U.S. DEPARTMENT Of COMMERCENatimai TechmncaI Imfamatimim u

AD-A030 967

Advances in Engine Burst Containmentand Finite Element Applicationsto Battle-Damaged Structure

Advisory Group for Aerospace Research & Development Paris France

Sep 76

AGARD-R44S /

L--I

IIIG1

j AGARD RFPORT No. 608

UI on

Advances in Engine Burst Containment-and Finite Element Applications to

Battle-Damaged StructureII /I

) SDEPAIRT'W'Nt OFr COMW RCt CRDISTRIBUTION AND AVAILAILITY

ON BACK COVER , ArI T1

-A,

AGARD-R 648

;4 NORTH ATLANTIC TREATY ORGANIZATION

ADVISORY GROUP FOR AEROSPACE RESEARCH ANID DEVELOPMENT -(ORGANISATION DU TRAITE DE L'ATLANTIQUE NORD)

AGARD Report No.648

ADVANCES IN ENGINE BURST CONTAINMENT AND

FINITE ELF-NIENT APPLICATIONS TO BATTLE-

~DAMAGED STRUCTURE

•. T 21 1976 .

-;k7

Pi

a- J

THE MISSION OF AGARD

The misqion of AGARD is to bring together the leading personalities of the NATO nations in th- fields ofscience and technology relating to aerospace for the following purposes:

Exchanging of scientific and technica: information.

Continuousl) stimulating advances in the aerospace sciences relevant to strengthening the .ommon defenceposture:

lnprovinr the co-operation among member nations in aerospace research and development.

Providing scientifiL and technical adAce and assistance to th. North Atlanvt Military Committee in thefield of aerospace research and development.

Rendering scientific and technical assistance, as requested. to other NA T bodies and to member nationsii c..nnection with rescaich -' id development problems in the aerospace field:

Providing asslstane to ,nember nations for the purpose of incicasing their - ientific and technical potential.

Re,.ornmending effect, : a)s for the member nations to ust their research and developm,:nt carabilitiesfor the common bcnefr, of the NATO community

The higncst authont. uithir AGARD is the National )elegates Board consisting of officiall.y appointed semirreprrsentatiscs from eath member nation The mission of AGARD is armed out through ,he Panels which are.omposcd of expcrts appoint-d b% the National Delegates. the Consultant and l.xchangc Program and tMe Aerosp-:cApplicatton% Studies Program. The results of AGARD work are reported to the member nations and the NATOAuthorities ihrough the AGARD s.ries -f publications of which this is one.

PartiLspatloit in AGARD atisaties is b% .nvitation only and is normall% limited to Citiens of the NATO, nations.

The content of tht% ph1Lation ha% been reprodukeddirectl, fromn matcrial %upp:icd b% AGARI) or the atthor,

'ubhhicd SeptelnbLI i ,

Sop right , A(,ARI) !. e.All Right Re..%ncd

ISBN o. 35.122- 3

Oumu mt 4W.t-

=Prited hi It g:uq mil b1 1:dtg arsd Rr,.du, rg,,in I idIlarrrd I,,a(. V ( 'arta,' t L,,nd,, It IP iill)

REI4RI DOCUMENTATION PAGE

.Recient's Refre Originat's R itfmel 3. Further Reference 4.Security CLAufichtionof Documnent

AGARD-R-648 ISBN 92-835-1224-3 UNCLASSIFIED

$.Origintr Advisory Group for Aerospace Research and Development -North Atlantic Treaty Organization7 rue Ancelle, 92200 Neuilly sur Seine. France

6.Ttle A DVANCES IN ENGINE BURST CONTAINMENT AND FINITE

ELEMENT APPLICATIONS TO BA ITLE-DAMAGED STRUCTURE

7. P ted atthe 42nd Structures and Materials Panel Mecting. Ottawa, April 1976.

8.Author(s) 9.Date

Various September 1976

10. Author-, Addres l.PagesVarious

22

k,2.1)iatribtion Statement This document is distributed in a.cordance with AGAR6poicies and regulations. which are outlined on tl-:Outzid,. Back Covers of all AGARD publications.

13. Keywords/Descriptors

Damage Bursting Structural analysisAircraft !mpact strength ContainrntntAircraft engines Weapons effects

14. Abtact

Aircraft must be designed to sustain damage arising from the impact of a vancy of types ofprojectiles s,..h as military weapons and debris from engine disintegration Recognising theneed for the collection and dissemination of information on this topic. th.- AGARD Structuresand Maierials Pane! has set up a Working Group on the Impact Damnge Tolerance of Structurescharged with the task of producing a D~esign Manual. A soecialist meeting was held in Ankara,Turke, itv September 1975 and tht, L.onference proct-edings have been published ,s tAGA RD-CP-! 86.

Two further papers were presented to a private session of the Working Group in April 1976.The paper by Messrs. Bristow. Davidson and Gerstle reviews some recent research into theapplication of fragment impact studies to an understanding of ergine burst fragmeni impactsand the initial development of an engine buist containment system. The paper by Dr Huangdescribed a method of analysis of a battle-damaged structure using the NASTRAN structural.nalys, program supplement -d by prepro.-essors designed to automatical% generate input C-fata:a .patching technique" is then used in the development of a finite element model trulrepre5senzing a battle-damaged structure.

1 NOR

L ___ _____ _____ ____

PREFACE

Aircraft must be designed ro s.1st:-n damna~e arising from thr imr-.act of a viriety oflype% of prujectiles such as military weapons and debris fromn engine disintegratija. Re-cognioing the need for the collect.on and disserninw~tion of information on thi!- topic, theAGARD Structurts a: J Materials Panel has set up a Working Group on the Impact DamageTolerance of St~uctur.-s charged with the task of p.oducing a De-ign Manual. To simulatethe collection of datz a specialist meeting was held in Ankara, Tu::.-y ir. Sepsensber 1975and the coniference proceedings have tcen pL'blisled as AGARD-CP.186.

Two further papers. which became availabl: too I ite to be presented at this meeting.were presented to a private su-is~na of trxc Wor~ing Grot.1 in April 1976. These were wellreceived by the audience at tti s ses~ion: it was ..-nsidere I that thest would be of immediatevalac to these cc-icerncid with the problemi of impac! damige tole, -ice and therefore thisAGARD Report has ')Ln published to give them wider cirtu~ation.

The paper by Messrs. R.i.Bristow. (".D.Davidson and iMH.Gerstle on ".%idvjnces. in[Engine Burst Containment*' reviews sonic rccent research into the itpplication of fragment I -

impict stil~dics to an understanding of engine bur-st fragment impacts and the initial develop-incnt of an engine burst contairment system. The paper by Dr. Njo. C Huang on -Finite A~

Emcnt Applications to Battle-Damagri Structurc- describes a method of analysis of abattle-damaged structure using the NASTRAN structural analysis ?,ropzam supplementedIby preprocessors designed to automatically generate input data: a -patchi~ng technique" isPten used in the development of a finite element model truly repiesenting i battle-damaged

structure.FN.F.IIARPUA.7Chairman. Working Group onImipact Damage folcrance of Structo'res

CONTENTS

PREFACE ii

Reference

ADVANCES !N ENGINE BURST CONTAINMENTby R.J.iristow. C.DDavidson and J.1s.Gevtk- I

FINITE ELEMENT APPLICATICNS TO BATLE-DAMAGEDSTRUCiURE

by P.C.Huang 7

ADVANCES li' CIiGINE BURST CONTAINMENTby

A. J. Briustow, C, D, Davidson and J. H. CerstieSenior Engin,-ers

P.O. Box 3999Skattle, Wathingtoa 98124

USA

SUMMARY

This paper prosents a partial review of recent ressarch by The Boeing Company Into the applicadlonof fragment impact studies to an understanding of engne burst fragment Impacts and the iritial dev~lop-meal of in engine burst containment system using du~ont iteviar material. All trt work to date has in-volved translational accelerators but spin pit testing Is planned for the near .zture. The program. hasnot yet 7esulted In a s"L.cfactcry cotitaiment system. However, It Is hiped that the basic data accumulated-ill provide 0vr buildlrg blocks in the~ development of design criteria f .r such a system.

In addition to a sumisnry ,if program accomplishments. this raper delves into several areas where un-expected results occurred atujd where Information was ",)tsined that my Influence future fragrent containmentefforts. One of these areis 'nvolves spinning fraWmets. None k~ the predicted adverse effects on Kevltzfabric were found. Another area concerns thermal ef fectb. It was found t. t the efficiency of the barrierin stopping fragments w-As influenced by the tes.perature of the Kevlar.

WIstOD :;trol h oigdmg.Teesuishv rne rmmtoodpoeto ihts

Stnce the early A95Oui a group at Boeing has been Involved In the study of iupact phenoaenology and

velocities of up to 32.0OQ fps down to free-falling rocks with test velocities as low as It Ps. Deriiagthis period, may sh.'elding concepts were developed. The usual method was to perform a number of pirametricto-sts. .Jescribe the d_ %ago (or other parameter of interes.L* In terms of a mathematical model involv.ng pro-jectIe, target. intE.r..eVt and environuenta' chairacteristics. The resulting model was then used to optiwizo-the shield design or to Ievelop desir.. critorna.

1.1itweight shields jere found to ,equitr tu.o properties: shear resistance locally aroun.' the ImpacLtpoint and tensile strentgth co bring the fragment to a stop, after tlzt Initial impact. It was also noted that1otn teasii- and shear ft~tingth requirements i._'uld be reeiced by uLing a shield eatenial with good elasticity.Keeping. the&, requiremett fr. mind. blast container weigtts were reduced by subumituting fiberglrss fabricI for che normal steel shield ,material. (Elasttcity was increa..ed by decigning the overall .ortalner to "give"

locally.) IUhile the ffberglab.s w.,s found to be very efficient. for containing blast waves and residual gases,

3 its poor bending and ctit resistamc. made It unsuitable fot containing fragments. Nylon was found to esve Igood smaW.l fragment su~pping ab~ity, but lacked the t-ntsile atrenith required to contain blast wave* or

jCurrenLly. a new (tunric is 1- io'eotigats... lis tnev fabric, called "Kevlar" by its manufacturer.duPont, is receiving wide public.., d ippears destimn. to find a nunber of military and commercial appli- -

cations. The purpose of this paper toj summirize thxe results of Boting's study of the abi.sty of Keviarto stop fraigments. Potertial i.scs - -rulting shietdA would be to protect explosive storage tanks, tocontuxn ttagasnts trom burst electrical po-er gencratirg turbines in vatiomary nu,-lear power Plants. .and

3 for~ t~a as shields against the blast atvt fragments of HE warheads. C rarti ..lar Interest io 'IGARD would 1'e

3 tt.e use of lcvlar shieldrs around ;et tngtnes to contain fragments ir the case of an Angire burst due tobattle damage. The sizes find shapes of fragments are quite variable iepcm'ng on ti - type of ,ngine hndtht failure mode. Types of disc failure modes are slown In Figure Figure 2 Is A" photo of some of the

resulItirig fragm-nts.

BACKGROUND

In 1966. Boeing carried out ballistic containment t~~eon titanium to provide localized shielding onthe Boeing 737. in 1971. tests were initiated to finti a w en roving fiber3 lass configuration to protect ahelicopter fuselage from engine blade failursts. The woven roving shields wce retained until the enginemanufacturer rectified the blade weckneqs. In 1972. the ncept of a lfghtweCht flexible energy a'.sorbingbarrier utilizing a fabric was conc.-Ivoed. The concept wao 'or the barricr to pceform in a x.-nner similar

to a net, giving on Impact .and then absorb~ng the energy In tevqli loadinr of the fibers. A series ofballistic impact tests was carried out on "S" plass fabric shf.r. and while the results were encouraging.they were inconsistent. This inconsistcntcy was Oue to the susce .tbility of the monofilament glass tohandling damage and the poor cut resistant properties of the filaments.

As a rr~nult ol' a material search, che duPont =aterial Kcvl.r vs.. proposed since it showed good mechani-cal properties (Table 1). The majority of Boeing activity since h's been based on Kevlar material, and thispaper primarily deals with work done in the conainment field w'.n this arimid fiber.

aA

SWield Design A"roach

Mhay ballistic barriers Mve been developed in the past. Nowever, these developments have generallyInvolved silid barrier materials. By using flexible fabric materials, a uumber of additional variableswere introduced; type of weave, fabric stretch, layer Interactions., restraint of individual fibers arotndti-4 barrier periphery. To Investigate all of these variables in a spin pit would be ecoromically pro-hIbltiv&. For this reasoc. it was planned that the majority 4f the development work would be conductedin a ballistic impact laboratory and that developnt would proceed along lines established for solidbarriera. In addition to economics, the ballistic tab approach had the advantages of gr, ater controlover fragment sizes, velocities and impact angles.

In the isct laboratory, the many containment shfcl. configurations could be reduced to the besttwo or three which could then be spin pit tested. The apin pit would only be used to validate and con-firm the laboratory results and to Impart to the fragment a realistic ro:.LIonal velocity.

The early shield designs were corrugated with variation in the depth and spacning of the corrugations.A f ace plate of .028 CIS sheet was provided and the corrutstions were held in pos-tion by wir.-. Flatslields were also designed of Kevlar and impact tested. Th., containment performance (f this flat uhieldwhen based on a weight per unit area was found to be equally effective to the corrugated type shield.Since the flat shield was less costly to manufacture and. als&. had installational advantages, tih! re-malader of the recent shield development program was based on the flat design configuration. The othervariables that were investigated included varying the shield weighf,, the material weave, the effects ofievlar ballistic felt and the effect of locating acoustic lininZ bejind the shield. In addition, barrierswere tested vtth matrix around the Kevlar such that the barrier cou;4 replace some of the existing struc-ture. Altogether. theiv were over 20 different ohleld design conflprations.

Test Eqit

The layout of the impact test facility which is ,hown sche. mtica' in Figure 3 and by photo InFigure 4 was used for the shield testing; it consisted of a 2.5 or 6 In. diameter bore launching barrel,60 iti. 1"g, with a threaded-on breach. The projestile to be launched (Figure 5) was placed in a sabotassembly and loaded into the barrel. The impact velocity was varied by the weight and type of powderctarge. The launcher muzzle extreded into a plenum where the burved gases were vented and a trap wasincorporated to r-train the sabot halves, As the projectile traveled downraae. it passed through twopripted electrical ktrccit sheets, spaced two feet apart, trggerlng the circuit of the electronic velo-city analyzer. The Impact chamber had a movable steel cover with an acrylic window to provide access formoutting of the shields and viewing ports for photogeaphy. The projectile travel from muzzle to shieldwas approximately 10.0 ft.

T e tasidual velocity of the projectile was measured by the number of witness sheets penetrated.(Figure 6 shows witness sheets and a "flat" shield mourted on Its support bracket).

h metal block was used to simulate the fragment and was varied from a 3/4. in. to 3 in. cube withvariations such as changing the aspect ratio (I x I x 3) and cnanging its density by using differentmaterials. The reasons ior using a cube were that it has a cutting edge which would more represer.t enginefragmnts than a sphere, is easy to manufacture, and is inexpensive. It's size can bt varied easily torept,-ent different size frageents. Alsa. the co*p2ct shape of the cube resulted in reasonably consistenttest data. It was the intent to impact the shield at energy levels and velacities which would be expectedfrom an engine. e.g.. no blade velocities in excess of 1200 fps and no blade root velocity greater than900 fps.

Test Results

Figure 7 shows the contain-nt performance of Kevlar shie!ds for two projectile sizes, plotting shieldweights (uninstalled) against contai".i energy. Also shown are the containment energy levels for othermaterials from tests done prcvlou.ly by other agenci'cs and Boeing. These results reflect the flat shieliconfiguration. It was rstablis!,ed that the corrugated type shield and the varyLz types of material weaveshowed no advantage. A balliStic felt-faced shield showed about 5% Improvement which is within the datascatter. The placing of acoustic lining material (either polvi-ide or alumnum) behind the shield showedno significant change in containment performance.

Empirical .4odel

Vupiri.-sl and seal-empirical models have been developed in the past for many types ef impact phenomena;armo" plate, structural damage. n. reoroid impact, and hail to mention only a few. These models have bewnfourd to be extremely useful in dosigning and optimizing hardware associated with the phenomena. For thisreasov, an empirical model was developed to represent the laboratory containz.ent barrier data.

The basic barrier test data was first plotted as a function of a barrier layer and witness sheet pene-tration as shon in Figure 8. (The witness sheets were a act of 1/16 inch aluminum plates Olaced behind thetest barrier. Enough plates were used to assure that the cube projectile would stop somewhere in the set.)The point whe:e the typical "S" shaped curve crosses the t undary t.tweern barrier and wltncss sheet penc'ra-tion is the ballistic limit (incipient penetration completely through the barrelr).

----

An empirical model wait then developed to express the ballistic limit as a function of several of themost important variables; fragment size. velocity. impact angle and number of layers of Kevlar. Th* re-sulting model was:

W -AV' D~' (sinG) _/61-

where W & real density of Kevlar -lb/ft 2

0.067 M. N - number of V-cvlar layersI-(9.6 oz/yj2 fabric)j

V *Impact velocity at ballistic lmit - fps

o - cube size - in~ches

e -*Impact angle seasoirtt from cutbe flifght path to4barrier surface

A

9A & 5 - Constanto

In orde' to make the model more usable, the nomogram of Figure 9 vas dp'veloped. Knowing the fragmnt

size, velocity az'l Impacting angle, the Figure 9 nomogramn makes It very vasy to determine the number of

Ktevlar layers required to stop the fragacrnt.

The barrier wc.ght is seen to depend on the square of the velocity of the fragatent. This appears reasivi-

able since the ftscment kinetic energy depends on the square of the vcloctty. However, the model showsthat the dependent. ean fragment size is only the 3/4. power of the cube dimension. This would Indicate adependenc& on %as& of anly mass to the 114 power. In other words. the barrier wc,3ht Is not stronglydcp~..r, on fragment mass. An Increase In fragment mass by a factor of 16 would require a bprrier weightIncrease of only two.

ESpinnag !Ljawntv th~ n~be etb ooivsiaosta h rgetsi epnil o

W~hen an engine bursts tire resulting fragments have both linear and rotational energy components. The

rato of these types of enwrgy varic-a from all rotational for a complete disc to virtually all linear for

thk rotationtal energy would oevetely limit the usefulnress of fabric barriers. They theorized that therapidly spinning fragment would ac:t like a "buzz saw** to cut through the fibers.

series was set up In which the projectile was made to spin before hitting the barrier. This was .n ore cdtrin fso cli nedcueth rgett awtruhtebareats

F ~ished by plA.ing a 5~ wooden ramp In front of the barrier. 111Uon hitting the ramp, the edge& of the aubeprojectle dug into the wood xn-3 caused the projectile to roll up the ratit. The resulting spin rate wasJeduLed ir,= high-speed motion pictures (tiorinalty 10.000 frames p-er second). Figure 10 shows the resultsof some of the testing. Consider-able scatter is shownm. However. with spir cites varyine from 9000 to24.000 R2I. o degradation in barrier efficiency was fe..ed.

In a complete.y different type of test. some of the effects of spinning projectiles were also noted.In this test series, the effects of projectile geometry vere being studied. The cube projectiles were

replaced by rods one by one by tuo kr three tiches long. fired end-way- Int- the targez as #Nown In

Figure 11. This rotation was, In effect. a high s."ed !.pin by the~ time the rod stnip-A. as shown inV Figure 11.The conclusiov to oe drawn froct thisey tests. and to be later verified ir. a spin pit, is that spin has

little. if anv. detrimen'al effect on tit. ability of a fabric barrier to *'jp famns

V c=j crature Effects

in an attuai, instalat.. thte engine .ontxinmen barrier wit; experience a range of temPerAturCa.The efie~t of th.-se va~latton- under statt. .ondixions was known fro-_ woraL done h% duPont (Figure 12).This work indl~ated thae. Ketlar pr.-pvrtlis were eonstant until a .ertain tcmrer.tur, is reachet and then *

fall o.ft gratuel.. at higher r~.-aus. It was 1c.Aded to .ondukt a gest t. =zk sure that the barrierwould fun,vtion pr~peri% at t, highest teupersture anti.Apoed. 320 0 F. Thez results were rourprising. In-stead of getting - s.ightli reduced ballistic limit as expekted, the W.vlar material appeared toi havebetter ,.roperties .3t hig~her temperatures. 11:, high '...,erature t. ..P were expamnded and indluded some at~old tev-orratures. The results vtre as ,&howr- in Figur-e 13. Abc-ve -. om teeperature, the bAl.islti linir.steadily improved up tc. about 420P . Above this temperature, the eft&ilen~v of the barrier rapidly fell

thi tecr. vria~e n te*;=tatmcine wath e t te~c trie Mt t rait p rte off ectsrIt was stumised that the m. teasing ballisti. it ittmeruewsda osranaeeft;

that strain rate effects tncreased faster with teaptrature titan the stat i, strcnpth decreased. Ini order

adifferent temperatures. The miaroinc wast capable of ttrsin rates from 0.05 to '.5 inlinmIn. Therslsare shtmwn in Figure 14. Strain rate effects do Indeed Appear ito be indicated.

4

Considerable more study is required, but it appears that the variation In ballistic lin.t with temp-erature may be used to advantage In tme design of barriers. However. It appears that temperature cyclingreduceb the Kevlar properties In sorne as yet unknotn manner. Figure 15 shows tests wherein test specimenswere cycled various numbern of times prior to test. In all cases the capability of the barriet was reduced.As shown In the figure. few cycling data points are available, hence there Is much uncertaint) an to theovmrall effects of temperature cycling.

FlrIVjE PIAN~S

The end objective of our work a. Boeing Is to understand design criteria for a lightweight engine burp!containment shield. Present effort focuses on tbree areas; in~reased energy containment. attachment desi,n.a=d spit. pit verification of laboratory Impact testing.

The manner In which the barrier is attached to the aiircraft (or to Itself) is closely related to I-stallation design problems. However. it Is also Importan~ for the development of the barrier itself. Ithas been found that as the size of the fragment goes up, the greater Is the dependency of the ballistl. limiton the method of attachment.

All testing to date has been with translational accelerators ottly. Thi~. year the plan .s to verifythese results in a spin pit. A disc with blades will be made to separate into three .zr four segments.These segmentst will then Impact a barrier which Is designed on the basiif of translational tests.

iX)CLUSiO!Ns

P'rogress has been made in the development of technology fo: engine burst ,ontaInment. However, a numberof major problents must be solved before a barrier of this weight ian be placed in an air.traft. Thesc prob-lems include pra..tical attachment concepts, temperature cycling degradation. arnd the vxtrapolation of currentcontainment designs to the higher energy levels.

ROM SEGMEPET AIIM-*f S -

HU2 OR SECTOR SHAFT

Figure I Disc Failure Mocks - Figute 2. Engine 3umi Frapnsefts

Table 1. En"in Butst Contamnment Material Properties

cOuluwOul coomwousjt L54V -S- CLASS IMAAZ

rEWWAn S15(uGNT IV"'

FIANET- "SAN on an

MEIGHT FATIO on 5S

LOPSTWITM IPI -$"MUE - NWCk usGOSE N.

I T W OF AVE TbG4.ASI. XODcnoimsa

-I- CO~dNOLPANELS AItR -RANGE NO. 2 1 COND

OFIE AND TARGI.T STORAGE S T'014AwL 4AREAI PREPARATION AREA AREA TOAGE

LL 4I"I

i I1. LAUjNCHER POSITI1N 4. TEST SECT10ON2. PLIENUM. BAFFLE. EXHAUST DUCT. AND SABOT TRAP S. RANGE COVER &BACKSTOPI YfLOCITY MEASUREMENT

Figure 3. Impact Mechanics Laboratory - Floor Plan

IAFiure 4. Impact Mechanics Laboratorv - No. 2 Test Rang Figure S. Simulate Engin Burst Fragmei ts

20 ~ ~~ - 0 TTAtADES Sd.ATtD)LANL FM.C 41- STIL LAC)

M AASA C EPof.TSTL C(

0 GLASS1.0-* KnOf STEEL COMOSITI

vr(s1woGooust spa fT Tnlts1.K *i"STL

UMNSTALLIEO WEIGHTiAREA (LSSFM'

~I-&lue 6.Test S~iecsen in Range Figure 7. Containment Pe-formmnce

FRAGMENT IMPACT ANGLE -I DECREESPENETRATION

SHIELD "'I. .

aSINCUS-- :: z

C IL

VELOCITY

Figure 8. lest Data FRpm 9. Empirica Engine 8"e Containmeat Mode!

6o

WITNESS SHEETS 7PENETRATED PENETP'ION WITHOUT+~l HI SPIN(

to WITHOUT SHIELD SPI/Au+ WIHU SHEL 1 -Ml /

, W /,-U '. tMo '

'I . CU r =m .

i~IN ~U 92 PM \ ,'-TJMBLE RATE-960 RPM

t D BARRIER , *cgAT pjesr ,elu~nec

1.5 2480 CPBE 5 RPM 1 4o

VELOCITY "IME AFTER IMPACT-I SEC

Figure 10. Effect of Spin Figure 11. Tumbling of Long Projectiles

14 - --.

<1 IX

w 2

t- z 10 - --Ml

p2- 482-F- -

150 158 RP0i

-100 0 100 200 2I00 400 500

0 1000 2000 30340

VE-OUT TSPECIMEN TEMPERATURE- F

Figure 12. Temperature Effect - Static Loading Figure 13. Effect of Temperature

zZ " EI MTHATS I/ AL CRCEDONSA - ., I / J/' Lt..

010

T32UEF

toa ISO

Fiue1. eprtueSri.Rt fecsFgr VI5. teprtr ycigEfc

0x 100 2014000

CJT

,TEMPERATURE - ..

Figure~~~~~~~~~ 141eprtr/tri2aeEfcsFiue1.Tmeaur0yln fet

7

FINITE ELEMENT APPLICATIONS TO BATTLE DAMAGED STRUCTURE

byDr. Pao C. Huang, ConsultantNaval Surface Weapons Center

White Oak LaboratorySilver Spring, Maryland 20910

United States of America

Summary

This paper introduces a 'Patching Technique"for the development of a finite elementmodel truly representing a battle-damagedaircraft. The applications of preprocessorsPING and BING to the automatic generation ofinput data for NASTRN analyses are alsobriefly shown. Finally, the importance ofmodeling technique is addressed.

(1) INTRODUCTION

The utilization of the finite element method in the design and analysis of a complexmissile structure is an important step toward an optimized system. The great speed andprecision of these techniques permit a missile to he thoroughly and repeatedly analyzedand modified, providing a more efficient weapon system. These same techniques areperfectly applicable to the structural vulnerability analysis of a battle-damagedvehicle which, in general, has a complicated configuration and requires detailed stressand deformation data. However, the success of this computer approach depends largely onthe modeling technique used to develop a "inite element model which accurately simulatesthe real structure.

In addition, the generation of such a model and its associated data output must bequick, economical, and routinized.

(2) FINITE ELEMENT PROGRAM - NASTRAN

NASTRAN, a general purpose finite element program for structural analysis developedby NASA, can be used as an example of a program for which data generation can be anoverwhelming task. NASTRAN is versatile and applies to a large class of static anddynamic problems as follows:

a. Static response to concentrated and distributed loads, thermal expansion, andenforced deformation

b. Dynamic response to transient loads and random excitation" c. Determination of real and complex eigenvalues for use in vibration analysis andstability analysis.

However, preparing NASTRAN input for a complicated problem is a large effort initself. It includes the layout of a grid mesh; calculations of grid point locations;gener-:ion of element properties and their connections to grid points; preparation ofinput sheets; and, finally, punching of input cards. All this work is extremelv timeconsuming and requires the effort of a team o: specialists. Furthermore, possibleoccurrences of human errors tooether with unavoidable redesign cycles would make theanalysis formidable.

To remedy these difficulties, special input generation proorams have been developedby the Naval Surface Weapons Center, White Oak Laboratory to automatically generatefinite element models and their associated NASTFAN inpul for missile structures.

(3) PPEPROCESSORS FOR NASTRAN

A Planform Input Generator (PING, [refcrence 11 which can rapidly develoo a finite *element model for an arbitrary wing planform of a missile, and a Body input Generator(BING) Ireference 21 which performs the same function for any type f missile shell bodyare currentl,. operational at NSWC/WOL. Utilizing these t.'c preprocessors, a completemissile such as shown in Figure 1 can be easily and qAick.ly generated for rigorousstructural analyses. These two preprocessors contain mrny useful features which can -eutilized to facaltate the development of a model to truly simulate a real missile.

The efficiency of PING is demonstrated in Figur, 2 where a finite element model of adouble wedge, solid, sweptba,:r wing is shown as an example. This finite element modelhas 277 qrid points, 260 plate elements, and 1400 degrees of freedom. PINq was employedin the development of this model. it only took iive manhours and a few seconds ofcomputer time to produce 1137 input cards for NASTRAN analysis.

Without PING it would take at least 360 r nhours to produce the same number of cardsmanually. The capai'ility cf developinc a finite elem -" ndel ravidly and economically

SW

is an essential requirement for such complicated work as the vulnerability study ofbattle damaged structures.

Figures 3 to 8 show some of the sample finite element models developed by PING andBING. Figure 3 demonstrates the useful feature of a mixed mesh which can be used toreduce the degrees of freedom in lIfss important areas. Figure 4 shows a model of acurved delta wing. Figure 5 ill,,trates how patches with a circular or elliptical holecan be used with a finer mesh aroundi the hole for more detailed stress distributon.Figure 6 shows a wing with an elliptical hole which can be developed in two stages.First, a wing model wit-hout hole isi developed and the elements around the hole areremoved; then a patch with the hole can be developed to fit in the onening. Figure 7shows a complicated but realistic missile body generated by BLIX. It has a nose coneand a cylindrical shell with an engine housing. Figure 8 shows a finite element modelof a BQM 34A wing. This built-up w,.ng has three spars connecting two curved pax.els.Again, d mixed mesh is used ,n the wing panels to reduce -he degrees of freedom withoutsacrificing the accuracy i the important wing root area.

Using the preprocessors and NASTRAN, r-liable analytical results can now be obtainedeasily and iapidly. Figure 9 shows the good correlation of the theoretical andexperimenftal data. The latter were provided by the Naval Research Laboratory from acompl.te vibration test of the actual missile. The theoretical results were obtained byMASTAN on a finite element model developed by PING and BING. It only took 40 manhoursto complete t-he vibration model and 1200 seconds of CDC 6500 computer time to obtain thefirst six vibration modes. With these powerful analytical tools, elaborate structuralanalysis ce-ainly becomes practical in the development of a complicated missile system.

(4) PATCHING TECNIQUE

On a demanding task such as the vulnerability study of battle damaged structures,precise stress data around the damaged area will be required to determine t-hepossibility of crack propagation which may lead to catastropnic failure. Finite elementanalysis techniques wil) be of great value in furnishing detailed information for such astudy. Therefore, an elaborate finite element model must be developed to accuratelyrepresent the damaged structure. Using PING, t-his task can be easily achieved. Figure10 shows tne procedure for t-he development of a damaged wing structure. First, a finiteelement model of the undamaged wing must be generated and the plate elements around thedamage remove.a. Then patcnes of damaged components having smaller plate elerents can begenerated to fit the cutout. Finally, these patches are inserted into the properlozations to complete the finite element model for the damaged wing. This modelingprocedure is aes-gnated as the *Patching Technique. The co-=cn grrdpoints on thecutout oundaries are not joined by compatibility conditions but by using a commongridpoint number, therefore, no constraint equations are required. This feature notonly eliminates many input data cards, but also yields a much better finite elementmodel.

(5) ANALYSES OF BATTLE DAM1AMD STRUCTU-S

A d.maged BOZI 34A wing was analyzed by NASTPIN under la loadinG. The finite elemo-ntmodel of the undamaged wing had 846 ;rid points, 849 plate elements, and 4602 degrees offreecom. This model not only too.. into account the curvatures of the skin panels, butalso the varying skin thickness. Five PING runs and approximately 64 manhours weretaken to complete this job. The damaged wing had a slot 1.5 inches wide in the upperskin panel and extended five inches diagonally inward from the front spar to a pointwhe.e the middle spar cap was complete-y cut. In addition, all three spars were damagedto different degrees. Using the patching technique, the finite _lement model for thisdamaged version was quickly generated by removing elements and adding patches. Thefinal model is shown in Figure 11. The NASTRPM analyses showed significantly higherstresses in the root area of the damaged wing than those in the corresponding region ofthe u.'damaged wing. However, the most critical point was found at the slot tip with apeak stress of 13,003 psi as indicated in the plot of iso-stress-lines. The actualdamaged wing was tested to destruction, which occurred at a 3g loading. A crack wasfirst formed at the tip of the slot then propagated toward the wing root at the rearspar. This was adequatelY predicted in the stress contour plot whereby a peak. stress of30000 psi was obtained at 3g level; a value wnic-h was thought to be the strength of thematerial at that state.

Tne structural response ' a damaged semimonocoque fuselage was analyzed from amudel by the same patching technique. A vert-:al slot between the second and thirdframes near tne tail end was cut in the fuselage skin. The cut extended from the crowndown to a point at 108 degrees on each side. The patch had a width coverinc the lastthree Loays of the fuselage and a circumferential length longer than that of tne slot.ts can ne seen in Figure 12, much smalle- elements are used in t-he patch especiallyaround the tip region where high stress concentratin is anticipated. This finiteelement model was subsequently analy7ad by NASTICv for an ig load condition. Properresultant forces were placed on the -our cut boundaries to transmit air pressures frortne nose. wings, and tail assembly which were omitted in this truncated fuselage.:tress concentrations were found around t-he slot tip area with sufficient intensity suchLnat a crack would form at this point. Catastrophic failure seems certain to occuroecause taere is no crack arresting structural member in the propagation path betweenframes, ind tho peak stress of 26000 psi shown in Figure 13 would Increase rapidly asthe crack enlarges.

- ---- £

9

To demonstrate the importance of the patching technique which provides stressconcentrat .on data, another analysis was made on a model obtained simply by rew.ving theplate elements between the two frames to simulate the damage inflicted on tht- fuselage.This simple model, Figure 14, had a cutout wider than the slot but was bounded on foursides by smooth edges. NASTRAN analysis revealed much h.ghex stresses than those of theuLdamaged model, however, no aanger could be detected for any catastrophic failure inthis analysis. This contradictory conclusion, of course, was not totally unexpected,especially when the simple model had a structurally sound configuration. However, itdirl point out the danger of under-modeling and the importance of a True simulation.

The study of structural vulnerbility of battle-damaged aircraft by projectile hitscan be conveniently treated in two p.-ase. In the impact fracture phase damage criteriaof instantaneous failure, i.e. the uj.cer bound of vulnerabili-y, are of interest. ThEnthe assessment of local damages to the structure must be determined for subsequentsurvival a" lyses. The prediction methods in these areas are inadequate at the presenttime, however, some guidelines are available for a quick estimate. In the continuedflight phase where residual strength, loss of control, dynamic and aerodynamicinstabilities are a concern, the finite element techniques become extremely valuable.To attack these problems two distinct finite element irodels should be employed. A"Static Model" with sufficient damage details can be developed for residual strengthanalysis while a "Kinematic Model", being m-ch simpler than the former, can be used inthe analyses of structural response. The kinematic model is definitely the cheaper oneto run, yet would yild sufficiently accurate results in stiffness or dynamic analysis.For completeness, many different analyses muzt. be performed on a large structure, thesemodels must be carefully designed for optimization in terms of accuracy and economy.

(6) CONCLUSION

Using the preprocessors PING and BINC together with NASTRAN, finite elementtechniques can now be applied easily and rapidly to the study of structuralvulnerability. Precise response data can be generated, analyzed for the residualstrength and performance of the damaged vehicle.

Patching technique has been applied to the development of a finite element model fora damaged vehicle. The procedure is found convenient and efficient. Good stressconcentration data have been obtained.

Modeling to truly represent a damaged vehicle -s an important task. It should notbe over- or under-designed for economical or accuracy reason.

References

1. Huang, Pao C., and 'Iatra, John P. , Jr., "Plan form Input renerator (PING), A NASTRPNPreprocessor for Lifting Surfaces-Theoretical Development, User's Manual, andProgram Listing," "NOLTR 73-109, Naval Ordnance Laboratory white Oak, Silver Spring,!-ID, Dec 1973

2. Huang, Pao C., and Matra, John P., Jr., "Missile Body Input Generator (BING), ANSTRAN Preprocessor, Theoretical Development, User's Manual, and Prograr Listing," A

NSWC/WOL/TR 75-9, Naval Surface Weapons Center, White Oak Laboratory, Silver Spring,VID, Mar 1975

10 4

~v

DATAN GENERATION~T P G NSTRAN- UFTNG SMACES

BIN GBIRUTE

mm~~~s~ms °" - o-mi

FI A COMPLE FINTE ELEMENT MODELFOR ELADORTE STRUCTURAL ANALYSIS

DATA GENTION 0 TRAN PING 0NASTRAN

BASIC FLArPOARIT? 6 GRID POINTSCOR TIO !lPRESSURE I

WFDIiTWLC'AD

277 GRID POINTS1 113F INSANS DRAG

LOAD

11 MESH LAYOUT

TIOE I NO OF TIME 1 0 OFMUAN URI CARDS MUAN HOUR'] CiS

COSPUTATION 20 0 3 0

WRITE- UPS IN I 1PROGRAM FORIIATS 4

CAI! PUNCHING 24 1137 5 I 29VERIFICATION ANDCORRECTION 11D.ES7 16 114 5 3INPUT CARDS REO'D13113FOR NASTRAN 1137___ 1137__

CARDS MANUALLY1213PUNCHED 1251__32

TO0TAL TIME REOVIRED 360 5

FIG. 2 EFKI30ECY OF PWN/NASTRAN

II - -4

REQUIRED INPUTS CARDSNASTRAN PING

NODES 277 GRIOICARDS 1

ELEMENTS 230 COUAD2 CARDS I30 CTLA2

236 PUIIAD2 A30 PTIlA2 i 7 7

797

" I

11385 DEGREES OF FREEDOMI

SFIG.3 SPMW SOW WING PL FOM

ItY-AXIS

k-l2

±12 X-AXIS

Mcr

T,~

0.1

FIG. 4 KAMMIE FINTE EBlEMN M O F A CYUNDRI WING

- -5

12 "

1 .4

\ \

I I 4 4 ' \

FIG. E WING WITH ELLIPTICAL HOLE

7- -l-

FIG. 7 FUSLAGE WITH ENGW HOUSING MUM BY BIN

~2I

B 34A WiG ACTUAL CROSS SECFlON

FfUlTE ELEMENIT MOEL DEVELOPED BY PW FNITE ELEMENT MODEL CROSS SECTION

FrG. 8 W ELEMEW MOEL OF BULT-UP WING DEVELOPED BY PING

14

f- - ' - - -,.

f 36.2 H10 ',". f 41.1H ",1; f: 45-2 Nz " .TESTDATAf-31Hz TESTDATAf 32Hz TESTDATA f, -2zFI. 9 OPRSN -TH HN AADA

DATA OF A VIDRATION IMUE DEVELOPED BY PMIG AND BWIM

.+ S K IN

FIITE OLMINT MiODEL Vt WNIG PATCH OF PALMAGED COMPONENTS

ROOT SECTION GiNERTA BY PING GiNMc"' BY PiN TOFIT THE CUT-OUT,.

IG. 10 ' Of W E I-f

M FO- --

BQM 34A DRONE ISO-STRESS-LINES FOR~ DAMAGED COVER7OF THESOM 34 AWING

I 6 L3AOftG

- ~33000 PSI

000 Ps

s 000 PSI

10 000 PSI

FUITE ELEMENT MODEL. OF DAMWiEO SOM 34 A WING 1XIM 34 A DAMAGED WIMIG WWIIE

'~~S~Co DU to x cMTY.o6

FIG. 11 VUUERABIIT STUOV! OF A DAMAGED MISSIL WIG

4- & A

F1NITEE REMOELT ODELTmF~uC U

A DAAE FSLG

FIG 12PVL~fO K LMN OE

FOR~~~~ ~ ~ A~ DAAE UEAG IHB

4, II)

VOW&E W.T m~#E FRAMES

FOR NASTRAN ANALYSIS

OSTORTION OF DAMAGED AREA UNDER 1-G LOAD CONDITIONHIG SiRES AT DAMAGED 'IP INICATESI ~CRFCA CONDION FOR W" PROPAGATION

FIG. 13 VIINERABLITY STUDY OF A DAMAGED MISSILE BOY

11,59 P PRINCIPAL STRESS

/i

0- 10,O000PSIB 400JPSi

!-1z 2000 PSI I , M SE)-,0- z 1000 psI"

FIG. 14 DAMAGE MODEL WITH SMOOTH SIDES

NATO +,OTAN' DISTRIBUTION OF UNCLASSIFIED

.NATONAL DISTRIBUTION CENTRES -BELGIUMi ITALY !

C7ordonnter AGARD .2VSL Aeonautica MAitareEtat.RD o de hI Force rf enn Uffpcio del Deleto nial diu oAGARDCaserne Price Bauto in 3. Mmrazahu AdenauerPaeDaIlly. 1030 Bruxelles RornaEUR

CANADA LUXEMBOURGDfence Scentific Information Sery be p s a i fmfDeament of National Defence NETIERLANDSOttawa. Oetau I Forcet del and legation to AGARDD laRe Ntiod Aerospce Laboratory. NLR

D~ENMARK P.O. Box 126Danih Defence Rt.zh ksd Delft9stetbtogadet Kastr Dlf ccopn* 0 NORWAY

Norimlian Offence Research Establidunent

FRANCE Main LibraryO.N.E.R.A. (Direction) P.O. Box 2519 Avenue de !i Division Leclerc N-2007 Kjelkr92 Chitillon isous Ragneux PORTUGAL

GERMANY dtreccao do Servico de MaterialZentrahstelle rr Luft. und Raumfahrt- d Forca Acresdcumentation trd -information Rua de Escola Politecnics 42D.8 Mnchen 86 LisboaPostfach 860680 Attn. AGARD National Delegate

TURKEYGREECE Department of Research and Developnent (ARGE)

Hellenic Armed Forces Command Ministry of National Deftuce. AnkaraD Branch. Athens UNITED KINGDOM

ICELAND Defence Research Irformation CentreDirector of Avatior, Station Square llotwec/o Flugnd St. Mary CrayRokjvik Orpington. Kent BR5 3RE

UNITED STATESNationa! Aeronautics anr. Space Administration (NASAl.Langley Field. Virgnia 23365Attn Report Distribution and Storage Unit

THE UNITED STATES NATIONAL DISTRIBUTION CENTRE (NASAl DOES NOT HOLDSTOCKS OF AGARD PUBLICATIONS. AND APPLICATIONS FOR COPIES SHOULD BE MADE

DIRECT TO TIlE NATIONAL TECHNICAL INFORMATION SERVICE (NTIS) AT THE ADDRESS BELOW

PURCIAc AGENCIS

Microfiche or Photocopy Mkrofiche MicroficheNational Technical Space Documentation Service Technology ReportsInformation Service (NTIS) Europ.-an Space Agency Centre ()TI)5215 Port Royal Road 114. Avenue Charles de Gaulle Station Square HouseSpringfield Q2200 Neusly sur Seine. France St. Mary CrayVirginia 22151. USA Orpington. Kent BRS 3RF

England

Requests for nticrofiche or photocopies -if AGARD documents should include the AGARD srial ,iumber. tith, author or editor, andpublication date. Requests to NIS should include the NASA accesion report number. Full bihographtcal references and atstracts

of AGARD publications are pven in the foilowing journals

Scientific and Technical Aerospace Reports (STAR). Government Reports Annou,.ements (GRA).published by NASA S.;entific and Technical published by the National TtchnicalInformation Facility Information Services. SpringfieldPost Office Box 8757 Virginu 22151. USABaltimore/Wadihngton International AirportMaryland 21240. USA

Pnnted by Technical EdirMtg asnd Rqpruuctk LidlIlford hi)us.. 7 9 axdotte St. lawdon WIP iHD

ISBN 92-h35-1224-3