Professional Documents
Culture Documents
WestphalianUniversityGelsenkirchen
WestphalianEnergyInstitute
InternalReportRevisionApril9
th
,2012
PrinciplesofLeakDetection
GerhardGeiger,Prof.Dr.Ing.,gerhard.geiger@whs.de
WestphalianUniversityGelsenkirchen
WestphalianEnergyInstitute
NeidenburgerStr.43
D45897Gelsenkirchen
Phone:+49(0)2864882680
Mobile:+49(0)1723703473
Preface
Preface
Pipelinenetworksarethemosteconomicandsafestmethodoftransportformineraloil,gasesand
otherfluidproducts.Pipelineshavetomeethighdemandsforsafety,reliabilityandefficiency.Most
pipelines,regardlessofwhattheytransport,aredesignedwithalifespanofaround25years.When
theydobegintofail,theydososlowlybeginningwithleaksatpoorconstructionjoints,corrosion
pointsandsmallstructuralmaterialcracks,andgraduallyprogresstoacatastrophicending.Butthere
arealsootherreasonsforleakdisasterssuchasaccidents,terrorism,sabotageandtheft.
Theprimarypurposeofleakdetectionsystems(LDSSystems)istoassistpipelinecontrollersinde
tectingandlocalizingleaks.LDSSystemsprovideanalarm,anddisplayotherrelateddatatothepipe
linecontrollersinordertoaidindecisionmaking.Pipelineleakdetectionsystemsarealsobeneficial
becausetheycanenhanceproductivityandsystemreliabilitythankstoreduceddowntimeandre
ducedinspectiontime.LDSSystemsarethereforeanimportantaspectofpipelinetechnology.
Thisreportpresentsanoverviewaboutthemostcommonlyusedprinciplesforleakdetection(and
leaklocalization).ThemainfocusisoninternalLDSsystems,whichutilizefieldinstrumentation(for
exampleflow,pressureandfluidtemperaturesensors)tomonitorinternalpipelineparameters.A
significantpartofthisreportisdedicatedtomodelbasedleakdetectionwhichisusuallycalledReal
TimeTransientModel(RTTM)basedLDS,andinparticularExtendedRTTMbasedLDSwhichcom
binescomputerbasedmodelingandsimulationtechniqueswithstatisticalleakclassification(orleak
signatureanalysis).Inthis,thesecondedition,chaptersaboutinstrumentationissues,datacommu
nication(SCADA)andleakmonitoringinshutinconditionshavebeenadded.Furthermore,en
hancementsofPipePatrolStatisticalLineBalance(SLB)haveledtosubstantialrevisionsoftherelated
chapter.
Acomparisonofallpresentedprinciplesandmethodsisalsoincludedinthetext,whichmayhelpthe
readertoselectaleakdetectionprinciplethatismostsuitableforaparticularapplication.Character
isticsarelistedonaninformativebasis,andwhiletheyaretosomeextentsubjectivebynatureevery
efforthasbeenmadeinthisdocumenttopresentobjectivefacts.
Finally,thereaderwillfindacomprehensivelistofdefinitionsthatarerelevantinthefieldofleak
detectionattheendofthisreport.
GerhardGeiger,Reken,Germany
geiger_g@tonline.de
Contents
ii
Contents
PrinciplesofLeakDetection.....................................................................................................................i
Preface......................................................................................................................................................i
Contents...................................................................................................................................................ii
ListofFigures...........................................................................................................................................iv
ListofTables............................................................................................................................................iv
SymbolsandUnits....................................................................................................................................v
1 Introduction....................................................................................................................................1
2 RegulatoryFramework...................................................................................................................3
2.1 TRFL(Germany).......................................................................................................................3
2.2 APIRP1130(USA)...................................................................................................................4
3 InstrumentationIssues...................................................................................................................6
3.1 PerformanceMeasures...........................................................................................................6
3.2 Flowmeters[ADEC].................................................................................................................7
4 SupervisoryControlandDataAcquisition(SCADA).......................................................................9
4.1 ComponentsofaSCADASystem.............................................................................................9
4.2 DataAcquisition....................................................................................................................10
4.3 DataCommunicationEquipment(DCE)................................................................................10
5 Pressure/FlowMonitoring............................................................................................................12
5.1 PressureMonitoring..............................................................................................................12
5.2 FlowMonitoring....................................................................................................................12
5.3 Summary................................................................................................................................12
6 RarefactionWaveMethod...........................................................................................................14
6.1 Summary................................................................................................................................14
7 BalancingMethods.......................................................................................................................16
7.1 SomeCommentsonDefinitions............................................................................................16
7.2 UncompensatedMassBalance.............................................................................................17
7.3 CompensatedMassBalance..................................................................................................18
7.4 UseofVolumetricFlowMeters.............................................................................................19
7.5 BalancingatStandardConditions.........................................................................................19
7.6 VolumeBalancing..................................................................................................................19
7.7 Summary................................................................................................................................20
8 StatisticalLeakDetectionSystems...............................................................................................22
8.1 HypothesisTesting................................................................................................................22
8.2 SignatureAnalysis..................................................................................................................23
8.3 Summary................................................................................................................................24
9 LeakMonitoringduringShutInConditions.................................................................................25
9.1 PTPressureTemperatureMethod.....................................................................................25
9.2 DPDifferentialPressureMethod.......................................................................................26
9.3 Summary................................................................................................................................26
10 LeakLocalization...........................................................................................................................28
10.1 GradientIntersectionMethod..............................................................................................28
10.2 TimeofFlightMethod..........................................................................................................28
11 RTTMRealTimeTransientModel.............................................................................................30
11.1 CompensationApproach.......................................................................................................32
11.2 FlowResidualApproach........................................................................................................33
11.3 PressureResidualApproach..................................................................................................34
12 PipePatrolStatisticalLineBalance(SLB).......................................................................................35
12.1 PipelineInventoryCompensation.........................................................................................35
12.2 StatisticalLeakClassification.................................................................................................36
12.3 Summary................................................................................................................................36
Contents
iii
13 PipePatrolExtendedRealTimeTransientModel(ERTTM).........................................................38
13.1 PipelineObserver..................................................................................................................38
13.2 LeakSignatureAnalysis.........................................................................................................38
13.3 LeakLocation.........................................................................................................................38
13.4 PipePatrolERTTM/PCLeakMonitoringinPumpingConditions.......................................39
13.5 PipePatrolERTTM/SCLeakMonitoringinShutinConditions..........................................43
13.6 Summary................................................................................................................................45
14 ComparisonofAllMethods..........................................................................................................48
14.1 FunctionalityandInstrumentation........................................................................................48
14.2 FieldofApplication................................................................................................................48
14.3 PerformanceParameters......................................................................................................48
A Bibliography..................................................................................................................................50
B Definitions.....................................................................................................................................51
AbouttheAuthor..................................................................................................................................64
ListofFiguresandTables
iv
ListofFigures
Figure1:SCADAsystemforanexampleapplication..........................................................................................................................................9
Figure2:Conditionalprobabilitydensityfunctions(PDF)...............................................................................................................................22
Figure3:LeakdetectionduringshutinbyPTmethod....................................................................................................................................25
Figure4:LeakdetectionduringshutinbyDPmethod....................................................................................................................................26
Figure5:Leaklocationbygradientintersectionmethod................................................................................................................................28
Figure6:RTTMtocalculatelocalprofiles;modelusingpressure(andtemperature)readings......................................................................31
Figure7:RTTMtocalculatelocalprofiles;modelusingflow(andtemperature)readings.............................................................................31
Figure8:CompensatedmassbalancewithRTTMbasedcompensation..........................................................................................................32
Figure9:Flowresidualapproach....................................................................................................................................................................33
Figure10:Pressureresidualapproach............................................................................................................................................................34
Figure11:PipePatrolStatisticalLineBalance(SLB)..........................................................................................................................................35
Figure12:PipePatrolERTTM/PCheadstationmonitoring...........................................................................................................................39
Figure13:PipePatrolERTTM/PCsubstationsmonitoring.............................................................................................................................40
Figure14:PipePatrolERTTM/PCsectionalmonitoring,substationsfullyinstrumented..............................................................................41
Figure15:PipePatrolERTTM/PCsectionalmonitoring,substationswithoutflowmeasurement................................................................42
Figure16:PipePatrolERTTM/SCheadstationmonitoring...........................................................................................................................43
Figure17:PipePatrolERTTM/SCsubstationsmonitoring.............................................................................................................................44
ListofTables
Table1:SymbolsandunitsPart1....................................................................................................................................................................v
Table2:SymbolsandunitsPart2...................................................................................................................................................................vi
Table3:Functionalityandinstrumentationofpressureandflowmonitoring.................................................................................................12
Table4:Fieldsofapplicationofpressureandflowmonitoring.......................................................................................................................12
Table5:Performanceparametersofpressureandflowmonitoring...............................................................................................................13
Table6:Functionalityandinstrumentationofrarefactionwavemethod.......................................................................................................14
Table7:Fieldsofapplicationforrarefactionwavemethod.............................................................................................................................14
Table8:Performanceparametersofrarefactionwavemethod......................................................................................................................15
Table9:Functionalityandinstrumentationforbalancingmethods................................................................................................................20
Table10:Fieldsofapplicationforbalancingmethods.....................................................................................................................................21
Table11:Performanceparametersforbalancingmethods.............................................................................................................................21
Table12:FunctionalityandinstrumentationforstatisticalLDS.......................................................................................................................24
Table13:FieldsofapplicationforstatisticalLDS.............................................................................................................................................24
Table14:PerformanceparametersforstatisticalLDS.....................................................................................................................................24
Table15:Functionalityandinstrumentationforleakmonitoringduringshutinconditions...........................................................................26
Table16:Fieldsofapplicationforleakmonitoringduringshutinconditions.................................................................................................27
Table17:Performanceparametersforleakmonitoringduringshutinconditions.........................................................................................27
Table18:FunctionalityandinstrumentationforPipePatrolSLB......................................................................................................................36
Table19:FieldsofapplicationforPipePatrolSLB............................................................................................................................................37
Table20:PerformanceparametersforPipePatrolSLB....................................................................................................................................37
Table21:FunctionalityandinstrumentationofPipePatrolERTTM................................................................................................................45
Table22:FieldsofapplicationofPipePatrolERTTM.......................................................................................................................................46
Table23:PerformanceparametersofPipePatrolERTTM..............................................................................................................................47
Table24:Comparisonofallmethods..............................................................................................................................................................49
SymbolsandUnits
SymbolsandUnits
Massflowingeneral kg/s
I
M
Massflowatinlet kg/s
Leak
M
Leakflow kg/s
O
M
Massflowatoutlet kg/s
, p P
Pressureingeneral Pa
P Probabilityingeneral
FA
P
Probabilityoffalsealarm
MA
P
Probabilityofmissedalarm
Q
Flowingeneral kg/s, m
3
/s
I
Q
Flowatinlet kg/s, m
3
/s
O
Q
Flowatoutlet kg/s, m
3
/s
s Onedimensionalcoordinatealongthepipeline m
Leak
s
Positionofleak m
t Time s
down
t
Downstreamarrivaltimeofrarefactionwavecausedbyaleak s
up
t
Upstreamarrivaltimeofrarefactionwavecausedbyaleak s
T Temperatureingeneral K
G
T
Temperatureofground K
V Volumeingeneral m
3
Pipe
V Volumeofthepipeline m
3
V
Volumeflowingeneral m
3
/s
I
V
Volumeflowatinlet m
3
/s
s
V
Volumeflowatstandardconditions sm
3
/s
O
V
Volumeflowatoutlet m
3
/s
x Actualmeasurementofanyphysicalvariable
x Flowresidualatinlet kg/s
ref
x
Referencevalueforcalculationofrelativemeasurementerror
true
x
Truevalueofanyphysicalvariable
y
Flowresidualatoutlet kg/s
z Pressureresidualingeneral Pa
Table1:SymbolsandunitsPart1
SymbolsandUnits
vi
Densityofthefluid kg/m
3
Meandensityofthefluidalongthepipeline kg/m
3
o
StandarddeviationoftheGaussiandistribution
Table2:SymbolsandunitsPart2
1 Introduction
1 Introduction
ModernleakdetectionsystemssuchasthePipePatrolfamilyfromKROHNEOilandGasmonitor
pipelinescontinuously,bylogicallytestingwhetheraleakhasoccurredornot(leakdetection).In
caseofaleaktheyalsocalculatetheleakflowandtheleakposition(leaklocalization).
Somecountriesformallyregulatepipelinesafety,forexampletheGermanrulesarelaiddownin
TechnischeRegelnfrRohrfernleitungsanlagen[TRFL].Othercountriesprovideguidelinesandrec
ommendedpractices,forexampleleakdetectionisspecificallyaddressedby[APIRP1130]inthe
USA.[APIRP1130]definesthefollowingimportantrequirementsofanLDS
1
:
Sensitivity:AnLDSshouldensurethatthelossoffluidasaresultofaleakisassmallaspossible.This
placestworequirementsonthesystem:itshoulddetectsmallleaks,anditshoulddetectthemquick
ly.PipePatrolExtendedRealTimeTransientModelERTTM(Chapter13)isabletodetectleakage
below1%ofnominalflowratewithinlessthanaminute.PipePatrolStatisticalLineBalance(SLB,
Chapter12)mayalternativelybeusedifdemandforsensitivityisreducedand/orleaklocalizationis
notrequired.
Reliability:TheusermustbeabletotrusttheLDS.Thismeansthatitshouldcorrectlyalarmanyreal
leak,butitisequallyimportantthatitdoesnotgeneratefalsealarms.PipePatrolERTTMtherefore
usesleaksignatureanalysisbasedonwellknownandprovenstatisticalprocedures.
Accuracy:TheLDSshouldreporttheleaklocation(andleakflow)accurately.Thisenablestargeted
actionsforrepairandreestablishingsafety.
Robustness:TheLDSshouldcontinuetooperateinnonidealcircumstances.Forexample,incaseofa
transducerfailurethesystemmustdetectthefailureandcontinuetooperate(possiblywithneces
sarycompromisessuchasreducedsensitivity).TakingPipePatrolERTTMasanexample,thisERTTM
basedLDSoptionallyusesthemuchsimplerPipePatrolStatisticalLineBalance(SLB)asabackupif
pressureortemperaturesensorsfail.
Theseperformancemeasurescanbefoundin[APIRP1130].Buttherearefurtherrequirements:
UniversalApplicability:AmodernLDSisexpectedtobeuniversallyapplicable.Forexample,PipePa
trolcanbeusedwithequalsuccessonliquidorgaspipelines,andoperateseffectivelyonsingle
productandmultiproductpipelineswithorwithoutseparationpigs.Dragreducingagent(DRA)cre
atessomechallenges,butnoinsurmountableproblems.
Wideoperatingrange:Leakdetectionisimportantthroughoutthewholeoperatingenvelope,includ
ingunsteady,ortransient,states.Itshouldcontinuetofunctioninhighlytransientstatessuchas
startupandshutdownandmoderatelytransientstatessuchasflowadjustmentaswellasthesteady
state.SomeLDSsalsoprovideamodelthatworksundershutinconditionssuchasPipePatrolE
RTTM,seeChapter13.
Accordingto[APIRP1130],LDSsaredividedintotwogroups:
- Externalsystemsusingdedicatedmeasurementequipment,suchasasensorcables
- Internalsystemsusingexistingmeasurementsensorsforflow,pressureandsoon
1
Thesedefinitionsformerlyhadbeenpartof[API1155],whichhadbeenwithdrawnbyAPI.Valuabledefinitionsanddiscussionsarenow
includedintoannexCfrom[APIRP1130].
1 Introduction
Externalsystemsprovideexcellentperformancebutinmanycasescannotoperatecontinuously.
Investmentandoperationalcostsareusuallyveryhighbecausetheyneeddedicatedmeasurement
equipmentsuchassensorcablesthatmustbelaidalongthepipeline.Forthatreason,externalsys
temswillonlybeusedincriticalapplications,forexamplewhenpipelinescrossnaturereserves.
Internalsystemsusuallyruncontinuously.Sensitivityisgenerallylower,butsoareinvestmentand
operationalcosts.Forthisreason,internalsystemsareverycommonandarerequiredbylawfor
mostpipelinesincountriessuchasGermany.Thefocusofthissurveythereforeisoninternalsys
tems.
InternalLDSsuseexistinginstrumentatione.g.forflow,pressureandtemperature.Instrumentation
issuesanddatacommunication(SCADA)topicsarethereforeveryimportantandarediscussedin
Chapters3and4,respectively.
ThissurveydescribestheprinciplesofinternalLDSandleaklocationasfollows:
- Chapter5: Pressure/FlowMonitoring
- Chapter6: RarefactionWaveMethod
- Chapter7: BalancingMethods
- Chapter8: StatisticalLeakDetectionSystems
- Chapter9: LeakMonitoringduringShutInConditions
- Chapter10:LeakLocalization
Thesemethodsandprincipleshavebeenwellknownfordecades.Greatimprovementsinthefieldof
computerhardwareandsoftwaremakeitpossibletosimulatesophisticatedtransientmodelsofthe
pipelineinrealtime.Chapter11thereforedescribesindetailprinciplesofusingRealTimeTransient
Models(RTTMs)forleakmonitoring.
Chapter12describesPipePatrolStatisticalLineBalance(SLB),whichcombinesthemassbalancewith
statisticalmethodsandoptionallyRTTMtechnology.Chapter13describesPipePatrolExtendedReal
TimeTransientModelERTTM,thepremiumleakdetectionsolutionofKROHNEOil&Gas.
Chapter14providesacomparisonofavailableleakdetectionmethodswhichmayhelpthereaderto
selectaLDSprinciplewhichismostsuitableforaparticularapplication.Listedinformationisprovid
edonaninformativebasisandsubjectivebynature.Lastbutnotleasttheinterestedreaderwillfind
withinAppendixBacomprehensivelistofdefinitionswhicharerelevantinthefieldofleakmonitor
ing.
2 RegulatoryFramework
2 RegulatoryFramework
2.1 TRFL(Germany)
GermanTRFLstandsforTechnischeRegelnfrRohrfernleitungsanlagen[TRFL],whichwasfirstly
publishedin2003andrevisedin2010.[TRFL]appliestomostGermanpipelinestransportingliquids
orgases.Itdemands:
a) Twoautonomous,continuouslyoperatingsystemsthatcandetectleaksinsteadystatecondi
tions.Thesetwosystemsmustrelyondifferentphysicalvariables.(Forgasandbrinepipelines
underwelldefinedoperatingconditions,onlyonesystemmaybenecessary.)
b) Oneofthesesystems,orathirdone,mustbeabletodetectleaksintransientconditions.
c) Onesystemtodetectleaksinpausedflowconditions.
d) Onesystemorproceduretodetectgradualleaks.
e) Onesystemorproceduretodetecttheleakposition.
2.1.1 InstallationsAccordingtoTRFLa)andb)
[TRFL]requirestwoautonomous,continuouslyoperatingLDSsthatcandetectleaksinthesteady
state.Eitherofthesesystems,orboth,orathirdone,mustbeabletodetectleaksintransientcondi
tions.
Specialattentionshouldbepaidtothedifferencebetweenthesteadystateandthetransientstate.
Steadystateconditionsexistwhenallrelevantphysicalvariablessuchasflow,pressure,temperature
anddensityaresufficientlyconstantalongthepipelinetoensurethatnowaveeffectscanbeob
served.Transientconditionsexistwhenthephysicalvariableschangesignificantlywithtime,sowave
effectsarepresent.Reasonsforthesechangesincludeproductcompressibilityandpipeelasticity,
togetherwithspecialoperationalconditionssuchas:
- Startingandstoppingpumpsorcompressorsduringstartupandshutdown
- Valveoperationanywherebefore,alongorbeyondthemonitoredpipelinesegment
- Floworpressurecontrolactions
- Changesoftargetthroughput
- Specialcasessuchascavitation
Experienceshowsthatgaspipelinesareusuallyina(moderately)transientstateasaresultofthe
highgascompressibility.Liquidpipelinesareoftenoperatedinasteadystate,butoncloseexamina
tiontransienteffectscanalsofrequentlybeobserved.
PipePatrolExtendedRealTimeTransientModel(ERTTM)fromKROHNEOil&Gasoffersastateof
theartleakdetectionsystemcapableofoperatinginbothsteadystateandtransientconditions;for
detailsrefertoChapter13.
2.1.2 InstallationsAccordingtoTRFLc)
[TRFL]requiresthateachpipelinehasonesystemtodetectleaksinpausedflowconditions,see
Chapter9.Inthiscontext,pausedflowjustmeans"flowequalorclosetozero";nothingissaidabout
howthiswillbeachieved.Iftheflowisblockedbyvalveslockingpressurewiththepipelinesegment,
itissaidtobeintheshutincondition.PipePatrolERTTMChapter13usesamodelbasedpressure
temperaturemethod(PTmethod),whichcanbeappliedtopipelinesinshutinconditions.
2 RegulatoryFramework
2.1.3 InstallationsAccordingtoTRFLd)
Gradualleaks(forexamplecausedbycorrosion)havetwoimportantcharacteristics:leakflowusually
is(very)small,anditdevelopsslowly.InstallationsaccordingtoTRFLa)arenotwellsuitedtodetect
ingthistypeofleak,so[TRFL]requiresadedicatedLDSforthispurpose.
AnexternalLDScanbeused,butthisisveryexpensive.InternalLDStechniquesincludingthepres
suretemperaturemethodChapterdescribedinChapter9.1anddifferentialpressuremethodde
scribedinChapter9.2canalsobeusedtodetectgradualleaks.
Ifadequate,leaktightvalvesarepresent,thePipePatrolERTTM/SCmodelbasedpressure
temperaturemethodcanalsobeapplied,seeChapter13.5.
2.1.4 InstallationsAccordingtoTRFLe)
TheTRFLadditionallyrequiresasystem(orotherprocedure)tolocateleaksrapidly,enablingtarget
edactionsforrepairandreestablishingsafety.Thisfunctioncanbeintegratedintooneofthesys
temsinstalledtocomplywithsectiona)forexample,PipePatrolERTTM,Chapter13.Detailsofleak
localizationaredescribedinChapter10.
2.2 APIRP1130(USA)
ThefirsteditionofAPI(AmericanPetroleumInstitute)RecommendedPractice(RP)1130Computa
tionalPipelineMonitoringforLiquidPipelineswaspublished2007[APIRP1130].APIRP1130does
notdirectlyimposelegalrequirementsonpipelineoperatorsinthesamewayasTRFL,but:
- Givesatechnicaloverviewofleakdetectiontechnology
- DescribesinfrastructuresupportforCPM
2
- DiscussesCPMoperation,maintenanceandtesting
Itprovidesthenecessarytechnicalinformationforconscientiousoperatorsandpipelinecontrollers
tomanagetheirpipelinessafely.
[APIRP1130]coversliquidpipelinesonly.LDSsaredividedintotwogroups:
- Externalsystemsusingdedicatedmeasurementequipment,suchasasensorcables.
- Internalsystemsusingexistingmeasurementsensorsforflow,pressureetc.AllLDSsintroduced
inthissurveyarepartofthisgroup.
[APIRP1130]alsodefinescriteria(ormeasures)forcomparingLDSsfromdifferentmanufacturers
3
:
- Sensitivity:Thesensitivityisacompositemeasureofthesizeofaleakthatasystemiscapableof
detecting,andthetimerequiredforthesystemtoissueanalarmintheeventthataleakofthat
sizeshouldoccur.Volumeormasslostbetweentheoccurrenceofaleakanditsdetectionisa
moreobjectivemeasureofperformancethanthesmallestdetectableleakflow.PipePatrolE
RTTMtypicallydetectsleakagebelow1%ofnominalflowrateinlessthanoneminute,resulting
inaleakvolumethatistypicallylessthan50liters.
- Reliability:Reliabilityisameasureoftheabilityofaleakdetectionsystemtorenderaccurate
decisionsaboutthepossibleexistenceofaleakonthepipeline,whileoperatingwithinanenve
lopeestablishedbytheleakdetectionsystemdesign.Itfollowsthatreliabilityisdirectlyrelated
2
CPM=ComputationalPipelineMonitoring.ThisisasoftwarebasedLDS.
3
Thesecriteriawereoriginallypublishedin[API1155]whichisbeenwithdrawnbyAPI.Usefuldefinitionsanddiscussionshavebeen
movedintoanAnnexof[APIRP1130].
2 RegulatoryFramework
totheprobabilityofdetectingaleak,giventhataleakdoesinfactexist,andtheprobabilityof
incorrectlydeclaringaleak,giventhatnoleakhasoccurred.
- Accuracy:Accuracycoversestimationofleakparameterssuchasleakflowrate,totalvolume
lost,typeoffluidlost,andleaklocationwithinthepipelinenetwork.Theseleakparameteresti
matesshouldbeasaccurateaspossible.
- Robustness:Robustnessisameasureoftheleakdetectionsystemsabilitytocontinuetofunc
tionandprovideusefulinformationevenunderchangingconditionsofpipelineoperation,orin
conditionswheredataislostorsuspect.Asystemisconsideredtoberobustifitcontinuesto
functionundersuchnonidealconditions.
3 InstrumentationIssues
3 InstrumentationIssues
Fortheinternalleakdetectionprinciplespresentedinthenextchapters,instrumentationincludes:
- Flowmeters
- Pressuresensors
- Temperaturesensors
[API1149]addressestheeffectofmeasurementuncertaintyonleakdetectionsystemsensitivity.It
considerssteadystateflowinliquidpipelinesusingpipecharacteristicstoestablishtheenvironment
inwhichthemeasurementsaremade.Itthenconsidersmeasurementaccuracyanduncertaintyfrom
theinstrumentvendorsinordertodeterminetheoreticalleakdetectionsensitivitybasedonmeas
urementquality.Theresultisatheoreticallimittoleakdetectionsensitivitybasedoninstrument
accuracyandpipelinecharacteristics.
3.1 PerformanceMeasures
Ingeneral,applyingaspecificLDSprincipleimpliesspecificperformancerequirementsfortherelated
measurementsystems.Thestaticperformanceofmeasurementsystemsismainlydefinedbytwo
measures:accuracyandrepeatability.Thedynamicperformanceisusuallydefinedbythesettling
time.
3.1.1 Accuracy
Accuracyisthedegreeofclosenessofmeasurementsxofaquantitytothatquantity'struevalue
x
tuc
.Formostmeasurementsystems,accuracyisspecifiedbydeclaringmaximummagnitudeval
uesfortheabsolutemeasurementerror
4
true
E x x .
Alternatively,accuracycanbeexpressedrelativetoareferencevalue:
true
ref ref
x x E
e
x x
= .
x
c]
indicatesachosenreferencevalue,usuallythemeasurementrange(x
c]
= x
mux
- x
mn
)or
thetruevalue(x
c]
= x
tuc
)
5
.Maximumerrorvaluesmaybespecifiedglobally,orasafunctionof
actualmeasurementx.
3.1.2 Repeatability
Repeatabilityistheclosenessofagreementbetweenindependentmeasurementreadingsobtained
withthesamemeasurementsystemunderthesameconditions(forexamplesameflowandsame
environmentalconditions).Veryoftenrepeatabilityisdefinedasthevaluebelowwhichtheabsolute
difference|x| |x|k] - x|k - 1]|betweentwosuccessivesinglemeasurementsx|k]and
x|k -1]obtainedunderthesameconditionsmaybeexpectedtoliewithaspecifiedprobability
suchas95%.
Repeatabilitydoesnotimplyaccuracy:theabsolutedifferencexmaybeverylow,butitispossible
foreverysinglemeasurementx|k]andx|k - 1]tobeincorrect.Ontheotherhand,accuracyimplies
repeatability:ifameasurementsystemisaccurate,x|k]andx|k - 1]showsmalldeviationsfrom
thetruevalue,andthereforethescatterofreadingsmustalsobesmall.Requiringagivenaccuracyis
thereforealwaysamoresevereconstraintthenrequiringagivenrepeatability.
4
Thisistrueonlyforadeterministicapproach.Withinastatisticalcontext,uncertaintyshouldbeusedinstead.
5
Inmanycasestheactualvaluexcanbeusedforx
tuc
asanapproximation.
3 InstrumentationIssues
3.1.3 SettlingTime
Accuracyandrepeatabilitydescribethestaticperformanceofameasurementsystem.Themostim
portantparameterdescribingthedynamicperformanceisthesettlingtimeI
s
,thetimerequiredfor
theresponsecurvetoreachandstaywithinarangeofcertainpercentage(usually5%or2%)ofthe
finalvalue.Fastmeasurementsystemsarecharacterizedbysmallsettlingtimes.
3.2 Flowmeters[ADEC]
Withtheexceptionofpressuremonitoringmethod(Chapter5.1),rarefactionwavemethod(Chapter
6)andleakmonitoringinshutinconditions(Chapter9),flowmetersarethemostimportantmeas
urementinstrumentsforleakdetection.Severaldifferenttypesofflowmeterareusedonpipelines
including:
- Orificeplates(differentialpressure),
- Turbinemeters
- Positivedisplacementmeters
- Directmassflowmeters(Coriolistype)
- Ultrasonicmeters
3.2.1 OrificePlate
Theflowmetersmostcommonlyinstalledonpipelinesaresharpedgedorificeplates,adifferential
pressuretypeofmeter.Althoughthesemetersareverycommoninapplicationssuchasthemeas
urementofnaturalgas,theiruseasaccurateinstrumentationforpipelineleakdetectionisquestion
able.Thebiggestproblemistheirmeasurementuncertainty.Thebasicuncertaintyinthedischarge
coefficientofawellinstalledorificeplateinfactoryconditionisaround0.5%.Thebestcaseuncer
taintyinafiscalgasapplicationwithallnecessarysecondarymeasurementstakenintoaccountis
closerto1%,conditionalonregularinspectionandoccasionalreplacementoftheplate.Ininstalla
tionsnottofiscalstandarditisreasonabletoexpectaccuraciesofaround35%.
Thereadingofanorificeplatedependsonthedensityaswellastheflow.Itiscommonpracticein
lowaccuracymeasurementstoperformasimplesquarerootextractioninatransmitterusingan
assumedconstantvalueofdensity.Thisisnotnormallysufficientlyaccurateandstableforleakde
tection,andaflowcomputershouldthereforebeconsidered.
3.2.2 TurbineMeter
Turbinemetersareflowmeasuringdeviceswithfanlikerotorsthatsensethevelocityofflowing
fluidinaclosedconduit.Thefluidforceontherotorcausesittorotateataratethatdependsmainly
onvolumetricflowrate,althoughalsoonfrictionforcesandotherfactors.Oversomerangeofcondi
tions,therateofrotationisnearlydirectlyproportionaltotherateofflow.Turbinemetersareused
extensivelyonpipelines,especiallythosecarryingpetroleumhydrocarbons.Thetotaltheoretical
uncertaintyofmassflowinafiscalturbinemetersystemisaround0.25%,althoughthisisanupper
boundandpracticalfiguresaslowas0.05%canbeachieved.Turbinesaresensitivetoviscosity,and
mosttypesperformlesswellaboveabout20cP.Theirsettlingtimeisfarfasterthanmightbeex
pected.
3.2.3 PositiveDisplacementMeter
Positivedisplacementmetersmeasureflowbydynamicallytrappingfluidparcelsofknownvolumein
oneormoremeasuringchambers.Thebiggestuncertaintyfactorwiththistypeofmeteristheun
measuredfluidslippagethroughitssealsandclearances.Thisvariesasthemeterwears,soregular
provingandfairlyfrequentservicingarebothimportant.Althoughgaspositivedisplacementmeters
exist,themainpipelineapplicationforthismetertypeisrelativelyviscousliquids.Inthiscase,the
expecteduncertaintyisagainaround0.25%forafiscalsystem.
3 InstrumentationIssues
3.2.4 CoriolisMassMeter
Coriolisdirectmassmetersareslowlygainingacceptanceandbeingincorporatedintothepipeline
industry.Theuncertaintyoftheseinstrumentsisapproximately0.5%ofreadingorbetter.Thistype
ofmeterhastheadvantageofprovidingdirectmassmeasurement,whichmeansthatadditional
measurementsoftemperature,pressure,anddensitymeasurementoranequationofstatetode
terminefluiddensityarenotnecessary.Coriolismeterscanbeappliedtobothliquidandgas,andare
insensitivetoviscosity.Theirprincipaldisadvantageisthattheyareverycumbersomeandexpensive
inmediumsizes,andlargesizesareonlystartingtobeavailable.Itisalsomoredifficulttochecka
massmeteragainstaprover,whichisbynatureavolumetricreference.
3.2.5 UltrasonicMeter
UltrasonicflowmetersareavailableinDopplerandTransitTimetypes,butonlythesecondofthese
iswidelyconsideredsuitableforpipelineapplications.TransitTimeultrasonicmetersrelyonaccu
ratetimingofultrasonicimpulsesthatcutacrossthepipediagonally,andwhosetraveltimeacross
themeterthereforedependsontherateofflow.Ultrasonicmetersdonotdirectlymeasuretheav
eragevelocitydistributioninthepipe,andsorequiremultiplepathstointegratethevolumeflow
properly.Generallymorepathsimplybettervelocitydistribution(orReynoldsnumber)compensa
tion,althoughsomemetershavemultipleequivalentpathswhosefunctionistocompensatefor
installationeffects.Ultrasonicmetersareavailableforbothgasandliquid.Thelatestgasultrasonic
technologygivesthebestperformanceofanymeasurementprinciple,withuncertaintyaslowas
0.2%.Liquidultrasonicmetersgiveuncertaintiessimilartoturbines,butwithahigherviscositycut
off.Oneofthemajoradvantagesofultrasonicmetersislittleornopressureloss.
4 SupervisoryControlandDataAcquisition(SCADA)
4 SupervisoryControlandDataAcquisition(SCADA)
InternalLDSsrequirefieldinformationsuchasflow,pressureandtemperaturethatwillbeprovided
bymeansofmeasurementstationslocatedatinlet,outletandintermediatestations.Measurement
stationsattheinletandoutletarecalledheadstations,intermediatestationsarecalledsubstations.
ThisfieldinformationwillusuallybeprovidedbyaSupervisoryControlandDataAcquisition(SCADA)
system;thisisacomputerbaseddatacommunicationsystemthatmonitors,processes,transmits,
anddisplayspipelinedataforthepipelinecontroller[ADEC].SCADAsystemsmaybeuseddirectlyfor
leakdetection,mayprovidesupportforanLDS,oranLDSmayoperateindependentlyofSCADA.
Generally,apipelineLDSwillusethedatageneratedbyaSCADAsystem.
4.1 ComponentsofaSCADASystem
Regardlessofthemanufacturer,SCADAsystemsaregenerallybasedonsimilarcomponents.
SubstationB SubstationA
HeadStation
Outlet
HeadStation
Inlet
ControlRoom
RS RS RS
MS
LDS
DCE
Figure1:SCADAsystemforanexampleapplication
Figure1showsanexamplewithtwoheadstations(inletandoutlet)andtwosubstations.Theinlet
headstationisapumpingstation,andtheoutletheadstationisavalvestation(forexamplewitha
pressurecontrolvalve).Twosubstationswithvalvespermitfastflowshutoffincaseofanemergen
cy.TheSCADAsystemcollectsrealtimedatafromfieldinstrumentsusingremotestations(RSs)at
substationsandoutletheadstation.RSsacquiredatafromfieldinstrumentssuchasflowmeters,
oftenviaanalogue420mAcurrentlooporcommunicationprotocolssuchasModbus.Programma
blelogiccontrollers(PLCs)usuallyformpartofRSswhenlocalmonitoringandcontrolfunctionsare
necessary.RSsaresometimescalledremoteterminalunits(RTUs)ifonlydataacquisitionisprovided.
Amasterstation(MS)attheinletheadstationisconnectedtotheRSsusingdatacommunication
equipment(DCE).Inmostcases,theMSprovidesanEthernetTCP/IPLocalAreaNetwork(LAN)in
ordertoconnectequipmentfordatapresentationandhumaninteraction;throughthis,thepipeline
controllermonitorsandcontrolsthesystem.
IntheexampleofFigure1theLDSacquiresrequiredfieldinformation(flow,pressure,temperature
andsoon)fromSCADAbutotherwiseoperatesindependently.DataistransferredfromSCADAtothe
4 SupervisoryControlandDataAcquisition(SCADA)
10
LDSusingadedicatedlink,e.g.Modbusoverserialconnection(RS232orRS485),Modbusover
EthernetTCP/IP,orOLEforProcessControl(OPC).NowadaysOPCisbecomingmoreandmorethede
factoindustrystandard.TheLDSperformsleakmonitoring;thediagnosticresults(leakno/yes,leak
flowandpositionincaseofaleak)aretransferredbacktotheSCADA.Usuallyresultsarepresented
tothepipelinecontrollerwhothentakestheappropriateactions.ForadditionaldiscussionofSCADA
systemdesignfactorsandtheireffectsonthequalityandtimelinessofthedatarequiredbyanLDS,
see[APIRP1130].
4.2 DataAcquisition
Dataacquisition(DAQ)convertsanaloguefieldsignalsintodigitalvaluesfordataprocessing.DAQis
partofRSandMSwhenfieldsignalsareanalogue,veryofteninformofa420mAcurrentloop.The
qualityoftheconversionheavilydependsontheDAQresolutionwhichisthesmallestincrementof
signalchangethatcanbedeterminedbythedevice,andthefullscalerangeofthefieldsignal.Ifthe
completefieldsignalrangeisassignedto420mAandtheDAQresolutionisspecifiedwithnbits,
thentheabsoluteresolutionisgivenby
20 4 16
2 2
n n
mA mA mA
E = =
yieldingarelativeresolutionof
1
20 4 2
n
mA mA
c
E
= =
.
ExperienceshowsthatDAQresolutionshouldben = 12bitorbetter.Itisalsoimportanttousethe
full420mArange.
4.3 DataCommunicationEquipment(DCE)
Datacommunicationequipment(DCE)includesphoneandradiomodemsaswellasfiberoptic,mi
crowaveandsatellitetransmissionequipment.MessagingbetweentheRSandMSisknownasthe
communicationsprotocolincludinghandshaking,errordetection,anderrorrecovery.Leakmonitor
ingissafetyrelevant;itthereforecouldbenecessarytoprovideredundantcommunicationlinksto
ensurethatleakmonitoringwillcontinueincaseofaconnectionfailure.
4.3.1 Topology
Topologyisthegeometricarrangementofnodesandlinksthatmakeupanetwork.ForaSCADA
system,thesetopologiesarecommonlyused:
- PointtoPoint.Thisisacommunicationlinkbetweenonlytwostations.Eitherstationcaninitiate
communicationwiththeother,oronestationcanactasmaster.
- PointtoMultipoint.Thisisacommunicationlinkamongthreeormorestationswithonestation
beingacommunicationarbitrator(master)thatcontrolswhentheotherstations(remotesta
tionsactingasslaves)cancommunicate.
- MultipointtoMultipoint.Thisisacommunicationlinkamongthreeormorestationswherethere
isnocommunicationarbitrator(master)andanystationcaninitiatecommunicationswithany
otherstation.
4.3.2 CommunicationProtocol
Acommunicationprotocolgovernstheformatofdatatransmissionbetweentwoormorestations,
includinghandshaking,errordetection,anderrorrecovery.Leakmonitoringissafetyrelevant,so
specialattentionisrequiredconcerningerrordetectionandrecovery.Transmissionmodemaybe
4 SupervisoryControlandDataAcquisition(SCADA)
11
halfduplex(informationissentinonedirectionatatimeoverthelink)orfullduplex(informationis
simultaneouslysentandreceivedoverthelink).
TheprotocolisconsideredpolledwhentheMSrequestsdatafromeachdeviceconsecutively.When
thelastdeviceisscanned,theMSwillautomaticallyrequestinformationfromthefirstone,creating
acontinuouspollingcycle.Thedurationofthiscycle,theSCADAsystempollingtime,hassteadily
improvedovertheyearsandcannowbelessthan0.25seconds.SCADAcommunicationsmayalsobe
nonpolled.Forexample,RSsmayreportwithoutbeingpolledonatimescheduledbasisorwhen
fieldconditionschange.
LDSsthatrelyontheSCADAsystemtoreceiveoperatingdataaredirectlyaffectedbythepolling
time.Longerpollingcyclestypicallytranslatetodegradedleakdetectionsensitivity.Mostmodern
SCADAsystemsincludequalitycheckingsoftwaretoassessthevalidityofthedatabeforeanycalcu
lationsarecomputedanddisplayed.FormodelbasedLDS,timetaggingmayberequired.
4.3.3 LinkMedia
Fieldinformationistransferredusingspecificphysicallinkmediasuchas:
- Publictransmissionmediasuchaspublicswitchedtelephonenetworks(PSTN),privateleased
lines(PLLs)anddigitaldataservices(DDS)
- Atmosphericmediasuchasmicrowaveradio,VHF/UHFradioandgeosynchronoussatellites
- Fiberopticalcable
Thechoiceoflinkmediadependson:
- Datatransmissionneedsoftheapplication
- Remotesiteandcontrolcenterlocations
- Distancebetweenstations
- Availablelinkmediaservices
- Projectbudget
5 Pressure/FlowMonitoring
12
5 Pressure/FlowMonitoring
Aleakchangesthehydraulicsofthepipeline,andthereforechangesfloworpressurereadingsafter
sometime[Krass/Kittel/Uhde].Localmonitoringofpressureorflowatonlyonepointcantherefore
providesimpleleakdetection.Itrequiresnotelemetry,forexampletocompareflowrateatinletan
outlet,aslocalmonitoringofpressureorflowrateissufficient.Itisonlyusefulinsteadystatecondi
tions,however,anditsabilitytodealwithgaspipelinesandmultiproductliquidpipelinesisextreme
lylimited.Itdoesnotprovidegoodsensitivity,andleaklocalizationisnotpossible.
5.1 PressureMonitoring
Ifaleakoccurs,thepressureinthepipelinewillfallbyanamountp.Aspressuresensorsarealmost
alwaysinstalled,itisnaturaltousethemforleakdetection.Thepressureinthepipelineissimply
comparedagainstalowerlimitafterreachingthesteadystatecondition.Whenthepressurefalls
belowthislowerlimit,aleakalarmisraised.
5.2 FlowMonitoring
Thesensitivityofthepressuremonitoringmethoddependsontheleaklocation.Neartheinletand
theoutletofthepipelinealeakleadstolittleornochangeinpressure.Thiscanbeavoidedbyflow
monitoring,wheretheflowismeasuredforchange.Thetwomethodscanbecombined.
5.3 Summary
Thenextsubchapterssummarizethemostimportantrequirementsandcharacteristics.SeeChapter
14foracomparisonofallmethodsintroducedinthissurvey.
5.3.1 FunctionalityandInstrumentation
Table3:Functionalityandinstrumentationofpressureandflowmonitoring
6
Bothmethodsprovideleakdetection,butnoleaklocalization.(Fordetailsaboutleaklocalizationsee
Chapter10.)Forpressuremonitoringonlyonepressuresensorisrequired,andforflowmonitoring
onlyoneflowmeterisrequired.Nostatisticalprocedureisinvolvedsoaccuracyisimportant(instead
ofrepeatability).Therearenospecialrequirementsconcerningthedynamictransmissionbehavior.
5.3.2 FieldsofApplication
Pumping Dynamics
Pressure/FlowMonitoring
PressureMonitoring PC,SC Steady Liquids+Gases a),c)
FlowMonitoring PC Steady Liquids+Gases a)
Application Method Medium TRFL
Table4:Fieldsofapplicationofpressureandflowmonitoring
7
6
LD=Leakdetection,P=Pressuresensor,Q=Flowsensor.
7
PC=Pumpingconditions,SC=Shutinconditions
5 Pressure/FlowMonitoring
13
Pressuremonitoringisabletodetectleaksinpumpingconditionsaswellasinshutinconditions;for
shutinconditions,thiswillbetrueifthepipelinevalvessealtightlyenough.(Forshutinoperation,
thereisamodifiedversioncalledthepressuretemperaturemethod,whichtakesintoaccountthe
actualfluidtemperature.FordetailsseeChapter9.)
Incontrast,flowmonitoringisonlyabletodetectleaksinpumpingconditions.Bothmethodsare
restrictedtothesteadystate,assmallchangesinpressureorflowwillcauseafalsealarm.Either
methodiscapableofmonitoringgasandliquidpipelines.Pressuremonitoringmeetsthefollowing
requirementsofTRFL:
- TRFLa),acontinuouslyworkingsystem,whichcandetectleaksinsteadystateconditions
- TRFLc),asystemtodetectleaksinshutinconditions
Incontrast,flowmonitoringonlyachievesTRFLa).
5.3.3 PerformanceParameters
Table5:Performanceparametersofpressureandflowmonitoring
Bothmethodswillworkwithoutmalfunctionifpressureandflowstaysufficientlyconstantindaily
operation.Thisistrueforsomeliquidpipelines,buthardlyeverforgaspipelines.
Thesesimplemethodsnormallydonotusesophisticatedstatisticalmethodstopreventfalsealarms
(Chapter8).Theonlywaytoavoidfalsealarmsisthereforetosetwidealarmlimits.Thiscausesa
shorttimetodetectaleakwithinliquidpipelines.Ingaspipelinespressurechangesareratherslow,
soleakdetectionisslow.Bothmethodsdetectsuddenleaksaswellasgradualleaksofsufficientsize.
6 RarefactionWaveMethod
14
6 RarefactionWaveMethod
Asuddenleakcaused,forexample,bycarelessuseofanexcavator,leadstoanegativepressure
wavepropagatingatthespeedofsoundcupanddownstreamthroughthepipeline.Suchawave,
calledararefactionwave,canberecognizedusinginstalledpressuretransmitters,givingaleak
alarm.Itisalsopossibletocalculatetheleaklocationbytimingthearrivalofthepressurewaveat
twoormorepointsonthepipeline(Chapter10).
6.1 Summary
Thenextsubchapterssummarizethemostimportantrequirementsandcharacteristics.SeeChapter
14foracomparisonofallmethodsintroducedinthissurvey.
6.1.1 FunctionalityandInstrumentation
Table6:Functionalityandinstrumentationofrarefactionwavemethod
8
Atleasttwopressuretransmittersareneededforleaklocalization;onepressuretransmitterallows
leakdetectiononly.(FordetailsaboutleaklocalizationseeChapter10.)Ineithercase,theselected
transmittersmustbecapableofdetectingrapidchangesinpressureandthereforemustbeveryfast
(verylowsettlingtime).Ontheotherhand,repeatabilityratherthanaccuracyisimportant.
6.1.2 FieldsofApplication
Pumping Dynamics
RarefactionWave(NegativePressureWave)
RarefactionWave PC,SC Steady Liquids a),c),e)
Application Method Medium TRFL
Table7:Fieldsofapplicationforrarefactionwavemethod
9
Therarefactionwavemethodisabletodetectleaksinpumpingconditionsaswellasinshutincon
ditions.Itisonlyabletodetectleaksinthesteadystatecondition,andsmallvariationsinpressure
caneasilyleadtofalsealarms.Rarefactionwavemethodsaremostusefulinliquidpipelines,aspres
surewavesarequicklyattenuatedingaspipelines.ThistechniquemeetsthefollowingTRFLrequire
ments:
- TRFLa),acontinuousworkingsystem,whichcandetectleakswithinsteadystateconditions
- TRFLc),asystemtodetectleaksinshutinconditions
- TRFLe),asystemorproceduretodetecttheleakposition
8
LD=Leakdetection,LL=Leaklocalization,P=Pressuresensor
9
PC=Pumpingconditions,SC=Shutinconditions
6 RarefactionWaveMethod
15
6.1.3 PerformanceParameters
Table8:Performanceparametersofrarefactionwavemethod
Thistechniquewillworkwithoutmalfunctionifpressurestayssufficientlyconstantindailyoperation,
whichistrueforsomeliquidpipelinesbuthardlyeverforgaspipelines.Sophisticatedstatistical
methodstopreventfalsealarms(Chapter8)willnotnormallybeused.Theonlywaytoavoidfalse
alarmsisthereforetosetwidealarmlimits.Thiscausesaveryshorttimetodetectaleakwithinliq
uidpipelines.Thismethodonlydetectssuddenleaksofsufficientsize.
7 BalancingMethods
16
7 BalancingMethods
Balancingmethodsarebasedontheprincipleofconservationofmass.Inthesteadystate,summed
overasufficientlylongtimeperiodt,themassH
I
enteringaleakfreepipelineatinletwillbal
ancethemassH
0
leavingitatoutlet.Inthemoregeneralcase,thedifferenceinmassatthetwo
endsmustbebalancedagainstthechangeofmassinventoryofthepipelineH
Ppc
.Overanygiven
periodoftimet,wecanthereforesay
0
I O Pipe I O Pipe
M M M M M M A A = A A A A = (1)
ifthereisnoleak,or,alternatively,
0
Pipe Pipe
I O I O
dM dM
M M M M
dt dt
= =
(2)
whendividingEq.(1)bytandt = Jt u.Eq.(2)istheinstantaneousversionofEq.(1).Here,H
andH
0
aremassflowatinletandoutlet,respectively,andJH
Ppc
Jt isthechangeofmassinvento
ryperunittime.Themassinthepipeisgivenby
Pipe Pipe
M V =
whereI
Ppc
isthevolumeofthepipeline,and
( )
0
1
L
s ds
L
}
isthemeanproductdensityalongthepipelineoflengthI.p(s)denotesthelocaldensityprofile
alongthepipelinewithspacecoordinates, u s I.
Anyadditionalmassimbalanceindicatesaleak.ThiscanbequantifiedbyrearrangingEq.(1)and
addingatermforleakmassyielding
Leak I O Pipe
M M M M A = A A A (3)
whereH
Lcuk
denotesthemasslostbytheleakduringt,or,usingEq.(2)
Pipe
Leak I O
dM
M M M
dt
=
(4)
whereH
Lcuk
denotestheinstantaneousleakmassflow.Theseequationsarevalidforliquidandgas
pipelinesinanyconsistentmassunits.
7.1 SomeCommentsonDefinitions
Balancingmethodsareverycommonsomanyreferencesaddressthistopic.Examplesofdefinitions
foundintheliteratureinclude:
- Massbalance
- Materialbalance
- Linebalance
- Volumebalance
- Modifiedorcompensatedvolumebalance
7 BalancingMethods
17
Unfortunately,someofthelisteddefinitionsaremisleading.Volumebalance,forexample,might
sometimesbeconfusedwithmassbalance.Butthereisnoprincipleofconservationofvolume,so
0
I O
V V A A = (5)
forleakfreepipelinesevenfort andidealsteadystateconditions.(I
I
andI
0
arethevol
umesenteringandleavingthepipelineduringt,respectively.)
Definitionsusedwiththissurveyconsiderstrictlythephysicalfactstherebyinsuringconsistency.
7.1.1 MassBalance
Balancingmethodsingeneralaremassbalancemethodsbecausetheprincipleofconservationof
massisused.Ifvolumeflowismeasuredinsteadofmassflow,thesemethodsgenerallyhavetocon
siderthedensityattheinletandtheoutlet,seeChapter7.4.Ifchangesinthemassinventoryofthe
pipelineH
Ppc
arenotconsideredthesearecalleduncompensatedmassbalancemethods(Chapter
7.2),andotherwisecompensatedmassbalancemethods(Chapter7.3).
7.1.2 VolumeBalance
Ifdensitiesp
I
andp
0
atinletandoutletareequal,then
0
I O
V V A A =
forleakfreepipelinesandt .Thisisonlytrueforsingleproductpipelineswherepressureand
temperatureatinletandoutletareequal.Themethodrelyingontheseassumptionsiscalledthe
volumebalancemethod,buttherestillisnorelationequivalenttoEq.(3)and(4),sochangeofinven
torycannotbecompensated.Anotherapproachuses
.
I O
V V V const A A = A =
whichmeansthatleakfreevolumeimbalanceisconsideredtobeconstant.Aleakchangestheoffset
Ileadingtoaleakalarmdeclarationusingstatisticalprocedures.Fordetailsonvolumebalancesee
Chapter7.6.
7.1.3 LineBalance
LineBalanceisagenerictermcoveringallbalancingmethods(massbalanceorvolumebalance).
7.2 UncompensatedMassBalance
Supposingthataleakwasallowedtocontinueforaninfinitelylongperiod,themassenteringand
leavingthepipelinewouldincreaseindefinitely.Themassinventoryofthepipeline,ontheother
hand,remainswithinafixedrange.H
Ppc
thereforebecomesnegligible,andEq.(3)reducesto
Leak I O
M M M A = A A (6)
whichisthebasicformulaofuncompensatedmassbalance.
Leak
M A isnowanestimateforthetrue
value
Leak
M A inEq.(3).Overafiniteperiodt,thisequationisanapproximation.Furthermore,er
rorsinmassflowcalculationmustbeconsidered,finallyyieldingtheleakdeclarationrule
Leak
No Leak
Leak I O
t
M M M
t
> A
A = A A
s A
(7)
whereyistheleakmassflowthreshold.Foroperationwithoutfalsealarms,thetimeperiodtmust
besufficientlylongfortheflowinandoutofthepipelinetobelargeincomparisonwiththechange
inpipelineinventory.Inthefollowingcases,averylargevaluewillberequired:
7 BalancingMethods
18
- Startingandstoppingpumpsorcompressorsduringstartupandshutdown
- Valveoperationanywherebefore,alongorbeyondthemonitoredpipelinesegment
- Floworpressurecontrolactions
- Changesoftargetthroughput
- Specialcasessuchascavitation
- Mostgaspipelines,mostofthetime
ForthecalculationofH
I
andH
0
inEq.(7),massflowsH
I
andH
0
willbeusedinmostcasesusing
M M t A = A
whereH
isthemassflowmeanvalueduringt.
7.3 CompensatedMassBalance
Unliketheuncompensatedmassbalance,thecompensatedmassbalancetakesaccountofchanges
inpipelineinventory;introducingH
Ppc
inEq.(7)thereforegives
Leak
No Leak
Leak I O Pipe
t
M M M M
t
> A
A = A A A
s A
(8)
whereinventorychangeH
Ppc
isgivenby
( )
0
1
L
Pipe Pipe Pipe
M V s ds V
L
(
= =
(
}
(9)
Itisnotpossibletodeterminethedensityprofilep(s)alongthepipelinedirectly.Threeindirectap
proachesaredescribedbelow.
7.3.1 MeasurementofPressureandTemperaturealongthePipeline
Aquantitynofpressurep
andtemperatureI
transmittersmustbeinstalledsufficientlyclosely,
1 i n.ThepipelineisthensplitintonsegmentsofknownvolumeI
ateachtransducerpair.
Foreverysegment,thedensityp
iscalculatedusingathermodynamicequationofstate
10
appropri
atetotheproduct.Equivalentrelationsforgasesarealsoavailable.Finally,meandensityp inEq.(9)
willbecalculatedusingp
.
7.3.2 DeterminationUsingaSteadyStateModel
Thereare(reasonablysimple)mathematicalmodelsavailableforliquidpipelinesaswellasforgas
pipelines.Inliquidpipelines,asimplelineardecreaseinpressurecansometimesbeassumedalong
thepipeline
11
;temperatureofthefluidcanbeassumedtoequalgroundtemperatureforlongpipe
lines.Correspondingflowequationsforgaspipelines,(assumingforexampleisothermaloradiabatic
flow)areavailable[Bohl].Usingthesestatemodelstogetherwiththecorrespondingequationof
statepermitsthecalculationofdensityprofilep(s)andpipeinventoryusingEq.(9).
7.3.3 DeterminationUsingaRealTimeTransientModel(RTTM)
Themostaccuratemethodistouseapipelinemodelthatcoverstransientaswellassteadystate
conditions.Thisallowsthedensitytobedeterminedateverypointalongthepipelineforthesteady
stateandtransientoperationseeChapter11.
10
EOS(forEquationOfState)orPVTequation(forP=Pressurep,V=SpecificVolumeu,T=TemperatureI)aresynonyms.
11
Thisassumesahorizontalpipelineofconstantinternalroughnessandcrosssectionalarea.Othercasesrequireamodifiedapproach.
7 BalancingMethods
19
7.4 UseofVolumetricFlowMeters
Itisnotalwayspracticaltomeasurethemassflowinandoutofthepipelinedirectlyforexample,
directmassmetersareonlyavailableinalimitedrangeofsizes.Itispossibletosubstitutevolumetric
flowmeters,buttheindicatedvolumeflowmustbemultipliedbylinedensitytoderivethemass.
Dependingontheapplication,theoptionsforobtainingdensityinclude:
- Thedensityoffluidsofknowncompositionisknownandconstant,andthereforecanbestored
inalookuptable
- Thedensitycanbedirectlymeasured
- Thedensityforcrudeoilanditsproductscanbedeterminedasafunctionofpressureandtem
peratureusing[APIMPMS11],providedthatareferencedensityisavailable
- ThedensityofnaturalgascanbecalculatedusingequationsofstatesuchasAGA8iftherelevant
parametersareknownorcanbemeasured
Conversioncalculationsfromvolumetricflowtomassflowforcommonlyusedvolumetricflowme
tersareusuallyimplementedinaflowcomputer.
7.5 BalancingatStandardConditions
Wherevolumetricflowmetersareused,itcanbeconvenienttoexpressthepipelinebalanceEq.(6)
intheformofstandardvolumeI
s
insteadofmassH.ThestandardvolumeofafluidofmassHis
definedasthefluidvolumeatsomefixedandagreedtemperatureI
s
andpressurep
s
,suchas
1.01325barand15C(ISOstandardconditions),andisgivenby
s
s
M
V
. (10)
Thestandarddensityp
s
issimplytheproductdensityatstandardconditions.Itisfixedforpure
products,andcanotherwisebecalculatedfrommeasurementsusinganequationofstate.Forun
compensatedmassbalance,usingEq.(7)togetherwithEq.(10)yields
, , ,
Leak
No Leak
s
s Leak s I s O
s
t
V V V
t
> A
A = A A
s A
(11)
whereI
s,I
andI
s,0
denotevolumeenteringandleavingduringtimeperiodtreferredtostand
ardconditions,andI
`
s,Lcuk
representstheleakvolumeatstandardconditions.y
s
isthecorrespond
ingalarmthreshold.Forcompensatedmassbalance,Eq.(11)canbemodifiedaccordingly.
7.6 VolumeBalancing
Balancingmethodsingeneralrelyonthemassconservationprinciple;flowmetersmusttherefore
providemassflowsdirectly(usingCoriolismeters)orindirectlyviavolumeflowcombinedwithpres
sureandtemperature,seeChapter7.4.Thereare,however,someapplicationswherevolumetric
flowmeterscanbeusedbyapplyingvolumebalancing
12
.
12
Intechnicalliterature,terms"massbalancing"and"volumebalancing"areoftennotusedconsistently.
7 BalancingMethods
20
7.6.1 ApproximatelyEqualDensityatInletandOutlet
Inthiscase,assumingsingleproductoperation,densityatinletandoutletisequal,sop
I
= p
0
= p ,
andEq.(7)becomes:
Leak
No Leak
Leak I O
t
V V V
t
' > A
A = A A
' s A
(12)
whereyistheleakvolumeflowthreshold.Thisonlyhappensiftemperature(andpressureforgases)
atinletandoutletareequal.Again,foroperationwithoutfalsealarms,thetimeperiodtmustbe
sufficientlylonggiventheflowinandoutofthepipeline,seeChapter7.2.ForthecalculationofI
I
andI
0
,volumeflowsI
I
andI
0
willbeusedinmostcasesusing V V t A = A
whereV
isthemean
volumeflowduringt.
7.6.2 BatchChangeforMultiProductPipelines
Duringabatchchangeformultiproductpipelines,theproductenteringthepipelineisnotthesame
astheproductleaving.Inthiscase,massflowsattheinletandoutletcandiffersignificantlyevenin
theabsenceofsignificanttransienteffects.ItmaythenbepreferabletousevolumebalanceEq.(12)
insteadofmassbalanceEq.(7).
7.7 Summary
Thissectionsummarizesthekeyrequirementsandcharacteristicsofthemethodsintroducedinthis
chapter.SeeChapter14foracomparisonofallmethodscoveredbythissurvey.
7.7.1 FunctionalityandInstrumentation
Table9:Functionalityandinstrumentationforbalancingmethods
13
Allbalancingmethodsrequireatleasttwoflowmeters,oneattheinletandtheotherattheoutlet.
Allmassbalancemethodsrequiremassflow,eitherdirectlyorindirectlymeasured,seeChapter7.4.
Volumebalancingrequiresvolumetricflowmeters.Allmethodsprovideleakdetection,butnoleak
location.(FordetailsaboutleaklocalizationseeChapter10.)Whenthechangeinpipelineinventory
iscompensated,additionalpressureandtemperaturesensorsarealsoneeded.
Therearenospecialrequirementsfordynamictransitionbehavior(settlingtime)exceptforRTTM
basedcompensationwhichrequiredfastinstrumentsforflowandpressuretofollowtransientef
fects.Formostversionsaccuracyisimportant(insteadofrepeatability).Volumebalanceonlyhasto
considerrepeatabilitybecausethismethodusesstatisticalprocedurestodetectchangesinthevol
umeimbalance.
13
LD=Leakdetection,Q=Flowsensor,T=Temperaturesensor,P=Pressuresensor,T
G
=Groundtemperaturesensor
7 BalancingMethods
21
7.7.2 FieldsofApplication
Pumping Dynamics
BalancingMethods
MassBalanceuncompensated PC Steady Liquids a)
MassBalancecompensated
DirectpandTmeasurement
PC Steady+LowTransient Liquids+Gases a)
MassBalancecompensated
Steadystatemodel
PC Steady+LowTransient Liquids+Gases a)
MassBalancecompensated
RTTM
PC Steady+Transient Liquids+Gases a),b)
VolumeBalance PC Steady Liquids a)
Application Method Medium TRFL
Table10:Fieldsofapplicationforbalancingmethods
14
Balancingmethodscanbeusedonlyinpumpingconditions:useinshutinconditionsisnotpossible.
Uncompensatedmassbalanceandvolumebalanceareonlyabletomonitorinsteadystatecondi
tions.Compensatedmassbalanceisabletomonitorforleaksinthepresenceofmoderatetransients.
Uncompensatedmassbalanceandvolumebalancearepracticallylimitedtoliquidpipelines;com
pensatedmassbalancecanmonitorgaspipelineswithsomesuccess.Balancingmethodsmeetthe
followingTRFLrequirements:
- TRFLa),acontinuouslyoperatingsystemthatcandetectleaksinsteadystateconditions
OnlytheRTTMcompensatedmassbalancemeetsthisTRFLrequirement:
- TRFLb),acontinuouslyoperatingsystemthatcantodetectleaksintransientconditions
7.7.3 PerformanceParameters
Table11:Performanceparametersforbalancingmethods
Allbalancingmethodsachieveamediumdetectionlimitwhereaccurateflowmeasurementsare
available.Uncompensatedmassbalanceandvolumebalancehavelongdetectiontimes,whilecom
pensationforchangeofinventoryhelpstoshortenthedetectiontime.RTTMcompensatedmass
balanceshowsthebestresults.Leakdetectiontimeislongerforgasesbecauseofthedynamiciner
tiaofpressureandflow.Allbalancingmethodsdetectsuddenleaksaswellasgradualleaksofsuffi
cientsize.
14
PC=Pumpingconditions
8 StatisticalLeakDetectionSystems
22
8 StatisticalLeakDetectionSystems
StatisticalLDSsusestatisticalmethodstodetectaleak.Thisgivesanopportunitytooptimizethe
decisionifaleakexistsinthesenseofsomechosenstatisticalcriteria.Howeveritmakesgreatde
mandsonmeasurements.Theyneedtobestationary(inastatisticalsense)forexample.Statistical
LDSsarethereforeverywellsuitedtosteadystateconditions,butputonlyprovidelimitedsensitivity
intransientstatesunlesstheyareadapted,forexampleusingaRealTimeTransientModels(RTTMs).
PipePatrolERTTM(Chapter13),forexample,combinesRTTMtechnology(Chapter11)withsophisti
catedstatisticalmethods.PleaserefertoChapter13formoredetailsonthistopic.
Statisticalmethodscanimprovetheperformanceofallleakdetectionmethodsintroducedinthis
survey.ThisChapterdescribesstatisticalLDSsbasedonuncompensatedmassbalance,Chapter7.2,
becausethesesystemsarecommon[Zhang].
8.1 HypothesisTesting
StatisticalLDSsbasingonhypothesistestingusemethodsandprocessesfromdecisiontheory[Bar
kat].Thehypothesistestforleakdetectionbasedontheuncompensatedmassbalanceuses
| | | | | |
I O
M i M i M i A
(13)
accordingtoEq.(2),whereH
|i]denotessampleioftheinstantaneousimbalancebetweeninlet
andoutletmassflow.Thesesamplescanbeusedtodecidebetweentwohypotheses,E
0
andE
1
:
0
1
: No leak
: Leak
H
H
Thestatisticalbehaviorofeveryindividualsampleisdescribedbyconditionalprobabilitydensity
function
( )
0
| p M H A
forhypothesisE
0
(noleak)and
( )
1
| p M H A
forhypothesisE
1
(leak).AGaussi
an(normal)distributionisgenerallyassumed:
Figure2:Conditionalprobabilitydensityfunctions(PDF)
8.1.1 LikelihoodRatioTest(ProbabilityRatioTest)
Instatistics,acauseleadstoapossibleresultwithsomeprobability.Howeverasetofresultsimplies
aparticularcausewithsomelikelihood.Forexample,ifacoincomesupheadssixtimesinarowit
wouldbepossibletocalculatethelikelihoodthatthecoinisbiased.Althoughthetermprobability
ratioissometimesusedforthetypeoftestdescribedhere,likelihoodratioismoreappropriate.
8 StatisticalLeakDetectionSystems
23
InsteadofusingonesinglesampleH
|i],awholecollectionofdata
| |
1 N i
M M M M i ( A A A A A
M
willnormallybeusedtoincreasetheperformanceofthehypothesistest.Thelikelihoodratiotestis
thengivenby
( )
( )
( )
1
1
0
0
|
(Leak)
(No Leak)
|
p H
H
H
p H
>
A
s
M
M
M
, (14)
where
( )
|
i
p H M
istheconditionalprobabilitydensityfunction(PDF)of AM
forhypothesisE
,
and
( )
A M
isthelikelihoodratioof AM
.Thevalueofymustbechosentosatisfyappropriate
statisticalcriteria,including:
- Probabilityoffalsealarms.Afalsealarmoccursifthereisaleakdeclarationbutthereisnoreal
leak.FalsealarmprobabilityP
PA
shouldbeassmallaspossible.
- Probabilityofmissedalarms.Amissedalarmoccursiftheisnoleakdeclarationbutthereisareal
leak.MissedalarmprobabilityP
MA
shouldalsobeassmallaspossibletoo.
Theproblemofchoosingyappropriatelymustbesolvedinallprobabilityratiotests.Detailscanbe
foundin[Kroschel]forexample.
AlikelihoodratiotestsuchasEq.(14)requiresaprioriknowledgeofPDFs
( )
|
i
p H M
.Inmostcas
es,includingleakdetection,thesePDFsarenotknowninadvance.Tosolvethisproblem,ageneral
izedlikelihoodratiotestcanbeusedinstead.Here,PDFs
( )
| ,
i i
p H M
haveparameters
i
suchas
themeanp
andstandarddeviationo
ofanormaldistribution.Thesecanbeestimatedbysome
statisticalprocedure,forexamplebyapplyingthemaximumlikelihoodmethod.Fordetails,referto
[Kay].
8.1.2 SequentialProbabilityRatioTest(SPRT)
Thelikelihoodratiotestabovereliesonevaluatingawholecollectionofdata AM
inasinglestep,so
applyingittoonlineleakdetectionschemeswoulddependonestablishingandmaintainingdata
buffers.Thisinconveniencecanbeavoidedusingasequentialprobabilityratiotest(SPRT),see
[Wald].Here,(AM
)willbecomputedrecursivelyyielding
| | ( ) | | ( ) | |
( )
1 , k f k k A A = A A A M M M
.
8.2 SignatureAnalysis
Themethodspresentedsofarcanbeimprovedbyusingstatisticalclassificationtechniquestoana
lyzethesignatureoffieldsignalsinfluencedbythepresenceofaleak.Asanexample,threedifferent
imbalancesignaturescanbedefinedfor:
a) Stepsignatureforsuddenleaks
b) Driftsignatureforgradualleaksorsensordrift
c) Allothertypesofsignature
8 StatisticalLeakDetectionSystems
24
Statisticalclassificationalgorithms(seee.g.[Kay])nowdeterminethesignatureof AM
byassign
menttoanyoftheclassesa),b)orc).Usingthistechniqueleadstoimprovedleakdetectionsensitivi
tywhileatthesametimepreventingfalsealarms.PipePatrolERTTMusesthisapproach,seeChapter
13.2.
8.3 Summary
Thissectionsummarizesthekeyrequirementsandcharacteristicsofthemethodsintroducedinthis
Chapter.SeeChapter14foracomparisonofallmethodscoveredbythissurvey.
8.3.1 FunctionalityandInstrumentation
Table12:FunctionalityandinstrumentationforstatisticalLDS
15
Statisticalmethodsbasedontheuncompensatedmassbalanceneedtwoflowmeters,oneatthe
inletandtheotherattheoutlet.Theyprovideleakdetection,butnoleaklocation(fordetailsofleak
localizationseeChapter10).Therearenospecialrequirementsfordynamicbehavior(settlingtime).
StatisticalLDSsonlyhavetoconsiderrepeatabilitybecausethesemethodsusestatisticalprocedures.
8.3.2 FieldsofApplication
Pumping Dynamics
StatisticalLDS
MassBalanceuncompensated
HypothesisTesting
PC Steady+LowTransient Liquids+Gases a)
Application Method Medium TRFL
Table13:FieldsofapplicationforstatisticalLDS
16
StatisticalLDSscanbeusedinpumpingconditions,butnotinshutinconditions.StatisticalLDSsare
abletooperateinmoderatelytransientstates,butwithincreasedleakdetectiontime.Statistical
LDSsprovidemoderateperformanceongaspipelines.Theymeetthefollowingrequirements:
- TRFLa),acontinuousworkingsystem,whichcandetectleakswithinsteadystateconditions
8.3.3 PerformanceParameters
Table14:PerformanceparametersforstatisticalLDS
StatisticalLDShavealowalarmlimit,buttimetodetectiscomparativelylong.Statisticalmethods
detectsuddenleaksaswellasgradualleaksofsufficientsize.
15
LD=Leakdetection,Q=Flowsensor
16
PC=Pumpingconditions
9 LeakMonitoringduringShutInConditions
25
9 LeakMonitoringduringShutInConditions
Therearetwomainpipelineconditions:
- Pumpingconditions,wheretheproductwillbetransportedbymeansoffluidflow
- PausedFlowconditions,wherefluidflowis(near)zero.
Thefocusofthemethodspresenteduptonowwasonpumpingconditions,butleakmonitoringin
pausedflowconditionsisalsoimportantifpausetimescannotbeneglected.Inparticular,thisap
pliestomultiproductpipelineswherepausetimesbetweentwobatchesmaybesignificant.Insome
applicationsvalveswillbeusedtoblockthefluidflowinthemonitoredsegment.Thisspecialpaused
flowconditionwillbecalledshutinorblockedlinecondition.
9.1 PTPressureTemperatureMethod
Inshutinconditions,valveswilllockapressureintooneormoresectionsofthepipeline.Itispossi
bleforconsiderablepressurechangestooccurinthiscaseasaresultofthermaleffects,butanyrap
idorunexpectedfallinpressureindicatesthataleakhasoccurred.
, p 0
0
/
/
p bar
C 0
p
0
Figure3:LeakdetectionduringshutinbyPTmethod
Afterclosingthevalves,duringtemperatureequalization,transienteffectsdecayandfluidpressurep
andfluidtemperature0approachtheirequilibriumvalues.Ifthereisnoleak,theresultingpressure
trendonlydependsonfluidtemperatureasshowninFigure3,wherethegreennoleaklineindicates
apressuredropasaconsequenceofatemperaturedrop.Leaktestingisperformedbybalancing
changesinthemeasuredpressureinthetestsectionagainsttheoreticalpressurechangescalculated
fromthemeasuredtemperatureinthetestsection.ThisisthereforecalledthePTmethod.
ThePTmethodcanbeusedforhydrostatictestingwherethepipelinesegmentisfilledwithwater.
Hydrostatictestingisregulatedinmanycountries,forexampleincaseofanewpipelineinstallation,
apipelinerelocation,replacementofexistingpipelinesegments,orwhenthereareotherchangesto
apipelinesystemwhichmayaffectintegrity[VdTV1051].
9 LeakMonitoringduringShutInConditions
26
9.2 DPDifferentialPressureMethod
Forpipelines,thePTmethodcanbeenrefinedtothedifferentialpressuremethod(DPmethod).
1 1
,
i i
p 0
,
i i
p 0
Figure4:LeakdetectionduringshutinbyDPmethod
Sometimeafterthevalvesareclosedthepressurep
andtemperature0
ofsegmenticomecloseto
theirequilibriumvalues.Wenowobservethedifferentialpressurepacrossthetightlyclosedvalve
V.Ifthereisnoleak,pstayssufficientlyconstant(greenline);incaseofaleak,thereisasignificant
pressuregradient(redline).UsingdifferentialpressureinsteadofabsolutepressureasforthePT
methodleadstohighersensitivitybutcanonlybeusedincaseintermediatevalvestationsarepre
sent.
9.3 Summary
Thissectionsummarisesthekeyrequirementsandcharacteristicsofthemethodsintroducedinthis
Chapter.SeeChapter14foracomparisonofallmethodscoveredbythissurvey.
9.3.1 FunctionalityandInstrumentation
Table15:Functionalityandinstrumentationforleakmonitoringduringshutinconditions
17
Bothmethodsprovideleakdetection,butnoleaklocalization.(Fordetailsaboutleaklocalizationsee
Chapter10.)ThePTmethodrequirespressureandtemperaturesensorsonbothsidesofthepipe
line,whereasDPmethodrequiresahighernumberofvalvestationswheredifferentialpressure
acrosstightlyclosedvaleswillbemeasured.Therearenospecialrequirementsconcerningthedy
namictransitionbehavior(settlingtime).Bothmethodsneedonlytoconsiderrepeatabilitybecause
thechangeof(differential)pressureisanalyzed.
17
LD=Leakdetection,T=Temperaturesensor,P=Pressuresensor,DP=DifferentialPressuresensor
9 LeakMonitoringduringShutInConditions
27
9.3.2 FieldsofApplication
Pumping Dynamics
LeakMonitoringduringShutInConditions
PTMethod SC Steady Liquids c),d)
DPMethod SC Steady Liquids c),d)
Application Method Medium TRFL
Table16:Fieldsofapplicationforleakmonitoringduringshutinconditions
18
Bothmethodsareabletodetectleaksinshutinconditions,ifthepipelinevalvessealtightlyenough.
Theyarerestrictedtosteadystateconditionsassmallchangesinpressurewillcauseafalsealarm.
Bothmethodsareparticularlyusefulforliquidpipelines,andmeetthefollowingrequirementsof
TRFL:
- TRFLc),asystemtodetectleaksinshutinconditions
- TRFLd),asystemorproceduretodetectgradualleaks
9.3.3 PerformanceParameters
Table17:Performanceparametersforleakmonitoringduringshutinconditions
Bothmethodsprovideasensitivealarmthreshold(inparticularwhenusedforgradualleakdetec
tion),buttimetodetecttheleakisverylong.Bothmethodsareabletodetectsuddenleaksaswell
asgradualleaksofsufficientsize.
18
SC=Shutinconditions
10 LeakLocalization
28
10 LeakLocalization
Whenaleakisdetected,itisimportanttolocateit.Anexactleaklocationgivestheopportunityto
takeswiftcontainingactiontominimizeharmtopeopleandtheenvironment.Localizedrepairscan
thenbecarriedoutcosteffectively.
10.1 GradientIntersectionMethod
Thegradientintersectionmethodisbasedonthefactthatthepressureprofilealongthepipeline
withitslengthLwillchangesignificantlyifaleakoccurs
19
.
I
V
O
V
dt
dV
Leak
Figure5:Leaklocationbygradientintersectionmethod
Thedashed,greenlineinFigure5showsthelinearpressuredropinaleakfreeliquidpipeline.The
pressureprofileismorecomplexforagaspipeline,butasimilarprincipleapplies.Ifaleakoccurs,
thepressureprofiledevelopsakinkattheleakpoint(continuous,redline).Theleaklocationcan
bedeterminedbycalculatingtheintersectionpointofthepressureprofilesupstreamanddown
streamoftheleak.Theclassicgradientintersectionapproachcalculatesthegradientofbothlines
usingtwopressurereadingsneartheinletandtwopressurereadingsneartheoutlet.Themodel
basedgradientintersectionmethod,asusedbyPipePatrolERTTMLDSChapter13,calculatesthe
twogradientswiththehelpoftherealtimetransientmodel,computedfromflowandpressure
measurementsatinletandoutlet.
Thismethodonlyachievesaccurateresultsifthepipelineisinthesteadystate.Thecauseordevel
opmentoftheleak(suddenorgradual)doesnotmatter.
10.2 TimeofFlightMethod
Asuddenleakcaused,forexample,bycarelessuseofanexcavator,leadstoararefactionwave
propagatingatthespeedofsoundcupanddownstreamthroughthepipelineofgivenlengthI.
Suchawavecanberecognizedusinginstalledpressuretransmitters,givingaleakalarm.Theleak
positioncanbedetermined
20
ifthemomentt
down
(downstream)andt
up
(upstream),whenthisneg
ativewavepassesthetransmittersismeasured.Settingt t
down
- t
up
,theestimatedleakloca
tionis:
19
Thisistrueforliquidpipelineswithconstantroughnessheightk
R
,aconstantcrosssectionA,andahorizontalbuiltpipeline.Themethod
mustbemodifiedinothercases.
20
Thespeedofsoundisnotconstantinliquidpipelinesundermultiproductconditionsoringaspipelines.Themethodhastobemodified
accordingly.
10 LeakLocalization
29
( )
1
2
Leak
s L c t = A . (15)
Thetimeofflightmethodneedsanidentifiablerarefactionwave.Resultswillbegoodifaleakis
sufficientlylargeandsudden.Smalland/orgradualleakscannotbelocatedbythismethod.Inpracti
caluse,itislimitedtosteadystateconditions.Itisabletolocateleaksinpumpingorinpausedflow
conditions.PipePatrolERTTMusestheRTTMbasedgradientintersectionmethod,whichcompen
satestransientsleadingtogoodresultseveninhighlytransientstates.
11 RTTMRealTimeTransientModel
30
11 RTTMRealTimeTransientModel
RTTMmeansRealTimeTransientModel.SomeLDSsofthePipePatrolLDSFamilybyKROHNEOil&
GasarebasedonRTTM,alsoknownasthePipelineObserver.TheKROHNEflagshipisPipePatrol
ExtendedRealTimeTransientModel(ERTTM),whichcombinesRTTMtechnologyusedforthere
sidualmethod(Chapter11.2andChapter11.3)withleaksignatureanalysistopreventfalsealarms,
Chapter13.PipePatrolStatisticalLineBalance(SLB,Chapter12)mayalternativelybeusedifdemand
forsensitivityisreduced,ifleaklocalizationisnotrequired,orasafallback.
RTTMsystemsbuildmathematicalmodelsoftheflowwithinapipelineusingbasicphysicallawssuch
as:
- Conservationofmass
- Conservationofmomentum
- Conservationofenergy
Whencombinedwithanequationofstate,introducedinChapter7,RTTMsystemsmodeltransient
andsteadystateflowinapipeline.Atransientstatemeansthatsuddenchangesinflow,pressure,
temperatureanddensitymayoccur.Thechangespropagateaswavesthroughthepipelinewiththe
speedofsoundcofthefluid.Forexample,transientstateoccursinapipelineduring:
- Startandstopofpumpsorcompressorsduringstartupandshutdown
- Valveoperationanywherebefore,alongorafterthemonitoredpipelinesegment
- Floworpressurecontrolaction
- Changesofthroughput
- Specialeffectssuchascavitation
Gaspipelinesarealmostalwaysinatransientstate,becausegasesareverycompressible.Evenin
liquidpipelinestransienteffectscannotbedisregardedmostofthetime.
AnRTTMmakesitpossibletocalculatemassflow,pressure,densityandtemperatureateverypoint
alongthepipelineinrealtimewiththehelpofmathematicalalgorithms.Thesesolutionsarecalled
localprofiles.
11 RTTMRealTimeTransientModel
31
F P T T
RTTM
T P F
P
I
Temperature(T)ofGround
Temperature(T)ofFluid
Pressure(P)
Flow(F)
Inlet Outlet
T
I
T
G
T
O
P
O
( )
M s
( ) p s ( )
s ( )
T s
Figure6:RTTMtocalculatelocalprofiles;modelusingpressure(andtemperature)readings
21
ThesimplestpossibilityforRTTMisshowninFigure6.Inthiscaseonlypressure(andtemperature)at
theheadstationsarefedintotheRTTM,alongwithgroundtemperatureifthepipelineisburied.
Calculationofthelocalprofilesneedsprocessmeasurementsattheinlet(subscriptI)andoutlet
(subscriptO)ofthepipelinethesepointsareknowntogetherastheheadstations.Variouscom
binationsofmeasurementarepossible,asweshallseeinamoment.Avalueofgroundtemperature,
T
G
isalsoneeded,assumingthatthepipelineisunderground.Ifgroundtemperaturecanbeassumed
constantalongapipelineinpractice,onesensormaybeusedtomeasurearepresentativevalue.
Itisalsopossibletoimplementthemodelusingflowattheheadstationsinsteadofpressure:
F P T
RTTM
T P F
Temperature(T)ofGround
Temperature(T)ofFluid
Pressure(P)
Flow(F)
Inlet Outlet
T
I
T
O
( )
M s
( ) p s ( )
s ( )
T s
T
T
G
I
Q
O
Q
Figure7:RTTMtocalculatelocalprofiles;modelusingflow(andtemperature)readings
22
21
Subscriptsareusedasfollows:"I"=inlet,"O"=outlet,"G"=ground.The"^"isusedtoindicatethatthevaluesarenotmeasured,but
calculated.Theadditionof(s)indicatesthatthesearenotsimplepointvalues,butprofilesandthereforefunctionsofthedistancealong
thepipeline.
22
denotesfloweitherasmassfloworvolumeflow
11 RTTMRealTimeTransientModel
32
11.1 CompensationApproach
ItwasmentionedinChapter7.3thatacompensatedmassbalancecalculationneedstointegratethe
densitypalongthepipelineinordertodeterminethemassinventoryofthepipe.TheRTTMpro
videsthenecessaryinformationtodosoaccurately,asshowninFigure8.
F P T T
RTTM
T P F
P
I
Inlet Outlet
T
I
T
G
( )
s
PipelineInventory
Compensation
I
M
O
M
Pipe
dM dt
+ -
-
Leak
M
P
O
T
O
Figure8:CompensatedmassbalancewithRTTMbasedcompensation
23
Inthisimplementation,theRTTMcalculatesthedensityprofilep(s)basedonpressureandtempera
tureattheheadstationsandgroundtemperature.PipelineinventorytheniscalculatedusingEq.(9)
( )
0
1
L
Pipe Pipe Pipe
M V s ds V
L
(
= =
(
}
whereI
Ppc
= A IdenotesthevolumeofthepipelineoflengthI,andp istheestimatedmean
densityalongthepipeline
24
.LeakswillbedetectedusingEq.(8)oralternatively
Leak
No Leak
Pipe
Leak I O
dM
M M M
dt
>
=
s
. (16)
Eq.(16)istheinstantaneousversionofEq.(8);Figure8isthereforeanexampleofancompensated
massbalancemethod,inthiscasewithRTTMbasedcompensation.
PipePatrolStatisticalLineBalance(SLB,Chapter12)usesthistechnologytogetherwithstatistical
methods(Chapter8).
23
Theflowneedstobemassflowhere;ifvolumeflowgiven,massflowhastobecalculatedseechapter0.
24
Forsimplicity,crosssectionAhadbeenassumedtobeconstantalongthepipeline.
11 RTTMRealTimeTransientModel
33
11.2 FlowResidualApproach
TheflowatheadstationsinFigure9isnotnecessarytocalculatethelocalprofiles,aspressureis
usedforthispurpose,seeFigure6.TheRTTMcalculatesflowattheendpointsofthepipeaswellas
everywhereelse.Itisthereforepossiblechecktothedifferencebetweenmeasuredandcalculated
flow.Adifferencebetweenthetwoindicatesachangeinthedynamicsofthepipelineinother
words,asuspicionthattheremaybealeak.
F P T T
RTTM(noleak)
T P F
P
I
Inlet Outlet
T
I
T
G
I
M
O
M
+
-
P
O
T
O
I
M
O
M
y x
+
-
Flow-Residuals
Figure9:Flowresidualapproach
25
BothoftheFlowResidualscanbeusedasleakindicators:
I I
O O
x M M
y M M
ThenoleakhypothesisH
0
istrueiftheindicatedflowsagreesufficientlycloselywiththemodel.The
leakpresenthypothesisH
1
istrueifthereisapositiveresidualattheinletand/oranegativeresidual
attheoutlet.Mathematically:
0
1
: No leak 0, 0
: Leak 0, 0
H x y
H x y
~ ~
> <
Weinsistontheappropriatesignsfortheresidualsbecauseapositiveresidualattheoutlet,forex
ample,wouldindicatethatmorefluidwasleavingthepipelinethanexpected.Inotherwords,the
casesx<0andy>0wouldindicateanegativeleak.Thistellsussomethinginterestingaboutthe
performanceoftheinstrumentsorthevalidityoftheRTTM,butitisnotaphysicallyrealisticbasisfor
declaringaleakalarm.
PipePatrolERTTM(Chapter13)usesthistechnologytogetherwithstatisticalmethods(Chapter8)
forheadstationmonitoringduringpumpingconditions,seeChapter13.4.1.
25
Theflowhereneedstobemass;ifvolumeflowgiven,massflowhastobecalculatedseesection0.
11 RTTMRealTimeTransientModel
34
11.3 PressureResidualApproach
Ifapipelineislongenough,substationswithpressuresensorswilloftenbeincluded.Theindicated
pressurescanbecomparedwiththosecalculatedusingtheRTTMmethod,givingpressureresiduals
asfollows:
, 1
i i i
z p p i n s s
NotethatinFigure10temperatureandflowmeasurementatthesubstationsareunnecessary.
F P T
RTTM(noleak)
T P F
PI
Inlet
TI TO PO
PressureResiduals
P P
1
p
1
p
2
p
2
p
z
1
z
2
Substation1 Substation2 Outlet
SegmentS2 SegmentS1 SegmentS3
T
TG
Figure10:Pressureresidualapproach
26
ThenoleakhypothesisH
0
istrueiftheindicatedpressuresagreesufficientlycloselywiththemodel.
TheleakpresenthypothesisH
1
istrueifthereisanegativeresidual.Mathematically:
0
1
: No leak 0
: Leak 0
i
i
H z
H z
~
<
Again,weinsistontheappropriatesignfortheresidualbecauseapositiveresidualwouldindicatea
negativeleak,whichimpliesaninstrumentproblem.
26
Forabetterpointofviewonlytwosubstationsareshown.Themethodisabletohandleasmuchsubstationaspresentatthepipeline.
12 PipePatrolStatisticalLineBalance(SLB)
35
12 PipePatrolStatisticalLineBalance(SLB)
PipePatrolStatisticalLineBalance(SLB)combinesbalancingmethods(Chapter7)withstatisticalclas
sificationmethods(Chapter8).
I
Q
Pipe
dM dt
O
Q
Leak
Q
Figure11:PipePatrolStatisticalLineBalance(SLB)
27
PipePatrolSLBoffersoptionstohandledifferentapplicationneeds.Instrumentsshownwithred
symbolsinFigure11aremandatory;instrumentsshownwithgreensymbolsareusedifrequiredfor
aspecificapplication.
Dependingonthechosenoptions,PipePatrolSLBcanbeclassifiedasabalancingmethod(Chapter7)
and/oranRTTMbasedmethod(Chapter11).Itappliescompensatedmassbalancing(Chapter7.3)or
uncompensatedmassbalancing(Chapter7.2)oralternativelyvolumebalancing(Chapter7.6).Esti
matesoftheleakmassorvolumewillbeanalyzedbyastatisticalleakclassificationimplementing
themethodsofastatisticalLDS(Chapter8).
IncontrasttoPipePatrolERTTM(Chapter13),PipePatrolSLBdoesnotprovideleaklocalization.If
thereishighdemandforsensitivity,PipePatrolERTTMshouldalsobeusedinstead.Ontheother
hand,PipePatrolSLBmaybeusedwithoutpressureandtemperaturesensorsifInventoryCompensa
tionisomitted;PipePatrolSLBcanthereforeserveasabackupsystemforPipePatrolERTTMincase
ofafailureofapressureortemperaturesensor.
12.1 PipelineInventoryCompensation
PipePatrolofferstwodifferentpossibilitiesofpipelineinventorycompensation:
12.1.1 RTTMBasedCompensation(CompensatedMassBalanceLDS)
Initsmostsophisticatedversion,PipePatrolSLBappliesRTTMbasedinventorycompensationusing
thecompensationapproach(Chapter11.1).ItisthenacompensatedmassbalanceLDSrequiringfull
instrumentation:directorindirectmassflowmeters,pressureandtemperaturesensorsatinletand
outlet.ItthenprovidessensitivityclosetothatofPipePatrolERTTM.
27
denotesfloweitherasmassflow(forcompensatedanduncompensatedmassbalance)orvolumeflow(forvolumebalance).Ifneces
sary,thetransformationofthevolumeflowtomassflowisdonewithinPipePatrolseeChapter0.
12 PipePatrolStatisticalLineBalance(SLB)
36
12.1.2 NoCompensation(UncompensatedMassBalanceorVolumeBalanceLDS)
Inventorycompensationoptionallycanbeomittedcompletely;PipePatrolSLBisthenanuncompen
satedmassbalanceLDS.Thisreducessensitivitysignificantly,butrequirementsforinstrumentation
arealsoreduced:onlyflowmetersattheinletandoutletarerequired.Inthisconfiguration,PipePa
trolSLBthencanserveasabackupsystemforPipePatrolERTTMincaseofafailureofapressureor
temperaturesensor.
Volumebalancingusingvolumetricflowmetersisalsopossible,seeChapter7.6;PipePatrolSLBis
thenavolumebalanceLDS.Thisisparticularlyusefulduringbatchchangeformultiproductpipe
lines,wheredifferentproductsareenteringandleavingthepipeline.Inthatcase,massflowsatinlet
andoutletwoulddiffersignificantlyevenintheabsenceofsignificanttransienteffects.
12.2 StatisticalLeakClassification
AnLDSthatgeneratesfalsealarmscannotbetrusted,soitisakeytasktoeliminatethem.PipePatrol
SLBpreventsthemusingstatisticalleakclassification(seeChapter8),whichexecutesaftertheinven
torycompensation.Thisbooststhereliabilityandtherobustnessofthesystemwithoutcompromis
ingsensitivity.Falsealarmsareprevented,evenwithlowalarmthresholds.
12.3 Summary
Thissectionsummarizesthekeyrequirementsandcharacteristicsofthemethodsintroducedinthis
Chapter.SeeChapter14foracomparisonofallmethodscoveredbythissurvey.
12.3.1 FunctionalityandInstrumentation
Table18:FunctionalityandinstrumentationforPipePatrolSLB
28
Allconfigurationsneedatleasttwoflowmeters,oneattheinletandtheotherattheoutlet.They
provideleakdetection,butnoleaklocation.(FordetailsaboutleaklocalizationseeChapter10.)If
thechangeinpipelineinventoryiscompensated,additionalinstruments(pressureandtemperature
sensors)arerequired.PipePatrolSLBusesstatisticalleakclassification;instrumentrepeatabilityis
thereforeofprimaryinterest,ratherthanaccuracy,whichreducesspecificationrequirements.In
strumentationmustbefast(lowsettlingtime)forRTTMbasedcompensatedmassbalanceonlyin
ordertofollowtransienteffectscorrectly.
28
LD=Leakdetection,Q=Flowsensor,T=Temperaturesensor,P=Pressuresensor,T
G
=Groundtemperaturesensor
12 PipePatrolStatisticalLineBalance(SLB)
37
12.3.2 FieldsofApplication
Pumping Dynamics
PipePatrolStatisticalLineBalance(SLB),KrohneOil&Gas
MassBalancecompensated PC Steady+Transient Liquids+Gases a),b)
MassBalanceuncompensated PC Steady+LowTransient Liquids a)
VolumeBalance PC Steady+LowTransient Liquids a)
Application Method Medium TRFL
Table19:FieldsofapplicationforPipePatrolSLB
29
Allconfigurationscanbeusedinpumpingconditions,butuseundershutinconditionsisnotpossi
ble.Allconfigurationsarecapableofhandlingmoderatelytransientstatesandliquidpipelines,but
heavilytransientstatesandgaspipelinescanonlybehandledbyaddingRTTMbasedinventorycom
pensation.Volumebalancewillbeusedmainlyformultiproductliquidpipelines.Allconfigurations
ofPipePatrolSLBarecapableofoffering:
- TRFLa),acontinuouslyoperatingsystem,whichcandetectleaksinsteadystateconditions
Ifpipelineinventorycompensationisused,PipePatrolSLBisalsoabletooffer:
- TRFLb),acontinuouslyoperatingsystem,whichisabletodetectleaksintransientstates
12.3.3 PerformanceParameters
Table20:PerformanceparametersforPipePatrolSLB
Allconfigurationsoffermediumtolowalarmthresholds.Timetodetectaleakislongwithoutpipe
lineinventorycompensation,butshortenssignificantlyifitisavailable.Timetodetectaleakislonger
forgaspipelinesbecauseofthedynamicinertiaofpressureandflow.Allconfigurationsdetectboth
suddenandgradualleaksofsufficientsize.
29
PC=Pumpingconditions
13 PipePatrolExtendedRealTimeTransientModel(ERTTM)
38
13 PipePatrolExtendedRealTimeTransientModel(ERTTM)
PipePatrolERTTMisKROHNEsflagshipLDS,whichfusestheRTTMtechnologydescribedinChapter
11withleaksignatureanalysisdescribedinChapter13.2inauniquemanner.Forthisreasonitis
calledExtendedRTTM[Geiger].PipePatrolERTTMisabletomonitorpipelines:
- Duringpumpingconditions(PipePatrolERTTM/PC,PumpingConditions,seeChapter13.4)
- Duringshutinconditions(PipePatrolERTTM/SC,ShutInConditions,seeChapter13.5).
13.1 PipelineObserver
PipePatrolERTTMusestheRTTMresidualapproachespresentedinChapter11.2(forpumpingcon
ditions)andChapter11.3(forshutinconditions).WithinthecontextofPipePatrolERTTM,the
RTTMmodulesinFigure9andFigure10arecalledpipelineobserver
30
,becausetheyareobserving
thepipelinebycalculatingthelocalprofilesforflow,pressure,fluidtemperatureanddensity.
UseoftheRTTMpipelineobservercompensatesthetransientbehaviorofthepipeline.Eveninheavy
transientstates(forexampleduringpipelinestartup)residualsstayclosetozeroinleakfreecondi
tions.Sensitiveleakdetectionisthereforepossibleintransientstates.
13.2 LeakSignatureAnalysis
AnLDSthatgeneratefalsealarmscannotbetrusted,soitisakeytasktoeliminatethem.PipePatrol
ERTTMusesleaksignatureanalysis(seeChapter8.2),whichexecutesafterthepipelineobserver,to
preventthem.Inthissecondstageresidualsareanalyzedforleaksignatures:
- Suddenleak.Thisclassicalleakdevelopsquicklyforexamplebyexternaldamagetothepipe
line.Itcausesadynamicsignatureinresiduals.Whensuchaleakrecognized,aleakalarmwillbe
reportedandtheleaklocationandleakflowaredetermined.
- Sensordriftorgradualleak.Thesemayoccurbycontaminationoftheflowmeterorbysmall
leakscausedbycorrosion.Theyresultinindistinguishable,slowsignatures.Whendriftisrecog
nized,asensoralarmisreportedandtheapparentleakflowisdetermined.
Thisbooststhereliabilityandtherobustnessofthesystemwithoutcompromisingsensitivityand
accuracy.Falsealarmsareprevented,evenwithlowalarmthresholds.
13.3 LeakLocation
PipePatrolERTTMlocatesleaksusingtwomethods,bothintroducedinChapter10:
- Modelbasedgradientintersectionmethod
- Modelbasedtimeofflightmethod
Themodelbasedgradientintersectionmethod([Billmann])calculatesanestimates
Lcuk
usingthe
amplitudesofthemassresidualsxandyattheheadstations.Themodelbasedtimeofflightmeth
odanalysesresidualsxandytodetectastep.Isasteprecognizeddownstreamattimet
down
iny
andrecognizedupstreamattimet
up
inx,theleaklocationcanbedeterminedbythetraveltime
differencet t
down
-t
up
.
30
SometimesalsocalledVirtualPipeline.
13 PipePatrolExtendedRealTimeTransientModel(ERTTM)
39
13.4 PipePatrolERTTM/PCLeakMonitoringinPumpingConditions
Flowispresentduringpumpingconditions,sofloweitherasvolumeflowormassflowisavailable.
Thisisalsotrueforunblockedpausedflowconditionswhenflowisclosetozero.
13.4.1 HeadStationMonitoring
ThisbasicschemeappliestheRTTMflowresidualapproach(Chapter11.2).
F P T
PipelineObserver
T P F
Inlet Outlet
Segment
LeakSignatureAnalysis
x y
Leak
Signatures
LeakAlarm
SensorAlarm
Leakflowandlocation
PI TI
I
M
I
M
O
M
TO PO
O
M
T
TG
Figure12:PipePatrolERTTM/PCheadstationmonitoring
31
TheexampleinFigure12showsheadstationmonitoringbasedontheflowresidualapproachde
scribedinChapter11.2.WiththehelpoftheRTTM,PipePatrolERTTMcomparesthemeasuredflow
atinletandoutletwiththecalculatedflowassumingaleakfreepipeline.Theflowresiduals,which
areusedbytheleaksignatureanalysis,are
I I O O
x M M y M M
yieldingaleakflowestimate[Billmann]
Leak
M x y =
wherechangeofpipelineinventoryisimplicitlyconsidered.UseoftheRTTMpipelineobserverthere
forecompensatesfortransientbehaviorofthepipeline.Eveninheavilytransientconditions(for
exampleduringpipelinestartup)x,yand
Leak
M
stayclosetozeroinleakfreeconditions.Sensitive
leakdetectionisthereforepossibleintransientconditions.
Thedynamicfreeresidualsarenowpassedtothesecondstage,theleaksignatureanalysis.Itstasks
accordingtoChapters13.2and13.3areto:
- Managealarms
- Determineleakrateandleaklocation
31
Ifnecessary,thetransformationofthevolumeflowtomassflowisdonewithinPipePatrolseeChapter13.4.5.
13 PipePatrolExtendedRealTimeTransientModel(ERTTM)
40
Ifpressureand/ortemperaturesensorsfail,thepipelineobserverinFigure12cannotcontinueto
calculate;inthatcase,PipePatrolSLBChapter12.1.2serveasabackupLDSbutwithreducedsensi
tivityandwithoutleaklocalization.
13.4.2 SubstationMonitoringwithoutFlowMeasurement
UsingthepressureresidualmethodChapter11.3allowsforhandlingsubstationswithpressure
measurementsasshownbelow.
F P T
PipelineObserver
T P F
PI
Inlet
TI TO PO
P P
1
p
1
p
2
p
2
p
Substation1 Substation2 Outlet
SegmentS2 SegmentS1 SegmentS3
LeakSignature
Analysis1
LeakSignature
Analysis2
SubstationEvaluation
z
1
z
2
T
TG
Leak
Signatures
Leak
Signatures
LeakAlarm
SensorAlarm
Leakflowandlocation
Figure13:PipePatrolERTTM/PCsubstationsmonitoring
32
TheRTTMpipelineobserverusespressureandtemperaturesensorsattheheadstationstocalculate
thelocalprofiles,includingthepressureprofilealongthepipe.Anydiscrepancybetweenthecalcu
latedandtheobservedpressureatthesubstationsindicatesachangeinthepipelinedynamics:in
otherwords,aleak.Thepressureresidualateachstationis:
, 1
i i i
z p p i n = s s
Theseareusedbytheleaksignatureanalysistodetectaleakandfinditslocation.Thiskindofpipe
linemonitoringiscalledsubstationmonitoring.Thepipelineobservercompensatesforanytransient
behaviorofthepipeline.
Thecompensatedresidualsarenowpassedtothesecondstage,theleaksignatureanalysis.Itstask
istomanagealarmsforeachindividualsubstation.Resultsofleaksignatureanalysisarecombinedby
substationevaluation,whichgroupssubstationalarmsanddeterminesleakflowrateandlocation.
Usually,headstationmonitoring(Figure12)andsubstationmonitoring(Figure13)willbecombined
toprovideleakmonitoringforpipelinesectionsconsistingofanumberofsegmentsformedbyhead
stationsandsubstations.
32
Twosubstationsareshownforclarity,butthemethodisabletohandleanynumberofsubstations.
13 PipePatrolExtendedRealTimeTransientModel(ERTTM)
41
13.4.3 SectionalMonitoringforSubstationswithFlowMeasurement
TheexampleinFigure14showsapipelinesectionconsistingoftwosubstationswithflowmeasure
ment.
I
M
(1)
O
M
(2)
I
M
(1)
I
M
O
M
(2)
O
M
(3)
I
M
(3)
O
M
1
M
2
M
Figure14:PipePatrolERTTM/PCsectionalmonitoring,substationsfullyinstrumented
33
Whereflowmeasurementisavailableatsubstationsinadditiontopressureandtemperature,the
pipelinecanbedividedintoindependentsegmentsasshowninthediagram.IndependentRTTM
pipelineobserversandERTTMLeakClassifiersmaybeappliedinparalleltoeverysegment,each
usingthemethodsalreadyintroducedinearliersections.Theshorterlengthofthemonitoredsec
tionscomparedtotheoveralllengthofthepipelineleadstoseveraladvantages:
- Significantlylowersmallestdetectableleakrate
- Significantlyshortertimetodetectaleak
- Significantimprovementinaccuracyofleaklocation
Thesegmentevaluationchoosesthesegmentthatshowsthemostsignificantleaksignature,deter
mineswhetheraleakalarmorasensoralarmispresent,andreportstheleaklocationandflowif
appropriate.
ThismethodachievesbetterperformancethanthemethodshowninFigure13,especiallyongas
pipelines.Thedisadvantageisthecomplexinstrumentationneededatthesubstations.
33
Twosubstationsareshownforclarity,butthemethodisabletohandleanynumberofsubstations.Ifnecessary,thetransformationof
volumeflowtomassflowisdonewithinPipePatrolseeChapter13.4.5.Groundtemperatureisomittedforclarity.
13 PipePatrolExtendedRealTimeTransientModel(ERTTM)
42
KROHNEOilandGasisabletoconfigurethismethodtoassignpipelinesegmentsdynamically.For
example,intheeventofatransmitterfailureatsubstation1itispossibletoskipthisstationand
performleakdetectionbetweentheinletstationandsubstation2.
13.4.4 SectionalMonitoringforSubstationswithoutFlowMeasurement
Thedisadvantageofthecomplexandexpensiveinstrumentationfromchapter13.4.3canbeelimi
natedbyvirtualflowmeasurement,asshowninFigure15.
I
M
(1)
O
M
(2)
I
M
(1)
I
M
O
M
(2)
O
M
(3)
I
M
(3)
O
M
Figure15:PipePatrolERTTM/PCsectionalmonitoring,substationswithoutflowmeasurement
34
ThefunctionalityisnearlythesameasshowninFigure14,exceptthatdirectflowmeasurementis
replacedbyvaluescalculatedintheRTTMpipelineobserverfromaneighboringsegment.
EachRTTMpipelineobservercalculatesflowateverypointalongitsassociatedsegment,including
theinletandtheoutlet.Themeasuredflowattheheadstationsiscomparedwiththecalculatedflow
intheusualway.Atintermediatestations,thecalculatedoutletflowforthesegmentupstreamis
simplycomparedwiththecalculatedinletflowforthesegmentdownstream,andviceversa.Asonly
oneRTTMpipelineobservercalculatestheflowattheheadstations,arealflowmeasurementisstill
neededattheselocationsforcomparison.
Iftemperaturemeasurementsatsubstationsarenotpresent,calculatedfluidtemperaturesfroma
neighboringsegmentcanbeusedinstead.
34
Twosubstationsareshownforclarity,butthemethodisabletohandleanynumberofsubstations.Ifnecessary,thetransformationof
volumeflowtomassflowisdonewithinPipePatrolseeChapter13.4.5.Groundtemperatureisomittedforclarity.Temperaturemeas
urementforsubstationscanbereplacedbygroundtemperaturemeasurementinmostcases.
13 PipePatrolExtendedRealTimeTransientModel(ERTTM)
43
13.4.5 FlowCalculation
PipePatrolERTTM/PCbasicallyrequiresmassflowasshowninFigure12,Figure14andFigure15,
butitisnotalwayspracticaltomeasurethemassflowinandoutofthepipelinedirectlyforexam
ple,directmassmetersareonlyavailableinalimitedrangeofsizes.Itisalsopossibletousevolu
metricflowmeters,whereusuallyflowcomputerscalculatethemassflowbycombiningmeasured
volumeflowwithothermeasurements.Correspondingflowcalculationsarestandardized.
Forspecialcases,wherevolumetricflowmetersareusedbutflowcomputersarenotavailablefor
conversionfromvolumeflowtomassflow,PipePatrolisabletoperformtherequiredcalculations
internally,e.g.using[APIMPMS11].
13.5 PipePatrolERTTM/SCLeakMonitoringinShutinConditions
PipePatrolERTTM/PCpresentedinChapter13.4islimitedtopumpingconditionswherenonzero
flowispresentorunblockedpausedflowconditionswhereflowisclosetozero.Ifvalvesareusedto
blocktheflow,thepipelineisinshutinorblockedlineoperation,seeChapter9;inthiscase,PipePa
trolERTTM/SCshouldbeusedinstead.PipePatrolERTTM/SCisamodelbasedpressure
temperaturemethod,whichisvalidforbothliquidandgaspipelines.Therelevantvalvesmustbe
leaktight,andthisshouldbeconsideredwhenchoosingthem.
13.5.1 HeadStationsMonitoring
ThisbasicschemeappliestheRTTMpressureresidualapproachChapter11.3.
P T
PipelineObserver
T P
Inlet Outlet
Segment
LeakSignatureAnalysis
zI zO
Leak
Signatures
LeakAlarm
SensorAlarm
Leakflowandlocation
TI
I
p
I
p
0
O
M =
TO
O
p
T
TG
0
I
M =
O
p
Figure16:PipePatrolERTTM/SCheadstationmonitoring
Itisclearwithoutmeasurementthattheflowattheinletandoutletshouldbezero.TheRTTMpipe
lineobservercanusethistocalculatethelocalprofiles,includingtheexpectedpressureatthetwo
headstations.Itispossibletocomparethesewiththemeasuredvalues,givingpressureresidualsz
I
andz
O
:
I I I O O O
z p p z p p
TheRTTMpipelineobserverisabletocompensatetransientbehaviorofthepipelineinshutincondi
tions.Inaddition,thethermodynamicequationofstateintheRTTMcompensatesfortheeffectof
temperaturechangesonpressure.
13 PipePatrolExtendedRealTimeTransientModel(ERTTM)
44
ThecompensatedresidualsarepassedtotheERTTMleaksignatureanalysis,asintroducedinChap
ter13.2.Leakrateandleaklocationwillbedeterminedwhennecessary.
13.5.2 SubstationsMonitoring
Substationmonitoringforshutinissimilartosubstationmonitoringforpumpingconditions.Details
canbefoundinChapter13.4.2.
P T
PipelineObserver
T P
PI
Inlet
TI TO PO
P P
1
p
1
p
2
p
2
p
Substation1 Substation2 Outlet
SegmentS2 SegmentS1 SegmentS3
LeakSignature
Analysis1
LeakSignature
Analysis2
SubstationEvaluation
z
1
z
2
T
TG
Leak
Signatures
Leak
Signatures
LeakAlarm
SensorAlarm
Leakflowandlocation
0
I
M =
0
O
M =
Figure17:PipePatrolERTTM/SCsubstationsmonitoring
35
Thepressureprofilefortheentirepipeiscalculatedusingonepipelineobserver.Calculatedpres
surescanthenbecomparedwithmeasuredvaluesatthesubstations.
Iftherelevantvalvesarecompletelyleaktight,evenverysmall,gradualleakswillberecognized.In
thiscase,PipePatrolERTTM/SCmeetstherequirementsofTRFLd).
Usually,headstationmonitoring(Figure16)andsubstationmonitoring(Figure17)willbecombined
toprovideleakmonitoringforsectionsconsistingofanumberofsegmentsformedbyheadstations
andsubstations.
35
Twosubstationsareshownforclarity,butthemethodisabletohandleanynumberofsubstations.
13 PipePatrolExtendedRealTimeTransientModel(ERTTM)
45
13.6 Summary
Thissectionsummarizesthekeyrequirementsandcharacteristicsofthemethodsintroducedinthis
Chapter.SeeChapter14foracomparisonofallmethodscoveredbythissurvey.
13.6.1 FunctionalityandInstrumentation
Table21:FunctionalityandinstrumentationofPipePatrolERTTM
36
PipePatrolERTTMprovidesleakdetectionandleaklocationinbothpumpingandshutinconditions
(fordetailsofleaklocalizationseeChapter10.)PipePatrolERTTM/PCHeadstationmonitoringre
quiresmeasurementsofflow,pressureandtemperatureattheheadstations.Ifgroundtemperature
isreasonablyconstantalonganundergroundpipeline,asinglerepresentativemeasurementatsome
pointalongthepipelineissufficient.Flowmetersatsubstationsarenotnecessary,butsubstations
mustbeequippedwithpressuresensors.Sectionalmonitoringrequiresflow,pressureandtempera
tureforthesubstations,butusingvirtualflowtechnologyrequirementsforthesubstationscanbe
reducedtopressureandtemperatureorevenpressurealone.
PipePatrolERTTM/SCHeadstationmonitoringneedsmeasurementsofpressureandtemperatureat
theheadstations.Again,groundtemperatureatsomepointalongthepipelineissufficient.ForSub
stationmonitoring,substationsmustbeequippedwithpressuresensors.
PipePatrolERTTMusesleaksignatureanalysis;instrumentrepeatability,ratherthanaccuracy,is
thereforethemainconcern,whichreducestherequirements.Instrumentationmustbefast(low
settlingtime)inordertofollowtransienteffectscorrectly.
36
LD=LeakDetection,LL=LeakLocation,Q=Flowsensor,T=Temperaturesensor,P=Pressuresensor,T
G
=Groundtemperaturesensor
13 PipePatrolExtendedRealTimeTransientModel(ERTTM)
46
13.6.2 FieldsofApplication
Pumping Dynamics
PipePatrolERTTM,KrohneOil&Gas
PipePatrolERTTM/PC
HeadStationsMonitoring
PC Steady+Transient Liquids+Gases a),b),e)
PipePatrolERTTM/PC
SubstationMonitoring
PC Steady+Transient Liquids+Gases a),b),e)
PipePatrolERTTM/PC
SectionalMonitoring
PC Steady+Transient Liquids+Gases a),b),e)
PipePatrolERTTM/PC
SectionalMonitoringVirtualFlow
PC Steady+Transient Liquids+Gases a),b),e)
PipePatrolERTTM/SC
HeadStationsMonitoring
SC Steady+LowTransient Liquids+Gases c),d),e)
PipePatrolERTTM/SC
SubstationMonitoring
SC Steady+LowTransient Liquids+Gases c),d),e)
Application Method Medium TRFL
Table22:FieldsofapplicationofPipePatrolERTTM
37
PipePatrolERTTM/PCprovidesleakdetectioninpumpingconditions;PipePatrolERTTM/SCprovides
leakdetectioninshutinconditions.Allconfigurationsareabletomonitorinsteadyand(moderately)
transientstates.Evengaspipelinescanbemonitoredwithoutproblems.PipePatrolERTTM/PCper
formsthefollowingfunctions:
- TRFLa),acontinuouslyfunctioningsystem,whichcandetectleaksinsteadystateconditions
- TRFLb),acontinuousworkingsystemwhichisabletodetectleaksintransientstate
- TRFLe),asystemorproceduretodetecttheleakposition
PipePatrolERTTM/SCperformsthefollowingfunctions:
- TRFLc),asystemtodetectleaksinshutinconditions
- TRFLd),asystemorproceduretodetectgradualleaks
- TRFLe),asystemorproceduretodetecttheleakposition
37
PC=PumpingConditions,SC=ShutinConditions.
13 PipePatrolExtendedRealTimeTransientModel(ERTTM)
47
13.6.3 PerformanceParameters
Table23:PerformanceparametersofPipePatrolERTTM
AllconfigurationsofPipePatrolERTTMprovidealowalarmthreshold,andPipePatrolERTTM/SC
canofferaverylowalarmthresholdifrequired.PipePatrolERTTM/PCprovidesveryshorttimesto
detectaleakforliquidpipelines.Timetodetectaleakbecomeslongerforgaspipelinesbecauseof
thedynamicinertiaofthefluid.AllconfigurationsofPipePatrolERTTMareabletodetectandlocate
suddenleaksaswellasgradualleaksofsufficientsize.
14 ComparisonofAllMethods
48
14 ComparisonofAllMethods
Thischaptercomparesallmethodspresentedsofar:
- Chapter5: Pressure/FlowMonitoring
- Chapter6: RarefactionWaveMethod
- Chapter7: BalancingMethods
- Chapter8: StatisticalLeakDetectionSystems
- Chapter9: LeakMonitoringduringShutInConditions
ThePipePatrolLDSfamilyfromKROHNEOil&Gasisalsoincludedinthecomparison,namely:
- Chapter12:PipePatrolStatisticalLineBalance(SLB)
- Chapter13:PipePatrolExtendedRealTimeTransientModel(ERTTM)
ThiscomparisonmayhelpinterestedreaderstoselecttheLDSprinciplebestsuitedtoaparticular
application.Characteristicsarelistedonaninformativebasis,andwhiletheyaretosomeextentsub
jectivebynatureeveryefforthasbeenmadetopresentobjectivefacts.Forallthesemethods,typical
characteristicsaregivenunderthefollowingheadings.
14.1 FunctionalityandInstrumentation
Amethodmayprovideleakdetection(LD),leaklocalization(LL),orboth(LD+LL).Instrumentre
quirementsareindicatedasfollows:
StaticBehavior:"Repeatability"meansthatrepeatabilityisofprimaryinterestincontrastto"Accura
cy"whereabsoluteaccuracyisalsorequired."Repeatability"islessstrict(andthereforelessexpen
sive)than"Accuracy",seeChapter3.1.
DynamicBehavior:"None"(nospecialrequirementsforsettlingtimeI
S
);"Fast"(I
S
typicallybelow
1s);"VeryFast"(I
S
typicallybelow0.1s).FordetailsrefertoChapter3.1.3.
14.2 FieldofApplication
Thisindicatesallowedoperationalconditions("PC"=PumpingConditions,"SC"=ShutinConditions,
"Steady"=SteadyState,"Transient"=TransientState)."LowTransient"meansthatonlysmalltransi
enteffectsareallowedforcorrectoperation,whichnormallyexcludesstartupandshutdown.Al
lowedfluidsare"Liquid","Gases",and"Liquid+Gases".PleaserefertoChapter2.1fordetailsof
compliancetoTRFL:"a)","b","c)","d)",and"e)".
14.3 PerformanceParameters
Alarmthresholdsareclassifiedas"High"(typically5%leakrateormore),"Medium"(typicallybe
tween1%and5%),"Low"(typicallyaround1%)and"VeryLow"(significantlowerthan1%).
Timetodetectisclassifiedas"VeryLong"(halfadayormore),"Long"(typicallylongerthanonehour
busnotmorethanhalfaday),"Medium"(typicallybetweentenminutesandonehour),"Fast"(typi
callybetweenoneminuteandtenminutes),and"VeryFast"(lessthanoneminute).
Detectableleaktypesinclude"Sudden","Gradual",and"Sudden+Gradual".
14 ComparisonofAllMethods
49
Table24:Comparisonofallmethods
A Bibliography
50
A Bibliography
[ADEC] ADEC:TechnicalReviewofLeakDetectionTechnologies,Volume1.Alaska
DepartmentofEnvironmentalConservation,1999.
[APIRP1130] APIRP1130:ComputationalPipelineMonitoringforLiquidPipelines.Rec
ommendedPracticefromtheAmericanPetroleumInstitute,2007.
[API1149] API1149:PipelineVariableUncertaintiesandTheirEffectonLeakDetecta
bility.AmericanPetroleumInstitute,1993.
[API1155] API1155:EvaluationMethodologyforSoftwareBasedLeakDetectionSys
tems.AmericanPetroleumInstitute,1995.(WithdrawnbyAPI).
[APIMPMS11] APIManualofPetroleumMeasurementStandardsChapter11:Physical
PropertiesData.AmericanPetroleumInstitute,2007.
[Baehr] Baehr,H.D.,Kabelac,S.:Thermodynamik.Springer,14.Auflage,2009.
[Barkat] Barkat,M.:SignalDetectionandEstimation.ArtechHouse,1991.
[Billmann] Billmann,L.:MethodenzurLeckberwachungundRegelungvonGasfernlei
tungen.FortschrittsberichteVDIReihe8,VDIVerlag.
[Bohl] Bohl,W.:TechnischeStrmungslehre.VogelVerlag,12.Auflage,2002.
[Geiger] Geiger,G.:LeakDetectionandLocatingASurvey.35
th
AnnualPSIGMeet
ing,15October17October2003,Bern,Switzerland.
[Kay] Kay,StevenM.:FundamentalsofStatisticalSignalProcessing,Volume2.
PrenticeHall,1998.
[Krass/Kittel/Uhde] Krass,W.,Kittel,A.,Uhde,A.:Pipelinetechnik.VerlagTVRheinland,1979.
[Kroschel] Kroschel,K.:StatistischeNachrichtentheorie.SpringerVerlag,1996.
[RFVO] Rohrfernleitungsverordnung.InTRFLTechnischeRegelnfrRohrfernlei
tungsanlagen.CarlHeymannsVerlag,2010.
[TRFL] TRFLTechnischeRegelnfrRohrfernleitungsanlagen.CarlHeymanns
Verlag,2010.
[VdTV1051] VdTV1051:WasserdruckprfungvonerdverlegtenRohrleitungennach
demDruckTemperaturMessverfahren(DTVerfahren).VdTVMerkblatt.
[Wald] Wald,A.:SequentialAnalysis,JohnWileyandSons,NewYork,1947.
[Zhang] Zhang,X.J.:StatisticalLeakDetectioninGasandLiquidPipelines.Pipes&
PipelinesInternational,JulyAugust1993,p.2629.
B Definitions
51
B Definitions
Accuracy:WithinthecontextofLDS,performancecriteriaby[APIRP1130].Ameasuredescribingthe
accuracyofleakflowandpositioncalculation.Formeasurementsystemsingeneral,thedegree
ofclosenessofmeasurementsxofaquantitytothatquantity'struevaluex
tuc
.Seealso:Accu
racy,Reliability,Repeatability,Robustness,andSensitivity.
API1149:"PipelineVariableUncertaintiesandTheirEffectonLeakDetectability."Publishedbythe
AmericanPetroleumInstitute.Allowsforthecalculationofatheoreticallimittoleakdetection
sensitivitybasedoninstrumentaccuracyandpipelinecharacteristics.
API1155:"EvaluationMethodologyforSoftwareBasedLeakDetectionSystems."Publishedbythe
AmericanPetroleumInstitute.Definesperformancemetricslikesensitivity,accuracy,reliability
androbustness.Withdrawn.Seealso:APIRP1130.
APIMPMSChapter11:"APIManualofPetroleumMeasurementStandardsChapter11:Physical
PropertiesData."PublishedbyAmericanPetroleumInstitute.Describesrelationstocalculate
densityoffluids(crudeoil,refinedproductsetc.)fromtemperatureandpressure.
APIRP1130:RecommendedPractice"ComputationalPipelineMonitoringforLiquidPipelines".Pub
lishedbytheAmericanPetroleumInstitute.Givesatechnicaloverviewaboutleakdetection
technologies,describesinfrastructuresupportsforCPM,anddiscussesCPMoperation,mainte
nanceandtesting.Definesperformancemetricslikesensitivity,accuracy,reliabilityandrobust
ness(fromAPI1155).Seealso:TRFL.
Batch:Acertainquantityoffluid(product)plannedforpipelinetransportfrominlettooutlet.See
also:Fluid,Pipeline,Product.
CommunicationProtocol:Asetofconventionsgoverningtheformatandtimingofdatatransmission
betweencommunicationdevices,includinghandshaking,errordetection,anderrorrecovery.
Synonym:Protocol.Seealso:SupervisoryControlandDataAcquisition(SCADA).
CompensatedMassBalance:Massbalancemethodwhichcompensatesforchangeofmassinvento
ryinapipeline.Seealso:LineBalance,MassBalance,UncompensatedMassBalance,andVolume
Balance.
Compressibility:Attributeofafluid,representingtherateofchangeofdensitywithrespecttopres
sureandtemperature.Itiscompensatedby(E)RTTMSystems.
ConservationofMass:Physicallawthatstatesthatmasstransportedbyapipelinewillbeconserved
incasethatthereisnoleak.Seealso:MassBalance.
CoriolisMassMeter:FlowmeasuringdevicedirectlymeasuringmassflowbasingontheCoriolis
principle.Seealso:Flowmeter,OrificePlate,PositiveDisplacementMeter,TurbineMeter,Ultra
sonicMeter,andVolumetricFlowMeter.
B Definitions
52
C(Continued)
CPMSystem:Computationalpipelinemonitoringsystemaccordingto[APIRP1130].Synonymfor
softwarebasedLDS.Seealso:LeakDetectionSystem(LDS).
DataAcquisition(DAQ):Theprocessofacquiringanaloguefielddataandconvertingthemintodigi
talvalues.Thequalityofthisprocessismainlyspecifiedbytheresolutiongiveninbits.Seealso:
Resolution,SupervisoryControlandDataAcquisition(SCADA).
DataCommunicationEquipment(DCE):Equipmentthatprovidesthefunctionsrequiredtoestablish,
maintain,orterminateaconnectionandthesignalconversionandcodingrequiredforcommuni
cationbetweendataterminalequipmentanddatacircuits.Examplesincludemodems,linedriv
ers,coaxialcable,satellitelinks,etc.Seealso:SupervisoryControlandDataAcquisition(SCADA).
DensityMeter:Asensorthatmonitorstheactualdensityofafluid.Seealso:Flowmeter.
DetectionLimit:Synonym:LeakDetectionLimit.
DifferentialPressureMethod(DPMethod):Differentialpressureacrossatightlyclosedvalvebe
tweentwotightlyclosedsegmentsisanalyzed.Seealso:PressureTemperatureMethod(PT
Method).
DigitalDataService(DDS):Aspecialwidebandwidthprivateleasedline(PLL)thatusesdigitaltech
niquestotransferdataathigherspeedsandlowererrorratethanvoiceband,analogPLLs.See
also:PrivateLeasedLine(PLL),PublicSwitchedTelephoneNetwork(PSTN).Seealso:Supervisory
ControlandDataAcquisition(SCADA).
DragReducingAgent(DRA):Addedtoliquidstoreducepipewallfriction.
Drift:Synonym:MeasurementDrift!
EOS:EquationofState.Synonyms:PVTEquation,ThermodynamicEquationofState.
ExternalLDS:Accordingto[APIRP1130],LDSsusingdedicatedmeasurementequipmentsuchas
sensingcables(e.g.vaportube,fiberopticcables).PipePatrolERTTMLDSandPipePatrolSLBLDS
donotbelongtothatLDSfamily.Seealso:InternalLDS.
ERTTMLDS:LDSbasedonExtendedRTTMtechnology.ThistechnologycombinesRTTMtechnology
withtheleaksignatureanalysis.Seealso:RealTimeTransientModel(RTTM),RTTMLDS,Leak
SignatureAnalysis,PipePatrolERTTMLDS.
FalseAlarm:Aleakalarmthatisraisedwhennorealleakispresent.FalsealarmprobabilityP
PA
specifiestheexpectednumberoffalsealarms,e.g.peryear.Seealso:LeakAlarm,MissedAlarm.
B Definitions
53
F(Continued)
Flow:Collectivetermformassflow(e.g.inkg/s),volumeflow(e.g.inm
3
/h),andvelocity(e.g.in
m/s).
Flowmeter:Devicemeasuringvolumeand/ormassflow.Seealso:CoriolisMassMeter,OrificePlate,
PositiveDisplacementMeter,TurbineMeter,UltrasonicMeter,andVolumetricFlowmeter.
FlowComputer:Deviceusedtopreprocessflowfieldsignalse.g.bycalculatingmassflowfrom
volumeflowforvolumetricflowmeters.Seealso:Flowmeter,Volumetricflowmeter.
FlowMonitoring:Asimpleleakdetectionmethodwhereflowismeasuredatasinglelocationinthe
pipeline;flowchangesincaseofaleak.Seealso:PressureMonitoring.
FlowResidual:Differencebetweenmeasuredflowandthevaluecalculatedbypipelineobserver
assumingaleakfreepipeline.UsedbyPipePatrolERTTM/PC.Seealso:PipelineObserver,Pres
sureResidual,Residual,andPipePatrolERTTM/PC.
Fluid:Asubstancethatiscapableofflowingthroughapipeline,includingallgasesandliquids.See
also:Batch,Product.
FullDuplexProtocol:Amodeofoperationforapointtopointlinkwithtwostations,inwhichmes
sagescanbesentinbothdirectionsatthesametime.Seealso:HalfDuplexProtocol,Supervisory
ControlandDataAcquisition(SCADA).
Gaussian(Normal)Distribution:Wellknownformofprobabilitydensityfunction,firstpublishedby
Gauss.Alsocallednormaldistribution.Veryoftenusedfortheunderlyingstatisticalmethodsof
StatisticalLDS.Seealso:StatisticalLDS.
GradientIntersectionMethod:Methodofleaklocationbasedoncharacteristicchangesinpressure
profilealongthepipeline.PipePatrolERTTMusesamodelbasedversion,whichisabletolocate
aleakevenintransientconditions.Seealso:LeakLocalization,TimeOfFlightMethod.
GradualLeak:Aslowlydevelopingleakidentifiablebyadriftsignature,oftenwithaverylowleak
flow.[TRFL]demandsdedicatedLDStodetectgradualleaks.PipePatrolERTTMraisessensor
alarmforgradualleaksofsufficientsize.Seealso:SensorAlarm,SuddenLeak.
HalfDuplexProtocol:Amodeofoperationforapointtopointorpointtomultipointlinkwithtwo
stations,inwhichmessagescanbesentinonedirectionortheotherbutnotbothatthesame
time.Seealso:FullDuplexProtocol,SupervisoryControlandDataAcquisition(SCADA).
HeadStation:Measuringstationatthepipelineinletoroutlet.Seealso:MeasuringStation,Substa
tion.
HeadStationMonitoring:ThisconfigurationofPipePatrolERTTMusesmeasurementsfromthe
headstationsofapipelinetocalculatetheflowresidualsandmonitorthepipelineforleaks.See
also:PipePatrolERTTMLDS,SectionalMonitoring,andSubstationsMonitoring.
B Definitions
54
H(Continued)
HypothesisTest:Familyofmethodfromstatisticaldecisiontheory,oftenusedbyStatisticalLDS.
Examplesarelikelihoodratiotestandsequentialprobabilityratiotest.Seealso:LikelihoodRatio
Test,SequentialProbabilityRatioTest,andStatisticalLDS.
Imbalance:Differencebetweenmass(orvolume)enteringapipelineandmass(orvolume)leavingit,
observedoversometimeperiodt.Changeofmassinventorymayadditionallybeconsidered.
Seealso:MassBalance.
Inlet:Thepointwherethefluidentersthepipelineinforwardoperation.Seealso:Outlet.
InternalLDS:Accordingto[APIRP1130],LDSsthatuseexistingmeasurementsensorsforflow,pres
sure,temperatureetc.PipePatrolERTTMLDSandPipePatrolSLBLDSbelongtothatLDSfamily.
Seealso:ExternalLDS.
Inventory:Synonym:PipelineInventory.
InventoryCompensation:Synonym:PipelineInventoryCompensation.
KROHNEOil&Gas:NetherlandssubsidiaryofKROHNEMesstechnikDuisburgGmbH&Co.KG.
PipePatrolistheLDSFamilyofKROHNEOil&Gas.Seealso:PipePatrolLDS.
LeakAlarm:Declarationofaleakevent.PipePatrolERTTMLDSwillraisethistypeofalarmincaseof
asuddenleak.Seealso:PipePatrolERTTMLDS,SuddenLeak.
LeakDetection:Processofdecidingifthereisaleakornot.UsuallyLDSareusedforthispurpose.
Seealso:LeakDetectionSystem(LDS),LeakLocalization,andLeakMonitoring.
LeakDetectionLimit:Theoreticalvalueofthesmallestdetectableleakfloworrate.Synonym:Detec
tionLimit.Seealso:LeakFlow,LeakRate.
LeakDetectionSystem(LDS):Aninternalorexternalbasedsystemtodetectleaksinpipelines,which
mayalsocalculateleaklocation.InGermany,LDSmustmeettherequirementsofthe[TRFL].See
also:CPMSystem,LeakDetection,andLeakLocalization.
LeakFlow:Massorvolumeflowoftheleak.Seealso:LeakRate.
LeakLocalization:Processofdeterminingthepositionofaleak,commonlycalculatedbyanLDS.See
also:LeakDetection,LeakDetectionSystem(LDS),LeakMonitoring.
LeakMonitoring:Monitoringprocessinvolvingleakdetectionandleaklocalization.Seealso:Leak
Detection,LeakLocalization.
B Definitions
55
L(Continued)
LeakRate:Leakflowrelatedtoareferencevalue,e.g.nominalflowofthepipeline.Seealso:Leak
Flow.
LeakSignature:Specificsignatureinsignalswhichoccursincaseofaleak,e.g.step(forsuddenleaks)
ordrift(forgradualleaks).Seealso:LeakSignatureAnalysis.
LeakSignatureAnalysis:MethodusedbyPipePatrolERTTMtoavoidfalsealarmswhereresiduals
areanalyzedforleaksignatures.Incaseofaleak,aleakalarm(forasuddenleak)orsensoralarm
(foragradualleakormeasurementdriftandoffset)israised.Seealso:ERTTMLDS,LeakAlarm,
LeakSignature,SensorAlarm,PipePatrolERTTMLDS.
LikelihoodRatioTest:Statisticalmethodtodecidebetweentwopredefinedhypotheses(e.g.leak
no/yes)basedonasetofmeasuredvalues.SomestatisticalLDSusethisprincipleforleakdetec
tion.Thesequentialversioniscalledsequentialprobabilityratiotest.Seealso:HypothesisTest,
SequentialProbabilityRatioTest,andStatisticalLDS.
LineBalance:Generictermcoveringallbalancingmethods(massbalanceandvolumebalance).See
also:MassBalance,CompensatedMassBalance,UncompensatedMassBalance,andVolume
Balance.
LocalProfiles:Flowandthermodynamicvaluessuchasmassflow,pressure,temperatureanddensi
ty,calculatedforeverypointalongthepipeline.
MassBalance:Methodsthatusetheconservationofmassprincipleforleakdetection.Synonym:
MaterialBalance.Seealso:Imbalance,LineBalance,CompensatedMassBalance,Uncompen
satedMassBalance,andVolumeBalance.
MasterStation(MS):ASCADAdevicesuchasaPLCwithI/Omodulesthatsendsdatatoandcollects
datafromremotestations.Seealso:ProgrammableLogicController(PLC),RemoteStation(RS),
andSupervisoryControlandDataAcquisition(SCADA).
MaterialBalance:Synonym:MassBalance.
MeasurementDrift:Verylowfrequencyerrorinmeasurements.PipePatrolERTTMraisessensor
alarmformeasurementdrift(flow,pressure,temperature)ofsufficientsize.Synonym:Drift.See
also:MeasurementError,MeasurementOffset.
MeasurementError:DeviationE x - x
tuc
betweenanactualmeasurementxandtheunknown
truevaluex
tuc
iscalledabsolutemeasurementerror.Ifdividedbysomereferencevaluex
c]
(e.g.x
tuc
ormeasurementrangex
mux
-x
mn
),itiscalledrelativemeasurementerror.Seealso:
MeasurementDrift,MeasurementOffset.
MeasurementOffset:Constanterrorinmeasurements.PipePatrolERTTMraisessensoralarmfor
measurementoffsets(flow,pressure,temperature)ofsufficientsizes.Synonym:Offset.Seealso:
MeasurementDrift,MeasurementError.
B Definitions
56
M(Continued)
MeasurementStation:Astationatonespecificpointofthepipeline,equippedwithsensorsand
SCADAdevices.Atinletandoutlettheyarecalledheadstations,otherwisesubstations.Seealso:
HeadStation,Substation,andSupervisoryControlandDataAcquisition(SCADA).
MissedAlarm:Afailuretodeclareanalarmwhenthereisarealleak.MissedalarmprobabilityP
MA
specifiestheexpectednumberofmissedalarms,e.g.peryear.Seealso:FalseAlarm,LeakAlarm.
Modbus:Aserialcommunicationsprotocolfirstpublishedin1979forusewithitsprogrammable
logiccontrollers(PLCs).Oneofthedefactostandardcommunicationsprotocolsintheindustry.
Seealso:OLEForProcessControl(OPC),SupervisoryControlandDataAcquisition(SCADA).
MultiProductPipeline:Pipelinethattransportseveraldifferentproducts.Seealso:SingleProduct
Pipeline.
MultipointToMultipoint:Communicationlinkbetweenthreeormorestationswherethereisno
communicationarbitrator(master)andanystationcaninitiatecommunicationswithanyother
station.Seealso:PointtoPoint,PointtoMultipoint,andSupervisoryControlandDataAcquisi
tion(SCADA).
NominalFlow:Flowatnominalconditions,e.g.inm
3
/h.
Offset:Synonym:MeasurementOffset.
OLEForProcessControl(OPC):Communicationstandardspecifyingthecommunicationofrealtime
plantdatabetweencontroldevicesfromdifferentmanufacturers.Seealso:Modbus,Supervisory
ControlandDataAcquisition(SCADA).
OrificePlate:Differentialpressuregeneratingflowmeteringdevice.Orificeplatesarenotdirectly
eithermassorvolumetricflowmeters.Seealso:CoriolisMassMeter,Flowmeter,OrificePlate,
PositiveDisplacementMeter,TurbineMeter,UltrasonicMeter,andVolumetricFlowmeter.
Outlet:Theendofapipelinewherethefluidleavesinforwardoperation.Seealso:Inlet.
PausedFlowConditions:Conditionswherenofluidispumpedthroughthepipeline.Flowmaybe
unblockedorblocked(shutinconditions)byvalves.[TRFL]assumesspecialLDSforthesecondi
tions.Seealso:PumpingConditions,ShutinConditions.
Pipeline:Alongdistancelineforfluid(product)transport.Seealso:Fluid,Product.
PipelineController:Apersonwhoisresponsibleforthemonitoringanddirectcontrolofapipeline.
Synonym:PipelineOperator.
B Definitions
57
P(Continued)
PipelineInventory:Massstoredwithinapipelinesegment.Synonym:Inventory.Seealso:Pipeline
InventoryCompensation.
PipelineInventoryCompensation:Methodtocompensateforchangesofpipelinecontentsinmass
balance.PipePatrolSLBoptionallyappliesRTTMbasedcompensation.Synonym:InventoryCom
pensation.Seealso:MassBalance,PipelineInventory,andPipePatrolStatisticalLineBalance
(SLB)LDS.
PipelineObserver:SWModulewithinPipePatrolERTTM(andSLB),whichcalculatesresidualsbased
onmeasuredvalues.Itisusedtoeliminatecompressibilityandelasticityeffects.Itmayalsobe
usedforvirtualflowcalculation.Synonym:VirtualPipeline.Seealso:LeakSignatureAnalysis,
PipePatrolERTTMLDS,PipePatrolStatisticalLineBalance(SLB)LDS,andVirtualFlow.
PipelineOperator:Synonym:PipelineController.
PipelineRupture:Occurswhenthereisasignificantbreachofthepipewallormajorlossofcontain
mentoftheproductwithinthepipeline.Mayleadtoalarge,suddenleakflow.Synonym:Rup
ture.
PipelineSection:Asubdivisionofacompletepipelinenetwork,whichmayconsistofoneormore
pipelinesegments.Seealso:PipelineSegment.Synonym:Section.
PipelineSegment:Asingleconnectionbetweentwonodesinapipelinenet.Ingeneralasubdivision
ofapipelinesection.Synonym:Segment.Seealso:PipelineSection.
PipePatrolERTTMLDS:PipePatrolExtendedRealTimeTransientModelLDSbyKROHNEOil&Gas
basedonERTTMtechnology(pipelineobserver)andleaksignatureanalysis.ProvidesLeakde
tectionandlocationinpumpingconditions(PipePatrolERTTM/PC)andshutinconditions
(PipePatrolERTTM/SC)forsteadyandtransientstatesasdefinedin[APIRP1130]and[TRFL].
Raisesleakalarmforsuddenleaks,andsensoralarmforgradualleaksormeasurementdriftor
offsets.Differentconfigurationsareavailablefordifferentpipelinenetstructures:Headstation
monitoring,substationmonitoring,andsectionalmonitoring.PipePatrolStatisticalLineBalance
(SLB)mayserveasabackupLDSincaseofpressureand/ortemperaturesensorfailure.Seealso:
ERTTMLDS,HeadStationsMonitoring,LeakAlarm,LeakSignatureAnalysis,PipelineObserver,
PipePatrolStatisticalLineBalance(SLB),SectionalMonitoring,SensorAlarm,andSubstations
Monitoring.
PipePatrolERTTM/PC:ConfigurationofPipePatrolERTTMforpumpingconditions.Seealso:
PipePatrolERTTM,PumpingConditions.
PipePatrolERTTM/SC:ConfigurationofPipePatrolERTTMforshutinconditions.Seealso:PipePa
trolERTTM,ShutinConditions.
PipePatrolLDS:FamilyofLDSbyKROHNEOil&Gas.AtthistimeconsistsofPipePatrolERTTMLDS
andPipePatrolStatisticalLineBalance(SLB).Seealso:KROHNEOil&Gas,PipePatrolERTTMLDS,
PipePatrolStatisticalLineBalance(SLB)LDS.
B Definitions
58
P(Continued)
PipePatrolStatisticalLineBalance(SLB)LDS:LDSbyKROHNEOil&Gasprovidingleakdetectionus
ingmassandvolumebalancetogetherwithRTTMbasedpipelineinventorycompensation(pipe
lineobserver).Differentconfigurationsforuncompensatedmassbalance,compensatedmass
balanceandvolumebalanceareavailable.MayserveasabackupLDSforPipePatrolERTTMin
caseofpressureand/ortemperaturesensorfailure.Seealso:LineBalance,PipelineInventory
Compensation,PipelineObserver,PipePatrolERTTM.
PointtoMultipoint:Anetworkwhereconnectionsexistbetweenonemasterstationandmultiple
remotestations.Seealso:MultipointtoMultipoint,PointtoPoint,andSupervisoryControland
DataAcquisition(SCADA).
PointtoPoint:Anetworkwhereaconnectionismadebetweentwoandonlytwoterminalinstalla
tions.Seealso:MultipointtoMultipoint,PointtoMultipoint,andSupervisoryControlandData
Acquisition(SCADA).
Poll:Procedurewheremasterstationsendsamessagetoaremotestationthatrequirestheremote
stationtoreturnaresponsetothemasteroranotherremotestation.Seealso:SupervisoryCon
trolandDataAcquisition(SCADA).
PollingTime:TimeintervalbetweentwoconsecutivepollsonaSCADAcommunicationlink.Syno
nym:ScanTime.Seealso:Poll,SupervisoryControlandDataAcquisition(SCADA).
PositiveDisplacementMeter:Flowmeasuringdevicethatmeasuresflowbymovingtheliquid
throughmeasuringchambersofknownvolume.Belongstovolumetricflowmeters.Seealso:Cor
iolisMassMeter,Flowmeter,OrificePlate,TurbineMeter,UltrasonicMeter,andVolumetric
Flowmeter.
PressureMonitoring:Asimpleleakdetectionmethodwherepressureismeasuredatasingleloca
tiononthepipeline;pressuredropsincaseofaleak.Seealso:FlowMonitoring.
PressureResidual:Differencebetweenmeasuredpressureandthevaluecalculatedbypipelineob
serverassumingaleakfreepipeline.UsedforparticularconfigurationsbyPipePatrolERTTM/PC
andgenerallybyPipePatrolERTTM/PC.Seealso:FlowResidual,PipelineObserver,Residual,and
PipePatrolERTTM.
PressureTemperatureMethod(PTMethod):Pressurewithinatightlyclosedpipelinesegmentis
analyzed,andtemperaturecompensatedusingandequationofstateforthefluid.PipePatrolE
RTTM/SCusesamodelbasedversionofthismethod.Seealso:DifferentialPressureMethod(DP
Method).
PrivateLeasedLine(PLL):Adedicatedvoicebandtelephonelinebetweentwoormorelocations
primarilyusedfordatatransmission.Seealso:DigitalDataService(DDS),PublicSwitchesTele
phonenetwork(PSTN),andSupervisoryControlandDataAcquisition(SCADA).
ProbabilityDensityFunction(PDF):Functionofprobabilitytheory,whichenablescalculationof
probabilityforanevent.Seealso:Gaussian(Normal)Distribution.
Product:Afluidflowingthroughapipeline,includingallgasesandliquids.Seealso:Batch,Fluid,
Pipeline.
B Definitions
59
P(Continued)
ProgrammableLogicController(PLC):Digitalcomputerusedforautomation.Unlikegeneralpurpose
computers,thePLCisdesignedformultipleinputandoutputarrangements,extendedtempera
tureranges,immunitytoelectricalnoise,andresistancetovibrationandimpact.Maybepartof
SCADAmasterstationsandremotestations.Seealso:MasterStation(MS),RemoteStation(RS),
andSupervisoryControlandDataAcquisition(SCADA).
Protocol:Synonym:CommunicationProtocol.
PublicSwitchedTelephoneNetwork(PSTN):Thestandarddialuptelephonenetworkoriginallyused
forvoicecommunication.Seealso:DigitalDataService(DDS),PrivateLeasedLine(PLL),andSu
pervisoryControlandDataAcquisition(SCADA).
PumpingConditions:Conditionswherefluidispumpedthroughthepipeline.[TRFL]assumesspecial
LDSfortheseconditions.Seealso:ShutinConditions,PausedFlowConditions.
PVTEquation:Synonyms:EOS,ThermodynamicEquationofState.
RarefactionWave:Hydraulicwavelikeeffectincaseofaleakpropagatingwithspeedofsound.See
also:RarefactionWaveMethod.
RarefactionWaveMethod:LDSevaluatingrarefactionwavescreatedbysuddenleaks.Seealso:
RarefactionWave.
RealTimeTransientModel(RTTM):Simulationmodelofflowinpipelinescoveringthesteadystate
aswellasthetransientstate.Calculatedinrealtimebydigitalcomputers.Seealso:ERTTMLDS,
RTTMLDS.
Reliability:WithinthecontextofLDS,performancecriteriondefinedby[APIRP1130].Abilityofa
leakdetectionsystemtorenderaccuratedecisionsaboutthepossibleexistenceofaleakonthe
pipeline,whileoperatingwithinanenvelopeestablishedbytheleakdetectionsystemdesign.Di
rectlyrelatedtoprobabilityofafalsealarmandprobabilityofamissedalarm.Seealso:Accuracy,
Robustness,andSensitivity.
RemoteStation(RS):ASCADAdevicesuchasaPLCwithI/Omodulesthatislocatedinasiteremote
fromthemasterstationandthatcontrolsI/Opointsattheremotesite.Aremotestationaccepts
commandsfromandmaysenddatatoamasterstationviaanetwork.Seealso:Programmable
LogicController(PLC),MasterStation(MS),RemoteTerminalUnit(RTU),andSupervisoryControl
andDataAcquisition(SCADA).
RemoteTerminalUnit(RTU):Remotestationonlyacquiringdata.Seealso:RemoteStation(RS),
SupervisoryControlandDataAcquisition(SCADA).
B Definitions
60
R(Continued)
Repeatability:Closenessofagreementbetweenindependentmeasurementreadingsobtainedwith
thesamemeasurementsystemunderthesameconditions(sameflow,sameenvironmentalcon
ditionsetc.).Veryoftenrepeatabilityisdefinedasthevaluebelowwhichtheabsolutedifference
|x| |x|k] - x|k -1]|betweentwosuccessivesinglemeasurementsx|k]andx|k -1]ob
tainedunderthesameconditionsmaybeexpectedtoliewithaspecifiedprobability(e.g.95%).
Seealso:Accuracy.
Residual:Differencebetweenameasuredvalue(e.g.pressureorflow)andacorrespondingvalue
calculatedbypipelineobserverassumingaleakfreepipeline.ForPipePatrolERTTM,inputfor
leaksignatureanalysis.PipePatrolusesflowandpressureresidualsdependingontheconfigura
tion.Residualsareanalyzedbyleaksignatureanalysisforleakalarmdeclarationandleaklocaliza
tion.Seealso:FlowResiduals,LeakSignatureAnalysis,PipelineObserver,PipePatrolERTTMLDS,
andPressureResiduals.
Resolution:Withinthecontextofdataacquisition,thesmallestincrementofsignalchangethatcan
bedeterminedbyaconverternormallyspecifiedinbits.Seealso:DataAcquisition,Supervisory
ControlandDataAcquisition(SCADA).
Robustness:WithinthecontextofLDS,performancecriteriondefinedby[APIRP1130].Abilityto
continuetofunctionandprovideusefulinformationevenunderchangingconditionsofpipeline
operation,orinconditionswheredataislostorsuspect.Asystemisconsideredtoberobustifit
continuestofunctionundersuchnonidealconditions.Seealso:Accuracy,Reliability,andSensi
tivity.
RTTMLDS:LDSbasingonaRealTimeTransientModel(RTTM).Seealso:ERTTMLDS,RealTime
TransientModel(RTTM).
Rupture:Synonym:PipelineRupture.
ScanTime:Synonym:PollTime.
Section:Synonym:PipelineSection.
SectionalMonitoring:ThisconfigurationofPipePatrolERTTMsplitspipelinesintosegmentsbound
edbymeasurementstations(headstationsandsubstations),andmonitorseachsegmentinde
pendently.Substationsmayprovideflow,pressureandtemperature;usingthevirtualflow
method,flowandtemperatureofasubstationmaybereplacedbyvaluescalculatedfromthe
pipelineobserver.Seealso:HeadStationsMonitoring,PipelineObserver,PipePatrolERTTMLDS,
SubstationsMonitoring,andVirtualFlow.
Segment:Synonym:PipelineSegment.
Sensitivity:WithinthecontextofanLDS,aperformancecriteriondefinedby[APIRP1130].Compo
sitecriterion,combiningsmallestdetectableleakflowaswellastimetodetectaleak.Example:
Lostvolumeormassbyleakflowfrombeginningofleakflowtillleakalarm.Seealso:Accuracy,
Reliability,andRobustness.
B Definitions
61
S(Continued)
SensorAlarm:PipePatrolERTTMwillraisethistypeofalarmincaseofagradualleakormeasure
mentdriftandoffset.Seealso:GradualLeak,LeakAlarm,PipePatrolERTTMLDS.
SequentialProbabilityRatioTest(SPRT):Asequentialversionofthelikelihoodratiotest.Somesta
tisticalLDSusethisprincipleforleakdetection.Seealso:LikelihoodRatioTest,StatisticalLDS.
SettlingTime:Timerequiredfortheresponsecurveofasensortoreachandstaywithinarangeof
certainpercentage(usually5%or2%)ofthefinalvaluefollowingastepchangeinthemeasured
variable.
Shutdown:Transitionphaseofpipelineoperationwhereoperationmodechangesfrompumping
conditionstopausedflowconditions.Seealso:Startup,TransientState.
ShutinConditions:Pausedflowconditions,wherepumpsareswitchedoff,andflowisblockedby
valves.Seealso:PumpingConditions,PausedFlowConditions.
SingleProductPipeline:Pipelinewhereonlyonesingleproductistransported.Seealso:Multi
ProductPipeline.
SlackLine:Conditionswhereasegmentofaliquidpipelineisnotentirelyfilled,sometimesasaresult
ofvaporizationofthetransportedproduct.
SpeedofSound:Propagationspeedofpressure,densityandflowdynamicsinfluids.Leaklocation
byTimeofFlightMethodisbasedonthespeedofsound.
StandardConditions:Combinationoftemperature(e.g.10C)andpressure(e.g.1.101325bar)at
whichfluidstandardvolumesareexpressed.
Startup:Transitionphaseofpipelineoperationwhereoperationmodechangesfrompausedflow
conditionstopumpingconditions.Seealso:Shutdown,TransientState.
StatisticalLDS:FamilyofLDSwithafocusonstatisticalmethods.Seealso:HypothesisTest,Leak
SignatureAnalysis,LikelihoodRatioTest,andSequentialProbabilityRatioTest(SPRT).
StatisticalLeakClassification:StatisticalmodulewithinPipePatrolStatisticalLineBalance(SLB)for
preventingfalsealarmswhilemaintainingthehighestpossiblesensitivity.Seealso:PipePatrol
StatisticalLineBalance(SLB).
SteadyState:Processconditionsforapipeline,wherephysicalvalues(e.g.flowandpressure)do
NOTchangesignificantlyovertime.[TRFL]requiresanLDStobecapableofmonitoringpipelines
forleaksundersteadystateconditions.Seealso:TransientState.
Substation:Allmeasurementstationsotherthantheinletandoutlet(whicharecalledheadsta
tions).Seealso:HeadStation,MeasurementStation.
SubstationMonitoring:ThisconfigurationofPipePatrolERTTMusesmeasurementsfromthesub
stationsofapipelinetocalculatepressureresidualsandmonitorthepipelineforleaks.Seealso:
HeadStationsMonitoring,PipePatrolERTTMLDS,SectionalMonitoring.
B Definitions
62
S(Continued)
SuddenLeak:Asuddenleakidentifiablebyastepsignature,oftentheresultofapipelinerupture.
PipePatrolERTTMraisesleakalarmforsuddenleaksofsufficientsizes.Seealso:GradualLeak,
LeakAlarm.
SupervisoryControlandDataAcquisition(SCADA):ThefieldinformationnecessaryforinternalLDS
usuallywillbeprovidedbyaSupervisoryControlandDataAcquisition(SCADA)system;thisisa
computerbaseddatacommunicationsystemthatmonitors,processes,transmits,anddisplays
pipelinedataforthepipelinecontroller.SCADAsystemsmaybeuseddirectlyforleakdetection,
theymayprovidesupportforanLDS,oranLDSmayoperateindependentlyofSCADA.Generally,
apipelineLDSwillusethedatageneratedbyaSCADAsystem.Seealso:CommunicationProtocol,
DataAcquisition(DAQ),DataCommunicationEquipment(DCE),DigitalDataService(DDS),Mas
terStation(MS),MeasurementStation,Modbus,OLEForProcessControl(OPC),PrivateLeased
Line(PLL),ProgrammableLogicController(PLC),PublicSwitchedTelephoneNetwork(PSTN),and
RemoteStation(RS).
ThermodynamicEquationofState:Thermodynamicrelationbetweenpressure,temperatureand
density,whichistrueforafluid.Forcrudeoil,refinedproductsandlubricatingoils,correspond
ingrelationsandparametersaregivenby[APIMPMS11].Synonyms:EOS,PVTEquation.
TimeTag:ASCADAfeaturerecordingthetimethatameasurementoreventoccursalongwiththe
data.Seealso:SupervisoryControlandDataAcquisition(SCADA).
TimeofFlightMethod:Leaklocationmethodwherethedifferenceintraveltimeofasuddenpres
suredroppropagatinglikeawaveisdetectedatdifferentlocationsalongthepipeline.PipePatrol
ERTTMusesamodelbasedversion,whichisabletolocatealeakevenintransientconditions.
Seealso:GradientIntersectionMethod,LeakLocalization.
Topology:Geometricarrangementofnodesandlinksthatmakeupanetwork.Example:aring,bus,
orstarconfiguration.Seealso:SupervisoryControlandDataAcquisition(SCADA).
TransientState:Stateofapipelinewherephysicalvalues(e.g.flowandpressure)DOchangesignifi
cantlyovertime.[TRFL]requiresanLDStobecapableofmonitoringpipelinesforleaksintransi
entconditions.Seealso:SteadyState.
TRFL:TechnischeRegelnfrRohrfernleitungsanlagen(TechnicalRulesforPipelines).Germanregu
lationrequiringLDSsforpipelines.AppliestomostliquidandgaspipelinesinGermany.Seealso:
APIRP1130.
TurbineMeter:Flowmeasuringdevicewitharotorthatsensethevelocityofflowingfluidinaclosed
conduit.Belongstovolumetricflowmeters.Seealso:CoriolisMassMeter,Flowmeter,Orifice
Plate,PositiveDisplacementMeter,UltrasonicMeter,andVolumetricFlowmeter.
UltrasonicMeter:Flowmeasuringdevicebasingonthemeasurementoftimedelaysofultrasonic
impulses.BelongstovolumetricFlowmeters.Seealso:CoriolisMassMeter,Flowmeter,Orifice
Plate,PositiveDisplacementMeter,TurbineMeter,andVolumetricFlowmeter.
B Definitions
63
U(Continued)
UltrasonicMeter:Flowmeasuringdevicebasingonthemeasurementoftimedelaysofultrasonic
impulses.BelongstovolumetricFlowmeters.Seealso:CoriolisMassMeter,Flowmeter,Orifice
Plate,PositiveDisplacementMeter,TurbineMeter,andVolumetricFlowmeter.
UncompensatedMassBalance:Massbalancemethodwithoutcompensationforchangeofmass
inventoryinapipeline.Seealso:LineBalance,MassBalance,CompensatedMassBalance,and
VolumeBalance.
VirtualFlow:Unmeasuredflowatsubstationsboundedbytwosegments,calculatedbythePipeline
Observersofthesesegments.Seealso:PipelineObserver,SectionalMonitoring.
VirtualPipeline:Synonym:PipelineObserver.
VolumeBalance:Specialformofmassbalancewheredensityatinletandoutletareequalsothat
volumeimbalancesummedupoverasufficientlongtimeisideallyzeroforleakfreepipelines.
Anotherapproachistoconsidertheimbalancefortheleakfreepipelineandtousestatistical
methodstodetectchangesintheimbalancecausedbyaleak.Seealso:LineBalance,MassBal
ance,CompensatedMassBalance,UncompensatedMassBalance,andVolumeBalance.
VolumetricFlowmeter:Flowmeterwhereflowismeasuredasvolumeflow.Todeterminemass
flowthedensityofthefluidsneedstobeknown;flowcomputerarenormallyusedforthispur
pose.Seealso:Flowmeter,FlowComputer.
AbouttheAuthor
64
AbouttheAuthor
GerhardGeigerwasborn1954inKaufbeuren,Germany,graduatedfromtheUniversityofDarmstadt
in1979,andwasawardedaPhDin1985forhisworkonanewapproachtotechnicaldiagnostics.He
cofoundedanengineeringcompanyactiveinthefieldofmicroelectronicsandautomationin1987,
andremainedCEOuntil1995.Herehewasresponsiblefordevelopment,salesandmarketingof
pipelineleakdetectionsystems,amongotherthings.Hewasappointedtoaprofessorshipatthe
WestphalianUniversityGelsenkirchen(Germany)in1995.Sincethen,themainfocusofhisworkhas
beenmodelbasedpipelineleakdetectionmethods.Hehaspublishedmorethan50articlesabout
leakdetection,anddevelopedtheExtendedRealTimeTransientModel(ERTTM)technology.
In1998,ProfessorGeigerandKROHNEMesstechnikinDuisburgstartedaclosecooperation.Togeth
er,theydevelopedKROHNE'sverysuccessfulPipePatrolleakdetectionsystemfamilybasedonE
RTTMtechnology.