DESIGNING COMOSITE STRUCTURES FOR IMACT ERFORMANCE WHAT CAN WE LEARN FROM THE AEROSACE INDUSTRY? Dr. Robert N. Yancey, Mr. Jean Michel Terrier, Mr Jean Baptiste Mouillet Altair Engineering Abstract Understanding the impact performance of composites is critical in Automotive applications where designing for crash impact loads is standard. Although composite body structures are relatively new in the Automotive industry and there is not much data in regards to crash performance, there is a significant experience base in the Aerospace industry that can be referenced. Composite fan blades and helicopter blades have been used for many years in the industry and they have to be designed to withstand impact loads from a bird strike or other debris. Likewise, composite wing and fuselage structures have to be designed to continue safe operation after a bird or debris impact event. Hard landings or ditching in water are also design conditions for airplanes and helicopters. There are differences in the dynamic nature of the impacts and different materials used between the two industries. Generating the right material model for the type of material and impact loads is key to accurately simulating the material and structural response to any type of dynamic load. A review of lessons learned from the Aerospace industry will be provided along with some specific examples of how simulation methods have proved useful in understanding the response of composite structures to impact loads of Aerospace structures. The review will also discuss the unique needs of the Automotive industry for crash simulation of composites and how these are being addressed. Introduction The use of composite materials in the aerospace industry started in the early 1970 s as a primary tool to decrease mass. As composite materials have a very different behavior than metallic materials, a lot of effort was made to develop tools to understand them and ensure the eventual certification, including numerical tools. Those tools became critical for assessing the performance of composite materials in shock, crash or impact as tests were very expensive and difficult, if not impossible to perform. Nonetheless, composites were expensive and could not be manufactured in large volumes, use in the automotive industry has been limited. As the understanding of composites has increased and more rapid manufacturing methods have been developed, the interest of automotive designers rose, especially as fuel economy standards continue to increase. The purpose of this paper is to describe crash and impact calculation methodologies in aerospace applications and to show how they are being brought to the automotive industry. For all of the numerical simulation work performed in this paper, HyperWorks by Altair Engineering was used (HyperMesh for pre-processing, RADIOSS for processing and HyperView for the post-processing). age 1
Material arameters Identification Material identification is a key phase to assess the crashworthiness and impact resistance of a structure. For this type of application, a material law was designed for RADIOSS as part of the CRASURV European program, featuring the following properties: Accurate description of the stress vs strain curve in main directions (tension, compression, fiber directions, shear ) Identification of the energy absorption capacities of the material ossible strain rate dependency Existence of a straightforward method to accurately identify the parameters of the material through tests CRASURV material law The orthotropic composite material law is based on a visco-elasto-plastic modelling of composites; non-linear and strain rate dependent behaviors. The plastic flow threshold F(σ) is formulated as a Tsai-Wu quadratic function of the stress tensor : 2 ( W ) σ + F ( W ) σ 2F12 ( W ) σ 1σ 2 = Fii ii F + Where F<1 means the elastic phase and F=1 the plastic phase. The F i F ij and F ii coefficients describing the elastic/plastic transition envelope are dependent on the global plastic work W p according to the relations: F F i F ( W ) ii ( W ) c ii 1 = σ 1 = σ α 2 t ( W ) σ ( W ) c + i 1 t ( W ) σ ( W ) ( W ) = F ( W ). F ( W ) 12 11 p 22 p. 1, Where : In these expressions, direction i. t 0 t σ c σ t = σ c = σ 0 0 ( ) t & ε 1+ b W nit 1+ ln i c & ε 0 ( ) c n & ε 1+ b W ic i 1+ c ln & ε 0 c σ and σ are the initial tension and compression yield stresses in age 2
The first terms depending on the plastic work Wp describes the non-linear static behaviors t c b according to parameters i, b i, nic and nit respectively in tension and compression. The second terms describes strain rate dependency as a function of strain rate and reference strain rates. The parameters are identified from dynamic characterization tests. Ultimate damage and softening properties are defined for each direction i=1,2,4 with 3 additional parameters : ( 1ci, 2ci, resi) in compression and similar parameters in shear and tension. A graph of the generic behavior for a single direction is shown in Figure 1. Figure 1. Crasurv material law : generic behavior showing stress versus strain curve for a single direction Failure is defined through 3 sets of parameters: Tension failure in direction 1 and 2 : Maximum plastic work Transverse shear strains for delamination : σ σ 31 23 = G = G 31 23 ( 1 d3) γ 31 ( 1 d 3 ) γ 23 With : d 3 γ γ = t γ γ m t age 3
As the tensile failure is defined independently from plasticity, failure may be either brittle if stress at failure strain is lower than yield stress or ductile. It should be noted that energy absorption due to delamination can be modeled with plastic work. Indeed, multilayered compression tests for instance may include delamination. It is also possible to add the following failure criteria: Hashin uck Ladeveze Chang-Chang Several criteria can be applied to each material and combined, such as for instance Hashin for fiber directions, uck for matrix directions and Ladeveze for delamination. It s possible due to the compatibility in Radioss code of the material and rupture criteria libraries. rony viscosity can also be added to take into account the viscous properties of the material Material Law arameters The material parameters are identified by fitting the static test curves in traction, compression and shear tests, according to, for instance, the following norms : D3039-95 (tension) SACMA SRM 1-94 (compression) ASTM D2344-95 (shear) for instance. DCB / MMB and ENF tests for delamination tests. As described above, the purpose is, for each mode (tension / compression in each orthotropic directions / shear), to identify the following parameters: Eii or Gij the Young or shear modulus (depending on the mode), y, b, n It should be noted that the process is generic regardless of the loading mode. Though the resulting parameters will obviously change according to the loading mode and the type of composites, resulting in a completely different physics, the general equation and method remains the same. A method for identifying the parameters from tests, using RADIOSS and Hyperview was designed to obtain the parameters in a robust manner. age 4
It is based on the following process (for static values): 1. Identification of the modulus as initial slope of the stress vs strain curve 2. Identification of the yield stress which is the stress level where the curve stress/strain begins to be non linear 3. Calculation of the plastic work curve vs strain as : ε W p (ε) = σ. dε 0 σ2 2. E Where Wp is the plastic work, is the strain, the stress level and E the modulus calculated at step 1 1. Identification of n and b from the previous curve 2. Identification of the failure parameter directly from the curve Some representative model fits are shown in Figure 3. Figure 3. Identification process of a typical CFR from a 45 degree compression test Validation on Simple Structure : CMH-17 RRI As part of CMH-17 Crashworthiness Working Group, a simple corrugated C-structure was tested. The structure is a 2 x 3 inches corrugated C, the material is carbon / epoxy TORAYCA fabric, with a layup of 8 plies (0.079 inches thickness). An illustration of the test article is shown in Figure 4. Five tests were performed exhibiting a significant result dispersion. The material characterization process was performed using the process described above based on reference [8]. A mesh size of 1 mm was chosen, for a 1000 elements model. Due to the quasi static nature of the model, the crush velocity was increased to 2.5 m/s, resulting in some oscillations, which were filtered with an SAE600 filter. The results are shown in Figure 5. age 5
Figure 4. Corrugated C used for CMH-17 stage I Five tests were performed exhibiting a significant result dispersion. The material characterization process was performed using the process described above based on reference [8]. A mesh size of 1 mm was chosen, for a 1000 elements model. Due to the quasi static nature of the model, the crush velocity was increased to 2.5 m/s, resulting in some oscillations, which were filtered with an SAE600 filter. The results and synthesis are shown in Figure 5. Stress vs strain curves, experiments (red) vs simulation (blue) comparison Name description Average force (N) specimen1 Test 15980 specimen2 Test 16395 specimen3 Test 15256 specimen4 Test 15183 Baseline simulation Simulation 15374 Figure 5. Results synthesis age 6
The results were mostly sensitive to compression behavior. Indeed, a purely linear / elastic behavior in compression gave high pressure waves and a very brittle behavior, leading to wrong results. The oscillating response on the model under compression is shown in the mixed colors in the Max Stress plot on the right side of Figure 6. Though the non-linearity in compression looks small it had a very high influence on the calculation results Figure 6 : unstable behavior due to compression waves Bird Strike on Leading Edge Aerospace Applications Bird strike tests have been performed on five different commuter wing leading edges [1]. The approach was to perform pre-test simulations in order to gain a first reference for test preparation and to perform post-test correlation for the validation of the models. The test specimens are commuter aircraft composite (sandwich Kevlar-Nomex) wing leading edges. Each leading edge was instrumented with strain gauges, and each test was filmed with high speed cameras. Various test conditions were considered, with speeds ranging between 111 and 153 m/s, variable location and incidence. The impactor had a mass close to 2 lbs (0.908 kg +- 6%). All the tests were performed at CEAT. The numerical simulation performed well in predicting failure modes with good correlation with damage extent and strain gauge signals from the test, as can be seen figures 7 and 8. Though no obvious application to automotive industry can be evidenced from this type of application, the general capacities of the approach to model failure and energy absorption was validated. age 7
Figure 7. Failure extent comparison between pre-test simulation and experiment Figure 8. Typical strain vs. time curves, test (blue) vs. simulation (red) age 8
Crashworthiness of a Composite Helicopter As part of a cooperative ONERA / Eurocopter (now Airbus Helicopter) and Mecalog (now Altair Engineering) program, a modelling methodology was designed to improve the crashworthiness performances of the Tiger Helicopter. The part of the helicopter analysed is shown in Figure 9. Figure 9. FE Model of the Tiger Composite Structure Based on a similar methodology, the method incorporated crashworthiness specificities such as connexion modelling, contact, presence of liquid and crashworthiness performance specifications (limited deformation and acceleration for the crew, no fuel leak ).The methodology is divided into the following steps: Identification of material properties Validation of the crash subcomponents and simulation of the crash behaviour Two main subcomponents were considered for crash behaviour: a sinus beam, and a honeycomb panel as shown in Figure 10. age 9
Sinus beam Honeycomb panel Figure 10. Main Crashworthiness Components The simulation results compared with one of the test specimens is shown in Figure 11. The failure mode was accurately predicted with the simulation model and the energy absorption results compared favourably with test. age 10
Simulation Experiment Figure 11. Simulation vs. test of the sinus beam (courtesy Airbus Helicopter) The global structure was then tested with good correlation with the simulation results. The model is shown in Figure 12 with graphs comparing the simulation with test for both a Lagrangian and ALE model in Radioss. The model and conditions bear strong similarities with automotive crashworthiness, in terms of impact velocities and modelling difficulties (contact, connexions ) age 11
Figure 12. Global structure simulation (courtesy Airbus Helicopter) Formula 1 Crash Absorber Automotive Applications Formula 1 was the first field in automotive industry where composite materials were used for structural components, as the low-volume manufacturing methods and high price were found less critical. A typical use of composites was the design of crash absorber for Formula 1. This is highlighted in the blue box in Figure 13. age 12
Figure 13 Main Composite Sub-Components of a Formula 1 Car The methodology used for predicting performance was very similar to what was shown for the Helicopter crashworthiness example. Figure 14 demonstrates the results of the simulation vs test of a Jaguar R2 nose cone (courtesy of Ford Motor Company). The test vs. simulation correlation is excellent. Figure 14. R2 Nose Cone Simulation versus Test age 13
IndyCar Crash Box The crash box for the IndyCar made by Dallara is made from carbon-fiber reinforced plastic (CFR) [5-6]. Dallara has worked to develop the simulation and failure models of the composite during a crash event. Figure 15 shows the simulation model and a photograph of the test setup. Figure 16 shows the test results compared to the simulation model. The test results are shown as the blue solid line and the simulation results are shown in red. As shown, the correlation between test and simulation is quite good. Once the models are defined and verified, simulations can reliably be carried out for conditions that cannot be easily tested. Figure 15. Simulation Model and Test Set-Up for Composite Crash Box Figure 16. Simulation vs. Test Results for IndyCar Crash Box age 14
Conclusions The work performed for the aerospace industry can directly be exploited by automotive industry. In some cases, such as crashworthiness, the modelled phenomenon were highly similar. The aerospace experience brought: Evidence of the feasibility of crash simulation of composite sub structures Numerical models for typical composite materials, focusing on the most important properties: energy absorption capacity and failure modes. The aerospace industry has developed good methods, models, and procedures for analyzing impact events on composite structures. Bird strike and hard landing models have been developed over many years and have good correlation to test data. The motorsports industry has also developed good capability for analyzing composite structures in a crash event. For passenger automobiles, the impact events will be different than motorsports and aerospace but much of the knowledge of how to develop and implement the models can be derived for passenger automotive use. It is key that good simulation models of crash events for composite structures be developed for the widespread adoption of composites in mainstream automotive applications. Bibliography 3. Arnaudeau, F., Mahe, M., Deletombe, E., Le age, F., Crashworthiness of Aircraft Composite Structures, IMECE 2002. 4. Delsart, D., Joby, D., Mahe, M., Winkelmuller, G., Evaluation of Finite Element Modeling Methodologies for the Design of Crashworthy Commmercial Aircraft Fuselage, ICAS 2004. 5. Radioss Theory Manual, Altair Engineering, September 2014. 6. Delsart, D. (ONERA), Lassus, V. (Eurocopter), Chauveau (Mecalog), Crashworthiness of Composite Helicopters Towards Design Costs Reduction, Internal Report. 7. Simulation: The Connections between Speed and Safety in Racing, Concept to Reality, Summer/Fall 2012, enton ublishing, www.c2r.altair.com. 8. Speed and Safety in Motorsport, Altair ATC 2012, www.altairatc.com 9. JB Mouillet, J Santini : CMH-17 Crashowrthiness work group stage I 10. AGATE-W3.3-033051-131 age 15