1、NASA “IN/)-1522tt_TTECHNICAL NOTED-1522VARIABLE-AMPLITUDE FATIGUE TESTS WITH PARTICULARATTENTION TO THE EFFECTS OF HIGH AND LOW LOADSBy Eugene C. NaumannLangley Research CenterLangley Station, Hampton, Va.(.;,_ _ = rl LECOPYNATIONAL AERONAUTICS AND SPACE ADMINISTRATIONWASHINGTON December 1962Provide
2、d by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NATIONAL AERONADTICS AND SPACE ADMINISTRATIONTECHNICAL NOTE D-1522VARIABLE-AMPLITUDE FATIGUE TESTS WITH PARTICU
3、LARA_ENTION TO THE EFFECTS OF HIGH AND LOW LOADSBy Eugene C. NaumannSUMMARYVariable-amplitude axial-load fatigue tests were conducted on 2024-T3 and7075-T6 aluminum-alloy sheet specimens with a theoretical elastic stress-concentration factor KT of 4. The load schedules were designed to approximategu
4、st load statistics for tests on specimens of both alloys and maneuver loadstatistics for tests on specimens of 7075-T6 aluminum alloy. The test data wereanalyzed by assuming linear cumulative damage, and a limited statistical analysiswas used to strengthen conclusions. The value of the summation of
5、cycle ratiosZ _ with in of application of the highestwas found to vary changes frequencyNload level for eight-step tests and with the omission of the lowest load level_-_ nfor four-step tests. The variation in _ was not significant when the low-est load level for eight-step tests was omitted.INTRODU
6、CTIONFatigue tests which are designed to represent anticipated service loadingshave become increasingly important in recent years. Because the fatigue testsare often conducted on large components of new designs or on full-scale struc-tures, time and cost are considerations of prime concern. The test
7、 designer mustselect the anticipated load history and in most cases reduce it to a small numberof load levels which can reasonably be expected to give a realistic indication ofthe fatigue life. The reduction of a complex load history to a simple step testcan introduce variations in fatigue life due
8、to various testing techniques.Because of the prohibitive costs involved and the ad hoc nature of these fatiguetests, it has not been possible to determine which test techniques have a signif-icant effect on fatigue life.In order to help the test designer evaluate some of the suspected variables,the
9、Langley Research Center has conducted an extensive program of variable-amplitude fatigue tests in which many systematic changes in the load programwereProvided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-madeto determine their effect on the fatigue life o
10、f simple sheet specimens.Reference 1 presents the results of fatigue tests in which systematic variationswere madein such parameters as sequenceof loading, meanstress, and materialfor specimenstested by using loading schedules based on gust load statistics.Reference 2 presents results of tests in wh
11、ich load schedules based on statisticsof maneuverload peaks were used. The block size and range of loads representedwere systematically varied.The present phase of the investigation is concerned primarily with the effectof the lowest load level in the test schedule. This level normally contains one-
12、third or more of the load cycles to be applied in a test and, therefore, consumesa considerable portion of the testing time. Of secondary importance in thisinvestigation is the influence of the numberof load applications at the highestload level. In the present paper the results of additional variab
13、le-amplitudeaxial-load fatigue tests on 2024-T3 and 7075-T6 aluminum-alloy sheet specimensare combinedwith data presented in references 1 and 2 to ascertain whether omis-sion of the lowest load level or changesin frequency of occurrence at the highestload level have an appreciable effect on fatigue
14、life.SYMBOLSKTNnn8SaltSdSmeanSminVitheoretical elastic stress-concentration factorfatigue life, cyclesnumber of cycles applied at a given stress levelnumber of cycles at step eight of schedulealternating stress, ksistress at design limit load (43.6 ksi for 2024-T3 and 50.0 ksifor 7075-T6)mean stress
15、, ksiminimum stress, ksidiscrete gust velocity, fpsSPECIMENSEdge-notched sheet specimens of 2024-T3 and 7075-T6 aluminum alloy were usedin this investigation. The edge notches gave a theoretical elastic stress-concentration factor KT of 4. (See ref. 3-) This particular configuration wasused because
16、its fatigue behavior is reasonably close to the fatigue behavior of2Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-componentparts (ref. 4) and is the sameas the configuration used in refer-ences 1 and 2.The specimenswere madefrom part of a stock of
17、commercial O.090-inch-thick2024-T3 and 7075-T6 aluminum-alloy sheets retained at the Langley Research Centerfor fatigue tests. Sheet layouts and material properties are given in refer-ences 5 and 6, respectively. The appropriate tensile properties are given intable I.The specimennumberidentifies the
18、 specimenas to material, sheet numberjand location within the sheet. For example, specimenAll7N1-6 is 2024-T3 mate-rial (indicated by A) and was taken from the N1 position of sheet ll7. The 6indicates the position within the material blank (All7N1) from which the specimenblank was taken.Specimendime
19、nsions are shownin figure I. The rolled surfaces were leftas received and the longitudinal surfaces were machined and notched in both edges.The notch was formed by drilling a hole to form the notch radius. Residualmachining stresses were minimized by first drilling with a small drill and thengradual
20、ly increasing drill sizes (increment in diameter = 0.003 inch) until theproper radius was obtained. For consistency, drills were not used more than fourtimes before being resharpened or replaced. The notch was completed by slottingwith a 3/32-inch milling tool.!171 _. _ 058 rod-I500_J _ 375!L;-225o
21、4Figure i.- Sheet-specimen details.Burrs left in the machining processwere removed by one of two methods.Although the effect of changing deburringprocesses was expected to be small, thesame methods were used as had been usedpreviously in order that the present datacould be compared readily with exis
22、tingdata.The first method (ref. i) was usedfor specimens to be tested by using aload schedule based on gust load statis-tics and consisted of holding the speci-men lightly against a rotating cone ofO0 grade steel wool. The second method(ref. 2) was used for specimens to betested by using load schedu
23、les designedto approximate a maneuver peak loadhistory and consisted of holding thespecimen lightly against a slowlyrotating, pointed, bakelite dowel impreg-nated with a fine grinding compouI_d. Allspecimens were inspected, and only thosefree of surface blemishes in and near thenotch were tested.3Pr
24、ovided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-MACHINESAll of the tests in this investigation were conducted in four axial-loadfatigue machines (designated by numbers 6 to 9). Each of the machines is capableof two types of loading. One type of loading
25、 is mechanical, for which a beam isexcited to vibrate near resonance by a rotating eccentric mass driven at 1,800 cpmby an electric motor. The vibrating beam imparts axial forces to the specimenwhich acts as one of the supports. (See fig. 2.) The other type of loading ishydraulic and uses the same b
26、asic machine structure. The hydraulic systemincludes a hydraulic ram, attachable to the lower specimen grip, an electricallydriven hydraulic pump, a four-way solenoid valve, a semiautomatic electronicmechanism for load control, and a recorder for monitoring the loads. The mechan-ical drive system wa
27、s used for low-amplitude cycles which occur very frequently,and the hydraulic system, with cycling rates up to 20 cpm, was used for the lessfrequent high-amplitude loads. A complete description of the hydraulic andmechanical systems is given in references 1 and 6, respectively.-Strain-gaged weigh ba
28、r- Upper grip“_-I Monit oringEquipment_Recorder-t ControllerLower grip-I _-Flexure plotes-Spring-moss systemflexure ploteflexure pl_Prelood._ spring-Removoble pin andclevis shown rotated90 Hydroulicrom,Weight_.,_Ro vibr t ingbeomtaringeccentricmossFigure 2.- Schematic diagram of fatigue testing mach
29、ine.The loads on the specimen were monitored by utilizing weigh bars, equippedwith resistance wire strain gages, in series with the specimen. For mechanicalloading, the strain-gage output was monitored by using an oscilloscope and asso-ciated balancing apparatus. The hydraulic system utilized the sa
30、me strain-gageoutput to control the loads. The hydraulically applied loads were monitored ona strip-chart recorder with use of a second set of strain gages.4Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-The load-measuring apparatus was calibrated p
31、eriodically. The load on thespecimen was estimated to be maintained within 20 pounds of the desired loadfor the mechanical system and within 50 pounds of the desired load for thehydraulic system.LOADING SCHEDULESGust LoadsEight-step loading schedules were used in this investigation to approximatea g
32、ust load history on the specimen. The load schedules used are taken fromreference i and are presented in table II for 2024-T3 and 7075-T6 aluminum-alloyspecimens. Statistical data on the frequency of occurrence of atmospheric gusts(ref. 7) were used as the basis for the loading schedules. For conven
33、ience, ashortened tabulation of the statistical values used is presented in the followingtable:Gust velocity, Numberft/sec exceeding30 0.6327.5 I.Z725 . 2.822.5 6.820 2017.5 7215 27012.5 975i0 3,3007.5 13,9005 51,0002.5 175,0000 500,000In order to convert these data to loading schedules, the followi
34、ng assumptionswere made:(i) A 30-fps gust produced design limit load(2) Alternating stresses could be obtained from the following simplerelation:Salt = (Sd- Smean)_ 0With the use of the equation for alternating stress, the gust velocityspectrum was converted to a stress frequency spectrum for mean s
35、tresses of17.4 ksi and 0 ksi for 2024-T3 and 20 ksi and 0 ksi for 7075-T6.Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-Each stress frequency spectrum was divided into eight approximately equalstress bands and a discrete stress level was selected t
36、o represent each stressband. The discrete stress level was determined by numerically integrating thetheoretical damagefor each stress band, linear damageaccumulation being assumed,and then selecting a discrete value of stress that will produce the sameamagein the samenumberof cycles. This process is
37、 explained in detail in refer-ence i. The integrating process required an S-N curve for each material and meanstress; data for these S-N curves are taken from references i, 8, and 9 and arepresented in figures 3 and 4. For stress bands which are lower than the fatiguelimit (stress at which the fatig
38、ue life is 107 cycles) of the specimen, the dis-crete load level was selected at approximately the samerelative position withinthe stress band as had been calculated for higher stress bands.y_The summation of cycle ratios L _N where n is the number of cyclesapplied at a given stress level and N is t
39、he number of cycles to failure at thesame stress level; for each test block was made to be approximately 0.I_ so thatfailure would be expected to occur at the end of i0 test blocks. All stresscycles at a given level within a block were applied in one continuous sequence.The load levels within each b
40、lock were applied in a random manner by using asequence obtained from a table of random numbers. Each block had a differentrandom schedule until the twentieth block; thereafter_ the schedule for the first20 blocks was repeated. The same random schedule was used for all tests.Maneuver LoadsEight- or
41、four-step loading schedules based on the frequency of occurrenceof peak loads in maneuvering flight were also used in this investigation. Theload schedules used are taken from reference 2 and are presented in table llI.Load statistics for the frequency of positive load factor peaks (ref. i0) weretra
42、nsformed into a peak stress frequency spectrum. This transformation requiredthe following assumptions: (i) a design limit load factor of 7.3 and (2) a i g(level flight) stress equal to 7 ksi. The maneuver load statistics are presentedin the following table:Acceleration, Numberg exceeding7-3 197.0 23
43、6.0 1155.0 4304.0 1,2203.0 2,8002.0 5,6001.0 iO, 000As in the case of gust loads the spectrum was divided into stress bands and anumerical integration of theoretical damage was performed to select discrete loadProvided by IHSNot for ResaleNo reproduction or networking permitted without license from
44、IHS-,-,-I00 IOFaligue life, N, cycesFigure 3.- Results of constant-amplltude fatigue tests of 2024-T3 aluminum-alloy specimens.(Ticks represent scatter bands and numerals indicate number of tests in each group.6O50 -4OE2OqI0 .iII I0000_ lM3 _ IO40 Ref Ic, Ref 8a Ref 9r- O_d not falltI:; ilo_ ,o_ ,o
45、,Fatigue life, N, cyclesFigure 4.- Results of constant-amplitude fatigue tests o5 TOT_-T6 aluminum-alloy specimens.(Ticks represent scatter bands and numerals indicate number of tests in each group. )?Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-l
46、evels to represent each stress band. The S-N curve for maneuver loads requiredconstant minimum stress rather than a constant mean stress as in the case ofgust loads. Thls S-N curve is presented in figure _ and is taken from refer-ence 2. The same random sequence of loading used for the gust tests wa
47、s usedfor these tests. Maneuver load tests were conducted on 707_-T6 specimens only.7o6o5o2OI00 Ref 2I“ Did not fail4Fatigue life, N, cyclesFigure 5.- Results of constant-amplitude fatigue tests of 7075-T6 alumlnum-alloy specimens.(Ticks represent scatter bands and numerals indicate number of tests
48、in each group. )Test VariationsFor each of the load schedules presented in tables II and III_ a series oftests was conducted In which the lowest load level was omitted to determinewhether this level had an effect on the fatigue llfe. Whether the lowest loadlevel does or does not affect the fatigue l
49、lfe is important because the lowestload level contributes as many as 84 percent of the gust loads (33 percent formaneuver loads); thus, this load level materially influences testing time, andtherefore, testing costs.8Provided by IHSNot for ResaleNo reproduction or networking permitted without license from
