FATIGUE BEHAVIOUR OF COMPOSITE JOINTS WITH HEXAGON BOLTS Romn Strikov nd Jokim Schön Deprtment of Aeronutics, Royl Institute of Technology SE-1 44 Stockholm, Sweden Structures nd Mterils Deprtment, Aeronutics Division, FFA Swedish Defence Reserch Agency, SE-172 9 Stockholm, Sweden SUMMARY: An experimentl ftigue progrm ws crried out on composite joints with titnium bolts. Single, double, nd triple-row double lp joints were tested in cyclic loding with the stress rtio, R = -1. In order to study the development of ftigue dmge in the joints bolt movement, reltive displcement between composite pltes, nd strin guge mesurements were used. Ftigue tests hve shown tht the ftigue resistnce of multirow joints is better thn tht of single-row joints. Direct mesurements of dmge development showed significnt influence of reduction in the pre-stress in the bolts nd ftigue degrdtion of the bolt holes on reltive displcement between composite pltes nd bolt movement. There ws significnt effect on the strin evolution from the bolts during ftigue loding. KEYWORDS: Bolted Joints, Ftigue, Bolt movement, Strin INTRODUCTION Advnced composite mterils re experiencing incresing use in the erospce industry becuse of their high strength-to-weight nd stiffness-to-weight rtios. In ircrft structures the lod between different prts is trnsferred by mechniclly fstened or bonded joints. In comprison with dhesive joining, mechnicl joints re used in cses when thick lmintes hve to be joined, when there is requirement for different prts of the sme structure to be disssembled for inspection, or repir purposes. However, bolt holes re stress concentrtion fctor in the composite joint pltes, which could significntly reduced the mechnicl performnce of joined structures. On the other hnd, ircrft structures experience vst spectrum of extreme service conditions tht include the presence of ftigue loding. The certifiction of ircrft requires tht ircrft components mnufctured from composite mterils be ble to withstnd the presence of ftigue loding in service [1 3]. Therefore, it would seem tht the ftigue behviour of composite joints plys n importnt role in the design nd performnce of ircrft composite structures. As previous investigtions hve shown, bolt movement is n importnt prt of the overll behviour of composite bolted joints. However, only few methods of boltmovement mesurement, such s the use of potentiometer trnsducer, low-kilovolt rdiogrphy, nd shdow moire techniques, hve been mentioned in the published literture [4 ]. Since most of them re quite complicted to mount nd use, new simple techniques re needed in order to observe processes ssocited with the bolt/composite
interfce. According to Ref. [7], the presence of fsteners hve n influence on the evolution of strin between them under loding. Therefore, it my be of interest to study this effect involving strin guge mesurements during ftigue tests. As it would seem, more experimentl dt nd some development in experimentl methods re required to understnd more deeply the ftigue behviour of composite bolted joints in order to improve the totl mechnicl performnce of composite structures contining bolted connections. The objective of this investigtion is to obtin experimentl results on the ftigue behviour of composite joints with different configurtion nd geometry. Strin guge mesurements re to be crried out between bolt rows during ftigue tests. Bolt-movement mesurements re to be done to nlyse the ftigue process in the bolt/composite interfce. Reltive displcement between the joint pltes is to be mesured using extensometers. EXPERIMENTAL PROCEDURE Specimen pltes were mnufctured from crbon fibre/epoxy mteril system (HTA7/37). A qusi-isotropic ly-up, (45//9) nd (45//9), ws used for outer nd middle pltes. The composite pltes were joined by two, four, nd six titnium Hexgon bolts s shown in Fig. 1. 54 24.24.24 Η1 Η1 15 54 24 15 15 15 272 15 15 15 24 15 248 54 24.24 Η1 15 175 15 24 24 15 272 (c) Fig. 1: Specimens tested in ftigue. The nominl bolt dimeter ws mm. This gives the following geometric rtio of the lminte strip width, edge distnce nd thickness to the bolt dimeter: w/d = 4.5, e/d = 2.5, nd t/d = 1.4. The Hexgon bolts were tightened to torque of 9 Nm.
Strin guges were pplied on some of the double lp joints in order to mesure strin between different bolt rows. Figure 2 shows the loctions of the strin guges on the specimen surfce. The strin guges were glued on one of the outer pltes. 27 2 nd bolt row 1 st bolt row bolt No.3 bolt No.1 SG5 27 3 rd bolt row 2 nd bolt row 1 st bolt row bolt No.5 bolt No.3 bolt No.1 SG1 SG5 bolt No.4 SG4 SG3 bolt No.2 bolt No. SG9 SG8 SG4 SG3 bolt No.2 SG2 SG7 SG2 SG1 SG SG1 3 15 12 12 15 15 3 12 12 3 12 12 15 Fig. 2: Strin guge positions on the specimen surfce with four nd six bolts. In order to mesure reltive displcement between two composite pltes, extensometers were used s shown in Fig. 3. Extensometer holders for extensometer No.1 1 23 ~2 res of ttchment for extensometer holder holders for extensometer No.2 Lod direction holders for extensometer Fig. 3: Appliction of extensometers: for reltive displcement mesurements between composite pltes; for bolt-movement mesurements. To mesure bolt movement, specil fixture ws built nd ttched together with n extensometer to the specimen s shown in Fig. 3. The specimens were ftigue tested t different lod levels with the stress rtio, R = /, equl to -1. An luminium lterl support ws used to prevent buckling of the joints s shown in Fig. 4.
lterl support specimen teflon films loding grip Fig. 4: The testing mchine grips with the specimen nd test fixture. The specimen pltes nd the lterl support were interleved with Teflon sheets to decrese the friction force between the composite pltes nd the fixture set-up. All ftigue tests were performed under the lod control mode of the testing mchine to ssure tht constnt mximum nd minimum lternting lod will be pplied to the specimen. The testing mchine ws set to stop cyclic loding when the grip-displcement limit ws reched nd ftigue filure ws thus ssumed to hve occurred. This limit ws determined by dding.8 mm to the totl deflection mgnitude of the specimen during ftigue loding. The deflection ws monitored using n IBM PC. The bolt temperture ws mesured to mke sure heting did not occur in specimen. The mximum temperture limit ws set to +33 C. If heting occurred the frequency of the test ws reduced. All experiments were performed t room temperture ( 22 C) nd mbient reltive humidity ( 5 ). EXPERIMENTAL RESULTS AND DISCUSSIONS Ftigue resistnce of the joints The ftigue test results re presented in Fig. 5 together with the ultimte tensile strength nd strin for ech joint configurtion tken from [7]. The lines in the plots re from lest squres fit to the logrithmic ftigue dt., (MP) σ 4 35 3 25 2 15 1 5 two bolts four bolts six bolts 1 1 1 1 1 4 1 5 1 1 7 Number of cycles
, (%) ε.8.7..5.4.3.2.1 two bolts four bolts six bolts 1 1 1 1 1 4 1 5 1 1 7 Number of cycles Fig. 5: Ftigue test results. As seen, the joints with six bolts performed the highest resistnce to ftigue, followed by the joints with four bolts. The lowest ftigue resistnce ws shown by the joint with two bolts. In most of the ftigue tests, filure of the specimens occurred becuse of ftigue filure of fsteners. The specimens with two bolts filed fter one bolt broke (see Fig. ); wheres the joints with four nd six bolts filed fter t lest two fsteners frctured (see Fig. nd (c)). (c) (d) Fig. : Ftigue tested joints. Specimens with six bolts, tested t = 275 MP nd higher lod levels, filed either due to net-section filure occurred or due to ftigue filure of bolts followed by netsection filure (see Fig. (d)). It is believed tht fter the frcture of bolts the net-section trnsferred more lod. This eventully resulted in net-section secondry filure mode.
δ σ σ σ σ Reltive displcement between pltes Reltive displcement between two composite pltes ws mesured fter different number of cycles. The mesured results re presented s extensometer-displcement rnge,, which is the pek to pek reltive displcement between the pltes during the compression nd tension prts of lod cycle s shown in Fig. 7. The obtined results for two specimens with six bolts, tested t = 27 MP nd 35 MP, re shown in Fig. 7. Extensometer displcement, (mm).4.3.2.1 EXT -.1 -.2 -.3 1 2 3 4 5 Time, (s) EXT, (mm) δ 1.9 EXT1 EXT1 = 27 MP = 35 MP EXT2 EXT2 = 27 MP = 35 MP.8 2.7. 1.5 b1.4 2 4 8 1 1 4 1.2 1 4 1.4 1 4 Number of cycles Fig. 7: Extensometer displcement: nomenclture; the mesured results. The extensometer curves show tht chnges in reltive displcement between the pltes occur quite erly. At the beginning of ftigue loding the extensometers mesured decrese in the reltive displcement between the composite pltes which might be due to n increse in the coefficient of friction between the composite pltes [8]. This induced n increse in the friction force between the pltes, thereby cusing less displcement between the pltes. In ddition, ftigue debris ccumulted in the bolt holes could fill the clernce between the fsteners nd the hole surfce. This would reduce the spce for the bolt to move, nd s result, the reltive displcement between the pltes would decrese s well. As the number of cycles increses, the curves show slow increse in the reltive displcement between the composite pltes. This is cused by the growth of ftigue dmge in the bolt holes resulting in their elongtion. Removl of ftigue debris from the bolt holes would crete more vilble spce for the bolts to move, thus prtilly resulting in the slow increse in the reltive displcement between the composite pltes. At the sme time, the coefficient of friction could decrese due to wer of the pltes [8] nd the friction force decreses due to the reduction in the clmping force in the bolts. This would increse the reltive displcement between the composite pltes s well. The joints bolted by six fsteners filed fter t lest two bolts broke. In Fig. 7 point (1) nd (b1) reflect the first filure of bolts. As seen, this resulted in n incresed reltive displcement mesured by tht extensometer, which ws ttched on the sme hlf of the specimens were the bolt filure occurred. However, in one cse the second extensometer mesured drop in the reltive displcement between the pltes, wheres for nother joint the EXT2 curve does not show ny effect from the bolt filure on the extensometer mesurements. Point (2) corresponds to the second filure of bolt, which chnged the following extensometer mesurements. After this, the reltive displcement for the first
extensometer incresed, wheres the second extensometer mesured smll decrese in the reltive displcement. As cn be seen, even fter the second bolt frctured the specimen ws still ble to support ftigue loding until ftigue filure of bolt No. 5. Bolt-movement mesurements Since the joints mostly suffered from ftigue filure of bolts, it ws of interest to nlyse the ftigue behviour of the fsteners under cyclic loding. Bolt-movement mesurements were done on bolt No. 1 nd No. 3 from specimens with four bolts (see Fig. 2). One specimen with two bolts ws ftigue tested with bolt-movement mesurement on bolt No. 1. The obtined mesurement results re presented together with the gripdisplcement curves in Fig. 8 for specimens with two nd four bolts, which were tested t = 155 MP nd 127 MP, respectively. Rnge of grip displcement grip displcement.24 2.4.22 2.2.2.18 2.1 1.98.14 bolt No. 1 1.9.12.1 5 1 15 Number of cycles Rnge of bolt movement, (mm) Rnge of grip displcement 2 1.8 1. 1.4 1.2 grip displcement bolt No.1 bolt No. 3 2 1 5 4 1 5 1 5 8 1 5 1 1 Number of cycles 1.8.7..5.4.3.2.1 Rnge of bolt movement, (mm) Fig. 8: The mesured results of bolt movement: specimen with two bolts; specimen with four bolts. The grip displcement nd bolt movement re chrcterized by three different stges of evolution. At the beginning of the ftigue tests, ll curves show decrese in the
grip displcement nd bolt movement. The initil decrese hs the sme cuse s in the extensometer curves bove. The further evolution of the grip displcement nd the bolt movement is believed to be ffected by reduction in pre-stress in the bolts nd ftigue degrdtion of the bolt holes nd the composite round them. The reduction in the pre-stress in the bolts results in lower friction force between the composite pltes thn it ws t the beginning of cyclic loding. Therefore, less lod is trnsferred between the pltes by the friction force, which increses the displcement between the pltes nd more lod being trnsferred by the bolts. In ddition, t this stge of the ftigue tests very often ccumulted ftigue debris were observed trnsporting out of the bolt holes by visul inspection of the specimens. Ftigue prticles re cused by wer nd re-deposition of composite mteril round the bolt holes (see Fig. 9) nd bolt movement under loding []. Fig. 9: The middle plte of specimen with four bolts tested t = 1 MP. Becuse of the removl of ftigue debris, the clernce between the fsteners nd the bolt holes incresed, thus inducing more movement of the bolts during ftigue loding of the specimens. Filure of the joints with two bolts occurred fter ftigue filure of one bolt; wheres the joints with four bolts filed fter filure of two bolts. In Fig. 8 point (1) is mrked. The point corresponds to ftigue filure of bolt No. 2 (see Fig. 2) fter 94445 cycles. This induced more movement on bolt No. 1, but the bolt No. 3 curve shows drop in the bolt movement fter this number of cycles. Further evolution of the grip displcement nd the bolt movement show more grdul growth in the curves rising from dditionl dmge development in the specimen, except for bolt No. 3. Strin guge loops It ws found tht the influence of the bolt presence on the strin guges is more pronounced for those strin guges, which were locted between two bolts (see Fig. 2). Therefore, it my be interesting to nlyse the ftigue behviour of these strin guges s function of number of cycles. Figure 1 shows strin guge loops for specimens with four nd six bolts. The different loops hve been shifted 2 strin to fcilitte observtion. The loop rrows in the plots indicte their direction.
1 15 Lod, (kn) 5-5 -1 c t -2 2 4 8 1 Strin*1, (ustr) c t c t 1st cycle 1 cycles 25 cycles 34 cycles 42 cyces Lod, kn 1 5-5 -1-15 t t t t c c c c -2 2 4 8 1 Strin*1, (ustr) 2nd cycle 1 cycles 51 cycles 75 cycles 1 cycles 15 1 Lod, kn 5-5 -1 2nd cycle 1 cycles 3125 cycles 475 cycles -15-2 2 4 8 Strin*1, (ustr) (c) Fig. 1: Strin guge loops fter different number of cycles As the plots in Fig. 1 nd show, ll strin guge loops contin some points of inflection, which re mrked s ( ) nd ( ) for the tensile nd compression prts of the lod cycle, respectively. These points could be result of bolt-movement influence during cyclic loding. Due to lod trnsfer nd bolt bending, the bolts crete contct forces in the re between them, i.e., the re where strin guge No. 3 ws locted. The contct forces cused compressive deformtion in the region between the bolts. Therefore, the strin guge loops show n insignificnt increse in strin with further loding. Moreover, the lod level t which these points pper, decrese with incresing number of cycles. For exmple in Fig. 1, point ( ) in the first loop ppers t 4 kn, wheres fter 42 cycles the sme point cn be defined t 2 kn. It cn be expected tht t the beginning of the ftigue test some lod is trnsferred by the friction force between the specimen pltes. This will soften the inflection points. As ftigue loding proceeded, relxtion in the fsteners occurred, decresing the friction force nd thus inducing more lod being trnsferred by the bolts. As result, the bolt influence on the strin loop would occur t lower lod level within one lod cycle thn it ws t the beginning of the ftigue test. At this stge of the ftigue test the inflection points become shrper thn t its beginning. Figure 1(c) shows strin guge No. 8 loops for the specimen with six bolts. Anlysis of the mesurements found tht strin guge No. 8 behved similr to strin guge No. 3, especilly fter lrge number of cycles. Despite tht the hysteresis is not s pronounced s for strin guge No. 3, its shpe looks similr. However, it should be noted tht the mesurement on strin guge No. 8 were done only until 475 cycles,
when during visul inspection of the specimen two crcks in the surfce ply of the composite plte were found propgting under the strin guge, thereby cusing erroneous redings. CONCLUSIONS Specimens with two, four, nd six titnium bolts were tested in ftigue. The dominnt filure mode ws ftigue filure of bolts. Since for the joints with more thn two bolts filure occurred fter t lest two bolts broke, this would not be considered indictive of ctstrophic filure. Reltive displcement between composite pltes nd bolt movement under cyclic loding re gretly ffected by reduction in the pre-stress in the bolts, ftigue degrdtion of the composite mteril round the bolt holes, nd ftigue elongtion of the bolt holes. Strin guge mesurements between bolts showed tht the bolt presence hs significnt effect on the strin evolution during ftigue tests. References [1] Rpoff AJ, Dill HD. Certifiction of dmge tolernt composite structures. In: Proceedings of the Eighth DOD/NASA/FAA Conference on Fibrous Composites in Structurl Design (Prt 2), Lngley Reserch Center, NASA, September 199. p. 499 514. [2] Rubin AM. Common filure modes for composite ircrft structures due to secondry lods. Composites Engineering 1992;2(5-7):313 2. [3] Ashton HR. Dmge tolernce nd durbility testing for F/A-18 E/F composite mterils structures. In: Proceedings of the 37th AIAA/ASME/ASCE/AHS/ASC Structures, Structurl Dynmics, nd Mterils Conference, Slt Lke City, UT, April 15-17 199. p. 1 13. [4] Cooper C, Turvey GJ. Effects of joint geometry nd bolt torque on the structurl performnce of single bolt tension joints in pultruded GRP sheet mterils. Composite Structures 1995;32:217 22. [5] Rmkumr RL, Tossvinen EW. Strength nd lifetime of bolted lmintes. Ftigue in Mechniclly Fstened Composite nd Metllic Joints, ASTM STP 927, Americn Society for Testing nd Mterils, Phildelphi 198. p. 251 273. [] Sunders DS, Gle S, Deirmendsin G. The development of ftigue dmge round fstener holes in thick grphite/epoxy composite lmintes. Composites 1993;24(4):39 32. [7] Strikov R, Schön J. Qusi-sttic behviour of composite joints with protrudinghed bolts. Composite Structures 21;51(4):411 425. [8] Schön J. Coefficient of friction of composite delmintion surfces. Wer 2;237:77 89. [9] Sunders DS, Gle SC, Deirmendjin GK. The development of ftigue dmge round fstener holes in thick grphite/epoxy composite lmintes. Composites 1993;24(4):39 321.