Research - Ultrasonic Hull Thickness Measurements

An ultrasonic thickness (UT) gauge can be used to measure thickness of metal and many other solid materials so this instrument can be used to ... Home Products SiteRecorder MagnetometerBook Ordering CaseStudies MaryRose FireshipFirebrand Stella1RomanBoat TheBarrelWreck Coronation ResurgamSubmarine TrescoChannel A7Submarine TheSHIPSProject JamesEaganLayne HMSWhiting More... Resources Downloads TechniquesGuide -StartingaProject -UnderwaterSurveying -Tools Research -RNShipwrecksDB Publications Training Company Aboutus Projects Clients BooksforSale Techniques>Research>UltrasonicHullThicknessMeasurements Research-UltrasonicHullThicknessMeasurements Versiondate:16October2016 Introduction Anultrasonicthickness(UT)gaugecanbeusedtomeasurethicknessofmetalandmanyothersolidmaterialssothisinstrumentcanbeusedtomeasurethethicknessofmetalremaininginironandshiphullstructureandplating.BackgroundinformationonmakinghullthicknessmeasurementswithaUTgaugecanbefoundinTechniques:UltrasonicThicknessMeasurements. TheuseofUTgaugesformeasuringplatingthicknessonfloatingvesselsiscommonandtheexpectedresultsarewellknown.However,publicationsthatreferredtotheuseofultrasonicthicknessmeasurementsonshipwrecksoftenreportedthatproblemsoccurredwhenmakingthemeasurements.ThesetestsweredesignedtoinvestigatetheseproblemsandtohelpformulategoodpracticefortheuseofUTgaugesinunderwaterculturalheritageprojects. SponsorshipwasobtainedthroughtheUScharityresearchfoundationProMareandCygnusInstrumentstopurchaseaCygnusDIVEIIgaugeandasetofprobesforthisinvestigation. Fig1:Measuringhullplatingthicknessunderwater Fig2:TheCygnusDIVEIIunderwaterultrasonicthicknessgauge AimsandObjectives Theaimsoftheexperimentswere: Test1: Todeterminetheeffectsofdifferenttypesofultrasonicthicknessprobe Test2: Todeterminetheeffectofdifferentmetalsurfacepreparationmethods Test3: Todevelopanoperatingmethodology Test4: Todeterminemeasurementqualityontypicalcorrodedmetalsurfaces Test5: Toexperimentwiththeuseoflowercostepoxyputtyforfillinginconcretionholesattestpoints ExperimentsontheuseoftheUTgaugewereundertakenonlandusingironandsteeltestpieces,ontheforeshorewreckofaWWIIsteelammunitionbargeandonthehullofHMsubmarineA7. EquipmentandMethod TheinstrumentchosenforthisprojectwastheDIVEMk2underwaterultrasonicdigitalthicknessgaugemadebyCygnusInstrumentsLtd.,thisinstrumentisrobust,smallandlightweightsocanbewornonthediver’swrist(Fig.2). Thegaugeisautomaticinoperationsotherearenocontrolstoadjustbeforemakingameasurement,thisisanadvantagewhentimeonsiteisshortandalsomakestheinstrumentveryeasytouse.  Thegaugecanbefittedwithdifferenttypesofprobesoweundertookexperimentstodeterminetheeffectivenessofeachtypeoncorrodedplating. Thegaugedisplayshowntotheoperatoriseasilyunderstoodandprovidesessentialfeedbackaboutthequalityoftheultrasonicsignalsbeingreceivedandthemeasurementsthathavebeenmade. Anoptionalupgradetothebasicgaugeistheabilitytostoreupto5000thicknessmeasurementsinitsinternalmemory,arrangedingroupsselectedbytheoperator,themeasurementsareautomaticallyloggedwhenastablemeasurementismadesotheoperationishands-free.  OncethemeasurementshavebeenmadetheloggedmeasurementscanbeuploadedintoacomputerfordisplayandannotationusingtheCygnusInstrumentsCygLinksoftwarethatissuppliedwiththegauge. OnceanyannotationshavebeenaddedtheCygLinksoftwarecanbeusedtoproduceareportinPDFformatorthemeasurementscanbeexportedtoaspreadsheet. Test1-ChoiceofProbeType Method Threeprobesweretestedinthefollowingexperiments,onesinglecrystalprobeandtwotwincrystalprobesthattransmitatdifferentfrequencies.Thenumberofcrystalsfittedintotheprobedeterminesitsmoreofoperationandthefrequencydetermineshowthickamaterialcanbemeasuredandtheresolutionofthemeasureements.Eachtypeofprobeisfittedwithaplasticbandofaparticularcoloursoitiseasytotellthemapart: Probe1:S2C Singlecrystal2.25MHz,standardprobe,redband,3-250mmthicksteel Probe2:T5B Twincrystal5.0MHz,forheavilycorrodedsteels,yellowband,1.5-50mmthicksteel Probe3:T2C Twincrystal2.0MHz,Castiron,plastics,blueband,2.5-150mmthicksteel Fig3:Singleandmultipleechomodes Thesinglecrystalprobeisthetraditionaldesignthatworksin‘multipleecho’mode(Fig.3).Formultipleechomeasurementstheultrasoundpulseisfirsttransmittedfromtheprobeintothematerialundertest.Thetransmittedpulsereflectsoffthebackwallofthematerial,returnstothefrontfaceofthematerial,isdetectedbytheprobeandtheclockwithinthegaugeisturnedon. Thereturnedechothenreflectsoffthefrontface,againreflectsoffthebackwallwhereitisdetectedagainandthetimeelapsedisrecordedinthememoryoftheinstrument.Thereturnedpulsereflectsoffthebackwallforathirdtimeandisagaindetectedbytheprobe,simultaneouslythegaugebeginstotimethethirdecho.Withthismethodtheprobedetectsthreeequally-spacedreturnechoesfromasinglepulsetransmissionwherethetimespacingoftheechoesisproportionaltothematerialthickness. Thestrongestreflectionsoccurwithinthetestmaterialsothismethodwillignorecoatingthicknessandtheneedtodetectthreepulsesprovidesanin-builtmeansofqualitycontrolforthemeasurements. Thedrawbackofthismethodisthatalargeamountoftransmittedsignalhastobecoupledintothematerial,asignallevelthatissufficienttoovercometheabsorptionandscatteringthatoccursateachreflectionandwithinthetestmaterialitself. Thealternativeisatwincrystalprobethatusestwocrystaltransducerelements,onetotransmitthesignalandasecondreceivercrystalthatisfittedalongsideitintheprobe. Withthistypeofprobeathicknessmeasurementcanbemadewithasingletransmissionandbackwallecho(Fig.3).Thetransmitpulsefromtheprobepassesintothetestmaterial,reflectsoffthebackwallandisdetectedbackattheprobe,withthetimeofflightofthesignalproportionaltothethicknessofthatmaterial. Asasingleechoisusedtomakethemeasurementthereisnoinherentqualitycontrolofthemeasurementsasthethicknessofanycoatingswillbeincludedinthemeasurement,sothismethodshouldonlybeusedonbaremetal. However,onlyasinglereflectionisrequiredtomakeameasurementsolesspowerhastobecoupledtothetestmaterialforameasurementtobemade. Thismethodisacousticallymoresimplethanthemultiple-echomethod,ismorelikelytoworkundermarginalconditionsbutprovidesnobuilt-inqualitycontrol. Theprobesusedfortheteststransmitteddifferentacousticfrequencies. Highfrequencypulsestravelashorterdistancethroughthetestmaterialthanlowfrequencypulsesbuttheresolutionofthehighfrequencymeasurementsisfiner,solowerfrequencieswillworkbetteronthickermaterialsbutwillgivelessprecisemeasurements. Forplatethicknessmeasurementswherethedistancestobemeasuredaresmallwewouldtendtousethehigher(5.0MHz)frequencyprobebutthelower(2.0MHz)probewastriedaswelltoseeifitwouldcouplethetransmitpulsestothetestmaterialmorereadily. Forbothprobesthereportedresolutionofthemeasurementsis±0.05mmontheHighsettingand±0.1mmontheLowsettinginthegauge. Fig4:Makingtestmeasurements Ultrasonicthicknesstestingforheritageapplicationsdiffersfromtypicalapplicationsofthistechniqueinthatmeasurementsareattemptedonironandsteelwithsurfacesthatmaybeheavilycorroded. Thereareanumberofproblemsthatoccurwhenmeasuringthethicknessofcorrodedmetal. Oneofthemostsignificantproblemsisthedifficultyincouplingortransferringthetransmittedacousticsignalsfromoneobjecttoanother,suchasfromtheprobetothetestmaterial. Attheinterfacebetweentheprobeandthematerialsomeoftheacousticsignalwillbeabsorbedorscattered,sohowwelltheprobeiscoupledtothetestmaterialmakesabigdifferencetotheabilitytomakeameasurement. Acouplantisneededbetweentheprobeandthetestmaterialwhichthewateritselfprovideswhenunderwaterandagelisusedonthesurface,sothisformspartoftheacousticpathbetweenprobeandmaterial. Themetalsurfacetheprobetouchesmaybeunevenorpittedwhichwillreducetheamountofacousticsignalthatcanbecoupledintothetestmaterial. Thebackfaceofthematerialmayalsobeunevensomuchoftheacousticsignalmaybescatteredratherthanreflectedback. Apoorcouplingbetweenprobeandmaterialalsorequiresthereturnechotobestrongersothatenoughsignalstrengthcancouplebackintotheprobetobedetected. Theoldestmaterialsprovidethebiggestdifficultiesasthematerialbeingmeasuredmaycontainsolittlesoundmetalthatitistoothintobemeasuredreliably. Thehigherresolution5MHztwincrystalprobecannotmeasuremetalthinnerthan1.5mm. Anumberofdifferenttestswererunusingthethreeprobeswhichaimedtoevaluatetheiruseonarangeoftypicalironandsteelhulls,theresultsoftheteststodeterminethequalityofthemeasurementsaredescribedbelowinTest4. Theotheraspectconsideredwastheeffectivenessofeachprobeinactuallymakingameasurementasitsoonbecameapparentthatoncorrodedmetalsurfacesitwasdifficulttoobtainanymeasurementsusingthesinglecrystalprobeandthemultipleechomethod. Arangeofmetaltestpiecesweretriedthathadarangeofagesanddegreesofcorrosion(Fig.4). Asexpected,allthreeprobesprovidedreliablemeasurementsonclean,smoothandcorrosion-freemetal. Onpittedsteelthetwincrystalprobes(probes2&3)wouldgiveameasurementquicklywherethesinglecrystalprobe(probe1)wouldneedtobecoaxedintomakingameasurementbymovingtheprobearoundthemetalsurfaceuntilasuitableplacewasfound. Thismeantthatoftenalargerareaofmetalhadtobecleanedbeforeameasurementcouldbemadeusingprobe1,causingmoredisturbancetotheconcretionthanwasdesired. Onamaterialwithmoreadvancedcorrosionthetwincrystalprobewouldprovideameasurementwhenthesinglecrystalwouldnot. Onthethinnestmetalthehigherfrequencytwincrystalprobe(probe2)wasmorereliablethanthelowerfrequency(probe3). Theresultsofthesetestsareinagreementwiththepreviouspredictionbasedonthelimitationsofeachprobeanditsmethodofmakingameasurement. Test1Conclusions Althoughthesinglecrystalprobe(probe1)couldbeusedtomakethicknessmeasurementsofsomeofthepartlycorrodedmetaltestpiecesthetwincrystalprobes(probes2&3)couldbeusedonmetalthatwasinworsecondition. Thetwincrystalprobesoftenprovidedmeasurementsmorequicklyanddirectlythanthesinglecrystalprobe;asignificantpointwhenworkingindeepwaterwheretimeonsiteisatapremium. Onlyasmallareaofconcretionneededtoberemovedforthetwincrystalprobeswhereasalargerareawasoftenneededforthesinglecrystalbeforeitmadeasuccessfulmeasurement. Test2-SurfacePreparationMethods Fig5:Samplehullplatingshowingthingreyconcretionoververypittedandcorrodedsteelplate Method Anidealmethodofmeasuringplatehullthicknesswouldinvolvenopreparationofthematerialbeforemeasurementsothatthematerialwouldnothavetobedisturbedandnovaluabledivetimewouldbespentinpreparation. Ifsurfacepreparationmustbedonethenitneedstotakeaslittletimeaspossibleanditshouldalsonotrequiretheuseofcomplicatedorexpensivetoolsortoolsthatrequireextensiveoperatortraining. Previousworkhasshownthatanysurfaceconcretionlayerhastoberemovedfromthetestareaanditbecleaneddowntobaremetal. Russelletal.evensuggestthatpreparationshouldincludebuffingandgrindingtomakethesurfaceassmoothaspossible,howeverthiswouldbeunwelcomeasitwouldrequiretheuseofpowertoolsunderwaterandwouldremovesomeofthethicknessofmaterialthatwasbeingmeasured. DuringworkonthesubmarineAE2in86mwaterdepthapneumaticneedlegunwasusedtoremoveconcretion,alogisticallycomplicatedmethodthatrequiredanoperatortrainedintheuseofthetool,equipmenttopowerthetoolplustheproblemofmanaginglonghosesindeepwater.Thesesurfacepreparationtestsweredesignedtoestablishhowlittlepreparationcouldbedonewhilestillbeingabletomakeuseablemeasurements.  AllthreeoftheultrasonicprobessuppliedwiththeCygnusunitweretested,thesinglecrystal(multipleecho)probeandthetwotwincrystal(single echo)probes.Fivemethodsofsurfacepreparationwereevaluatedonasteelblocktestpiece(Fig.4): Surface1.Unaltered-oneareawithintheconcretedrecesswasleftinitsuntreatedstatewithanycorrosionorconcretionleftintact Surface2.Softbrush-thisareawasbrushedwithasoftplasticbrushthenrinsedwithfreshwater Surface3.Wirebrush-thisareawasbrushedvigorouslywithawirebrushthenrinsedwithfreshwater Surface4.Hammerandchisel-concretionwasremovedfromasmallareaandcleaneddowntobaremetaljustusinga20mmchiselandlumphammer Surface5.Drill-anelectricdrillwasusedtomakeholesofincreasingdiametersthroughtheconcretiondowntothebaremetalbelow. Results Fig6:UTmeasurementabouttobemadeinasmallholechiselledthroughtheconcretiononthehullofA7 WithboththeunalteredSurface1andSurface2cleanedwithasoftbrushnousefulmeasurementswereabletobetakenofthematerialwitheitherthesingleortwincrystalprobes.Theprobesdidnotreceiveanyultrasonicsignalthroughtheconcretion. WithSurface3cleanedwithawirebrushnoconsistentmeasurementsweremadeofthecorematerial. Thesinglecrystalprobedidnotreceiveanyultrasonicsignalthroughtheconcretionbutthetwincrystalprobesdidoccasionallyregisterareading. Howeverthethicknessrecordedwasmuchsmallerthantheknownthicknessoftheobjectandlessthanhalfthethicknessrecordedwithcallipers. Itispossiblethatthesemeasurementsmayrepresentameasurementoftheconcretiononly,orthatthelayerednatureoftheconcretionhascausedtheacousticsignaltobedeflectedandthereforegivenafalsereading. ThehammerandchiselwasusedtoremoveconcretionfromSurface4. Concretionwasmosteasilyremovedusinggentletapswiththechiselheldatrightanglestothemetal,withthemetalsurfacethencleanedasmuchaspossibleofanyremainingconcretionusingmorelighttapsonthesurfacewiththechisel. Thismethodproducedthemostreliable,repeatablemeasurementsusingthetwincrystalprobebuttheresultswereinconsistentwiththesinglecrystalprobe.Forthetwincrystalprobetheconcretionwasremovedoveranareajustlargeenoughtoadmittheprobesoitcouldtouchthesurfaceofthebaremetal. Forthesinglecrystalprobealargerareawasneededonthelowerqualitysamplesbeforeaplacecouldbefoundwhereasuccessfulmeasurementcouldbemade. Surface5.Drill-Thismethodwhenusedwiththetwincrystalprobeproducedsomeresults,howevertheywerenotrepeatableinallcases. Toleranceswereoftenwithinthe0.5mmrange,butmeasurementswhichwereoutlierscouldbeinerrorbyasmuchas3mmsoitisnotclearwhatdistancetheinstrumentwasactuallyrecording. Nosuccessfulmeasurementscouldbemadeusingthesinglecrystalprobe. Underourtestconditionswefoundthatmeaningfulmeasurementscouldnotbeobtainedfrompreparationmethods1‐3;unaltered,softbrushedandwirebrushed. Method5useddrillsofincreasingsizestoputholesthroughtheconcretion,wherealltheholesweresmallerthanthediameterofthefrontfaceoftheprobes. Thismethodwastriedsimplybecauseitwouldbelessdestructivetothesurroundingmaterialthanusingahammerandchisel. Themethodwasunlikelytoworkbecauseofthedifficultyintransmittingtheultrasonicsignalfromtheprobe,downthesmallholethroughtheconcretion,intothebaremetal,reflectedoffthebackwallofthemetalthenbackoutoftheholetotheprobe. Usingalargediameterdrillcauseditsowndifficultyasthepointedendofthedrillwouldbewithinthebasemetallayerbeforetheholehadbeendrilledclear,andtheconicalblindholeinthemetalstoppedanythicknessmeasurementsbeingmade. Thismethodmayworkifaflatfacedendmillingcutterisusedtomakethehole,oneslightlylargerindiameterthanthediameteroftheprobefacesotheprobecansitflushwiththebaremetal. Drillingtheconcretiontomakeaholethroughitmaybelessdisruptivetothematerialthanhittingitwithahammerandchiselbutwouldrequiretheprovisionofapowersourceandhoseswhichwouldmakethismethodmorelogisticallydifficult. Drillingmayalsobeabletopenetrateparticularlyhardconcretionsthataredifficulttoremovewithachisel. Method4usingachiseltoremoveanareaofconcretionandexposethebaremetalbeneathproducedthemostreliableresultsonthesurfaceswetested. Theconcretiononthetestblockwasquicklyandeasilyremovedwithhandtoolssothismethodisalsoverysimple. Unlikemethodsrequiringtheuseofunderwaterpowertools,thismethodrequiresnosurfacesupport,powerpackorhoses. Fig7:60mmholeinthe6mmthickconcretiononA7showingbrightbutunevenmetalunderneath Onthemetalstructureswehavetestedsofartherehasbeenawidevariationonthenatureandcompositionoftheoverlyingconcretionscale,withsomemuchthickerthanothersandsomemorefirmlyattachedtothebasemetal.Althoughthehammerandchiselmethodworkedonthesesurfaces,powertoolsmayneedtobeusedonotherwrecksiftheconcretionsareveryfirmlyattachedtothebasemetal. Test2Conclusions Thetestsshowedthatreliablemeasurementscouldonlybemadewiththeconcretionremovedfromthebaremetal. Inallcaseswefoundthattheultrasonicsignalfromtheprobedidnotpenetrateeventhethinnestconcretionlayer,suggestingthatthelayerednatureofthecorrosionlayerortheinterfacebetweenbasemetalandcorrosionlayeraresuchthattheydonotallowthetransmissionofultrasonicsignalsatthefrequenciesusedbytheUTgauge. IthasbeenreportedthatmeasurementshavebeenmadethroughparticularlytoughconcretionsontheAE2submarinebutthisreportedthethicknessoftheconcretionandthebasemetal,sothiswasoflittlepracticaluse. AfterremovaloftheconcretiontheunderlyingmetaltendstobepittedandunevensoitmaybethisthatlimitstheeffectivenessofanyUTgauge(Fig.5). TheUTprobeneedstobecoupledtothebasemetalonaflatandsmoothsurfacesothattheprobedoesnotmoveandsothereisnogapbetweenprobeandmetaltocauseacousticscatteringandreverberation. Eventhesmallestvariationinthesurfaceofthebaremetalwillhindertheprobefrombeingreplacedinexactlythesamepositionandorientationwhentakingthesamemeasurementtwice,sothevariationinrepeatmeasurementsmaybemoretodowithsurfaceroughnessthanactualvariationinmetalthickness. Inpracticewhatthismeansisthatonroughmetalsurfacesthereisalimittohowpreciselywecanmeasurethethicknessofthemetal. Test3-MeasurementProcedure Method ThedifficultconditionsontheA7submarinesitewithitspoorvisibility,lowlightanddeepwaterledtooperationalproblemswhenmakingmeasurements. TherewerejustafewminutesavailabletomakeUTmeasurementsonanydive,lowvisibilityhamperedrecordingworkandnitrogennarcosisaffectedtheabilityofthediverstothinkclearly. ThereforeitwascrucialthatthediversusingtheUTgaugebecamefamiliarwithitsuseonthesurfaceandinshallowwaterbeforemakinghullthicknessmeasurementsatdepthonthesubmarine. TheUTgaugewaspurchasedwiththeadditionaldataloggingoptionsothemeasurementsmadeunderwatercouldberecordedbytheinstrumenttheninspectedlaterinmoredetailonthesurface. Automaticloggingwastobeusedinpreferencetorecordingonadiveslateasitwouldbequickerforonedivertodo,astheseconddiverofthepairwouldbeoccupiedholdinglightstoilluminatetheworkarea. Oncethetestareahadbeencleanedbacktobaremetalthemeasurementscouldbemade. Theinitialplanwasthatthedivermakingthemeasurementswastoplacetheprobeonthemetalsurface,waituntilthemeasurementstabilisedsufficientlyforthegaugetologit,thenmovetheprobeawayfromthemetaltoresetthegaugeforthenextmeasurement. Thisprocesswouldberepeatedatleastthreetimesateachmeasurementpointsoanideaofmeasurementqualitycouldbeassessed. Inpracticebothmethodshadtobeused,bothautomaticandmanuallogging. Theproblemwastheroughnessofthemetalsurfaceallowedtheprobetowobbleveryslightly,whichcausedinstabilityinthemeasurementsfromthegauge,andifthemeasurementswerenotsufficientlystabletheywouldnotbeautomaticallylogged. FortheAutoLogfeaturetorecordameasurement,theDIVEgaugeneedstoreceive2secondsofstablemeasurements,so16measurementsintotalatarateof8persecond.Stablemeasurementsshowachangelessthanorequaltotwotimestheresolutionsetting,withresolutionsetto0.1mm(Low)loggingwilltriggerwhenthereadingsfluctuatenomorethan0.2mmfor2seconds. OnceameasurementhadbeensuccessfullyloggedthediverhastoreleasetheprobefromthesurfaceforaperiodbeforetheAutoLogwouldrecordanothermeasurement. Thedisplayonthegaugereportsthequalityofthemeasurementsbycolouringthetextshowingthecurrentthicknessmeasurement,withredrepresentingapoormeasurementandyellowagoodmeasurement. Whenthemeasurementsweresufficientlygoodtobeloggedayellowbarisshownacrossthescreenandthelastloggedmeasurementvalueis showninthecornerofthedisplay.Onceasinglemeasurementismadeitisessentialthatthedivermovestheprobeawayfromthemetalsurfacefor2secondsbeforeattemptingthenextmeasurement.IfmeasurementsareattemptedtooquicklythentheAutoLogfeatureinthegaugedoesnotre-armandthesubsequentmeasurementisnotlogged.Oncethiswasdiscoveredandtheoperatingprocedurewaschangedtheproblemnolongeroccurred. Wedidexperimentwithplacingathinlayerofsoft,plasticmaterialovertheprobefaceinthehopethatitwouldhelpdampenanyrockingmotionoftheprobeontherough,corrodedmetalsurface. Unfortunatelythesoftmaterialactedjustlikeanyothercoatingandstoppedanysingleechomeasurementsbeingmade. Test3Conclusions Indiscussionwiththemanufacturer,theconclusionwasthatthesmallvariationofmeasurementvaluesfromthesamesampleareaisaninherentfeatureofthesingleechomethodandifitwerepossibletousemultipleechothiswouldeliminatethisproblem. Usingthemultipleechomethodtheprobecanberotatedinthesameplaceonthetestsiteandthereadingremainsconsistentasthesoundpulseisbeingsentintoandoutofthematerialinastraightline. Whereaswiththetwincrystalprobeandthesingle-echotechniquethesoundisbeingsentinaVshapedpath,sowhentheprobeismovedslightlythesoundpathisalteredandthemeasurementvaluechangesaccordingly.Onceasinglemeasurementismadeitisessentialthatthedivermovestheprobeawayfromthemetalsurfacefor2secondsbeforeattemptingthenextmeasurement. Test4-MeasurementQuality Method Theinitialtestsusingthegaugeshowedavariationinthemeasurementsmadeatasinglepointonmetalwitharoughorpittedsurfacewhichisasurfacetypetypicallyfoundoncorrodedironandsteelfromshipwrecks. Thisvariationinvaluesissignificant;ifmeasurementsmadeatasinglepointvarythenitaffectsourconfidenceinanymeasurementsmadeonanyotherpartofthesamehull. Ifthesamethicknessmeasurementsarerepeatedatalaterdateandareusedtoestimaterateofcorrosionthenitisessentialthatthemeasurementerrorisestimated,otherwiseinherentvariationinthemeasurementsmaybemisinterpretedasanactualchangeinmetalthickness. Testswereundertakenbytheprojecttodeterminethequalityofmeasurementsunderdifferentconditionsasameansofestablishingthemeasurementerror. Theabilityofthegaugetodetectthe‘true’thicknessmeasurementisdependantonanumberoffactors. Usingthesingleechomethod,thethicknessmeasurementisdefinedbytheleadingedgeofthefirstechoorfirstreturnfromthebackfaceofthematerialundertest. Withhomogenousmaterialthathassmoothfrontandbackfacesitisusuallystraightforwardtodeterminewherethispointisintherecord,butitbecomesmoredifficultasthequalityofthematerialandsurfacesdegrade. Theechoreturnsfromsmoothfacedhomogenousmaterialtendtobeamuchhigheramplitudethanthebackgroundnoisebutasthematerialdegradesthestrengthandqualityofthereturnechoesreduce. Amplifyingthereturnechoesallowsthemtobedetectedbutthisprocessalsoamplifiesthebackgroundnoise,eventuallytheechosignalbecomessoweakthatitislostinthebackgroundnoise. Theabilitytomakeathicknessmeasurementiscontrolledbytheratiobetweentheamplitudeoftheechosignalandtheamplitudeofthenoise,knownastheSignaltoNoiseratioorSNR. Figure8:TheeffectsofSignaltoNoiseRatio(SNR)ontheabilitytomakeameasurement(1.Good,2.Adequate,3.Poor) -Figure8showstheeffectsofdecreasingSignaltoNoiseratioonhullthicknessmeasurements.Eachofthethreetime-seriesplotsshowsonesingleechomeasurementandeachshowsthereturnedsignallevelovertime.Formeasurement1thegaugerecordsthebackgroundnoiselevelforaperiodoftimebeforethefirstreturnechoisseenasapeakontheplot,thetimebetweenzeroandthefirstreturnisthetimethesignaltakestotravelfromthefrontwallofthematerialtothebackwallthenbacktothefrontwallagainwhereitisdetected.Anytimebeforethefirstreturnarrivesthegaugewilljustberecordingthebackgroundnoise.Whenthefirstreturnarrivesthegaugedetectsitusingabuilt-insetofrulestodecideatpreciselywhattimethesignalisconsideredtohavearrived. -ForMeasurement2inFigure8thebackgroundnoiseisstrongerintheperiodbeforethefirstreturnarrives.Infact,thebackgroundnoiseisthesamebutthesignalhasbeenamplifiedsothattheweakerfirstreturncanbedetected,unfortunatelybyamplifyingthesignalyoualsoamplifythenoise.Againthegaugeusesitsrulestodeterminewherethemeasurementhasbeenmadethenreportsthemeasurementvalue. -ForMeasurement3thereturnsignalfromthebackfaceofthematerialisweakerstillsothegaugehastoamplifyitfurthertobeabletodetectit.Herethesignallevelissoweakitishardtotellwherethenoisestopsandthereplysignalstartssothegaugeisoperatingatthelimitsofitscapability. Aswecanseeabove,theabilitytomakeameasurementdegradeswithmaterialqualityonaslidingscalewithrepeatable,clearmeasurementsatoneendofthescaleandnomeasurementspossibleattheother. Inbetweenthetwoextremesliearangeofmaterialsandsurfaceswheretheabilitytomakeameasurementisinconsistentandthereliabilityofthemeasurementsislesscertain.  TheSNRistheratiobetweenthebackgroundnoiseandthesignallevelreceivedbythegauge,sothisisaffectedbyanumberoffactors: Signalpowertransmittedbytheprobe Theamountofsignalcoupledintothefrontsurfaceofthetestmaterial Theamountofsignalreflectedbackbythebacksurfaceofthematerial Theamountofsignalcoupledbackintotheprobe Thenoiseintheinstrument Thesignalleveltransmittedbytheprobeissufficientfornormaloperationandmakingthetransmittedpowerhigherwouldbringinotherunwantedside-effects.Theinstrumentwillbedesignedtominimiseitsownself-noisesothistoocannotbeimproved.Theotherfactorsaretheonesthataffecttheabilitytomakethicknessmeasurementsoncorrodedshipwrecks.Theabilitytogetsoundfromtheprobeandintothetestmaterialisdirectlyrelatedtotheroughnessofthematerialsurfacesotherougherthematerialtheharderitistogetthesignalintothematerial.Thesameappliesonthereturnjourneyasitithardertogetthesignalbackfromthematerialtotheprobewhenthesurfaceisrough.Theroughnessofthebacksurfaceofthematerialalsolimitshowmuchsignalisreflectedbackratherthanbeingscatteredawayinalldirections.Sotosummarise,asthesurfacesofthetestmaterialdegradeitbecomeshardertocouplethesignalintothematerial,theSNRdecreasesanditthenbecomesdifficulttomakereliablemeasurements. Asthematerialqualitydegradesthestrengthofthereturnechoesislikelytogetweaker(SNRgetssmaller),sotheprobabilityofsuccessfullymakingameasurementdecreasesandtheprobabilityincreasesthatanymeasurementthatismadeisincorrect. Thisisaninherentfeatureofallmeasuringinstrumentsandnotsomethingpeculiartoultrasonicthicknessgauges,ortheDIVEgaugeinparticular. UnfortunatelyitiscorrodedandroughhullplatingthatwewishtomeasuresooftentheseinstrumentswillbeworkingatthelimitoftheircapabilitywhendoingUCHwork.. Thefollowingtestshavebeendesignedtodemonstratewhathappenstothemeasurementsfromeachtypeofprobeasthequalityofmaterialdegradesfromidealtopoor.TheDIVEgaugedataloggingoptionrecordstheacousticresponsemeasuredbytheprobesothequalityofthesignalsusedforeachmeasurementcanbeinspectedafterthediveusingtheCygnusCygLinksoftware. Belowareaseriesoftestresultsshownasplotsofsignalstrengthagainsttimemeasuredbytheprobeandrecordedwithinthegauge. Eachplotrepresentsonesinglemeasurementandthereddotsontheplotsshowwherethegaugehasdeterminedthethicknessmeasurementtobe. Thetestsweredoneusingallthreeprobes: Probe1:S2C Singlecrystal2.25MHz,standardprobe,redband Probe2:T5B Twincrystal5.0MHz,yellowband Probe3:T2C Twincrystal2.0MHz,blueband Test4a Thefirsttestwastoinvestigatethebehaviouroftheprobesunderidealconditionstodemonstratethatanysignificantvariationinmeasurementsisnotcausedbytheinstrumentitself.  Themeasurementsweremadeona15mmthicksteeltestpiecewithsmoothsurfacesonboththebackandthefront. Thethreeplotsbelow(Fig.9)showtherecordsfromeachofthethreeprobesforasinglethicknessmeasurementofthesteeltestpiece. Figure9:Onemeasurementfromeachofthethreeprobesonthe15mmtestpiece,showinganidealsignalresponseforeachprobe Probe1showsanumberofevenlyspacedpeakswhichrepresentthemultipleechoeswithinthesteelmeasuredbythesinglecrystalprobe;thisismultipleechomodewheretheacousticsignalhasreflectedmultipletimesinsidethesteeltestsample. Eachofthepeaksrepresentsanechoreceivedattheprobe. Thethreereddotsatthetopofthefirstthreepeaksidentifythethreepeaksusedtodeterminethetimeofflightofthepulseandthusthethicknessofthesteel. Notethatthepeaksarenarrow,clearlydefinedandevenlyspacedasthismeasurementhasagoodSNR. Probe2showstheresponseforthehighfrequencytwincrystalproberecordingthesingleechofromthebackwallofthesteeltestpiece. Againthepeakintheresponseisnarrow,clearlydefinedandconsiderablylargerinsizethanthebackgroundnoise,soalsoshowsagoodSNR.Thereddotshowswherethemeasurementvaluehasbeenrecorded. Probe3showsthesameresultasprobe2exceptthateachpeakiswiderasthepeakwidthisafunctionofthefrequencyusedfortheultrasonicpulse. Makingthesamemeasurementmultipletimeswitheachprobegavetheexpectedhighlevelofrepeatabilityandlowmeasurementerror;themeasurementresolutioninthegaugewassetto0.1mmandallthemeasurementsfromeachprobeagreedtowithin0.1mm.  Test4b Byrepeatingthesametestonsteelblockapproximately35mmthickwithsurfacerustbutnopittingwecouldstarttoseesomevariationinrepeatmeasurements(Fig.10): Figure10:Fourmeasurementsmadewiththehighfrequencytwincrystalprobe(2)atthesamepointona35mmthicksteelblock Notethatthepeaksarelesswelldefined,especiallymeasurement2,andthebackgroundnoiselevelhasincreasedbetweenthetransmissionatT0andthefirstreturnaround35mmdistance. Themeasurementsreportedinthissetare35.0,35.1,34.8,35.2,buttheplotformeasurement4at35.2mmappearstobeofthehighestquality(narrowestandwithwell-definededges)soshouldbeconsideredtobethemostreliable. Asimilarresponsewasgivenbythelowfrequencytwincrystalprobe(3),butwiththewiderpeaksasseeninFig.5. Thesinglecrystalprobe(1)whenusedatthesametestpointonthesteelblockgavethreeclearmeasurementsandonethatwasofmarginalquality(Fig.11): Figure11:Fourmeasurementsmadewiththesinglecrystalprobe(1)atthesamepointona35mmthicksteelblock Forthefirstthreegraphsthethreepeaksthatarerequiredtomakeameasurementareclearlyvisiblebutthefourthpeakislostinnoise,thisgavemeasurementsof35.0,35.0and35.0mm. Inthefourthgraphthethirdmeasurementisontheedgeofthenoisebutthemeasurementisstillrecordedat35.6mm.Themultipleechomodeusedherebythesinglecrystalprobe(1)isontheedgeofbeingabletomakereliablemeasurements.ThreeofthemeasurementsagreetowithintolerancebutthefourthresultshowsthatitispossibleforthismethodtomakeanincorrectmeasurementwhenthematerialsurfaceispoorandtheSNRislow. Test4c Boththetestpieceandthesteelblockwereingoodconditionandgavetheexpectedgoodresultssothetestswerethenrepeatedonasamplemorerepresentativeofthekindofmaterialencounteredonshipwrecks. Thesampleusedwasan 8.5mm(±0.2mm)thicksectionofboilerplateofunknowndatethathadacorrodedandpittedsurface. Figure12:Fourmeasurementsmadewiththehighfrequencytwincrystalprobe(2)atthesamepointonan8mmthickboilerplate Therecordedmeasurementswere8.4,7.1,7.1,7.9mmwithtwoadditionalmeasurementsat8.4mmnotshown(Fig.12). NotetheincreasedamplitudeofthebackgroundnoisebetweenT0andthefirstreturn(reddot). Measurement1showsacleanechoreturnandofthefourthefirstislikelytobethemostreliablemeasurement.   Measurements2and3havedetectedatapointonasmallsignalpeakbeforethemainpeakwhilemeasurement4showsalesswelldefinedfirstreturn. Here,theautomaticdetectionofthefirstreturnhasnotbeensuccessfulbutinspectionofthetracesinpost-processinghasbeenabletoidentifythemostreliablevalueforthemeasurement. Figure13:Fourmeasurementsmadewiththelowfrequencytwincrystalprobe(3)atthesamepointonan8mmthickboilerplate Thelowfrequencytwincrystalmadethreemeasurementsof8.5,8.2,8.3and7.0mm(Fig.13). Inmeasurement4thegaugehasdetectedearlyonasmallerpeak. Notethatinthefirstmeasurementthebackgroundnoisebeforethefirstreturnisnowmorethanaquarterasstrongasthefirstreturn,sotheSNRisnowpoorandtheabilitytomakeareliablemeasurementisdegraded. Figure14:Fourmeasurementsmadewiththesinglecrystalprobe(1)atthesamepointonan8mmthickboilerplate Themeasurementsfromthesinglecrystalshowasmallervariationinvaluesthanreportedbythetwincrystalprobes:7.6,7.6,7.5,7.6(Fig.14). ThenoisesurroundingtheechoesisverylargesotheSNRispoor,hencetheconfidenceinthemeasurementsshouldbepoor,butthefourmeasurementsagreetowithin0.1mmsomaybemorereliablethanthefourplotssuggest.Althoughfourmeasurementsweremadesuccessfullyittookconsiderablylongertomakeeachmeasurementastheprobeneededtobepositionedinexactlytherightplaceontheroughmetalsurfacebeforethegaugewouldrecordareading. Theconclusionthatwecandrawfromthissetofmeasurementsisthatitishardertoobtainameasurementonthisqualityofmaterialusingthesinglecrystalprobebutwhenameasurementisobtainedthenitisreliable. Test4d Test4dwasdoneonthecorrodedsteelhullofaRoyalNavyAclasssubmarinelostin1914.Unfortunately,thequalityofthesteelhullontheHMsubmarineA7wasworsethantheboilerplatesampleandthispoorerqualityhadadetrimentaleffectonthehullthicknessmeasurements.FormoredetailsaboutthelocationofthetestpointspleaseseeExperimentsinUltrasonicHullThicknessMeasurementsontheA7Project. Measurementswereplannedfor8pointsonthehullofthesubmarinebutonly5werecompletedduetolackofavailabletimeonsite. Thetestsabovethatweredonepriortomakingmeasurementsonthesubmarineshowedthatthehighfrequencytwincrystalprobewaslikelytoproducethebestresultssothatiswhatwasusedonthehull.Eachsetofmeasurementsweremadeoneaftertheotherwithaminimaldelaybetweenthem,withtheprobemovedoffthehullandreplacedeachtime,withinanareaonlyslightlylargerthantheareaofthefaceoftheprobetomaximiserepeatability. Notethatthegraphsbelowshowaconsiderablevariationintheamplituderesponseforwhatshouldbethesamemeasurement. Test4d-UT5 TwoattemptsweremadeatrecordingathicknessatpointUT5asitwasdifficulttomakeareliablemeasurementonthefirstattempt. Fourmeasurementsweremadeonthesecondattempt(Fig.15): Figure15:FourmeasurementsmadeatpointUT5usingthehighfrequencytwincrystalprobe(2) Therecordedvalueswere3.8,3.6,3.8,3.7mm. Theshortestofthethreevaluesis3.6mmsoitwouldbetemptingtousethisvalueasitistheworstcasemeasurementforremaininghullplating. However,inspectingthetraceforthe3.6mmmeasurement(Fig.15:2)showsthatherethegaugehasdetectedatapointonasmallsignalpeakthatprecedesthemorereliablesignalpeak,whichresultsinashortmeasurement.Again,thisshowsthatwithpoorSNRmeasurementsitisusefultobeabletorecordthemeasurementsforlaterinterpretation. Test4d-UT1 DifficultieswereencounteredwhenattemptingtomakemeasurementsatpointUT1onthestarboardbow. Thevaluesreportedbythegaugewerenotstableasthesurfaceatthispointwasheavilypittedandmovingtheprobebythesmallestamountwouldchangethemeasuredvalueconsiderably,asshownbelow: Figure16:FourmeasurementsmadeatpointUT1usingthehighfrequencytwincrystalprobe(2) Threemeasurementsweremadeintwoattemptswithvalues3.4,5.4,5.1mm(Fig.16). Inthefirstgraphwecanseethatthegaugehasdeterminedthepointtobeatthelefthandedgeoftwopeaksofsimilaramplitudegivingavalueof3.4mm. Inthesecondmeasurementithasdeterminedthepointtobebetweenthetwopeaksandhasrecordedalongermeasurementof5.4mmbutthethirdmeasurementof5.1mmlooksthemostreliableofthethree. Thereisnoclearlydefinedfirstreturninanyofthethreeplotsastheinstrumentisattemptingtomakemeasurementswithinthebackgroundnoiseandisoperatingatthelimitofitscapability. Test4d-UT8 PointUT8onthestarboardsternprovidedthreemeasurementswitharangeofvalues:8.8,9.7and9.8mm(Fig.17). Inspectingtheplotswecanseethatinthefirsttwothegaugehasdetectedonnoisebeforethemainpeakandonlyinthelastplotcanweseeaclearlydefinedfirstreturnpeak. Themeasurementsweremadeinthesameareaoneaftertheotheryettheyshowalargevariationinresponse. ThevariationiscausedbychangesinthegeometrybetweentheprobeandmaterialandservestodemonstratethelimitontherepeatabilityofUTmeasurementsondegradedsteel. Therepeatabilityisnotcausedbylimitationsintheinstrumentbutisdeterminedbythequalityofthematerialundertest. Areliablemeasurementcouldbemadeatthislocationbutitrequiredtheinspectionoftheamplitudeplotstohelpdeterminewhichwasthemostreliablevalue. Figure17:ThreemeasurementsmadeatpointUT8usingthehighfrequencytwincrystalprobe(2) Testsweredoneusingtheboilerplatetoseeifin-situcorrosionlayersonthebackfaceofthemetalhadanyeffectbutnodifferencewasfoundinthequalityofthemeasurementsifthecorrosionlayerwasinplaceorhadbeenremoved. Asthepresenceofanycorrosiononthefrontfaceofthemetalseverelyattenuatesthesignalfromtheprobeitappearsthatcorrosionlayersdonotcoupleultrasonicsignalswell,sotheirpresenceonthebackwallofthemetalshouldhavenoeffect. Foralloftheteststhesoundspeedvalueformildsteel(5920ms-1)wasusedeventhoughforsomesamplesthematerialtypewasunknown. Thetestswereintendedtodeterminetheabilitytomakeultrasonicthicknessmeasurementsandtoinvestigatethequalityofthemeasurementsratherthantodeterminethedegreeoffitwiththetruesamplethickness. Test4Conclusions Thesetestshavedemonstratedthattheabilitytomakeathicknessmeasurementlessensasmaterialdegrades. Onpoorerqualitysamplesitisimportanttomakeanassessmentofthequalityofrepeatmeasurementsbeforeattemptingtomeasurevariationwithinasmallareaoracrossanentirewreck. Thesinglecrystalmultipleechomethodworksongoodtoaveragequalitymetalsurfacesbutonpoorerqualitysurfacesitishardertomakeameasurement. Thetwincrystalsingleechomethodwillproduceresultsonpoorerqualitymetalsurfacesbutitisessentialthataminimumofthreemeasurementsperpointaremadewhenusingthetwincrystalprobesoqualityandmeasurementerrorcanbeassessed. OnpoorqualitymaterialsthelowSNRofthesignalsbeingusedtomakemeasurementsmeansthattheabilitytoinspectloggedtracesusingtheCygLinksoftwareisessentialforpost-diveanalysisofmeasurementquality. Test5-BackfillingwithEpoxy Method OntheA7submarinetheholesintheconcretion scalemadeateachmeasurementpointwerebackfilledwithepoxyputtytominimiselocalcorrosionoftheexposedmetal. WessexrecommendedtheuseofAquastikepoxybutthisisexpensiveandcouldonlybeboughtinlargequantities,sothecheaperandmorereadilyavailableMilliputepoxywasusedinstead. EachpackofMilliputcontainstwo57gsticksofdifferentcolours. Tousetheputty,equalquantitiesofeachstickaremixedtogethertoasoftandstickypastewhichgraduallyhardensovertimewithoutshrinking. Theputtywillsticktometals,concretion,wood,ceramics,plasticsandglass,butmostimportantlyitwillalsoadheretowetandcorrodedironandsteel. Theepoxywillsetunderwaterwiththespeedofhardeningdependantontemperature;inwater12°CtheMilliputcouldstillbemouldedforuptoonehouraftermixing. Milliputcanbemixedunderwaterbutintestswefoundthatitwasdifficulttomixthetwopartsevenlysomixingonthesurfacejustbeforethedivewaspreferred. AlthoughthemanufacturersadvertisethatMilliputcanbeusedunderwaterwetestedabatchinseawaterforamonthbeforeusingitontheproject. Thetestsshowedthattheepoxywouldadheretocorrodedsteel,wouldsetcompletelyafterfourhoursandremainedvisiblyunaffectedbyseawaterafteraonemonthimmersion. Withthetestscompletetheepoxywasusedtorepairholeschiselledintotheconcretiononthesubmarine. Ballsofepoxy30mmindiameterweremixedonthesurfacejustbeforethestartofthedivebeforebeingplacedinare-sealableplasticbagwithalittleseawatertostopthemstickingtogether. Onceunderwater,whenaholeneededtobefilledaballofmixedepoxywastakenfromtheplasticbagandfirmlypushedintotheholeintheconcretionusinghandsratherthantools,ensuringthattheepoxywasfirmlypushedintothesidesoftheholetosecureitfirmlyandtostabiliseanylooseconcretionaroundtheedgesofthehole. Inspectionofthefirstbatchofepoxyonasubsequentdayshowedthatithadsetsolidandcouldnotbepriedfromtheholeintheconcretionwithouttools.Priortomakingsomerepeatthicknessmeasurementsoneoftheepoxypatcheswasprisedfromtheholewithadiveknifesomeweeksafteritwasplacedanditcameawaycleanlywithoutdisruptingthesurfaceofthemetalorthesurroundingconcretion. Test5Conclusions InconclusionwefoundthattheMilliputepoxyfulfilledallofourrequirementsforsealingandstabilisingholesinconcretion. ThecheaperMilliputepoxyputtyperformedwellenoughinthetestssothatthemoreexpensiveAquastikneednotbeused. Conclusions Theconclusionsofthetestsare: TheCygnusDIVEMk.2gaugeisasuitableinstrumenttouseforthicknesstestingforUCHwork RemoveallconcretionbeforemakinganyUTmeasurements,themethoddoesnotworkifyoudonot Thesimplehammerandchiselmethodworkedwellforremovingconcretiononallofthemetalsamplestested Usethesinglecrystalprobeongoodtoaveragequalitymetal Usethetwincrystalprobeonaveragetopoorqualitymetal Makeaminimumofthreerepeatmeasurementsateachsamplepoint Inspectthemeasurementplotsinpostprocessingtoensuregoodqualitycontrol Calculatethevariationinmeasurementsateachsamplepointandusethevaluewhencalculatingchangesinmetalthicknessovertimeoracrossasite UseMilliputepoxyforrepairingholesmadeinconcretion References HoltP.,2015,HMSubmarineA7,AnArchaeologicalAssessment,BARBritishSeriesno.613,Oxford,ISBN 9781407313740  RussellM.,ConlinD.,&MurphyL.,2006,AMinimum-ImpactMethodforMeasuringCorrosionRateofSteel-HulledShipwrecksinSeawater,InternationalJournalofNauticalArchaeology35.2:310–318 SmithT.,2008,ReportofOperationSilentANZACMaritimeArchaeologicalAssessmentOfHMASAE2,availableathttp://ae2.org.au/resources/research/,accessedOct.2016 WessexArchaeology,2012,UltrasonicThicknessMeasurementMethodologyDevelopmentandTesting.HMSubmarinesHollandNo.5andA1,Unpublishedreport,Ref:83800.23 RelatedPages UltrasonicThicknessMeasurements ExperimentsinUltrasonicHullThicknessMeasurementsontheA7Project ©3HConsultingLtd.RegisteredNo.03081823Allrightsreserved. Home|Contactus|