheres at both ends. The ratio of length to diameter was 2. The weight of the bird was 1.8kg and density was 950kg/m 3, thus the length of the bird model was 226mm and the diameter was 113mm. Upon impact, the bird splashes like a fluid. Smoothed Particle Hydrodynamics (SPH) method based on a free mesh is used in the simulation. In SPH, the bird is represented by a series of particles with unattached mass, which transmit the bird s momentum [6]. A function f(x) is constructed to solve the velocity and energy of the particles at any time. We can approximate the value of f(x) for particle i by: ( ) = f x i N m j j=1 ρ j f ( x j ) W( x i x j, h) where m j and ρ j are the mass and density of particle j, respectively. x i and x j are respectively the position of particle i and particle j. h denotes the smooth length. In the computational domain of particle i, a set of N distributed particles affect particle i. The radius of the computational domain R is given by R=k*h, with k=2. The interpolation Kernel W(x,h) is important for the accuracy of the calculation result. In this paper, the spline form proposed by Monaghan [6] was introduced as the interpolation Kernel which is expressed by: 1 3 x 2 + 3 3 x 0 x 2 h 4 h h < 1 W(x, h) = 1 1 πh 3 4 2 x 3 1 x h h < 2 x 0 h 2 (3) To avoid the numerical instabilities, a viscosity coefficient according to Monaghan 7 was introduced. The coupling effects between the bird (SPH) and the plate (FE) was realized by the contact definition in PAM-CRASH. Constitutive model of the bird was described by Murnaghan EOS equation 7, which is defined as: P = P 0 + B ρ γ 1 ρ 0 where, P 0 denotes the reference pressure, r 0 is the initial density, B and g are constants to be determined by optimization calculation. Throughout this paper, Table 1. Values of five parameters in Johnson-Cook equation for LY12 Parameter A/MPa B/MPa C m n Value 325 555-0.001 2.2 0.28 (2) (4) 588 Polymers & Polymer Composites, Vol. 21, No. 9, 2013
the values for B and g are, B = 128 x 10 6, g = 7.98 4. 2.3 The Calculation Model The jigs of the equipment were modelled with 8800 solid elements. The plate was a 600 mm by 600 mm square with a thickness of 10 mm, and it was modelled with 10200 Beltyschko- Tsky shell elements. Contact relation between the bird and the plate, the plate and the jigs was defined. Six positions on the jigs were fixed. The plate and the jigs were connected with the bolts using 24 rigid elements. The complete calculation model including the jigs, the plate and the bird (8280 SPH elements) are shown in Figure 1. the jigs and the bolts have a greater influence on S1 than the bird impact. This explains the difference in the first positive peak in the strain-time history. The boundary stiffness of the plate in the numerical simulation is Figure 1. Complete calculation model higher than that in reality. As a result, the simulated strain and displacement values are lower than the experimental results at the beginning of impact. As time goes on, the stress wave caused by the bird impact propagates to the 2.4 Model Validation Through Experiment The bird impact velocity was 70 m/s, and the impact location was the center of the plate. The strain at the test points was measured by the strain gauges (S1 to S4). As the strain gauges may be damaged during the test, additional four strain gauges (S5 to S8) were attached at the symmetric locations to ensure adequate data capture (see Figure 2). The position of the test points are depicted in Figure 2. Figure 3 shows the LY12 plate fixed in the test-bed. At two locations on the plate, the displacement was measured by the optical displacement sensors. The strain gauges and the displacement sensors are shown in Figure 4. 2.5 Results and Discussion Comparisons of strain-time history and displacement-time history of the test points between simulations and experiments are shown in Figure 5 and Figure 6. The figures show that the simulation results for changing trends and peak locations of strain and displacement of calculations agree with the experimental data reasonably well, which indicates that the simulation methods used are adequate. Gauge S1 is near the edge of the plate and at the beginning of impact process, Figure 2. The position of the test points Figure 3. The specimen, jigs and test-bed Polymers & Polymer Composites, Vol. 21, No. 9, 2013 589
Liu Fu, Zhang Jiazhen, Hu Zhongmin, and Zhang Mingyi Figure 4. The strain gauge and the displacement sensor edge of the plate, calculated results are closer to measured ones due to less boundary effect. 3. Bird impact on an aluminum alloy 2024-T3 and a GLARE plate Figure 5. Comparison of strain at test points between calculations and experiments 3.1 Bird Impact Simulation on an Aluminum Alloy 2024-T3 Plate For the bird strike simulation on an aluminum alloy 2024-T3 plate, the same calculation model as described in Section 2.3 for the LY12 plate was used. However, the plate thickness was reduced to 2.35 mm. The constitutive model for the aluminum alloy 2024-T3 material was again the same as used before for LY12 (see Section 2.1). The material parameter values for the Johnson-Cook equation fitted for 2024-T3 are shown in Table 2. The equivalent strain limit for 2024-T3 was measured as 0.19 by the SHTB experiments. The constitutive and SPH model for the bird was identical to the one described in Sections 2.2. Through iterative simulations, a critical velocity of 160 m/s for 2024-T3 was calculated. Figure 7 shows the equivalent strain fringe of the plate caused by the bird impact at different times. Figure 6. Comparison of displacement at test points between calculations and experiments 3.2 Bird Impact Simulation on a GLARE Plate In this study, GLARE3 4/3 was used, which has a lay-up of A/0/90/A/0/90/ A/0/90/A, whereby A denotes a layer of aluminum alloy and 0 and 90 denote the fibre orientation. The metal layers were aluminum alloy 2024-T3 with a thickness of 0.4 mm and the composite layers were glass/epoxy FM94 S2-Glass with a thickness of 0.125 mm. The total thickness of the laminate was 2.35 mm. The mechanical properties of the glass/ epoxy ply are given in Table 3 8. 590 Polymers & Polymer Composites, Vol. 21, No. 9, 2013
The constitutive model of the aluminum alloy 2024-T3 in the GLARE laminates was described by the Johnson-Cook equation (see Section 3.1). Failure used is the same as stated in Section 3.1. Continuum Damage Mechanics (CDM) theory according to Ladeveze was used to describe the damage mechanics of the glass/epoxy ply 9. This model includes three scalar damage variables: d f which quantifies fibre damage, d 2 which denotes damage arising from matrix micro cracking and d 12 which relates to fibre/matrix debonding, all of which take on values 0 d i 1. Y 12 and Y 2 are introduced to govern damage development. These two quantities are similar to energy release rates which drive crack propagation. Three scalar damage variables are then assumed to be functions of Y 12 and Y 2 : d 2 =f 2 (Y 2, Y 12 ), d 12 =f 2 (Y 2, Y 12 ) (5) The parameters of the CDM model for FM94 S2-Glass Layers are based on specimen experiments which are shown on Table 4 10. Table 2. Values of parameters in Johnson-Cook equation for 2024-T3 Parameter A/MPa B/MPa C m n Value 345 462 0.001 2.75 0.25 Table 3. Mechanical properties of glass/epoxy ply Parameter Value E 11 (GPa) longitudinal Young s modulus 45.6 E 22 (GPa) transverse Young s modulus 16.2 G 12 (GPa) in-plane shear modulus 5.83 v 12 Poisson s ration in 1,2 plane 0.278 v 23 Poisson s ration in 2,3 plane 0.4 X t (MPa) longitudinal tensile strength 1280 X c (MPa) longitudinal compressive strength 800 Y t (MPa) transverse tensile strength 40 Y c (MPa) transverse compressive strength 145 S (MPa) in-plane shear strength 73 ε 1t (%) longitudinal tensile ultimate strain 2.087 ε 1c (%) longitudinal compressive ultimate strain 1.754 ε 2t (%) transverse tensile ultimate strain 0.246 ε 2c (%) transverse compressive ultimate strain 1.2 γ 12u (%) in-plane shear ultimate strain 4 Figure 7. The equivalent strain fringe at different times Polymers & Polymer Composites, Vol. 21, No. 9, 2013 591
Liu Fu, Zhang Jiazhen, Hu Zhongmin, and Zhang Mingyi Table 4. Parameters of CDM model for FM94-S2-Glass Layers Parameter Value Y 120 ( Pa ) threshold for initial shear damage 324 Y 12C ( Pa ) slope of the assumed linear Y vs. d 12 relationship 3500 Y 20 ( Pa ) threshold for initial transverse damage 200.4 Y 2C ( Pa ) slope of the assumed linear Y vs. d 2 relationship 1700 Y S ( Pa ) cutoff value for brittle failure in transverse tension 686 Y R ( Pa ) cutoff value for brittle failure in shear 2105 d max maximum allowed value of d 2 and d 12 (d max 1) 0.99 e f tensile fibre initial strain i 0.057 e f tensile fibre ultimate strain u 0.067 γ compressive stiffness loss constant 4.187 10-10 d f tensile fibre ultimate damage u 0.99 Figure 8. Deformation and damage of plate 3.3 Comparison of the Simulations for the Aluminum Alloy 2024-T3 vs. the GLARE Plate Comparison of deformation and damage between the aluminum alloy 20124-T3 and the GLARE after bird impact is shown in Figure 8. Although the weight of GLARE plate reduced from 2.35 kg to 2.09 kg, there is no penetration on the GLARE plate at the same velocity. The result shows that GLARE has a better impact resistant performance than the aluminum alloy. 4. Conclusions Bird impact simulations on a framed square aluminum alloy LY12 plate was performed using PAM-CRASH, the dynamic responses of displacement and strain were in good agreement with the experimental results. Using the same FE and SPH models to simulate bird impacts on an aluminum alloy 2024-T3 and a GLARE3 4/3 plate with the same thickness showed that at the critical velocity for the aluminum alloy 2024- T3 plate, no penetration took place on the GLARE plate, which is 11% lighter than the aluminum plate. The results confirmed the superior impact resistant performance of the GLARE laminates. References 1. Dolbeer R.A., Wright S.E., Weller J., et al., Wildlife strikes to civil aircraft in the United States 1990-2008. FAA National Wildlife Strike Database, Serial Report Number 15, 2009. 2. Georgiadis S., Gunnion A.J., Thomson R.S., et al. Bird-strike simulation for certification of the Boeing 787 composite moveable trailing edge. Composite Structures, 86(1-3) (2008) 258-268. 3. Smojver I. and Ivancevic D., Numerical simulation of bird strike damage prediction in airplane flap structure, Composite Structures, 92(9) (2010) 2016-2026. 4. McCarthy M.A., Xiao J.R., McCarthy C.T., et al. Modeling of bird strike on an aircraft wing leading edge made from fibre met al laminates Part 2: modeling of strike with SPH bird model. Applied Composite Materials, 11(5) (2004) 317-340. 5. Guida M., Marulo F., Meo M., et al., SPH-Lagrangian study of bird impact on leading edge wing. Composite Structures, 93(3) (2011) 1060-1071. 6. Monaghan J.J. and Lattanzio J.C., A refined particle method for astrophysical problems. Astronomy and Astrophysics, 149(1) (1985) 135-143. 7. Monaghan J.J., On the problem of penetration in particle methods. Journal of Computational Physics, 82(1) (1989) 1-15. 8. Huang Z.M. and Zhang H.S., Current status and future trend of researches on the strength of fiberreinforced composites- summary of the results from a failure Olympics. Advances in Mechanics, 37(1) (2007) 80-98. 9. Ladeveze P. and LeDantec E., Damage modelling of the element ply for laminated composites. Composites Science and Technology, 43(3) (1992) 257-267. 10. McCarthy M.A., Xiao J.R., Petrinic N., et al., Modeling of bird strike on an aircraft wing leading edge made from fibre metal laminates Part 1: material modeling. Applied Composite Materials, 11(5) (2004) 295-315. 592 Polymers & Polymer Composites, Vol. 21, No. 9, 2013