1、oic1-fNATIONALADVISORY COMMITTEEFOR AERONAUTICSTECHNICAL NOTE 4050STUDIES OF STRUCTURAL FAILURE DUE TOACOUSTIC LOADINGBy Robert W. Hess, Robert W. Fralich,and Harvey H. HubbardLangley Aeronautical LaboratoryLangley Field, Va.WashingtonJuly 1957Provided by IHSNot for ResaleNo reproduction or networki
2、ng permitted without license from IHS-,-,-TECH LIBRARY KAFB, NMIv NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS Illllllllllllllllllllllllllllli-011bL72iTECHNICAL NOTE 4050STUDIES OF STRUCTURAL FAILURE DUE TOACOUSTICBy Robert W. Hess,and HarveyImmNGRobert W. l?ralich,H. HubbardSUMM4J3YSome discussion o
3、f the acoustic fatigue problem of aircraft struc-tures is given along with data pertaining to the acoustic inputs fromsome powerplants in common use. Comparisons are given for results of4 some fatigue tests of flat panels and cantilever beams exposed to bothrandom- and discrete-type inputs. In this
4、regard it appears that boththe stress level of the testbhence, no generalization csaincreasing the fatigue life,a panel due to curvature andbeneficial.and the type of model are significant;be made at this time. With regard toit was noted that increased stiffeningpressure differential is particularly
5、ofINTRODUCTIONIt is well-known that fatigue damage can occur to aircraft struc-tures which are exposed to inte=se acouProvided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NACA TN 4050FATIGUE LIFE _Effects of Overall Noise LevelSome fatigue results obtaine
6、d for panels exposed to both of thesetypes of,input are given in figure 3. Fatigue life as a function ofthe overall noise level is shown for an 0.032-inch gage flat panei11 inches by 13 inches mounted over a rectangular cutout in a rigidframe. The panel was attached by small round head bolts tighten
7、ed toa predetermined torque. This configurationwas chosen for the reasonthat it.facilitatedassembly and disassembly of models while stress con-centrations similar to those in a riveted Structurewere retained. Forthe solid points which represent fatigue data obtainedwith the siren,the curve has been
8、sketched in through the available points to indicatea general trend of the data. Fatigue life is very strongly dependent-onthe level of noise excitation, since it varies from under a minute toseveral hours in the noise-level range of the tests.Attention is called particularly to the open points whic
9、h are dataobtained with the jet. These data fall generally to the right of thecurve in figure 3; thus, a longer fatigue life is indicated. !thiS dif-ference in fatigue life is due in part to the fact that the panel.hasa higher.root-mean-sqyarestress level when excitedby the siren for agiven overall
10、noise level than when excitedby the air jet.Effects of Method of MountingDuring the discrete frequency tests with these simple panels, theopportunitywas taken to change the mannerof mounting to evaluatepos-sible effects on fatigue life. These mounting configurationsare shownschematically in figure 4
11、 along with some of the test results in bargraph fog. For all the mountings, the gage and size of panel were con-stant and the input noise levels were also constant. The-basic configu-ration A is the same as that for which data were presented in figure 3.Failures in the skin panel occurred first nea
12、r the bolt heads and theav,eragefatigue life tor this configurationis used as a reference inthe figure,ConfigurationB is the same as configurationA except that a layerof bonding material is placed between the pnel and the rigid frame.During testing, the panel first peeled away from t-hebonding and t
13、heri-”failure in the skin occurred near the bolt heads. This configurationlasted on the average about 50 percent longer than confkmtion f-l.An attempt was made to eliminate peeling by bonding both sides andclampithepanel between two rigid surfaces as in configurationC.In this case failureoccurred at
14、 the edge of the frame and the averagemodel lasted twice as long as configurationA.=.?Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NACA TN 4050 5In order to study the effects of curvature some panel models wererolled to an 8-foot radius and were f
15、astened for testing to a curvedrigid frame as indicated in configurationD. Failures were initiatedw near the bolt heads as in configurationsA and B. This condition resulted by so doing th fatigue life was doubled as indicatedby thebar of dashed lines. ConfiguratioE is the same as configurationDexcep
16、t that tests were made with a pressure differential of 6 pounds persquare inch across the panel. This high internalpressure caused the firstpanel frequency to nearly treble and the fatigue life was greatly increasedas shown, in spite of the much faster rate of ap-ication of stress cycles.As a matter
17、 of interest a 0.064-inch-gagepanel was tested in a con-figuration similar to configurationA for comparison. It was found that) doubling the gage thickness of the panel increased its fatigue life toabout twenty times that of configurationA. This finding was confirmedin both the jet and siren tests-A
18、 limited nuniberof other tests have been made on larger and morecomplex panels. In all cases failures came first in the stiffener ele-ments; thus, the importance“ofdetail design of the panel supportingstructure is emphasized. It was also noted that crack growth was mark-edly more rapid in bonded str
19、uctures than in riveted structures.Comparison of Random and Discrete Frequency TeatsFlat panels.- The rest of the paper will deal with comparisonsof fatigue life under discrete and random loading at the same root-mean-square (RMS) stress levels. The results of flat-panel tests are givenin figure 5.
20、Time to failure in hours is shown for various root-mean-square stress levels for an 0.032-inch-gagepanel. we solid pointswere obtained by means of discrete frequency excitation from a sirenwhereas the open potits were obtained with random excitation from anair jet. The curve is a least-square curve
21、through the solid points.It can be seen that at a given root-mean-squarestress level, failureoccurs in a shorter time when the panels are excited by the random jetnoise. The fact that the random noise of the jet is more destructivemay result from the fact that some of the peak stress responses are s
22、ev-eral times as great for a given root-mean-squarevalue”as they are forthe constant-level siren tests. This phenomenon is particularly notice-able at the lower stress levels wherethe differences in fatigue lifetend to be the greatest and the panel dsmping is relatively low. AtProvided by IHSNot for
23、 ResaleNo reproduction or networking permitted without license from IHS-,-,-6 NACA TN 4050higher stress levels the panel damping is greater and.the differencesin fatigue life tend to be smaller.Cantileverbeams.- Similar fatigue tests have been made for notchedcantileverbeams at various stress levels
24、 foboth random- and discrete-type inputs. A schematicdiagram of the model along with some of thetest results are illustrated in figure 6. “ .Here again the time to failure in hours is shown at various root-mean-square stress levels for:the beams tested in bending. The speci-mens were 3 inches long,
25、1 inch wide, and 1/4 inch deep with 3/16-inchnotches located 1/2 inch from the root. The open points were obtained_by means of an amplified tape recording of jet noise fed into a shakerattached to the tip. The solid points were obtained by applying a sinus-oidal load at the tip in a Sonntag bending
26、fatigue chie. There is a“”tendency”in these tests also for the random load to be relatively less “-destructive at the higher stress levels and more destructive at thelower stress levels than the sinusoidal load. For the data of figures 5and 6 the strain-gage locationswere arbitrary and hence the roo
27、t-mean-square stresses shown in figures 5 and 6 are not necessarily comparableA theoreticalprediction of the time to failure for the randomloading is given by the solid curve. This theory is essentially onegiven by Miles (ref. 1) and is based on Miners rule of linear accumu-lation of damage. “-Itcan
28、 be seen that this theoretical curve fits the .data fairly well at low stress levels but is very conservativeat thehigher stress levels.CONCLUDINGREMARKSThe problem of acoustic fatigue of aircraft structureshas beendiscussedwith particular emphasis on a comparison of the fatigue lifedue to discrete-
29、 and random-type loadings. -Inthis regard it appears “ “- -that both the stress level of the test and the type of model are signif- -icant; hen-ce,no generalizationscan be made at this time. With regardProvided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-
30、NACA TN 4050 74 to increasing the fatigue life, it was noted that increased stiffeningof a panel due to curvature and pressure differential is particularly,” beneficial.%Langley Aeronautical Laboratory,National Advisory Committee for Aeronautics,Langley Field, Va., March 7,_1957.IETERENCES1. Miles,
31、JohnW.: On Structural Fatigue UMer RsndomLoading. Jour.Aero. Sci., vol. 21, no. 11, Nov. 1954, PP. 753-762.2. Lassiter, Leslie W., Hess, Robert W., and Hubbard, Hsrvey H.: An Exper-imental Study of the Response of Simple Panels to Intense AcousticLoading. Jour. Aero. Sci., vol. 24, no. 1, Jan. 1957,
32、 pp. 19-24, 80.3. Howes, Walton L., and Mull, Harold R.: Near Noise Fieldof a Jet-a71 Engine Exhaust. I - Sound Pressures. NACATN 3763, 1956.J.Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NACA TN 4050TABLE IACOUSTIC FATIGUE PROBLEMwACOUSTICINPUTSS
33、TRUCTURALCHARACTERISTICSwSTRESSRESPONSEA NOISE SPECTRUM AT ARBITRARY POINTB. CORRELATIONSI. SPACE AND TIME2.REAL AND QUADRATURE POWER FORVARIOUS FREQUENCIESL oyNAMic RESPONSE TO UN I FORM LOADB. DYNAMIC IN FWENCE “COEFFICIENTS (COMPLEX) .A. STRESS AT ARBITRARY POINTB. STRESS DISTRIBUTION OVER SURFAC
34、EFATIGUEILl FE-.Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NACA TN koACOUSTIC INPUTSAIRCRAFT SURFACEJI Lxi4$O/llsILLLEVEL,DECIBELS+-1 +170 . 130#/*,/-/” /160 I 40/ /-, ,3150 / /404-4048121620AXIAL DISTANCE , x/dFigure 1THRUSTL8/su i+TURBOJET 2,6
35、00 3,800ROCKET 25,000INPUT AND RESPONSE CHARACTERISTICSSIREN AIR JETN(u)IIINPUTSPECTRUM #)1 I 1 I I .4J!-AD:cEdJL_“U(U-MU(IJLFREQUENCY FREQUENCYFigure 2Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NACA TN 4050FATIGUE LIFEOVERALL NOISELEVEL, DECIBE
36、LS160r CmaDo1ya71-Ooam Cm155 Ia71abY,*150 + oa.;STFR$IN ., ”: .t-145 a a71 a71-,* a711 “0 LABORATORY AIR JET _ 140 - SIRENIOL .01 1I I J.1 10 100TIME TO FAILUE, HOURSFigure 3RELATIVE FATIGUE L!FEPANEL GAGE AND SIZE CONSTANTAE&?Zi tBC?%+%RADIUS =8D=EDEFERENTIALPRESSURE=6P s I1,1“ I.532:+.:?:”&-=i7j -
37、 -.230-mmoI I ! I0 10 2030 100RELATIVE TIME TO FAILUREFigure 4.,.tProvided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NACA TN 405Q nFATIGUE OF FLAT PANELSI 2,000 ro AIR JETa71 SIREN1%*oa71 008,000 a71RMS STRESS,Psl Oo -0 % -a+4,000 0 0*I vL) I I Io .0110
38、TIME T: FAILURE, dOURSFigure 5FATIGUE OF NOTCHED CANTILEVER BEAMSFUNDAMENTAL FREQUENCY, I19CPS30,00020,000RMS STRESS,PslIO,oootO JET NOISE LOAD&STRAIN GAGEa71 SINUSOIDAL LOAD THEORY PL1 I I I Jo .01 . I 10 100TIME TO FAILURE, HOURSFigure 6NACA - Langley Field, va.Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-
