8/10/2019 1954_Gunn_Ultrasonic Defect Detection In

1/7

34

ULTR SONI DEFE T DETE TION IN SOLID SH FTS N ELE TRONI LO D ME SUREMENTS

y ] R. GUNN

To say that there is nothing new under the sun isan adage which is usually very accurate when oneanalyses most modern inventions. I will attempt todeal with two very recent acquisitions in the field ofengineering science and as we progress you w ll seethat although the equipment is highly complicated,it is basically only doing what has been done formany centuries, bu t of course; in a far more accuratemanner.

T ULTR SONI DEFE T DETE TOR

The usual method of testing certain metals fordefects is to give whatever is to be tested a sharpblow or ta p with a hammer and listen to the ring orsound emitted. This is done by the carriage andwagon examiner who taps wheels on the railways, byfitters working on locomotives who deliver a mightyblow with a four pound hammer on the crankpins ofthe driving wheels to endeavour to establish whether a th e pins are loose or b whether they are flawed.These audible sounds vary in frequency from 16cycles or vibrations per second to 16,000 cycles persecond, the higher the note the higher the frequency.I t is safe to say that in most cases the defect isrevealed by a dull note instead of a sha rp note, andthat this note is usually of lower frequency than thatemitted by a sound unflawed specimen. I t is aknown fact that every object when struck by anotherobject will tend to set up vibrations of a frequencypeculiar to that specific object, dependant on thenature of material of the object and on its shape andsize. I f the vibrations are audible the frequency mustfall between 16 and 16,000 cycles per second. I f theobject is defective, the structure of the material willbe different from the original, and this will cause adifferent frequency, usually lower, and the testerwill report that he has discovered a defect, and, innine cases out of ten, th e object will be discardedwithout knowing what, where or to what extent thedefect exists. In some cases this may be verywasteful and uneconomical.

ith the advancement of science some moreaccurate method of testing of expensive machinerywas evolved to meet the requirements of safety andeconomy. There are such testing apparatus as X-ray,magnetic crack detector and ultrasonic testing. Thelatter device is basically the same as the wheeltapping method except that the frequency of vibrations set up in the shaft are far beyond the audiblerange.

The apparatus consists of an electronic impulse

generator which can be tuned to predetermined

frequencies of t megacycles per second (or 500,000cycles per second) to 21 megacycles per second(2,250;000 cycles per second). Impulses at the abovefrequency are fed to a tuned quartz crystal, which isheld in contact with a smooth surface on the materialto be tested. These impulses are t ransmi tted to theobject for a specified time interval and then cut off.In one manufacturer s apparatus this space of time is1 microseconds or 1/400,000th of a second. Echoesreflected back from various parts of the material tobe tested strike the quartz crystal, re-energise it andanother section of the electronic apparatus picks up

and correlates the echo impulses with the originalsignal impulse. In order that the results may be.visually interpreted the information is fed to anoscillograph or cathode ray tube. This is effected inthe following manner.

The thyratons which cut off the high frequencyimpulses to the quartz crystals also influence a visualhorizontal line on the oscillograph screen, so thatinstead of it being a straight line it is in the formshown in Figure 1

- - _

IG

Each dash in the line represents a definite intervalof time. As the ultrasonic impulses travel with adefinite speed through the material, then, as wetrave l from left to right across the screen or timelines and multiply the time represented by thedashes by the speed of transmission of the vibrations,we can establish a specific distance. To quote a moreeasily conceivable case, if a car travels for 2 hours at30 miles per hour, the distance travelled will be

2 X 30 or 60 miles. In the case of the ultrasonic

8/10/2019 1954_Gunn_Ultrasonic Defect Detection In

2/7

8/10/2019 1954_Gunn_Ultrasonic Defect Detection In

3/7

36

t _ _ rc

IG S

e

deflection will reveal itself on the dial at a pointwhere the geometry of the shaft tells the operatorthat there should be no echo. Whereas testing fromthe other side of the shaft does not reveal the samechanges in geometry a crack will show up at the samespot in the shaft bu t obviously when tested from

the other side the distance will be read off on the dialat a point of L A from the end.

Having located a crack we must now investigate itto ascertain whether the shaft must be condemned ornot. This is done with an angle beam which emitsbeamed sound at 45 to surface of the tes ting head.Reference to figure 8 showing an enlarged port ionof the shaft shows how the angle beam is used.

__ 1 r

IG 6.

on the right and testing from that end the sound willtravel from right to left it is nevertheless recordedfrom left to right as the instument is only interested.in the time between the emitted sound and the echoand not in the direction it travels. t is to be noticedthat certain fillets and changes in section are pickedup in one direction and not in the other. This is dueto there being no reflecting surfaces in one directionfor instance from the pintle end the rays will passthrough the outside end of the bearing journal andas these do not leave the shaft they willnot meet anyreflecting surface. t is for this reason that thegeometry of the shaft must be known. Assumingthere is a crack in the shaft under the shell as shownin figure 7 the sound rays will be reflected from thecrack as well as the fillets etc. and a vertical

FIG

ith the angle beam if no fillet is encounteredand no crack is present no echo is obtained. thereis a crack in the shaft and its position is located withthe direct longtitudinal beam then by sighting theangle beam to hit the crack a reflected echo will bereceived at the transmitting head. y moving theangle beam longitudinally backwards and forwardson the shaft a distance on the shaft can be markedoff where echoes are received. By moving the headradially at the same distances as those marked thecrack can be traced round the shaft. By suitablyinterpreting the area on the shaft on which echoesare received the depth and radial length of thecrack can be calculated by simple trigonometrical orgeometrical methods. the crack is under the shella direct reflection ma y not be obtained due to the

8/10/2019 1954_Gunn_Ultrasonic Defect Detection In

4/7

37

shell interfering with the testing head as shown infigure 8 bu t this does not prevent testing from beingcarried out on the other side of the shaft as indicated.

I have omi tted to state that any ultrasonic wavesleaving the skin of the objec t being tes ted do no t

effect the dial indications at all. Thus any soundwaves leaving the steel shaft and entering the castiron shell do no t return and therefore the Cast ironshell does not influence the testing in any manner at ~ except that it gets in the way if a crack is found.

Anything welded to the shaf t will influence theresults very materially since welding is the actualfusing of one object to another so that ultrasonicallythe two objects are th e same mass of material andreflections will be picked up from the welding fillets.In this instance it is interesting to state that oneshaft at Tongaat came under suspicion of beingcracked until the operator of the ultrasonic testerlocated the crack on his screen measured back onthe shaft an d found a juice guard tacked to the shaftby welding. So accurate was his machine that byrunning the testing head radially round the end ofthe shaft he told me by interpreting the readings onthe screen how many tacks there were on the juiceguard and exactly where they were radially roundthe shaft circumference.

enefi ts der ived from Ultrasonic Testing

The benefit of being able to test anything withoutdestroying it as in a tensile test is obvious. is a

source of great satisfaction to me to k now t hathaving had all the shafts of two tandems ultrasonictested I have 50 mill roll shafts free of flaws andcracks. Had this testing been done during 1953 itmay have avoided the inconvenience and expense ofhaving two roller shafts fractured in operation. Wehave found by bitter experience that the crankpinsof slow speed horizontal mill engines are prone tofailure an d all these were tested. I believe that atIllovo Sugar state when testing axles of the railwaytrucks approximate ly 10 per cent. were found tohave been cracked just inside the wheel bosses.

During my period of service in the South AfricanRailways at Cape Town the Class 23 locomotivesbegan to suffer from severe breakages of driving andcoupled wheel axles as well as side rods. Thisbecame so chronic that special instructions wereissued that the life of axles on that particular classof locomotive was to be reduced. This mean t thatwhen a certain period had elapsed the axles were tobe discarded irrespective of their condition. This wasa very costly business. Now the problem has beensolved by ultrasonic. testing and good axles aregiven a fur the r lease of life the reby keeping thelocomotives in service longer and saving the expenseof a wheel change which involves 10 days solid

work.

To quote a case of saving at Tongaat we discovered that an impor ted shaf t which has had fourreshells gave the clearest screening of all 50 shaftsand was declared the best shaft at th e mill. ha dbeen pu t aside as a risky shaft due to its age bu t nowis a certainty to be reshelled for the fifth time.Another shaft from the 84 tandem was discoveredto have been cracked when it was being dressed upfor a reshell at an engineering firm in Durban. Thecrack was machined with a round nosed tool and itpersisted down to about depth so the shaft wasabandoned. During the ultrasonic testing the depthof the crack was measured with an angle beam andwas found to be only another deeper. As thejournal size of the shaft is less and the shell seat bymuch more than the depth of the crack it is proposedto turn ou t the crack and have the shaft reshelled.In itself that represents a saving of approximately600.

While ultrasonic testing is a comparatively simpleoperation it requires skilled operation and interpretation of the dial readings. is not considered that aset purchased by say th e Sugar Milling ResearchInstitute or an individual sugar company would bean economical purchase as the equipment is costly.On the other hand provided sufficient companiescan agree to having rolls tes ted at certain t imes sothat the operator can travel from one mill to theother the cost of having the services of the SouthAfrican Bureau of Standards is very reasonable.

lec tronic Load or Strain Measurements

To go back in time again we find in heavy engineering and also in comparatively modern aeroengineering that in many instances the loads appliedto various parts of machinery were gauged by theextension or lengthening of premeasured bolts andcotter pins. To quote two examples in locomotivefitting procedure in order to ensure that the pistonrod is t ight in the cross head the cotter pin is drivenin lightly and then marked it is then driven in afurther on the taper this compressing the taperpin and loading the piston rod so that it is rammed

tightly into the crosshead. In aero-engineering thebolt clamping the one web of a Bristol radial engine iscoupled up and its length is measured with a micrometer. is then t ightened up until it had stretched.060 inches and the f it ter is then assured that thecrank web is t ightened up to the required amount orin other words that the bolts have been sufficeintlyloaded.

is common knowledge that when any material isstretched or compressed the cross section area of thebody either becomes less or more as the case may be.This change in area follows a direc t relationship tothe tension or compression in the body and is afunction of Poissons Ratio. Working on the theory

8/10/2019 1954_Gunn_Ultrasonic Defect Detection In

5/7

STRAIN i

lO

that an increase in tensile loading will cause increasein length of a body with corresponding reduction inarea, the electronic engineers have evolved a methodof analysing strains by making use of the physicalproperties of cross section area changes coupled withits effect on elect rical resistance. Assume we havean accurately rolled piece of resistance wire of knownYoung s Modulus and Poisson s Rat io , if we s tretchth e wire there will be the attendant reduction incross sectional area. The electrical resistance of thewire will depend on the l ength of it and on its crosssection area. it is stretched, th e length is increasedand the area is reduced, the latter increasing theelectrical resistance and the former further increasingit as the wire is now longer. the wire is arranged asshown in figure 9 the effect of s tretch ing the wire isfurther increased.

38

I O V I V

The amount of change of resistance is microscopicbu t that does no t bother the electronic expertbecause he uses thermionic amplifier valves tomagnify the results into measurable quantit ies.

Nex t feature employed in load or st rain analysisis t he standard Wheatstone Bridge, which is knownto anyone who has studied basic physics or electricalscience. Figure 10 shows the circuit employed.

An alterna ting current of nov. at 100 cycles persecond is fed to the two ends of the bridge systemwhich comprises in a very simple form of twoadjustable resistance R I and R 2 These are madeadjust ab le to enable various ranges of testi ng to beperformed and St and Sf, which are the strain gaugewhich does the testing, an d the strain gauge elementwhich is not subjec ted to any strain bu t is used in a

fixed capac it y to balance the zero posit ion of theinstrument. The connection between R I and R 2 andbetween St and Sf are connect ed to the electronicamplifier for magnification and interpretation.

th e resistance values of H I St, then there willR 2 Sf

be no cu rr en t flowing through the bridge and thebridge will be in balance. To pu t it in another formSt R I X Sf and as RI> R 2 and Sf axe fixed, once

R2th e instrument has been zeroed, t he n a ny change in

St will cause the equat ion to be upset and thereforecurrent will flow to the electronic amplifier.

flI1PLIFIR

FIG 10

ethod tes ting for Strains and oadingBy preliminary calculations or by knowledge of the

approximate loading, th e correct ratio of RI> R 2 ,St and Sf is selected. The s train gauge St, which is asmall plastic instrument about 2 long X i wide, isfixed to the object to be tested by means of a culluloseadhesive at each end of it, so that if the objectstretches the strain gauge will stretch with it. Inorder that t empera ture will not affect the accuracyof the test, strain gauge Sf is fixed adjacent to St,bu t on this occasion Sf is only glued at one end sothat any stretching of the tes t piece will not effect Sf.The electrical wiring is completed and the instumentis brought to zero by adjusting rheostats an d po-ten tiometers to counterac t for length of leads, etc.

When the load is applied to the test piece it wille ither be compressed or lengthened a microscopicamount, and the information passed to the amplifierwill be detected, amplified and reflected on a dialfor a st at ic type of tester, or reproduced in a form ofa graph on sensatised paper for a dynamic testerwhich will record a continuous st ra in analysis of asequence of events, such as the Strains tak ing placein a piston rod or connecting rod of an engine.

r om t he reading of th e stat ic dial type or th echart of the dynamic tester, we can evaluate the

8/10/2019 1954_Gunn_Ultrasonic Defect Detection In

6/7

39

actual strain or elongation or compression which th estrain gauge has suffered. It is necessary to knowthe Young s Modulus of the material of the test piecein order to calculate t he actua l stress. This is doneby multiplying th e strain gauge factor by theYoung s Modulus. To obtain the actual loading th estress has to be mult iplied by the cross section areaof th e test piece.

I have referred to test piece mainly for want ofa better word. The test piece is actual machinery inmotion an not a laboratory test piece. Let us nowconsider where we can use this equipment in a sugarmill and factory. Fo r the practical engineer, there isnot very much use for it as the correct results requirequi te involved mathematics to be used and as th eengineer has to do the best with the machinery he hasand knowing th e stresses set up in individualcomponents do not make much difference to its

operation. However, the designer or manufacturerof th e machinery could derive great benefit from itsuse.

Take for example, th e case of mill headstocksalone. We have a firm in Glasgow supplying millswith hydraulic rams th e same size on the gearingside and on the pintle side, and the same supplierfurnishing other seemingly identical mills with largerrams on the gearing side to balance th e climbingeffect of the mill pinions thrusting away from eachother. By using strain gauges on the vertical cheeksof t he t op roller bearing guides on both sides of themill, the difference in loading on each side headstockcan be evaluated. There is another firm of millsuppliers which has th e cap inclined at an angle. Inorder to establish whether the angle is correct, straingauges placed at varying angles on th e headstocknear the top, will establish which angle gives themaximum reading and therefore at which angle thecap should be inclined. The loading on mill rollersand the effects of mill settings can be established bymeasuring th e compression of the actual mill rollerbearings. Most text books on mill extract ion seemto be indifferent regarding th e setting of th e trashplates, but a strain gauge analysis of th e loadingtaken by th e trashplate could easily be obtained byfixing strain gauges to th e dumb t urne r and to thedumb turner draw bolts. Fo r th e designer or researchengineer, invaluable information can be obtained byth e use of strain analysis.

The Railways Administration carried out a verythorough investigation into the stresses set up in alocomotive under load, pulling up a heavy gradient.In this case the s train gauges were affixed to th estrategic points on the steel frame, on th e couplingrods, connecting rods, slide bars and various otherpoints. TheIocomotive was tested after a completeoverhaul with everything tight and properly adjustedan d the stresses set up were calculated. Variousbushes were then turned down to represent con-

ditions of wear and the same series of tests werecarried out. The results were so impressive that anew code of maximum wear clearances was introduced as a result. A further series of tests werecarried out to ascertain th e loading when coupledwheels took again immediately after slipping. Theresults of these tests revealed abnormally heavyloading indeed. The strain gauges have been used totest such things as hammer blow on rails and impactloads on bridges.

I regret that I cannot refer you to any books onthe two instruments which I have discussed. Fo r th einformation regarding ultrasonic testing I indebted to Mr. B. Zenzinger of th e South African Bureauof Standards, who lent me Sperry s Technical Databooklet No. 50 - 755. My information regardingstrain gauges was obtained from participating in th eRailway testing of the locomotives.

Mr Grant th e Chairman, said that the paper wasof most useful practical value. He asked Mr. Gunnif old-age crystallisation would not deter him ofre-shelling an old shaft five times. Mr. Grant saidthat he understood that arrangements were beingconsidered for the Bureau of Standards to visitfactories to test shafts.

Mr Gunn said that he did not think that therewould be any difficulty in re-shelling th e shaft afifth time, because the ultrasonic testing, to a certain

degree, shewed th e crystalline state of th e shaft.He mentioned a case of another locally-producedshaft which was full of blow holes and inclusionfailures which would be more risky to re-shell.

Mr Walsh asked Mr. Gunn if he did not think. that customers should demand a tes t on new rollershafts, especially in this country. Defects couldinvolve firms in considerable expense.

Mr Gunn said that that was a very importantpoint because he knew of two such cases during thispast year where local shafts had proved to bedefective. The Bureau of Standards, situated in

Pretoria, was in a good position to test all locallyproduced shafts before they were sent down to th esugar belt.

Mr Heslop asked what th e range of metals whichcould be tested would be.

Mr Gunn replied that as far as he knew an ymetal could be tested. Different frequencies wererequired for different materials. As far as size wasconcerned the machine available in South Africawould test up to fifty-four feet.

Mr Lindemann inquired if the machine could beused on boilers to test for cracks in plates and thelike.

8/10/2019 1954_Gunn_Ultrasonic Defect Detection In

7/7

40

Mr Guoo said th t there were other testersavailable r ther th n an ultrasonic one. They h da machine which for instance could measure fatiguein rivets. was not an ultrasonic type bu t a

.magnetic one.

Mr Rault said th t we should not wait for theBureau of Standards to do our testing. He wonderedif therefore it could not be done within the indus tryor some local organisation r ther th n wait for theBureau of Standards. There were three hundred ormore rollers requiring testing in the industry andthe financial loss occasioned by breakdowns waslarge.

Mr Guoo pointed out th t not all mechanicalbreakdowns could be avoided by using the ultra -sonic tester. He h d been told th t any companycould hire testing time from the South AfricanBureau of Standards up to one hundred and fiftyhours. This he thought would be a more.economicalproject th n buying their own equipment. Suchtest ing would have to be done all t the same time

nu ry or February. r ouwes ekker said th t he h d had a dis-

cussion with the Bureau of Standards on the. possibility of the Sugar Milling Research Institute

having a contracting period of about one hundred-and fifty hours for the whole sugar industry. Heplanned to pursue the m tter further.