This action might not be possible to undo. Are you sure you want to continue?
Finite Element Methodologies Development to Simulate the Behaviour of Composite Fuselage Structure and Correlation with Drop Test
J.F.M. WIGGENRAAD, A.L.PJ, MICHIELSEN, D. SANTORO, F. LEPAGE, C. KINDERVATER, F. BELTRAN, M. AL-KHALIL
Within the frame work of Brite-Euram programme CRASURV 'Commercial Aircraft - Design for Crash Survivability', technology is being developed for the design of composite air frames with respect to crashworthiness. A significant part of the project consists of the design, fabrication and drop-testing of two representative composite fuselage sections, to generate the experimental data needed for the validation of new energy absorbing structural concepts, validation of simulation tools to model and analyse these structures subjected to crash loads.
he design for crash sur\ivabilitv has become of increasing importance, not in the least because so many crashes ba\e been demonstrated to be potentially survivable. Crashes on take-off and landing around tile airfields were shown to be the most conlmon sur\ivable crash scenarios [1 ]. The load cases tinder crash conditions are well established, hence, tire design of traditional metallic aircraft structure with 'crashworflliness' capabilities, i.e. capabilities to protect fire passengers up to a certain limit, is quite manageable. -foda\"s comnl.ercial aircraft are made
mostix of aluminiunl, a material with an
of aerospace research establishments, tmix'ersities, and computer code de\'elopers (which qualify as 'small and medium enterprises'). Tile specific task objectives were (1) to de~ clap appropriate material models, (2) to design and build representatix'e composite aircraft fuselage components, (3) to test these components, (4) to develop and validate computer codes to simulate crash behaviour, (5) to determine the effect ot the crash loading on the occupants, and ((~) to assess the newh de\'eloped methodolog)', and propose design guidelines.
defined by Aerospatiale. The principal difference between the two aircratt (with respect to crashworthiness) i>, that typical commuter aircraft ha\e a very small space behveen the cabin
floor and the Otlter shell, while the larger airliner contains a cargo hold
between fine cabin floor and tlne outer shell, hence contain> illore space to allow deformations in a crash. The ttb, elage sections developed for CRASL!RV were limited to tile sub-tloor structure,, including two frames. This paper prn\ ides an o\er\ie~.\ of tilt'
work related to the conllnuter fuselage
in]lerent capability' to absorb energy in crash st\clarions. Since the last two decades, composite materials are used more extensi\'eh to build aircraft structures. Applicatiim of composite materials to aircraft fuselages rapidly' becomes a feasible option, so the crashworflliness issue related to composite materials has now become a serious isstie. The objective of fineCRASURV project is to develop the technolog} for the design of con> posite air frames with maximum safety with respect to potentially survivable crash scenarios.
fhe ultimate goal of the project is to develop computer codes with which crash beha\iour of composite fuselage structures can be simulated. A significant part of the project consists of the design, fabrication and drop-testing of two representative composite fuselage sections, in order to generate the experimental data needed for the validation of tl~e new code developments. One fuselage section, representative of a conlnluter aircraft like the ATR-42/72, was defined by Alenia. The second fuselage section, representative of a larger airliner like the Airbus A320, was
The design and fabrication
:\n important design requirenlent for
conlposite fuselages is ol.wiouslv to provide a reasonable alllount of erastm'of thmess to the passengers. 'Copying' traditional a]uminium design concepts for composite structures is clearly not pro-
viding tills capability, because energy absorption due to plastic deformation
will not occur. Hence, new design con
The project team was assembled Prom representatives of the aircraft indus\r\.,
cepts for composite structtlres had to be developed first in a 'building block' approacil.
Air Transport /.9-" ~/
SAFE -..> \ "N
\ \ .....
/ ' ;
~ing frame \\
floor beam ," strtts ~i/
be a'p_~- " ~ / / sphce plate
Figure I. Design c o n c e p t composite commuter
It has been demonstrated, that composites can absorb an even superior amount :~f energy' compared to metals , but that the failure mechanisms which govern the energy absorption are very different, based on fibre fracture, friction, etc.. In order to obtain sufficient fibre fracture, it is important to consider crashworthiness not just as a materials aspect, but as a combined materials/ ~tructures/fabrication approach. A :ombination of carbon fibres and .tramid fibres was shown to give good results for components dedicated to energy absorption: carbon fibres for ,trength, stiffness and energy absorption and aramid fibres for post crash integrity. However, as energy absorplion needs to be accomplished in components which are loaded in compres<ion, providing stability against buckling of the crushing laminates is essential. Sine-wave beam sections and cruciforms with closed cross sections, assem:qed from laminates made of hybrid car9on/aramid reinforced epoxy prepregs, were shown to perform well, U~dividuallv, but also when incorporated in box structures. Material data and models for tlnese laminates were provided by NLR, DLR and the Uni\ ersitv of Liverpool.
Alenia was in charge of the design of the composite commuter fuselage structure. The baseline configuration was the current ATR-42/72 configuration, with ring frames, transverse floor beams, and inclined struts supporting the floor beams at points underneath the seats and connected to the frames below. However, to incorporate energy absorbing mechanisms into the structure, certain components had to be dedicated for this purpose, which changed the configuration. A new design concept was developed and discussed at several workshops by the partners responsible for the fabrication task of CRASURV: Alenia, Aerospatiale, NLR and DLR
]he struts were replaced by longitudinal sine-wave beams for energy absorption, based on the favourable results obtained in the box programme. These beams were placed at a 10-degree angle with respect to the vertical plane. It was thought that the rotation of the frames during the crash might lead to a rotation of the beams to a more vertical position. Moreover, crashes do not ah.vavs take place vertically, and lateral loading components can be expected to occur, so a sine-wave beam that is slightly loaded out-of-plane should still be capable of energy absorption. A prexious inclined sine-wave beam test had
shown this to be possible . Sizing of these sine-waves was performed bv DLR, using their analysis methods and database on sine-wave beams. The sinewave beams contained a trigger mechanism along their lengths, consisting of dropped plies in tile web near the bottom cap, to initiate crushing of the web at a dedicated location and at a well defined load. A concept was chosen, whereby the upper part of the ring frame was continued in the transverse floor beam, with the intention that this closed frame would remain intact and provide post-crash integrity, while the bottom part of the ring-frame (below the floor) was attached to the upper part by hinges, in order not to transfer bending moments that might induce fractures of the upper part of the frame, in order to provide controlled fracture of the bottom part of the frame, it was configured as two sections, connected by a splice plate at the impact point. The sine-wave beams and the frames were not connected in order to prevent the frames to interfere with the crushing beams, which are meant to be crushed by the ground. After a certain amount of rotation of the frames, they would dig into the sine-wave beam webs, thereby providing a certain amount of lateral stability to the beams. This design concept was subsequently evaluated at the component lexel: the sine-wave beam and the hinge-frame configuration with splice plate. NLR fabricated two specimens of each component, and tested one of each in static compression, while DLR tested the other two specimens 'dynamically'. In the first of the tests carried out by DLR, the function of the frame assembly with a collapsible splice plate could be proved. The splice plate between the two stiff frame sections broke very early after the impact and then the frames started to rotate around the hinges, exactly in the way they were designed. The excellent crushing behaviour of composite sine-wave beams in combi• VOL, } o No ~';J4 2001
A I R & SPACE E U R O P E
COMPOSITEFUSELAGE STRUCTURE SIMULAtiONS
D L R Sh.J ~?a, WE} BK siru~sturai +rlteuqt++Sert,or~ C R A S L ; R V S_+b T+:~sk4 I Test o~ C o m m u l e r Sine+',v ~,+'e Beams
nearly constant load levd of about
51) kN was observed, from the point
S i N 4 D [3,ma!t c Test:+ D L R 9OOr~f ,.• ~,+r~.... , .u+.. d ~. . . . ." IN3D S!atm
where the web, broken at the trigger line and translated downwar&~, contacts the lower flange again. Compared to the quasi-static compression test carried out at NLR, the crushing load i~ about 25'!;, lower !fi\*Im'2).
The l:inal assenlblv +,~f the o.Jrnrnt.ltt:r sub-floor structurG fij~'lm' 3, consisted ol a number of pre-cured parts, which were bonded togdher, while some part, were also bolted together. Ihe composite parts were the skin (two plates joined together by adhesive bonding along the impact line), with secondarily bonded Ishaped stringers, four C-section frame hal\es with bonded and bolted N>licc plates, two sine-wave beam>, two later al floor 1-beams, each COllsisting oi: four parts, and two rail tracks on top uf the sine-waves, of similar cross section a~, the floor beams. AlunQniunl brackets w e r e used for tile hinges, tile connection between the ends of the floor beams and the skin, and tile connections between the floor beams and the seat track.,,. Ihe connections between ~ine-wavus and skin, and between frames and ~,kin ~xere bonded and also bolted.
1 :~ ' [mnl]
Figure 2. Comparison of stah'c and dynamic sine-wave beam tests,
Figure 3. Final assembly of Commuter sub-floor structure. (Dec NLR)
nation with a ply drop-off trigger was demonstrated in tile second dynandc test carried out at DLR. The 350 mm high sine-wave beam (rather high compared to typical helicopter beams) failed along the trigger line and showed now damage far away from the trigger area. A
Figure 4. Test set up at CEAT.
The drop test
Because ot tilestructural geometry and the reqtlirenlentthat the ltlsda£e >ecLion must impact fileground in a vertical p o s i t i O i l , a g u i d a n c e s v M e l l l w a >
de\ ised bv C E A I which wa> able to litt
ing process with a
the specimen to the predetermined height, to release the structure by a pyrotechnic s}stem, and to guide the structure downward during its fall ifi~urc 4). The s t r u c t u r e \va~, lifted to 2.6 m above ground level in order to reach the vertical speed of 7 m/,+ on itnpact. To simulate the masses ot occupants, seats, and the upper part of the fusela,t~e, the shucture was loaded ~ith 40 k~ at a position near the skin, and with 300 k~
Air T r a n s p o r t
above each sine-wave beam. Part of the
.Guidance system also contributed to the oad above the hinge.
)ifferent types of sensors were set up at . arious parts of the commuter structure n order to provide the partners with 'est data to correlate with their analvsis "esults. Furthermore, the measurements vere intended to support the analysis of he structural behaviour and to estimate he structural energy absorption which s needed to limit the load and accelera:ion of the occupants.
Figure 5. Post-test configuration. (Doc, CEAT)
The overall energy absorption was very low, and not by the mechanisms intended. However, the experimental data were useful to modify numerical models, and to validate the code developments. The failed structure, after being pulled up (it had been crushed flat), is shown in
Failure m o d e
fhe actual irnpact velocity was 7.0c) m/s, the energy at impact 19866J. Xt the first impact, the splice plates .~,hich connected the two halves of the :rames did fail as predicted (after 7 ms). Ihe frames did not immediately rotate ibout the hinge points because of their :onnection to the skin, and started to :~uckle (after 9 ms). The various stringers impacted the ground consecutivel}, md high postbuckling stresses occurred n the frames. The upper part of the :rames then hit the sine-wave beams, :utting into the webs. Subsequently, :he frames broke, one half at the hinge e~el, the other half at 'stringer 3' level after 10 ms), followed by frame/skin :tebonding (10-15 ms), and outward )ending of the sine-wave beams (after 15 ms). T-he sine-wa~e beams hit the ground after 19 ms) at an angle of 10 degrees. I-he outer parts of the sine-wave beams outside the two frames) broke away at ibout the same time as the failure of the ,:rigger mechanism. However, the sine'.~'ave beams did not crush, but bent ~utwards (after 22 ms) until they were ;topped at the next stringers. They did i~ot absorb a significant amount of ~nergy. As a result, high forces were Iransferred to the lower flanges of the ;eat tracks, leading to distortions and (ractures. After 22 ms, the sine-wave >earns stopped crushing and absorbing ,,nergy.
Figure 6. Finite element mesh for pre-test analyses. (Doc. Principia)
32,000 nodes, and had 190,000 degrees of freedom. The constitutive model for the material of the isotropic shells was an elastic-plastic model with isotropic damage and failure. For some components, those not expected to be critical, a simpler elastic-plastic model with strain softening was used. The equivalent isotropic properties were determined by means of single element numerical tests, matching the response computed using the actual lay-up and the response of a single layer with isotropic properties. In the case of the sine-wave beam, calibration was refined further using the results of the dwlamic test performed bv DLR. Pre-test analyses
Partners Alenia, CASA, Dassault and Principia participated in the numerical analysis of the crash behaviour of the commuter structure.
Principia The mesh used for pre-test analyses is
shown in.fi\,ztrt' 0, with symmetry con-
ditions imposed. Bonded and riveted joints were modelled as tied contacts between the different components. The masses that were added to represent occupants and upper structure are not represented in the figure. To represent these masses, beam finite elements were attached to the upper flange of the seat track. The finite element model comprised approximately 30,000 elements,
ALR & SPACE EUROPE • V O L . 3 o N o 3/4
COMPOSITEFUSELAGE STRUCTURE SIMULAtiONS
pact would produce the peak deceleration at passenger level and would lead to the tearing of the sine-wave webs at the position of the trigger. Then, the edge of the web was predicted to contact the inner surface of the skin and to slide sideways, increasing the angle of the web Figure 7. Detail of sine-wave beam predicted by with the vertical. pre-test analyses at 60 ms. (Doc Principia) With respect to the behaviour expected b)' the designers, the most remarkable feature was that the web of the sinewave beam did not crush in the anah'sis as much as expected, but the bob tom of the web slid sideways. Hence, the behaviour obtained in the Ere-test analyses was more or less in agreement with the failure me&anism obserx'ed in the test. Test results could be matched bv introducing a more brittle beha\'iour in the material models, even if simple equivalent Figure 8. Lateral instability of the sine-wave beam, isotropic material (Doc. AlenJa) models were used to represent the were performed using the commercial actual laminates. It appeared that very. version of ABAQUS/Explicit . sophisticated material models, invol\'Fi~¢urc7 presents a view of the deformed ing anisotropic damage or plasticity, shape at 60 ms. The failure of the speciwere not needed, at least for representmen during the drop test, predicted by ing the kind of behaviour observed in the test for quasi-isotropic laminates. the Ere-test simulations, consisted of the The excessiveh, ductile bellax'iour of rupture of the frames at se\eral locamaterials used in Ere-test analyses was tions, followed bv the impact of the botthe result of the process of material tom of the sine-wave beams with the property calibration. ground. It was predicted that this iraTile main lesson for future >inuflations is that, due to the limitations to simulate tile actual physical phenonlena during crushing with regular shell finite elements, material models to be Llsed for this kind of analyses should be calibrated according to an expected failure mode. As an alternative, she/I-type finite elements able to properly represent the mechanism of delamination might bc used. However, these kind of element> are not available at the present in the most popular commercialh' axailable computer codes used for simulating crash tests.
Within the scope to assess effectixe design guidelines for crashworthv aircraft structures, Alenia aimed its D>tcrash comparison o~ numerical and experimental results towards the understanding of win)' the pre-test simulations did not predict the unstable beha\iour of the sine-wave beams. The post-test anal\sis, using PAX/CRASH, was performed in collaboration with CIRA at their own facilities. First, numerical runs at sub-component h'\el were performed in order to determine the energy absorbing charack'ristics of each of the main structural components, and to highlight possible weaknesses in the design assembl\'. Then, the sub-conlponent energy absorbing characteri~a tion was scaled up b\ simulating the complete structure. It was concluded frot;~ the Alenia-CIRA post-test numerical simulation that the pre-test simulation failed because erroneous assumptions had been made [or the coefficient of friction between the sine-wa\'e beams and the ground, for the disbanding/rivet allm\ ables between the skin and the h'ames, and tar the rotational weakness at the joint of seattrack and sine-wa\e beam ()),,~/m' $), It was also concluded, that the crash tests of the sub-components were not representative of the actual loading conditions within the overall assembl\ of the complete commuter ~tructurv.
,'~ design concept was de\eloped for a .~ub-floor structure of a fuselage, repre.~entative of an 'ATR-type' commuter , ircraft, made largely out of fibre reinlorced composite material, with the ~equirement for the structure to be , rashworth> As composite fuselages for :uch large aircraft satisfying this requirement have never been built I,efare, the exercise can be considered to be the first of its kind, and of a highly ,,xplorative nature. The structure was built and tested by dropping it at a verical velacitv of 7 m/s, while loaded vith dummy masses to represent the mssengers and the structural weight of he super structure. The test data was ;uccessfullv used to validate the con> ~uter softx~,'are developed for the simuation of crashing composite aircraft
4ructure. lhe design concept was not successful,
lack of lateral support, these beams slipped sideways, and escaped mast of the compressive loading. It indicates the importance of tile presence of rigidly connected lateral support structure to stabilize those components which are dedicated to absorb energy in a campressive mode. The experimental outcome points at the need to dedicate a future research programme entirely to the development of successful design concepts for crashvxorthv composite fuselage structures, now making use of the computer simulation capabilities developed within the CRASURV programme. •
(3) AL,PJ, Michielsen, J,EM. Wiggen-
raad, Review of Crashworthiness Research of C o m p o s t e Structures, Part 1: NLR-contribution, NLR CR 97046 L, 1997. (4) Hibbitt, Karlsson a n d Sorensen, Inc. ABAQUS/Explicit User's Manual, Version 5.8.
About the authors:
J.F.M. Wiggenraad and A.L.RJ. Michielsen (presently at NedTrain Consulting) work for NLR in the Netherlands, firstname.lastname@example.org (J,EM,Wiggenraad), D. Santoro works for Alenia in Italy, dsantoro@aeronautica,aleNa,it E I.epage works for CEAT in France. fabrice,lepage@ceat,fr C. Kindervater works for DLR in Germany, christof, kindervater@dlr,de F. Beltran (presently at IDOM Ingenieria) works for Principia, Spain, martinez@principia,es Marwan AI-Khalil works for Airbus, UK, marwan.al-khalil@baesystems,com
References (1) R,H.WM. Frijns, JF.M Wiggenraad, A,T.M. Schaefers, Assessment of Surviva ble Crash Scenarios During the Period 1980-1992, NLR CR 94012, 1994,
(2) A.F. Johnson, C,M, Kindervater, H.G.S,J, Thuis, J.F.M. Wiggenraad, Crash Resistant C o m p o s i t e Subfloor Struc tures for Helicopters, AGARD FVPSymposium: A d v a n c e s in Rotorcraft Technology, Ontario, C a n a d a , ] 996
n that the energy absorption capability .vas significantly less than foreseen. [his capability was to be provided by h e controlled crushing of longitudinal 4ne-wave beams. However, due to the
AIR, & SPACE L U R O P E
• VOL. 3 • No
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.