1、I NASA TECHNICAL NOTE ISA TN D-3075 -I - P. / FATIGUE OF REN 41 UNDER CONSTANT- AND RANDOM-AMPLITUDE LOADING AT ROOM AND ELEVATED TEMPERATURES by EWUY P, Phillips LungZey Resemcb Center Lungley Stutio, Humpton, Vd. NATIONAL AERONAUTICS AND SPACE ADMINISTRATION WASHINGTON, D. C. Provided by IHSNot fo
2、r ResaleNo reproduction or networking permitted without license from IHS-,-,-TECH LIBRARY KAFB, NM 0330067 FATIGUE OF REN 41 UNDER CONSTANT- AND RANDOM-AMPLITUDE LOADING AT ROOM AND ELEVATED TEMPERATURES By Edward P. Phillips Langley Research Center Langley Station, Hampton, Va. NATIONAL AERONAUTICS
3、 AND SPACE ADMINISTRATION For sale by the Clearinghouse for Federal Scientific and Technical Information Springfield, Virginia 22151 - Price $2.00 . Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-FATIGUE OF REI6 41 UNDER CONSTANT- AND RABDOM-AMPLITU
4、DE LOADING AT ROOM AND ELEVATED TEMPERATURES By Edward P. Phillips Langley Research Center SUMMARY Narrow-band random-amplitude and constant-amplitude bending fatigue tests were conducted on sharply notched Red 41 specimens at room temperature, 700 F (644 K), and 1400 F (1033 K). When compared on th
5、e basis of the root mean square of the nominal peak stresses, the random loading generally gave shorter lives than the constant-amplitude loading. Theoretical life predictions were made for the random-loading tests by using the Palmgren-Miner cumulative-damage rule and two different peak stress dist
6、ributions (the distribution determined from the tests and the classical Rayleigh distribution). estimated the fatigue life in practically all cases. The predicted lives based on the Rayleigh peak distribution were always less than those predicted by using the experimentally determined peak distribut
7、ion. For both types of loading in the long-life region, a loss of fatigue strength from that at room temperature occurred at TOO0 F (644O K) but no further loss occurred at 1400 F (1033 K). The predictions under- INTRODUCTION The prediction of the fatigue life of structures subjected to random loadi
8、ngs represents a challenging problem to present-day designers of aircraft and missiles. A particular vehicle may receive random loadings from several sources, each loading being characterized by a power spectrum of different mag- nitude and shape. At the present time no analytical methods are availa
9、ble to give accurate and consistent answers in this problem area even for materials for which the constant-amplitude fatigue properties are well-known. To make the problem even more difficult, the high-temperature environments in which many new vehicles must operate necessitate the use of new materi
10、als for which few fatigue data of any kind are available. A remedy to this situation is, of course, to conduct test programs employing representative random loadings and temperatures. One material which is being considered for high-temperature structural applications is Re the actual dimensions used
11、 for computing section properties were determined by measuring to the nearest 0.0001 inch (3 pm). cal elastic-stress concentration factor of approximately 7 for an axially loaded specimen. This edge-notched configuration corresponds to a theoreti- 2 Provided by IHSNot for ResaleNo reproduction or ne
12、tworking permitted without license from IHS-,-,-The Renk 41 material was obtained as nominal 3/16-inch (0.48-cm) thick sheet in the mill-annealed condition. The specimen blanks were sheared from the sheet with the longitudinal axis of the specimen parallel to the direction of rolling. following proc
13、edures: The blanks were heat treated before machining according to the (1) Heat to 1950 F (1339O K); maintain temperature for 30 minutes (2) Air cool to room temperature (3) Heat to 1400 F (1033O K); maintain temperature for 16 hours (4) Air cool to room temperature This heat treatment was used to o
14、btain maximum tensile strength. After heat treatment, the specimens were machined from the blanks, material being taken from all surfaces to remove the oxide film resulting from heat treatment. The finished specimens had a surface finish of approximately 0.0001 inch (3 pm) root mean square. The tens
15、ile properties and nominal chemical composition of Re therefore, an apparatus employing quartz-tube radiant heaters and reflectors was built at the Langley Research Center. control equipment was the same as that for the 700 F tests. shown in figure 5. The temperature- A test setup is RESULTS AND DIS
16、CUSSION Stress Response for Random Loading Typical power spectra of the shaker input (load) and the LVlYT output (stress) are plotted in figures 6(a) and 6(b) with logarithmic vertical scales. 4 . .- I Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-
17、Samples of the time histories of these signals are also shown in figure 6(c). Although the shaker input was rather wide band in frequency, the specimen responded only to frequency components near its natural frequency. This type of response resulted in a stress history which was of essentially const
18、ant fre- quency but variable in amplitude. Knowledge of the distribution of the peak stresses is important in char- acterizing the fatigue environment since the fatigue damage of a cycle of stress is generally considered to depend on stress amplitude rather than on the shape of the stress cycle. The
19、refore, distributions of peaks were obtained for sev- eral tests having different stress levels and temperatures. The distributions were obtained by recording the stress histories with a light-beam oscillograph and then manually measuring the distance of each peak from the mean over a 6.4-second sam
20、ple. The root mean square of the peak stresses (2 sP )112 was calculated from the measurements. For the first test, both positive and nega- tive peaks were measured. The analysis, however, showed that the positive and negative peak distributions were practically the same; thus, for subsequent tests,
21、 only the positive peak distribution was obtained. The results of the measurements are shown in figure 7 in terms of sP /py sJ? peaks exceeding Sp. from the distributions. An average distribution for all the tests is described by the dashed line drawn through the data. The solid line in the figure d
22、escribes the Rayleigh distribution. and in percent of No trend due to stress level or temperature was noted Test Results The results of the random-loading tests are presented in table I11 and are plotted in figure 8 in terms of cycles to failure and root-mean-square nominal stress, the quantity meas
23、ured in the test. The cycles to failure were not meas- ured in the tests but were computed as the product of time to failure and the natural frequency of the specimen. Curves were faired through the data for each temperature condition. The results of the constant-amplitude tests are presented in tab
24、le IV and are plotted in figure 9 in terms of peak nominal stress and cycles to failure. The cycles to failure were computed by the same procedures as those for the random-loading tests. Curves were faired through the data for each temperature condition and extrapolated to higher stress levels since
25、 information was required at these levels for the theoretical predictions of random-loading results. Theoretical Life Predictions Theoretical life predictions were made for the random-loading tests based on the Palmgren-Miner linear cumulative-damage theory. To make the calcula- tions, it was assume
26、d that each positive peak stress was followed by a negative peak stress of the same magnitude. This assumption was justified since the 5 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-positive and negative peak distributions were practically identic
27、al and since the stress time history had the appearance of a modulated s5ne wave. The cal- culations were made for various values of the root mean square of the peak stresses. Two life predictions were generated for each temperature condition; one using the measured peak distribution and, for compar
28、ison, one using the Rayleigh distribution. The method of calculation of predicted life for each distribution is as follows: To calculate the theoretical cycles to failure by using the measured peak stress distribution, figure 7 was divided into small, equal bands of ?e (?)I/* and the probability Pb
29、of peaks occurring in each band was determined. The probability of peaks occurring in each band was divided by the constant- amplitude life N at the band midpoint stress, and the resulting quotients were summed. The reciprocal of this sum gives the predicted life, or expressed mathematically is: To
30、calculate the cycles to failure by using the Rayleigh distribution, the same procedure was used except that the probability of peaks in each stress band was calculated by using the Rayleigh probability density function. upper cutoff level in stress magnitudes included in these calculations was taken
31、 as the level at which - pb did not change the sum of quotients by more than N 0.1 percent. This method of calculation of theoretical fatigue life is essen- tially the same as that explained in reference 2. The Comparison of Results Comparison of constant- and random-loa-ding. - In order to compare
32、the constant- and random-amplitude tests on the basis of peak stresses, the ordi- nates of figures 8 and 9 were converted to the root mean square of peak stresses and the faired curves were replotted in figures 10, 11, and 12. For the constant-amplitude tests, the root mean square of peak stresses i
33、s the same as the stress amplitude or Sp. mean-square stress was converted to the root mean square of peak stresses by the relation In the case of the random-loading tests, the root- 6 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-This relation was
34、 developed in reference 7 and is based on the theory of refer- ence 9 for the narrow-band response of a lightly damped single-degree-of-freedom system. When compared at equal values of the root mean square of peak stresses, the mean lives under random loading were generally shorter than those under
35、constant-amplitude loading, the difference increasing from high to low stress levels. The only case where life under random loading was longer than that under constant-amplitude loading was at 1400 F (1033O K) at the highest random- loading stress levels. (See fig. 12.) Comparison of experimental an
36、d predicted lives.- The fatigue-life predic- tions for the random-loading tests were made by using the Palmgren-Miner concept r-l of fatigue damage in which = 1 means total damage or fatigue failure. Two ZN life predictions were made for each temperature condition; one used the measured peak stress
37、distribution, and the other used the Rayleigh peak stress distribu- tion. (644 K), and 1400 F (1033 K) are plotted in figures 10, 11, and 12, respec- tively. always greater than those made by using the Rayleigh peak distribution; however, the difference between the life predictions was generally sma
38、ll. The results of the life predictions for room temperature, TOO0 F The life predictions made by using the measured peak distribution were In practically all cases, experimental lives were greater than those pre- dicted by using either the measured peak distribution or the Rayleigh peak dis- tribut
39、ion. Since the predicted lives represent = 1, this result means that r-l for practically all the random-loading tests, 1. A clear trend for for narrow-band random loading with a zero mean has not emerged from the literature. This point is illustrated in the following table : Reference 3 Reference 8
40、Reference 10 Reference 11 Reference 12 Reference 4 Reference 7 Present investigation Specimen material 2024 aluminum alloy SAE 4130 normalized steel 24 ST aluminum alloy 2024 aluminum alloy 2024-T3 aluminum alloy L.73 aluminum alloy 7075-T6 aluminum alloy Ren6 41 1 Bending Bending Bending Bending Be
41、nding Axial Bending Bending Central hole Edge notched Circumferential notch Circumferential notch Unnot c hed Unnotched Edge notched Edge notched - 7 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-A complete explanation cannot be given for the diffe
42、rences reported, but there is evidence that some of the differences in results could be due to the materials tested. For example, it has been reported in references 13 and 14 that for variable-amplitude tests with zero mean, 7075-T6 aluminum alloy gave consistently higher values of 2 than 2024-T3 al
43、uoninum alloy did. Of the investigations listed in the previous table, one used 7075 aluminum alloy and reported 1; 1, whereas four employed 2024 aluminum alloy and reported CN Temperature effects.- - For short lives, a progressive loss of fatigue strength occurred with increase in temperature, the
44、greatest loss occurring between 700 F (644 K) and 1400 F (1033 K). lives, however, the curves for TOO0 F and 1400 F begin to converge and in the case of constant-amplitude loading, fatigue strength at 1400 F is higher than that at TOO0 F. The constant-amplitude S-N curve for 1400 F is marked by a ve
45、ry sharp break at the knee; this result, however, is based on a rather small number of data points. (See figs. 8 and 9.) For longer The result that the fatigue strength did not progressively decrease with increase in temperature in the long-life region is not unique to Re et al.: Research Investigat
46、ion To Determine Mechanical Proper- ties of Nickel and Cobalt-Base Alloys for Inclusion in Military Handbook 5. ML-TDR-64-116, vol. I, U.S. Air Force, Oct. 1964. 2. McClymonds, J. C.; and Ganoung, J. K.: Combined Analytical and Experimental Approach for Designing and Evaluating Structural Systems fo
47、r Vibration Environments. Eng. Paper No. 3091, Missile and Space Systems Div., Douglas Aircraft Co., Oct. 1964. 3. Smith, P. W., Jr.; and Malme, C. I.: Fatigue Tests of a Resonant Structure With Random Excitation. J. Acoustical SOC. Am., vol. 35, no. 1, Jan. 1963, pp. 43-46. 4. Lowcock, M. T.; and W
48、illiams, T. R. G.: Effect of Random Loading On the Fatigue Life of Aluminum Alloy L.73. A.A.S.U. Rept. No. 225, Univ. of Southampton (Hampshire, England), July 1962. 5. Mechtly, E. A.: The International System of Units - Physical Constants and Conversion Factors. NASA SP-7012, 1964. 6. Weiss, V.; an
49、d Sessler, J. G., eds.: Aerospace Structural Metals Handbook. ASD-TDR-63-741, Vol. 11, U.S. Air Force, Volume I1 - Non-Ferrous Alloys. Mar. 1963. (Revised Mar. 1964.) 7. Fralich, Robert W.: Experimental Investigation of Effects of Random Loading on the Fatigue Life of Notched Cantilever-Beam Specimens of 7075-6 Alumi- num Alloy. NASA MEMO 4-12-59L, 1959. 8. Fral
