Fiber / Toughened Epoxy Composites under

Fatigue Behavior o Impact-Damaged Carbon Fiber / Toughened Epoxy Composites under Compressive loading Toshio Ogasawara, Hisaya Katoh, unao ugimoto, and Takashi Ishikawa Advanced d Composite Technology Center (ACE-TeC) C) Japan Aerospace Exploration Agency (JAXA) Tokyo Japan CDW-15 (15th Composites Durability Workshop) October 17-19, 21, Kanazawa, Japan

Outline Introduction Experimental procedure Materials Compression ater impact (CAI) and atigue testing Results and discussion tatic strength Fatigue lie (-N curves) Rough estimation o atigue lietime distribution ib ti o an impact-damaged specimen CAI atigue lietime prediction Conclusion

Background Fatigue behavior o impact-damaged carbon iber composite (CFRP) laminates has been studied since the late 197s. The atigue data o early CFRP materials were published in ATM-TP, NAA reports, and other reports o the literature.

Fatigue behaviors o impact-damaged carbon iber composites (CMH-17-3G, Vol. 3, Chapter 12.7) p The endurance limit at 1 6 cycles turned out as between 5 and 75% o the initial strength. Damage growth starts very close to the end o the specimen lietime -N curves or the IM7/977-2 and the T8H/F-655-2 material Damage growth as measured using C-CAN, versus atigue cycles or the T8H/ F655-2 material The ratio between the endurance limit at 1 6 cycles, and the initial static strength turned out as between.5 and.75. Thereore, sizing a structure (with these materials) using Ultimate Loads is expected to push atigue loads down to a level that is likely to limit atigue problems with low energy impact damage. Unrealistic atigue stress (greater than 75% o the static strength) was necessary to produce such measurements. This igure shows that, despite the log axis, damage growth starts very close to the end o the specimen lietime (between 85% and 95% or all cases investigated in this program), with a very high slope.

Background Damage growth under cyclic compressive loading is extremely sensitive to stress, and it starts very close to the end o the specimen lietime. A conventional damage tolerance methodology or aluminum alloys is not applicable or CFRP. No damage growth (allowable stress design) is becoming the most common concept or designing composite structures. For this methodology, atigue strength o composite laminates which have barely visible impact damage (BVID) is important to determine the allowable stress. BVID: The minimum impact damage surely detectable by scheduled inspection

Objectives Experimental results that clariy the atigue lietime distribution o impact-damaged CFRP laminates have not been published because they are sensitive or material suppliers as well as aircrat companies. tatistical methods or predicting CAI atigue lietime have not been understood suiciently. The objective o this study is to obtain the atigue lietime data o impact-damaged d carbon iber composites under compressive loading to elucidate the lietime prediction methodology based on statistical approaches.

Experimental procedures Materials Carbon iber / toughened epoxy composites Toray T8/39-2B Quasi-isotropic laminates [45//-45/9]4s (32 ply) (Elastic modulus 49.2 GPa) Low-speed Impact loading (in accordance with ATM D7136) Drop weight type impact tester pecimen geometry 1 mm width, 15 mm length 6 mm thickness Drop weight Weight 5.5 kg, Radius 5/8 inch Impact energy 6.7 J/mm Low speed Impact

Experimental procedure Compression ater impact (CAI) testing (ATM D7137) Test ixture ATM D7137 Electromechanical testing system (25 kn, 5885; Instron Corp.) Number o samples: 9 CAI atigue testing ervo-hydraulic testing system (5 kn, 884; Instron Corp.) Test ixture ATM D7137 ide: simple support Top & bottom: ixed support tress ratio R = σ mim / σ max=1 Frequency = 3 Hz inusoidal wave Run-out 1 7 Number o samples > 36

CAI strength distribution (Weibull plot) -F)) ln (-ln (1 1-1 -2-3 CAI strength (ATM D7137) CAI (m=17.5) Average 272MPa (.55%) CV 5.4% m=17.5 NHC (m=53.2) NHC strength (ATM D6144) Average 583 MPa (1.18%) 18%) CV 1.8 % m=53.2 2 3 4 5 67 trength (MPa) F ( ) = 1 exp( ( / ) m ) The Weibull shape parameter (m) or CAI strength is much lower than that or NHC strengths. th This result seems to derive rom the variation in impact damage size.

-N curves (peak compressive stress vs. atigue lie) Pea ak stress s (MPa) 3 2 1 Impact 6.7 J/m R = 1 = 3 Hz CAI strength (average 272 MPa) 73% o CAI strength (2 MPa) 1 1 1 1 2 1 3 1 4 1 5 1 6 1 7 Number o cycles to ailure pecimen ailed at 123 1.23 1 6 ( σ min = 22 MPa ) cycles CAI atigue lietime exhibits relatively lat -N curves and considerable scattering. Fatigue ailures occurred at the 73% o CAI strength, which agrees with the data ound in some literatures.

Fatigue lie distribution (Weibull plot) ln (-ln (1-F F)) 1-1 -2 23 MPa M =.6 22 MPa tress (MPa) 21 MPa Weibull parameters M N Number o samples 2.52 2.1 1 6 6 21.6 6.6 1 5 8 22.6 1.6 1 5 9 23 45.45 28 1 2.8 4 6-3 Impact t67j/ 6.7 J/m 1 2 1 3 1 4 1 5 1 6 1 7 F ( N ) = 1 exp( ( N / N ) M ) Number o cycles to ailure Weibull plots showing atigue lietime distribution or 21, 22, and 23 MPa are shown here. The Weibull shape parameters (M) are apparently independent o the stress.

Evaluation o damage growth using ultrasonic C-scan CAI-B-4 (σ min = 22 MPa, R=1, = 3 Hz, atigue lie 1.23 1 6 ) 34. 34. 34. cycle 1 5 cycles 1 6 cycles Damage growth is not clearly visible until 1 6 cycles, even though the lietime was 1.23 1 6. However, the pulse-echo intensity rom damage increases with the number o cycles.

Observations o damage growth using digital video recording CAI-A-4 (σ min = 2 MPa, R=1, = 3 Hz, atigue lie 1.67 1 6 ) 15 cycles (no growth) 1 cycles 3 cycles Final ailure (1.67 1 6 Cycles) Damage growth to transverse direction was observed very close to the end o the specimen lietime (1-15 cycles beore the inal ailure (1.67 1 6 )). Until then, no visible damage growth was observed.

Rough estimation o atigue lietime distribution o an impact-damaged specimen (1) We made two assumptions The mode-ratio (G II /G I, G III /G I ) at the critical point o a damage is independent o damage growth during a atigue test. tress redistribution attributable to damage growth is ignored. Based on dimensional analysis, the energy release rate G is roughly approximated by the ollowing relation. G 2 : Compressive stress (=P/A) a (1) a : Representative damage size Then strength o the impact-damaged specimen,, is given as G / a (2) c G c is a mixed mode racture toughness (see appendix). a

Rough estimation o atigue lietime distribution o an impact-damaged specimen (2) Paris law is applied or damage growth under cyclic compressive loading. da / dn Δ ) Δ = ( 1 1/ R ) n / 2 n ( Δ G ) = ( Δ a (3) Integrating Eq. (3) rom the initial damage size a to the inal size a, the atigue lie N under a constant stress amplitude Δ is given as n 2 N = Q /( Δ) n (4) For these calculations, l a -n <<a -n is assumed because n > 1. Q is a atigue parameter which depends on the material properties, damage geometries, damage size, specimen geometries, loading constraint, etc. CAI strength ( ) distribution is given as, F ( ) = 1 exp( ( / ) m ) (5) ubstituting Eq.(4) into Eq. (5) yields the atigue lie distribution unction. F ( N M M ) = 1 exp( ( N / N ) ) (6) = m /( n 2) n 2 n N = ( Q ) /( Δ)

Rough estimation o atigue lietime distribution o an impact-damaged specimen (3) This equation shows that two methods can be used to determine the damage growth exponent (n). A) Weibull shape parameters m ( CAI strength) and M (CAI atigue) F B) Weibull scale parameters (N ) and compressive stress () ( N ) = 1 exp( ( N / N ) M = m /( n 2) n 2 n N = ( Q ) /( Δ) Method A M m = 17.5 M =.6 (=21, 22 MPa) M ) = m /(nn 2) n=31 (6) Method B Pe ak stress (M MPa) n 2 n N = ( α ) /( Δ) Δ = ( 1 1/ R) 3 n=31 2 1 1 3 1 4 1 5 1 6 1 7 Weibull scale parameter, N

Typical exponent value in delamination growth rate reported in literatures Detailed investigations using DCB and ENF methods in literatures revealed delamination growth behaviors o carbon iber composites. Exponent values in delamination growth rate are reportedly between 2 and 5. n = 2 n = 32 M. Hojo et al., Comp. ci. Technol., 29 (1987), 273-92 J. chön et al., Comp. ci. Technol., 6 (2), 173-184

CAI atigue lietime prediction Pea ak stress s (MPa) Mean 3 F=.9 2 1 F=.1 B-basis value Impact 6.7 J/m R = 1 = 3Hz 1 1 1 1 2 1 3 1 4 1 5 1 6 1 7 tatic strength 272 MPa Fatigue strength Lietime F=1% B-value* 1X 1 6 1X 1 7 183 MPa 167 MPa (67 %) (61 %) 17 MPa (63 %) 155 MPa (57 %) * Number o samples 9 Number o cycles to ailure Fatigue lietime is predicted or ailure probability (F) o 1 %, 5 %, and 9%. Numerical results agree with the experimental data. The ratio between the endurance limit (B-value) at 1 6 cycles and the initial static strength was estimated as 61%. B-Basis value At least 9% o the population o material values is expected to equal or exceed this tolerance bound with 95% conidence.

Conclusion Fatigue lietime data o impact-damaged carbon iber / epoxy composites (Impact energy 6.7 J/mm) were obtained. The CAI strength and atigue lietime exhibit considerable scattering. A simple statistical model was proposed p or predicting the lietime o impact-damaged CFRP laminates. In spite o rough approximation, the estimates agreed with the experimental results. The ratio between the endurance limit (B-value) at 1 6 cycles and the initial static strength was estimated as 61%. In uture the eect o stress ratio on CAI atigue behaviors In uture, the eect o stress ratio on CAI atigue behaviors will be investigated.

Appendix Mixed mode racture toughness Gc A mixed mode ailure criterion is assumed as ollows. GI GII GIII + + 1 (1) G G G IC IIC IIIC The energy release rate o each mode (i = I, II, II) is given as a G i = κ (2) 2 i a κ i : constant (i =, I, II, III) Failure stress is assumed to be given as the ollowing equation. 2 κ a G c (3) ubstituting Eq.(2) into Eq. (1), a mixed mode racture toughness Gc is obtained as G c 1 κ I κ II κ III = κ + + (4) G IC G IIC G IIIC