An experimental study on fatigue crack growth in lap joints with multiple fastener holes

Transcription

1 An experimental study on fatigue crack growth in lap joints with multiple fastener holes T. Sasaki &T. Honda National nstitute of ndustrial Safety, Japan. Abstract Although a lot of studies have been conducted on the multi-site damage (MSD) problems, most of them focus on the strength of plates with MSD and the phenomenon of MSD seems to be not fully clarified yet, n this study, fatigue crack growth tests were conducted using bolted lap joint specimens with multiple fastener holes in order to investigate the phenomenon of MSD. Crack growth data were generated using notched and unnotched specimens at different clamping force levels. The results of fatigue tests revealed that the existence of initial notches at fastener holes was an important factor for the occurrence of MSD in this type of specimen. Furthermore, the stress distribution of the lap joint specimen was analyzed by an infrared stress imaging system to understand the nature of fatigue crack initiation and fatigue crack growth. Good agreement was obtained between the crack initiation sites and the highest stress regions in the unnotched specimen, 1 ntroduction Since the 1985 Japan Airlines Boeing 747 crash and the 1988 Aloha Airlines Boeing 737 incident, the phenomenon of multi-site damage (MSD) in fuselage lap joints has received a great deal of attention [1-8]. MSD is d@ined as growth of multiple cracks emanating from fastener holes. After some period of slow crack growth, MSD cracks may suddenly link up to form a,large crack that may cause catastrophic failure. n this study, to prevent failure of lap joints with multiple fastener holes caused by MSD, fatigue crack growth in lap joints was investigated through fatigue tests using lap joint specimens fastened with two rows of bolts. Not rivets but bolts were used as fasteners to control the clamping force. Two types of specimen, with and without initial notches at fastener holes,

2 226 Damage and Fracture Mechanics V were used to examine the effect of initial notches on MSD. Furthermore, the stress distribution of the specimen was measured using an infrared stress imaging system to understand the nature of fatigue crack initiation and fatigue crack growth in the lap joint specimen. 2 Fatigue test set-up 2.1 Specimen description Figure 1 shows the basic specimen configuration used in fatigue tests. t consisted of two 2024-T3 aluminum alloy plates 3 mm thick and one SUS304 stainless steel plate 6 mm thick, fastened with eight M8 hexagon socket head cap screws arranged in two rows. To avoid out-of-plane bending and to get the higher frictional force, the double shear type was adopted in this study. One drawback of this type of specimen is that cracks which initiate in the inner plate 6 mm thick cannot be visually observed. To conquer this problem, SUS304, which has higher fatigue strength than 2024-T3, was used for the inner plate. Table 1 and Table 2 show mechanical properties and chemical composition of 2024-T3 used in this study, respectively. To control the clamping force, a strain gage, which can measure the axial force of a bolt, was embedded in each hexagon socket head cap screw. All screws were calibrated before fatigue tests. Two types of specimen, with and without initial notches at fastener holes, were used. The initial notches were machined by means of electro-discharge machining into the upper row of fastener holes in the top plate that is in contact * 15, W.: _. _ _:. 3 : 300 Ln - s +~ N ++.=.._. -._. _ _. +._. _._._: m N ~ ;;1 (540) 1 w 3 Lower side bolts Uppser side bolts M8x4 M8x4 w nner plate Figure 1: Fatigue test specimen configuration.

3 Table 1: Mechanical properties of 2024-T3. Damage and Fracture Mechanics V 227 Tensile strength Yield strength Elongation MPa MPa % Table 2: Chemical composition of 2024-T3 (wt. %). Cu Mg Mn Cr Si Fe Ti Zn Zr with the bolt heads. The length of initial notch is nominally 2.5 mm on each side of the fastener hole. This length was chosen so that crack initiation might not be obscured by the fastener head. 2.2 Testing condition All specimens were loaded in tension on a closed-loop, servo-hydraulic material test system. The applied constant amplitude loading was P~,X = 53.3 kn and F ~in = 5.3 kn, which gave a remote plate stress range, Aa of 80 MPa and a stress ratio, R of 0.1. A frequency of 10 Hz and a sinusoidal waveform were selected for the loading cycle. All tests were carried out in air at room temperature. Crack growth was monitored optically using a traveling microscope, and the crack length and the number of load cycles were continuously recorded. Three different bolt axial force levels, F = 8 kn, 12 kn, 16 kn, were used for both notched and unnotched specimens. All bolts were tightened at the same axial force level in each specimen. 3 Fatigue test results To date six specimens have been tested, and these results are summarized in Table 3. The first part of specimen number in Table 3 represents the bolt axial force; the second part shows whether it is notched or unnotched: WN means with notches and WON means without notches ; and the last part represents the serial number for each test case, The crack growth nature was totally different between the notched specimens and the unnotched ones. The total fatigue lives of the notched specimens were almost occupied by the crack propagation lives while those of the unnotched specimens were dominated by the crack initiation lives. And less number of crack linkups was observed in the unnotched specimens than in the notctied specimens. n the notched specimens, the multiple cracks initiated from most notches and grew simultaneously. As a result, MSD did occur in the notched specimens.

4 228 Damage and Fracture Mechanics V Table 3: Fatigue life and number of crack linkups for each specimen. Total cycles to Crack propagation Number of failure cycles to failure crack inkups Specimen N~ ND N4N~ Nlinku~ F8k-WN , , F12k-WN-l 1,017, , F16k-WN- 1 1,073, , * F8k-WON- 1 1,746, , F12k-WON-l 1,291, , F16k-WON-l 1,496, , * During crack growth from initial notches, other fatigue cracks initiated from the same places as the crack initiation sites of the unnotched specimens, When loading was continued, the adjacent MSD cracks linked up one after another with forming a large crack, and finally the specimen failed. On the other hand, in the unnotched specimens, only a few cracks initiated from near the top of the upper side holes after longer period of loading comparing with the crack initiation lives of the notched specimens. They grew so rapidly that the specimens resulted in failure before other cracks initiated. As a result, MSD did not always occur in the specimens without initial notches. t should be noted that the crack propagation life of the unnotched specimen was shorter than that of the notched specimen at the same clamping force level although the unnotched specimen had less number of crack linkups than the notched specimen. Figures 2 and 3 present typical crack growth curves for the notched and the unnotched specimen, respectively. n these figures, U 1-U4 represent the upper side bolts (see Figures 6 and 7), and open symbols represent internal cracks (with two tips each) while solid symbols represent single edge cracks. Figures 4 and 5 present views of failure surfaces of the notched and the unnotched specimen, respectively. 4 Stress distribution analysis of the lap joint specimen 4.1 Procedure The stress distribution of the lap joint specimen without initial notches was analyzed using an infrared stress imaging system [9]. To me!isure the stress, the system employs thermoplastic effect, basically obtained by Lord Kelvin, which is formulated as: AT= ~ TAo = KmTAcr, K, where AT is the change of temperature, cz is the coefficient of linear thermal expansion, T is the absolute temperature, Acr is the change of the principal stress (1)

5 Damage and Fracture Mechanics V ,, ( El Ooul A U1+UZ+U3 U2 Linking Up 80 v UZ+U3 o U3 i A U1+U2+U3+U4 A U4 Linking up Linking up na@w QJ!? / ~ ~ AAAA nl 1! 1! 1 1 o Load cycles Figure 2: Specimen F8k-WN-l crack growth data (F = 8 kn all, with initial notches). 100 i m 80 1 E E i g 60 - g) x c) 40 Thecrack reached the U2bolt hole. ~o The crack reached the U4 bolt hole. ~; o ! Load cycles Figure 3: Specimen 12k-WON- 1 crack growth data (F = 12 kn all, w;thout initial notches).

6 230 Damage and Fracture Mechanics V Figure 4: Failure surface of specimen F8k-WN- 1 (F = 8 kn all, with notches). initial Figure 5: Failure surface of specimen F12k-WON-l (F = i2 kn all, without initial notches).

7 Damage and Fracture Mechanics V 231 Figure 6: Stress distribution (F= 12 kn all, without initial notches). Figure 7: Stress distribution ( F = OkN for the center bolt in the upper row while F=12 kn for all other bolts, without initial notches).

8 232 Damage and Fracture Mechanics V sum, pis the density, C,, is the specific heat at constant pressure, and Km is the thermoelasticity coefficient. The system used was JTG-8000 manufactured by JEOL nc. The sensitivity of this system is K that equates to 1 MPa in steel. Another lap joint specimen similar to Figure 1 was used for the stress distribution analysis. This specimen consisted of two 2024-T3 plates 5 mm thick and one 2024-T3 plate 10 mm thick fastened with six M8 hexagon socket head cap screws arranged in two rows. During infrared measurements, constant amplitude load was applied to the specimen so that a remote plate stress range, AO was 20 MPa and a stress ratio, R was Results Figure 6 shows the stress distribution of the lap joint specimen without initial notches in which the bolt axial force F = 12 kn for all bolts. n this case, the highest stresses were observed in the regions of just above the bolt heads. These regions coincided with the places where cracks initiated in the fatigue tests using the unnotched specimens. Moreover, the shorter fatigue propagation lives of the unnotched specimens can be, explained by the fact that the stresses around the crack paths of the unnotched specimen were higher than those of the notched specimen. Next, mean stresses on three segments y y in Figure 6 were respectively calculated. The result showed that three bolts in the load edge side (upper side) shared the applied load of about 55% while other three bolts in the free end side shared the applied load of about 45%. Terada et al. [10] reported the same phenomenon using the riveted single shear lap joint specimens. On the other hand, Figure 7 shows the stress distribution of the same specimen in which the bolt axial force F = O kn for the center bolt in the upper row while F = 12 kn for all other bolts. n this case, conspicuous stress concentration appeared not only in both sides of the loosened bolt but also in the regions of just above the bolt heads adjacent to the loosened bolt. 5 Conclusions Fatigue crack growth tests under constant amplitude loading were carried out using notched and unnotched lap joint specimens with multiple fastener holes in two rows at three clamping force levels. The crack growth nature was totally different between the notched and the unnotched specimens:, n the notched specimens, multi-site damage always occurred and adjacent MSD cracks linked up frequently while MSD didn t always occur in the uqnotched specimens. Nevertheless, crack propagation life of the notched specimen was shorter than that of the unnotched specimen at the same clamping force level, The higher clamping force seems to lengthen fatigue life. However, further studiqs are required to clarify the effect of clamping force because only limited numbers of tests were conducted in this study. The stress distribution of the specimen

9 Damage and Fracture Mechanics V 233 obtained by the infrared stress imaging system well explained the crack initiation sites in the unnotched specimen and the shorter crack propagation life of the unnotched specimen comparing with that of the notched specimen at the same clamping force level. References [1] Kobayashi, H., Arai, Y. & Nakamura, H,, Fatigue failure analysis of the rear pressure bulkhead of the crashed Boeing 747 Jumbo jet plane. Journal of the Society of Materials Science, Japan, 36(409), pp , (n Japanese) [2 Dawicke, D.S. & Newman, J. C., Jr., Analysis and prediction of multiple-site damage (MSD) fatigue crack growth. NASA-TP-3231, pp. 1-15, [3 Molent, L. & Jones, R., Crack growth and repair of multi-site damage of fuselage lap joints. Engineering Fracture Mechanics, 44(4), pp , [4] Nathan, A. & Brot, A., An analytical approach to multi-site damage. Proc. of The 17 1 Symp. of the nt. Committ. on Aeronautical Fatigue, Stockholm, pp , [5] Singh, R. Park, J-H. & Atluri, S. N., Growth of multiple cracks and their inkup in a fuselage lap joints. AAA Journal, 32(11), pp , [6] Moukawsher, E.J., Grandt, A.F., Jr. & Neussl, M. A., Fatigue life of panels with multiple site damage. Journal of Aircraft, 33(5), pp , [71.. Moukawsher, E.J., Heinimann, M.B. & Grandt, A.F., Jr., Residual strength of panels with multiple site damage. Journal of Aircraft, 33(5), pp , [8: Cherry, M. C., Mall, S., Heinimann, M.B. & Grandt, A.F., Jr., Residual strength of unstiffened aluminum panels with multiple site damage. Engineering Fracture Mechanics, 57(6), pp , [9: Kitsunai, Y., Honda, T. & Sasaki, T., The determination of stress concentration factors and stress intensity factors by means of themoelastic effect. Proc. of 4 1 [nt. Con. on Computer-Aided Assessment and Control, eds. H. Nisitani, M.H. Aliabadi, S-. Nishida & D.J. Cartwright, Computational Mechanics Publications: Southampton, pp , [10] Terada, H. & Okada, T., Stress analysis of riveted single lap joint by using thermography. Proc. of The l rsymp. on Thermographic NDT&E Technique, The Japanese Society for Non-Destructive nspection: Tokyo, pp , (n Japanese)