Numerical simulation of aircraft interior components under crash loads

International Journal of Crashworthiness Vol. 13, No. 5, October 2008, 511–521

Numerical simulation of aircraft interior components under crash loads S. Heimbsa,∗ , D. Vogta , R. Hartnacka , J. Schlattmannb and M. Maierc a EADS Innovation Works, Nesspriel 1, Hamburg, Germany; b Institute of Laser and System Technologies, Hamburg University of Technology, Denickestraße, Hamburg, Germany; c Institute for Composite Materials, Kaiserslautern University of Technology, Erwin-Schrödinger-Straße, Kaiserslautern, Germany

Downloaded By: [Heimbs, Sebastian] At: 15:42 9 September 2008

(Received 14 January 2008; final version received 16 April 2008) Aircraft interior components like overhead stowage compartments may be subjected to highly dynamic loads in case of turbulence or an emergency landing where the structural integrity of these lightweight components made of composite sandwich materials has to be maintained. Dynamic simulations of such overhead bins under crash loads were conducted in parallel with experimental full-scale testing to investigate the structural behaviour. The focus of these investigations is on modelling of the honeycomb sandwich materials, joints and baggage involved, which made testing on a coupon and component level necessary. Because the structural response of these overhead bins depends on a large number of parameters, uncertainty assessment methods were utilised for their characterisation. Despite the complexity of the models, a high level of correlation of simulation results and experimental data could be achieved making such numerical methods a useful tool for the development of cabin components for dynamic loads. Keywords: aircraft interior; baggage modelling; dynamic simulation; material modelling; sandwich structures; uncertainty analysis

Introduction Inside the passenger cabin of a commercial aircraft most components are made of composite sandwich structures [10,32]. This is because fibre-reinforced phenolic (FRP) R faces combined with a phenolic-impregnated Nomex honeycomb core are an established lightweight structure of high bending stiffness and strength with outstanding fire safety properties like low smoke emission, low toxicity and low heat release [30,41]. Examples are sidewall panels, ceiling panels, lavatories and overhead stowage compartments (OHSCs). In particular, these OHSCs may be subjected to high and uncertain loads due to heavy carry-on baggage. Typically, they are designed for static loads and tested statically for certification. But, the external loads acting on these overhead bins may be highly dynamic ranging from air turbulences to hard landings. Even in the worst case scenario, a survivable crash landing, the integrity of the OHSCs has to be assured, because a structural failure or detachment would make an evacuation difficult. The structural behaviour of the stowage bins under such abrupt deceleration pulses cannot be considered adequate when designed for static loads. Only aircraft seating systems have to fulfil dynamic design criteria for certification according to Federal Aviation Regulations [8]. An investigation of the behaviour of OHSCs under short-time dynamic loads is therefore reasonable and has been initiated in past studies.

∗

Corresponding author. Email: [email protected]

ISSN: 1358-8265 C 2008 Taylor & Francis Copyright DOI: 10.1080/13588260802221203 http://www.informaworld.com

Within the framework of the European research programme ‘Crashworthiness for Commercial Aircraft’, a survivable crash scenario was recreated by a drop test of an Airbus A320 fuselage section with OHSCs [11,16]. In addition to this drop test, the structural response of the stowage bins was investigated in dynamic sled tests conducted in forward and vertical load cases [20]. The respective numerical simulations with LS-DYNA were performed by Schweizerhof and Senger [34]. Because of the limited computational performance at that time, rather simple models with homogenised and isotropic shell elements for the sandwich structures were used. The authors themselves put the significance of the simulation results into question because of insufficient material data of the sandwich structures and joints as well as the baggage stiffness. Within the crashworthiness program of the Federal Aviation Administration [1,7] dynamic tests were conducted on narrow body fuselage sections with OHSC, both from Boeing 707 [3,21] and Boeing 737 [5,26] aircraft. This testing involved drop tests as a vertical load case and longitudinal deceleration tests as a forward loading, both representing survivable crash scenarios. The respective aims included the determination of the structural deformation of the overhead bins under dynamic loading as well as the maximum loads and fracture behaviour. Dynamic simulations were conducted with LS-DYNA by Byar et al. [5] and MSC Dytran by Jackson


512

S. Heimbs et al.

and Fasanella [18]. The overhead stowage bins were modelled with very low detail and the mass of the baggage was smeared into the bin’s mass or modelled with concentrated mass points. The sandwich material was modelled with isotropic shells without knowledge of the constitutive properties and literally using ‘best guess estimates’ [18]. Hence, these models did not allow for detailed investigations of the bin’s structural behaviour and demonstrated that a huge amount of data concerning material properties, joint load limits and baggage stiffness is necessary to effectively use numerical simulations as a predictive tool for cabin component development under crash loads. Therefore, in this study the structural response of overhead bins under crash loads is investigated in a more detailed way than it was done in the past. Dynamic finite element (FE) simulations with the commercial software LS-DYNA are used because in combination with today’s computational performance they have the potential to allow for detailed investigations of failure processes and parametric studies combined with a minor amount of time and cost compared to prototype testing. However, because the accuracy of the simulation results has been shown to depend on the accuracy of the FE-model and its input data, extensive test series were conducted on the sandwich material, joints, baggage and finally the whole OHSC under dynamic loads. But still any model will suffer from numerous uncertainties resulting primarily from the properties of the carry-on baggage and secondarily from the manufacture process by hand [40]. Therefore, uncertainty assessment tools were used to analyse and quantify these influences. Rather than evaluating the crashworthiness properties of a specific overhead bin design, the focus of this article is to establish the methods for characterisation and adequate numerical modelling

Figure 1. Overhead stowage compartments in aircraft cabin.

for dynamic simulations of such composite aircraft interior components. Overhead bins in current study This study was conducted on prototypes of two different OHSCs suitable for a wide-body aircraft, which were specially designed for this investigation. One is a fixed lateral bin and the other is a centre bin with two movable compartments for both gangways (Figure 1). The bins are both made of composite sandwich structures with FRP faces R and a Nomex honeycomb core, but they were produced in two different ways by the manufacturer Comtas, Hamburg. For the centre bin a mould was used on which the sandwich lay-up of inner FRP prepreg, honeycomb core and outer FRP prepreg were placed, sealed in a vacuum bag and co-cured in an autoclave, that is the prepregs of the faces were hardened and bonded onto the core at the same time. The lateral bin, on the other hand, was produced with the cut-and-fold technique [29]. This state-of-the-art manufacturing technique requires just one pre-made flat sandwich panel that is typically produced on a hot press in a co-curing process. If a stripe of one skin of the sandwich panel is cut out and the core material is removed to a certain degree, the panel can be bent along this line. This edge can be fixed by adhesive and a reinforcement layer of FRP. The complete housing of the overhead bin can be folded in this manner from just one panel. Both OHSCs were equipped with energy absorbers included in the support structures that were developed by the Technical Monitoring Association ¨ ¨ Rheinland [15]. Technischer Uberwachungs-Verein (TUV) These energy absorbers limit the maximum load acting on the cabin components and the respective sandwich panels.

International Journal of Crashworthiness

513

Figure 2. Tensile stress-strain curves of E-glass fibre-reinforced phenolic resin face material at different strain rates.


Material testing and modelling To characterise the mechanical properties of the composite sandwich material of the overhead bins under quasi-static and dynamic loading conditions, extensive tests were conducted on coupon level. The sandwich faces were made of E-glass fibrereinforced phenolic resin (GF/PF) with the material specification Stesalit PHG 600-68-50. Stiffness and strength data were determined by quasi-static compressive, tensile and shear tests. Because GF composites are known to exhibit a strain rate effect [4,27,38], dynamic tests for the determination of the rate-dependent tensile and shear properties were conducted on a drop tower facility [14]. The results showed a remarkable increase of strength (88% increase) and failure strain (53% increase) for the highest strain rate of 50 s−1 compared with the quasi-static values, although the maximum strain rates in the OHSC under dynamic loading are expected to be lower than 10 s−1 (Figure 2). The sandwich core structures in this study are a hexag R onal Nomex honeycomb for flat panels as well as an over R expanded Nomex honeycomb for curved surfaces. The core height was 15 mm for the centre bin and 10 mm for the lateral bin. The in-plane material directions of honeycomb cores are referred to as the L-direction (ribbon direction) and Wdirection (direction perpendicular to the ribbon), the outof-plane direction as the T-direction (thickness direction, Figure 3). For three-dimensional orthotropic honeycomb material models the complete stress-strain-behaviour for normal and shear loads in all three material directions has to be defined. Therefore, a complete quasi-static testing series was carried out on honeycomb specimens including compressive, tensile and shear testing in L-, W- and Tdirection. Also in this case dynamic testing was conducted to R quantify the strain rate effect of the Nomex honeycomb structures [14]. In accordance to similar studies by

Goldsmith and Sackman [9] as well as Aitken [2], a strain rate effect with minor stress-increases for compression and shear loading was established that could be used for ratedependent material modelling. A characteristic of co-cured sandwich materials is the unevenness of the faces, which is referred to as the telegraphing effect [33]. This effect can occur as a result of the autoclave pressure acting on the faces that are only supported by the cell walls and not supported in the open cells (Figure 4). Furthermore, during the autoclave process resin of the face prepregs flows into the honeycomb core resulting in resin fillets, which are responsible for the face-to-core bonding. An according micrograph is shown in Figure 4. The uneven face, the resin fillets and the low face thickness directly under the cell walls are visible. These phenomena lead to the fact that the mechanical properties of the GF/PF-faces in the co-cured sandwich are different to those of the laboratory GF/PF-specimens used for previous testing. Therefore, edgewise compressive and tensile tests were conducted on sandwich specimens to compare the results to pure GF/PF-specimens. The respective stressstrain-diagrams are shown in Figure 5. Although the initial

Figure 3. Sandwich structure with honeycomb core and its material axes.


514

S. Heimbs et al.

Figure 4. Uneven sandwich faces resulting from the co-curing process.

stiffness is not affected, the strength of the sandwich faces is drastically reduced because of the aforementioned effects. For this reason, these lower material properties were used for material modelling. The face-to-core bonding strength of the sandwich was characterised by flatwise tensile tests and transverse shear tests as well as climbing drum peel tests. During the tensile and shear tests the core failed and the bonding remained intact. This result was expected, because a design rule for sandwich constructions says that the bonding strength always has to be higher than the core strength. A debonding could only be achieved by peeling the skins off the core. The FE modelling of the sandwich structures in this study with LS-DYNA was done with the shell-solidshell approach as the best compromise between computational cost and accuracy [12,13]. The faces were modelled with four-node shell elements with uniformly reduced

integration [22], whereas the core was modelled with onepoint co-rotational solid elements. Neglecting peel loads, a perfect face-to-core bonding could be assumed. But, because the nodal degrees of freedom of shell and solid elements are incompatible, a direct connection with common nodes should be avoided. For this reason and to achieve a correct representation of the bending moment of inertia, an offset between face and core element was introduced and the connection was realised by a contact definition without failure option. For the GF/PF-faces the LS-DYNA composite material model MAT54 was used [22]. This model assumes a linear-elastic stress-strain-relationship up to failure, which is characterised by the failure criteria of Chang and Chang [6]. The corresponding material properties were taken from the experimental results. This material model does not incorporate a strain rate effect, so that this phenomenon had to be neglected in the model. The honeycomb core was modelled with the LSDYNA honeycomb material model MAT126 [22]. In this orthotropic model nonlinear elasto-plastic constitutive behaviour based on the experimentally determined stressstrain curves could be defined separately for all normal and shear stresses. These are considered to be fully uncoupled (ν = 0). To represent the strain rate effect a scale factor versus effective strain rate, which is the Euclidean norm of the deviatoric strain rate tensor, is defined that scales the stress curves.

Joint testing and modelling Past experiments showed that sandwich joints may be potential locations of failure, so that respective load limits have to be known. In this context, the load introduction points, that is inserts and metallic brackets, and the corner joints or L-joints of the overhead bins are of special interest. Therefore, test series were conducted on such sandwich joints on component level.

Figure 5. E-glass fibre-reinforced phenolic resin tensile and compressive stress-strain-diagrams: Sandwich specimens compared with pure E-glass fibre-reinforced phenolic resin specimens.


515

Figure 8. L-joint bending test: Experiment and simulation.


Figure 6. Insert pull-out test: Experiment and simulation.

Flatwise pull-out and in-plane shear-out tests were conducted on potted inserts to determine the failure loads (Figure 6). Pull-out tests were also conducted on metallic brackets, which are mounted on the stowage bins with counter plates to carry the forward loads. The primary test result was the load bearing capacity before failure occurring in the sandwich panel around the load introduction bracket (Figure 7). With these data, FE models of the load introduction points were developed and validated. In this context, the inserts and the connection of the bracket were modelled with the spot-weld option in LS-DYNA [22]. Spot-weld elements are solid elements, which connect two surfaces independent of the respective meshing. Failure loads can be defined for normal as well as shear loading and are combined in a quadratic failure criterion. These values could be taken directly from the insert test results. After reaching the failure limit, the spot-weld element is deleted. Although this modelling will not lead to a correct physical representation of the failure mode, which is a fracture of the potting material around the insert, the load bearing capacity of the joint is correctly represented.

Figure 7. Bracket pull-out test: Experiment and simulation.

The L-joints were characterised by bending and shear tests to separately analyse the respective failure loads. All different types of L-joints that appear in the OHSC were considered, that is simple butt joints, cut-and-fold joints and monolithic FRP joints. The main failure mode in the L-joints was crack opening and propagation at bonding surfaces. To represent this fracture in the numerical models, contact definitions with failure were introduced at the specific locations (Figure 8). A quadratic failure criterion involving normal and shear interface stresses is the basis of this contact type in LS-DYNA. The respective values were adjusted empirically in order to best fit the experimental results.

Baggage testing and modelling Carry-on baggage brought onboard by the passengers into the aircraft cabin is quite heterogeneous concerning its mass, stiffness, damping and density properties. In addition, it is placed nearly arbitrarily in the OHSC. Hence, the baggage introduces one of the major uncertainties regarding the load case of the overhead bins and should be treated with special care. In this study, specific loadings covering a typical carryon baggage configuration had to be identified. In a second step, the behaviour of this baggage under dynamic acceleration had to be characterised to quantify the energy absorbed by the baggage through elastic or plastic deformation. The recommendation of the International Air Transport Association [17] in terms of limiting the carry-on baggage size to 22 × 18 × 10 in. (56 × 45 × 25 cm) and the overall sum of dimensions to 45 in. (115 cm) is applied at most of the airlines worldwide. However, weight limitations are seldom checked and quite diverse depending on each airline. To get an idea about size and weight properties of carryon baggage under real operating conditions, data from his¨ toric surveys [24,25,31,36,37] and a current survey of TUV Rheinland were taken into account. In consideration of perpassenger volume provided by modern OHSCs, the values

516

S. Heimbs et al.


Figure 9. Sketch of test box with carry-on trolley cases mounted on the test sled and simulation model with medium level of detail of trolleys in box (sectional view).

are similar to the luggage densities mentioned by Jenkinson and Torenbeek [19,35]. On the basis of the the surveys mentioned, typical items of carry-on baggage could be identified, for example, trolleys, sports bags or coats. Various configurations of these carry-on baggage items in a simple box have been analysed under dynamic conditions by preliminary sled tests, in which the acceleration and contact force levels between baggage and box were recorded. As a result of these tests, trolley cases in combination with polystyrene foam blocks filling the gaps showed the best performance regarding reproducibility of the loading condition, repeatability of the force impulse applied to the OHSC as well as designing a worst case load (Figure 9). To create a validated numerical model of the chosen carry-on baggage configuration a reverse engineering approach was used. Five different modelling approaches with decreasing level of abstraction were investigated to conduct a convergence study. The simplest model consisted of a lumped mass design, which resulted in a force impulse F of the mass m under acceleration a defined by the equation of Newton’s second law: F =m·a

(1)

In the second more elaborate model all baggage items were represented by one body filling the whole box. A homogeneous distribution of the total mass over the body was assumed, including isotropic linear-elastic stiffness characteristics. The geometrical discretisation of the model with medium level of detail, which presumes an orthotropic linear-elastic material behaviour for the homogeneously modelled trolleys and isotropic linear-elastic properties for the filling material, is shown in Figure 9. On the basis of the third model, important structural parts like the wire frame reinforcement and the light alloy profiles of the trolley cases were modelled separately. Dis-

tinguishing the features with a higher level of detail resulted in the fourth model. The most complex modelling approach was the detailed modelling of each trolley case with its fabric, metal and polymeric parts down to the level of single rivets. The accelerations of the test sled at different tests were used as input data for the simulations. To compare the modelling approaches with respect to the compliance of the force impulse in the simulation and in the test, a measure of performance was introduced. Transforming the transient forces of simulation and test to the frequency domain using a discrete Fourier transformation, the magnitude of error summed over all frequencies could be obtained for amplitude and phase independently. This measure enabled a more accurate way to compare the transient curves than utilising the commonly used summed squared error approach. Calibrating the stiffness properties of the carry-on baggage in each of the competing models was done by using stochastic design improvement [23,28]. This simple heuristic optimisation method populates the design space with possible random samples of the modelled system and identifies a global optimum of the simulation parameters iteratively. For the five different models stochastic design improvements with several thousand single simulation runs were performed. The objective was to identify a parameter set for each baggage model that minimises the described error measure in frequency space of the force impulse from all real test results. The best fit was reached by the model with a medium level of detail (Figure 9). With this approach, an optimum baggage stiffness of 0.5 N/mm2 was identified. The difference between the simulation and the respective test (Figure 10) is of the same order of magnitude as the deviation occurring between repeated tests with the same baggage configuration. These validated trolley models with optimised stiffness parameters were used for the following global OHSC simulations.


517

load case, extracted from the high-speed video, is shown in Figure 11. A bending deformation of the stowage bin’s floor panel can be seen, which leads to a gap opening relative to the OHSC housing. However, the attachment points could carry the load and the structural integrity was maintained.


Figure 10. Comparison of contact force impulse between baggage and box from test and simulation.

Overhead stowage bin testing and simulation The loads acting on cabin components during a hard landing scenario are characterised by a highly dynamic deceleration pulse. With respect to a systematic investigation, the translatory motion of a landing aircraft was split up and two simplified load cases in forward and vertical direction were distinguished and investigated separately. Both deceleration pulses were approximated by triangular time-dependent functions. The peak decelerations and pulse durations for the OHSC were derived based on the dynamic design criteria for aircraft seating systems according to the Federal Aviation Administration regulation Emergency Landing Dynamic Conditions [8], which specifies the forward load case with a 16 g peak deceleration after 90 ms and the vertical load case with a 14 g peak deceleration after 80 ms. The decelerations acting on the OHSC during a hard landing are similar to these aircraft seat values, although not exactly the same. They were derived by scaling these pulses in correlation with data extracted from fuselage drop tests as well as evaluations of past survivable aircraft accidents, without giving more specific details here due to non-disclosure reasons.

Modelling Computer-aided design data of the centre and lateral OHSC were used as geometrical basis for the global models in LSDYNA. These models (Figure 12) incorporated the material and component models described previously

r r r r r

honeycomb sandwich material models, insert model, bracket model, L-joint models and baggage model

as well as specific details resulting from the manufacturing process. The energy absorbers were modelled from a combination of rigid body and discrete elements based on a force-deflection curve as input. The final model of the centre OHSC consisted of about 150, 000 elements and 38 different material definitions. The lateral OHSC model included about 75, 000 elements and 23 different materials. A large number of uncertainties arise in the real test, for example, from the size, weight or position of the carryon baggage items, the deceleration impulse and direction or even from the manufacturing process of the sandwich

Testing Full-scale dynamic tests of the OHSC prototypes were con¨ Rheinland. The overducted on the crash test rig of TUV head bins were filled with trolleys and mounted horizontally for the forward load case and upright for the vertical load case on a test sled. This sled was accelerated to a defined initial velocity and exposed to the specified deceleration impulse. Besides documentation of the structural deformations through high speed cameras, experimental data were recorded through acceleration sensors at the test sled and OHSC, displacement measurement of the energy absorbers and load cells at the mounting points. In this manner, different configurations from single bins up to assemblies with differently loaded, interacting bins were tested. As an example, the deformation of the centre OHSC in a vertical

Figure 11. Centre overhead stowage compartments on test sled in vertical load case after 0 ms and 60 ms (courtesy of Technischer ¨ Uberwachungs-Verein, Rheinland).

518

S. Heimbs et al.


Figure 12. Overhead stowage compartments prototypes for dynamic testing and LS-DYNA simulation models.

structures by hand. Therefore, the influence of such uncertainties on the global structural behaviour of the OHSC was taken into account from the beginning of the modelling and simulation work by means of uncertainty analysis methods [39].

Uncertainty analysis The calculation of the space and time discretised OHSC model in a dynamic simulation deterministically leads to one single solution. But in the real world, system behaviour is never exactly repeatable due to tolerances and natural scatter in the system properties and its boundary conditions [23]. An approach to simulate the consequences of scatter is a stochastic simulation of the system using the Monte Carlo method. For this procedure, the scattering system parameters are defined in a meta-data model with their specific probability density functions. From this metadata model, manifold random samples are drawn for all parameters with respect to their specific distribution function. Multiple duplicates of the deterministic model are created and every one of them is assigned with a set of drawn random samples. All models are computed in the numerical solver and the results are extracted to a uniform database format. The system behaviour of the entire population can be deduced from the analysis of the respective samples. The stochastic simulation results are evaluated using statistical methods. By calculating first and second order statistic moments, the expected system behaviour and the range of scatter is determined. With this information, a statement about the robustness of the system behaviour can be made regarding extreme values or interferences inbetween uncertain system parameters. Evaluating the stochastic simulation of the OHSCs revealed a noticeable discrepancy between the results of the nominal deterministic analysis and the expectation values of

the stochastic simulation. Furthermore, the full-scale tests showed results in the range of the most likely behaviour of the stochastic simulation, differing from the nominal analysis. The calculation of statistical measurements allows the dependencies between uncertain parameters and the system result to be determined. Conducting a principal component analysis of the nonlinear correlation matrix showed which parameters were of paramount influence on the system behaviour. Surprisingly, the conditions of usage showed the most important influence regarding critical load situations of the OHSC. Especially the total mass and stiffness of the baggage as well as the centre of gravity can result in adverse loads on the OHSC.

Simulation results Because each of the component models mentioned above was validated separately with respect to a correct representation of the respective stiffness and/or failure behaviour, the structural behaviour of the global OHSC models in dynamic simulations was expected to be in good correlation to the experimental sled test results. This correlation could be demonstrated in vertical and forward tests for a single bin. Consequently, the simulation models of the assembly, including the interactions betweens the components and the increased level of complexity, can be considered as validated as well. The simulation models could be used to identify critical areas with high stress concentrations, which were primarily located around load introduction points. These are shown in Figures 13 and 14 as the dark areas with high stress levels. Furthermore, the force levels in the attachment points as well as the energy absorber extraction could be predicted very precisely. As an example, a comparison of the experimentally and numerically determined force levels in the vertical attachments in a vertical load case of the centre and



519

Figure 13. Effective stress plot of two lateral bins (empty + filled) in forward load case (sectional view).

lateral OHSC are given in Figure 15, both filtered with a CFC60 filter. The plateau level in these curves is the result from the utilisation of the energy absorbers in the attachment struts. Because of this good correlation, the OHSC models and stochastic simulation methods have been used for pretest simulations to design further dynamic tests with different baggage loading configurations in a larger test setup. In this context, the interaction of two OHSC in the

forward load case was investigated experimentally and numerically (Figures 13 and 14). If both stowage compartments were filled equally, no interaction occurred. In case of different filling, the forward OHSC with less baggage was loaded by the global deceleration impulse and, in addition, by the contact force of the neighbouring heavier OHSC. Both in the forward and vertical load case the energy absorbers significantly reduced the load acting on the bin’s

Figure 14. Effective stress plot of two centre bins (empty + filled) in forward load case (transparent front view).

520

S. Heimbs et al.


Figure 15. Comparison of force levels in attachment points in vertical load case between test and simulation: Centre overhead stowage compartments (OHSC) (left) and lateral OHSC (right).

structure and on the test sled, representing the aircraft’s primary structure. This effect could be confirmed both in experiment and simulation. Overall, in this study dynamic nonlinear simulations showed a high potential as an efficient tool for the design process of cabin components under shorttime dynamic loads.

Conclusions The methods for a numerical analysis of the structural behaviour of composite aircraft overhead bins under transient dynamic crash loads were presented. Different components could be identified having a major influence on the deformation and failure behaviour. Those are the composite honeycomb sandwich material, the load introduction points in terms of inserts or brackets and the corner L-joints. A detailed knowledge of the respective failure mechanisms and failure loads is necessary for an adequate FE modelling, which makes testing on material and component level inevitable. Furthermore, the baggage items inside the overhead bin highly influence the load acting on the structure in terms of their mass stiffness and geometric properties. In a systematic procedure based on dynamic test results, an appropriate homogenised FE model was derived for the trolleys used in this study. On the basis of the validated component models, the structural deformation and failure behaviour of the overhead bins under forward and vertical crash loads could be predicted properly. In this way, critical areas with high stress concentrations and the influence of uncertainties, crucial for the robustness of the system, could be analysed in advance of the full-scale tests. Such dynamic simulations showed the potential to be a useful tool for the design of aircraft interior structures for crash loads. The behaviour for different load cases or different baggage configurations can be investigated numerically, minimising the need for expensive prototype testing. Even

different materials or designs can be evaluated efficiently with regard to weight reduction of overhead bins. Acknowledgement This investigation was accomplished in the framework of the 2nd aeronautical research programme of the Free and Hanseatic City of Hamburg. The authors thank M. Pein (Hamburg University of Technology) for performing the ¨ joint testing as well as M. Sperber and M. Demary (TUV Rheinland) for the dynamic baggage and OHSC testing and the development of the energy absorber element. References [1] Federal Aviation Administration Technical Center, Aircraft safety research plan, Federal Aviation Administration Technical Center, Atlantic City, NJ, 1991. [2] R. Aitken, Damage due to soft body impact on composite sandwich aircraft panels, Ph.D. diss., University of Auckland, New Zealand, 2000. [3] D. Ault, Longitudinal acceleration test of overhead luggage bins in a transport airframe section, U.S. Department of Transportation, 1992, Report DOT/FAA/CT-92/9. [4] S. Barré, T. Chotard, and M.L. Benzeggagh, Comparative study of strain rate effects on mechanical properties of glass fibre-reinforced thermoset matrix composites, Compos. Part A 27 (1996), pp. 1169–1181. [5] A. Byar, J. Awerbuch A. Lau, and T. Tan. Finite element simulation of a vertical drop test of a Boeing 737 fuselage section, 3rd Triennial International Fire and Cabin Safety Research Conference, Atlantic City, NJ, 2001. [6] F.K. Chang and K.Y. Chang, A progressive damage model for laminated composites containing stress concentrations, J. Compos. Mater. 27 (1987), pp. 834–855. [7] Crash Dynamics and Engineering Development Program, Federal Register, Vol. 49, No. 185, 1984. [8] Federal Aviation Regulations, FAR §25.562, Emergency landing dynamic conditions, U.S. Department of Transportation, 1988. [9] W. Goldsmith and J.L. Sackman, An experimental study of energy absorption in impact on sandwich plates, Int. J. Impact. Eng. 12(2) (1992), pp. 241–262.


International Journal of Crashworthiness [10] G.E. Green, P.G. Ketteridge, and P.H. Lynam, Uses of Nomex honeycomb in composite structures: a review, 1st Core Conference, Zurich, Switzerland, 1988. [11] S.M.R. Hashemi and A.C. Walton, A systematic approach to aircraft crashworthiness and impact surface material models, Proc. I MECH E Part G J. Aerosp. Eng. 214(5) (2000), pp. 265–280. [12] S. Heimbs, P. Middendorf, and M. Maier, Honeycomb sandwich material modeling for dynamic simulations of aircraft interior components, 9th International LS-DYNA Users Conference, Dearborn, MI, 2006. [13] S. Heimbs, Sandwichstrukturen mit Wabenkern: Experimentelle und numerische Analyse des Schädigungsverhaltens unter statischer und kurzzeitdynamischer Belastung, Ph.D. diss., Kaiserslautern University of Technology, Germany, 2008 (in German). [14] S. Heimbs, S. Schmeer, P. Middendorf, and M. Maier, Strain rate effect in phenolic composites and phenolic-impregnated honeycomb structures, Compos. Sci. Tech. 67(13), (2007), pp. 2827–2837. [15] D. Humfeldt, M. Harriehausen, J. Schroeder, M. Sperber, and M. Demary, Attachment for aircraft interior parts, European Patent EP1630098, 2006 [16] Industrial and Materials Technology, Crashworthiness for commercial aircraft programme, Area-3 Aeronautics, European Commission, 1992. [17] International Air Transport Association, Baggage Services Manual, 6th ed., 2003. [18] K.E. Jackson and E.L. Fasanella, Crash simulation of a vertical drop test of a B737 fuselage section with overhead bins and luggage, 3rd Triennial International Fire and Cabin Safety Research Conference, Atlantic City, NJ, 2001. [19] L.R. Jenkinson, P. Simpkin, and D. Rhodes, Civil Jet Aircraft Design, Arnold, London, 1999. [20] G.L.W.M. Knops, Aircraft seat and luggage bin sled tests, in Crashworthiness for Commercial Aircraft, European IMT Research Programme, TNO, Delft, Netherlands, 1994, Report No. 94.OR.BV.043.1/GKN AERO-CT92–0030. [21] T.V. Logue, R.J. McGuire, R.J. Reinhardt, J.W. Vu, and T.V. Vu, Vertical drop test of a narrow-body fuselage section with overhead stowage bins and auxiliary fuel tank system on board, U.S. Department of Transportation, Washington, DC, 1995, Report DOT/FAA/CT-94/116. [22] LS-DYNA, Keyword User’s Manual, Version 970, LSTC, Livermore, CA, 2003. [23] J. Marczyk, Principles of Simulation-Based ComputerAided Engineering, FIM Publications, Barcelona, Spain, 1999. [24] J.R. McCarty, SAWE technical papers 552: Industry revalidation of average baggage weight, International Society of Allied Weight Engineers, Los Angeles, CA, 1966. [25] ———, SAWE technical papers 1155: Carry on baggage review, International Society of Allied Weight Engineers, Los Angeles, CA, 1977.

521

[26] R.J. McGuire, Longitudinal acceleration test of overhead luggage bins and auxiliary fuel tank in a transport airplane airframe section, part II, Report DOT/FAA/AR-99/4, II, U.S. Department of Transportation, 2000. [27] J.F. Newill and J.R. Vinson, Some high strain rate effects on composite materials, 9th International Conference on Composite Materials, Madrid, Spain, 1993, pp. 269–277. [28] L. Nilsson and M. Redhe, An investigation of structural optimization in crashworthiness design using a stochastic approach, 8th International LS-DYNA Users Conference, Dearborn, MI, USA, 2004. [29] K. Noakes, Successful Composite Techniques, Osprey Publishing, London, 1992. [30] M.R. Nyden, J.E. Brown, and S.M. Lomakin, Flammability properties of honeycomb composites and phenolformaldehyde resins, Fire and Polymers II: Materials and Tests for Hazard Prevention, ACS Symposium Series 599, Washington, DC, 1994, pp. 245–255. [31] I. Richards and S.B. Abrahams, CAA Paper 88008: Passenger and hand baggage weight survey – Luton Airport, Civil Aviation Authority, 1988. [32] E.F. Schiantarelli, Core composites in Swissair aircraft, 1st Core Conference, Zurich, Switzerland, 1988. [33] B. Schoke, Schaumstoffe in der Innenausstattung von Zivilflugzeugen, Fachtagung Polymerschäume, 1997, pp. K.1– K.12 (in German). [34] K. Schweizerhof and W. Senger, Numerical simulation of hatrack sled tests, in Crashworthiness for Commercial Aircraft, European IMT research programme AERO-CT920030, CAD-FEM, Grafing, Germany, 1995, Report No. AERO 95–2. [35] E. Torenbeek, Synthesis of Subsonic Airplane Design, Delft University Press, 1982. [36] K.S. van Dyke, SAWE technical papers 196: Average baggage weights – Proposals for the ATA baggage subcommittee, International Society of Allied Weight Engineers, Los Angeles, CA, 1958. [37] J. Vassiliadis, SAWE technical papers 1990: Average passenger and hand baggage weight survey, International Society of Allied Weight Engineers, Los Angeles, CA, 1991. [38] J.R. Vinson and B.M. Powers, High strain rate effects in materials in sandwich constructions, Proceedings of the 3rd International Conference on Sandwich Construction, Southampton, UK, 1995, pp. 769–776. [39] D. Vogt, R. Hartnack, and M. Olbert, Uncertainty assessment with stochastic simulation in aircraft cabin development, 6th European LS-DYNA Users Conference, Gothenburg, Sweden, 2007. [40] D. Vogt, M. Olbert, and J. Schlattmann, Uncertainty management for computer-aided engineering in aircraft cabin development, International Workshop on Aircraft System Technologies AST 2007, Hamburg, Germany, 2007. [41] P. Willson, An update on phenolic resins for use in mass transport, marine, off shore and construction applications, in Composites in Fire 3, University of Newcastle, UK, 2003.