ACOUSTIC EMISSION, AN EXPERIMENTAL METHOD ‘I‘stis fat the Degree of M. S. MICHIGAN STATE UNIVERSITY Paul S. Shoemaker 1961 w 1" LIBRARY 1'? . Michigan State ‘ 9,.) University I This is to certify that the thesis entitled ACOUSTIC EMISSION , AN EXPERIMENTAL METHOD presented by PAUL S. SiOEMAKER has been accepted towards fulfillment of the requirements for MASTER OF SIIENCE degree in MECHANICS , Major professor Date m5 ; 1.9—6L g!" r! ABSTRACT Acousuc EMISSION, AN EXPERIMENTAL moo by Paul S. Shoemaker Acoustic emissions are sound pulses emitted from crystalline materials subjected to an applied stress. The purpose of this investigation is to exanine an experimental method that would enable one to study acous- tic eaission and to show soae experimental results. These acoustic emissions are at such low energy levels that aaplification is neces- sary to detect them. Great care must be taken in the aethod of detec- tion so that any experimental process that generates aechanical or electrical vibrations will not interfere or be confused with the actual acoustic emission signals. The transducer employed for sound pulse detection is an ammonium-dihydrogen phosphate piezoelectric crystal. The loading aechanism used to subject a specimen to stress must be completely silent. This silent load is produced by cooling contraction of a pre-heated aluminum rod. The results presented are in the fora of histograms with acoustic emissions per ainute versus specimen strain. There is ample verification that emission signals are detected and that the experiaental method is quite satisfactory. An analysis of the re- sults indicates that there is an obvious difference in acoustic eais— sion between different engineering aaterials; and some evidence that there could be a difference in the same engineering aaterials. There are obvious patterns present in the histograas that indicate possible correlation between acoustic eaission and engineering design properties. Certainly, there is evidence present that indicates a relation between acoustic emission and aetallurgical and crystal properties in aetals. ACOUSTIC EMISSION, AU EXPERIHEITAL HETHDD PAUL S. SHOEHAKER A THESIS Submitted to the School of Graduate Studies of Michigan State University of Agriculture and Applied Science in partial fulfillaent of the requirements for the degree of MASTER OF SCIENCE Department of Applied Mechanics 1961 WW5 'The author is indebted to Dr. Clement A. Tatro for his guidance and assistance throughout this research. It is certain that this investigation would have been almost impossible without his abilities. Mention must be made of the fact that the thermal loading method is an idea conceived by Dr. Tatro, and was develOped by the writer. The author is also grateful to the Applied Mechanics De- partment, Dr. Charles 0. Harris and the Division of Engineering Research for help in many ways. .Appreciation is also expressed to the author's wife, Beth, for her typing and proofreading. ii TABLE’ F‘CONTENTS CHAPTER PAGE ABSTRACT ACKNOWLEDGEMENTS.. ......... . .......... ..... ..... ii LIST OF FIGURES... ................ . ............. iv NOTATION.. ........ . ....... . ......... ... ........ v I INTRODUCTION..... ..... . .................... ..... 1 II EXPERIMENTAL METHOD... ......... ................. 2 Necessity Of Silence ....................... 2 Silent Loading Mechanism ..... . ........ ..... 5 Electronic Mechanism.. ..... .... ....... ..... 12 Noise Isolation ............ ................ 16 Specimens And Stress-Strain Recording Mechanism........... ...... . ............ .... 19 III EXPERIMENTAL PROCEDURE ................... ....... 28 IV PRESENTATION OF EXPERIMENTAL RESULTS............ 30 V SUMMARY AND CONCLUSIONS ..... .... ................ 39 BIBLIOGRAPHY ....... . .................... ........ bl iii LI Fm *— FIGURE PAGE 1 Mechanical Apparatus ................ ..... ..... . ..... ... .. . 6 2 Heat Tube Assembly... ......... . ............... . ......... ..... 8 3 Poly-Tube Insulator.... .................. . .......... . ........ 9 h Thermal Bar Cooling Rate..................................... 11 S C1013 Cold Rolled Steel, Stress-Strain Diagram... ............ 13 6 Electronic Recording Mechanism............... ..... ........... lb 7 Electronic Playback Hechanism.. .................... . ......... l7 8 Specimen Noise Isolation ................................. .... 18 9 Anechoic Chanber...... ........... . ....... .................... 9 10 Detail Of Specimen........................................... 20 11 I Specimen, Strain Extensometer And Crystal Assembly ......... .. 22 12 Load Cell And Wheatstone Bridge................... ........... 23 13 Strain Extensometer Linear Ca1ibration Curve................. ?6 1h Strain Extensometer And Wheatstone Bridge ............ ........ 27 15 Histogram And Stress-Strain Diagram of Run L/l/61, 20?b-Th Aluminum Specimen...................... ..... ............... 31 16 HistOQram And Stress¢Strain Diagram of Run b/3/6l, Annealed C1018 Cold Rolled Steel Specimen............... ..... ....... 32 17 Histogram Of Run 12/23/60, Shot Peened Steel Specimen. ....... 3h 18 Relative Signal Voltage or Run b/1/61, 20211-711 Aluminum Specimen................................................... 35 19 Relative Signal Voltage Of Run b/3/61, Annealed C1018 Cold Rolled Steel Specimen... ..... ......................... ..... 36 20 Photograph Of Oscilloscope Image For A Yielding Region Of Run 12/23/60, Shot Peened Steel Specimen................... 38 P1 Experimental Apparatus..... ...... ......... ..... .............. 15 iv I’o ADP Cfl °C ‘5 in. kg ksi psi HMS 38C. L0 SKL N0 TA TION_ cxtensometer beam radius attenuation angstroms ammonium dihydrogen phosphate speed of sound centimeters degrees centigrade modulus of elasticity degrees fahrenheit amplifier gain inches RIIOgrams hips per square inch critical section length meters millimeters pounds per square inch acoustic mismatch or impedence ratio rootcmean-Square seconds piezoelectric crystal sensitivity Spencer Kennedy Laboratories extensometer beam thickness signal voltage piezoelectric crystal stack element width deformation strain -6 micro (lO ) pi (3.1h) density stress ohms vi CHAPTER I IITRCDUCTION Acoustic emissions are sound pulses emitted from crys- talline nterials subjected to an applied stress. For many years it has been known that the metal tin, when stressed, emits somds that can be heard by the unaided ear. However, only quite recently has there been am investigation to study the sound pulses emitted from stressed crystalline metals such as steel, aluminum, copper, lead, zinc, etc. In 1950, Dr. Josef Kaiser (Rs 50), in Germany, presented a doctorate thesis on the subject and showed that these materials exhibited acoustic emission. The emissions, hosiever, were at such low energy levels that amplification was necessary for detec— tion. Since Dr. Kaiser's initial investigation, there has been a great deal of research in the acoustic emission phenomenon carried out by B. H. Schofield of Lessells and Associates, Incorporated, Boston, Hassachusetts (Sch 58,.PR 1-11) and by Dr. C. A. Tatro at Hichigan sun University (ER 57, .Ta 59, r: 60). Because the acoustic emissions are at such low energy levels, great care must be taken in the detection of the sound pulses so as not to confuse them with mechanical and electrical noise that may be present in the experimental apparatus. It was for this reason that a silent loading mechanism and noise isolation of the specimen hid to be deve10ped. It is the purpose of this investigation to ex- amine an experimental method that would enable one to study acoustic emission and to show some experimental results. CHAPTER II WWAL HETIKD 1. IECESSITY OF SILENCE To further impress on the reader the necessity of mechanical and electronic silence in the vicinity of the specimen under test, an approximation will be made of the magnitudes of stress and strain that accompany a detectable acoustic emission. The piezoelectric transducer chosen for this experiment was a IiS‘ .Z-cut ADP (a-onium ditwdrogen phosphate) crystal stack composed of eight elements in parallel. The crystal sensitivity is: S - .177 voltsémeter 2 newto meter The noise inherent in the electronic recording mechanism shown in Fig. 6 was approxintely 5 x 10"3 volts at the oscilloscope output. The sullest acoustic emission level judged detectable at the oscil- 2 loscope was about two times this or 10' volts. The pre-amplifier gain was 800 and the signal loss (attenuation) in cabling from the crystal stack to the pre—amplifier was approximately .5. Another loss in signal detection was the acoustic mismatch or impendence ratio at the interface between the crystal stack and the specimen. This is determined as follows (HuBo 55)! - dence ratio r - .2 (crystal) where: r impe pc (specimen) at interface p - density of nterial c - speed of sound in material 6 - 6.1 x 10 k 3 AC (ADP) {3 x sec. 3 pc (aluminum) - 13.8 x 106 .1393 g n x sec. on: (steel) - 39.1 x 106 3:33 _m_ n x sec. then, Aluminm: r .- 6.1 - .11112 13—38 Steel: r - 6.1 - .156 3?.‘1' In continuing, one would choose the aluminum specimen as an optimistic approach in stress detectability. The stress pulse received by the crystal stack from the specimen head is then given by: 0- V ( §)AGr 0- 10.2 .177 1. 9 x 0' . . 22 o - 21.2 x 10"2 newtons/m2 6 o - 30.8 x 10- pounds/in.2 where a - stress pulse at specimen head S - crystal stack sensitivity per element width - .177 voltsfimeter 2 newto meter - element 214:1: - 1/16 inch - 1.59 x 10“3 meters 2 - smallest detectable voltage level - 10' volts Cable Attenuation - .5 cat-<1: I - pre-amplifier gain - 800 r - impedence ratio - .11112. The stress pulse at the specimen head is, hovever, a smaller pulse than is actually coming from the specimen critical section. Referring to the specimen detail, Fig. 10, one may see that the pulse traveling h from the critical section toward the head will either spread over the entire head or will be kept focused to the one half inch shank diameter. Assuming the critical section stress pulse spreads over the entire head, the critical section stress is: o - o'x head area c.s. critical section area _ -6 2 _ 4. ob... 30.8 x 10 x i 2 8.7h x 10 psi Assuming the critical section stress pulse is kept focused to the one half inch shank diameter, the critical section stress is: at s - o'x shank area ° ‘ critical section area a - 30.8 x 10'"6 x . 2 - 2.18 x 10'“ psi c.s. .187 2 Again using an optimistic approach, a stress of 8.7h x 10-h psi in the critical section would lead to a strain of: -b -S s-o-8.th10 -8.7bx10 ‘E 10 Assuming all of this strain occurs in the specimen critical section, u in./in. ‘which is approximately .75 inches long, the total deformation due to one detectable acoustic emission is: A ' IL (8.7h x 10-11)(.7S) - 6.55 xlO"ll inches 16.65 x 10'11 cm .01665 i 'which is approximately one two-hundredths of an aluminum atomic dia- meter. It should be apparent that the above analysis is a rough approximation and should be valued accordingly. There is some 5 question as to whether the elastic modulus used to obtain strain from the critical section stress should be Young's modulus of elasticity (107 psi) or another elastic constant which more closely describes inter-crystal action. An acoustic emission does not originate from a plane in the cross section but originates in a single crystallite of a polycrystalline specimen, and is thus effectively a point source. The sound pulse emitted from the point source should, however, develop into a plane wave within the length it travels. The relative Igni- tudes arrived at above do, however, serve to point out the need for proper somd isolation of the crystal stack and specimen. Ismgine the confusion in detecting an acoustic emission when the noise level in- herent in the most quiet conventional testing machine is forty milli- volts (Ta 59) or four times the magnitude of the smallest detectable acoustic emission. 2. SILEUT LOADING HERMES! Basically and simply, the silent load is produced by cooling contraction of a pro-heated aluminum rod. The thermal loading apparatus is shown in detail in Fig. 1. The one inch diameter thermal bar is suspended by a ball and socket arrangement from a derrick-like frame- work fixed to the upper stationary head of the Riehle testing machine. Aluminum was chosen for the thermal bar because of the relatively large coefficient of thermal expansion, and was hung from a ball and socket Joint to minimize effects of moments produced on the specimen (assuring an uniaxial tension). The theml bar is heated over approximately six feet of its length by a cylinder-like furnace or heat tube which surrounds it. The heat tube contains two electric strip heaters, each three feet m M/f1eaf rub? 4663711701 Dar 1' I 1“ 1” 1““; 1 1 ,. ”"1 urper plate 0,: I 1; - 1- . 1 1: 11:1 .3 “1““ :‘e 1mg mac/11m3 ‘ 1:“ 1 1 111 1 9 1. ' 1 111,1 11 1: ,- '11 1 . 1 1.11 _ 1 ,111 11 1_:--f1ec31‘ woe alzqn- . 2 1 11:11 , . menf adjustmenf . 3211:. 1 1 1 1 1 x1- EWOUGPU framework 5‘7‘ ' 1'1 "' *1 51—1 (no:L p071" of +———L_ ' 11 1" , L4. - - 1 ‘: 1 1 ! fesfmg mac/Sme) 1 1 1 1 : . 1 1 1 1 1: 1 1 11 1 3 1 1T 1.5111, 1: 7—7 11—1 1 ’111 11 star/man; resting 1 1 1 31;, 1 machine frame 1 .1. 1 .‘ . 1 1. .; 1. 1" 1 ‘1 1 1 1 1" 1 ”1.1 ’1 4 1 ,~ ,' 1 1 11.1111 1““: /’—pu(g’~1ub€> 1119111/aror 1 4-- +- 1 / a 1 1 1? 1 1 1 1/ ”/11/101717 C911 ,1..- 1- .11-. -.._-, ... “’1'" I _ ,- . 1‘1 W 1*: ,/172e1‘:1! :11th 11‘ “1“ “—311" 1//‘ 4:1,-» (701:: C 79510! 1 1%» ”a“ 1-—»~GT1PC/"1011"C (ficmber 1 1 ; 1115;111:311. 1 9' “I 1,__1 1 11:. ,/ _' 151mm cwfengomefer 1_ 1 1 j} _- If, 1/“ 1 “j 1 11:: 7:-.. 1." ”SC-earner) 1 1 ‘1 :1 114‘: , 1 (19511119 macfime : ; ' ’--——-~ f/ ; adjustabm» head 1 1 7"" ”fr“ / : ....— :e—r '1 1 1 ‘v—— F'R— i ,1 1 A, 1-,_ _ 712 1-“ _1 5 1 PIP/719 ’ { console 119511179 macfu‘ne base F/GURE / ~ Mecfiam‘cal 1 1 1103C] . . . 1 1 mdzcatzon J App—€110 1115 7 long, that run end to end along the entire length of the six foot fur- nace. The heat tube itself is constructed of two layers of steaa pipe insulation secured rigidly by cheesecloth and wheat paste. Refer to Fig. 2. The prOperties of piezoelectric materials vary considerably with temperature and the least percent change in these properties oc- curs around rooa tenperature (20’ C, 68' F)(Br 53). For this reason, it is necessary to insulate the crystal stack fro: the heat produced by the theraal loading nechanisa. The insulator shown in Fig. 3 con- sists of thirteen quarter inch dianeter by six inches long by .008 inch wall thickness stainless steel tubes equally spaced between two ten- inch dianeter steel plates. The upper plate, attached closest to the heat source, is one inch thick and has six pie-like sections removed. The spoke-like upper plate ainiaizes radiation to the lower plate and serves to dissipate heat as it is conducted radially outward along the spokes. The stainless steel tubes provide the necessary strength properties required to support the loading, yet have a rela- tively low thermal conductivity. The lower plate is a solid disc, three quarters of an inch thick. This poly-tube arrangement adequately provides the necessary surface area to dissipate the heat. The as- siveneas of this insulator is not to insure adequate heat transfer, but to provide strength qualities necessary for ainisun deflection under an applied load. It aust be kept in mind that deflection is at a preniua because it is the principle load producing aechanisa. In other words, any deflection absorbed in the tension linkage is not i‘POsod on the specinen as it should be. Controlled deflection as opposed to controlled loading, is haven“ wash anon «m 350: hounded...” 33.13?“ am mmBE .3355 305084 5 mg 10 a unique advantage which the thermal loading aechanisn has over a dead weight silent loader. Once a weight is applied to a dead weight loader, the stress is iaaediately changed. This stress range is aoved through so rapidly, that acoustic emission recordings are alaost iapossible. One could conceivably aove through the yield point (an interesting area for acoustic eaission study) instantaneously with one load application. The therlal loader applies the load very much the sane way as a con— ventional testing aachine, although the straining rate is somewhat less. The lowest conventional testing machine deflection speed available is .012 inches per ainute, while that of the thermal loader is approxlaately .OOh inches per ainute. The thernal loader straining rate is taken fron.the cooling rate curve shown in Fig. h, and is calculated, assum- ing a constant slope. It may be seen that this curve is exponential, but is approxiaately linear over the cooling range necessary to pro- duce the required loads. The straining rate can be varied by siaply varying the cooling rate. The controlled deflection and low straining rate lake it possible to record the yield point region of steel over a period of about three ninutes. This ennbles one to obtain a very accurate picture of the stress-strain curve right up to fracture Ulthr out fracturing the Specilen. The magnitude of the load produced varies with speciaen material, critical cross section, critical length and aaount of therlal contraction. For example, a steel speciaen of .25 inches critical cross section and 2.5 inches critical length results in a load of about 3000 pounds when.subjected to approxiastely .15 inches of therlal contraction. This is well into the plastic region of this particular steel. 11. I. 80 THERMAL BAP COOLING PATE F/GUPE 4 r70 / P60 / / / / +50 / .3" L40 E 7 <1. ‘5. ~30 " F20 HO J L 1 l C? .05 .IO .15 .20 quth'nc Conrrwzc Non - inches 12 The noise inherent in the theraal loading aechanisa is practically nil. A stress-strain curve for a steel speciaen is shown in Fig. 5. This speciaen was initially loaded through the curve AK: and exhibited acoustic eaissions without on particular noise interfer- ence. The specimen was then reloaded froa F to C to D. Acoustic eaissions do not occur in a speciaen over the pro-loaded portion (1“ to C) and there could be no confusion in this loading region as to whether the signals were loading nechanisa noise or acoustic eaissions. In this loading region, no signals of any kind were encountered. However, inaediately upon reaching point C, the acoustic enissions began and continued to point D. The peculiar nose in the yield point region, (point B) of Fig. 5, is not aaterial for this report and no atteapt will be ude to explain it. 3. ELECTRONIC WIS! The electronic recording nechanisa which is necessary for collecting data is shown in Fig. 6. This instruaentation is shown photographed in Fig. 21. The stress pulse froa the speciaen is re- ceived by an anoniua dihydrogen phosphate piezoelectric crystal stack. This crystal stack is corposed of eight one-sixteenth by one half by one inch plates that fora into a one half by one half inch square by one inch high transducer (See Fig. ll). The signal froa the transducer is aaplified appmxiaately 800 tines by a Tektronix type 122 pro-amplifier. The amplified signal passes through a Spencer Remedy Laboratories frequency wave filter where all frequencies below 1500 - 2000 cycles per secorud are eliminated. The acoustic eaission sigaal is thus available for viewing with an oscilloscope and for permanent recording with a tape recorder. The acoustic signals are recorded on 13 m.c.~\.E :Q Eotm mxyvmm nfixxyw mxxmw UKXEV fl .lel d — q hx — m MQDGE \o\\\m SE éqmoSQ 233m ..mmmgm \Eflm cczom 28 $05 Q/ ()SX) scans 1h EmEDEoE 9.685% _ 820:?ng 0 MQDOE EVQ booq cchoE 0:: meg m ulmtfi . _ $0.50an l. " dQUk RCCQEQ anecqot.~E dtoclqOKSE " 00: E 0.63 \ xii. ” xquq J E052 \s ..... tobacco 1 :dECde, a .. Q.» DSQ anus totem Vtgmmiu dex/ viva: t#_ QQOumozbmg 00% u 0 - yr. m 529% em: 3&3 arm cqa x56e§$ W 803 oeummonQK QEUC T o I." .. I‘OTJJ to" 00.... FIGURE 21: 15 Experimental Apparatus 16 the even channels (2, h, 6) and voice comments and specimen loading data are recorded on the odd channels (1, 3, 5, 7) of a seven—channel Anpex (FRpllOO) tape recorder. A Bell & Howell microphone and Knight audio amplifier are used for voice recording. The oscilloscope used in this experiment was a Tektronix type 532 with a "D" type plug-in unit. Two types of playback or counting mechanism were used in analyzing the recorded data, as shown in Fig. 7. In one case, the tape recorded acoustic signal is played through the oscilloscope (for visual viewing) into a Hewlett-Packard model 523-8 electronic counter. Simultaneously, voice recording is played back through the audio amplifier and loudSpeaker. In this manner, the loads recorded on the voice channel are synchronized with the acoustic signals being counted. The other method of data analysis employs a Ballantine model 320 true root-mean—square voltmeter. The acoustic signal from the tape recorder is directed into the HHS meter where it may be visually observed as an EMS meter reading or recorded on a Sanborn oscillograph. Again, the loudspeaker announces the applied loads to the specimen in order that synchronization is available between acoustic signals and specimen stress properties. In both cases of data analysis, the SKL wave filter may be used for refiltering. h. NOISE ISOLATION The loading linkage is capable of introducing noise inter- ference if the specimen is not acoustically isolated from these sources. For this reason, acoustic mismatch materials are provided between the specimen heads and the loading mechanism as shown in Fig. 8. Washers lof clay-impregnated Bakelite one sixteenth of an inch thick are glued to the loading frames and provide the necessary noise absorption l7 EMEUEdE 500390 uiotoofiu. K MQDGE - ii}. mikwmcmmw. Emém «avg wwz 2o n .-- Ewisuhq w I Zoe‘onm _ _ 063% m “a bsolT iii a RIOS? T-..:----:,I.iilli illillia ill - -.--..,-:JJJ 1 i it --.! -- Il1--..MH.I..u--... - - _. -- i t 1. .._ _ ”.003 £36-- “ lililii . 3:0:qu _ moreoofi . 300T: _ m r not. _ _ oQU . _ Rwandan @ $32 ,_ _ W m K w 335 w _ :9, r M new; 60:..Qu l zoomznm .143; min ”7. , -._-...Mn..-...l.;_s--.. Enact 1:" illlll 292 aromas @3358 zlmmim o: moons i -E.C:QE< . scanning 9.0: Q. 1 1 iii Vii}! a! .....-itsri... {£511.}? 1.!!! 11:1: r-.. i. -rniillllllllflg. been RIQZX “ I --- i, A Tillie: .2551.-- 2 ton. . . i -. v .. ... 1. hotgow __ E mifl. ” _ 320235 _ @5056on m ..Q.. m “ 3:5. . aobnoocmor mmmm Evoke 888:.ch M . Em “ _ page do _ l - on cacao mam a -, r I - .. ..n 00: mm ..Emcsml 5203 cm: ._ . xmnan 18 /,/'C7gs fol C10 mp / C(a (m n /g 91 meg ated -/ Bakelite (Both / 5/ < ends) ///6 m. i” 2 _ FPH T‘ A “ (Ea-cm ends and 9; V crystal clamp) ‘ //.32 m. ' Loading ‘ Load: ng Force \ Force {J i ”f Loadl'rrg frame 1 I Y 77 ' I 7% j ‘r 7' /"/' _/ ,' , l [,7 7 . I '4. I ’ ', , .' / , . l v ,1 , x. I / , , 1' ,1 ‘1' ,/ / 1‘ , R . ,I,“ / , . . , ., ,r, . ‘ , .’ . C I ‘ a' 4‘ LL. 4‘ A . #4 J L7 C ,_ SP5 C //\//E N F/GUPE 8 NO/SE lSOLAT/O/V l9 originating in nut and bolt assemblies in the loading linkage. This clay-impregnated Bakelite provides an attenuation of approximately .60 (Ho 61). A good grade of one thirty-seconds of an inch thick felt is glued to the specimen shank close to the heads where the shank goes through the loading frame. The felt is also glued to the specimen head in the notches provided for mounting the crystal clamp. The felt pro- vided in these places eliminates any possible metal to metal contact which could be a noise generator. The transducer crystal stack is very sensitive to electro- static pickeup and it becomes necessary to construct an electrostatic shield around it. The electrostatic shield consists of two nested metal boxes completely surrounding the crystal stack as shown in Fig. 1. An anechoic chamber was constructed (Sc 50) which encloses the specimen and crystal in order to minimize any noise interference from general laboratory activity. Shown in Fig. 9 is a close-up of the chamber and Fig. 1 shows its placement among the apparatus. A foam rubber pad is placed between the anechoic chamber and its resting place to eliminate any mechanical vibration that may be transmitted from the ground or the testing machine. A small window has been placed in the chamber wall to allow a view of the specimen. Experimental tests were conducted during the early morning (between 12:00 midnight and 6:00 a.m.) to eliminate any possible mechanical or electrical noise which would be present during normal laboratory activity hours. 5. sprcmsns AND STRESS-3mm RECORDING uscmmsq Specimen geometry is shown in detail in Fig. 10 and in the ZMECMQW Q0 $EMQL 9 MQDQQ mp5 58 - :l 1.5.9:?qu GINO ®\ .4. 1:9 .‘ - _ .9.an A a . Q\ . .9 M a . n. er. 1 ...... a _ ‘1: 111.). N mu m u l t I: I illill mu ” m a \ m _ N \ ......«Km :4 I: I; .1 I | it m6 I II» I I It.ilw 21 photographs, Fig. ll. The specimens were machined from one inch dia- meter bars and the center mark was removed to provide a smooth surface for piezoelectric crystal mounting. Only one crystal was used in this experiment. However, provisions were made for crystal mounting at each end. The specimen used for Run h/1/6l was aluminum (202ii-Th) with a critical length of .150 inches and a critical section diameter of .1876 inches. The specimen used for Run 14/3/61 was annealed cold rolled steel (C1018) with a critical length of .768 inches and a critical section diameter of .1850 inches. The heat treatment for annealing the steel specimen was as follows: 115 minutes at 1600‘ F 70 minutes at 1150' r Cool in still air from 1150' 1“ (Approximately thirty minutes). A continuous load record was charted on one channel of a dual channel Sanborn stylus oscillograph. A signal was sent from a semi-conductor strain gage (gage factor of 118) mounted on a rod within the tension linkage, to the Sanborn recorder. The strain gage bridge circuit for this load cell is shown in Fig. 12. Shown in the top two arms are the active strain gage (sixty-five ohms) and a tem- perature compensating gage (sixtybfive ohms). A 5000 Oil potentiometer was placed in shunt with the compensator gage to provide a means of re—zeroing the oscillograph. The oscillograph was calibrated to give approximately three pounds per chart line and the chart was fifty lines wide. Because specimen load ranges were approximately zero to 3000 pounds, it was necessary to employ a rezoning technique. In this bridge circuit, it was necessary to put an 800 or: resistance 22 A 3282 sound 24 sensuoncofim £93m .8583 .3 menu: a 23 mobzm ocoamncoci UFO tad boon... N\ MQDQQ tbuQC MOQQm 00m.“ Qq‘Oq EmEUcooE é 02.508 5 rt .50 :oemcom/I doom awx: U not .n q/ 03% QOM\-O® 3002 8 730.86% was»? _ \ . ZQOQZVW m0 coco 956? AHOOQ .. c. / w t T as... / coco / moachzcoaoq \ cocomcoQEOo oocEUQ oucoQ EEEQQEQK monoEQEonoQ 2b in shunt with the active gage to balance the bridge. Stresses were then calculated from the load data, depending on the critical area of the specimen. Strain measurements were made with a unique extensoaeter shown in Fig. 11. This strain measuring device consists of a semi- conductor strain gage (gage factor of 118) mounted on a spring steel beam formed into a semi-circular arc of one inch radius. The beam width is one half inch and the thickness is one thirty-seconds of an inch. The beam is fastened to small Bakelite blocks which are glued to the specimen. The Bakelite blocks allow an acoustic mismatch so that the strained beam does not transmit a vibration into the specimen. The extensometer is flexible enough that it will introduce only negli- gible moments or additional stresses on the specimen. It also permits strain to be recorded without physically hampering the specimen by mounting a strain gage directly upon it. This method of measurement furnishes excellent strain data throughout the elastic and plastic ranges‘where a conventional strain gage application could not with- stand a yielding Specimen. An analysis of strain in the specimen versus strain in the beam shows the following relationship; e Where ‘b - strain in beam 2t '77; a“ lo es - strain in specimen t - beam thickness a - beam radius provided the higher order terms are neglected. The above equation shows a linear relationship. However, the higher order terms that have been dropped would theoretically introduce a small degree of 25 non-linearity. Fig. 13 shows that experimentally there is a linear relationship between strain in the beam and actual specimen strain. The curve was constructed (by actual measurement of extensometer indi- cated strain that accompanied the mechanical deflection of a call- brating frame. The continuous strain recording is nude on one channel of the dual channel Sanborn oscillograph (with continuous load recording on the other channel). The bridge circuit for the strain extensometer is shown in Fig. lb. This circuit is approximately the same as tilt for the load cell except that a fifty-five ohm resistor was added in series with the active gage to act as a desensitizer. It was found that the strain gage was too sensitive and this resistor, placed in series with it, reduces the gage factor by approxintely one half. The fifty-five ohm resistor placed in series with the compensator gage is a balance for the desensitizer. Strain measurements ranged from zero to h0,000 u. in./in. and again it was necessary to provide a means of re-zeroing the oscillograph. This was accomplist with a 10,000 olaa potentiometer. The oscillograph was calibrated to give approximately fourteen p. in./in. of strain per chart line. 26 some: 20cc Eocene c.sotm b.3855 3358. not xm oomol 00mm 00.? s 8.9 o _ _ _ _ 1 $0. 3 MGSQQ mine i no. 20: SEES osmzz one m: one/E E Z «on m n ms .49. L 0m. (U?) NO/1937j30 7VJIN7f-l33W 39:79 7W0 l 27 ombtm .on e3 dEotcdeoQ do: 0Q QGMCV \ :09 econmncogi Uta moneEoQOexm c.scoemmii .MQDQQ EDQQC ““0ng QQMIT ..va \nwlDO RN doom COOQ .9 omoo notomcoqEOo anchovwm mEEm, ZQQQZQWL % 23039: om 3N.c.sm:dmoQ ’ , mm @305 arrange. oonom 8:20p Rm.:.sw:omoQu SE Him 1&0, QB 230: one méomzmexm 230nm m I B E Dom coECon + boon CHAPTER III EXPERIMENTAL PROCEDURE The initial experimental procedure begins by eXpanding the thermal bar. The heat tube is turned on and allowed to heat the thermal bar until an elongation of approximately .30 inches is reached. This is about the maximum elongation attainable with this particular heating furnace. This amount of elongation indicates an average temperature of the thermal bar of approximately b50° F. As the bar expands, the adjustable head on the Riehle testing machine is moved down to maintain a predetermined tension on the specimen. (The reader may wish to refer to Fig. 1.) When the full expansion of the thermal bar is reached, the testing machine and heat tube are turned off. The contraction of the heated bar being cooled now begins to produce a load on the specimen. The actual test begins at this point and the tape recorder is turned on. The acoustic signal is being viewed on the oscilloscope (see Fig. 20) and simultaneously recorded on an even channel of the tape recorder. The loads produced may be observed at any time directly from the load indication dial of the testing machine. The laboratory is in complete silence, with the exception of these loads being monitored into an odd channel of the tape recorder every minute. Occasionally voice comments are added on the voice channel in the event of an interesting occurrence. Normally, thirty-five to forty minutes of natural cooling produces sufficient load (approxi- mately .15 inches of contraction) to put any Specimen well into the plastic region. The tape recorder is stepped and the testing machine is turned on when yielding is so great that fear of fracture is present. Specimen fracture is avoided to preserve expensive apparatus. There 28 29 is now a permanent record of acoustic emissions that may be replayed at any convenient time for careful inspection. Stress-strain data is available from the Sanborn recording which has been synchronized with the tape recorder by a remote timing switch. At the precise time the minute interval loads are monitored on the tape recorder, a mark is placed on the oscillograph chart. This easily enables one to correlate stress-strain data with acoustic data. The smallest acoustic signals, Judged detectable, are two times amplifier noise. This is attained by trigger level setting on the oscillosc0pe. The tape recorder records everything that comes through the SKL filter. The filtering frequency on the SKL is variable, and its choice of setting depends a good deal on acoustic experience with various Specimen materials. Frequencies below 2000 cycles per second were filtered for this particular experiment. In general, one must gain experience and become quite familiar with acoustic emission and the electronic apparatus to thoroughly understand all of the range settings. CHAPTER IV EBBSENIATION‘QE EXPERIMENTAL.RESULTS Significant results are presented in the form.of stress- strain curves superimposed on acoustic emission histograms. The re- sults of the 202h-Th aluminum specimen (Run_h/l/6l) are shown in Fig. 15, and the annealed C1018 steel specimen (Run h/3/61) in Fig. 16. The stress-strain diagrams are self-explanatory. However, to under- stand the histograms, some explanation may be required. The stress- strain curve and the histogram share a common abscissa, which is strain. Keeping in.mind that the loads are recorded in one minute intervals, the specimen strain is also related to time (via the stress- strain curve). The electronic counter is arranged to count the total number of acoustic emissions per minute (per load recording interval) that are greater than two times noise. The electronic counter is arranged so that it gives one square wave signal (to be counted) per oscilloscope sweep and the oscilloscope is arranged so that one acoustic emission will occur in one sweep length. Assuming, then, that only one acoustic emission occurs during a sweep and that the emission is greater than two times noise, all emissions are counted. This is not so however, and it must be kept in mind that some emissions are missed (Those that are less than two times noise and those that occur more than one per sweep). The acoustic emission.counts per minute, or per strain interval corresponding to a minute, are then plotted as the ordinate in the histogram. The steps in the histogram represent one minute intervals and are represented as strain intervals corres- ponding to one minute on the abscissa. This method of plotting is very descriptive because it represents relative acoustic emissions as 30 31 m.:..\.:.e «C 50.: m — - 11. m......... ... .. ............. n. w. ...d1 .... . . .fi . .. d .. .. WW0 IIIIIIO s............ ......w . V 3 ......filsll. 1. L m h Liomw :9 m 3 4% so % 0m 1 m. .m a ._ i Q. _ com mo V g . 4 Om. IOO :dECon 9 mQDOQ E3:.eE3< QKLVWON \®\\\¢ ESQ (15») $561.25 32 ~356de \ .2 <53 ES #0 QQQON .....ius. . ...fl 3. u . r. L‘. ..., . .. .. ..L.)_ . . . h r ... r Q MQDUE 63m SEQ $05 osmosis ERE Ewe 98 OQQU\ . ...} ~ . n. . i ' O~ M&.. . v 3 ' l OHSHODV 0 c3 1 [d4 III J 00m... 0...... 9. C._ .r... S . O. U ). 06m 110m 33 they occur on the stress strain diagram. The portions of the histo- gramnwhere counts are not recorded are merely portions where data was not available because changes in recording tape track were made at this time. Although one cannot be certain, one is relatively sure that there is no outstanding phenomenon occuring in these void areas because of the trend of the plot. The histogram shown in Fig. 17 is plotted with time in min- utes as the abscissa and again acoustic emission counts per minute as the ordinate. The data for this run was obtained in exactly the same ‘wmy as‘was the data for Runs h/1/6l and h/3/6l. Strain measuring instrumentation was not available at the time this test was made. How- ever, the time from test start may be correlated with specimen loads. The counts that disappear off the scale of the plot continue to a magnitude of 700 to 900 acoustic emissions per minute. This histogram appears to be more descriptive plotted to this scale rather than one that ranges from zero to 900. The specimen for Run 12/23/60 was steel that had been shot peened over approximately one and seven sixty-fourths of an inch of its critical length. The critical length 'was two inches and the critical diameter was one quarter inch. A relative description of the acoustic emission pulses are presented in a somewhat different form in Figs. 18 and 19. These oscillograph recordings are for the aluminum 202h~Th specimen (Fig. 18) and for the annealed steel 01018 specimen (Fig. 19). These charts were obtained by oscillograph recordings of the acoustic emission signals as seen by a root-meanrsquare voltmeter. The baseline of the chart represents the relative noise magnitude and the peaks or pulses represent relative acoustic emission voltage signals. Because of the J 1. .01 in ....A r.‘ .3... . it...“ we... ... e.-- ...... a, ’7. w .. Fr .-.i_... r a . is ...... ... 3J1. I ...—I 1 v . hawk pnuwoef (... . .sy . . . exfi. lan.‘ J .. . hfnr M.» E. ... w (9!. -. . .. .. .. . .1 .. - . .I i r o-.0 ‘_ ...!I L a)! , . . .- a . . . . Ill-HI . I . I u . . . J . LI. 1.. I. .Q...) .‘Ww- anflKi .n o .Jloo. I E...) INN”? if. I... .2 ‘5... o f. . )2 .... .honI» ..r viii-[r f.t.. . ..5. a r....&. s... \mf.l.n . .I .. .. v... of. o _ -..»... - . .. . is. .. , .. ...... 1H.~ . . . vy . AWN! ..At‘fi . .43. a! )- s. ......... J 3W1». I'. ermiir . . ... - . . . . ‘ .mlnn... . . M- .doa. 1...... s . 1 I. a u m .. ... .. .4 . . s .W. ”(a .. . .. .I- ..x . W . . . ‘9 I. D. . . . .’.. v u. " 5. .. ’4 -.*.J is‘ .s 2 s .. .. a. o LIL: I. .bll [I n n .It 4 o... . n 5.1 (I \o v. . . . . . p .. . oh. — . ....» . .I ..l .... .... .. . . . . It . . . a .. 3‘04? .o . ‘0‘“ o. . a...“ A..- o .. .....n..............w.. no“ .-.... ...”. .2. . .. ~... .7. v f . .. . .NI .1 D... r . ’1‘ as r o. ... 71.. . .rx6......fl...!1.._._.