Sei sulla pagina 1di 60

DetectingLeakagesin

UndergroundBuriedPipe
DepartmentofMechanicalEngineering
NationalUniversityofSingapore

Session2009/2010

Name:TeoHwiBee
MatricNumber:U067137N
Supervisor:AssocProfChewChyeHeng

Summary
Thereisgreatdifficultyinlocatingtheexactlocationofleakagesinwaterpipes.
Often,noisecreatedduetosharpbendsisdetectedinstead.Inthisproject;weshall
trytodifferentiatenoisecreatedbyleakageorbendsinpipes.Thereisaneedto
findoutthefrequencywindowatwhichturbulenceisdetectedsowecaneliminate
thesefrequencies.
Inthefirstpartoftheexperiment,thevibrationfrequencyspectrumisobtainedfor
waterflowingthroughapipewithoutanyleaks.Thepipehasasharpelbow
attachedtoittocreateturbulencenoise.Therangeoffrequenciesatwhich
turbulenceoccursisnoted.
Inthesecondpartoftheexperiment,aholeisdrilledintothepipetoallowleakage
tooccur.Similarly,thevibrationfrequencyspectrumisobtainedandcertainranges
offrequenciesareeliminated.Therangesoffrequencieswhichareomittedoutare
thosesuspectedtobeduetonoisecreatedbyturbulence.Theonlynoisethat
shouldbeobservedwillbeduetoleakageinthepipes.Inordertolocatetheexact
pointofleakage,correlationsweredone.
Insteadofusingdataloggedfromwaterpipingsystemdirectly,ascaleddownmodel
wassetuptosimulatethisproblem.

Acknowledgements
Theauthorwishestoexpresssincereappreciationoftheassistancegivenbymy
supervisor,AssocProfChewChyeHenginadvisingdirectionsoftheproject.Not
forgettingthehelprenderedandassistancefromthetechniciansofDynamicsLab,
AirConditioningLabespeciallyMrChengandMrDevanincarryingoutthework
successfully.

Contents

Summary ................................................................................................................................... 2
Acknowledgements................................................................................................................... 3
ListofFigures ............................................................................................................................ 6
ListofTables ............................................................................................................................. 7
ListofSymbols .......................................................................................................................... 8
1.Introduction .......................................................................................................................... 9
1.1Purpose ............................................................................................................................. 10
2.ProjectDirections................................................................................................................ 11
3.Theory ................................................................................................................................. 13
4.ExperimentProcedure ........................................................................................................ 14
4.1ExperimentalSetUp1(WithoutHole) ............................................................................. 15
4.2WaterFlowRate ............................................................................................................... 16
4.2.1CalculationofWaterFlowRate ..................................................................................... 17
4.2.2ResultsforDifferentWaterFlowRate........................................................................... 18
4.3GeneralObservations ....................................................................................................... 19
4.3.1LocatingTurbulenceforDifferentLocationsofAccelerometers................................... 23
4.3.2LocatingTurbulenceFrequencyforDifferentFlowRate............................................... 26
4.4ResultsforSetUp1........................................................................................................... 27
5.ExperimentSetUp2(WithHole)........................................................................................ 29
5.1ResultsforDifferentWaterFlowRates ............................................................................ 31
5.2LeakageFrequenciesforDifferentHoleSizes................................................................... 32
5.3Locatingconsistencyinleakagefrequency....................................................................... 34
5.4LocatingPointofLeakage ................................................................................................. 35
6.Discussion............................................................................................................................ 38
6.1Limitations........................................................................................................................ 39
6.2OtherPossibleErrors ........................................................................................................ 41
7.Conclusion........................................................................................................................... 42
8.Recommendations .............................................................................................................. 43
AppendixISpectrumsforDifferentFlowRates .................................................................... 44

AppendixIILocatingforconsistencyinturbulencefrequency ............................................. 46
AppendixIIILeakageFrequencyfor1&2revolutionwaterflow ......................................... 51
AppendixIVLeakFrequencyfor1mmhole........................................................................... 54
AppendixVLeakageFrequencyfor1.5mmHole................................................................... 57
Bibliography ............................................................................................................................ 60

ListofFigures
1.Layouttodeterminethepositionofaleakfromburiedpipewaterdistribution
pipe
2.SchematicDiagramofPipeSetUp
3.Closeupviewofvalveand90elbow
4.ActualExperimentalSetUp
5.DifferentWaterFlowRateChannel1
6.DifferentWaterFlowRateChannel2
7.FrequencySpectrumofPipe
8.FrequencySpectrum1revolution,1mdistanceapart,channel1
9.FrequencySpectrum1revolution,1mdistanceapart,channel2
10.FrequencySpectrum1revolutionflow,0.8mdistanceapart
11.FrequencySpectrum1revolutionflow,0.9mdistanceapart
12.FrequencySpectrum1revolutionflow,1mdistanceapart
13.SchematicDiagramofSetUp2
14.Frequencyof1mmhole
15.Frequencyof1.5mmhole
16.Timespectrumfor1revolutionflow,1mmhole
17.TimeSpectrumfor1revolutionflow,1.5mmhole,channel1
18.TimeSpectrumfor1revolutionflow,1.5mmhole,channel2
19.Vibrationpausedinbetweenmeasurements

ListofTables
1.Waterflowratefordifferentrevolutionofvalve
2.Distancewithcomparisontorangeofturbulence
3.WaterFlowrateandrangeofturbulencefrequency
4.Rangeatwhichturbulenceoccurs
5.RangeofFrequenciesfor1revolutionand2revolutionwaterflow
6.RangeofLeakFrequencyfor1mmand1.5mmhole

ListofSymbols
D1

Distanceoftheleakagepoint,m

wavespeed,m/s

timedifference,s

1.Introduction
Todayswaterutilityoperatorshavearangeofequipmentsandtechniquesto
measure,analyse,monitorandreduceleakageinpipes.Inrecentyearstherehas
beenasurgeindevelopmentoftoolsandequipmenttosupportthistask.
However,thereisstillabiggapinsuchtechnologies,complementarytechnology
andequipmentforlocatingandpinpointingleaksinhardtoreachsituations,suchas
undergroundburiedpipes.Thussystematicdetectionmethods,exactwatermeter
readingandfastleakrepairsareessentialtoreducetheamountofleakages.
Thereareseveralconventionaltechnologiesforlocalizing,locatingandpinpointing
leaksindistributionnetworks.Themostcommonlyusedmethodwillusuallyinvolve
a2stepapproach.Firstly,waterauditingisdone,whichmonitorsthenighttime
minimumflowrateintheinletandoutletofapipelinesystem.Theauditscannot
findthepreciselocationoftheleaksbutestimatethetotalleakagefromthepipeline.
Next,acousticmethodisusedwherebysoundorvibrationbywaterleakingfrom
pipeisdetected.Otherconventionalmethodsincludetracergas,thermography,
flowandpressuremodeling.
Therearealsolessconventionaltechniqueswhichareinvariablycalledonwhenthe
othertechnologiesfail:

Correlationwithlowfrequencyhydrophones

Inversetransientanalysis

Groundpenetratingradar(GPR)

Lowfrequencyhydrophonesworksonthesamebasisasacoustics.Hydrophones
havepiezoelectrictransducersthatconvertssoundsignaltoelectricalsignal,since
soundisapressurewave.
Inversetransientanalysisisbaseduponpressuredatacollectedduringthe
occurrenceoftransienteventsandminimizationofthedifferencebetweenthe
observeddataandthecalculatedpressures.Insystemswithoutleaks,important
phaseshiftsofthepressurewaveoccuratallsiteswithinthepipe.Forpipeswith
leaks,pressurewaveandtransienteventdampingoccursandmathematicalanalysis
ofthisphenomenonallowsthepositionoftheleaktobedetermined.
Groundpenetratingradar(GPR)isarapid,highresolutiontoolfornoninvasive
subsurfaceinvestigation.GRPproduceselectromagneticradiationthatpropagates
throughthegroundthenreturnstothesurface.Theradarwavesaredependent
uponthedielectricconstantduetochangesinsubsurfacematerial.Thetraveltime
oftheelectromagneticwavefromtheemittedtoreflectedwaveisafunctionofthe
depthandelectricpropertiesofthemedia.

1.1Purpose
Differentiatebetweennoisecreatedtoduetoturbulenceandnoisecreateddueto
leakages

10

2.ProjectDirections
Initially,theintentionoftheprojectistoinvestigatethepossibilityofusingground
penetratingradartoreplacetheexistingacousticmethoddetectionsysteminwhich
PUBisusing.AsGRPalsodetectsundergroundwatercontent,itisapossible
alterativetechnologyforleakdetection.TheuseofGRPinvolvesaworkerto
manuallynavigatethedeviceaccordinglytothewaterdistribution.Dependingon
theantennaeandthematerialpropertiesofthesoil,itcansuccessfullypenetrate
downtoabout10m.Dataarestoredandmapintoimageswhichdisplaysthewater
contentofthesoil.Themethodislabourintensiveassurveyisneededtobedone
throughoutthewholenetworksystem.
AftersomediscussionwithPUB,acousticloggersystemwasdecidedeventuallyas
theyhaveexistingloggersaroundtheirwaternetworksystem.Theyarecurrently
usingthePhocus2SystemandEurekaDigitalsystemtodetectleakages.
Phocus2systemisanacousticloggersystem.Anareaissurveyedbyinstallinga
numberofloggersatavailablehydrantsandvalves,Phocus2recordsacousticnoise
overaselectedperiod.EurekaDigitalsystemisaleaknoisecorrelator.Transmitters
areplacedat2pointsonasuspectedleakposition.Thenoisecorrelatoroperatesby
comparingthenoisedetectedat2differentpointsinthepipeline.

11

Bothsystemsbasicallyworksalongthesameprincipleandbothoperateswithan
integratedaccelerometer.Themainproblemthatariseswithsuchloggersisthatitis
difficulttodifferentiatebetweennoisecreatedduetoturbulenceorduetoleak.

12

3.Theory
Atypicalmeasurementlayouttodeterminethelocationofaleakinaburiedpipeis
showninfigure1.

Figure1:Layouttodeterminethepositionofaleakfromburiedpipewaterdistributionpipe

Ifaleakissuspected,theacousticsensors,typicallyaccelerometerareplaceseither
sideoftheleak.Theaimistodeterminethepositionoftheleak,whichinthiscaseis
thedistanceD1fromtransmitterA.Thisdistanceisrelatedbytheequationbelow:

Where isthewavespeed

isthetimedifference

Thustodeterminetheleak,thesevariablesneedtobeknown.Thewavespeedc,
canbemeasuredexperimentallybytappingonthepipe,thenmeasuringthetimeat
whichthesoundtravel.Toestimate

,thecrosscorrelationofthesignalsfromthe

sensorsisgenerallyused.Howeverthequalityofthisestimatedependsuponthe
typeandpositioningofthesensorsandtheprocessingofthesignals.

13

4.ExperimentProcedure
2setupsaredonetoinvestigatenoisecreatedduetoturbulenceandleakage.In
thisexperiment,insteadofusingtheactualpipingusedinwaterworks,a1inch
copperpipeisusedtosimulateactualbehavior.
Copperpipeof1inchdiameterwasusedtogetthefrequenciesofvibrationwhen
thepipeiswithorwithouthole.Accelerometersareplacedatbothendsofthepipe
toobtainthevibrationmotionofthepipeanddistancesatwhichaccelerometers
areplacedonthepipearevaried.A90elbowbendwasplacedtocreatesomeform
ofturbulence.2experimentsareconductedtoinvestigatethevibrationspectrumof
thepipe.Firstsetupwaswhenwaterflowthroughanormalcopperpipeandthe
turbulencefrequencyisinvestigated.Asforthesecondexperiment,aholewas
drilledintothepipetosimulatewaterleakagewhiletheotherendofthepipeis
sealed.Hence,leakagefrequencyisinvestigated.

14

4.1ExperimentalSetUp1(WithoutHole)

SetUp1:Togetfrequencyspectrumforwaterflowincopperpipe

valve

Accelerometer2

Accelerometer1

CH2

CH1

90elbow

1.3mpipe

Figure2:SchematicDiagramofPipeSetUp

Figure3:Closeupviewofvalveand90degreeelbow

15

Channel2

Channel1

Figure4:ActualExperimentalSetUp

4.2WaterFlowRate

Waterflowrateiscontrolledbythenumberofrevolutionsdonebythetap.A
markerwasusedtogivearoughestimateofthewaterflowrate.Timeisrecorded
togettheaverageflowrate.
Table1:Waterflowratefordifferentrevolutionofvalve

Revolutions

Timetaken/s

AverageTime/s

VolumeFlowrate/m3s1
104

49.23

48.35

48.79

2.07

25.5

25.45

25.475

3.96x104

24

24.93

24.465

4.13x104

16

4.2.1CalculationofWaterFlowRate

1inch=0.0254m

Timetakenforwatertoflowthroughthepipeisestimatedtobeslowestat3.17s.
Hence,whenobtainingfrequencyspectrumforthepipe,realtimecaptureof
approximately5swasdone.Arealtimecaptureof5sshouldbesufficienttocollect

17

vibrationsdatamadebythepipe.Thefrequenciesmadebythepipeshouldbe
relativelylowandoflowamplitude.

4.2.2ResultsforDifferentWaterFlowRate

Comparingfigure5and6,itcanbeseenthatahigherwaterspeedresultsingreater
vibrationmagnitude.Thisisexpectedofgreaterwaterspeed.Soundcreatedwillbe
louderandofhighermagnitude.Anotherobservationthatwecanseebelowisthat
mostofthepeaksaregroupedatalowfrequencyrangeforall3waterflowrate.
Forthepowerspectraobtainedfromchannel2,thereisananomalyfor1revolution
waterspeed.Themagnitudeshouldbelowercomparedtoafasterwaterspeed.
However,itcanbesuspectedthatitmaybeduetonoisefromtheenvironment
duringrecordingoroutletnoisefromthepipe.Itcanbeobservedthatthereare
clustersofpeaksofhigherfrequenciesinthegraphofchannel2butnotchannel1.

18


Figure5:DifferentWaterFlowRateChannel1

Figure6:DifferentWaterFlowRateChannel2

4.3GeneralObservations

Afteranalyzingmanyspectrums,therearesomeconsistentobservationsthatcanbe
broughtup.Theseobservationscanbeappliedtobothexperiment1and2.
19


Figure7:FrequencySpectrumofPipe

Firstobservationisthatmostspectrumshasinitialspikeat0frequency,thereforeit
isdifficulttoanalysetherestofthespectrumathigherfrequencies.Thismaybedue
toincorrecttimingatwhichmeasurementsaretaken.Theinitialspikemaybedue
tolowturbulencefrequencyatwhichiscausedeitherbythevalveortheelbow.
Thereisalargeinitialchangeinmomentum.Thisoccurrenceiscommonformany
graphsandwhenthishappens,weshallomitandfocusonthepeaksdueto
turbulence.

20


Figure8:1revolution,1mdistanceapart,channel1

Figure9:1revolution,1mdistanceapart,channel2

Aswecomparetheresultsbetweenchannel1and2andzoomintooneofthepeaks
inchannel1whichhappensat1.488kHzhavingamagnitudeof8.53E03Vrmsand

21

channel2at1.488kHzhavingamagnitudeof6.02E03Vrms.Thereisadecreasein
magnitudeduetolostofenergywhenwavepassesthroughadistance.Thepeak
magnitudewaschosenasitisduetotheturbulencecausedbythe90elbow.
Zoomingintothecircledoutfrequenciesaboveatfigure8,itcanbededucedthat
thesehighmagnitudesarenoisescausedbytheturbulenceoftheelbow.Compared
tochannel2,atthesefrequencies,themagnitudesareslightlylowerduetoenergy
losses.
Figure9hasacircledoutregionandwecanseethatthereareusuallyhigh
magnitudesathigherfrequencies.Thisshouldnotbethecaseandwecansuspect
thatthesehighmagnitudesareduetotheoutletnoise,noisecreatedwhenwater
gushesoutofthepipe.Sincechannel2accelerometerisnearertotheoutlet,itis
abletopickuptheloudnoiseswhenwaterisescaping.
Hence,secondobservationisthatinitialpeaksareduetoturbulenceorleakages.
Thirdobservationisthatpeakathighfrequenciesforchannel2isduetooutlet
noise.

22

4.3.1LocatingTurbulenceforDifferentLocationsof
Accelerometers

Ideally,accelerometersaretobeplacedatacertaindistance.Sincethepipeisabout
1.3mlong,theaccelerometersareplacedabout1mapart.Itcannotbeplacedany
furtherasitwillcollectnoisefromtheenvironmentforchannel2andforchannel1,
itwilldirectlycollectnoisefromthe90degreeelbow.
Distanceatwhichaccelerometersareplacedarevariedfrom1mto0.8m.Placing
theaccelerometerstoocloseisnotfavourableaswellbecausethesoundcollected
willattunesuchthatbotharerecordingthesamething.Inthiscase,wearetryingto
getthemaximumdistanceatwhichsounddiesoffsotheaccelerometerswillnotbe
placedatthatparticulardistance.Belowaresomeofthedistancesvariedfrom0.8m
to1m.

Figure10:1revolution,0.8mdistanceapart

23

Whenaccelerometersareplaced0.8mdistanceapart,theturbulencefrequency
occursatarangeof1.328kHzto1.584kHz.

Figure11:1revolutionflow,0.9mdistanceapart

Theaccelerometersareplaced0.9mdistanceapart,wenowobserveasimilartrend,
theinitialpeaksatlowfrequenciesforchannel1isduetoturbulence,thesepeaks
decreasesatchannel2.Hence,wecaninterpretthatturbulencewhichistheboxed
outregionoccursat1.248kHzto1.840kHz.Therearelotsofpeaksat1.92kHz
onwards.Thesepeaksareduetonoisecreatedbytheoutlet.

24


Figure12:1revolution,1mdistanceapart

Fromthegraphabove,itisobservedthattherangeatwhichturbulenceoccursover
816Hzto1.97kHz.
Table2:Distancewithcomparisontorangeofturbulence

Distanceatwhichaccelerometersareplaced Rangeofturbulencefrequency/kHz
1m

0.8161.97

0.9m

1.2481.84

0.8m

1.3231.584

Fromtable2,atalldistancesinwhichaccelerometersareplaced,vibrationcanbe
detected.Accelerometerscannotbeplacedanyfurtherduetotheshortpipelength.
Theoptimumdistanceatwhichaccelerometersshouldbeplacedisprobablyata
distanceof0.9m.Thedistanceof0.8mmaybetooclose.Therangeofturbulence
frequencyobtainedwhenaccelerometersareplacedcloselyistoonarrow.

25

4.3.2LocatingTurbulenceFrequencyforDifferentFlowRate

Ideally,aslowerflowratewillbebetterinlocatingturbulenceandleakage.Asthere
islittleexcitationonthepipe,thenoisemadewillbesmaller.Turbulenceand
leakagepeakswillbemoreobvious.
Belowisthetableatwhichdifferentflowrateisusedandthecorresponding
estimatedturbulencefrequency.Comparingdifferentflowrates,theslowestwater
speedgivethemostappropriaterangeatwhichturbulenceoccurs.Forahigher
waterspeed,therangeatwhichturbulenceoccuristoowidespread.
Table3:WaterFlowRateandRangeofTurbulence

Rangesatwhichturbulenceoccurs

ID5(1rev,0.9m)

1.248kHz

1.84kHz

ID6(2rev,0.9m)

368Hz

1.87kHz

ID7(3rev,0.9m)

688Hz

1.54kHz

RefertoAppendixIforthespectrumsabove.

26

4.4ResultsforSetUp1

Readingswererecordedbythetaperecorderandsignalsaresentthroughthe
oscilloscope.Timecaptureof5sofvibrationisrecordedandthefrequencyspectrum
rangeisuptoafrequencyof6.48kHz.Anyfrequenciesabovethatareundetectable
andthenoisemagnitudeisverysmall.Averagesof20aredone.
Revolutionsofthevalvesanddistanceatwhichtheaccelerometersareplacedare
varied.Inordertochecktheconsistencyinourdeductionwenotedthefrequencies
ofpeaksforchannel1andchannel2.Notethatwewilleliminatetheoutletnoise
pickedupbychannel2andestimatethepeakfrequencyforchannel2whichis
createdduetoturbulence.Anotherconditionisthatthemagnitudeatwhichthis
occursforchannel2mustbelowerthanchannel1.Since1revolutionwaterflow
speedgivesabetterrange,analysiswillbecarriedoutforthisspectrum.Thetime
captureof5swillbebrokenupinto1sintervalof5partstocheckiftheturbulence
frequencyisconsistent.

27

Table4:Rangeatwhichturbulenceoccurs

Rangesatwhichturbulenceoccurs

ID2(1rev,1m)5seconds

1.22kHz

1.48kHz

1stsecond

1.28kHz

1.41kHz

2ndsecond

928Hz

1.36kHz

3rdsecond

928Hz

1.46kHz

4thsecond

1.38kHz

1.41kHz

5thsecond

928Hz

1.07kHz

ID5(1rev,0.9m)5seconds 1.248kHz

1.84kHz

1stsecond

2ndsecond

1.22kHz

1.58kHz

3rdsecond

1.02kHz

1.87kHz

4thsecond

1.34kHz

1.41kHz

5thsecond

1.15kHz

1.46kHz

RefertoappendixIIforthespectrums
Fromthetableabove,wegettheaveragesforchannel1andchannel2respectively.
Frequencyturbulenceoccursatarangefrom1.15kHzto1.48kHz.Thoughthereare
certainanomalies,likeID5spectrumforthe1stsecondwherebyweareunableto
spottheturbulencefrequencyrangebecauseitdoesnotfulfilltheconditionthat
magnitudeofvibrationofchannel1ishigherthanchannel2.Thisoccurrencecanbe
duetovariousreasons.Thereasonswillbediscussedlater.Ifwetakenoteofthese
rangeatwhichturbulenceoccurs,weshouldbeabletoeffectivelyspotwhere
leakagesoccurbeliminatingsuchfrequencies.

28

5.ExperimentSetUp2(WithHole)
valve

hole

sealed

0.3m

0.7m
CH2

CH1

0.5m
1.3mpipe

Figure10:SchematicDiagramofSetUp2

Copperpipeof1inchdiameterwasusedtogetthefrequenciesofvibrationwhen
thepipeiswithhole.Accelerometersareplacedatbothendsofthepipetoobtain
thevibrationmotionofthepipeanddistancesatwhichaccelerometersareplaced
onthepipearevaried.A1mmholewasdrilledtostimulatewaterleakage.Thehole
isdrilled0.5mawayfromtheelbow.1mmholewaschosenasitisthesmallesthole
thatispossiblewiththeavailableequipments.Initially,abiggerholewasdrilledand
thepipewasvibrationvisiblyhardthusthewholeexperimentwasredonebydrilling
asmallerhole.Theendofthepipeissealedsonoisemadebythepipeshouldonly
beduetotheleakonly.Inthissecondpartoftheexperiment,wearetoinvestigate
thefrequencyrangeofleakage.

29

Fromtheresultsofsetup1,aslowerflowrateismoreappropriatewhenmeasuring
excitationofthepipeduetoleakage.Therefore,ouranalysisshallbelimitedto1to
2revolutionofwaterflowspeed.Ideally,therangeatwhichleakageoccursshould
besmallerthanturbulencefrequency.Theexcitationduetoturbulenceisgreater
thanleakage.

30

5.1ResultsforDifferentWaterFlowRates

Analysisisdoneincomparisonfor1revolutionand2revolutionwaterspeed.The
distanceatwhichtheaccelerometersarevariedfrom1mto0.8mdistance,itwill
notreallyaffectthefrequencyobtainasideallyrangeoffrequencyforleakage
shouldberoughlyconsistentforsamewaterflowrate.
Table5:RangeofFrequenciesfor1revolutionand2revolutionwaterflow

Rangeoffrequencyatwhichleakageoccurs/kHz

ID1(1rev,1mdist)

0.944

1.28

ID5(1rev,0.9mdist)

0.848

0.900

ID8(1rev,0.8mdist) 0.480

0.804

ID3(2rev,1mdist)

0.144

0.368

ID6(2rev,0.9mdist) 0.688

0.704

ID9(2rev,0.8mdist) 0.480

0.702

RefertoappendixIIIforthespectrums
Fromthedataabove,itlooksliketheresultsareveryinconsistent.Wearenotable
todistinguishadifferencebetweennoisecreatedbyturbulenceorleakage.Mostof
thetime,wehavetoassumethattheleakfrequencyislow.Often,themagnitudeof
channel1andchannel2coincides.

31

5.2LeakageFrequenciesforDifferentHoleSizes

1mmholewasdecidedasitisthesmallestpossibleholethatcanbedrilledintothe
pipe.Theholeissupposedtobeassmallaspossibleinordertosimulatelittle
excitationtothepipe.Whereasnoisemadebyturbulenceismuchhigherthanthe
leak.Intheexperiment,acomparisonwasdonebetween2holesizes,1mmand
1.5mmholerespectively.

Figure11:Frequencyof1mmhole

32


Figure12:Frequencyof1.5mmhole

Fromthe2spectrumsabove,1mmholehasanestimatedleakfrequencyof1kHzto
1.58kHz.Thoughthemagnitudeofchannel1isnothigherthan2butitisalsohighly
impossiblethattheleakfrequencyis2.5kHz.Theassumptionhereisthatthepipe
lengthisrelativelyshortsothereislittledampingeffect.Thissameassumptionis
assumedforfigure14.Theleakfrequencyisfrom688Hzto1.37kHz.
Thespectrumsresultsaredifferentfromwhatweareexpectedof.Thebiggerhole
leakageshouldoccuratahigherfrequencycomparedtothesmallerholeleakage.
Moreover,thevibrationmagnitudeforthesmallerholeisgreaterthanalargerhole
whichdefiesexpectation.

33

5.3Locatingconsistencyinleakagefrequency

Inordertocheckiftheleakagefrequencyislessthan1kHz,wewillanalysethetime
capturespectruminpersecondinterval.Theslowestwaterflowrateisusedfor
analysis.
Table6:RangeofLeakFrequencyfor1mmand1.5mmhole

Rangeofleakfrequency/kHz

ID1(1rev,1mmhole,1mdist)
1stsecond

0.128

2ndsecond

0.048

0.208

3rdsecond

0.048

4thsecond

0.064

5thsecond

0.048

1stsecond

0.528

0.704

2ndsecond

0.480

0.688

3rdsecond

0.528

0.672

4thsecond

0.528

0.816

5thsecond

0.528

0.688

ID26(1rev,1.5mmhole,0.9dist)

RefertoAppendixIVfor1mmholespectrums.
RefertoAppendixVfor1.5mmholespectrums.
Fromtheanalysis,theaveragefrequencyatwhich1mmholeleakageis67.2Hzto
208Hz.Whereastheleakageforthebigger1.5mmholeisfromarangeof412Hzto
714Hzwhichishigherthanthesmallerhole.Aswhatwehaveanalysedhereis
differentfromwhatwehavefoundoutearlieron.Perhaps,forlowerfrequencywe
havetouseatimecaptureforasmallerintervaltoobtainamorereliableresult.
34

5.4LocatingPointofLeakage

Inordertolocatepointofleakage,atimecaptureofspectrumsattheexactsame
timeintervalisneeded.Next,wehavetosieveoutacommonpatternbetweenboth
spectrumsfromchannel1and2toobtainthetimedifference.Thisisalsoknownas
correlation.Bynotingthetimedelayandmultiplyingitwiththespeedofsoundin
copperpipe,wecanlocatetheleakagepoint.Distancebetweenthe2
accelerometersneedtobenotedaswell.Throughsomeanalysis,wezoomedinto
thetimecapturespectrumtotheclearestpossibleview.

peakduetoleakage

Figure13:Timespectrumfor1revolutionflow,1mmhole

Aboveisthetimebasedspectrumfor1mmholeat1revolutionwaterflowrate,
from0.53sto0.55sinterval.Itseemslikeasimilarpatterncannotbefound,thus
unabletoobtainthetimedelay.Byzoomingintothepeaks,wenoticedthatthe
peakoccursatanearliertimeinchannel2comparedtochannel1andthisshould

35

notbethecaseaschannel1isnearertotheleakage.Channel1shoulddetectthe
leakagefirst.Hence,furtherexplainsthatthisresultiswrong.Differenttimingsof
thespectraisinvestigatedaswellbuttherearenoconvincingresults.

Figure14:TimeSpectrumfor1revolutionflow,1.5mmhole,channel1

Figure15:TimeSpectrumfor1revolutionflow,1.5mmhole,channel2

36

For1.5mmhole,bycomparingtheabove2timebasedspectrums,wecanestimate
toshiftthespectraforchannel2alongthisdirectiontoobtainthesamepatternas
channel1.Hence,therewillbeatimedelayof0.02s.Speedofsoundincopperis
3901m/s,hencetheholeisestimatedtobeatadistanceof78.02m,whichisvery
incorrect.
Throughseveralanalysis,wezoomedinthetimecapturespectrumtothesmallest
possibleavailablescale.However,wearestillunabletocorrelateorfindanysimilar
patterninbothchannel1and2.Iftherearenosimilarpatternavailable,wecannot
obtainthetimedelayneithercanwecalculatetheexactleakagepoint.Hence,
locatingleakpointisnotreallypossiblethroughthisexperimentset.Therearefar
toomuchvariationinwaterpressures,hencedeterminingthepeakduetoleakageis
difficult.Therearemanyreasonscanattributetothisoccurrence.Further
explanationwillbeelaboratedbelowunderthediscussion.

37

6.Discussion

Generally,mostspectrumshaveagreatrangeandlotsofthemarechosendueto
manyassumptions.Manyspectrumshaveinitialspikeinattenuationandtherefore
wefocusontheareawhichwethinkaresignificant.
Forthefirstexperiment,theturbulencefrequencieshoveraround1kHz.However,
thereweremuchinconsistenciesintheexperimentalresults.Generally,channel2
hasahighermagnitudethanchannel1.Thereisasuspicionthatchannel2collected
outletnoiseswhenwaterisgushingoutofthepipe.Thepeaksspectrumforboth
channel1and2arenotconsistent.Unknownpeaksatcertainhighandlow
frequenciesrangearespotted.
Forthesecondexperiment,itisdifficulttosieveouttheleakagefrequency.Either
thefrequencyistoolowsincethewholesetupisscaleddownorthat
environmentalnoisesarerecorded.Belowaresomeofthelimitationsofthe
experiments.

38

6.1Limitations
Numerouslimitationswerefacedwhengatheringthedatapresentedinthisreport.
Manylimitationsaffectedtherepeatabilityofthedatacollectedinvariousways.
Lackofequipmenttomimictheactualconditionsofwaterpipingsystemsalso
resultedinalossofaccuracyoftheresults.
Thelimitationsarelistedbelow:

Thelengthofthepipeistooshort;alongerpipewillbemorereliable.Wave
travelstooquicklyandpipegetsexcitedbysurroundingvibrationsveryeasily.

Lengthofpipeisonly1.3msotherecanbeverylittlevariationinthe
distancingthe2accelerometers

Waterflowspeedmaynotbeslowenoughtodetectdifferenceinturbulence
andleakages.

Noisesintheenvironmentmaybelouderthanthenoisecreatedbythe
turbulenceandpipesitself.

Waterpressuremaynotbeconsistentduringmeasurements

Equipmentsmaynotbesensitiveenoughtocapturelowfrequencies
measurements

Turbulencemaybecausedbythehalfopenedvalveratherthan90degree
elbow

39

Pipesarefreelysupportedandnotfullysupportalongitsaxiswillresultin
higherverticalamplitudetothepipeaxisarenotidealforleakagelocation
becausetheyarenarrowbandincharacterandpropagationvelocitymay
varywithfrequency.

Vibrationisnotcapturedfullybytheaccelerometersattimes(seefigure18
below)

Astheresultsarebadlyaffectedbythelimitations,filteringandconditioningofthe
signalshastobedoneinordertoseeanycorrelation.Thesefactorsarealsowhythe
resultspresentedmayappearseeminglyunconvincingandinconsistent.

Figure16:Vibrationpausedinbetweenmeasurements

40

6.2OtherPossibleErrors

Thoughtherearemanylimitationscouldaffecttheresultsofbothexperiment,
theremaybeotherpossibleerrorsthatresultedinunreliableresults.

Timewindowselectedforanalyzingofdata.Asdatacapturedmayhave

possiblepausesinbetween.Awrongtimeintervalisselectedwhenanalyzing.

Assumptionatwhichfrequencyofturbulenceandleakageoccursmaybe

wrong.Asalways,ouranalysismaybewrongandthusresultsareincorrect.Some
possiblefactorsanddetailsmaybeleftoutduetolackinexperience

Humanreactiontimeerrorwhenrecordingvibration.Numberofaverages

takenmaybeinsufficient.

Itisdifficulttocomeupwithatheoreticalspectrumofthewaterflowingina

pipe,hencelittlecomparisoncanbedone.Waterflowingthroughthepipeis
turbulent.

Valveand90degreeelbowareclosetogether,hence,difficulttoidentify

vibrationduetovalveorelbow.

41

7.Conclusion

Fromthisproject,thereisabetterunderstandingindetectingleakagesinpipe.Itis
evidentthatarangeoffrequencywasobtainedforbothleakageandturbulence.
Thoughsomeoftheexperimentalresultsdonotconcurwiththeoreticalexpectation
butthereweremanylimitationsanderrorthroughoutthewholeexperimental
process.Mucheffortisdoneintryingtoreducesucherrors,however,resultsmay
notbeveryconvincingunlessmoretestingisdone.
Perhaps,oneofthemoresignificantfindingsisthatturbulencefrequencyranged
around1kHzandthatleakagefrequencyisaround100to800Hz.Leakagefrequency
isdependentonholesize.Anotherfindingisthathighwaterflowspeedisnotgood
indeterminingbothleakageandturbulencefrequencyrange.Amuchlongerpipeis
neededtospaceouttheaccelerometer.
Lastly,inordertoobtainbetterresults,furtherimprovementscouldbedonetothis
experiment.Webelievethatthroughcontinualefforts,weareabletocorrectly
determineanddistinguishbetweenturbulenceandleakfrequency.

42

8.Recommendations

Therecommendedsolutionsarearrangedinorderofeaseofimplementation:
1) Installinganadditional/largeraccelerometertocheckconsistencyofdata.
2) Notedowntimingsatwhichtheenvironmentalconditionsaremore
favorable,quietestperiodtoconductdatacollection.Experimentscould
possiblybeconductedatnight.
3) Increasenumberofdatacollected.
4) Ifthelimitationsoffluctuatingpressurecannotbeavoided,wecaninstalla
pressuregaugemeteronthepipetorecordthepressurewhenthe
experimenttoplota3Dgraphforbettercorrelation.
5) Longerpipe,biggerdiameterpipecouldbeused.Actualpublicwatersystem
wouldbepreferred.
6) Collectresultofpipeburiedingroundratherthanpipewhichisfreely
supported.
7) Ensurepressureofwaterisconstantthroughoutthewholeexperiment.Have
apropervalveordigitalsystemofmeasuringwaterflowrate,inputofflow
meter.
8) Fluidsshouldnotcontainanyothersubstancessuchasairbubblesorsmall
particles.
9) Measuretheidealwaterflowratewhichresultsintheleastvibrationofthe
pipe.
43

AppendixISpectrumsforDifferentFlowRates

44

45

AppendixIILocatingforconsistencyinturbulencefrequency

46

47

48

49

50

AppendixIIILeakageFrequencyfor1&2revolutionwaterflow

51

52

53

AppendixIVLeakFrequencyfor1mmhole

54

55

56

AppendixVLeakageFrequencyfor1.5mmHole

57

58

59

Bibliography
Dunegan,H.(2004).LocationofLeaksinPipebyUseofAcousticEmissionModelRatio
Techniques.
Liston,D.L.(1992).Leakdetectiontechniques.JournaloftheNewEnglandWaterWorks
Association,103108.
M.JBrenna,P.J.SomeRecentResearchResultsontheuseofAcousticMethodstoDetect
WaterLeaksinBuriedPlasticWaterPipes.InstituteofSoundandVibrationResearch,
UniversityofSouthhampton.
R.Long,M.L.(2003).Attenuationcharacteristicsofthefundamentalmodesthatpropagate
inburiedironwaterpipes.Ultrasonics41,509519.
Riehle,H.V.(1991).TenYearsofExperiencewithLeakDetectionbyAcousticSignalAnalysis.
AppliedAcoustic33,119.

SeungYeupHyun,Y.S.Y.S.(2007).Thelaboratoryscaleddownmodelofaground
penetrationradarforleakdetectionofwaterpipes.MeasurementScienceandTechnology
18,27912799.

60

Potrebbero piacerti anche