1、NASA TECHNICAL NOTE NASA TN D- e. I pe - 3879 INVESTIGATION OF FATIGUE CRACK GROWTH IN Ti-sA1-1Mo-lV (DUPLEX-ANNEALED) SPECIMENS HAVING VARIOUS WIDTHS by C. Michael Hudson Langley Research Center Langley Station, Hampton, Va. NATIONAL AERONAUTICS AND SPACE ADMINISTRATION WASHINGTON, D. C. MARCH 1967
2、 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-TECH LIBRARY KAFB, “I 0130b07 NASA TN D-3879 INVESTIGATION OF FATIGUE CRACK GROWTH IN Ti-8A1-1Mo-1V (DUPLEX-ANNEALED) SPECIMENS HAVING VARIOUS WIDTHS By C. Michael Hudson Langley Research Center Langle
3、y Station, Hampton, Va. NATIONAL AERONAUTICS AND SPACE ADMINISTRATION For sale by the Clearinghouse for Federal Scientific and Technical Information Springfield, Virginia 22151 - CFSTI price $3.00 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-INVES
4、TIGATION OF FATIGUE CRACK GROWTH IN Ti-8Al-1Mo-1 V (DUPLEX-ANNEALED) SPECIMENS HAVING VARIOUS WIDTHS By C. Michael Hudson Langley Research Center SUMMARY Axial-load fatigue-crack-propagation tests were conducted on 2-, 4-, 8-, and 20-inch-wide (5.1-, 10.2-, 20.3- , and 50.8-cm) sheet specimens made
5、of Ti-8A1-1Mo-1V (duplex-annealed) titanium alloy to study the effect of specimen width on fatigue crack growth. Both longitudinal- and transverse-grain specimens having a width of 8 inches (20.3 cm) were studied. and at five alternating stresses with amplitudes ranging from 2 to 25 ksi (14 to 173 M
6、N/m2). fatigue-crack-growth curves for different specimen widths. Schijve et al. (Rept. NLR-TR M.2142) found a similar result in tests on clad 2024-T3 aluminum-alloy specimens. The resulting data were successfully correlated by using both Paris and a slightly modified form of McEvily and Illgs crack
7、-propagation analyses (see book entitled “Fatigue - An Interdisciplinary Approach,“ Syracuse Univ. Press, 1964, and NACA TN 4394, respec- tively). A simple power function relating rate and stress-intensity factor was empiri- cally adjusted to fit the test data. Tests were conducted at a mean stress
8、of 25 ksi (173 MN/m2) At a given stress level, there was relatively little difference between the Wilhem (ASTM Preprint No. 38, 1966) found the transition of the specimen fracture surfaces from the tensile to the shear mode to be associated with the change in slope of the semilog plot of rate of cra
9、ck growth against stress-intensity factor; however, inspec- tion of the fracture surfaces of specimens tested in the present investigation showed no such correlation. Cracks grew slightly faster in the transverse grain direction than in the longitudinal direction. INTRODUCTION The fatigue process in
10、cludes three general phases: crack initiation, crack propaga- The crack propagation phase frequently occupies a Consequently, parameters which affect fatigue crack growth tion, and residual static strength. major portion of the process. Provided by IHSNot for ResaleNo reproduction or networking perm
11、itted without license from IHS-,-,-can have an important effect on total fatigue behavior. One parameter which may affect fatigue crack growth is the width of the part through which the cracks propagate. The present investigation was conducted to study the effects of this parameter. Axial-load fatig
12、ue-crack-growth tests were conducted on sheet Ti-8A1- 1Mo- 1V (duplex-annealed) titanium-alloy specimens. (This alloy was selected because of its potential for use in supersonic aircraft construction.) These specimens ranged in width from 2 to 20 inches (5.1 to 50.8 cm). Tests were conducted at one
13、mean stress and five alternating stresses for each specimen width. The data were correlated by using both McEvily and Illgs (ref. 1) and Paris (ref. 2) crack-growth analyses. In reference 3, Wilhem reported a correlation between the stress-intensity factor and the transition of the fracture surfaces
14、 from a tensile to a shear mode. The fracture surfaces of the specimens tested in this investigation were inspected to see whether a similar correlation existed. The effects of grain direction on fatigue crack growth were also investigated. SYMBOLS The units used for the physical quantities defined
15、in this paper are given both in the U.S. Customary Units and in the International System of Units (SI). Factors relating the two systems are given in reference 4 and those used in the present investigation are presented in appendix A. a one-half of the total length of a central symmetrical crack, in
16、. (cm) c,c* constant in crack propagation equation accounting for mean load, loading frequency, and environment E Youngs modulus of elasticity, ksi (GN/m2) e elongation in 2-inch (5.1-cm) gage length, percent Ak range of fluctuation of stress-intensity factor, ksi-in1/2 (MN/m3/2) Neuber technical st
17、ress - concentration factor K finite width factor (see appendix B) Kw 2 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-kmax kmin L N n Pa m max min R stress -intensity factor corresponding to maximum stress, ksi -in 1/ (MN/m3/ 2) stress-intensity fa
18、ctor corresponding to minimum stress, ksi-inll2 (MN/m3/2) denotes longitudinal -grain specimen number of cycles exponent in fatigue -crack -growth equation amplitude of alternating load, kips (N) mean load, kips (N) maximum load, Pm + Pa, kips (N) minimum load, Pm - Pa, kips (N) ratio of minimum str
19、ess to maximum stress amplitude of alternating net stress, Pa /(w - x)t, ksi (MN/m2) amplitude of instantaneous alternating net stress, Pa/(w - 2a)t, ksi (MN/ m 2, mean net stress, Pm,/(w - x)t, ksi (MN/m2) maximum gross stress, Pm,/wt, ksi (MN/m2) instantaneous maximum net stress, Pm,/(w - 2a)t, ks
20、i (MN/m2) minimum gross stress, Pmin /wt, ksi (MN/m2) denotes transverse-grain specimen t I specimen thickness, in. (mm) 3 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-W X (Y specimen width, in. (cm) length of crack-starter notch, in. (cm) tangent
21、 correction factor for finite width of panel effective radius of curvature at tip of a fatigue crack, in. (cm) ultimate tensile strength, ksi (MN/m2) yield strength (0.2-percent offset), ksi (MN/m2) SPECIMENS All specimens were made from Ti-8A1-1Mo-1V (duplex-annealed) titanium-alloy sheet 0.050 inc
22、h (1.27 mm) in thickness. All the material used in this investigation was obtained from the same mill heat. Tensile properties and the nominal chemical compo- sition of the material are listed in table I. Details of the duplex-annealed heat treatment are also presented in table I. The configurations
23、 of the crack propagation specimens are shown in figure 1. The width and length of the specimens are as follows: I Width I Length I The 2-, 4-, and 20-inch-wide (5.1-, 10.2-, and 50.8-cm) specimens were made with the longitudinal axis of the specimen parallel to the grain of the sheet. The 8-inch-wi
24、de (20.3-cm) specimens were made with the longitudinal axis of the specimens normal to the grain of the sheet. The width at the ends of the 20-inch-wide (50.8-cm) specimens was reduced to accommodate the 12-inch (31-cm) grips available on the testing machines. Doubler plates, 3/32 inch (2.38 mm) thi
25、ck, were bonded to the ends of these 20-inch (50.8-cm) specimens (by using an unfilled epoxy resin) to reduce the stress. The inte- rior ends of these doubler plates were scarfed to zero thickness in 2 inches (5.1 cm) to provide gradual load transfer from the doublers to the test specimens. A 0.1-in
26、ch-long (0.25-cm) central notch was cut into the center of each specimen to initiate the fatigue cracks. These notches were cut by using a spark discharge 4 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-Grip line T 3“ (76cd 24“ (61cm.I stress roiSe
27、r . Grip line f 4“ I Detail of SIES rdser Detail of stress raiw Grip line T 5“ (12.7cm) I Figure 1.- Specimen configurations. Specimen thickness, 0.050 inch (1.27 mm). 5 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-process. The heat generated by t
28、his process is localized and was not expected to affect the bulk of the material through which the fatigue cracks propagated. To mark intervals along the path of the crack, a reference grid (ref. 5) was photo- graphically printed on the surface of the specimens. Some specimens bearing the grid were
29、subjected to tensile tests and metallographic examination, which indicated that the grid had no detrimental effect on the material. TESTING MACHINES Two types of axial-load fatigue-testing machines were employed in this investiga- tion: a *20-kip-capacity (89-kN) subresonant machine which had an ope
30、rating frequency of 1800 cpm (30 Hz) and a combination hydraulic and subresonant machine. The combina- tion machine could apply alternating loads up to 66 kips (294 kN) hydraulically or 39 kips (173 kN) subresonantly. The operating frequencies were 40 to 60 cpm (0.7 to 1 Hz) for the hydraulic unit a
31、nd approximately 820 cpm (13.7 Hz) for the subresonant unit. Loads were monitored continuously by measuring the output of a strain-gage bridge cemented to a dynamometer in series with the specimen. The maximum error anticipated for this monitoring system was *1 percent. TEST PROCEDURE Axial-load fat
32、igue tests were conducted under a positive mean stress of 25 ksi (173 MN/m2). Alternating stresses ranged from 2 to 25 ksi (14 to 173 MN/m2). Both the mean and the alternating stresses were based on the original net area of the speci- mens. The mean load and the amplitude of the alternating loads we
33、re kept constant throughout the crack-propagation portion of each test. Fatigue cracks were sometimes initiated at an alternating stress level slightly higher than the level used in the crack growth portion of the test to expedite testing. The alternating stress was subsequently reduced to the desir
34、ed level after crack initiation. Fatigue crack growth was observed through a 10-power monocular telescope while illuminating the specimen with stroboscopic light. The number of cycles required to propagate the crack to each grid line was recorded so that the crack growth rates could be determined. T
35、ests were terminated when the cracks reached a predetermined crack length. In all tests, specimens were clamped between lubricated guides (ref. 6) to prevent buckling and out-of-plane vibrations during testing. surfaces of the specimens and guides. A 1/2-inch-wide (1.27-cm) slot was made across the
36、width of the guide plate to allow visual observation of the crack growth region. Light oil was used to lubricate the 6 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-RESULTS AND DISCUSSION The results of the fatigue-crack-propagation tests are prese
37、nted in table II, which gives the number of cycles required to grow the crack from a half-length a of 0.15 inch (0.38 cm) to the specified half-lengths. The number of cycles given in table II is the mean number of cycles required to propagate cracks of equal length on both sides of the central notch
38、. Included in table II are data from reference 7 for longitudinal-grain sheet specimens having a width of 8 inches (20.3 cm). gation and reference 7 was from the same mill heat. The material tested in both this investi- The ratios of the clear height to the specimen width were somewhat small for the
39、 4- and 20-inch-wide (10.2- and 50.8-cm) specimens (1.5 and 1.3, respectively). These small ratios were not expected to affect crack growth significantly in this investigation, however, because the crack lengths were relatively short in comparison with the speci- men widths (2a/w 0.5). from these sp
40、ecimens with the data from the 2- and 8-inch-wide (5.1- and 20.3-cm) spec- imens (having ratios of clear height to width of 3 and 2, respectively) indicated that there was no significant effect. Successful correlation (to be shown subsequently) of the data The cracks generally grew symmetrically abo
41、ut the specimen center line, with eccentricities seldom exceeding 0.05 inch (1.27 mm). determined graphically by taking the slopes of the fatigue-crack-growth curves (plotted on a linear scale) at various crack lengths. figure 2. grain sheet specimens (ref. 7). ence between the curves for different
42、specimen widths. in tests on clad 2024-T3, that specimen width had a relatively slight effect on crack propagation. Fatigue-crack-growth rates were The fatigue-crack-growth data are presented in At a given stress level there was relatively little differ- Schijve et al. (ref. 8) also found, Included
43、in figure 2 are the data for the 8-inch-wide (20.3-cm) longitudinal- For the 8-inch-wide (20.3-cm) specimens, the cracks consistently propagated more rapidly in the transverse grain direction than in the longitudinal direction. differences between these curves were slight at all five stress levels.
44、However, the The fatigue-crack-growth data were analyzed by using a modified form of McEvily and Illgs method (ref. 1). The principle behind this method is that the rate of fatigue crack growth is an explicit function of the product of the theoretical stress-concentration factor for the crack (qN) a
45、nd the instantaneous maximum net stress (S$;la). In the present investigation, the theoretical stress-concentration factors for cracks were com- puted by using the following equation (see appendix B): Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-I
46、1 I I I I I II 1111111111111ll11l1l11l11ll1l1111111l1 I111 II IIIII I $= 25ksi (173MN/m2) L I 0 I ,I LA ,I 3 I 10 10 I o4 N.cycIes 5 10 I06 I ,I ,I ,I 2 3 5 I06 10 IO I o4 10 Sb= 2 ksi ( 14MN/m2) I ,I ,I ,I 3 5 I06 10 I o4 IO N.cycles 7 10 I Figure 2.- Fatigue-crack-propagation curves for Ti-8AI-lMo
47、-IV (duplex-annealed) specimens tested at Sh = 25 ksi (173 MN/m2). 8 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-500 1000 I I ,_. 103 P I Rote. ln/cycle -6 10 -I 10 -7 10 IO Rate, nm/cycle I I I J 500 1000 1500 KTNS: ,hsi Figure 3.- Variation of
48、fatigue-crack-growth rate with KT in place of SkaX resulted in good corre- lation of the data (fig. 3) and thereby indi- cated that alternating-stress amplitude was a predominant factor affecting fatigue crack growth. Frost and Dugdale (ref. 11) reported a similar finding from tests on sheet specimens made of an aluminum alloy, mild steel, and copper. The fatigue-crack-growth data were also analyzed by using Paris stress- intensity analysis method (ref. 2). The data fell within a reasonably small scatter band on a semilog plot of rate of crack growth against Ak. (See fig. 4.) The f
