Beruflich Dokumente
Kultur Dokumente
i
EPA/600/R-06/028
December2005
WaterDistributionSystemAnalysis:
FieldStudies,ModelingandManagement
AReferenceGuideforUtilities
U.S.EnvironmentalProtectionAgency
OfficeofResearchandDevelopment
NationalRiskManagementResearchLaboratory
WaterSupplyandWaterResourcesDivision
Cincinnati,Ohio
AReferenceGuideforUtilities
Notice
Anyopinionsexpressedinthisdocument/referenceguideforutilitiesarethoseoftheauthor(s)and
donot,necessarily,reflecttheofficialpositionsandpoliciesoftheU.S.EnvironmentalProtection
Agency(EPA).Anymentionofproductsortradenamesdoesnotconstituterecommendationforuse
byEPA. ThisdocumenthasbeenreviewedinaccordancewithEPAspeerandadministrativereview
policiesandapprovedforpublication.
ii
AReferenceGuideforUtilities
Foreword
TheU.S.EnvironmentalProtectionAgency(EPA)ischargedbyCongresswithprotectingtheNations
land,air,andwaterresources.Underamandateofnationalenvironmentallaws,theAgencystrives
toformulateandimplementactionsleadingtoacompatiblebalancebetweenhumanactivitiesand
theabilityofnaturalsystemstosupportandnurturelife.Tomeetthismandate,EPAsresearchprogram
isprovidingdataandtechnicalsupportforsolvingenvironmentalproblemstodayandbuildinga
science knowledge base necessary to manage our ecological resources wisely, understand how
pollutantsaffectourhealth,andpreventorreduceenvironmentalrisksinthefuture.
TheNationalRiskManagementResearchLaboratory(NRMRL)istheAgencyscenterforinvestigation
oftechnologicalandmanagementapproachesforpreventingandreducingrisksfrompollutionthat
threatenhumanhealthandtheenvironment.ThefocusoftheLaboratorysresearchprogramison
methodsandtheircost-effectivenessforpreventionandcontrolofpollutiontoair,land,water,and
subsurfaceresources;protectionofwaterqualityinpublicwatersystems;remediationofcontaminated
sites,sedimentsandgroundwater;preventionandcontrolofindoorairpollution;andrestorationof
ecosystems. NRMRLcollaborateswithbothpublicandprivatesectorpartnerstofostertechnologies
thatreducethecostofcomplianceandtoanticipateemergingproblems.NRMRLsresearchprovides
solutionstoenvironmentalproblemsby:developingandpromotingtechnologiesthatprotectand
improvetheenvironment;advancingscientificandengineeringinformationtosupportregulatory
and policy decisions; and providing the technical support and information transfer to ensure
implementationofenvironmentalregulationsandstrategiesatthenational,state,andcommunity
levels.
ThispublicationhasbeenproducedaspartoftheLaboratorysstrategiclong-termresearchplan.Itis
published and made available by EPAs Office of Research and Development to assist the user
communityandtolinkresearcherswiththeirclients.
SallyGutierrez,Director
NationalRiskManagementResearchLaboratory
iii
AReferenceGuideforUtilities
TableofContentsOverview
Overviewofdistributionsystem
infrastructure,generalwaterqualityissues
andmanagementapproaches
Summarydiscussionofvarioushydraulic
andwaterqualitymodelingtools
Outlineforplanningandexecutionofa
tracerstudytoperformdistributionsystem
evaluation
Techniquesforcalibrationandvalidationof
distributionsystemmodels
Frameworkofavailableoptionsfor
monitoringdistributionsystemwaterquality
Introductiontotheuseofgeospatial
technologyforwaterdistributionsystems
Realworldapplicationsofdistribution
systemmodelingapproaches
Chapter1
Chapter2
Chapter3
Chapter4
Chapter5
Chapter6
Chapter7
iv
AReferenceGuideforUtilities
TableofContents
Notice .......................................................................................................................................................ii
Foreword.................................................................................................................................................. iii
TableofContentsOverview....................................................................................................................iv
TableofContents .....................................................................................................................................v
ListofTables............................................................................................................................................ix
ListofFigures..........................................................................................................................................ix
AcronymsandAbbreviations................................................................................................................ xii
Acknowledgments..................................................................................................................................xv
1.0 Introduction..................................................................................................................................1-1
1.1 DistributionSystem-InfrastructureDesignandOperation..................................................1-3
1.1.1 KeyInfrastructureComponents.....................................................................................1-3
1.1.1.1 StorageTanks/Reservoirs.....................................................................................1-3
1.1.1.2 PipeNetwork........................................................................................................1-3
1.1.1.3 Valves....................................................................................................................1-3
1.1.1.4 Pumps...................................................................................................................1-4
1.1.1.5 HydrantsandOtherAppurtenances....................................................................1-4
1.1.2 BasicDesignandOperationPhilosophy.......................................................................1-4
1.1.2.1 Pipe-NetworkConfigurations...............................................................................1-5
1.1.2.2 MultipleSourceConfiguration ............................................................................1-5
1.1.2.3 ImpactofSystemDesignandOperationonWaterQuality................................1-5
1.2 WaterQualityProblemsandIssues.......................................................................................1-6
1.3 RegulatoryFramework...........................................................................................................1-7
1.4 AssessmentandManagementofWaterQuality...................................................................1-8
1.5 AdvancedToolsforWaterQualityManagement ................................................................1-11
1.6ReportOrganization................................................................................................................1-11
1.7 Summary..............................................................................................................................1-11
References.....................................................................................................................................1-13
2.0 ModelingWaterQualityinDrinkingWaterDistributionSystems...............................................2-1
2.1 DistributionSystemNetworkHydraulicModeling................................................................2-1
2.1.1 HistoryofHydraulicModeling.......................................................................................2-1
2.1.2 OverviewofTheoreticalConcepts................................................................................2-2
2.1.3 BasicHydraulicModelInputCharacterization..............................................................2-3
2.1.3.1 PipeNetworkInputs.............................................................................................2-3
2.1.3.2 WaterDemandInputs...........................................................................................2-3
2.1.3.3 TopographicalInputs............................................................................................2-5
2.1.3.4 ModelControlInputs ...........................................................................................2-5
2.1.3.5 ExtendedPeriodSimulation(EPS)SolutionParameters....................................2-5
2.1.4 GeneralCriteriaforModelSelectionandApplication..................................................2-6
2.1.4.1 DevelopingaBasicNetworkModel.....................................................................2-6
2.1.4.2 ModelCalibrationandValidation.........................................................................2-6
2.1.4.3 EstablishingObjectivesandModelApplication..................................................2-6
2.1.4.4 AnalysisandDisplayofResults............................................................................2-7
2.2 ModelingWaterQualityInDistributionSystemNetworks....................................................2-7
2.2.1 HistoryofWaterQualityModeling................................................................................. 2-7
2.2.2 TheoreticalConceptsforWaterQualityModeling........................................................2-8
2.2.3 WaterQualityModelInputsandApplication ..............................................................2-10
2.3 HydraulicandWaterQualityModelingSoftware................................................................2-11
2.3.1 EPANETSoftware ........................................................................................................2-11
2.3.2 CommercialHydraulic-WaterQualityModelingSoftware..........................................2-12
2.4 AdditionalModelingTools....................................................................................................2-12
2.4.1 StorageModelingTools...............................................................................................2-12
2.4.1.1 CFDModels ........................................................................................................2-13
2.4.1.2 CompartmentModels........................................................................................2-14
2.4.1.3 PhysicalScaleModels........................................................................................2-14
2.4.2 TransientAnalysisSoftware.........................................................................................2-15
v
AReferenceGuideforUtilities
2.4.3 OptimizationTools........................................................................................................2-15
2.4.3.1 OptimizingCalibration........................................................................................2-15
2.4.3.2 DesignOptimization ...........................................................................................2-16
2.4.3.3 OptimizationofOperation..................................................................................2-17
2.4.4 ProbabilisticModels.....................................................................................................2-17
2.5 SummaryandConclusions .................................................................................................2-17
References.....................................................................................................................................2-19
3.0 TracerStudiesforDistributionSystemEvaluation.....................................................................3-1
3.1 PlanningandDesigningaDistributionSystemTracerStudy ..............................................3-2
3.1.1 EstablishingStudyObjectivesandTime-Line...............................................................3-2
3.1.2 FormingaStudyTeam...................................................................................................3-2
3.1.3 DefiningStudyAreaCharacteristics..............................................................................3-2
3.1.4 SelectingTracerMaterial...............................................................................................3-3
3.1.4.1 Fluoride.................................................................................................................3-3
3.1.4.2 CalciumChloride..................................................................................................3-4
3.1.4.3 SodiumChloride..................................................................................................3-5
3.1.4.4 OtherChemicalsThatMaybeAddedasTracers................................................3-5
3.1.4.5 NaturallyorNormallyOccurringTracers............................................................3-5
3.1.4.6 ComparisonofTracers.........................................................................................3-6
3.1.5 SelectingFieldEquipmentandProcedures..................................................................3-8
3.1.5.1 InjectionPump(s)..................................................................................................3-8
3.1.5.2 TracerStorageandDosageRateMeasurement.................................................3-8
3.1.5.3 DistributionSystemFlowRateMeasurement.....................................................3-9
3.1.5.4 FieldMeasurementofTracerConcentration.......................................................3-9
3.1.6 DevelopingaDetailedStudyDesign..........................................................................3-10
3.1.7 AddressingAgencyandPublicNotification................................................................3-11
3.2 ExecutingaTracerStudy.....................................................................................................3-11
3.2.1 Procurement,Setup,TestingandDisinfectionofStudyEquipment.........................3-11
3.2.2 InstallationofFieldEquipmentandTesting.................................................................3-12
3.2.3 TracerDosageandInjectionDurationCalculations....................................................3-13
3.2.4 DryRunsandPlannedTracerInjectionEvent(s).........................................................3-13
3.2.5 RealTimeFieldAssessments,Sampling,andAnalysis..............................................3-13
3.2.6 EquipmentDe-Mobilization,InitiationofDataCollection,Reduction,andVerification
Process.........................................................................................................................3-14
3.3 TracerStudyCosts...............................................................................................................3-14
3.4 Summary,ConclusionsandRecommendations.................................................................3-15
References.....................................................................................................................................3-17
4.0 CalibrationofDistributionSystemModels..................................................................................4-1
4.1 HydraulicandWaterQualityModelCalibration....................................................................4-2
4.1.1 HydraulicModelCalibration..........................................................................................4-2
4.1.2 WaterQualityModelCalibration....................................................................................4-2
4.2 StaticCalibrationandDynamicCalibration...........................................................................4-4
4.2.1 Steady-StateCalibrationMethods.................................................................................4-4
4.2.1.1 C-FactorTests .......................................................................................................4-4
4.2.1.2 Fire-FlowTests ......................................................................................................4-5
4.2.1.3 ChlorineDecayTests............................................................................................4-5
4.2.2 DynamicCalibrationMethods.......................................................................................4-6
4.3 ManualCalibrationandAutomatedCalibration....................................................................4-7
4.4 CaseStudies...........................................................................................................................4-7
4.4.1 Case1-Small-Suburban,Dead-EndSystem ...............................................................4-8
4.4.2 Case2-Large-SuburbanPressureZone....................................................................4-10
4.5 FutureofModelCalibration.................................................................................................4-13
4.5.1 CalibrationStandards ..................................................................................................4-13
4.5.2 TechnologicalAdvances..............................................................................................4-14
4.6 SummaryandConclusions .................................................................................................4-14
References.....................................................................................................................................4-15
vi
AReferenceGuideforUtilities
5.0 MonitoringDistributionSystemWaterQuality...........................................................................5-1
5.1 EstablishingMonitoringObjective(s).....................................................................................5-1
5.1.1 RegulatoryDrivenMonitoring.......................................................................................5-2
5.1.2 SecurityRelatedMonitoring..........................................................................................5-2
5.1.3 ProcessControl-RelatedMonitoring.............................................................................. 5-3
5.1.4 WaterQualityCharacterization......................................................................................5-3
5.1.5 Multi-PurposeUseofMonitoringData..........................................................................5-3
5.2 MonitoringTechniques ..........................................................................................................5-3
5.2.1 ManualGrabSampling..................................................................................................5-3
5.2.2 Automated/OnlineMonitoring.......................................................................................5-4
5.3 MonitoringEquipmentOverview..........................................................................................5-4
5.3.1 PhysicalMonitors...........................................................................................................5-4
5.3.2 ChemicalMonitors.........................................................................................................5-4
5.3.3 BiologicalMonitors......................................................................................................... 5-4
5.4 EstablishingMonitoringRequirements..................................................................................5-5
5.4.1 MonitoringParameters...................................................................................................5-5
5.4.2 NumberandLocationofMonitors................................................................................5-5
5.4.2.1 NumberofMonitors.............................................................................................5-6
5.4.2.2 OptimalMonitorLocations ..................................................................................5-6
5.4.3 MonitorCharacteristics..................................................................................................5-7
5.4.4 AmenabilitytoRemoteMonitoringandSCADAIntegration........................................5-8
5.4.4.1 OnlineSampling/ControlDevices........................................................................5-9
5.4.4.2 SCADAorRemoteMonitoringNetwork .............................................................5-9
5.4.4.3 FieldWiringandCommunicationMedia.............................................................5-9
5.5 EngineeringandEvaluatingaRemoteMonitoringSystem ...............................................5-10
5.5.1 RemoteMonitoringSystemEvaluation.......................................................................5-11
5.6 MonitoringCaseStudies......................................................................................................5-11
5.6.1 RuralCommunityApplication......................................................................................5-11
5.6.2 WashingtonD.C.RemoteMonitoringNetwork...........................................................5-11
5.6.3 TucsonWaterMonitoringNetwork.............................................................................5-12
5.7 SummaryandConclusions .................................................................................................5-14
References.....................................................................................................................................5-15
6.0 GeospatialTechnologyforWaterDistributionSystems .............................................................6-1
6.1 HistoryofGeospatialDataManagement...............................................................................6-2
6.1.1 Mapping,Surveying,andRemoteSensing..................................................................6-2
6.1.2 CADD ..............................................................................................................................6-4
6.1.3 GIS..................................................................................................................................6-4
6.1.4 DEMs...............................................................................................................................6-5
6.1.5 DatabaseManagementSystems...................................................................................6-6
6.1.6 FacilityManagement......................................................................................................6-6
6.2 GISPrinciples .........................................................................................................................6-6
6.2.1 GISFeatures...................................................................................................................6-6
6.2.2 Topology.........................................................................................................................6-7
6.2.3 MapProjections,Datum,andCoordinateSystems......................................................6-8
6.2.4 GISDatabaseDesign......................................................................................................6-8
6.2.5 ManagementofGIS.......................................................................................................6-8
6.3 GeospatialDataManagementintheWaterIndustry............................................................6-9
6.3.1 CADD ..............................................................................................................................6-9
6.3.2 GIS..................................................................................................................................6-9
6.3.3 CIS ..................................................................................................................................6-9
6.3.4 SCADA............................................................................................................................6-9
6.3.5 LIMS................................................................................................................................6-9
6.3.6 SupportTechnology......................................................................................................6-9
6.4 IntegrationofGeospatialDataManagementandModeling...............................................6-10
6.4.1 ModelIntegrationTaxonomy.......................................................................................6-10
6.4.2 IssuesinIntegratingGISandWaterDistributionSystemModels..............................6-11
vii
AReferenceGuideforUtilities
6.5 UseofGISinWaterUtilities-CaseStudies.........................................................................6-11
6.5.1 UseofGISatLVVWD...................................................................................................6-11
6.5.1.1 PressureComplaintResolution..........................................................................6-13
6.5.1.2 WaterMainBreakAnalysis.................................................................................6-13
6.5.2 Geo-codingforDemandForecastingandAllocationatDenverWater ......................6-13
6.6 Summary..............................................................................................................................6-17
References.....................................................................................................................................6-18
7.0 Real-WorldApplicationsPlanning,Analysis,andModelingCaseStudies..............................7-1
7.1 AnalysisofWaterborneOutbreakGideon,Missouri..........................................................7-1
7.1.1 GideonCaseStudyOverview........................................................................................7-1
7.1.2 TheGideonWaterSystemSetup..................................................................................7-1
7.1.3 EPAFieldStudy..............................................................................................................7-2
7.1.4 DistributionSystemEvaluation......................................................................................7-2
7.1.5 CaseStudySummaryandConclusions.......................................................................7-3
7.2 ReconstructingHistoricalContaminationEvents-DoverTownship(TomsRiver),NJ........7-4
7.2.1 CaseStudyOverview.....................................................................................................7-4
7.2.2 OverallModelingApproach...........................................................................................7-5
7.2.3 SimulationTechniques...................................................................................................7-6
7.2.4 SimulationResultsandConclusions.............................................................................7-7
7.3 ApplicationofWaterDistributionSystemModelinginSupportofaRegulatoryRequirement......7-9
7.3.1 IDSERequirementsOverview......................................................................................7-10
7.3.2 ExampleApplicationofModelingintheIDSEProcess..............................................7-10
7.4 UseofWaterDistributionSystemModelsinthePlacementofMonitorstoDetectIntentional
Contamination......................................................................................................................7-13
7.4.1 RedTeam-BlueTeamExercise.....................................................................................7-13
7.4.2 ApplicationofOptimizationModel..............................................................................7-14
7.4.3 CaseSummary.............................................................................................................7-15
7.5CaseStudyUseofPipelineNetModel................................................................................7-15
7.5.1 Overview.......................................................................................................................7-16
7.5.2 ModelCalibration.........................................................................................................7-16
7.5.3 MonitoringSiteLocationMethodology......................................................................7-16
7.5.4 ResponseandMitigationTools....................................................................................7-17
7.5.4.1 ConsequenceAssessmentTool .........................................................................7-17
7.5.4.2 IsolationTool.......................................................................................................7-17
7.5.4.3 SpatialDatabaseDisplayTool ............................................................................7-17
7.5.5 CaseSummary.............................................................................................................7-17
7.6 UseofThreatEnsembleVulnerabilityAssessment(TEVA)ProgramforDrinkingWater
DistributionSystemSecurity ...............................................................................................7-18
7.6.1 TEVAOverview.............................................................................................................7-18
7.6.1.1 StochasticModeling...........................................................................................7-19
7.6.1.2 ImpactAnalysis...................................................................................................7-19
7.6.1.3 ThreatMitigationAnalysis..................................................................................7-19
7.6.2 ApplicationofTEVAtoaWaterDistributionSystemforOptimalMonitoring...........7-19
7.6.2.1 SimulationOverview..........................................................................................7-20
7.6.2.2 TEVAAnalysisApproach ...................................................................................7-20
7.6.2.3 TEVAAnalysisResults........................................................................................7-21
7.6.3CaseSummary..............................................................................................................7-22
7.7 FieldTestingofWater-DistributionSystemsinSupportofanEpidemiologicStudy ........7-22
7.7.1 CaseStudyOverview...................................................................................................7-22
7.7.2 FieldWork.....................................................................................................................7-23
7.7.2.1 C-FactorandFire-FlowTests..............................................................................7-23
7.7.2.2 TracerTestandContinuousMeasurements......................................................7-23
7.7.3 AdditionalTestProcedures..........................................................................................7-24
7.7.4 CaseStudySummary..................................................................................................7-25
7.8 ChapterSummary................................................................................................................7-25
References.....................................................................................................................................7-25
viii
AReferenceGuideforUtilities
ListofTables
Table Page
Table1-1. SelectedRulesandRegulationsDealingwithDistributionSystems(NotInclusive).....1-9
Table2-1. AvailableHydraulicandWaterQualityNetworkModelingSoftwarePackages..........2-13
Table2-2. ExampleCFDModelingSoftwarePackages.................................................................2-14
Table2-3. ExampleTransientModelingSoftwarePackages.........................................................2-15
Table3-1. TracerCharacteristics(adaptedfromTeefy,1996) .......................................................... 3-7
Table3-2. EquipmentCosts............................................................................................................3-14
Table3-3. RepresentativeLaborHoursforaRangeofStudies....................................................3-15
Table4-1 Calibration/InputRequirementsforWaterQualityModels.............................................4-3
Table4-2 DraftCalibrationCriteriaforModeling(basedonECAC,1999)...................................4-13
Table5-1. FederalDistributionSystemWaterQualityMonitoringRequirements..........................5-2
Table7-1. MasterOperatingCriteriaUsedtoDevelopOperatingSchedulesfortheHistoricalWater
DistributionSystem,DoverTownshipArea,NJ(fromMasliaetal.,2001)....................7-6
Table7-2. FieldDataCollectedDuringFire-FlowTestatSiteH02.................................................7-24
ListofFigures
Figure Page
Figure1-1. WaterTreatmentProcessattheMillerPlantontheOhioRiver(Adaptedfrom:GCWW
2005).................................................................................................................................1-1
Figure1-2. DistributionSystemInteractionsthatAffectWaterQuality(Adaptedfrom:MSU,2005)..1-3
Figure1-3. TotalNumberandProportionofU.S.WaterborneDiseasesAssociatedwithWater
DistributionSystemDeficiencies.....................................................................................1-6
Figure1-4. EvolutionofFederalDrinkingWaterRegulations-Timeline..........................................1-8
Figure2-1. SimpleLink-NodeRepresentationofaWaterDistributionSystem................................2-2
Figure2-2. AFlowChartforEstimatingFutureWaterDemandBasedonLand-UseMethodology...2-5
Figure2-3. EPANETGraphicalOutputShowingFlowandPressure...............................................2-7
Figure2-4. SampleEPANETTimeSeriesPlotsofFlow,Pressure,andTankWaterLevel. .............2-7
Figure2-5. EPANETSampleTabularOutputs(attime10.00hrs).....................................................2-8
Figure2-6. IllustrationoftheEvolutionofHydraulicandWaterQualityModels.............................2-9
Figure2-7. GraphicalOutputfromaCFDModelShowingTracerConcentrationinaTank. .......2-14
Figure2-8. ALargePhysicalModelofaTank(Source:BureauofReclamationLaboratory).......2-14
Figure2-9. GraphicalOutputBasedon3-DLaserInducedFluorescencewithaPhysicalScale
ModelShowingMixinginTank(Source:GeorgiaTech).............................................2-15
Figure2-10. NegativePressureTransientAssociatedwithaPowerOutage. ..................................2-16
Figure2-11a. VelocityFieldataJunction. ...........................................................................................2-18
Figure2-11b.TracerConcentrationataJunction..............................................................................2-18
Figure3-1. FlowCalibrationTube......................................................................................................3-9
Figure3-2. AutomatedMonitoringStation.....................................................................................3-10
Figure3-3. TracerInjectionSetup(StorageTank,CalibrationTubeandFeedPump)..................3-11
Figure4-1. ConceptualRepresentationofCalibration......................................................................4-1
Figure4-2. EffectsoftheInitialWaterAgeontheModeledResults. ................................................ 4-3
Figure4-3. SchematicofStandardTwo-GageC-FactorTestSetup. ................................................ 4-4
Figure4-4. SchematicofParallelHoseC-FactorTestSetup.............................................................4-4
Figure4-5. Fire-FlowTestSetup. .......................................................................................................4-5
Figure4-6. AHydrantBeingFlowedwithaDiffuserasPartofaFire-FlowTest..............................4-5
Figure4-7. SchematicRepresentationofSmall-SuburbanDead-EndSystem................................4-8
Figure4-8. EmpiricalRelationshipBetweenChlorideandConductivity..........................................4-8
Figure4-9. SampleChlorideDataUsedatOneStationforCalibration...........................................4-9
Figure4-10. ComparisonofModelVersusFieldResultsforContinuousMonitorLocationCM-18at
VariousCalibrationStages..............................................................................................4-9
ix
AReferenceGuideforUtilities
Figure4-11. CalibrationofLoopedPortion....................................................................................4-10
Figure4-12. SchematicRepresentationofCase2StudyLocation..................................................4-11
Figure4-13a.ModeledFlowsComparedtoMeasuredFlowsBeforeCalibration.............................4-11
Figure4-13b.ModeledFlowsComparedtoMeasuredFlowsAfterCalibration................................4-11
Figure4-14a.ChlorideConcentrationforCalibrationEventatContinuousMonitorLocationCM-59. 4-12
Figure4-14b.ChlorideConcentrationforValidationEventatContinuousMonitorLocationCM-59...4-12
Figure5-1. WastewaterInjection:FreeChlorineandAssociatedGrabSampleResults. ................5-6
Figure5-2. TheoreticalExampleofBenefitsfromMonitors.............................................................5-6
Figure5-3. Pareto-OptimalCostEffectivenessDiagram....................................................................5-7
Figure5-4. FortReno#2RemoteSamplingSystem......................................................................5-12
Figure5-5. WASARemoteMonitoringSystemLayoutandDataTransmissionScheme.............5-12
Figure5-6. MonitoringDataforFortRenoTank..............................................................................5-13
Figure5-7. CityofTucsonWaterQualityZoneMap.......................................................................5-13
Figure5-8. ContinuousWaterQualityMonitoringStation.............................................................5-13
Figure6-1. MercatorsMapoftheWorldin1569(Whitfield,1994)..................................................6-2
Figure6-2. LandsatThematicMapperImagesoftheMissouriRiverFloodplainNearGlasgow,
Missouri.(USGS,1993). ...................................................................................................6-4
Figure6-3. TypicalInputsandResultsofCurrentGISPackages......................................................6-5
Figure6-4. DigitalTerrainModelofMountSt.HelensafterEruptionin1980(R.Horne,2004)......6-6
Figure6-5. MapofPressureZoneShowingThreeTypesofGISVectorData. ...............................6-6
Figure6-6. RegionalLandCoverCharacterizationasaRasterDatabase(USGS,1992). ................6-7
Figure6-7. TriangulationofElevation(Z)Data..................................................................................6-7
Figure6-8a. TypicalRepresentationofaPipeSectioninGIS. ..........................................................6-11
Figure6-8b. TypicalRepresentationofaPipeSectioninaNetworkModel....................................6-11
Figure6-9. LVVWDDistributionSystemGrowth............................................................................6-12
Figure6-10. ConceptualRelationshipModelforIntegration............................................................6-12
Figure6-11. One-to-OneRelationshipBetweenGISandNetworkModelingData..........................6-12
Figure6-12. PressureComplaintResolutionGISParcel/AccountSearchWindow......................6-13
Figure6-13. PressureComplaintResolutionParcelandHydrantLocation. .................................. 6-14
Figure6-14. PressureComplaintResolutionModelandFieldPressureComparison..................6-14
Figure6-15. WaterMainBreakAnalysisValveIsolation.................................................................6-15
Figure6-16.WaterMainBreakAnalysisImpactedCustomerList.................................................6-15
Figure6-17. WaterMainBreakAnalysisComparisonofJunctionPressures-BeforeandAfter
Shutdown......................................................................................................................6-16
Figure6-18. GISGeo-coding-MeteredSalesDemandAllocationProcedure................................6-17
Figure7-1. ComparisonofEarlyConfirmedCasesofSalmonellaPositiveSampleVersusthe
EstimatedDistributionofTankWaterDuringtheFirst6HoursoftheFlushingProgram. 7-3
Figure7-2. InvestigationArea,DoverTownship,OceanCounty,NJ(modifiedfromMasliaetal.,
2001).................................................................................................................................7-4
Figure7-3. DistributionSystemZonesWoburn,MA(May1969)..................................................7-5
Figure7-4. Three-DimensionalRepresentationofMonthlyProductionofWater,DoverTownship
Area,NJ(fromMasliaetal.,2001)..................................................................................7-6
Figure7-5. ArealDistributionofSimulatedProportionateContributionofWaterfromtheParkway
Wells(22,23,24,26,28,29)toLocationsintheDoverTownshipArea,NJ,July1988
Conditions(fromMasliaetal.,2001). ..............................................................................7-8
Figure7-6. SimulatedProportionateContributionofWaterfromWellsandWellFieldstoSelected
Locations,DoverTownshipArea,NJ,July1988Conditions(fromMasliaetal.,2001)....7-8
Figure7-7. EstimatedUpper97.5PercentCredibilityLimitforAnnualPerchlorateIntakebyOne
Plaintiff(Grayman,2004). ................................................................................................7-9
Figure7-8. AverageWaterAgeintheDistributionSystemOverLast24Hoursofa2-Week
Simulation......................................................................................................................7-11
Figure7-9. MinimumWaterAgeintheDistributionSystemOverLast24Hoursofa2-Week
Simulation......................................................................................................................7-11
Figure7-10. DiurnalWaterAgeatNodeJ-456..................................................................................7-12
Figure7-11. MinimumChlorineResidualinDistributionSystemOverLast24Hoursofa2-Week
Simulation......................................................................................................................7-12
x
AReferenceGuideforUtilities
Figure7-12. ZonesRepresentingPotentialMonitoringLocationsforIDSEBasedonModeling....7-12
Figure7-15. ContaminantConcentrationJustDownstreamofContaminantIntroductionLocation
Figure7-16. ContaminantConcentrationFarDownstreamofContaminantIntroductionLocation
Figure7-20. HypotheticalSystemShowingHighScoreAreas(>27)OverlainwithHospitalsand
Figure7-21. DisplayofLow-VelocityPipes,OversizedPipes,andCurrentMonitoringStationsUsing
Figure7-23. Comparisonof24-Hour,48-Hour,andReal-Time,ContinuousContamination
MonitoringSystemsfortheReductioninMeanInfectionsfora24-HourContaminant
Figure7-24. Comparisonof24-Hour,48-Hour,andReal-Time,ContinuousContamination
Figure7-26. ContinuousRecordingPressureLoggerMountedonBrassShutoffValveandHydrant
Figure7-29. ArrivalTimesoftheCalciumChlorideTraceratMonitoringLocationsinHadnotPoint
Figure7-13. WaterDistributionSystemCharacteristics....................................................................7-13
Figure7-14. AllowableContaminantIntroductionLocations...........................................................7-14
(Node121)......................................................................................................................7-14
(Node143)......................................................................................................................7-14
Figure7-17. MonitoringLocationsSelectedbytheOptimizationModel. ........................................ 7-15
Figure7-18. HypotheticalWaterDistributionSystemShowingPipelines........................................7-16
Figure7-19. ConceptualDiagramShowingtheRankingandPrioritizationMethodology. ............. 7-16
Schools..........................................................................................................................7-17
theSpatialDatabaseDisplayTool. ................................................................................7-18
Figure7-22. ThreatEnsembleVulnerabilityAssessmentFramework. ..............................................7-18
Attack..............................................................................................................................7-21
MonitoringSystemsfortheReductionintheMaximumNumberofInfectionsfora24-
HourContaminantAttack..............................................................................................7-21
Figure7-25. WaterDistributionSystemsServingU.S.MarineCorpsBase,CampLejeune,NC....7-22
AdapterCapUsedforFire-FlowandC-FactorTests. ....................................................7-23
Figure7-27. LocationofFireHydrantsUsedinFire-FlowTestatSiteH02.......................................7-23
Figure7-28. HoribaW-23XDDualProbeIonDetectorInsideFlowCell...........................................7-24
WTPArea,May25,2004...............................................................................................7-25
xi
AReferenceGuideforUtilities
AcronymsandAbbreviations
AC AlternatingCurrent
ACCNSCM ArsenicandClarificationsto
ComplianceandNewSource
ContaminantMonitoring
ADAPT ArealDesignandPlanningTool
ADEQ ArizonaDepartmentof
EnvironmentalQuality
Al
2
(SO
4
)
3
AluminumSulfate
AM/FM AutomatedMapping(orAsset
Management)/Facilities
Management
AMR AutomatedMeterReading
ASCE AmericanSocietyofCivilEngineers
ATSDR AgencyforToxicSubstancesand
DiseaseRegistry
AWWA AmericanWaterWorksAssociation
AwwaRF AwwaResearchFoundation
C CoefficientofRoughness
ClO
4
PerchlorateAnion
CaCl
2
CalciumChloride
CAD Computer-AidedDesign
CADD Computer-AidedDesignand
Drafting
CDC CentersforDiseaseControland
Prevention
CFD ComputationalFluidDynamics
CIS CustomerInformationSystem
CM ContinuousMonitoring
COGO CoordinatedGeometry
CRT CathodeRayTube
CWS ContaminationWarningSystem
DBP DisinfectionBy-Products
DBPR1 Disinfectant By-Product Rule -
Stage 1
DBPR2 Disinfectant By-Product Rule -
Stage 2
DC Direct Current
D.C. DistrictofColumbia
DEM DigitalElevationModel
DIME DualIndependentMapEncoding
DLG DigitalLineGraph
DSOP DistributionSystemWaterQuality
OptimizationPlan
DSS DistributionSystemSimulator
DTM DigitalTerrainModel
DWQM DynamicWaterQualityModel
EBMUD EastBayMunicipalUtilityDistrict
EDM ElectronicDistanceMeasurement
EMPACT EnvironmentalMonitoringforPublic
AccessandCommunityTracking
EOSAT EarthObservationSatellite
EPA U.S.EnvironmentalProtection
Agency
EPS ExtendedPeriodSimulation
ESSA EnvironmentalScienceServices
Administration
EWS EnvironmentalWarningSystem
FC FecalColiform
FeCl
3
FerricChloride
FOH FederalOccupationalHealth
GA GeneticAlgorithm
gal Gallon
GBF GeographicBaseFile
GC GasChromatograph
GCWW GreaterCincinnatiWaterWorks
GIS GeographicInformationSystem
GPD GallonsPerDay
gpm GallonsPerMinute
GPS GlobalPositioningSystem
GRASS GeographicResourcesAnalysis
SupportSystem
GUI GraphicalUserInterface
GWR GroundWaterRule
HAA HaloaceticAcid
HAA5 ThefiveHaloaceticAcids
HACCP HazardAnalysisCriticalControl
Point
HGL HydraulicGradeLine
HSPP HealthandSafetyProjectPlan
ICR InformationCollectionRule
IDSE InitialDistributionSystem
Evaluation
IESWTR InterimEnhancedSurfaceWater
TreatmentRule
ILSI InternationalLifeSciencesInstitute
I/O Input/Output
ISE IonSelectiveElectrode
ISO InsuranceServicesOffice
LCR LeadandCopperRule
xii
AReferenceGuideforUtilities
LIFO LastIn/FirstOut
LIMS LaboratoryInformation
ManagementSystem
LIS LandInformationSystem
LT1ESWTR LongTerm1EnhancedSurface
WaterTreatmentRule
LT2ESWTR LongTerm2EnhancedSurface
WaterTreatmentRule
LVVWD LasVegasValleyWaterDistrict
ma Milli-Amperes
MCL MaximumContaminantLevel
MCLG MaximumContaminantLevelGoal
MDL MinimumDetectionLimit
MDNR MissouriDepartmentofNatural
Resources
MDOH MissouriDepartmentofHealth
MGD MillionGallonsperDay
mg/L milligramsperliter
MIT MassachusettsInstituteof
Technology
MOC MasterOperatingCriteria
MRDLG MaximumResidualDisinfectant
LevelGoals
MS MassSpectrometer
MSU MontanaStateUniversity
NaCl SodiumChloride
NAD27 NorthAmericanDatumof1927
NAD83 NorthAmericanDatumof1983
NAPP NationalAerialPhotography
Program
NASA NationalAeronauticsandSpace
Administration
NFPA NationalFireProtectionAssociation
NHAP NationalHighAltitudePhotography
NIPDWR NationalInterimPrimaryDrinking
WaterRegulations
NJDHSS NewJerseyDepartmentofHealth
andSeniorServices
NMWD NorthMarinWaterDistrict
NPL NationalPrioritiesList
NPWA NorthPennWaterAuthority
NOM NaturallyOccurringOrganic(and/
orInorganic)Matter
NRC NationalResearchCouncil
O&M OperationsandMaintenance
OCMS OnlineContaminantMonitoring
System
ODBC OpenDatabaseConnectivity
ORD OfficeofResearchandDevelopment
ORP OxidationReductionPotential
PAB3D AThree-DimensionalComputational
FluidDynamicsModeldevelopedby
AnalyticalServices&Materials,Inc.
PC PersonalComputer
PDD PresidentialDecisionDirective
PHRP PublicHealthResponsePlan
PL PublicLaw
POE PointofEntry
POGA ProgressiveOptimalityGenetic
Algorithm
psi PoundsPerSquareInch
PVC PolyvinylChloride
PWS PublicWaterSystem
QA QualityAssurance
QAPP QualityAssuranceProjectPlan
QC QualityControl
RDBMS RelationalDatabaseManagement
Systems
RDWR RadoninDrinkingWaterRule
SAN StyreneAcrylonitrile
SCADA SupervisoryControlandData
Acquisition
SCCRWASouthCentralConnecticutRegional
WaterAuthority
SDMS SpatialDatabaseManagementSystem
SDWA SafeDrinkingWaterAct
SDWAA SafeDrinkingWaterActAmendments
SMP StandardMonitoringProgram
SNL SupplyNodeLink
SOP StandardOperatingProcedure
SPC StatePlaneCoordinates
SSS SystemSpecificStudy
SVOC SemivolatileOrganicCompound
SWTR SurfaceWaterTreatmentRule
SYMAP SynagraphicMapping
T&E TestandEvaluation
TCE Trichloroethylene
TCR TotalColiformRule
TEVA ThreatEnsembleVulnerability
Assessment
THM Trihalomethane
TIGER TopologicallyIntegratedGeographic
EncodingandReferencing
xiii
AReferenceGuideforUtilities
TIN TriangulatedIrregularNetwork
TIROS1 TelevisionandInfraredObservation
Satellite1
TOC TotalOrganicCarbon
TT TreatmentTechnique
TTHM TotalTrihalomethane
TV Television
U.S. UnitedStates
UF Ultrafiltration
UHF UltraHighFrequency
UV Ultraviolet
UV-Vis Ultraviolet-Visible
USGS UnitedStatesGeologicalSurvey
UTM UniversalTransverseMercator
VHF VeryHighFrequency
WASA WaterandSewerAuthority
WATERS WaterAwarenessTechnology
EvaluationResearchandSecurity
WQP WaterQualityParameter
WRC WaterResearchCentre
WSSM WaterSupplySimulationModel
WSTP Wells,StorageTanks,andPumps
WTP WaterTreatmentPlant
xiv
AReferenceGuideforUtilities
Acknowledgments
EPAwouldliketoacknowledgetheprincipalauthorsofthedocument,WaterDistributionSystem
Analysis:FieldStudies,Modeling,andManagement,aReferenceGuideforUtilities. Theauthors
were Mr. Srinivas Panguluri, P.E. (Shaw Environmental, Inc.), Dr. Walter M. Grayman, P.E. (W.M.
GraymanConsultingEngineer),andDr.RobertM.Clark,P.E.,D.E.E.(EnvironmentalEngineeringand
Public Health Consultant). The work was performed under EPA Contract No. EP-C-04-034, Work
AssignmentNo.0-08and1-08,withShawEnvironmental,Inc.
EPAacknowledgesthesignificantcontributionsfromMr.DavidJ.HartmanandDr.YeonghoLee,
P.E.,oftheGreaterCincinnatiWaterWorks.Mr.HartmanandDr.Leeservedasexternalreviewers
anddirectlycontributedtothedevelopmentofportionsofthisdocument. GreaterCincinnatiWater
Works also participated significantly in the field tracer tests that served as a foundation for the
developmentofthisdocument.
EPAacknowledgesthepeerreviewsbythefollowingutilitypersonnel:Mr.WilliamR.Kirkpatrickand
Mr.RonaldB.HunsingerofEastBayMunicipalUtilityDistrict,andMr.ArnoldStrasser,P.E.,ofDenver
Water.
EPAalsoacknowledgesthepeerreviewsprovidedforthisdocumentbyDr.FranA.DiGiano,P.E.,of
UniversityofNorthCarolina,Mr.MorrisL.Maslia,P.E.,D.E.E.,oftheAgencyforToxicSubstances
andDiseaseRegistry,Dr.EdwardA.McBean,P.E.,ofUniversityofGuelph,Dr.VanessaL.Speight,
P.E.,ofMalcolmPirnie,Ms.L.MichelleMooreofNationalDrinkingWaterClearinghouse,andMr.E.
RadhaKrishnan,P.E.,ofShawEnvironmental,Inc.
EPAalsogratefullyacknowledgesthefollowingcontributorstothisdocument:Mr.DanielR.Quintanar
ofTucsonWater,Ms.LauraB.Jacobson,P.E.,Mr.SridharKamojjala,P.E.,andMr.MaoFang,P.E.,of
LasVegasValleyWaterDistrict,Mr.MorrisL.Maslia,P.E.,D.E.E.,oftheAgencyforToxicSubstances
andDiseaseRegistry,Dr.WilliamB.Samuels,P.E.,ofSAIC,Dr.MarkW.LeChevallierandDr.KalaK.
FlemingofAmericanWater.
EPAContributors
Ms.ShirleyJ.GibsonfunctionedasProjectOfficerofEPAContractNo.EP-C-04-034,Ms.LucilleM.
GarnerservedasWorkAssignmentManagerandMr.CraigL.Patterson,P.E.,servedastheAlternate
WorkAssignmentManagerandtechnicalreviewerforthisdocument.Mr.RoyC.Haughtservedas
EPATechnicalAdvisor.Mr.JonathanG.Herrmann,P.E.,Mr.KimR.Fox,P.E.,D.E.E.,andDr.James
A.Goodrichperformedtechnicalreviewsofthedocument. Mr.StephenM.HarmonwastheEPA
QualityAssuranceManager,andwasresponsibleforthequalityassurancereviewofthedocument.
ThedocumentwasalsoreviewedbymanyotherEPAreviewersincludingthefollowing:Mr.BlakeL.
Atkins,Mr.WilliamH.Davis,Ms.JeanE.Dye,Mr.RobertJ.Janke,Mr.BruceA.Macler,Ms.JillR.
Neal,Dr.LewisA.Rossman,P.E.,Mr.KennethH.Rotert,Dr.IrwinJ.Silverstein,P.E.,D.E.E.,andMs.
ElinA.Warn.
xv
Chapter1
Introduction
DrinkingwaterutilitiesintheUnitedStates(U.S.)
andthroughouttheworldfacethechallengeof
providingwaterofgoodqualitytotheirconsumers.
Frequently,thewatersupplyisderivedfromsurface
waterorgroundwatersourcesthatmaybesubjectto
naturally occurring or accidentally introduced
contamination(ILSI,1999;Gullicketal.,2003).In
othercases,routineupstreamwastedischargesor
purposefulcontaminationofthewatercandiminish
thequalityofthewater.Thetreatedwatermaybe
transmittedthroughanetworkofcorrodedor
deterioratingpipes. Allofthesefactorscanresult
indegradationinthequalityofthewaterdelivered
tocustomers.
IntheU.S.,drinkingwaterqualityhastocomply
withfederal,state,andlocalregulations. Thisis
basedonselectedphysical,chemical,andbiologi-
calcharacteristicsofthewater.TheU.S.Environ-
mentalProtectionAgency(EPA)haspromulgated
manydrinkingwaterstandardsundertheSafe
DrinkingWaterAct(SDWA)of1974.Theserules
andregulationsrequirethatpublicwatersystems
(PWSs)meetspecificguidelinesand/ornumeric
standardsforwaterquality.TheSDWAdefinesa
PWSasasystemthatservespipedwatertoatleast
25personsor15serviceconnectionsforatleast60
dayseachyear.Forthepurposesofthisreference
guide,PWSsarereferredtoasutilities.
TheSDWAhasestablishedtwotypesofnumeric
standards. Thefirsttypeofstandardisenforceable
andreferredtoasamaximumcontaminantlevel
(MCL). Theothernon-enforceablestandardis
referredtoasamaximumcontaminantlevelgoal
(MCLG). MCLGsaresetatalevelatwhichnoknown
oranticipatedadversehumanhealtheffectsoccur.
Whereitisnoteconomicallyortechnologically
feasibletodeterminethelevelofacontaminant,a
treatmenttechnique(TT)isprescribedbyEPAinlieu
ofestablishinganMCL. Forexample,Giardiaisa
microbialcontaminantthatisdifficulttomeasure.To
ensureproperremoval,experimentalworkhas
establishedoptimumtreatmentconditionsforthe
waterataspecifiedpH,temperature,andchlorine
concentrationforaspecifiedlengthoftimetoachieve
afixedlevelofinactivation.
CompliancewithMCLandTTrequirementsis
typicallyensuredbyrequiringthatwaterutilities
periodicallymonitorvariouscharacteristicsofthe
treatedwater.Insummary,theEPAGuidelinesand
Standardsaredesignedtoensurethatdrinking
waterisadequatelytreatedandmanagedbywater
A Reference Guide for Utilities
Removingcontaminantsfromdrinkingwatercanbe
expensive. Dependinguponthetypeandlevelof
contaminant(s)presentinthesourcewater,utilitiescan
choosefromavarietyoftreatmentprocesses. These
individualprocessescanbearrangedinatreatment
train(aseriesofprocessesappliedinasequence).
Themostcommonlyusedtreatmentprocessesinclude
coagulation/flocculation, sedimentation, filtration, and
disinfection. Somewatersystemsalsouseionex-
change,membraneseparation,ozonation,orcarbon
adsorptionfortreatment. Thebasictreatmentoptions
arebrieflydiscussedlaterinthischapter.Asan
example,Figure1-1depictsthewatertreatment
processimplementedbytheGreaterCincinnatiWater
Works(GCWW)attheMillerPlantontheOhioRiver.
Presettling
Finalsettlingoccurs,
removesmostsolids
waterprepared
Furthersettling forfinaltreatment
occursinreservoir
pHadjusted
Settlingaidsadded
OhioRiver
Intake
Pumps
Todistribution
system
Sandand
Granularactivated
gravelfilter
water
Furnace
cleanscarbon
forreuse
organics
carbonremoves
pHadjusted,
chlorineadded,
fluorideadded
Reservoir
Figure 1-1. Water Treatment Process at the Miller
Plant on the Ohio River (Adapted from: GCWW 2005).
utilities to support public safety, protect public
health,andpromoteeconomicgrowth(Clarkand
Feige,1993).
Disinfectionofdrinkingwaterisconsideredtobeone
ofthemajorpublichealthadvancesofthe20th
century.Thesuccessfulapplicationofchlorineasa
disinfectantwasfirstdemonstratedinEngland. In
1908,JerseyCity(NJ)initiatedtheuseofchlorinefor
waterdisinfectionintheU.S. Thisapproachsubse-
quentlyspreadtootherlocations,andsoontherates
ofcommonepidemicssuchastyphoidandcholera
droppeddramatically.Today,disinfectionisan
essentialpartofadrinkingwatertreatmenttrain.
Chlorine,chlorinedioxide,andchloraminesaremost
1-1
Non-potablewaters(e.g.,sea,river,andlakewater)
withoutadequatetreatmenthavebeenusedforfire
protectionformanyyears,oftenwithdisastrousresults.
However,reclaimedwastewater(incaseswhereits
qualityisbettermanagedthantheaforementioned
unregulatedsources)hasbeeneffectivelyusedfor
providingfireprotection(AwwaRF,2002).St.Peters-
burg,FL,hasbeenoperatingsuchasystemtobolster
fire-protectioncapacitysince1976. Thereclaimed
waterhydrantsaredistinguishedfrompotablewater
hydrantsbycolorandtheirspecialvalves. Ifthe
reclaimedwatersystemisdesignedforfireprotection,
thepotablewaterpipingcanhaveaverysmalldiameter
andinvestmentscanbemadeinhigherqualitypipe
materials,which,withmuchshorterresidencetimein
thesystem,wouldvastlyimprovethequalityofthe
wateratthetap.Withthisinmind,whereretrofitting
oneofthetwosystemsisnecessary,itmightbewiserto
usetheexistingpotablewatersystemforthereclaimed
waterandretrofitwithnew,high-quality,smaller,
potablewaterlines(Okun,D.,1996).
treatmentplantandtheconsumer.Theselongtravel
timesandlowvelocitiesmaybedetrimentalto
meetingthedrinkingwaterMCLs. Longresidence
timesmayleadtoformationofDBPs,lossofdisinfectant
residuals,bacterialgrowth,andformationofbiofilm.
1.1.2.1 Pipe-NetworkConfigurations
Thebranchandgrid/looparethetwobasicconfigura-
tionsformostwaterdistributionsystems.Abranch
systemissimilartothatofatreebranchwithsmaller
pipesbranchingofflargerpipesthroughoutthe
servicearea. Thistypeofsystemismostfrequently
usedinruralareas,andthewaterhasonlyone
possiblepathwayfromthesourcetotheconsumer.A
grid/loopsystemisthemostwidelyusedconfigura-
tioninlargemunicipalsystemsandconsistsof
interconnectedpipeloopsthroughouttheareatobe
served. Inthistypeofsystem,thereareseveral
pathwaysthatthewatercanfollowfromthesourceto
theconsumer.Transmissionmainsaretypically20to
24inchesindiameterorlarger. Dual-servicemains
thatservebothtransmissionanddistributionpurposes
arenormally12to20inchesindiameter. Distribu-
tionmainsareusually6to12inchesindiameterin
everystreet.Servicelinesaretypically1inchin
diameter.Specificpipesizescanvarydependingon
theextentofthedistributionsystemandthemagni-
tudeofdemand.Loopedsystemsprovideahigh
degreeofreliabilityshouldalinebreakoccur,
becausethebreakcanbeisolatedwithlittleimpact
onconsumersoutsidetheimmediatearea(Clarkand
Tippen,1990;Clarketal.,2004).
1.1.2.2 MultipleSourceConfiguration
Manysystemsservecommunitieswithmultiple
A Reference Guide for Utilities
sourcesofsupply,suchasacombinationofwellsand/or
surfacesources. Inagrid/loopedsystem,thisconfigu-
rationwillinfluencewaterqualityinadistribution
systemduetotheeffectofmixingofwaterfromthese
differentsources. Theseinteractionsareafunctionof
complexsystemhydraulics(Clarketal.,1988;Clark
etal.,1991a). Waterqualitymodelscanbevery
usefulindefiningmixingandblendingzoneswithin
waterutilitydistributionnetworks. Mixingofwater
inanetworkcanresultintasteandodorproblemsor
otherwaterqualityproblemsandcaninfluence
maintenance,repair,andrehabilitationprocedures.
1.1.2.3 ImpactofSystemDesignandOperation
onWaterQuality
Basedonthedesignandconfigurationofaparticular
system,therearemanyopportunitiesforwaterquality
tochangeaswatermovesbetweenthetreatmentplant
andtheconsumer. Theseunwantedchangesmay
occurduetovariousreasonsincluding:failuresatthe
treatmentbarrier,transformationsinthebulkphase,
corrosionandleachingofpipematerial,biofilm
formation,andmixingbetweendifferentsourcesof
water.Manyresearchershaveinvestigatedthefactors
thatinfluencewaterqualitydeteriorationonceit
entersthedistributionsystem. Ithasbeenwell
documentedthatbacteriologicalgrowthcancause
taste-and-odorproblems,discoloration,slimebuildup,
andeconomicproblems,includingcorrosionofpipes
andbio-deteriorationofmaterials(WaterResearch
Centre,1976). Bacterialnumberstendtoincrease
duringdistributionandareinfluencedbyseveral
factors,includingbacterialqualityofthefinished
waterenteringthesystem,temperature,residence
time,presenceorabsenceofadisinfectantresidual,
constructionmaterials,andavailabilityofnutrients
forgrowth(Geldreichetal.,1972;LeChevallieretal.,
1987;Mauletal.,1985aandb;ZhangandDiGiano,
2002;Camperetal.,2003).
Itisdifficultandexpensivetostudytheproblems
causedbysystemdesignandconfigurationinfull-
scalesystems. Forexample,oneapproachto
studyingresidualchlorinelevelsindead-endor
low-flowsituationswouldbetoconstructapilot-
scalepipesystemtosimulatethephenomena.
Anotherapproachwouldbetousemathematical
hydraulicandwaterqualitymodelsforsimulation.
Foreitheroftheseapproachestowork,theymust
beproperlyconfiguredand/orcalibratedtoclosely
simulateafull-scalesystem. Acombinationof
theseapproachesmaybeusedtoassessvarious
operational and design decisions, to determine the
impactsresultingfromtheinadvertentordeliberate
introduction of a contaminant into the distribution
system,andtoassistinthedesignofsystemsto
improvewaterquality.
1-5
Figure 1-3. Total Number and Proportion of U.S. Waterborne Diseases Associated with Water Distribution
System Deficiencies.
A Reference Guide for Utilities
Inpipes,ithasbeenfoundthatchlorinecanbelost
throughboththeinteractionwithNOMinthebulk
phaseandwithpipewallsthemselvesintransporting
finishedwater.Thismechanismforlossofchlorinemay
beevenmoreseriousthanlongresidencetimesin
tanks. Thepipewalldemand,possiblyduetobiofilm
andtubercles,mayuseupthechlorineveryrapidlyina
distribution system. Maintaining adequate levels of
disinfectantresidualmayrequireroutinecleaning/
replacementofpipesandintensivetreatment(Clarket
al.,1993a).
in1989andaSalmonella outbreakinGideon,
Missouri,in1993. Thesetwoexamples,discussed
laterinChapter7,illustratetheimportanceofthe
multiple-barrierconcept. Inbothcases,thewater
sourcewasun-disinfectedgroundwaterandthe
utilitysinfrastructurewasbreached,allowing
contaminantstoenterthesystem. Thiscontamination
resultedinmajorwaterborneoutbreaks.Waterquality
modelingwasusedinbothcasestoidentifythe
sourceoftheoutbreaksandtostudythepropagation
oftheoutbreakthroughthedistributionnetwork
(Clarketal.,1993aandb).
1.2 Water Quality Problems and
Issues
DrinkingwatertreatmentintheU.S.hasplayeda
majorroleinreducingwaterbornedisease. For
example,thetyphoiddeathrateforaparticularyear
inthe1880swas158per100,000inPittsburgh,PA,
comparedwith5per100,000in1935. Suchdramatic
reductionsinwaterbornediseaseoutbreakswere
broughtaboutbytheapplicationofdrinkingwater
standardsandengineeringmultiplebarriersof
protection. Themultiple-barrierconceptincludesthe
useofconventionaltreatment(e.g.,sandfiltration)in
combinationwithdisinfectiontoprovidesafeand
aestheticallyacceptabledrinkingwater.Theresidual
disinfectantlevelsservedtoprotectthewaterquality
withinthedistributionsystempriortoitsdeliveryto
theconsumer(Clarketal.,1991b).
DespitethepassageoftheSDWA,waterborneout-
breaksstilloccur. Twoextensivelystudiedexamples
ofwaterbornediseaseintheU.S.wereanEscherichia
coli O157:H7 (E. coli) outbreakinCabool,Missouri,
OneusefuloutcomeoftheoutbreaksinMissouriis
that the ensuing investigative studies have typically
ledtothedevelopmentandenhancementofscientific
analysistechniques. Forexample,theGideon
Salmonella outbreakconclusionswerebasedon
statisticalstudiesperformedbyCentersforDisease
ControlandPrevention(CDC)andcorroboratedby
waterqualitymodelingperformedbyEPA.Thestudy
providesanexampleofhowtoolssuchaswater
qualitymodelscanbeusedtoreliablystudycontami-
nantpropagationinadistributionsystem(Clarketal.,
1996). BoththeGideonandCaboolincidentswere
associatedwithsourcewatercontamination,inad-
equatetreatment,andbreechesinthedistribution
system.
Thesetypesofproblemsarenotjustisolatedincidents
ofinfrastructurebreakdowns. Infact,severalprob-
lemswithdrinkingwatersystemsintheU.S.have
beenidentifiedbyresearchers. TheNationalResearch
Council(NRC,2005)examinedthecausesofwater-
borneoutbreaksreportedbyvariousinvestigators
between1971and2004. Figure1-3presentsthetotal
numberandproportionofwaterbornediseases
associatedwithdistributionsystemdeficiencies
1-6
OnDecember16,1974,theU.S.Congresspassedthe
SDWA,whichauthorizedtheEPAtopromulgate
regulationswhichwouldprotecthealthtotheextent
feasible,usingtechnology,treatmenttechniques,and
othermeans,whichtheAdministratordeterminesare
generallyavailable(takingcostsintoconsidera-
tion)(SDWA,1974).Asaresult,asetofregulations
waspromulgatedin1975whichbecameeffectiveJune
24,1977. ThesewereknownastheNationalInterim
PrimaryDrinkingWaterRegulations(NIPDWR).The
NIPDWRestablishedMCLsfor10inorganiccontami-
nants,sixorganiccontaminants,turbidity,coliform,
radium-226,radium-228,grossalphaactivity,and
man-maderadionuclides. TheNIPDWRalsoestab-
lishedmonitoringandanalyticalrequirementsfor
determining compliance.
(extractedfromtheNRCreport). Asthefigure
reveals,overallthereisageneraldecreaseinthe
totalnumberofwaterbornediseaseoutbreaks
duringthereportedperiod. However,thereisa
generalincreaseinthepercentageofoutbreaksthat
areassociatedwithdistributionsystemdeficiencies.
TheNRCreportattributesthisincreaseinpercent-
ageofoutbreaks(attributabletodistributionsystem
deficiencies)tothelackofhistoricalregulatory
focusondistributionsystems.
1.3 Regulatory Framework
Concernsaboutwaterbornediseaseanduncon-
trolledwaterpollutionresultedinfederalwater
quality legislation starting in 1893 with the
passageoftheInterstateQuarantineActand
continuingto1970underthestewardshipofthe
U.S.PublicHealthService(AWWA,1999).Even
thoughsignificantadvancesweremadetoeliminate
waterbornediseaseoutbreaksduringthatperiod,
thefocusofdrinkingwaterconcernsbeganto
changewiththeformationoftheEPAinlate1970.
Bythe1970s,morethan12,000chemicalcom-
poundswereknowntobeincommercialuseand
manymorewerebeingaddedeachyear.Manyof
thesechemicalscausecontaminationofgroundwa-
terandsurfacewater,andareknowntobecarcino-
genicand/ortoxic.ThepassageoftheSDWAof
1974wasareflectionofconcernsaboutchemical
contamination. Inthissection,abriefoverviewof
theregulatoryframeworkispresented. Adetailed
historyoftheevolutionofthefederaldrinking
waterregulationsisbeyondthescopeofthis
document.
EarlyinthehistoryoftheSDWA,themajorfocusof
EPAwastoimplementtheActandtoinitiatethe
regulatoryprocess. ThefirstMCLestablished
A Reference Guide for Utilities
undertheSDWAwastheTTHMRulein1979.
However,afterseveralyearsofdevelopingregula-
tions,itbecameobviousthattherulemaking
processmustextendbeyondafocusonMCLsatthe
treatmentplantandintothedistributionsystem.
ManywaterutilitiesintheU.S.usingsurface
supplieswereexperiencingwaterborneoutbreaks,
especiallyfromGiardia. The1986SDWAAmend-
mentslaidthegroundworkforthepromulgationof
theTotalColiformRule(TCR)andtheSurface
WaterTreatmentRule(SWTR)in1989.The1986
SDWAAmendmentsalsosetforthanaggressive
plantoeliminateleadfromPWSsandresultedin
thepromulgationoftheLeadandCopperRule
(LCR)in1991. Theseactionsthereforeextended
theSDWAbeyonditsfocusonthetreatmentplant
andintothedistributionsystem(Owens,2001).
Asummaryoftheevolutionoffederaldrinkingwater
regulationsincethepassageoftheSDWAin1974is
presentedinFigure1-4.Inadditiontotherulesand
regulationspromulgatedundertheSDWA,security
hasrecentlybecomeanissueforthewaterutility
industry. Securityofwatersystemsisnotanewissue.
Thepotentialfornatural,accidental,andpurposeful
contaminationofwatersupplieshasbeenthesubject
ofmanystudies. Forexample,inMay1998,President
ClintonissuedPresidentialDecisionDirective(PDD)
63thatoutlinedapolicyoncriticalinfrastructure
protection,includingournationswatersupplies.
However,itwasnotuntilafterSeptember11,2001,
thatthewaterindustryfocusedonthevulnerabilityof
thenationswatersuppliestosecuritythreats.In
recognitionoftheseissues,PresidentGeorgeW.Bush
signedthePublicHealthSecurityandBioterrorism
PreparednessandResponseActof2002(Bioterrorism
Act)intolawinJune2002(PL107-188). Underthe
requirementsoftheBioterrorismAct,community
watersystems(CWSs)servingmorethan3,300people
arerequiredtopreparevulnerabilityassessmentsand
emergencyresponseplans. CWSsarePWSsthat
supplywatertothesamepopulationthroughoutthe
year.
Table1-1summarizesthekeyrequirementsofthe
regulationspresentedinFigure1-4fromadistribu-
tionsystemcomplianceperspective.
Manyofthetoolsandtechniquesdiscussedinthis
referenceguidecanassistincomplyingwiththe
rulesandregulationsandsecurityissuesdiscussed
above.Waterqualitymodelingtechniquescanbe
used to identify points in the distribution system
thatexperiencelongretentiontimes,whichcanin
turnrepresentlocationsinthesystemthatmay
experiencechlorineresidualloss,excessive
formationofDBPs,andtheformationofbiofilms.
Chlorineresidualloss,inconjunctionwithbiofilm
1-7
Figure 1-4. Evolution of Federal Drinking
Water Regulations - Timeline.
A Reference Guide for Utilities
Meetingandbalancingtherequirementsofthe
variousregulationscanprovideasignificantchal-
lengetowaterutilities.Insomecases,regulations
provideguidanceorrequirementsthatcouldresultin
contradictoryactions. Forexample,theSWTR
requirestheuseofchlorineorsomeotherdisinfectant.
However,chlorineorotherdisinfectantsinteractwith
NOMintreatedwatertoformDBPs.Similarly,raising
thepHoftreatedwaterwillassistincontrolling
corrosionbutmayincreasetheformationofTTHMs.
Variousanalyticaltools,suchaswaterqualitymodels,
canprovidetheutilitywithinformationandan
understanding that helps the utility in balancing the
contradictoryrequirementsofsomeregulations.
formation,mayresultinthesporadicoccurrenceof
coliforms(indicatororganismsassociatedwith
bacteriologicallypollutedwater). Modelscanbe
usedtodefinemixingzoneswhereblendingwater
fromtwoormoresourcesresultsinwaterquality
problems. Specifically,waterqualitymodeling
toolsmayassistutilitiesincomplyingwiththe
TCR,SWTR,IESWTR,LT1ESWTR,andLCR.
Modelingcanassistinidentifyingpartsofthe
systemwithhighTTHMformationpotential
(DBPR1)andmeetingtheInitialDistribution
SystemEvaluation(IDSE)requirementsofthe
DBPR2(seetheIDSECaseStudyinChapter7). In
addition, modeling techniques can assist in
tracking contamination from cross-connections and
other accidental or deliberate contamination events
suchasawaterborneoutbreak.
1.4 Assessment and Management
of Water Quality
Waterutilitiestreatnearly34billiongallonsofwater
everyday(EPA,1999). Generally,surfacewater
systemsrequiremoretreatmentthangroundwater
1-8
A Reference Guide for Utilities
Table 1-1. Selected Rules and Regulations Dealing with Distribution Systems (Not Inclusive)
Regulation KeyDistributionSystemRequirements
SDWA Gives EPA the authority to establish national primary and secondary drinking water regulations
(MCLs and MCLGs).
NIPDWR The NIPDWR which was adopted at the passage of the SDWA required that representative
coliform samples be collected throughout the distribution system.
TTHM Established a standard for TTHMs as 0.1 mg/L.
86SDWAA Established the MCLG concept.
TCR Regulates coliform bacteria which are used as surrogate organisms to indicate whether or not
treatment is effective and system contamination is occurring.
SWTR Requires using chlorine or some other disinfectant.
LCR Monitoring for compliance with the LCR is based entirely on samples taken at the consumers tap.
ICR Provided data to support the interim and long-term enhanced SWTR, and Stage 2 DBP rule.
96SDWAA Has many provisions dealing with distribution systems, including the role that surface water
quality can play in influencing the quality of distributed water.
IESWTR Provisions to enhance protection from pathogens, including Cryptosporidium, and intended to
prevent increases in microbial risk while large systems comply with the DBPR1.
DBPR1 This standard applies to all Has lowered the standard for TTHMs from 0.1 mg/L to 0.08 mg/L.
community water supplies in the U. S. and requires monitoring and compliance at selected points
in the distribution system.
LT1ESWTR Provisions to enhance protection from pathogens, including Cryptosporidium, and prevent
increases in microbial risk for systems serving less than 10,000 people while they comply with
the DBPR1.
systemsbecausetheyaredirectlyexposedtothe
atmosphere,runofffromrainandmeltingsnow,and
otherindustrialsourcesofcontamination. Water
utilitiesuseavarietyoftreatmentprocessestoremove
contaminantsfromdrinkingwaterpriortodistribu-
tion. Theselectedtreatmentcombinationisbasedon
thecontaminantsfoundinthesourcewaterofthat
particularsystem. Thegeneraltechniquesinclude:
Coagulation/Flocculation: This process
removesdirtandotherparticlessuspendedin
thewater.Inthisprocess,alum,ironsalts,and/
orsyntheticorganicpolymersareaddedtothe
watertoformstickyparticlescalledfloc,
whichattractthesuspendedparticles.
Sedimentation:Inthisprocess,theflocculated
particlesaregravity-settledandremovedfrom
thewater.
Filtration:Manywatertreatmentfacilitiesuse
filtrationtoremovethesmallerparticlesfrom
thewater.Theseparticlesinclude:claysand
silts,naturalorganicmatter,precipitatesfrom
othertreatmentprocessesinthefacility,iron
andmanganese,andmicroorganisms. Filtration
clarifiesthewaterandenhancesthe
effectivenessofdisinfection.
Disinfection: Waterisdisinfectedatthewater
treatmentplant(orattheentrytothe
distributionsystem)toensurethatmicrobial
contaminantsareinactivated. Secondary
disinfectionispracticedinordertomaintaina
residualinthedistributionsystem.
Oncethetreatedwaterentersthedistributionsystem,
anumberofprocessesmayoccurthatcanadversely
impactthewaterqualitydeliveredtoconsumers. As
thewaterentersanetworkofburiedpipes,valves,
joints,meters,andservicelines,itissubjectto
disruptionssuchaswaterhammer(transientpressure
shockwave),aging(atdeadendsandlargetanks),
corrosion,cross-connections,leachingoftoxic
chemicals,intrusionofpathogens,andpipeline
breaks.Someoftheseeventsmayberegularoccur-
rences,suchaswateraging,lossofchlorineresidual
indeadends,ordepositionofsedimentationin
stagnantareas. Othersmayberareorunusualevents.
Anyoftheseeventscancausethewaterqualityto
deteriorateandposeapotentialpublichealthrisk.
Someroutinedistributionsystemdesignchangesand
maintenanceoroperationalproceduresthatcanhelp
topreventorreducetheeffectsofsucheventsinclude
the following:
TankMixing:Inadequatemixinginatankcan
leadtostagnantareascontainingolderwater
1-9
2-2
P V
2 2
P V
h
L
+ h
M
(Equation2-2) Z
1
+
1
+
1
+ h Z = +
2
+
2
+
P 2
2g 2g
waterqualitymodeling.ThoughtheEPSsolution
doesintroducesomeapproximationsandignoresthe
transientphenomenaresultingfromsuddenchanges
(e.g.,apumpbeingturnedon),thesemorerefined
assumptionsaregenerallynotconsideredsignificant
formostdistributionsystemstudies.
ConservationofEnergy:Theconservationofenergy
principlerequiresthatthedifferenceinenergy
betweentwopointsinanetworkmustbethesame
regardlessofflowpath. Forhydraulicanalysis,this
principlecanberepresentedintermsofheadas
follows:
where
Z
1and2
=elevationatpoints1and2,respectively,inft(m)
2 2
P
1and2
=pressureatpoints1and2,respectively,inlb/ft (N/m )
3 3
=fluid(water)specificweight,inlb/ft (N/m )
V
1and2
=velocityatpoints1and2,respectively,inft/s(m/s)
2 2
g=accelerationduetogravity,inft/sec (m/sec )
h
P
=pumpingheadgain,inft(m)
h
L
=headlossinpipes,inft(m)
h
M
= headlossdue to minor losses,in ft(m)
Pipe-frictionheadloss:Theequationmostcommonly
usedinmodelingsoftwareforcomputationofpipe-
frictionheadlossistheHazen-Williamsequation
representedasfollows:
2.1.3BasicHydraulicModelInput
Characterization
Buildinganetworkmodel,particularlyifalarge
numberofpipesareinvolved,isacomplexprocess.
Thefollowingcategoriesofinformationareneededto
constructahydraulicmodel:
Characteristicsofthepipenetworkcomponents
(pipes,pumps,tanks,valves).
A Reference Guide for Utilities
Wateruse(demands)assignedtonodes
(temporalvariationsrequiredinEPS).
Topographic information (elevations assigned
tonodes).
Controlinformationthatdescribeshowthe
systemisoperated(e.g.,modeofpump
operation).
Solutionparameters(e.g.,timesteps,tolerances
asrequiredbythesolutiontechniques).
Commonlyusedmethodsfortheseinputsarebriefly
describedinthefollowingsubsections.
2.1.3.1 PipeNetworkInputs
Constructionofthepipenetworkanditscharacteris-
ticsmaybedonemanuallyorthroughuseofexisting
spatialdatabasesstoredinGISorCADpackages.
Mostcommonly,GISorCADpackagesareusedin
thisprocessandaredescribedinmoredetailin
Chapter6. Theinitialstepinconstructinganetwork
modelistoidentifypipestobeincludedinthe
model. Nodesareusuallyplacedatpipejunctions,or
atmajorfacilities(tanks,pumps,controlvalves),or
wherepipecharacteristicschangeindiameter,C-
value(roughness),ormaterialofconstruction.Nodes
mayalsobeplacedatlocationsofknownpressureor
atsamplinglocationsoratlocationswherewateris
used(demandnodes). Therequiredpipenetwork
component information includes the following:
pipes(length,diameter,roughnessfactor),
pumps(pumpcurve),
valves(settings),and
tanks(crosssectioninformation,minimumand
maximumwaterlevels).
2.1.3.2 WaterDemandInputs
Waterconsumptionorwaterdemandisthedriving
forcebehindtheoperationofawaterdistribution
system. Anylocationatwhichwaterleavesthe
systemcanbecharacterizedasademandonthe
system. Thewaterdemandsareaggregatedand
assignedtonodes,whichrepresentsanobvious
simplificationofreal-worldsituationsinwhich
individualhousetapsaredistributedalongapipe
ratherthanatjunctionnodes. Itisimportanttobe
abletodeterminetheamountofwaterbeingused,
whereitisbeingused,andhowthisusagevarieswith
time(Walskietal.,2003).Demandmaybeestimated
byacountofstructuresofdifferenttypesusinga
representativeconsumptionperstructure,meter
readingsandtheassignmentofeachmetertoanode,
andtogenerallanduse.Auniversaladjustmentfactor
shouldbeusedtoaccountforlossesandother
unaccountedwaterusagesothattotalusageinthe
2-3
Futurewateruse:Fordesignandplanningpurposes,a
watersystemmustbeexaminedunderfutureconditions.
Insituationswhereasystemislargelycurrentlybuilt
out,futuredemandsmaybeestimatedbydeveloping
globalorregionalmultipliersthatareappliedtocurrent
demands.However,innewordevelopingareas,existing
waterusedoesnotprovideausefulbasisforestimating
futuredemands. Alternativeapproachesusepopula-
tion-based projections, socioeconomic modeling, and
land-usemethods(JohnsonandLoux,2004).
Inestimatingfuturedemandsforuseinanetwork
model,themostappropriatemethodisgenerallythe
land-usemethod. Theland-usemethodisbasedon
mappinglandusesandthenapplyingawater-usefactor
toeachland-usecategory.Whenappliedtoexisting
situationsorinhistoricalreconstructionofwater
systems,aerialphotographsaremostcommonlyusedas
thebasemapforidentifyingland-usecategories. For
developmentoffuturedemands,landusemapscanbe
obtainedfromplanners. Theland-usemethodologyis
depictedinFigure2-2.
Q C
ki kj@x=L
C
ij@x=0
=
k
(Equation2-6)
Q
kj
k
where
C
ij@x=0
= theconcentrationat thestartofthelink
connectingnodeitonodejinmg/L(i.e.,wherex=0)
C
kj@x=L
= theconcentrationat theendof a link,in mg/L
Q
kj
= flowfrom ktoi
A Reference Guide for Utilities
Modelingthemovementofacontaminantwithinthe
distributionsystemsasitmovesthroughthesystem
fromvariouspointsofentry(e.g.,wellsortreatment
plants)towaterusersisbasedonthreeprinciples:
Conservationofmasswithindifferentiallengths
ofpipe.
Completeandinstantaneousmixingofthewater
entering pipe junctions.
Appropriatekineticexpressionsforthegrowthor
decayofthesubstanceasitflowsthroughpipes
andstoragefacilities.
Thischangeinconcentrationcanbeexpressedbythe
following differential equation:
AccordingtoEquation2-5,therateatwhichthemassof
materialchangeswithinasmallsectionofpipeequals
thedifferenceinmassflowintoandoutofthesection
plustherateofreactionwithinthesection. Itisas-
sumedthatthevelocitiesinthelinksareknown
beforehandfromthesolutiontoahydraulicmodelof
thenetwork. InordertosolveEquation2-5,oneneeds
toknowC
ij
atx=0foralltimes(aboundarycondition)
andavaluefork
ij
.
Equation2-6representstheconcentrationofmaterial
leaving the junction and entering a pipe:
Equation2-6statesthattheconcentrationleavinga
junctionequalsthetotalmassofasubstanceflowing
intothejunctiondividedbythetotalflowintothe
junction.
sourcewillprovideamixedvolumecalculatedas50
percentfromthedesignatedsource.
2.2.3WaterQualityModelInputsand
Application
Inadditiontothebasichydraulicmodelinputs
describedinSection2.1.3,thewaterqualitymodels
requirethefollowingdataelementstosimulatethe
behaviorinadistributionsystem:
WaterQualityBoundaryConditions-Awater
qualitymodelrequiresthequalityofall
externalinflowstothenetworkandthewater
qualitythroughoutthenetworkbespecifiedat
thestartofthesimulation.Dataonexternal
inflowscanbeobtainedfromexistingsource
monitoringrecordswhensimulatingexisting
operationsorcouldbesettodesiredvaluesto
investigateoperationalchanges.Initialwater
qualityvaluescanbeestimatedbasedonfield
data.Alternatively,bestestimatescanbemade
forinitialconditions.Thenthemodelisrunfor
asufficientlylongperiodoftimeundera
repeatingpatternofsourceanddemandinputs
so that the initial conditions, especially in
storagetanks,donotinfluencethewaterquality
predictionsinthedistributionsystem.The
waterageandsourcetracingoptionsonly
requireinputfromthehydraulicmodel.
ReactionRateDataFornon-conservative
substances,informationisneededonhowthe
constituentsdecayorgrowovertime.
Modelingthefateofaresidualdisinfectantis
oneofthemostcommonapplicationsof
networkwaterqualitymodels.Thetwomost
frequently used disinfectants in distribution
systemsarechlorineandchloramines(areactant
ofchlorineandammonia). Freechlorineis
morereactivethanchloramineanditsreaction
kinetics have been studied more extensively.
Studieshaveshownthattherearetwoseparate
reactionmechanismsforchlorinedecay,one
involving reactions within the bulk fluid and
anotherinvolvingreactionswithmaterialonor
releasedfromthepipewall(Vasconcelosetal.,
1997). Bulkdecayistypicallyrepresentedasa
firstorderexponentialdecayfunctionwitha
singledecaycoefficientspecifiedtorepresent
thedecayovertime.Insomecircumstances,this
functiondoesnotadequatelyrepresentthe
observeddecaycharacteristics,andmore
complexformulationsmaybeusedtodescribe
thedecay. Wallreactionrepresentsthe
disinfectantdecayduetocontactwith
oxidizeablesubstancesatthepipewall,suchas
corrosionproductsorbiofilm.Themostwidely
usedapproachforrepresentingwalldemand
considerstwointeractingprocessestransport
2-10
Storagetanksareusuallymodeledascompletely
mixed,variablevolumereactorsinwhichthechanges
involumeandconcentrationovertimeareasfollows:
Manyalgorithmsandmethodsexistforthenumerical
solutionoffluidflowsdescribedbytheNavier-Stokes
equations.Thesealgorithmscanbeclassifiedas
EulerianorLagrangianandaseithertime-drivenor
event-driven.InaEulerianmethod,themovementof
thefluidisviewedfromastationarygridasthewater
movesthroughthesystem.Onthecontrary,ina
Lagrangianmethod,theanalysisisviewedfroma
frameworkthatismovingwiththeflow.Time-driven
methodsassessthesystematfixedtimesteps. Event-
drivenmethodsevaluatethesystemonlywhenthereis
adiscretechangeinwaterqualitysuchasapulseof
waterwithdifferentconcentrationsenteringorleaving
apipe. VariousmethodologiescombineeitherEulerian
orLagrangiansolutions(orhybridcombinationsof
thesetwocases)witheithertime-drivenorevent-driven
procedures.
ofthedisinfectantfromthebulkflowtothe
wallandinteractionwiththewall(Rossmanet
al.,1994). Recentstudieshavesuggestedthat
thisformulationmaynotadequatelyrepresent
theactualwalldemandprocessesandthat
furtherresearchisneeded(Clarketal.,2005;
Graymanetal.,2002;DiGianoandZhang,
2004). Therehasbeenlittlestudyonthenature
ofthewallreactioninchloraminatedsystems.
Alimitedamountofmodelingofthegrowthof
DBPs(mostnotablyTHMs)hasbeenperformed
assuminganexponentialgrowthapproachinga
maximumvaluecorrespondingtotheTHM
formationpotential. Boththeformation
potentialandthegrowthrateconstantmustbe
specifiedinthistypeofmodel(Clarketal.,
A Reference Guide for Utilities
1996). Therehasbeenextensiveresearchon
biofilmformationindistributionsystemsand
thishasledtothedevelopmentofseveral
theoreticalmodelsofthisphenomenon(Powell
etal.,2004).However,thesemodelsare
generally quite complex involving many
parametersthataredifficulttodetermine,and
thusarenotreadyforinclusioninageneral
waterdistributionsystemmodel.
Thefollowingsectionprovidesanoverviewof
availablesoftwareforhydraulicandwaterquality
modeling.
Distributionsystemwaterqualitymodelsaregenerally
limitedtotrackingthedynamicsofasinglecomponent
(e.g.,chlorine,waterage)atatimewhentheselected
componentistransportedthroughoutthenetworkof
pipesandstoragetanks. Suchmodelsdonotconsider
interactionsbetweendifferentcomponentsinthe
flowingwaterorcomplexreactionsbetweencompo-
nentsthataretransportedwiththewaterandsurface
componentsthatarefixedtothepipewall.Thiscanbea
significantlimitationwhenmodelingreactivecompo-
nents,forexamplewhenchlorineresidualismodeledfor
acasewheretherearemultiplesourceswithsignificant
differencesinwaterqualitycharacteristics. Another
morecomplexexamplethatisnotadequatelyrepre-
sentedbythesingle-speciesmodelismodelingofDBP
formation. Asolutiontothisdeficiencyisageneral-
purpose,multi-speciescapabilitythatisbeingaddedto
EPANET(Uberetal.,2004).Thisadditionwillallow
userstoprogramtheirownchemical/physical/biological
reactionsinEPANETwithalmostunlimitedinteraction
capability between multiple species.
2.3 Hydraulic and Water Quality
Modeling Software
Avarietyofsoftwarepackagesareavailableto
performhydraulicandwaterqualitymodeling.A
majorityofthesepackagesutilizetheEPANET
formulationasthebasiccomputationengine. Afull
discussionofindividualsoftwarepackagesisbeyond
thescopeofthisdocument. Thefollowingsubsec-
tionsbrieflydescribetheEPANETmodeland
summarizethefeaturesofotheravailablesoftware.
2.3.1EPANETSoftware
EPANETwasinitiallydevelopedin1993asa
distributionsystemhydraulic-waterqualitymodelto
supportresearcheffortsatEPA(Rossmanetal.,1994).
ThedevelopmentoftheEPANETsoftwarehasalso
satisfiedtheneedforacomprehensivepublic-sector
modelandhasservedasthehydraulicandwater
qualityengineformanycommercialmodels.
2-11
A Reference Guide for Utilities
EPANETcanbeusedforbothsteady-stateandEPS
hydraulicsimulations. Inaddition,itisdesignedto
bearesearchtoolformodelingthemovementandfate
ofdrinkingwaterconstituentswithindistribution
systems. EPANETcanbeoperatedintheSI(metric)
orBritishsystemsofmeasurement.
ThewaterqualityroutinesinEPANETcanbeusedto
modelconcentrationsofreactiveandconservative
substances,changesinageofwaterandtraveltimeto
anode,andthepercentageofwaterreachinganynode
fromanyothernode.OutputsfromEPANETinclude:
color-codednetworkmaps,
timeseriesplots,and
tabularreports.
ExampleoutputsfromEPANETwerepreviously
presentedinFigures2-3,2-4,and2-5.
2.3.2CommercialHydraulic-WaterQuality
ModelingSoftware
InadditiontoEPANET,thereareseveralcommercial
softwarepackagesthatarewidelyusedintheU.S.and
internationally.Mostofthesepackagesarebasedon
theEPANETformulationandincludevalue-added
componentsaspartsofGUIthatincreasethecapabil-
ityofthesoftware. Examplesofsuchvalue-added
componentsthatarepartofoneormoreofthe
commerciallyavailablesoftwarepackagesinclude:
Scenariomanager:Manageinputsandoutputs
ofagroupofmodelruns.
Calibration optimization: Utilize genetic
algorithm optimization technique to determine
modelparametersthatbestfitasetoffielddata.
Design optimization: Utilize genetic algorithm
optimizationtechniquestoselectpipesizesthat
minimizecostsorotherselectedobjectives.
IntegrationwithGISorCAD:Waterdistribution
modeldirectlyintegrateswithGISorCADto
assistinconstructingorupdatingmodeland
InadditiontothestandarduseofEPANETina
Windowsenvironmentusingthegraphicaluser
interface(GUI),thefunctionalityofEPANETcanbe
accessedthroughtheEPANETtoolkit.Thetoolkitisa
seriesofopensourceroutinesavailableinbothVisual
BasicandC(programminglanguage)thatcanbeused
asisormodifiedandaccessedfromausersown
computerprogram. Thispowerfulcapabilityhasbeen
widelyusedthroughouttheworldtosupportboth
researchandspecificapplicationsinthefieldofwater
distribution system analysis.
displayresults.
Flexible output graphics: Provides convenient
waystomodifyparametersforgraphical
displaysofoutputdata.
Energymanagement:Calculatesenergyusefor
aselectedalternative.
Automatedfire-flowanalysis:Assessesthe
availabilityoffireflowatarangeofnodesand
determineswhetherasystemmeetsfire-flow
requirements.
Watersecurityandvulnerabilityassessment
methods, skeletonization, and demand
allocation tools.
Table2-1providesasummarylistingofmajor
commercialsoftwareandaWeblinkwhereadditional
detailsmaybeobtainedonspecificfeaturesand
current version availability/pricing.
2.4 Additional Modeling Tools
Inadditiontostandardhydraulicandwaterquality
modelingofdistributionsystems,thereareseveral
otherrelatedtypesofmodelsthatcanbeusedto
assesshydraulicandwaterqualitybehaviorin
distributionsystems.Theseinclude:storagemodeling
tools,transient(waterhammer)modelingtools,
optimizationtools,andprobabilisticmodels. Eachof
thesetypesofmodelsarebrieflydescribedand
demonstratedinthefollowingsections.
2.4.1StorageModelingTools
Animportantaspectofwaterqualityandcontaminant
propagationindrinkingwaterdistributionsystemsis
theeffectofsystemstorage. Mostutilitiesusesome
typeofgroundorelevatedstoragesystemtoprocess
waterduringtimeperiodswhentreatmentfacilities
wouldotherwisebeidle. Itisthenpossibleto
distributeandstorewateratoneormorelocationsin
theserviceareaclosesttotheuser.
Theprincipaladvantageofdistributionstorageisthat
itequalizesdemandsonsupplysources,production
works,andtransmissionanddistributionmains.Asa
result,thesizesorcapacitiesoftheseelementsmaybe
minimizedandpeakpowertariffperiodsforpumping
canoftenbeavoided.Additionally,systemflowsand
pressuresareimprovedandstabilizedtobetterserve
thecustomersthroughouttheservicearea.Finally,
reservesuppliesareprovidedinthedistribution
systemforemergencies,suchasfirefightingand
poweroutages.
Inmostmunicipalwatersystems,lessthan25percent
ofthevolumeofthestorageintanksisactivelyused
(onadailybasis)underroutineconditions.Asthe
2-12
AReferenceGuideforUtilities
truck,costswillbehigherifthesetypesofitemsare
notreadilyavailable. Ifthestudyteamelectsto
analyzesamplesin-houseratherthanusinganoutside
laboratory,theteamshouldbalancethecostoflabor,
andthecostofadditionalreagentsandchemicals
againstthecostofperformingtheanalysesatan
outsidecommerciallaboratory.Laborcostsmaybe
evenmorevariablethanequipmentandmaterialcosts
andareafunctionofthesizeandcomplexityofthe
study.Inordertoprovideaneasybasisforcompari-
son,thelaborcostsarepresentedinlaborhours(Table
3-3)andincludeacombinationofengineersand
technicians. Laborhourshavebeenestimatedforlow,
medium,andhigh-endstudies. Theseestimatesare
obtainedfromactualfieldstudies,asdescribedbelow.
Thisapproachshouldallowutilitiestomakesite-
specificcostestimates.
Table 3-3. Representative Labor Hours
for a Range of Studies
Activity Low-End Medium High-End
Planning 27 274 480
Setup - 150 520
Fieldstudy 51 604 370
Laboratory
analysis
8 160 120
Post-study
assessment
24 212 740
Total 110 1,400 2,230
Atypicalexampleofalow-endtracerstudyis
providedbytheSweetwaterAuthoritydistribution
systeminSouthernCalifornia(seesecondsidebarin
Section3.1.4.5,page3-6). TheSweetwatersystem
coversaserviceareaof28squaremiles. Theutility
wasabletotakeadvantageofanaturallyoccurring
tracerandusedgrabsamplestakenat28existing
dedicatedsamplingsitesoveraperiodof5days. A
studyperformedinthe21-square-mileCheshire
serviceareaoftheSouthCentralConnecticut
RegionalWaterAuthorityin1989(seesecondsidebar
inSection3.1.4.1onpage3-4)providesanexample
ofamedium-leveltracerstudy.Inthiscase,the
normalfluoridefeedwasshutoffforaperiodof7
days(andthenturnedbackon)andgrabsampleswere
takenatintervalsofafewhoursat23sitesovera
periodof14days. Anexampleofahigh-endstudyis
providedbyatwo-phasedfieldinvestigationcon-
ductedintwosuburbanareasofGCWW.Thefirstarea
isasmall(<1squaremile)dead-endsystem,andthe
secondarea,a12-square-milepressurezone. A
calciumchloridetracerwasinjectedandmonitored
using a combination of automated conductivity
metersandgrabsamples. Inthesmallerarea,20
meterswereusedandmonitoringwasconductedover
a24-hourperiod. Inthesecondarea,33meterswere
usedandtwoseparatetracerinjectionswerecon-
ductedoveraperiodof5days. Includingboth
studies,atotalof725grabsamplesweretakenand
analyzedforconductivity,chloride,andcalcium.
Flowwasmonitoredatfourlocationsusingultrasonic
flowmeters.
Table3-3presentsestimatedlaborhoursforthese
typesofstudies. Theyaredividedintotheplanning
phase(asdescribedinSection3.1);setup,fieldwork,
andlaboratoryanalysisthattogethermakeupthe
executionphase(seeSection3.2);andthepost-study
modeling,assessment,andreportphase. Asillus-
tratedinthistable,thereisasignificantvariationin
thelaborhoursrequiredtoconductatracerstudy.For
example,thelow-endlaborcostsresultedduetothe
followingstudycharacteristics:naturallyoccurring
tracerwasused,nonewequipmentwaspurchased,
existingroutinemonitoringsiteswereused,andonly
alimitedpost-studyassessmentwasmade. The
medium-sizedstudyincludedthefollowingcharacter-
istics: achemicalthatwasroutinelyadded(fluoride)
tothewaterdistributionsystemwasusedasthetracer
(byshuttingitoff),thestudyrequiredamuchlonger
periodtocomplete,andsinceitwasthefirstmajor
tracerstudyinthedistributionsystem,itrequired
significant planning. The high-end study included
thefollowingcharacteristics: itwasthefirstmajor
tracerstudyemployingwide-scaleuseofcontinuous
monitors; a non-naturally occurring, non-routinely
addedchemicalwasinjectedasatracer;andsignifi-
canttimewasrequiredforacquiringandinstallingthe
equipment. Forpurposesofthisstudy,averydetailed
post-studydataassessmentinvolvingprocessingof
tracerstudydata,pipenetworkmodelcalibrationand
reportpreparationrequiredsignificantlaborexpendi-
tures. Examplesofmodelcalibrationeffortsassoci-
atedwithtracerstudiesarepresentedinChapter4.
3.4Summary,Conclusionsand
Recommendations
Tracersandtracingtechniqueshavebeenusedfor
manyyearsinanumberofengineeringapplications
toestimatestreamvelocityandretentiontimein
waterandwatersupplyunitprocesses.Morerecently,
tracershavebeenusedforcalibratingdrinkingwater
distributionsystemhydraulicandwaterquality
models. Forthepurposesofthisdocument,itis
assumedthattracerstudiesareusedtocalibrateand
validatenetworkmodels. Thecalibratedandvali-
datednetworkmodelsarethenusedtoestimateother
parameterssuchaswaterageandtraveltimes.
However,thedatafromatracerstudycanbedirectly
usedtoestimatesomespecificparameterssuchas
waterage(DiGianoetal.,2005). Acomprehensive
3-15
AReferenceGuideforUtilities
summaryofpotentialusesandregulatoryapplications
fortracerstudiesisprovidedinthefirstsubsectionof
thischapter.Drinkingwatertracersmightinclude
chemicalsthatareinjectedintoawaterdistribution
pipe,thetemporaryshutoffofachemicaladditive
currentlybeingaddedtotreatedwater(suchas
fluoride),orsignificantchangesinconcentrationof
disinfectants,DBPs,ornaturalcompounds. Thetracer
methodology selected would significantly impact the
overallcostsofthestudy.Probably,themostexpen-
siveoptionwouldbetoinjectachemicaltracer,
monitoritusingleasedorpurchasedonlineinstru-
mentation,andconductthestudyusingcontractor
staff.Theleastexpensiveapproachwouldbetotake
advantageofanaturaltracer,monitortheprogressof
thetracerbygrabsampling,andconductthestudy
usingprimarilyin-housestaff. Onceatracerinjec-
tionmethodologyhasbeenselected,carefulplan-
ningandexecutionwillensurethesuccessofthestudy.
Whenplanningatracerstudy,ifthespecificsteps
outlinedinthischapterarefollowed,theyshould
greatlyincreasethepotentialforasuccessfulstudy.
Thesestepsinclude:establishingclearstudyobjec-
tives,formingastudyteam,definingthestudyarea
characteristics,carefullyselectinganappropriate
tracer,selectingtheproperfieldequipment,develop-
ingkeyplanningdocuments,andensuringthatthe
publicandaffectedagenciesarenotified. Applica-
tionofadistributionandwaterqualitymodelduring
theplanningstageishighlyrecommendedtosimulate
theapproximatebehaviorthatwillbeexpected
duringtheactualtracerevent.
Duringtheexecutionphaseofthestudy,thefollow-
ingissuesshouldbeaddressed: procurementof
equipmentandmaterials;setup,testinganddisinfec-
tionoftheprocuredequipment;availabilityof
analytical instrumentation and laboratory facilities;
and,finally,theinstallation,testing,andoperationof
fieldequipment. Duringtheexecutionphase,itis
importanttoreviewandunderstandhowtracer
dosagesandinjectiondurationaretobeimplemented.
Dryrunsarehighlyrecommendedasameansof
debuggingtheprocedurespriortoafullstudy.
Distributionsystemtracerstudieshavebeenconducted
forover15years,butrecenttechnologydevelopments
haveimprovedtheefficiencyofthesestudiesand
providepromiseforgreatlyexpandedapplicationsin
thefuture. Specificcomponentsthatwillfuelthis
expanded use include the following: continuous
monitorsthatcanbeeasilyadaptedforuseindistribu-
tionsystemsarebeingdevelopedandtested,inpartin
responsetowatersecurityconcerns;automatedmeter
reading(AMR)equipmentisbeinginstalledbymany
utilitiesandcouldprovidemoredetailedtemporaland
spatialconsumptiondataforhydraulicmodels;
advancedanalysissoftwareisevolvingthatwill
facilitatetheuseoflargeamountsofcontinuousdatain
calibratingdistributionsystemmodels;andwith
increasedavailabilityofthesetechnologies,costsare
expectedtodecreasesothatlargerutilitiescanaffordto
purchaseandroutinelyusetheequipment,andconsult-
ingengineerscanaffordablyoffertheseservicesto
smallerutilities.
Duringthefieldstudy,itisimportantthatthestudy
teambeabletoassesstheprogressofthetracer,inreal
time,asitpropagatesthroughthesystem. Concise
andconsistentcommunicationsbetweentracerstudy
teammembers,testcoordinator,andwaterutilitystaff,
iscriticalalalltimesduringthetest.
Inthefuture,itishighlylikelythatadvancescur-
rentlyonthehorizonwillresultinsignificant
increaseduseofbothonlinetracer(orwaterquality)
monitorsandflowmonitoringinstrumentation. As
theon-linetechnologybecomesmorewidelyusedin
drinkingwater,theuseofnetworkwaterquality
modelswillalsobemorewidelyaccepted. Online
monitoring in conjunction with water quality
modelingwillprovideanin-depthunderstandingof
themannerinwhichwaterqualitychangescanbe
monitoredinadrinkingwaterdistributionsystem.
Also,giventhecurrentclimateofconcernover
distributionwaterqualityfrombotharegulatoryand
securityviewpoint,itisreasonabletoassumethat
therewillbeincreasedinterestinapplyingthistypeof
technologyinthewaterindustry.
3-16
AReferenceGuideforUtilities
References
Boccelli,D.L.,F.Shang,J.G.Uber,A.Orcevic,D.
Moll,S.Hooper,M.Maslia,J.Sautner,B.Blount,and
F.Cardinali. TracerTestsforNetworkModel
Calibration.Proceedings,ASCE-EWRIAnnual
Conference. 2004.
Boccelli,D.,andJ.Uber.IncorporatingSpatial
CorrelationinaMarkovChainMonteCarloAp-
proachforNetworkModelCalibration.Proceedings,
ASCE-EWRIWorldWater&EnvironmentalRe-
sourcesCongress,Anchorage,AK. 2005.
Boulos,P.F.,W.M.Grayman,R.W.Bowcock,J.W.
Clapp,L.A.Rossman,R.M.Clark,R.A.Deininger,and
A.K.Dhingra. HydraulicMixingandFreeChlorine
ResidualsinReservoirs,Journal of AWWA,88(7):48-
59. 1996.
Clark,R.M.,W.M.Grayman,R.Males,andA.F.Hess.
Modeling Contaminant Propagation in Drink-
ingWaterDistributionSystems,Journal of Environ-
mental Engineering, ASCE,119(2):349-364. 1993.
Clark,R.M.,W.M.Grayman,J.A.Goodrich,R.A.
Deininger,andA.F.Hess.FieldTestingDistribution
ofWaterQualityModels,Journal of AWWA,
83(7):67-75. 1991.
Clark,R.M.,G.Smalley,J.A.Goodrich,R.Tull,L.A.
Rossman,J.T.Vasconcelos,andP.F.Boulos.Manag-
ingWaterQualityinDistributionSystems: Simulat-
ingTTHMandChlorineResidualPropagation,
Journal of Water Supply Research and Technology
Aqua,43(4):182-191. 1994.
Clark,R.M.,andJ.A.Coyle. MeasuringandModel-
ingVariationsinDistributionSystemWaterQuality,
Journal of AWWA,82(8):46-53. 1990.
DiGiano,F.A.,andG.Carter.TracerStudiesto
MeasureWaterResidenceTimeinaDistribution
SystemSuppliedbyTwoWaterTreatmentPlants.
Proceedings,AWWAAnnualConference.2001.
DiGiano,F.A.,W.Zhang,andA.Travaglia.Develop-
mentofthemeanresidencetimefromtracerstudiesin
distributionsystems.Journal of Water Supply:
Research and Technology - Aqua54:1-14.2005.
EPA.The Stage 2 DBPR Initial Distribution System
Evaluation Guidance Manual. 2003a.
EPA.Drinking Water Advisory: Consumer Accept-
ability Advice and Health Effects Analysis on Sodium.
Availableat:http://www.epa.gov/safewater/ccl/pdfs/
reg_determine1/support_cc1_sodium_dwreport.pdf.
2003b.
Ferguson,B.A.andF.A.DiGiano.Impactoftempo-
raryswitchesfrommonochloraminetofreechlorine
onwaterqualityindistributionsystems.Proceed-
ings,AWWAAnnualConference.2005.
Grayman,W.M.,R.A.Deininger,A.Green,P.F.Boulos,
R.W.Bowcock,andC.C.Godwin.WaterQualityand
MixingModelsforTanksandReservoirs,Journal of
AWWA,88(7):60-73. 1996.
Grayman,W.M.,L.A.Rossman,C.Arnold,R.A.
Deininger,C.Smith,J.F.Smith,andR.Schnipke.
Water Quality Modeling of Distribution System
Storage Facilities.AwwaRFandAWWA.Denver,CO.
2000.
Grayman,W.M.UseofTracerStudiesandWater
QualityModelstoCalibrateaNetworkHydraulic
Model,Current Methods,1(1):38-42,HaestadPress,
Waterbury,CT.2001.
Grayman,W.M.,L.A.Rossman,R.A.Deininger,C.D.
Smith,C.N.Arnold,andJ.F.Smith.Mixingand
AgingofWaterinDistributionSystemStorage
Facilities,Journal of AWWA,96(9):70-80. 2004.
Hatcher,M.D.,W.M.Grayman,C.D.Smith,andM.
Mann. MonitoringandModelingoftheSweetwater
AuthorityDistributionSystemtoAssessWater
Quality.Proceedings,AWWAAnnualConference.
2004.
Maslia,M.L.,J.B.Sautner,C.Valenzuela,W.M.
Grayman,M.M.Aral,andJ.W.Green,Jr.Useof
Continuous Recording Water-Quality Monitoring
EquipmentforConductingWater-DistributionSystem
TracerTests:TheGood,theBadandtheUgly.
Proceedings,ASCE-EWRIWorldWater&Environ-
mentalResourcesCongress,Anchorage,AK. 2005.
Panguluri,S.,R.Krishnan,L.Garner,C.Patterson,Y.
Lee,D.Hartman,W.Grayman,R.ClarkandH.Piao.
UsingContinuousMonitorsforConductingTracer
StudiesinWaterDistributionSystems.Proceedings,
ASCE-EWRIWorldWater&EnvironmentalRe-
sourcesCongress,Anchorage,AK. 2005.
Sautner,J.B.,M.L.Maslia,C.Valenzuela,W.M.
Grayman,M.M.Aral,andJ.W.Green,Jr.Field
TestingofWaterDistributionSystemsatU.S.Marine
CorpsBase,CampLejeune,NorthCarolina,in
SupportofanEpidemiologicalStudy.Proceedings,
ASCE-EWRIWorldWater&EnvironmentalRe-
sourcesCongress,Anchorage,AK. 2005.
3-17
AReferenceGuideforUtilities
AWWA.Standard Methods for the Examination of
Water and Wastewater, 20
th
Edition. Editedby:
LenoreS.Clrsceri,ArnoldE.Greenberg,andAndrew
D.Eaton,AmericanPubicHealthAssociation/AWWA/
WaterEnvironmentFederation,pp2-44to2-45.
1998.
Teefy,S.M.,andP.C.Singer.PerformanceTesting
andAnalysisofTracerTeststoDetermineCompliance
ofaDisinfectionSchemewiththeSWTR.Journal of
AWWA,82(12):88-98. 1990.
Teefy,S.M.Tracer Studies in Water Treatment
Facilities: A Protocol and Case Studies.AwwaRF
andAWWA,Denver,CO.1996.
Vasconcelos,J.J.,L.A.Rossman,W.M.Grayman,P.F.
Boulos,andR.M.Clark. KineticsofChlorine
Decay,Journal of AWWA,89(7):54-65. 1997.
Vasconcelos,J.,P.Boulos,W.Grayman,L.Kiene,O.
Wable,P.Biswas,A.Bahri,L.Rossman,R.Clark,and
J.Goodrich. Characterization and Modeling of
Chlorine Decay in Distribution Systems.AwwaRF.
1996.
3-18
AReferenceGuideforUtilities
Chapter4
CalibrationofDistributionSystemModels
Waterdistributionsystemmodelscanbeusedina
widevarietyofapplicationstosupportdesign,
planning,andanalysistasks. Sincethesetasksmay
result in engineering decisions involving significant
investments,itisimportantthatthemodelusedbean
acceptablerepresentationoftherealworldandthat
themodelerhaveconfidenceinthemodelpredic-
tions. Inordertodeterminewhetheramodelrepre-
sentstherealworld,itiscustomarytomeasure
varioussystemvalues(e.g.,pressure,flow,storage
tankwaterlevels,andchlorineresiduals)duringfield
studiesandthencomparethefieldresultstomodel
predictions. Ifthemodeladequatelypredictsthefield
measurementsunderarangeofconditionsforan
extendedperiodoftime,themodelisconsideredtobe
calibrated. Iftherearesignificantdiscrepancies
betweenthemeasuredandmodeleddata,further
calibrationisneeded. Therearenogeneralstandards
fordefiningwhatisadequateorwhatisasignificant
discrepancy.However,itisrecognizedthatthelevel
ofcalibrationrequiredwilldependontheuseofthe
model. Agreaterdegreeofcalibrationisrequiredfor
modelsthatareusedfordetailedanalysis,suchas
designandwaterqualitypredictions,thanformodels
usedformoregeneralplanningpurposes(e.g.,master
planning).
Allmodelsareapproximationsoftheactualsystems
thatarebeingrepresented. Inanetworkmodel,both
themathematicalequationsusedinthemodelandthe
specificmodelparametersareonlynumericalapproxi-
mations. Forexample,theHazen-Williamsequation
usedtodescribefrictionheadlossisanempirical
relationshipthatwasderivedbasedonlaboratory
experiments(WilliamsandHazen,1920).Further-
more,theroughnessparameter(C-factor)usedinthe
Hazen-Williamsequationthatmodelersassigntoeach
pipeisnotknownwithtotalcertaintybecauseitis
notfeasibletoexamineandtesteverypipeinthe
system.Thegoalincalibrationistoreduceuncer-
taintyinmodelparameterstoalevelsuchthatthe
accuracyofthemodeliscommensuratewiththetype
ofdecisionsthatwillbemadebasedonmodel
predictions.
Thetypesofmodelcalibrationassociatedwithwater
distributionsystemanalysiscanbecategorizedin
severalways. Thenomenclaturedependsuponthe
adjustedparametersandthetechniqueemployed. In
general,calibrationcanbecategorized(orreferenced)
asfollows:
Hydraulicandwaterqualitymodelcalibration.
Theconceptofcalibrationcanbecomparedtofine
tuninganoldfashionedtelevision(TV)set. Oneknob
ontheTVisusedfortuningthechannelwhileother
knobsareadjustedtoimprovecolor,sharpness,contrast,
andhue. However,incalibratinganetworkmodel,there
arefarmoreknobstoadjustasillustratedinFigure4-1.
Adjustment knobs
Field data
Initial model results
Model results after calibration
Figure 4-1. Conceptual Representation of Calibration.
Someoftheknobsmaybeusedtoadjustroughness
coefficientsforpipes,otherknobstoadjustdemands
assignedtonodes,whilestillotherknobsmaycontrol
valvepositions,pumpcurves,orotherparametersthat
arenotknownwithcompletecertainty.Calibratinga
modelisanarduoustaskbecausetherearemanyknobs
thatcanbeadjusted.Findingthecombinationof
parametersthatresultsinthebestagreementbetween
measuredandmodeledresultsisdifficult. Thisprocess
iscomplicatedbythefactthattheremaynotbeasingle
bestsetofparameters. ExtendingtheTVanalogy,the
knobsmaybeadjustedinordertogetthebestreception
foronechannel. However,whenthechannelischanged,
theknobsmayneedtobeadjustedtoimprovethe
receptionforthenewchannel.Similarly,withanetwork
model,asetofparametersmaygivethebestmatchfor
onesetofdatawhileotherparametersmaygivebetter
resultsforanothersetofdata.Therefore,itisrecom-
mendedthatamodelerfirstcalibratethemodelusing
oneormoresetsoffielddataandthenvalidateitwith
anindependentsetoffielddata.
4-1
AReferenceGuideforUtilities
Static(steadystate)ordynamic(extended
period simulation) calibration.
Manual or automated calibration.
Hydrauliccalibrationreferstotheprocessofadjust-
ingtheparametersthatcontrolthehydraulicbehavior
ofthemodel. Similarly,waterqualitycalibration
relatestotheprocessofadjustingparametersusedin
thewaterqualityportionofthemodel. Staticor
steady-statecalibrationrelatestocalibrationofa
modelthatdoesnotvaryovertime,orusingdatathat
iscollectedrepresentingasnapshotintime.Dynamic
orEPScalibrationusestime-varyingdatainthe
calibrationprocess. Manualcalibrationreliesupon
theusertoinvestigatetheeffectsofarangeof
possibleparametervalues. Automatedcalibration
employsoptimizationtechniquestofindthesetof
parametersthatresultsinthebestmatchbetween
measuredandmodeledresults.
Itshouldbenotedthatthespecificapplication
methodandavailabilityofsomeofthesetechniques
willvarydependinguponthesoftwareusedfor
modelingandtheavailablenetworkmodelinforma-
tion. Therefore,onlythegeneraltechniquesem-
ployedineachofthesetypesofcalibrationare
discussedinthefollowingsections. Then,some
examplecasestudiesarepresentedtoillustratetheir
use. Thefinalsectioninthischapterdiscussesfuture
trendsincalibrationandthepossibilityofgeneral
calibration standards.
4.1HydraulicandWaterQuality
ModelCalibration
Hydrauliccalibrationisessentialforanymodel
simulationtobemeaningful. Furthermore,the
distributionsystemwaterqualitymodelsworkin
concertwiththehydraulicmodelandutilizetheflow
and velocity information calculated by the hydraulic
model. Thus,ifthehydraulicmodelisnotproperly
calibratedandresultsininaccurateflowandvelocity
estimates,thewaterqualitymodelwillnotperform
correctly.Infact,waterqualitymodelingisvery
sensitivetotheunderlyinghydraulicmodel. Fre-
quently,ahydraulicmodelthathasbeencalibrated
sufficientlyforapplicationssuchasmasterplanning
mayrequireadditionalcalibrationbeforeitis
appropriateforuseinwaterqualitymodeling. The
followingsubsectionsdescribetheparametersand
techniquesemployedforhydraulicandwaterquality
model calibration.
4.1.1HydraulicModelCalibration
Hydraulicbehaviorreferstoflowconditionsinpipes,
valvesandpumps,andpressure/headlevelsat
junctionsandtanks. Parametersthataretypicallyset
andadjustedincludepiperoughnessfactors,minor
losses,demandsatnodes,thepositionofisolation
valves(closedoropen),controlvalvesettings,pump
curves,anddemandpatterns.Whenintiallyestablish-
ingandadjustingtheseparameters,careshouldbe
takentokeepthevaluesfortheparameterswithin
reasonablebounds. Forexample,iflocalexperience
showsthattheroughnessfactorfora20-yearold
ductileironpipetypicallyfallswithinarangefrom
100to130,avaluethatisnotwithinorclosetothat
rangeshouldnotbeusedjusttoimprovetheagree-
mentbetweenthemeasuredandmodeleddata. Useof
unreasonablevaluesmayleadtoabettermatchfor
onesetofdata,butwilltypicallynotprovidearobust
setofparametersthatwouldapplyinothersituations.
Propercalibrationrequiresthatadjustmentsbemade
tothecorrectparameters. Acommonmistakeoccurs
whenadjustmentsareincorrectlymadeinonesetof
parametersinordertomatchthefieldresultswhilethe
parametersthatareactuallyincorrectareleftun-
touched. Thisprocessisreferredtoascompensating
errorsandshouldobviouslybeavoided. Field
verificationofsuspectparameters(e.g.,openorclosed
valves)canreduceconfusioncreatedbycompensat-
ingerrors.
Anexampleofcompensatingerrorsisanadjustmentin
roughnessfactorsinordertocompensateforaclosed
isolationvalveinthesystemthatisrepresentedasopen,
orpartiallyopen,inthemodel. Inthiscase,unreason-
ablylowvaluesfortheHazen-Williamsroughness
coefficientsaretypicallyintroducedinordertoforcea
largeheadlossinthepipesthatareactuallyclosed.
Thoughthismayresultinapproximatingthepressure
measurementsmadeinthefield,itwillintroduceother
errorsinflowandvelocitycalculations. Compensating
errorsmayalsoresultfromincorrectlyadjusting
demandsorotherparameters.
4.1.2WaterQualityModelCalibration
Subsequenttothepropercalibrationofahydraulic
model,additionalcalibrationofparametersinawater
qualitymodelmayberequired. Thefollowing
parametersareusedbywaterqualitymodelsthatmay
requiresomedegreeofcalibration:
InitialConditions:Definesthewaterquality
parameter(concentration)atalllocationsinthe
distributionsystematthestartofthe
simulation.
ReactionCoefficients:Describeshowwater
qualitymayvaryovertimeduetochemical,
biologicalorphysicalreactionsoccurringinthe
distribution system.
SourceQuality:Definesthewaterquality
characteristicsofthewatersourceoverthetime
periodbeingsimulated.
4-2
AReferenceGuideforUtilities
Table 4-1. Calibration/Input Requirements for Water Quality Models
Thedetailsofcalibrationdependupon
thetypeandapplicationofthewater
qualitymodel. Calibrationrequire-
mentsforeachtypeofmodelingare
describedbelowandsummarizedin
Table4-1.
Waterage:Noexplicitwater
quality calibration can be
performedbecausethereareno
reactioncoefficients. Estimates
ofinitialwaterageintanksand
reservoirsaredesirableinorder
toshortenthelengthofthe
simulation. Sourcewaterageisusuallysetto
zeroforallsources.Wateragecanbeespecially
sensitivetoinflow-outflowratesfortanks,
mixingcharacteristicsoftanks,andtraveltimes
indead-endpipes.
Whenmodelingatank,animportantparameteristhe
initialageofthewaterinthetankatthestartofthe
simulation. Thisvaluecannotbemeasuredinthefield
butcanbeestimatedbydividingthetankvolumeby
thevolumeofwaterthatisexchangedeachday.
Frequently,modelerswilljustassumethattheinitial
ageiszeroandrunthemodelforalongperioduntilit
hasreachedadynamicequilibrium.Thisoccurswhen
theinitialwaterinthetankhasbeenflushedout
entirelythroughthefillanddrawprocess.Thefollow-
ingfigure(Figure4-2)showstheeffectsoftheinitial
waterageonthemodeledresults.Asillustrated,a
goodinitialestimateforwaterage(120hoursinthis
case)resultsinamuchshortertimeperioduntilthe
dynamicequilibriumisreached.Infact,inthiscase
whentheinitialagewasinputaszerohours,themodel
didnotevencomeclosetoreachingdynamicequilib-
riumduringthesimulationperiodandwouldhave
requiredamuchlongerrundurationtoreachthesame
point.
0
20
40
60
80
100
120
140
0 10 20 30 40 50 60
Hours
W
a
t
e
r
a
g
e
(
h
o
u
r
s
)
Initial age 0 hours Initial age 120 hours
Figure 4-2. Effects of the Initial Water Age on the
Modeled Results.
ModelApplication
Initial
Conditions
Reaction
Coefficients
Source
Quality
Waterage YES NO UsuallyNO
Sourcetracing YES NO UsuallyNO
Conservativeconstituent YES NO YES
Reactiveconstituent YES YES YES
Sourcetracing: Noexplicitwaterquality
calibrationcanbeperformedbecausethereare
noreactioncoefficients. Estimatesofinitial
conditionsintanksforpercentageofwater
comingfromasourcearedesirableinorderto
shortenthelengthofthesimulation. Valuesfor
sourcesareusuallysettozeroforallsources
exceptforthespecificsourcebeingtraced.
Conservativeconstituents:Noexplicitwater
qualitycalibrationcanbeperformedbecause
therearenoreactioncoefficients. Estimatesof
initialconditionsintanksforconcentrationsof
the conservative constituents can usually be
determinedfromfielddataandaredesirablein
ordertoshortenthelengthofthesimulation.
Valuesforsourcesaresettothetypical
concentrationsfoundinthesource.
Reactive constituents: For reactive constituents,
boththeformofthereactionequationandthe
reactioncoefficientsmustbeprovided. When
modelingchlorineorchloraminedecay,the
mostcommonformulationisafirstorderdecay
equation including both bulk and wall decay
coefficients.Valuesforthesecoefficients
typicallyrequirelaboratoryandfieldanalysis
andcalibrationinordertomatchmodelresults
totheconcentrationsmeasuredinthefield.
Correspondingly,THMs,agroupofDBPs
formedwhenwaterischlorinatedor
chloraminated,generallyincreasein
concentrationwithtime(Vasconcelosetal.,
1996). Thisprocessisfrequentlyrepresentedas
afirstordergrowthfunctionthatasymptotically
approachesalimitingvaluerepresentativeof
maximumconcentrationreachedwhenallofthe
NOMhasreactedorallofthechlorinehasbeen
consumed. Boththelimitingvalueandtherate
ofgrowthmustbedeterminedinthiscase.
Waterqualitymodelingisverysensitivetothe
hydraulicrepresentationofthesystem.Toreiterate,
hydrauliccalibrationthatmaybesufficientforsome
hydraulicsimulationmayrequireadditionalcalibra-
tionwhenusedasabasisforwaterqualitymodeling.
4-3
AReferenceGuideforUtilities
4.2StaticCalibrationand
Dynamic Calibration
Justaswaterdistributionsystemmodelscanberun
inasteady-stateoranextendedperiodmode,
calibrationcanbeperformedineitherastaticmode
usingasteady-statemodelorinadynamicmode
usinganextendedperiodmodel. Acommon
approachistoperformastaticcalibrationfirst
followedbyEPS,toenhancethestaticcalibration
throughadynamiccalibration. Theoptionsand
proceduresforthesetwotypesofcalibrationare
describedbelow.
4.2.1Steady-StateCalibrationMethods
Thetwomostcommonapproachesusedincalibrat-
ingasteady-statehydraulicmodelareC-factortests
andfire-flowtests.Forwaterqualitymodelsof
chlorine/chloramines,atestprocedureforestimating
bulkandwalldemandmaybeemployed. Inallof
thesecases,fielddataiscollectedundercontrolled
conditionsandthenappliedtodeterminethemodel
parametersthatresultinthebestfitofthemodelto
thefielddata.
4.2.1.1 C-FactorTests
C-factortests(sometimescalledheadlosstests)are
performedtoestimatetheappropriateC-factorstobe
usedinahydraulicmodel. TheC-factorrepresents
theroughnessofthepipeinthewidelyusedHazen-
Williamsfrictionequation.Typically,suchtestsare
performedonasetofpipesthatarerepresentativeof
therangeofpipematerials,pipeage,andpipe
diametersfoundinthewatersystemthatisbeing
studied. Theresultsofthetestsarethenusedtoassign
C-factorsforotherpipesofsimilarcharacteristics.
Inafieldtest,ahomogeneoussectionofpipebetween
400and1,200feetlongisinitiallyisolated. Subse-
quently,flow,pipelength,andheadlossaremeasured
inthefield.Typically,nominalpipediametersare
TheunderlyingconceptforaC-factortestisthatall
factorsintheHazen-Williamsfrictionequationcanbe
measuredinthefieldandtheequationcanthenbe
solvedfortheunknownC-factor.Itcanalsobeused
toaccountforminorlossesthatoccurthroughdistri-
butionsystemcomponents(e.g.,valves,fittings). The
followingequationistheHazen-Williamsequation
(Equation2-3)arrangedtosolveforroughness.
C=8.71VD
-0.63
(H/L)
-0.54
(Equation4-1)
where
C=roughnessfactor
V=velocityinfeetpersecond
D=pipediameterininches
H=headlossinfeet
L=pipelengthinfeet
Flowed
Fire Hydrant #1
Length
X
Hydrant
Fire Hydrant #2
Flow
Pitot Gage
Closed Valve
Figure 4-3. Schematic of Standard Two-Gage C-
Factor Test Setup.
Differential
pressure gage Flowed
Pitot Gage
Fire Hydrant #1
Length
X
Hydrant
Fire Hydrant #2
Flow
Closed Valve
Small diameter hose
Figure 4-4. Schematic of Parallel Hose C-
Factor Test Setup.
takenfromsystemmapsandthesevaluesareused
alongwithflowratetocalculatevelocity.Thereare
twoalternativemethodsfordeterminingheadlossin
thepipesection: atwo-gagemethod(Figure4-3)and
aparallelhosemethod(Figure4-4). Withthetwo-
gagemethod,pressureisreadathydrantslocatedat
theupstreamanddownstreamendofthesectionand
usedalongwithelevationdifferencebetweentheends
tocalculateheadloss. Withtheparallelhosemethod,
asmall-diameterhoseisusedtoconnectthetwo
hydrantstoadifferentialpressuregagetodirectly
measurethedifferenceinpressure.Thetwoend
hydrantsshouldbespacedfarenoughapartandthere
shouldbesufficientflowsothatthereisapressure
dropofatleast15poundspersquareinch(psi)fora
two-gagetestora3-psipressuredropforaparallel
hosetest(McEnroeetal.,1989). Inbothcases,a
hydrantdownstreamofthetestsectionisopenedto
induceflowandasufficientpressuredrop. Multiple
downstreamhydrantsmayalsobeemployedtoinduce
agreaterflowandlargerpressuredrop.Typically,a
pitotgage(asshowninFigures4-3and4-4)is
attachedtotheflowinghydrantstomeasuretheflow
rate. Itisimportanttoensurethatallflowbetween
hydrantsisaccountedfor(i.e.,anyconnectionsthat
maybleedwaterintooroutofthetestsection). The
two-gagemethodisthemorecommonlyused
approach.Theparallelhosemethodrequiresmore
4-4
AReferenceGuideforUtilities
specializedequipment,butisinherentlymore
accurateandmaybeusedwhenalargepressuredrop
cannotbeachieved.Notethatthevalveisclosed
downstreamoftheflowinghydrant.
Asnotedabove,anassumptionismadethatthepipe
diameterhasnotdiminishedfromitsoriginalnominal
diameterduetotuberculationonthepipewalls. If
thatassumptionisnotvalid,thecalculatedC-factors
willbelowerthanexpected. IfverylowC-factorsare
calculatedbasedonafieldC-factortest,itisrecom-
mendedthatfurtheractionsbetakeninorderto
determinetheeffectivediameterofrepresentative
pipes.Theseactionscouldincludedirectinspection
ofsamplepipesoruseofcalipersinsertedintothe
pipetomeasuretheeffectivepipediameter.
4.2.1.2 Fire-FlowTests
Fire-flowtestsareroutinelyperformedbywater
utilitiestodeterminetheabilityofthesystemto
deliverlargeflowsneededtofightfires. Insucha
test,firehydrantsareopened,theflowthroughthe
hydrantsmeasuredandpressuresmeasuredatadjacent
hydrants(seeFigure4-5).Thehighdemandscaused
bytheopenhydrantsleadtohighflowsandincreased
headlossinpipesintheareaaroundthehydrants.
Undertheseconditions,thesystemisstressedandthe
capacityofthesystemtodelivertheseflowsisvery
sensitivetotheroughnessofthepipes.
Thesefire-flowtestscanalsobeveryeffectiveasa
calibrationmethodology.Inthiscase,inadditionto
thestandardinformationroutinelycollectedaspartof
afire-flowtest(flowsandpressures),informationis
collectedonthegeneralstateofthesystemsuchas
pumpandvalveoperation,tankwaterlevels,and
generalsystemdemand. Thedistributionsystem
modelisthenrununderthesystemconditions
observedduringthetestandadjustmentsmadein
roughnessfactors(orotherparameters)sothatthemodel
adequatelyrepresentsthedatameasuredinthefield.
P=42 psi
P=55 psi
Q=400 gpm
P=36 psi P=52 psi
Figure 4-5. Fire-Flow Test Setup.
Figure 4-6. A Hydrant Being Flowed with a Diffuser
as Part of a Fire-Flow Test.
Figure4-6illustratesanexamplesetupforafire-flow
test. Thediffuserattachedtothehydrantinthefigure
includesapitotgageusedtomeasuretheflow. The
cagediffusestheflowandpreventsanyobjectsinthe
streamfrombeingprojectedoutathighspeed.
InthecaseshowninFigure4-5,onlyasinglehydrant
isopened,withtheflowmeasuredatthathydrantand
pressuremeasurementsmadeatfourhydrants.
Additionalhydrantsmaybeflowedandmonitoredas
partofafire-flowtestforcalibration(seeCaseStudy
inSection7.7).
4.2.1.3 ChlorineDecayTests
Chlorinebulkreactionandwallreaction(ordemand)
testingprocedurescanbeusedtodeterminethe
reactionparametersusedinwaterqualitymodels.
Bottletestsmeasuretherateofchlorinereactionthat
occursinthebulkflowindependentofwalleffects.
Thisprocedureisperformedbyfirstmeasuringthe
chlorineatarepresentativelocationsuchasinthe
effluentfromawatertreatmentplant. Thenseveral
bottlesarefilledwiththesamewaterandkeptata
constanttemperature. Separatebottlesaresubse-
quentlyopenedatintervalsofseveralhours(ordays)
andthechlorinecontentismeasured. Theresulting
recordofchlorineatdifferenttimesisusedtoestimate
thebulkreactionrate.SeeAWWA(2004)foramore
completeprotocolforthistest.
Thepurposeofthechlorinedecayfieldtesting
procedureistoestimatethechlorinewalldemand
coefficientforrepresentativepipesinthedistribution
system. Themethoddescribedhereinvolvesthe
measurementofchlorineconcentrationsinapipe
segmentundercontrolledflowconditionsanduseof
theresultingchlorinemeasurementstodeterminethe
wallreactionrateforthatpipesegment. Themethod
isdesignedtobecomplementarywithC-factortesting
sothatitcanbeconductedinconjunctionwithaC-
factortest.Themethodisconsideredtobeexperimental
4-5
AReferenceGuideforUtilities
andfeasibleonlyforpipesthatareexpectedtohave
relativelyhighwallreactionvalues,suchassmaller
diameterunlinedcastironpipes.Forthesmaller
diameterunlinedcastironpipes,pipesectionswitha
lengthintherangeof1,500to2,000feetwillbe
requiredtoestimatewalldemand. Forothertypesof
pipesthattypicallyhavelowwalldecayfactors(e.g.,
plasticandnewpipes),therequiredlengthofthepipe
maybesolongastomakethistestimpractical. Other
factorsthatshouldbeconsideredinselectingsites
include the following:
Abilitytomeasureflowinthepipe.
Abilitytovalveoffthepipesegments.
Presenceofareasonablechlorineresidual
(preferably>0.4mg/L)attheupstreamendof
thepipesegment.
Abilitytovaryflowinthepipeovera
reasonableflowrange(e.g.,fora6pipe,a
rangeofflowsof100to500gpmwouldbe
desirable).
Abilitytoestimatetheactualpipediameterfor
thepipesegment.
Fortheselectedpipesegment,majorlateral(s)and
downstreamsegmentsshouldbevalvedofftocontrol
flowinthepipe.Twoorthreesamplingpointsshould
beestablishedalongthesegmentofinterest(up-
stream,downstream,andanoptionalmidpoint).
Typically,thesewouldbetapsonfirehydrants.Prior
tothetesting,thetapsshouldberunforseveral
minutestocleanouttheline. Theapproximatetravel
timethroughthepipeshouldbecalculatedand
chlorinemeasurementstakenfromupstreamto
downstreamsothatapproximatelythesameparcelof
waterissampledateachstation. Flowmeasurements
canbemadeatanylocationwithinthesegment.
Thetestshouldberepeatedforthreeflowvalues:a
lowflowrate,amediumflowrate,andahighflow
rate. Duringeachflowtest,chlorineresidualshould
bemeasuredateachofthetwoorthreesampling
points. Sincerelativelysmallvariationsinchlorine
concentrationareexpected,agoodqualityfield
chlorinemetershouldbeemployedandthreerepli-
catesshouldbetakenateachsamplingpointforeach
flowtest. Followingthefieldanalysis,aspreadsheet
canbeusedtobackcalculatetheresultingwall
reactioncoefficients,orawaterdistributionmodel
canbeusedtodeterminethewallreactioncoefficient
throughtrialanderror.
4.2.2DynamicCalibrationMethods
Dynamiccalibrationmethodsareassociatedwiththe
useofanEPSmodel. Thedynamiccalibration
methodsinclude: (1)comparisonofmodeledresults
Ifmeasuredandmodeledrecordsoftankwaterlevels
donotagreewell,therelationshipbetweenthetwo
tracescanprovidecluesastothepotentialproblems.
Intheexampledepictedbelow,thetimingofthefill
anddrawcyclesinthemeasuredandmodeledresults
arequiteclosebutthemodeledandmeasureddepthof
thefillcyclesvarysignificantly.Thissuggeststhatthe
systemdemandsmaybeinerror,resultinginan
incorrectamountofflowenteringthetank.
Inthesecondexampleillustratedbelow,themagnitude
ofthechangeinwaterlevelisquitecloseinthe
modeledandmeasuredresults,butthetimingofthefill
anddrawcyclesdiffer.Thisistypicallycausedby
errorsinthepumpingcontrolsinthemodel,resulting
inpumpsbeingturnedonandoffatthewrongtime.
tomeasurementsmadeinthefieldovertime,and(2)
tracerstudies. Inbothcases,modelparametersare
adjustedsothatthemodeladequatelyreproducesthe
observedbehaviorinthefield. Tracerstudiesare
discussedindetailinChapter3.
Comparisonofmodeledandmeasureddatacanbe
usedforcalibrationofbothhydraulicandwater
qualitymodels. Themostcommonlymeasured
hydraulicdataaretankwaterlevels,flows,and
pressures. Frequently,thisinformationisroutinely
reportedthroughSCADAsystemstoadatabaseand
canbeextracted. Inothercases,continuousflow
metersorpressuregagesmustbeinstalledtocollect
dataduringatestperiod. Generally,tankwaterlevel
dataandflowmeasurementsarethemostusefulform
ofdataforcalibratinganextendedperiodmodel.
Underaveragewateruseconditions,temporal
variationsinpressuremeasurementstypicallyvary
overarelativelysmallrangeandthenonlyin
responsetovariationsintankwaterlevels.Asaresult,
theyarelessusefulincalibratingmodelparameters.
Ifpressuremeasurementsaregoingtobeusedfor
4-6
AReferenceGuideforUtilities
dynamiccalibration,thesystemmustbestressedby
conductingfire-flowtestsduringthetestingperiod.
Theprimarymodelparametersthatareadjusted
duringdynamiccalibrationare:demandpatterns,
pumpschedulesandpumpcurves,controlvalve
settings,andtheposition(openorclosed)ofisolation
valves.
Dynamiccalibrationproceduresusingtracerstudy
dataisdiscussedviaacasestudyinSection4.4of
thischapter.Dynamiccalibrationcanalsobeusedfor
calibratingwaterqualityparameters,suchasthewall
demandcoefficientforcomputingchlorineresiduals.
Generally,waterqualityfieldstudiesareperformedin
conjunctionwithfieldhydraulicstudiesorwitha
tracerstudy.Forchlorinemodels,measurementsof
chlorinearetakenatfrequentintervalsinthefieldat
representativesites. Thesemayincludededicated
samplingtaps,hydrants,tankinlet/outlets,orother
accessiblesites. Continuouschlorinemetersmayalso
beused. Duringthemodelcalibrationprocess,the
modelisfirstcalibratedforhydraulicparameters,and
waterqualitycoefficientsaresubsequentlyadjusted
sothatthemodelresultsmatchthefielddata.
4.3ManualCalibrationand
Automated Calibration
Theaforementionedprocessofadjustingmodel
parameterssothatthemodelreproducesthehydraulic
and/orwaterqualityresultsmeasuredinthefieldcan
involveasignificantamountofeffortinlargeor
complexsystems.Asdiscussedearlierinthischapter,
therearemanyparametersthatcanbeadjustedinthe
modelandthecombinationsofpossibleparameter
valuescansometimesappeartobequiteoverwhelm-
ing. Typically,amanualtrialanderrorapproachis
used.Themostinfluentialparameterscanbeidenti-
fiedbasedonsensitivityanalysisandthenadjustedto
seeiftheyimprovetheresults. Thisprocessis
continueduntilanacceptablelevelofcalibrationis
achieved or until budgetary constraints dictate
closure. Itisnotunusualformany(dozensoreven
hundreds)separatemodelrunstobemadeinthis
process.
Anextensiontothemanualcalibrationprocessisan
automatedapproachthatallowsthecomputerto
searchthroughdifferentcombinationsofmodel
parameters(witharealmofrealisticvalues)andto
selectthesetofparametersthatresultsinthebest
matchbetweenmeasuredandmodeledresults. The
developmentofthistypeofprogramhasbeenthe
topicofmanystudiesoverthepast25years(Walski
etal.,2003).
Automatedmethodsrequireaformaldefinitionofan
objectivefunctionformeasuringhowgoodaparticu-
larsolutionis.Generally,thevalueofasolutionis
measuredbyastatisticthatreflectsthedeviation
betweenmeasuredandmodeledresultsinflowand
pressure. Acommonlyusedobjectivefunctionis
minimizationofthesquarerootoftheweighted
summationofthesquaresofthedifferencesbetween
observedandpredictedvalues. Theweightingisused
toestablisharelationshipbetweentheerrorsassoci-
atedwithflowandpressure. Forexample,theuser
maychoosea1-psierrorinpressurepredictiontobe
equivalentinvaluetoa10-gpmerrorinflow.
Inmostautomatedmethods,theuseralsogroups
pipesbycommoncharacteristics,suchasage,
material,andnodes,intocommondemandcharacter-
isticssuchasresidentialorcommercial.Theuserthen
specifiesarangeofallowablevaluesforpiperough-
nessfactorsorarangeofmultipliersappliedtothe
existingroughnessfactors.Similarly,arangeof
allowabledemandmultipliersisalsospecified,asare
potentialpipeswhereanexistingisolationvalvemay
beclosed. Theoptimizationroutineisthenapplied
andtheroughness,demands,andisolationvalve
positionsareselectedthatresultintheminimumerror.
Thoughmanualcalibrationstillremainsthepredomi-
nantmethodology,automatedcalibrationmethodsare
becomingmoreavailableincommercialmodeling
packages. Itislikelythatastheautomatedcalibra-
tionmethodsarerefined,thetechnologywillexpand
forroutineusewithEPShydraulicandwaterquality
models.
4.4CaseStudies
Inordertoillustratesomeofthecalibrationmethods
describedearlierinthischapter,twocasestudiesare
presentedinthissubsection. Thetwocasestudiesare
similaringeneralmethodologybutdifferinthe
overallscaleandspecificsofthestudyarea. Inboth
cases,thedistributionsystemmodelthatwasusedas
astartingpointforthecalibrationexercisewaspartof
askeletonizedmodelextractedfromunspecified
portionsoftheGCWWdistributionsystem.
Mostlargerurbanwatersystemsgenerallyhaveat
leastaskeletonizedmodeloftheirdistribution
system. Itshouldbenotedthat(asdiscussedin
Chapters2and3ofthisreport)askeletonizedmodel
denotesamodelthatincludesonlyamajorsubsetof
actualpipesratherthanallpipesinthedistribution
system. Theextractedsystemmodelwasmodified
andconvertedtoEPANETformatforuseinthis
project. Themodificationsincluded:additionofkey
pipes,updatestoconsumerdemanddata,andan
interconnectionbetweenthecasestudyareaandthe
fullsystembyafixedgradenode(reservoir). These
portionsofthebasemodelhadbeenpreviously
calibratedusingvariousdynamiccalibrationmethods
andwereusedforroutinewaterutilitywork. Forthe
4-7
AReferenceGuideforUtilities
purposesofcalibration,separatefieldstudieswere
conductedineachstudyarea.
Inbothfieldstudies,afood-gradeconservativetracer
(calciumchloride)wasintroducedintothesystemand
itsmovementthroughthesystemwasmonitoredby
bothgrabsamplingandcontinuousmonitoring(CM)
stations installed at key locations in the distribution
systems. TheCMstationswereinstalledathydrants
whichwereleftpartlyopenforthedurationofthe
studytominimizetraveltimebetweenthemainand
samplinglocation. Eachopenhydrantwasaddedasa
newdemandnodeintheEPANETnetworkmodel.
Additionally,severalultrasonicflowmeterswere
installedtoprovidecontinuousflowmeasurementsat
keylocations.Thegeneralprocedures,methodology,
andinstrumentationusedinthesefieldstudiesare
consistentwiththosepresentedinChapter3.
4.4.1Case1-Small-Suburban,Dead-End
System
Thissystemispartofalargerpressurezone. Itwas
selectedbecauseoftherelativelycompactsizeand
simplestructure,fedbyasinglesupplypipewithno
additionalstorage. Asaresult,themovementofthe
tracerwasrelativelyrapidthroughthesystemandit
couldbemonitoredwithcontinuousmetersplacedat
severallocations. Thegenerallayoutofthissub-
system,thelocationoftheinjectionsite,andthemon-
itoringlocationsforthisstudyareshowninFigure4-7.
Thecalciumchloridetracerwasinjectedastwo
pulses,atwo-hourpulsefollowedbya2.5-hour
periodofnoinjectionandthenfollowedbyahigher
concentrationpulseoftwohoursduration. The
injectionrateandtheresultingconcentrationofthe
tracerinthedistributionsystemjustdownstreamof
Compliancewithstateandfederalregulationsduringa
tracerstudyisobviouslyquiteimportant. Inorderto
ensurethatthetracerwillnotexceedallowablelevels,
itisnecessarytomonitorinformationsuchastherate
ofinjectionofthetracer,theflowinthereceivingpipe,
andtheresultingconcentrationinthereceivingpipe.
Frequently,asafetyfactorfortheinjectionrateis
includedtoaccountforuncertainty.Inthisfieldstudy,
thetracerinjectionratewasverylowandtheflow
meterontheinjectionpumpprovidedapproximate
values. Chlorideconcentrationsweremonitoredata
suitablelocationapproximately100feetdownstream
oftheinjectionpointwithatraveltimeofapproxi-
mately10minutes. Duetounexpectedvariationsin
flowthroughthepipe,delayinmeasurements,and
CM07
CM09
CM12
CM16
CM20
CM19
CM18
CM17
CM15
CM14
CM13
CM10
CM11
CM01
CM06
CM05
CM08
CM02 CM03 CM04
Flow gage
Conductivity meter (CM 1-20)
Injection point
Figure 4-7. Schematic Representation of
Small-Suburban Dead-End System.
theinjectionpointwerecarefullymonitoredtoensure
that the resulting chloride concentration did not
exceedthesecondarymaximumcontaminantlevel
(MCL)of250mg/Lforchloride.
Themovementsofthetracerpulsesweremonitored
by using both manual sampling and continuous
conductivity meters located throughout the distribu-
tionsystem. Additionally,fourultrasonicflowmeters
wereinstalledinthestudyareatoprovidecontinuous
flowmeasurementsatkeylocationswithinthe
distribution system.
Inpreparationforthecalibrationprocess,theconduc-
tivityreadingswereconvertedtochlorideconcentra-
tions using a relationship developed in the laboratory.
Figure4-8showstherelationshipbetweenconductiv-
ityandchlorideandthebest-fitlinearandpolyno-
mialrelationshipsbetweenthem. Thisconversionwas
necessarybecauseconductivityisnotatrulylinear
parameterand,asaresult,cannotbesimulated
exactly inawaterdistributionsystemmodel. The
converted continuous concentration readings were
thencomparedtothemanuallycollecteddatafor
qualitycontrolpurposes. Figure4-9showsthe
resultingchloridedatasetthatwasusedatone
locationasabasisforevaluatingmodelpredictionsas
partofthecalibrationprocess.
Thepreliminaryresultsindicatedsomediscrepancy
betweentheEPANET-modelpredictedvaluesandthe
relatedcomputations(associatedwithtracertravel
time),chloridevaluesexceedingthetargetlevelwere
experiencedforabriefperiodbeforetheinjectionrate
Figure 4-8. Empirical Relationship Between
wasadjusted.
Chloride and Conductivity.
4-8
Level1
Level2
Level3
Level4
Figure 4-10. Comparison of Model Versus Field Results
for Continuous Monitor Location CM-18 at Various
Calibration Stages.
AReferenceGuideforUtilities
ofthejunction(wherethemodelpredictionsare
comparedwiththefieldvalues),bothconcentration
andpredictedtimeoftracerarrivalmightnotbein
perfectagreementduetolocalvariationindemands,
localflowvelocities,anddilutionimpacts. Thesharp
tracerfrontsobservedinthisfieldstudymadeit
difficulttoemployquantitativestatisticalmeasures
(e.g.,meanerror,standarddeviation,rootmeansquare
error). Therefore,agraphical(visual)approachwas
consideredtobemoresuitableformodelcalibration
Figure 4-9. Sample Chloride Data Used at One
Station for Calibration.
actualfield-measuredvalues,indicatingtheneedfor
modelrefinementandre-calibrationtoimprovethe
predictioncapabilityoftheEPANETmodel.There-
fore,EPANETmodelingwasperformedtoevaluate
thefollowingfourlevelsofmodelrefinements:
Level1(priortocalibration): Askeletonized
EPANETmodelwasusedwiththeoriginal
hourlydemandpatternprovidedbyGCWWand
atime-stepinjectionpatternof60minutes.
Level2: ThesameasLevel1,butarefined10-
minutetime-steppatternforinjectionwasused
alongwiththeconversionoftheoriginalhourly
demandpatternsto10-minutepatterns.
Level3: ThesameasLevel2witharefined
demandpatternforeachnodeusingthefield-
measuredflowdata,additionofdemandnodes
representingwaterdemandofthepartiallyopen
hydrants,adjustmentforalargeindustrialuser
ofwaterinthestudyarea(basedondata
obtainedduringthestudy),andtheresidential
waterbillinginformationprovidedbyGCWW.
Level4: ThesameasLevel3withadetailed
all-pipe(non-skeletonized)EPANETmodel.
Theresultsofthefour-stagemodelrefinementand
calibrationprocessareshowninFigure4-10fora
continuous monitoring location (CM-18) located on
themainfeederpipe. Asillustrated,theimprove-
mentsinthedemandestimatesandinclusionofthe
systemdetailsintheall-pipemodelresultedinavast
Duringthecalibrationandrefinementprocess,various
modelinputssuchasflow,demand,andpipe
characteristicswereadjustedtoimprovethemodel
prediction. TheEPANETmodelwasconsideredtobe
calibratedfortheareawhenthefielddatamatchedthe
model-predictedoutputtoanacceptabledegreebased
onvisualobservation. Dependinguponthelocation
improvementinthemodelspredictionabilityforthat
monitoringlocation. Similarresultswerefoundfor
mostmonitoringlocationsonthemainpipe.
4-9
AReferenceGuideforUtilities
inthisapplication. Forexample,ifthepredictionof
thearrivaltimeforthetracerdiffersbyevenafew
minutesfromtheobservedarrivaltime,useofthese
standardmeasuresoferrorcouldresultinahigh
number,eventhoughthepredictioncouldbeviewed
graphicallyasverygood.
Thecalibrationoftheloopedportion(referringto
theportionofthenetworkonthebottomrighthand
sideofFigure4-7)ofthisnetworkprovedtobemore
difficultandtheresultsforsomemonitoringlocations
ontheloopedpipingwerelesssatisfactory.Themost
problematic were continuous monitoring locations
CM-02andCM-04. MonitoringstationCM-02was
locatedneartheconfluenceoftwoseparateloops,
with the actual monitored connection being slightly
offsetfromthejunctionnode. Examinationofthe
modelresultsshowedthatflowreachedthatjunction
frombothdirectionsandsmallvariationsinthe
amountofflowineachoftheloopsresultedinvery
differenttraveltimes.AsillustratedinFigure4-11,
thiscomplextravelpatternalongwiththeoffset
locationofthemonitoringstationresultedinpoor
predictionoftraveltimetothatstation. Also,
monitoringstationCM-04islocatedattheendofa
dead-endpipesectionandtraveltothisnodeis
stronglyinfluencedbydemandsattheveryfarendof
thedead-endsection.AsillustratedinFigure4-11,
thisresultedinapoormatchofthepeakconcentra-
tionduringthesecondpulse. Itisalsopostulatedthat
dispersion,whichisnotrepresentedinEPANET,may
havehadaninfluenceonthepeakconcentrationdue
totheverylowvelocitiesinthedeadendpipe. In
somecases,thiscouldalsobecausedbyinaccurateC-
factorsasappliedtothedistributionsystem. However
(asillustratedinFigure4-11),formonitoringlocation
CM-03locatedinthemainpartoftheloopingsystem,
themodelandfieldagreementwasquitegood.
Case1dataillustratesthat,dependinguponthelevel
ofrefinementandcalibration,thereisasignificant
variationinthecapabilityofamodeltoaccurately
representthesystem. Ingeneral,thepartsofthe
networkthatareconfiguredastrees(mainstemwith
branches)aremoreeasilycalibratedbymaking
adjustmentsindemands. Forloopingpartsofthe
systemandatdead-ends,resultsareverysensitiveto
CM-02
CM-03
CM-04
Figure 4-11. Calibration of Looped Portion.
smallvariationsindemandsandsystemconfigura-
tion,leadingtothepossibilityofsignificantpredic-
tionerrorsatsomelocations.Uncertaintyindemand
estimatescanbeamajorsourceoferrorinthemodel
estimates.
4.4.2Case2-Large-SuburbanPressureZone
SimilartoCase1,afieldstudyandcalibration
exercisewascarriedoutinalarge-suburban,pressure
zone. Thisareawasselectedinordertodemonstrate
theapplicationoftracerstudiesandcalibration
techniquesinamorecomplexarea. Theselectedarea
containedmultiplepumpsandtanks. Theselected
distributionsystemareaisrepresentativeofrelatively
complex,well-griddedsystemsfoundinmanylarger
watersystems. Thelayoutofthesystem,thelocation
oftheinjectionsite,andthemonitoringlocationsare
showninFigure4-12.
Twoseparatetracerstudieswereperformedinthis
zone.Thefirststudywasusedtofurthercalibratethe
skeletonizedmodelreceivedfromthewaterutility.
Thesecondstudyservedasavalidationeventtotest
theveracityofthecalibratedmodel.Inthecalibration
event,thetracerwasintroduceddirectlyintothemain
feedlineservicingtheentirearea(characterizedby
higherflow/higherpressure).Inthevalidationstudy,
thetracerwaspulsed. Atotalof34continuous
conductivitymeterswereinstalledinthesystem.
Fourflowmetersweretemporarilyinstalledtoprovide
flowmeasurementsatkeylocations.
Duringthecalibrationstudy,thecalciumchloride
tracerwasinjectedintothemainfeedlineservingthe
4-10
Figure 4-13a. Modeled Flows Compared to Measured
Flows Before Calibration.
Figure 4-13b. Modeled Flows Compared to Measured
Flows After Calibration.
AReferenceGuideforUtilities
EPANETmodeltoaccommodateforthisdevelop-
ment.Anotherpossibilityforthediscrepancywas
thatthedemandinthisregionwassignificantly
higherthantheaverageresidentialdemandmodeled
inthearea.Tosimulatethispossibility,asensitivity
analysiswasperformedinwhichthemodeleddemand
inthisregionwasdoubled.Themodel-predicted
resultsimprovedsignificantlyforthisregionbasedon
thesethreeadjustments.
InRegion2(CM52,CM53,CM55andCM56),an
oppositephenomenontothatinRegion1was
observed.Thefielddataindicatedthatthetracer
arrivedseveralhoursafterthemodelsprediction.
Onepossibleexplanationwasthatthisregionhad
lowerdemandthantheaverageresidentialdemand
modeledinthisarea.Theflowmeterdataupstreamof
thislocationsupportedthistheoryastheEPANET
predictedflowinthispipewasmuchhigherthanthe
fieldobservedflow(seeFigure4-13a).Tosimulate
thispossibility,thelocaldemandinthisregionwas
reducedby30percentinthemodel.Theresultant
flowmatchedtheflowmeterdata(seeFigure4-13b).
Also,similartoRegion1,itwasfoundthatapotential
flowpathhadagainbeenleftoutdueto
skeletonizationofthemodel,whichaffectedCM52.
Thispipelinkwasaddedtothemodelusingthe
appropriatepipeparameters. Distributionmains
betweenCM55andCM53werealsofoundtohave
beenupgradedsincetheEPANETnetworkmodelwas
Injection Point
Flow Monitoring Location
Continuous Monitoring Sample Location
CM-54
CM-53 Region 2
CM-52
CM-51 CM-41
CM-42 Region 1
CM-56
CM-45
CM-55 CM-50
CM-43
CM-44
Q21
CM-49 CM-46
CM-47
CM-57 CM-48
CM-37
CM-61
CM-59 CM-40
CM-58
CM-64 CM-32
Q23
CM-36
Q22
CM-60
CM-31
CM-33
CM-62
Q20
CM-39
CM-63 Region 3
CM-35
CM-34
CM-38
Figure 4-12. Schematic Representation of Case 2
Study Location.
areaforaperiodof6hours.Inthevalidationstudy,
thetracerwaspulsedbyfillanddrawcyclesina
storagetankatthesamelocation.Inbothcases,a
targetchlorideconcentrationof190mg/Lorlower
wassetinordertosafelynotexceedthe250mg/L
secondaryMCLforchloride.
Duringthecalibrationprocess,initialEPANET
modelsimulationswerereviewedindetailto
determinetheflowpatternsaroundvariousmoni-
toringlocationsandtoattempttoidentifycauses
fordiscrepanciesintheobservedandpredicted
values.Acarefulexaminationoftheareasof
significantdiscrepanciesindicatedthatthesewere
primarilylimitedtothreegeographicsub-regions
withintheskeletonizednetwork.Inadditionto
thesethreesub-regions,therewereafewisolated
locationswherethepredictedtracerpatterndidnot
matchtheobservedtracerpatternfromthefield
study.Themodelingteamcarefullyexaminedeach
oftheseregionsandaddressedthezonalissues
accordingly.Thethreesub-regionsareshownin
Figure4-12.
InRegion1(CM42,CM43,andCM44),thefielddata
indicatedthatthetracerarrivedatthesecontinuous
monitoringlocationsseveralhoursbeforethemodels
prediction.Oncloserinspection,itwasfoundthata
potentialflowpathexistedwhichwasnotincludedin
theskeletonizedmodel.Whilethepipediameterwas
small,itsignificantlyalteredthehydraulicwaterflow
pathtothatregion.Thismissingpipe-linkwasadded
tothemodel,usingtheappropriatepipeparameters.
Furthermore,themodelingteaminvestigatedtheGIS
databasetoseeiftherewereanysubstantialchanges
intheseareassincethetimewhentheoriginalwater
demandpatternsweredevelopedfiveyearsago.The
updatedGISinformationindicatedapresenceof
recenthousingdevelopmentinthatregion.There-
fore,additionaldemandnodeswereenteredintothe
4-11
Figure 4-14a. Chloride Concentration for Calibration
Event at Continuous Monitor Location CM-59.
Figure 4-14b. Chloride Concentration for Validation
Event at Continuous Monitor Location CM-59.
AReferenceGuideforUtilities
developedforthisarea.TheEPANETmodelpipesfor
thislocationwereupdatedusingthenewerinforma-
tion.Themodel-predictedresultsimprovedsignifi-
cantlyforthisregionbasedonthesethreeadjustments.
InRegion3(CM34andCM35),thefielddata
indicatedthatthetracerarrivedatlocationsCM34
andCM35severalhoursafterthemodelspredicted
arrivaltime.However,thefield-verifiedtracerarrival
timematchedthepredictedtracerarrivaltimeat
locationCM33whichisslightlyupstreamofthese
locations. Also,areviewofthewaterflowpatternin
thisregionindicatedthatthewatertraveledfrom
CM33towardsCM34andCM35(atalltimes). Based
onthedemandsintheEPANETmodel,thepipe
lengths,andtheregionalwaterflowinformation,the
delayintracerarrivalatCM34andCM35couldnot
beexplained. Acloserinspectionoftheregion
revealedacomplexgridofinterconnectedpipesin
thisregion,whichwereskeletonizedastwoparallel
pipes. Thisskeletonizationeliminatedanumberof
differentpossiblehydraulicflowpathsbetweenCM
33andCM34/CM35. Also,intheEPANETmodel
inputs,itappearedthatthedemandclosetoCM34
andCM35wassetartificiallyhigher(toaccountfor
theoveralldemandintheskeletonizationprocess).
Thismodelsetupresultedinthepredictedfastertracer
arrivaltimesatCM34/CM35thanthoseobservedin
thefield.Toaccountforthisanomaly,afewpipe
segmentsfromthemasterplanwereaddedtothe
skeletonizedmodelofthisregiontobettersimulate
theactualgriddemandsnearCM34andCM35. This
modeladjustmentresultedinbetterpredictionofthe
tracerarrivaltimes.
Duringthecalibrationprocess,asdemandswere
adjusted,amassbalancewasperformedforeachhour
toensurethatthenetwaterdemandinthestudyarea
remainedthesame,i.e.,theincreaseinthedemandat
certainnodeswasbalancedbythereduceddemandat
othernodestoeliminateanynetimpactonwater
demand. Inthefinalrefinements,amultiplierof2.0
wasusedforthebasedemandinRegion1,anda
multiplierof0.7forthebasedemandinRegion2.
Theserefinementsshowedsomeimprovementinthe
modelsabilitytocorrectlypredictthetracerarrival
timeandconcentration. Thesecalibrationefforts
resultedinarelativelywell-calibratednetworkmodel.
However,somelocalproblemsremained,especiallyin
loopedareasandareasthatwerebranchedofffromthe
mainlines.
ThesubstantialchangesmadetotheEPANET
skeletonizedmodelrepresentingthelargearea
necessitatedavalidationprocess. Therefore,the
calibratedEPANETmodelinputfilefromthefirst
eventwasusedtovalidatethemodelscapabilityto
predicttheresultsduringthesubsequenttracer
addition. Forthepurposesofthisvalidation,thedata
fromthesecondsetofpulsedinjectionswasmodeled
usingthecalibratedEPANETnetworkmodelforthe
studyareatoseehowthepredictedresultscompared
with the continuous monitoring data collected during
thisevent. Themodeledandmeasuredconcentrations
arecomparedinFigure4-14afortheEPANET
calibration. AsimilarcomparisonisshowninFigure
4-14bforthevalidationstudy.
Additionallyforthepurposesofthisanalysis,the
EPANETpredictionsfromthevalidationeventwere
comparedwiththefieldresultsforeachmonitoring
siteandeachsitewasgivenagradeasfollows:
Verygoodmatch(within20percentofthe
actualconcentrationandwithin1hourofthe
actualtracerarrivaltime)
Moderatematch(within30percentofthe
actualconcentrationandwithin5hoursofthe
actualtracerarrivaltime)
Poormatch(greaterthan30percentofthe
actualconcentrationorgreaterthan5hoursof
theactualtracerarrivaltime).
Ofthe34monitoringsitesinthisstudyareaforthe
validationevent,15receivedagradeofverygood
match,14wereinthemoderatematchcategory,and5
receivedthelowestgradeofpoormatch. Ingeneral,it
wasfoundthatbettermatchesoccurredonlargerpipes
servinglargepopulations,whilethepoorestmatches
occurredinmorelocalizedloopsservingfewer
4-12
customers. Theseresultsare,ingeneral,quitesimilar
totheresultsobtainedforthecalibrationevent,and
mostproblemsrepeatedlyoccurredatthesame
locationsforbothevents. Thevalidationevent
resultsconfirmthefactthatthecalibratedEPANET
networkmodelcannowbeusedtopredictthe
outcomeofaseparateeventtothesamedegreeof
accuracy.
4.5FutureofModelCalibration
Calibrationcontinuestobeamajorfocusofmost
modelingefforts.Itcanprovideamodelthatmaybe
usedwithgreaterconfidenceandproduceresultsthat
arecommensuratewiththeimportantdecisionsthat
aremadebasedontheapplicationofthemodel.
However,thereissignificantroomforimprovements
in calibration methodologies and in developing a
standardizedsetofcalibrationprotocols.Thishasled
toanactiveresearchprograminthisareathatis
expectedtocontinueintothefuture.
4.5.1CalibrationStandards
Thefollowingissuesareraisedfrequentlyinthefield
of distribution system modeling:
extentofcalibrationneededforvarious
applications, and
standardsforcalibration.
Thoughtheseareveryreasonablequestions,straight
forwardanswersareusuallynotreadilyavailable.
Thereisgeneralagreementinthemodelingprofession
thattheamountanddegreeofcalibrationrequiredfor
amodelshoulddependupontheintendeduseofthe
AReferenceGuideforUtilities
model(EngineeringComputerApplicationsCommit-
tee[ECAC],1999).Someapplicationssuchasdesign
andwaterqualityanalysistypicallyrequireahigh
degreeofcalibration,whileotheruses,suchasmaster
planning,canbeperformedwithamodelthathasnot
beencalibratedtosuchahighstandard. However,
therearenouniversallyacceptedstandards.
IntheUnitedKingdom,thereareperformancecriteria
formodelingdistributionsystems(WaterAuthorities
AssociationandWRc,1989). Theseareexpressedin
termsoftheabilitytoreproducefield-measuredflows
andpressureswithinthemodel,asshownbelow.
Flow
1. 5percentofmeasuredflowwhenflowsare
morethan10percentoftotaldemand
(transmissionlines).
2. 10percentofmeasuredflowwhenflowsare
lessthan10percentoftotaldemand
(distribution lines).
Pressure
1. 0.5m(1.6ft)or5percentofheadlossfor85
percentoftestmeasurements.
2. 0.75m(2.31ft)or7.5percentofheadlossfor
95percentoftestmeasurements.
3. 2m(6.2ft)or15percentofheadlossfor100
percentoftestmeasurements.
In1999,theAWWAEngineeringComputerApplica-
tionsCommitteedevelopedandpublishedasetof
draftcriteriaformodeling. Thesewerenotintended
astruecalibrationstandards,butratherasastarting
pointfordiscussiononmodelingneeds.Thesecriteria
aresummarizedinthefollowingtable(Table4-2).
Table 4-2. Draft Calibration Criteria for Modeling (based on ECAC, 1999)
Intended
Use
Levelof
Detail
Typeof
Simulation
Numberof
Pressure
Readings
1
Accuracyof
Pressure
Readings
Numberof
Flow
Readings
Accuracyof
Flow
Readings
Long-Range
Planning
Low Steady-State
orEPS
10%ofNodes 5psifor
100%Readings
1%ofPipes 10%
Design Moderateto Steady-State 5%- 2%of 2psifor90% 3%ofPipes 5%
High orEPS Nodes Readings
Operations LowtoHigh Steady-State
orEPS
10%- 2%of
Nodes
2psifor90%
Readings
2%ofPipes 5%
Water
Quality
High EPS 2%ofNodes 3psifor70%
Readings
5%ofPipes 2%
1
Thenumberofpressurereadingsisrelatedtothelevelofdetailasillustratedinthetablebelow.
LevelofDetail NumberofPressureReadings
Low 10%ofNodes
Moderate 5%ofNodes
High 2%ofNodes
4-13
AReferenceGuideforUtilities
Atthispoint,thereisnoclearmovementtoward
establishingcalibrationstandards. However,itis
likelythattheneedforfurtherguidanceinthisarea
willincreaseastheextentandsophisticationof
modeling continues to expand.
4.5.2TechnologicalAdvances
Researchiscontinuingintwoareasthatstrongly
influencethelikelihoodofimprovedcalibrationof
waterdistributionsystemsmodels:monitoring
technology and optimization techniques. The
available optimization techniques (and those under
development)havebeenbrieflydiscussedinthis
chapterandinChapter2. Activeresearchand
development areas include optimization techniques
forwaterqualitycalibration,EPSmodels,anduseof
tracerdata. Areasofresearch,development,and
experimental applications in monitoring technology
includelessexpensivemetersthatcanbeinserted
intopipesinthedistributionsystemandautomated
monitoringforuseinconjunctionwithtracerstudies
(asdiscussedinChapter3).
4.6SummaryandConclusions
Waterdistributionsystemmodelscanbeusedfora
numberofpurposes. Manyoftheseusesresultin
engineering decisions that involve significant
investments.Itisthereforeimportantthatthemodel
representtherealworld.Calibrationtechniquescan
beusedtoensurethatthemathematicalrepresentation
ofthesystem,ormodel,adequatelysimulatesthe
system.
Calibratingamodelisadifficulttaskbecausethere
aremanyparametersthatcanbeadjustedandfinding
thecombinationofparametersthatresultinthebest
agreementbetweenmeasuredandmodeledresultsis
oftenchallenging. Itisrecommendedthatthemodel
becalibratedusingonesetormoreoffielddataand
subsequentlyvalidatedwithanindependentsetof
fielddata.
Calibrationofwaterdistributionsystemmodelscan
beviewedinmanydimensions. Hydrauliccalibration
isusedtoadjusttheparametersassociatedwith
hydraulicsimulations,whilewaterqualitycalibration
isappliedtoreactionratesandotherparametersthat
controlthewaterqualitysimulation.Staticorsteady-
statecalibrationmethodsareusedwithsteady-state
modelsanddatacollectedatinstantaneoussnapshots
intime,whiledynamiccalibrationisconductedwith
extended-periodsimulationmodelsandtime-series
data. Manualcalibrationtechniquesinvolve manual
applicationofmodelsinatrial-and-errormode,while
automatedcalibrationusesthepowerofthecomputer
tosearchawiderangeofsolutionsandtoselectthe
setofparametersthatbestachieveastatedobjective.
Automatedmethodscanreducemuchofthetedium
Duringthecalibrationprocess,itisimportantto
eliminatevarioussourcesoferrorsinmodeling. Asa
firstpass,amodelershouldcheckfortypographical
errors,accuracyofaffectedpipinglayoutandmaterial,
generalsystemflow,velocityvalues,anddistribution
systemdemands. Thereafter,oneshouldlookintoother
sourcesoferrorssuchasskeletonization,valveposi-
tion,geometricnodeplacementanomalies,SCADA
dataerrors,andpumpperformance.
associatedwithcalibrationbutrequirethemodelerto
formallydefineaquantitativeobjectivefunctionfor
measuringhowwellthemodelmatchesthefielddata.
Suchautomatedmethodsarebecomingmoreavail-
ableincommercialmodelingpackages.
Twocasestudiesarepresentedinthischapter.The
casestudiesdifferintermsoftheoverallscaleofthe
studyarea.Inbothcases,thedistributionsystem
modelthatwasusedasastartingpointforthe
calibrationexercisewaspartofaskeletonizedmodel.
Theresultsdemonstratetheneedforadequatemodel
calibration.
Theextentofcalibrationandcalibrationtechniques
areamajorissueinmostmodelingefforts. Thereis
significantpotentialforimprovementsincalibration
methodologies and in standardization of calibration.
Thishasledtoanactiveandcontinuingresearch
programinthisimportantarea.
4-14
AReferenceGuideforUtilities
References
AWWA.Computer Modeling of Water Distribution
Systems (M32).AWWA,Denver,CO.2004.
ECAC.CalibrationGuidelinesforWaterDistribu-
tionSystemModeling.Proceedings,AWWAImTech
Conference,NewOrleans,LA. April18-21,1999.
McEnroe,B.M.,D.V.Chase,andW.W.Sharp.Field
TestingWaterMainstoDetermineCarryingCapac-
ity.MiscellaneousPaperEL-89,U.S.ArmyEngineer
WaterwaysExperimentStation.Vicksburg,MS.
1989.
Vasconcelos,J.J.,L.A.Rossman,W.M.Grayman,P.F.
Boulos,andR.M.Clark. Characterization and
Modeling of Chlorine Decay in Distribution Systems.
AWWAandAwwaRF,Denver,CO.1996.
Walski,T.M,D.V.Chase,D.A.Savic,W.M.Grayman,
S.Beckwith,andE.Koelle.Advanced Water Distri-
bution Modeling and Management. HaestadPress,
Waterbury,CT.pp268-278. 2003.
WaterAuthoritiesAssociationandWRc.Network
Analysis A Code of Practice. WRc,Swindon,UK.
1989.
Williams,G.S.,andA.Hazen.Hydraulic Tables. John
Wiley&Sons,NY.1920.
4-15
AReferenceGuideforUtilities
Chapter 5
Monitoring Distribution System Water Quality
Monitoringawatersupplysystemanditsvarious
componentsfacilitatesthegatheringofdataaboutthe
stateofthesystem(physical,operationalandwater
quality). Ifthestateofthesystemhasminimal
changesintimeorspace,asimplemonitoringsystem
maybesufficienttodefineandmanagethesystem
characteristics. However,ifthereispotentialfor
significantvariationinthestateofthesystem,the
monitoringsystemmustbeadequatelydesignedto
capturethatvariability.Thereafter,dependingupon
thetypeandmagnitudeofvariability,anappropriate
responsecanbeprovidedtorestorethenormal
systemstate. Thischapterwillfocusonmonitoring
waterquality-relatedparametersinadistribution
system.
Inadistributionsystem,waterqualitymayvarydue
tofactorssuchasnormalpatternsinwaterconsump-
tion,seasonalvariations,sourcewaterquality,
componentsofthedistributionsystem,operationof
thesystem,retentiontimeinstorage,traveltimein
thepipingsystem,ortheconditionofthesystem
itself. Variabilitymayalsoresultfromunusual
occurrences, such as intentional/accidental intrusions
ofcontaminants,orchemicalprocessessuchas
nitrification. Designofawaterqualitymonitoring
programmusttakeintoaccountboththenatureofthe
variabilityandthemannerinwhichmonitoringdata
willbeused. Inotherwords,theobjectiveofthe
monitoringprogrammustbedefinedalongwith
appropriateoutputorreportingrequirements.
Ingeneral,monitoringsystemscanbedefinedbased
ontheusesorneedsofthemonitoringprogram,the
generaltypeofmonitoringtobeperformed(manual
grab sampling and/or continuous automated online
monitoring),orthespecificmonitoringequipment
characteristics. Itisimportanttofirstestablishaclear
objective(s)formonitoring.Thereafter,depending
upontheavailabilityoffunding,need,andexpertise,
oneshouldselecttheappropriatesampling
technique(s)andmonitoringequipment. Oncean
appropriatemonitoringsystemhasbeenselectedand
implemented,itisimportanttooperateandmaintain
theprogramtoachieveoptimalresultsandbenefits.
However,thesystemshouldbeflexibleenoughso
thatitcanbemodifiedincaseitdoesnotmeetthe
original objective(s).
Thischapterdiscussesthevariousdriversorobjec-
tivesformonitoringfollowedbyasummaryof
availablemonitoringtechniques.Anoverviewof
monitoringequipmentispresentedfollowedby
guidelines for establishing monitoring requirements
(e.g.,selectionofparameters,numberandlocationsof
monitors,andmonitorcharacteristics). Someguid-
anceforengineeringandevaluatingremotemonitor-
ingsystemsisalsopresented,alongwithsomeEPA-
sponsoredmonitoringcasestudies. Thechapter
concludeswithasummaryandalistingofreferences.
Therecentstudiesinvolvingtheuseofonlinecontinu-
ousmonitoringsystemshaveresultedinlargestreams
ofdatathatdocumenttheminute-by-minutechanges
inwaterqualitythatexistatvariouspointsinthe
waternetworks. Theapplicationofthistechnology
hasthepotentialforprovidingnewinsightsastohow
waterdistributionsystemsmaybeoperatedand
designedtoimprovewaterquality.However,these
systemswillrequirearelativelyhighlevelofsophisti-
cationintermsofdatamanagement,includingthe
capabilitytogeneratereal-timereports,graphicaland
visualrepresentationofinformation,andcompliance
reportsformeetingdrinkingwaterstandards. Someof
thesedatastreamsmaywellrevealexcursionsinwater
qualitythatconstituteviolationsofcurrentorfuture
drinkingwaterstandards,orasecurity-relatedinci-
dent. Thistypeofinformationmayputpressureon
drinkingwaterutilitiesandregulatoryagenciestotake
remedialaction,possiblyonanemergencybasis,even
whensuchactionsmaynotbefullyjustified(or
warranted).However,carefulplanningandnegotia-
tionswithappropriateregulatoryauthoritiestodefine
thesepotentialexcursionsandthepropercorrective
actiontobetakenwouldpreventanymisunderstand-
ingsandminimizeoreliminatethepotentialfor
unjustifiedenforcementorresponseactions.
5.1EstablishingMonitoring
Objective(s)
Inordertodefineandimplementaneffectivemonitor-
ingplan,clearobjectivesmustbeestablished.
Collectingdatajustforthesakeofaccumulating
informationisnotcosteffective. Indrinkingwater
systems,thereareseveralspecificreasonstocollect
dataand,typically,themonitoringsystemistailored
tomeetoneormoreoftheseneeds. Theobjectivesof
monitoringdistributionsystemscanbebroadly
classifiedintothefollowingfiveuses:
regulatory driven monitoring,
security related monitoring,
processcontrolrelatedmonitoring,
5-1
AReferenceGuideforUtilities
Table 5-1. Federal Distribution System Water Quality Monitoring Requirements
Regulation Monitoring Requirement(s)
TCR Samplesmustbecollectedatsitesthatarerepresentativeofthewaterquality
throughoutthedistributionsystembasedonasiteplanthatissubjecttoreviewby
theprimacyregulatoryagency.
Theminimumnumberofsamplesthatmustbecollectedpermonthdependsonthe
populationservedbythesystem.
Foreachpositivetotalcoliformsample,therearerepeatsamplingrequirements,
additionalanalyses,andanincreasednumberofroutinesamples.
SWTRandIESWTR DisinfectantresidualsmustbemeasuredatTCRmonitoringsites.
LCR Allsystemsservingapopulation>50,000peoplemustdowaterqualityparameter
(WQP)monitoring.
ThenumberofsamplesitesforPb/CuandWQPmonitoringisbasedonsystemsize.
DBPR2 TheIDSErequirementofDBPR2inturnrequirestheestablishmentofaStandard
TheSMPwillrequireoneyearofdataonTHMsand MonitoringProgram(SMP).
HaloaceticAcids(HAAs).Thenumberofsamplinglocationsisbasedonutilitysize
andsourcecharacteristics.Modelingcanreducesamplingrequirements.
waterqualitycharacterization(e.g.,general,
baseline,orotherresearch-relatedmonitoring),
and
multi-purpose(acombinationofabove)useof
monitoring data.
Thefollowingsubsectionspresenttheoverallscope
ofeachofthesefiveobjectives.
5.1.1 Regulatory Driven Monitoring
Variousfederal,state,orothergovernmentalagencies
haveregulationsthatspecifydistributionsystem
monitoringrequirements. Anoverallreviewoffederal
regulationsimpactingdistributionsystemswas
presentedinChapter1. Thespecificfederaldistribu-
tionsystemmonitoringrequirements(existingand
proposed)aresummarizedinTable5-1.Insome
cases,stateshaveimposedmorestringentcriteriaand
monitoringrequirements.
5.1.2 Security Related Monitoring
Assessmentsperformedbyutilitiesandvarious
researchstudieshaveidentifiedthatwaterdistribu-
tionsystemsarevulnerabletointentional(oracciden-
tal)contamination. Inadditiontohardening
systemsinordertodeterintentionalcontamination,
monitoringaspartofanearlywarningsystem(EWS)
hasemergedasalogicalapproachtocopewith
potentialcontaminationevents. Therearenoexisting
orproposedstandardsforsuchmonitoring. However,
itiswellrecognizedthatmonitorswillneedtobe
sufficientlysensitivetoabroadrangeofpotential
contaminantsandappropriatelylocatedtodetecta
contaminationeventwithinareasonabletime.
Additionally,asdetailedinEPAsResponseProtocol
Toolbox(EPA,2003-2004),monitorsmustbean
integralpartofanemergencyresponsemanagement
planinordertobeeffective. Extensiveresearchand
developmentisunderwayonmonitordevelopment,
calibration,andplacementinresponsetotheper-
ceivedsecuritymonitoringneeds.
Currently,EPAhasanongoingtestprogramtoevalu-
atethepotentialofsensorsmonitoringroutineonline
waterqualityparameters,suchaspH,oxidation
reductionpotential(ORP),freechlorine,totalorganic
carbon(TOC),conductivity,andturbidity,toserveas
rapid detection devices for detecting contamination
eventsindistributionsystems. Onlinemonitorswere
selectedbecauseresponsetimeiscriticalforachieving
theobjectiveofprovidingearlywarning. Bothbench-
andpilot-scalestudiesarebeingconductedatthe
WaterAwarenessTechnologyEvaluationResearchand
Security(WATERS)CenterwithintheEPAsTestand
Evaluation(T&E)FacilityinCincinnati,Ohio. The
bench-scalerunsaredesignedtoidentifythedetection
thresholdofeachsensorforspecificcontaminants.
Thepilot-scalerunsaredesignedtoevaluateoverall
responseoftheselectedsensorsbyinjectingknown
quantities of potential contaminants into the distribu-
tionsystemsimulator(DSS). Forthispurpose,several
pilot-scaleDSSshavebeenfabricatedandusedfor
thesetestruns. Thesensordataarecollectedcontinu-
ouslyandarchivedelectronicallytoestablishstable
baselineconditionsandtoalsorecordsensorresponses
toinjectedcontaminants. Grabsamplesarecollected
periodicallybeforeandafterinjectionofcontaminants
toconfirmthesensorresults.
5-2
AReferenceGuideforUtilities
5.1.3 Process Control-Related Monitoring
Monitorscanalsobeusedinadistributionsystemto
providereal-timeornearreal-timeinformationon
waterqualitythatcanthenbeusedtocontroltreat-
mentprocessesatatreatmentplantorinthedistribu-
tionsystem. Theuseofcontinuouschlorinemonitors
inthedistributionsystemtocontroldisinfectantfeed
ratesattheplantoratin-distributionsystembooster
chlorinationstationsareexamplesofthistypeof
monitoring(Uberetal.,2003).
5.1.4 Water Quality Characterization
Informationfromlong-termmonitoringofdistribution
systemscanbeusedtodevelopbaselinetrendsin
waterqualityforthatsystem. Suchinformationis
usefulinevaluatingawatersupplysystemandfor
planningupgradesormodificationstosystemdesign
or operation.
Additionally,ifthisinformationisappropriately
distributed,itbuildsconsumerconfidenceandhelps
tokeepcustomersuptodateaboutthewaterquality
sothattheycanusethisinformationtomakedeci-
sionsaboutprotectingtheirhealth. Currently,there
arenostandardsorguidelinesforthistypeofmonitor-
ing. However,forthisinformationtobeusefuland
cost-effective,aregularprogramforexaminingand
analyzingthecollectedinformationisessential.
5.1.5 Multi-Purpose Use of Monitoring Data
Monitoringcanbeanexpensiveundertakinginterms
ofcapitalcosts,aswellasoperationandmaintenance
(O&M)costs,includinglabor.Costsincludethe
purchaseandupkeepofequipment,laboratory
analysis,labor,andconsumablesupplies.The
investment in monitoring and automated monitoring
systemsisjustifiableiftheresultingdataareusedfor
morethanoneobjective. Forexample,ifdata
collectedforsecuritypurposescanalsobeusedfor
processcontrol,itshouldbeeasytojustifypoten-
tially large investments in automated monitoring
equipment. Monitoringsystemsshouldbeproperly
designedinordertomeetmulti-purposerequirements.
5.2MonitoringTechniques
Thetwomajorfactorsindesigningandimplementing
aneffectivemonitoringprogramaresampling
techniquesandequipmentselection. Thissection
focusesonavailablemonitoringtechniques. Samples
canbecollectedandanalyzedintwoways:grab
samplesand/orbyautomatedonlinemonitoring.
Automatedmonitors(continuousordiscrete)are
sometimessupplementedwithautomatedsamplers
thatcancollectbothdiscreteandcompositewater
samplesforfurtheranalysisatalaterdate/time. Grab
samplesarecollectedmanuallyandanalyzedinthe
fieldorinthelaboratory.Grabsamplesarelabor-
intensiveincomparisontoautomatedsamplingand
providesnapshotinformationaboutthesystematthe
timeofsamplecollection. Automatedmonitoring
usesonlineinstrumentation,anddataiscollectedby
meansofsensorsandautomateddataloggers. They
canalsobetiedtoaSCADASystem.High-end
monitorsrequireahighercapitalexpenseforthe
purchaseandmaintenanceofsensors,dataacquisi-
tion,datacommunication,datastorage,anddata-
processinghardwareandsoftware.However,thistype
ofmonitoringprovidesacontinuoustime-series
profileofchangesinwaterquality.Bothautomated
andgrabsamplingcanbeincorporatedintoacompre-
hensivemonitoringplan. Thesetechniquesare
furtherdiscussedinthefollowingsubsections.
5.2.1 Manual Grab Sampling
Historically,routinewaterqualitymonitoringin
distributionsystemshasbeencarriedoutthrough
manualgrabsamplesfollowedbyanalysisinthefield
orinthelaboratory.Essentially,allregulatory
monitoringisstillcarriedoutbythismethod. For
example,samplesrequiredforlargecommunitywater
supplysystemsundertheSWTRaremanually
Theequipmentroutinelyrequiredinamanualgrab
samplingprogramincludesfieldsamplingequipment
(e.g.,chlorinemeter),safetyequipment(vests,raingear,
andflashlights),andlaboratoryequipment. Consum-
ablesuppliesincludesamplingcontainers,reagents,
andmarkingpens. Oneshouldidentifytheneedsand
availability of equipment and supplies and investigate
varioussourcesforequipment. Becauseequipment
malfunctionorlossispossible,someredundancyin
equipmentisappropriate. Someimportantfunctionsto
considerwhenestablishingafieldsamplingprogram
include the following:
Establishasystematicandorganizedmethodfor
allsamplinganddatarecording. Takenotesto
documentallaspectsoftheprocess.
Providetrainingtosamplingcrewsandspecify
thesetrainingrequirementsinthesampling
programplan.
Contingency planning is important; therefore,
considerthepotentialforequipment
malfunction,illnessofcrewmembers,
communicationproblems,severeweather,
malfunction,andcustomercomplaints.
Establishacommunicationsprotocolto
coordinateactions. Ameansofcommunication
isneededtorespondtounexpectedevents.
Alternativesincluderadios,cellularphones,
walkie-talkies,oracoordinatorinavehicleto
circulateamongfieldcrews.
Calibratefieldanalyticalequipmentbeforeand
during the sampling activity.
5-3
AReferenceGuideforUtilities
collectedatsiteswithinthedistributionsystemand
testedfordisinfectantlevelsinthefield.Samplestaken
tosatisfytherequirementsoftheTCRarealsomanually
collectedinthefieldandsubsequentlyanalyzedinthe
laboratory.Manualsamplingis labor-intensiveandthe
numberofsamplesthatcanbecollectedislimitedby
availabilityofpersonnelandanalysiscosts.However,
theyarespecifiedbysomeregulations. Potentially
importanteventsthatmayoccurbetweentheroutine
grabsamplesmaybelost(e.g.,processupset). Also,
thereisapotentialfordismissingunusualgrab
samplingresultsassometypeofmanualmonitoring
error(Hargesheimeretal.,2002).
5.2.2 Automated/Online Monitoring
Asstatedinthereport,OnlineMonitoringfor
DrinkingWaterUtilities(Hargesheimeretal.,2002),
Thereisanevolutionfromgrab-samplemonitoring
toonlinemonitoringassampling,analysis,data
processing,andcontrolfunctionsbecomemore
automated.Onlinemonitoringrequiresamechanism
formovingthesamplewaterfromthedistribution
systemtoaninstrument,appropriateinstrumentation
foranalyzingthewater,amechanismforcommunicat-
ingtheresults,andameansofassessingtheresultsof
the monitoring. Additionally, the instrumentation
mustbeperiodicallycalibratedandmaintainedfor
quality control/quality assurance.
Inthepast,distributionsystemonlinemonitorswere
typicallyhousedinacontrolledenvironmentwith
samplelinesfromthedistributionsystemtothe
instrument. Thisresultedinmostinstrumentation
beinglocatedatfacilitiessuchastanksandpump
stations. Theinstrumentationwassometimescon-
nectedtoaSCADAsystemsothatresultscouldbe
communicatedtoacentraloffice.Morerecently,
someinstrumentationisavailablethatisdesignedfor
installationinmanholesorfordirectinsertioninto
watermains.
TheAmericanSocietyofCivilEngineers(ASCE),in
concertwithotherleadingorganizations,enteredinto
acooperativeagreementwiththeEPAtodevelop
standardsdocumentsandguidanceaimedatenhancing
thephysicalsecurityofthenationswaterandwaste-
water/stormwatersystems. Underthisagreement,
ASCEisleadingtheefforttodevelopguidelinesfor
designing an online contaminant monitoring system
(OCMS). TheInterimVoluntaryGuidelinesfor
DesigninganOCMSwerepublishedinDecember
2004(ASCE,2004). Thisdocumentprovidescompre-
hensiveinformationonseveraltopics,including
rationaleforOCMSandsystemdesignbasics,selec-
tionandsitingofinstruments,dataanalysis,anduseof
distribution system models.
5.3MonitoringEquipment
Overview
Ingeneral,monitorscanbecategorizedbythetypes
ofparameters(contaminants,agents,andcharacteris-
tics)thatthemonitorisusedtomeasure. Forestab-
lishingwaterquality,themonitorsaredesignedto
measureoneormoreparametersthatrepresent
physical,chemical,and/orbiologicalcharacteristics
ofthesystem. Typically,inmanualgrabsampling
programs,hand-heldphysicaland/orchemical
parametermeasuringdevicesareused. Thesehand-
helddevicesarecarriedtothesamplinglocation
alongwithappropriatecontainerstocollectwater
samplesforperformingmorecomplexchemicaland
biologicalanalysesinalaboratory.Theonline
samplingdevicesaremorecomplexdevicesthatare
designedtoautomaticallymeasure,record,and
displayspecificphysical,chemical,orbiological
parameters. Abriefoverviewofthesedevicesis
presentedinthefollowingsubsections.
5.3.1 Physical Monitors
Physicalmonitorsareusedtomeasurethephysical
characteristicsofthewaterinadistributionsystem.
Theyincludeavarietyofinstrumentationthat
measuresvariousmacrocharacteristics,suchasflow,
velocity,waterlevel,pressure,andotherintrinsic
physicalcharacteristics. Examplesofintrinsic
physicalcharacteristicsincludepH,turbidity,color,
conductivity, hardness, alkalinity, radioactivity,
temperature,fluorescence,UV254,andORP.In
general,physicalmonitorstendtoberelatively
inexpensive, quite durable, and readily available.
5.3.2 Chemical Monitors
Chemicalmonitorsareusedtodetectandmeasure
inorganicororganicchemicalsthatmaybepresentin
thewater.Awiderangeofchemicalsmaybeof
interest,andalargevarietyoftechnologiescanbe
used. Aspecifictechnologyormultipletechnologies
mustbeproperlyselectedforaparticularchemicalor
agroupofchemicalsofinterest. Examplesof
chemicalmonitorsinclude,butarenotlimitedto
residualchlorinemonitor,TOCanalyzer,andgas
chromatograph/massspectrometer(GC/MS). Typi-
cally,thesamegeneraltypeoftechnologymaybe
available in either automated online monitoring
capabilityortosupportmanualgrabsampleanalysis.
5.3.3 Biological Monitors
Biologicalmonitors(biomonitors)includebio-
sensorsandbio-sentinels. Bio-sensorsdetectthe
presenceofbiologicalspeciesofconcern,suchas
someformsofalgaeorpathogens. Thegeneral
operatingprinciplesofbio-sensorsmayinclude
photometry,enzymatic,and/orsomeformofbio-
chemicalreaction. Thebio-sentinelsusebiological
5-4
AReferenceGuideforUtilities
organismsassentinelstodeterminethelikely
presenceofchemicaltoxicityinawatersample.
Ingeneral,bio-sentinelscannotbeusedtoidentify
thepresenceofaspecifictoxiccontaminantrather
onlythatthereissomeformoftoxiccontaminant
present. Mostbio-sentinelsoperatebyobservingthe
behaviorofselectedorganisms. Examplesofsuch
organismsinclude:fish,mussels,daphnia,het-
erotrophicbacteria,andalgae. Whenthesentinel
organismsensesthepresenceoftoxicity,itreactsin
someunusualmanner.Bio-sentinelinstruments
respondtothesereactionsandnotethatanunusual
eventisoccurring. Thisapplicationissomewhat
analogoustotheuseofindicatororganisms(e.g.,total
coliforms)toindicatethewaterqualityinthedistri-
bution system.
Whilebio-sensorscanbedirectlyappliedindistribu-
tionsystemswithoutpretreatmentofthesample,the
bio-sentinelsaretypicallyusedinsourcewaters. This
isbecausemostorganismsaresensitivetothe
presenceofchlorine(orotherdisinfectants)inthe
water.Therefore,ifabio-sentinelisproposedtobe
usedfordistributionsystemmonitoring,thewater
mustbede-chlorinatedpriortoenteringthebio-
sentinelinstrument. Dechlorinationmayaffect
detectionreliabilityandthechemicalcharacteristics
ofthewater.Also,thebio-sentinelsrequireapro-
tectedhousingenvironmentalongwithsomesortof
nutritionalsupplytokeepthesentinelorganismalive
and healthy.
5.4EstablishingMonitoring
Requirements
Selectionofthetypes,numbers,andlocationsof
monitorsisdependentonthenatureofthemonitoring
programdesired.Theserequirementsdependupon
the overall monitoring objectives and the distribution
systemsite-specificrequirements. Forexample,a
monitorusedforregulatorypurposesmayneedto
monitordifferentconstituentsthanoneusedaspartof
aprocesscontrolorsecuritysystem. Similarly,a
differentmonitormaybeneededforautilitythatuses
chlorineasthedisinfectantcomparedtoonethatuses
chloramine. Thesite-specificmonitoringrequire-
mentscanbeevaluatedandrepresentedinthe
followingterms:
monitoringparameters,
numberandlocationofmonitors,
nonitorcharacteristics(e.g.,detectionlimits,
samplingfrequency,cost,falsenegatives/false
positives), and
amenabilitytoremotemonitoringandSCADA
integration.
Theserequirementsarefurtherdiscussedinthe
following subsections.
5.4.1 Monitoring Parameters
Theparameterstobemonitoreddependstronglyupon
thespecificuseofthemonitoranduponutility-
specificsituations. Forregulatorypurposes,the
regulationstypicallyspecifytheminimumsetof
parametersthatmustbesampled. Foreachsystem,the
regulating authority typically also specifies the
monitoringlocationsandfrequency.Autilitymay
choosetoanalyzethewaterforadditionalparameters
and/orincreasethefrequencyofmonitoringinorder
toaddressotherwaterqualityconcerns.
Forsecuritymonitoring,therearenoregulationsor
standards. Utilitiescanchoosewhetherornotthey
wanttoperformsuchmonitoringandselectthe
parameterstheywillmonitor.Generally,such
monitoringwillbelimitedbybudgetsandby
technology.Researchanddevelopmentisbeing
conductedonsecuritymonitoringsystems,in
conjunction with event detection platforms, that
measurestandardparameters,suchasTOC,pH,
turbidity,conductivity,chlorine,ORPandtempera-
ture. Forbothprocesscontrolandsecurity-related
monitoring,instrumentresponsetimeiscritical.
Therefore,onlinemonitorsaretypicallyusedinthese
typesofapplications. Theparametersmonitoredvary
widelydependinguponthetypeofprocessand/or
security monitoring.
Thegoalofonlinemonitoringforsecuritypurposesis
toautomaticallyanalyzethedatatodetermine(1)
whetherthereisanindicationofunusualcontamina-
tioninthesample;and(2)whatthelikelycontami-
nantis,basedonthewaterqualitysignatureofthese
parameters.
5.4.2 Number and Location of Monitors
For selecting monitoring locations in distribution
systems,therearetworelateddecisions:(1)howmany
monitorstoplaceinthesystem,and(2)whereto
placethem. Thenumberofmonitorsisgenerally
controlledbythemonitoringobjective(e.g.,regula-
toryrequirement)orbybudgetaryfactors,whilethe
locationofmonitorsisamorecomplexissuethatcan
beaddressedinmanyways. Forexample,forcompli-
ancewiththeTCRandtheSWTR,therearespecific
requirementsastothenumberofsamplesthatmustbe
taken. Formostotheruses,thenumberofsampling
points(orthenumberofmonitorsinstalled)is
controlledbybudgetaryandfinancialconstraintsand
throughcomparisontothebenefitsassociatedwith
themonitors. Thefollowingsubsectionssummarize
anapproachthatcanbeusedwhentheestablished
objectivesdonotclearlydefinethenumberand
locationofmonitors.
5-5
AReferenceGuideforUtilities
EPAsresearchincontaminationwarningsystems(CWSs)attheT&EFacilityis
developingdatabasedonbench-andpilot-scaleexperimentsthatrevealhow
traditionalwaterqualityparameters,ifmonitoredonline,canserveastriggers
forcontaminationevents. Figure5-1showstheresponseofseveralinstruments
totheinjectionofsecondarywastewaterintoaDSS.
Figure 5-1. Wastewater Injection: Free Chlorine and Associated Grab
Sample Results.
5.4.2.1 Number of Monitors
Toselecttheoptimumnumberofmonitorsfora
distributionsystem,theoreticallyonecanperforma
simplecost-to-benefitanalysis.Iftheoveralllife
cyclebenefitsofeachmonitorexceeditslifecycle
costs,analysiswouldsuggestthatthemonitoris
justified.Lifecyclecostsrepresentboththecapital
andoperationalcostsforthemonitors.Depending
uponthelocation-specificrequirements,asthe
numberofmonitorsincrease,theremaybeeconomies
ofscaleortheunitcostmayactuallyincrease
disproportionately. Theunitcostsincreasewhenthe
additionalmonitorsareplacedinlessconvenient
locationswhereservicingand/ordatacommunication
costsarehigher.Frequently,budgetaryconstraints
mayalsolimitthenumberofmonitorsthatcanbe
deployed,evenifbenefits
justifytheircosts.
Figure5-2isagraphical
representationofbenefits
associatedwithincreasingthe
numberofmonitorsina
distribution system. This
graph illustrates that typically
afterabasicnetworkof
monitorshasbeenestablished
foradistributionsystem,the
incremental benefits gained
by installing additional
monitorsfollowthelawof
diminishingreturns. The
actualdevelopmentofsucha
graphisdifficultbecauseof
the need to explicitly quantify
benefits. Inthecaseofwater
security-related monitoring,
onecouldmeasurethevalue
basedonpopulationor
sensitivefacilities(e.g.,
hospitals)protectedbyuseof
onlinemonitors.Forother
typesofmonitoringsitua-
tions, quantification of
benefitsismoredifficult.
Thoughaformalcost-benefitanalysismaynotbe
feasible,thisdiscussionprovidesageneralframework
thatcaninformallyguidethedesignofamonitoring
network.
5.4.2.2 Optimal Monitor Locations
Historically,monitors/sensorshavebeenplacedin
distributionsystemstomeetregulatoryrequirements.
Theirlocationshavebeendeterminedbasedonease
ofaccessandageneralintuitiveassessmentof
representativelocations. Leeetal.(1991)proposeda
methodforlocatingmonitors,basedontheconceptof
coverage,whichisdefinedasthepercentageoftotal
demandthatissampledbyasetofmonitors. Various
otherresearchersfurtheraddressedthisissueusing
alternativemathematicalmethods(Kessleretal.,
1998). Thoughwidelycited,thesemethodologies
haverarelybeenappliedinactualpractice.However,
followingtheattacksofSeptember11,2001,therehas
B
e
n
e
f
i
t
s
NumberofMonitors
beenarenewedinterestinthedevelopmentof
monitoringtechnologyandplacementofmonitorsin
thedistributionsystemasamechanismfordetecting
intentional contamination of distribution systems.
Manycurrentstudiesareapplyingoptimization
techniquestodeterminetheoptimalplacementfor
monitorsindistributionsystemsbasedonadefined
objectivefunction. Ostfeld(2004)andOstfeldand
Figure 5-2. Theoretical Example of Benefits from
Salomons(2004)providereviewsofpastworkinthis
Monitors. areaandpresentexamplemathematicalformulations
5-6
AReferenceGuideforUtilities
A17
th
centuryItalianeconomist,VilfredoPareto,
developedamethodforcomparingalternatives. Based
onhiswork,asituationisdefinedasbeingPareto-
optimalifbyreallocationyoucannotmakesomeone
betteroffwithoutmakingsomeoneelseworseoff. This
canbeappliedtoevaluatingmonitorsbyexaminingthe
diagram(Figure5-3)wherevariousmonitoringoptions
arecomparedintermsoftheircostandsomemeasureof
effectiveness. JustlookingatalternativesAandB,we
cansaythatAisbetterthanBbecauseitcostslessand
ismoreeffective. Bycomparingallpotentialalterna-
tives,wecandefineaParetofront. Allalternatives
locatedonthatfrontarebetterthanalternativeslocated
totherightandbelowthefront. Thisprovidesauseful
conceptual mechanism for evaluating alternative
monitoringschemes.Foradditionalinformationonthe
workofPareto,seeJohansson(1991). Formoredetails
ontheapplicationofParetosconceptsintheareaof
optimizationrelatedtowaterdistributionsystem
analysis,seeWalskietal.,2003).
Figure 5-3. Pareto-Optimal Cost Effectiveness
Diagram.
Cost
E
f
f
e
c
t
i
v
e
n
e
s
s
Paretofront
1
2
3
4
5
6
A
B
usinggeneticalgorithmsolutiontechniques. Their
methodologyfindsanoptimallayoutofanearly
warningdetectionsystemcomprisedofasetof
monitoring stations aimed at capturing contamination
fromexternalsources,nodes,ortanksunderEPS
conditions. Berryetal.(2004)developedanoptimi-
zationprogramthatconsidersthemaximumvolume
ofcontaminatedwaterexposureataconcentration
higherthanadefinedsafelevel. Themethodusesan
integer programming optimization technique to place
alimitednumberofperfectsensorsinthepipesor
junctionsofawaternetworksoastominimizethe
expectedamountofexposuretothepublicbefore
detection,assumingtheattackoccursonatypicalday.
Watsonetal.(2004)usemixed-integerlinearpro-
grammingmodelsforsensorplacementoverarange
ofdesignobjectives. Usingtwocasestudies,they
Bahaduretal.(2003)describesanapproachusing
PipelineNetinwhichGISdataandhydraulicmodel
resultsareusedtoguidethemanualplacementof
monitorsinordertofulfillsomegeneralcriteria. Ina
casestudyconductedwithpersonnelatawaterutility,
25potentialmonitoringsiteswereidentifiedand
subsequentlyreducedtotwobestsitesusingtheGIS/
PipelineNetframework.Thisapproachismoreclosely
relatedtothetraditionalmethodsforlocatingmoni-
torscomparedtotheoptimizationtechniquesde-
scribedinthissection.
showedthatoptimalsolutionswithrespecttoone
designobjective(e.g.,populationexposed)are
typicallyhighlysub-optimalwithrespecttoother
designobjectives(e.g.,timefordetection).The
implicationisthatrobustalgorithmsforthesensor
placementproblemmustcarefullyandsimultaneously
consider multiple, disparate design objectives.
Ingeneral,theoptimizationmethodsdescribedabove
areexperimentalapproachesthathavebeenapplied
onlytohypotheticalorsmallwatersystemsandare
basedonassumptionsabouttheavailabilityof
monitoring technology, ability to define explicit
objectivefunctions,andlimitedincorporationofthe
variabilityofwatersystemoperation. Further
researchanddevelopmentisneededbeforethis
technologyisreadyforroutineuse.
5.4.3 Monitor Characteristics
Thefollowingcharacteristicsofmonitorsmustbe
evaluatedpriortoselectinganappropriatedevice:
Minimumdetectionlimit(MDL)The
minimumdetectionlimitisthelowest
concentrationorvalueatwhichthemonitorcan
dependably detect the constituent of interest.
TheMDLcanvaryfordifferentconstituents,
differenttechnologies,orfordifferent
implementationsofthesametechnologyand
constituent.
Falsenegatives/falsepositivesTwoformsof
errorsassociatedwithamonitorarefalse
positivesandfalsenegatives.Afalsepositive
existswhenamonitorreports,incorrectly,thatit
hasdetectedaconstituentwherenoneexistsin
reality.Afalsenegativeexistswhenamonitor
reports,incorrectly,thataconstituentwasnot
detectedwhen,infact,itwaspresent. False
positivescanleadtounneededresponses,and
repeatedfalsepositiveswillleadtoalackof
confidenceintheinstrument. Lackofdetection
associatedwithafalsenegativeresultsinno
responsetoarealcontaminationeventandcan
exposeconsumerstocontaminantsinthe
system.
5-7
AReferenceGuideforUtilities
Theheightenedlevelofconcernovertheneedto
protectwaterdistributionsystemshasledtothe
initiationofresearchintothedevelopmentofCWSsfor
bothsourceandfinishedwaters(Clarketal.,2004a).
CWSsareintendedtoreliablyidentifylowprobability/
highimpactcontaminationeventsinsourceordistrib-
utedwater.TheInternationalLifeSciencesInstitute
(ILSI)developedareport(ILSI,1999)focusedonthe
developmentofenvironmentalwarningsystems(EWSs)
forsourcewater.Thesamedevelopmentprinciples
applytodistributionsystems. EWSsappliedto
distributionsystemsarecommonlyreferredtoasCWSs.
ThefollowingdesignrequirementsforEWSswere
identifiedbyILSIintheirreport:
provideswarninginsufficienttimetorespondto
acontaminationeventandpreventexposureof
the public to the contaminant,
capable of detecting all potential contamination
threats,
remotelyoperable,
identifiesthepointatwhichthecontaminantwas
introduced,
generatesalowrateoffalsepositiveandfalse
negative results,
provides continuous, year-round surveillance,
producesresultswithacceptableaccuracyand
precision,
requireslowskillandtraining,and
beaffordabletothemajorityofpublicwater
systems.
AkeyaspectofaneffectiveEWSwillbetheneedforit
tooperateinaremotemonitoring andreportingmode.
SamplingfrequencyTherateatwhicha
monitoranalyzesandreportsavalueisthe
samplingfrequency.Thismayvaryfromafew
secondsorlessforaninstrumentsuchasa
pressuregagetoanhourormoreforinstruments
thattakelongerperiodstoperformtheanalysis
suchasagaschromatograph. Forgrab
sampling,thisdelaymaybeevenhigher.Some
instrumentscanbesetfordifferentsampling
frequencies. Morefrequentsamplingmayresult
inhigheroperatingcosts,shorterbatterylife,
increaseddatastoragerequirements,or
increased communication needs.
AmenabilitytoSCADAintegrationThe
monitorsabilitytobeonlineandintegrated
intosomesortofSCADAorremotedata
acquisitionsystemiscriticalifmultipleremote
locationsaremonitoredsimultaneously.Most
currentonlinemonitorshaveanalog(e.g.,4-20
mA,1-20V)ordigitalsignal(e.g.,RS232,
RS485)outputsthatprovidetheabilityto
remotelycollectandstoredataatacentral
locationforanalysis.
OperationandmaintenancerequirementsThe
operationalrequirementsofmonitorscanvary
significantlyandmaystronglyimpactthe
selectionprocess. Issuesincludetheelectrical
needs,expendablematerialneeds(e.g.,
reagents,wearrelatedcomponents),temperature
andhumidityrequirements,needstohandle
wastestreamsfromthemonitor,andother
factorsrelatedtothehousingofthemonitor.
Similarly,themaintenancerequirementsofthe
monitorswillalsoimpacttheselectionprocess.
Issuessuchashowfrequentlyatechnicianmust
servicethemonitorinthefieldandthelevelof
expertiserequiredtoservicethedeviceare
important considerations when evaluating
monitors.
CombinationsofmonitorsTheabilityofa
monitoringsystemtoreliablydetecta
contamination event generally increases with
multiplemonitorsworkingintandem. For
example,asinglemonitorthatreportsasignal
slightlyabovethenoiselevelmayeasilybe
dismissed. However,ifmultiplemonitorsat
severallocationsincloseproximityorseveral
instrumentsatthesamelocationmonitoringfor
differentparametersalldetectapotentialevent,
amoreforcefulandrapidresponseislikely.An
ongoingareaofresearchisthedevelopmentof
dataminingalgorithmsthatcandifferentiateor
detectasignalabovebackgroundlevelsthatare
notnormallyobservedinthemonitoredsystem.
CostsThecostofmonitoringsystemscan
varyoverseveralordersofmagnitude. Asingle
simpleinstrumentmonitoringforaphysical
parametersuchasconductivitymaycostless
than$1,000. Thecostofamulti-parameter
physicalmonitoristypicallyinthe
neighborhoodof$10,000. Morecomplex
instrumentssuchasaTOCmonitororaGC/MS
costintherangeof$25,000to$90,000. The
costofmorecomplexinstrumentsora
monitoring station containing multiple
instrumentscaneasilyexceed$100,000in
capitalcost. Installationandongoing
maintenancecostsarefrequentlysite-specific
andvaryaccordingtoenvironmental
conditions.
5.4.4 Amenability to Remote Monitoring and
SCADA Integration
Foracomprehensivenetwork-widewaterquality
remotemonitoringprogram,itisessentialtoensure
thatthesystemanditsmonitoredcomponentsare
amenabletoremotemonitoringandSCADAintegra-
5-8
tion. TheSCADAcomponentaddstheelementof
controltothemonitorednetwork. Mostutilitieshave
somesortofSCADAfunctionalitytoautomateand
monitorthekeywatertreatmentand/ordistribution
operations. Thecontrollogicistypicallytriggered
basedonaspecifiedtimeand/orevent. Forexample,
thepumpsmaybesettofilladistributionsystemtank
atmidnightandwhenthetanklevelmonitordetects
thatthetankisfull(aneventreportedthroughthe
SCADAsystem),thecontrollogictoturnoffthe
pumpsisinitiated. Thistypeofcontrollogiccanbe
enhancedtoperformcontrolfunctionsbasedon
detectionofwaterqualitychangeinthedistribution
system. However,toachievethisfunctionality,one
needstounderstandthefollowingthreemajor
componentsofaremotemonitoringand/orcontrol
system(orSCADA):
onlinesamplinginstruments(e.g.,pH,ORP)
and/orcontroldevices(e.g.,pump,valves),
SCADAorremotemonitoringnetwork,and
fieldwiringandcommunicationsmedia.
Thesecomponentsarediscussedbrieflyinthe
following subsections.
Electricpowerisgenerallyrequiredforoperatingthese
components. Ifelectricpowerisnotreadilyavailable
atthedesiredlocationwhereamonitoristobein-
stalled,considerthecostsforinstallingasuitable
powerapparatus(e.g.,asolarpanel,batterypack).
5.4.4.1 Online Sampling/Control Devices
Onlinesampling/controldevicescanbethemost
expensivecomponentofaSCADAsystem.The
sensors,switches,monitors,andcontrollersusedina
SCADAsystemmayvarywidely,dependinguponthe
parametersthatneedtobecontrolledand/orobserved.
Thecostforonlinesamplingdevicescanrangefroma
fewhundreddollarstoover$100,000. Controlunits
suchassamplefeedpumpsorshut-offvalvesareless
expensive(Pangulurietal.,1999). Costsassociated
withmaintenanceandcalibrationoftheonline
sensorswhenplanningtheacquisitionandimplemen-
tationofaremotemonitoringnetworkshouldalsobe
considered.
5.4.4.2 SCADA or Remote Monitoring Network
LargerutilitiestypicallyusesometypeofSCADA
systemforwaterdistributionsystemcontrolthatcan
easilybeintegratedtoincludeonlinesampling
instrumentationinacost-effectivemanner. Also,
recentadvancesinelectronichardwareandsoftware
technologieshaveresultedinseveralcost-effective
SCADAalternativesforsmallersystems. Amicropro-
AReferenceGuideforUtilities
SensorsandTransducers: Asensorrespondstoa
physicaland/orchemicalstimulus,suchasthermal
energy,flow,light,chemical,pressure,magnetism,or
motion. Atransducertakesthemeasuredphysicaland/
orchemicalphenomenon(e.g.,pressure,temperature,
humidity,andflow)andconvertsittoanelectrical
signal. Ineachcase,theelectricalsignalsproducedare
proportionaltoaphysicaland/orchemicalquantity
beingmeasuredbasedonapre-definedrelationship.
Theelectricalsignalsgeneratedbytransducersoften
requireconditioning. Dependinguponthetrans-
ducer,asignalconditionercanbeusedtoperformone
ormoreconditioningfunctions,suchasnoisefiltration,
amplification, linearization, isolation, and excitation.
cessor-basedsmartSCADAsystemcanbeusedin
remotelocationsbysmallsystemoperatorswhere
directonlinecommunicationisexpensive. Smart
systemshavehigherinitialcosts,butoverallcostsare
reducedsincethecommunicationcosts(e.g.long-
distancephonecosts)arenegligiblebecausemostof
theburdenistransferredfromthemaincomputerto
theindividualSCADAunitattheremotesite
(Pangulurietal.,1999).NewerSCADAunitsare
fairlyinexpensive,withcapitalcostsrangingbetween
$500(PCcard-basedunitsandremotedatacollection
nodes)and$5,000(independentPC-basedfull
SCADAunits).
Thedataacquisitionhardwareprocessesthedigital
and analog inputs/outputs from various online
samplingandcontroldevices. Formonitoring
systems,thehardwaretypicallyprocessestheanalog
datameasuredfromvariousinstrumentsandtransfers
ittoacomputersystemfordisplay,storage,and
analysis. Inamonitoring/controlsystem(SCADA)
scenario,thehardwarewouldprocessbothanalogand
digitalinputs(typicallyfromafielddevice)and
outputs(toperformcontrolfunctionality). The
applicationsoftwareprovidestheoperatorthe
display,control,andanalysis(trendsandreports)of
collected data.
5.4.4.3 Field Wiring and Communication Media
Dependinguponavailability,cost,userpreference,
andtherelativelocationofthesensorstothedata
acquisitionsystem,thecommunicationmediacanbe
eitherwired(e.g.,direct,phoneline)orwireless(e.g.,
radio,cellular). Infieldenvironments,distributed
input/output(I/O)istypicallyemployed. Aremote
dataacquisitionhardwareunitemployedatthefield
location performs the appropriate signal conditioning
andtransmitsthedatatoacentralhub.Morerecently,
meshorgridcomputingsystemsareusedinremote
locationstoaddredundancyincasesoflinkfailures.
Thefieldwiringbetweenthesensorandtheremote
dataacquisitionhardwareunitisusuallydirect
wire.
5-9
AReferenceGuideforUtilities
Dependingupontheareacoveredandavailability,in
somecasesitmaybepreferabletousesomeformof
radiocommunicationdevices. Theavailableradio
communicationdevicesoperatemostlyinthevery
highfrequency(VHF)orultrahighfrequency(UHF)
range. TheVHFfrequenciesrangebetween30-300
megahertz(MHZ)andtheUHFfrequenciesrange
between300-1,000MHZ. InU.S.,mostofthe
availableVHF/UHFradiofrequenciesarelicensed.
The unlicensed bands available include the industrial,
scientificandmedicaldevicechannelswithfrequency
rangesbetween902-928and2,400-2,484MHZ.
Theunlicensedbandsdonothaveanyconnectionor
monthlyfeerequirements.
Typically,directwireandphoneline(including
cellular)communicationmediaareinexpensive. The
primarylimitationsassociatedwithselectingthe
communication media include installation and
operatingcosts,whichcanvarybetween$200(fora
simpletelephoneorcellularmodem)andseveral
hundreddollarsforasatellite-basedsystemper
location. Ongoingmonthlyoperatingcostscanrange
from$25foraphonelinetoapproximately$200per
monthforsatellite-basedserviceswithintheU.S(per
monitored location).
5.5EngineeringandEvaluatinga
RemoteMonitoringSystem
Onceallofthebasicrequirementshavebeenestab-
lished(e.g.,objectives,parameters,location)as
outlinedintheprevioussection(Section5.4)andthe
requirementsindicateaneedforasystem-wideremote
monitoringprogramforwaterquality,thefollowing
additionalsite-specificneedsshouldbeevaluatedfor
waterqualitymonitoringinadistributionsystem
(Pangulurietal.,1999):
Whatarethecomplexitiesofthedistribution
system(size,location)?
Whatlocationsarebestsuitedforsamplingand/
orcontrolsysteminstallation?
Issufficientflowandwaterpressureavailable
foronlineinstruments?
IsthereanexistingSCADAsystemavailable?
Whattypesofcommunicationmediaare
available at the selected locations?
Howmanyparametersaregoingtobe
monitored and/or controlled at each location?
Whatothersite-specificinformation(e.g.,
availabilityofpower,access,security)willbe
needed?
Additonalfactorstobeconsideredare(Haughtand
Panguluri 1998):
systemfeatures(e.g.,easeofoperation,
customization, networkability, operator
security),
cost(initial,training,serviceagreements,and
operationandmaintenance),and
vendorsupport(hardwareandsoftware
upgradesandremotediagnosis).
Itisimportantthateachsiteisevaluatedindividually
forappropriateSCADAsystemselection.Thecostof
SCADAsoftwarehasplummetedoverthepastfew
years. Forexample,thecostofonecommercially
availablegraphical(Windows-based)SCADAsoftware
packagehasdroppedfrom$30,000intheearly1990s
to $2,000 today.
Priortoselectingandimplementingaremotemonitor-
ingnetwork,oneshouldevaluatetheoptionscare-
fully.Engineeringaremotemonitoringsystemisa
difficulttaskthattypicallyinvolvesmanyfactors:
multi-dimensional objectives, changing needs, rapid
Besidestheaforementionedimmediateneeds(e.g.,ease
ofoperation,customization,networkability),SCADA
systemfeaturesinclude:
Scalability:Thisallowsforfuturegrowthwith
respecttoadditionofI/Oblockswithmore
channelsoradvancecapabilities. TheseI/O
channelsareusedtocommunicatewithvarious
fieldmonitoringinstruments(sensors)andcontrol
devices.
LocalMemory: TheSCADAhardwaremustalso
containsufficientlocalmemorytostorethe
monitoreddataforextendedperiodsoftimein
caseofcommunicationfailures.
Remoteoperationanddiagnosis: Intheeventof
brownoutsorblackouts,thefieldSCADAunits
shouldnormallyself-bootuponresumptionof
powersupply.ThefieldSCADAunitsshouldalso
allowforremotediagnosis.
Call-outfeature:Thisfeatureallowsthesystems
softwaretonotifyappropriatepersonnelif
problemsdevelopwithatreatmentsystemor
waterquality.Thisfeaturecangreatlyenhance
operatorresponseinemergencysituationsand
preventcostlyshutdownsandlossofwaterand/or
waterquality.
OpenDatabaseConnectivity(ODBC): This
featureallowsforopencommunicationwithother
databasesandtoolsthatcanbeintegratedto
provideadditionalfeatures. Thedatathencan
alsobeusedfornetworkmodeling.
5-10
AReferenceGuideforUtilities
distribution system,
fieldevaluationofsensorsandremote
monitoring technologies for inclusion in the
network,
developmentofeffectivemethodstopublish
real-timedatathatenhancedconsumer
confidence,
evaluationofcostsassociatedwith
implementingsuchsystems,and
identification of potential problems and
suggestionsforremedialactionswhen
implementingremotemonitoringnetworks.
Freechlorine,pH,temperature,andturbiditywere
selectedasthemonitoredparametersbasedonthe
availabilityofonlinesensortechnologies. The
selectionwasbasedonthepremisethatthesewere
parameterswhichcouldbereliablymonitored
continuouslyandtheselectedinstrumentsrequired
limitedmaintenance. Additionally,WASAusedtheir
SCADAsystemtotrackvariousoperatingparameters
withinthedistributionsystem. Duringtheevalua-
tion,itwasclearthatuseoftheexistingSCADA
systemtomanagethemonitoreddataprovidedclear
advantagesoverotheravailablesystems. Usingthe
existingSCADAsystemminimizedlong-termon-site
supportcosts.
Aftersuitablelocation(s)wereidentified,customized
samplingandmonitoringsystemswerebuilt. The
remotemonitoringsysteminWashingtonD.C.was
implementedinthreephases. Inthefirstphase,a
remotemonitoringsystemwasinstalledattheFort
Reno#2tank(Figure5-4),whichprovidedsecurity
andeasyaccesstothedistributionsystem.Subse-
quently,basedoninitialsuccessatthislocation,two
othersites(BryantStreetandBluePlains)were
selected and added
totheremote
monitoring network
inthesecondphase.
Thethirdphase
involved the
developmentofa
Web-basedapplica-
tion to publish the
real-timedatain
ordertoenhance
consumerconfi-
dence.
Figure5-5showsthe
relationship between
theSCADAsystem
andthetransmission
ofthedata. Al-
Figure 5-4. Fort Reno #2
Remote Sampling System.
thoughWASAsSCADAsystemusedaproprietary
operatingsystem,itprovidedapersonalcomputer
(PC)linkwhichwasusedtodumpdataintoaregular
PCforfurtherprocessing. Thehardware-basedfeature
enabledtightsecurity;anauthorizedendusercould
onlycopytherelevantdatapublishedonthePCand
couldnotdirectlyaccesstheSCADAsystem.This
featurealsoeliminatedanypotentialinterference
betweenthesamplingsystemdataandotherdistribu-
tionsystemoperationsdata. Unfortunately,theEPA
fundingforthisstudywasterminatedand,asaresult,
thesystemsandtheWebsitearecurrentlynotopera-
tional. Theoverallproject,however,diddemonstrate
thatsuchsystemscouldbedevelopedandoperated.
Figure5-6showssomeoftheoutputdatafortheFort
Renotankwhichindicatesthelossofdisinfectant
chlorinelevelsatnight.Clearly,thistypeofinforma-
tioncanbeusedtoimprovesystemoperationsto
bettermaintainthewaterquality.
5.6.3 Tucson Water Monitoring Network
BasedonagrantreceivedfromtheEPAsEnvironmen-
talMonitoringforPublicAccessandCommunity
Figure 5-5. WASA Remote Monitoring System Layout and Data
Transmission Scheme.
5-12
Figure 5-6. Monitoring Data for Fort Reno Tank.
Tracking(EMPACT)Program,thecityofTucson
implementedacomprehensivewaterqualitymonitor-
ingprogram.ThecitysEMPACTgoalsincludedthe
following: implementing enhanced monitoring of the
utilitys potable distribution system, providing the
communitywithnearreal-timewaterqualityinforma-
tiononTucsonWatersWebsite
(www.cityoftucson.org/water),andcreatingcommu-
nitypartnershipstobetterinformwaterconsumers
aboutwaterqualityandresourceissues.Thewater
quality monitoring and data collection tools provided
throughEMPACTalsoenablestheutilitytotrackand
respondtoreal-timechangesinsystemwaterquality.
TucsonWatersdistributionsystemconsistsofone
centraldrinkingwaterdistributionsystemthatserves
themajorityofthecustomersandtenisolated
drinkingwaterdistributionsystems.Alleleven
drinkingwaterdistributionsystemscoveraservice
areaof300squaremilesandserve680,000customers
intheTucsonmetropolitanarea.Thetwotypesof
sourcewaterthatsupplythecentraldistribution
systemarenativegroundwaterandrenewable
rechargedsurfacewaterfromtheColoradoRiver. The
sourcewaterthatsuppliesthetenisolateddistribution
systemsisgroundwater.
Forthepurposesofmonitoring,thecentraldistribu-
tionsystemisdividedintotenwaterqualityzones
andeachisolateddistributionsystemisconsideredan
individualwaterqualityzone.Figure5-7showsthe
zonemap.Awaterqualityzoneisdefinedasanarea
ofthedistributionsystemthatissimilarinwater
qualitycharacteristics,waterpressure,geographical,
andpoliticalboundaries.Eachwaterqualityzonehas
asetnumberofdedicatedsamplingstationsand
points-of-entry(POE). Thededicatedsampling
stationsmonitorthequalityofthedrinkingwaterin
thedistributionsystembeforedeliverytothecus-
tomer.ThePOEsareusuallyindividualwellsthat
representthewaterqualityofasinglewellorinafew
AReferenceGuideforUtilities
cases,combinedPOEsystemsthatrepresentthe
collectiveblendedwaterqualityfromagroupofwells
thatdirectlysupplyTucsonsdrinkingwater.
Intotal,thereare262dedicatedsamplingstationsand
approximately154activePOEslocatedwithinthe
multipledistributionsystems.Inaddition,22online
waterqualitystations(formonitoring:chlorine
residual,totaldissolvedsolids,pH,andtemperature)
arelocatedthroughoutthecentraldistributionsystem
atstrategiclocations,suchasreservoirs,wellsites,
andboosterstations,asoneoftheprimaryobjectives
oftheEMPACTprogram.
Figure 5-7. City of Tucson Water Quality Zone Map.
Figure5-8depictsacontinuouswaterquality
monitoringstation.Themonitoringfrequencyranges
fromtri-annually(forgrabsamplelocations)toevery
60seconds(forcontinuousmonitoringstations),
depending on the location and specific monitoring
programthatisbeingutilizedforthatlocation.
Thecomprehensivewaterqualitymonitoringprogram
encompassestheentiredistributionsystem. Source
watersaremonitoredandsampledaccordingtothe
ArizonaDepartmentofWaterResourcesandthe
Figure 5-8. Continuous Water Quality Monitoring Station.
5-13
AReferenceGuideforUtilities
ArizonaDepartmentofEnvironmentalQuality
(ADEQ)regulations,whilethedrinkingwateris
monitoredaccordingtoEPAandADEQregulations
anddrinkingwaterstandards.Drinkingwaterisalso
evaluatedagainstasetofconsumer-establishedwater
qualitygoals. Special-purposesamplesaretakento
characterizeandtrackchangingtrendsinwater
qualityforbothsourcewateranddrinkingwater.All
datasetsareutilizedtotrackandmonitorchangesin
waterqualitytolearnthebaselinewaterquality
operatingparameterlevelsandalsotobeableto
identifyandreactappropriatelywhenacontamina-
tioneventoccurs.Mostoftheanalysisisconducted
bytheutilityswaterqualitylaboratoryandallthe
resultsaretrackedthroughtheWaterQualityManage-
mentDivision.
All262dedicatedsamplinglocationsaremonitored
atleastonceeachmonthfortotalcoliformand
chlorineresidual,while26otherparametersare
monitoredonceeverythreemonths. Basedonthe
waterqualitymeasurementscollectedeachmonth
fromthese262samplinglocations,thetrendsinwater
qualityconditionsaredeterminedforeachwater
qualityzoneandforthedistributionsystemasa
whole.Thisinformationcanbefoundontheafore-
mentionedWebsiteintheWaterQualitysection
underTucsonsWaterQualityandWaterQualityin
MyNeighborhoodlinks.Thewaterqualityinforma-
tiondisplayedontwointeractivemapsshowsdata
chartsandtablesforeachlocationthatissampled
undertheWaterQualityprogram.Inaddition,the
informationprovidedtoallTucsonwatercustomersin
theannualwaterqualityreportorconsumerconfi-
dencereportisbasedonPOEmonitoringdata.
5.7SummaryandConclusions
Distributionsystemmonitoringisintendedtoidentify
thespatialandtemporalvariationsinwaterquality
thattakeplaceinadrinkingwatersystem. Monitor-
ingdatacanbeusedtosatisfyvariousobjectives,
suchasregulatoryrequirements,securityrequire-
ments,orprocesscontrolrequirements. Thecostsof
implementingsuchasystemcanbestbejustifiedif
theresultingdatacanbeusedformorethanoneofthe
aforementioned objectives.
Amonitoringprogramcanimplementeitherroutine
grabsamplingorcontinuousmonitoring. Acom-
bined approach, utilizing both continuous and grab
samplingdata,mayprovetobeveryeffectiveasthe
basisforacomprehensivesystem-widemonitoring
plan. Inthepast,distributionsystemonlinemonitors
weretypicallyhousedinacontrolledenvironment
withsamplelinesfromthedistributionsystemtothe
instrument. Thisresultedinmostinstrumentation
beinglocatedatfacilitiessuchastanksandpump
stations. Theinstrumentationwassometimescon-
nectedtoaSCADAsystem,sothatresultscouldbe
communicatedtoacentraloffice.Recently,some
instrumentationhasbeendesignedforinstallationin
manholesorfordirectinsertionintowaterdistribu-
tionsystempipes.
Vulnerabilityassessmentsperformedbyutilitiesand
variousresearchstudieshaveidentifiedthatwater
distributionsystemsarevulnerabletointentionalor
accidental contamination. In addition to hardening
systemstomakeitmoredifficulttocontaminatea
system,monitoringaspartofaCWShasemergedasa
logicalapproachtocopewithpotentialcontamina-
tionevents. Monitorscanalsobeusedinadistribu-
tionsystemtoprovidereal-timeornearreal-time
informationonwaterquality.Thedatacanthenbe
usedtocontroltreatmentprocessesatatreatmentplant
orinthewaterdistributionsystem.However,thistypeof
programmaynotbepracticalforsmallsystems.
SCADAiswidelyusedinindustrialenvironmentsand
bylargerwaterutilitiestocontrolandmonitortheir
individualfacilityoperations.However,water
utilitiestypicallydonotuseavailableSCADA
systemsforconventionalwaterqualitymonitoring.
Waterutilitiestypicallymonitorwaterquality
parametersbyperforminggrabsamplingonasched-
uledorrandombasisthatprovidesaperiodicsnap-
shotoftheoverallsystem. Currentdrinkingwater
regulationsrequireallpublicwatersystemsto
implementwaterqualitymonitoringfortotalcoliform
toensurethatgoodqualitywaterisprovidedto
consumers(EPA,1996).Sincetheregulationsdonot
clearlyspecifythatreal-timemonitoringofwater
qualityisrequired,utilitieshavebeenreluctantto
installandoperatesuchdevices.
Aftertheeventsof9/11,utilitieshavebecomemore
interestedinthepotentialforcontinuouswater
qualitymonitoring. SCADAsystemscanassistinthis
function by constantly monitoring water quality
withindrinkingwaterdistributionsystems. These
systemscanpotentiallyreducetheriskofsecurity
relatedthreatsorevennon-securityrelatedthreats,
anddetectundesirablewaterqualitychangeswithina
system(Meckesetal.,1998).
Usersshouldevaluatemonitoringdataappropriatelyfor
errorsandinconsistenciesbeforecommencingactions
basedonacquireddata. Eachcomponentinamonitor-
ingsystemisapotentialsourceoferror.Forexample,a
remotemonitoringsystemcouldhavedataerrorsforone
ormoreofthefollowingreasons:instrumenterrorsand
spikes,SCADAdataerrorsrelatedtosystemfailure,
backfillingduetocommunicationfailure,timingerrors,
ormissingdata. Itisimportanttovalidatedataand
understandroutinechangesinwaterqualitydueto
system-specificoperations.
5-14
AReferenceGuideforUtilities
Monitoring equipment should be chosen appropri-
atelyafterestablishingthemonitoringrequirements.
Theindividualmonitorcharacteristics,costs,and
amenabilitytoSCADAintegrationarekeytoeffective
implementation. Eachsystemshouldbeindividually
examinedandengineeredforimplementation.
References
ASCE. Interim Voluntary Guidelines for Designing
an Online Contaminant Monitoring System. Pub-
lishedbyAmericanSocietyofCivilEngineers,
Reston,VA.2004.
Bahadur,R.,W.Samuels,W.M.Grayman,D.Amstutz,
andJ.Pickus. PipelineNet: AModelforMonitoring
IntroducedContaminantsinaDistributionSystem.
Proceedings,ASCE-EWRIWorldWater&Environ-
mentalResourcesCongress,Philadelphia,PA.2003.
Berry,J.,W.Hart,C.Phillips,andJ.Uber.AGeneral
Integer-Programming-BasedFrameworkforSensor
PlacementinMunicipalWaterNetworks.Proceed-
ings,ASCE-EWRIWorldWater&Environmental
ResourcesCongress,SaltLakeCity,UT.2004.
ClarkR.M.,W.M.Grayman,S.G.Buchberger,Y.Lee,
andD.J.Hartman. DrinkingWaterDistribution
Systems: AnOverviewinWater Supply Systems
Security. EditedbyL.W.Mays,McGraw-Hill,NY,pp
4.1-4.49. 2004a.
Clark,R.M.,S.Panguluri,andR.C.Haught. Remote
MonitoringandNetworkModels: TheirPotentialFor
ProtectingU.S.WaterSupplies,inWater Supply
Systems Security,editedbyL.W.Mays,McGraw-Hill,
NY,pp14.1-14.22.2004b.
Clark,R.M.,G.S.Rizzo,J.A.Belknap,C.Cochrane.
WaterQualityandtheReplacementandRepairof
DrinkingWaterInfrastructure:TheWashington,DC
CaseStudy. Journal of Water Supply Research and
Technology - Aqua,48(3):106-114. 1999.
EPA.Drinking Water Regulations and Health
Advisories.OfficeofWater,EPA822-B-B-96-002.
October,1996
EPA.ResponseProtocolToolbox.Overviewand
Modules1through6. Canbedownloadedat: http://
cfpub.epa.gov/safewater/watersecurity/
home.cfm?program_id=8#response_toolbox. 2003-
2004.
Themonitoringcasestudiespresentedinthischapter
demonstratethemannerinwhicheffectivemonitoring
systemscanbeimplementedinsmall,medium,and
largedistributionsystems. Ifthedataareusedfor
respondingtoacontaminationthreat,itisimportant
tounderstandthemovementofwaterinthesystem.
Goodrich,J.,J.Adams,andB.Lykins,Jr.Ultrafiltra-
tionMembraneApplicationforSmallSystem.EPA
NationalRiskManagementResearchLaboratory.
1993.
Haught,R.C.,andS.Panguluri. Selectionand
ManagementofRemoteTelemetrySystemsfor
MonitoringandOperationofSmallDrinkingWater
TreatmentPlants.Proceedings,FirstInternational
SymposiumonSafeDrinkingWaterinSmallSystems,
May10-13,1998,Washington,D.C.USA.1998.
Hargesheimer,E.,O.Conio,andJ.Popovicova
(Editors). Online Monitoring for Drinking Water
Utilities.AwwaRFCRSProAqua.2002.
InternationalLifeSciencesInstitute(ILSI). Early
Warning Monitoring to Detect Hazardous Events in
Water Supplies.ILSIPress,WashingtonD.C.1999.
Johansson,P. An Introduction to Modern Welfare
Economics. CambridgeUniversityPress,Cambridge,
UK. 1991.
Kessler,A.,A.Ostfeld,andG.Sinai.Detecting
Accidental Contaminations in Municipal Water
Networks.Journal of Water Resources Planning and
Management,ASCE.124(4):192-198. 1998.
Lee,B.,R.Deininger,andR.Clark.Locating
MonitoringStationsinWaterDistributionSystems.
Journal of AWWA,83(7):60-66. July1991.
Meckes,M.C.,J.S.Mattingly,G.J.Papadopoulos,M.
Dosani,andS.Panguluri.RealTimeWaterQuality
MonitoringofaWaterDistributionSystemUsing
RemoteTelemetry.Proceedings,AWWADistribution
SystemSymposium,Austin,Texas.September20-22,
1998.
Ostfeld,A.OptimalMonitoringStationsAllocations
forWaterDistributionSystemSecurityinWater
Supply Systems Security,editedbyL.W.Mays,
McGraw-Hill,NY,pp16.1-16.15.2004.
5-15
AReferenceGuideforUtilities
Ostfeld,A.,andE.Salomons.OptimalLayoutof
EarlyWarningDetectionStationsforWaterDistribu-
tionSystemsSecurity. Journal of Water Planning
and Resources Management,ASCE.130(5):377-385.
2004.
Panguluri,S.,R.C.Haught,M.C.Meckes,andM.
Dosani,1999. Remotewaterqualitymonitoringof
drinkingwatertreatmentsystems.Proceedings,
AWWAWaterQualityTechnologyConference,
Denver,CO.November1999.
Uber,J.,F.Shang,M.Ploycarpou,andZ.Wang.
FeedbackControlofBoosterChlorinationSystems.
AwwaRF,Denver,CO.2003.
Walski,T.M.,D.V.Chase,D.A.Savic,W.Grayman,S.
Beckwith,andE.Koelle.Advanced Water Distribu-
tion Modeling and Management. HaestadPress,
Waterbury,CT.2003.
Watson,J.,H.J.Greenberg,W.E.Hart.AMultiple-
ObjectiveAnalysisofSensorPlacementOptimization
inWaterNetworks.Proceedings,ASCE-EWRIWorld
Water&EnvironmentalResourcesCongress,Salt
LakeCity,UT.2004.
5-16
A Reference Guide for Utilities
Chapter6
GeospatialTechnologyfor
WaterDistributionSystems
Section 6.1 provides a brief summary of the history and development of geospatial data management
based on information extracted from many sources. This history is included in order to provide a context
for the current geospatial data management methodologies in use today by utilities. Readers who are
already familiar with the history may choose to skip Section 6.1 of this chapter.
Geospatialdataidentifiesthegeographiclocationand
characteristicsofnaturalorconstructedfeaturesand
boundariesontheearth.Thisinformationmaybe
derivedfromvarioussourcesofdata,including
remotesensing,mapping,andsurveyingtechnolo-
gies. Moresimply,geospatialdataisanyinformation
inorontheearththathasawherecomponent. This
canbeahouseaddress,astreetintersectiononamap,
apumpstationwithacoordinatelocationstoredina
facilitieslist,orthelocationofthesamplingtapona
diagramofapumpstation. Thus,everyobjecthasa
geospatialdatacomponentbasedonitslocation.
Geospatialdataprovidesamechanismforincorporat-
inggeographiclocationsofvariousfunctionsand
facilitiesinadistributionsystemsanalysis. Thecost
ofincorporatingmapdataintothewaterdistribution
systemsdisciplineisdecreasing,whichenablesa
wideraudienceofuserstoperformpowerfulspatial
analysesovertime,suchasmasterplandevelopment,
pipebreakanalysis,andlocationalinformationon
sensitivesubpopulations(e.g.,nursinghomes,
schools). Asthesetoolsanddatasetsbecomemore
commonlyusedandsharedamongengineers,new
efficiencieswillberealizedthatwillhaveapositive
impactonwaterdistributionsystemmanagement.
Watersystemsarebynaturequitegeographically
extensiveandthelocationofaparticularcomponent
orfeaturemaysignificantlyaffectitsperformance.
Sourcewatershedscancoverhundredsorthousandsof
squaremiles. Similarly,distributionsystemscan
covervastareas. Theoperationofawatersystem
entailsmovingwaterfromonelocationtoanother.
Elevation(Z),thethirddimensionoflocation(along
withtheXandYdimensionsofaCartesiancoordi-
natesystem),isanimportantfactorindesigningand
operatingawatersystem. Thisillustratesthatthe
managementofawatersystemisinherentlya
geospatial issue.
Becauseofthespatialnatureofwatersystems,many
aspectsofmanagingawatersystemconsistofusing,
managing,anddisplayinggeospatialdata. Thishas
ledtoavarietyofmechanismsrangingfrommapsand
plans to sophisticated, computerized database
managementsystems.Thefollowingisalistofsome
ofthecomputerizeddatasystemsthatwaterutilities
typicallyuseformanagingtheirspatialdata. These
systemswillbediscussedingreaterdetaillaterinthis
chapter.
GISGeographicInformationSystem.
CADDComputer-AidedDesignandDrafting.
AM/FMAutomatedMapping(orAsset
Management)/Facilities Management.
CISCustomerInformationSystem.
G
IS
C
A
D
D
GPS
SC
A
D
A
G
IS
C
A
D
D
GPS
SC
A
D
A
A
M
/F
M
CIS
D
E
M
A
M
/F
M
CIS
D
E
M
DEMDigitalElevationModel.
GPSGlobalPositioningSystem.
SCADASupervisoryControlAndData
Acquisition.
LIMSLaboratoryInformationManagement
System.
LISLandInformationSystem.
RDBMSRelationalDatabaseManagement
System.
SDMSSpatialDataManagementSystem.
6-1
6 links 3 nodes
Figure 6-8b. Typical Representation of a Pipe Section in a
Network Model.
6.4.2IssuesinIntegratingGISandWater
DistributionSystemModels
Asanevolvingtechnology,therearestillissuesin
trulyintegratingGIStechnologywithwaterdistribu-
tionsystemmodels. Theseissuesprimarilyrevolve
aroundthelevelofdetailrequiredinthetwosystems
andtheproceduresforupdatingthemodelandthe
database.
Inmostcases,awaterutilityGISisusedformany
purposes including mapping, facility management,
planning,andmodelingsupport. Asaresult,there
maybeagreatdealofdetailintheGIS. Forexample,
itmayincludehydrants,shutoffvalves,watermeters
andhouseholdconnections,airreleasevalves,and
otherappurtenances. Ontheotherhand,typically,
waterdistributionsystemmodelsdonotexplicitly
includemanyofthesecomponents. Thisisshown
graphicallyinFigure6-8. Inthiscase,onlyjunction
nodesandpipesareincludedinthemodelrepresenta-
tion. Asaresult,theGISrepresentationincludes17
linksand11nodes,andthemodelrepresentation
includes6linksand3nodes. Thedisparitybetween
thetworepresentationscanincreasebyanotherorder
ofmagnitudeifwatermetersandcustomerconnec-
tionsareincludedintheGIS. Variousapproachesare
takentodealwiththissituation.
Averydetailedmodelisconstructedthat
includesalloftheelementsintheGIS.This
solutioncanresultinaverylargemodelwithan
excessivenumberofnodesandlinks.
TheGISrepresentationgoesthrougha
consolidation (skeletonization) procedure to
eliminateunneedednodesandtoaggregatethe
resultinglinksinordertoconstructthemodel
representation.Thoughthisresultsinamore
appropriatemodel,itaddsanintermediatestep
betweentheGISandthemodel.Additionally,
aftertheconsolidationprocess,thereisno
longeraone-to-onecorrespondencebetween
GISandmodelfeatures. Thislackof
correspondenceleadstoissuesrelatedtostoring
modeloutputintheGISandupdatingthemodel.
Multiple representations are maintained within
theGISfordifferentuses. Thedetailed
representationisthecomplete,basecaseand
usedforfacilitymanagementwhilethe
skeletonizedversionisusedformodeling.This
approachhasthelimitationthatrequires
changesininformationtobemadeinmultiple
databases.
Thebasiclink-nodenetwork(asusedinthe
model)ismaintainedasthebasecaseintheGIS
andassociatesothercomponents(suchas
hydrants)withlinksratherthanstructurally
embeddingtheminthenetwork.
A Reference Guide for Utilities
17 links 11 nodes
Figure 6-8a. Typical Representation of a Pipe Section in GIS.
Inanyoftheoptionsdescribedabove,proceduresfor
updatingtheGISareessential.Therearemanyissues
associatedwithGISupdatingsuchasauthorizationof
specificuserstomakechanges,natureofthechanges
aspermanentorpartofawhat-ifmodelingscenario,
andfrequencyofreplicationofthetwodatabasesif
separatemodelandGISfilesaremaintained. Allof
theseshouldbecarefullyspelledoutpriortodesign-
ingandimplementingaGIS.
6.5 Use of GIS in Water Utilities -
Case Studies
ThissectionpresentsthepotentialusesofGISinthe
waterutilityindustry.CasestudiesfromtheLasVegas
ValleyWaterDistrict(LVVWD)andDenverWaterare
presented.
6.5.1UseofGISatLVVWD
Between1989and2004,LasVegasgrewfasterthan
anyothermetropolitanareaintheU.S. Asaresult,
LVVWDhasmorethandoubleditsservicearea
populationduringthisperiod. In1989,theservice
areapopulationwas558,000andin2004itroseto
1,209,000,representinganincreaseof651,000people
servicedbyLVVWD(JacobsenandKamojjala,2005).
Figure6-9isaGISrepresentationoftheLVVWD
distributionsystemgrowthbetween1989and2004.
Toaddressavarietyofissuesrelatedtothisrapid
growth,LVVWDintegratedthefunctionsofmaster
planning, operational planning, and development
reviewbyintegratingitsGISdatawithmodeling,
SCADA,andenterprisedata(suchasCIS,AM/FMand
LIMS). Figure6-10presentstheconceptualrelation-
shipmodelofthesefunctionsandpotentialintegra-
tionbenefits(Jacobsenetal.,2005).
Duringtheprocessofintegration,LVVWDdeveloped
aone-to-onerelationshipbetweentheGISspatialdata
anditsnetworkmodel(JacobsenandKamojjala,
2005). Anexampleofthisone-to-onerelationshipis
6-11
A Reference Guide for Utilities
allocationaspartofthisanalysis. Specifically,
existingdemandswerespatiallyallocatedusing
addressgeo-coding(Strasseretal.,2000).
DenverWaterhasbeenacompletelymetered
systemsince1992. Historicconsumptiondatais
availablefrom1993tothepresent. Consumption
datawasextractedfromthebillingsystem,and
importedintoanMSAccessdatabaseforusein
varioustasksincludingthetreatedwaterstudy.
Theextracteddataincludedbothcustomeraddress
andcustomerclassinformation(e.g.,singlefamily,
multi-family,commercial,industrial,andpublic).
Thecustomerclassinformationallowedthe
demandinformationtobeaggregatedbycustomer
class. Becausetheserviceareawaslarge(over250
squaremiles),itwasimportanttoidentifythe
locationoftheconsumptiondemandpoints. This
iswheretheuseofGISbecameveryimportant
(Strasseretal.,2000).
DenverWaterusedtheaddressgeo-codingfeature
availablewithinArcInfowhichallowedforeach
customerordemandpointtobeidentifiedonabase
map. Usingthisprocess,adotisplacedonthebase
maprepresentingeachcustomerthatcouldbe
positivelygeo-coded. Thematchratewasover93
percent. Thoseaccountsthatcouldnotbegeo-coded,
mostlylargeaccountsrepresentingmastermeter
accounts,andwholesalecustomers,wereenteredin
thesystemmanually.Resultsfromthegeo-coding
processforonepressurezoneareillustratedinFigure
6-18. Aqualitycontrolcheckwasperformedonthe
resultsofthisgeo-codingprocessbyreconciling
consumerdemandswithDenverWatersannual
statisticalreport(Strasseretal.,2000).
A Reference Guide for Utilities
6.6 Summary
Useandmanagementofgeospatialdataisanimpor-
tantaspectofthedesignandoperationofwater
systems. Thiscanbeaccomplishedthrougharangeof
systemsundertheoverallumbrellaofSDMSutilized
tocollect,store,andusethespatialdata. This
umbrellacoversnotonlythebroadtopicsofGISand
CADDthatarewidelyrecognizedasgeospatialdata
systems,butalsosystemssuchasSCADAandLIMS
thathaveaspatialcomponentassociatedwithall
data.
Theareaofspatialdatabasemanagementiscontinu-
ingtoevolvewithinthewaterindustry.Justasthe
capabilities of the various individual components
withintheSDMSumbrellacontinuetoexpand,the
integrationofthevarioussystemsisanactiveareaof
development. Waterdistributionsystemanalysisisa
significantbeneficiaryoftheseimprovementsand
integration. Asaresult,modelscanbebuiltmore
quicklyandingreaterdetail. Informationonfacili-
tiesanddemandscanberoutinelyupdated. The
resultsofamodelapplicationcanberapidlydis-
playedandviewedalongwithotherspatialdata.The
prospectofreal-timeapplicationofmodelstoassistin
systemoperationunderroutineconditionsorunder
emergencyconditionsisgettingcloser.
Figure 6-18. GIS Geo-coding - Metered Sales Demand
Allocation Procedure.
6-17
AReferenceGuideforUtilities
7.1.3 EPA Field Study
OnJanuary14,1994,anEPAfieldteam,inconjunc-
tionwiththeCDCandtheStateofMissouri,initiated
afieldinvestigationthatincludedasanitarysurvey
andmicrobiologicalanalysesofsamplescollectedon
site. Asystemevaluationwasalsoconductedin
whichEPANETwasusedtodevelopvariousscenarios
toexplainpossiblecontaminanttransportinthe
Gideonsystem. PriortotheGideonoutbreak,a
similarwaterbornediseaseoutbreakinCabool,
Missouri,andsubsequentadvancementsinwater
qualitymodelingfirmlyestablishedtheuseofwater
qualitymodelstoanalyzesuchevents.
Thekeyanalysiswasfocusedonaflushingprogram
conductedearlierbytheutilityinresponsetotaste
andodorcomplaints. Asequentialflushingprogram
wasconductedonNovember10,1993,involvingall
50hydrantsinthesystem. Theflushingprogramwas
startedinthemorningandcontinuedthroughthe
entireday.Eachhydrantwasflushedfor15minutes
atanapproximaterateof750gallonsperminute
(gpm). Itwasobservedthatthepumpatoneofthe
wellswasoperatingatfullcapacityduringthe
flushingprogram(approximately12hours),which
wouldindicatethatthemunicipaltanksweredis-
charging during this period.
Duringtheevaluation,itwashypothesizedthatthe
tasteandodorproblemsmayhaveresultedfroma
thermalinversionthathadtakenplaceduetoasharp
temperaturedroppriortothedayofthecomplaint. If
stagnantorcontaminatedwaterwerefloatingonthe
topofatank,athermalinversioncouldhavecaused
thiswatertobemixedthroughoutthetankandtobe
dischargedintothesystemresultingintasteandodor
complaints(Fenneletal.,1974). Asaconsequence,
the utility initiated the aforementioned city-wide
flushingprogram. Turbulenceinthetankfromthe
flushingprogramcouldhavestirredupthetank
sedimentsthatweresubsequentlytransportedintothe
distributionsystem. Itislikelythatthebulkwater
and/orthesedimentswerecontaminatedwithSalmo-
nellaserovarTyphimurium. DuringtheEPAfield
visit,alargenumberofpigeons(birddroppingsare
knowntocontainSalmonella)wereobservedroosting
ontheroofofthe100,000galmunicipaltank.
7.1.4 Distribution System Evaluation
TheEPAstudyteamevaluatedtheeffectsofdistribu-
tionsystemdesignandoperation,demand,and
hydrauliccharacteristicsonthepossiblepropagation
ofcontaminantsinthesystem. Giventheevidence
fromthelabsamplesandtheresultsfromthevalve
inspectionoftheprivatetank,itwasconcludedthat
themostlikelycontaminationsourcewasbird
droppingsinthelargemunicipaltank. Therefore,the
analysisconcentratedonpropagationofwaterfrom
In1991,ajointworkshopsponsoredbytheEPAand
AwwaRFrecommendedtheapplicationofwaterquality
modelingtechniquestoevaluatewaterbornedisease
outbreaks. Thefirstopportunitytoattemptthistypeof
applicationaroseasaresultofanoutbreakthatoccurred
betweenDecember15,1989,andJanuary20,1990,in
Cabool,Missouri,population2,090(Geldreichetal.,
1992). Duringtheoutbreak,residentsandvisitorsto
Caboolexperienced243casesofdiarrhea(85bloody)
andsixdeaths. Theillnessanddeathswereattributedto
thepathogenicagentE. coli. serotypeO157:H7. Atthe
timeoftheoutbreak,thewatersourcewasuntreated
groundwater.Shortlyaftertheoutbreakwasidentified,
EPAwasinvitedtosendateamtoconductaresearch
studywiththegoalofdeterminingtheunderlyingcause
oftheoutbreak.
Exceptionallycoldweatherpriortotheoutbreak
contributedtotwomajorwatersystemlinebreaksand43
watermeterreplacementsthroughoutthecityarea. The
sewagecollectionlinesinCaboolwerelocated(forthe
mostpart)awayfromthedrinkingwaterdistribution
linesbutdidcrossorwereneartowaterlinesinseveral
locations. Atthetimeoftheoutbreak,stormwater
drainedviaopenditchesalongthesidesofthestreets
androads. Duringheavyrainfalls,sewagewasobserved
tooverflowmanholecovers,andtooverflowstreetsin
severallocations,parkinglotsandresidentialfounda-
tions.
TheDynamicWaterQualityModel(DWQM),developed
byEPA,wasappliedtoexaminethemovementofwater
andcontaminantsinthesystem(Graymanetal.,1988).
Steady-statescenarioswereexamined,andadynamic
analysisofthemovementofwaterandcontaminants
associatedwithmeterreplacementandthelinebreaks
wasconducted.Typicaldemandpatternsweredeveloped
fromavailablemeterusageforeachserviceconnection,
anditwasfoundthatthewaterdemandwas65percentof
theaveragewellproduction,indicatinginaccurate
meters,un-metereduses,andahighwaterlossinthe
system.
Themodelingeffortrevealedthepatternofillness
occurrencewasconsistentwithwatermovementpatterns
inthedistributionsystemassumingtwowaterline
breaks. Itwasconcludedthatsomedisturbanceinthe
system,possiblythetwolinebreaksor43meterreplace-
ments,allowedcontaminationtoenterthewatersystem.
Analysisshowedthesimulatedcontaminantmovement
covered85percentoftheinfectedpopulation.
TheapplicationofDWQMprovedtobeavitalstepin
completingtheanalysisoftheoutbreak. Thenext
opportunitytoapplywatermodelingtechniquesoc-
curredin1994asaresultofawaterborneoutbreakin
Gideon,Missouri(Clarketal.,1996). Intheintervening
period,EPAhaddevelopedEPANETandGideon
provided an opportunity to test its application.
7-2
thelargemunicipaltankinconjunctionwiththe
flushingprogram. Otherpossiblesourcesofcontami-
nation,suchascrossconnectionswerealsostudied.
Thesystemlayout,demandinformation,pump
characteristiccurves,tankgeometry,flushingpro-
gram,andotherinformationneededforthemodeling
effortwereobtainedfrommapsanddemographic
informationandnumerousdiscussionswithconsult-
ingengineersandcityandMDNRofficials.EPANET
wasusedtoconductthecontaminantpropagation
study(Rossmanetal.,1994).
TheEPANETnetworkmodelwascalibratedby
simulatingflushingatthehydrantsassuminga
dischargeof750gpmfor15minutes. TheCfactors
(piperoughnessseeChapter4)wereadjusteduntil
theheadlossinthemodelmatchedheadlosses
observedinthefield. Afterthecalibration,the
hydraulicmodelwassimulatedfor48hours. Thereaf-
ter,theflushingprogramwassimulatedstartingat8a.m.
onday3,bysequentiallyimposinga750gpm
demandoneachhydrantfor15minutes. Utilizingthe
TRACEoptioninEPANET,thepercentagesofwater
frombothmunicipaltankswerecalculatedateach
nodeoveraperiodof72hours.
Duringthesimulationoftheflushingprogram,the
pumpatoneofthewellswasoperated(asprevi-
ouslyobserved)atfullcapacity,whichwasover
800gpm,andthenrevertedtocyclicoperation.
Thesimulationresultsshowedthatthetank
elevation fluctuated for both municipal tanks, and
both the tanks discharged during the flushing
program. Attheendoftheflushingperiod,nearly
25percentofthewaterfromthelargemunicipal
tankpassedthroughthesmallmunicipaltank
whereitwasagaindischargedintothesystem. The
modelpredicteddramaticpressuredropsduringthe
flushingprogram. Basedontheinformation
available,itwasfeltthatthesemodelingresults
replicated the conditions that existed during the
flushingprogramcloselyenoughtoprovideabasis
forananalysisofwatermovementinthesystem.
Datafromthesimulationstudy,themicrobiological
surveillancedata,andtheoutbreakdatawereutilized
toprovideinsightintothenatureofbothgeneral
contaminationproblemsinthesystemandthe
outbreakitself. Thewatermovementpatternsshowed
themajorityofthecollectedsamplesthatweretotal
coliformandfecalcoliform(FC)-positiveoccurredat
pointswithinthezoneofinfluenceofthesmalland
largetanks. Duringboththeflushingprogramandfor
largepartsofnormaloperation,theseareaswere
predominatelyservedbytankwater,whichconfirmed
thebeliefthatthetanksarethesourceofthefecal
contaminationsincetherewerepositiveFCsamples
AReferenceGuideforUtilities
priortochlorination. Figure7-1showsthecompari-
sonofearlyconfirmedcasesofSalmonellapositive
sampleversustheestimateddistributionoftankwater
duringthefirstsixhoursoftheflushingprogram.
Figure 7-1. Comparison of Early Confirmed Cases of
Salmonella Positive Sample Versus the Estimated
Distribution of Tank Water During the First 6 Hours
of the Flushing Program.
20% or more of Small Tank water
20% or more of Large Tank water
Gideon Schools reflects increase in absentee level
Hydrant with confirmed Salmonella
Residences with confirmed case
Homes called as part of CDC survey
7.1.5 Case Study Summary and Conclusions
DatafromtheCDCsurveyoftheoutbreak,incombi-
nationwiththeEPANETsimulatedwatermovement,
wereutilizedtoestablishthepossiblesourceof
contamination. AnoverlayoftheCDCdataonthe
watermovementsimulationsshowedthattheareas
servedbythesmallandlargetanks(duringthefirst
sixhoursoftheflushingperiod)coincidedwiththe
earliestrecordedinfectiouscases. Furthermore,the
earliestrecordedcasesandthepositiveSalmonella
hydrantsamplewerefoundintheareathatwas
primarilyservedbythelargetank,butoutsidethe
smalltanksareaofinfluence.
Theinvestigatorsconcludedthatduringthefirstsix
hoursoftheflushingperiod,thewaterthatreachedan
infectedresidentandtheGideonSchool(theearliest
reportedinfections)wasalmosttotallyfromthelarge
tank. Basedontheresultsofthestudy,itappeared
thatthecontaminationhadbeenoccurringovera
7-3
AReferenceGuideforUtilities
periodoftime,whichisconsistentwiththepossibil-
ityofbirdcontamination. Itislikelythatthecon-
taminantwaspulledthroughthesystemduringthe
flushingprogram. TheapplicationofEPANETtothe
outbreakprovedtobeavitalpartofthestudy.
7.2Reconstructing Historical
ContaminationEvents-Dover
Township(TomsRiver),NJ
Thiscasestudyisfocusedonevaluatingthedistribu-
tionofchemicallycontaminatedsourcewaterina
distribution system. The supporting investigations
forthiscasestudywereprimarilysponsoredbythe
ATSDR.Theinvestigationsinvolvedseveralother
organizations. Themajorcontributorsincludedthe
NewJerseyDepartmentofHealthandSeniorServices
(NJDHSS),theMultimediaEnvironmentalSimula-
tionsLaboratoryattheGeorgiaInstituteofTechnol-
ogy,EPAsNationalRiskManagementResearch
Laboratory,andtheU.S.GeologicalSurvey.
Key Phrases to Characterize Case Study:historical
reconstruction,hydraulicandwaterqualitymodeling,
exposuremodeling,contaminationassessment,source
tracing,sourcecontribution,modelcalibration,
sensitivity analysis, genetic algorithm.
7.2.1 Case Study Overview
InAugust1995,respondingtoanevaluationre-
questedbytheATSDR,theNewJerseyDepartmentof
Health(nowNJDHSS)determinedthatthechildhood
cancerincidencerateinDoverTownship(andthe
TomsRiversection)washigherthanexpectedforall
malignantcancerscombined(brainandcentral
nervoussystemcancer,andleukemia,Berry,1995).In
March1996,NJDHSSandATSDRdevelopedaPublic
HealthResponsePlan(PHRP)describingactions
theseagencieswouldtaketoinvestigatetheunex-
pectedincreaseinchildhoodcancersandenvironmen-
talconcernsinDoverTownship(NJDHSSand
ATSDR,1996).ThePHRPincludedalistofseveral
evaluations. Oneofthekeyevaluationswasto
identify potential environmental exposure pathways
relativetotwoNationalPrioritiesList(NPL)sitesin
DoverTownship(Figure7-2)Ciba-GeigyandReich
Farm. Figure7-2alsoshowsthetwopublicwater
supplywellfields(ParkwayandHolly)thatwere
identifiedaspotentialroutesofexposure. Thesewell
fieldsarenotonlylocatedinthevicinityofthe
aforementionedNPLsites,butarealsoinareaswhere
thestatisticallyhigherchildhoodcancerrateswere
established.
Theensuingevaluationsrevealedthepresenceofa
previously unidentified compound, styrene acryloni-
trile(SAN),inthegroundwaterfromtheParkwaywell-
fieldthatcouldbetracedtotheReichFarmNPLsite.
Figure 7-2. Investigation Area, Dover Township, Ocean
County, NJ (modified from Maslia et al., 2001).
Similarly,asearchofhistoricalrecordsrevealed
contamination (primarily semivolatile organics
[SVOCs])oftheHollywellfieldsthatcouldbe
tracedtotheCiba-GeigyNPLsite. Furthermore,
oneofthehypothesesfortheepidemiologiccase-
controlstudywasthatthehighercancerincident
ratewasrelatedtothehigherexposuretopublic
water supplies with documented contamination (the
ParkwayandHollywellfields).ToassistNJDHSS
withthecontaminateddrinkingwaterexposure
assessmentcomponentoftheepidemiologicstudy,
ATSDRdevelopedawaterdistributionmodelfor
thestudyareausingtheEPANETsoftware.This
networkmodelwasusedtosimulatehistorical
characteristicsofthewaterdistributionsystem
servingDoverTownshipfrom19621996.Because
therewasalackofhistoricalcontaminant-specific
dataduringmostoftheperiodrelevanttothe
epidemiologicstudy,themodelingeffortfocused
onestimatingthepercentageofwaterthatastudy
subjectmighthavereceivedfromeachwellthat
suppliedwatertotheimpactedarea. Thefollowing
subsectionspresentabriefoverviewofthewater
distribution modeling effort (both hydraulic and
waterquality)followedbyasummaryoffindings
and conclusions.
7-4
PriortotheATSDRsanalysisofwellfieldcontamina-
tioninDoverTownshipandthepotentiallinkagesto
childhooddiseases,anotherstudyinWoburn,Massa-
chusetts,heraldedtheeraofsuchanalyses. Though
boththemodelandgraphicalpresentationsareprimi-
tivebytodaysstandards,theywereeffectiveinprovid-
ingaquantitativebasisforassessingthespreadof
contaminantsinthedistributionsystem. Thefollowing
isabriefdescriptionoftheWoburnanalysis.
InMay1979,theMassachusettsDepartmentofEnviron-
mentalQualityEngineeringdiscoveredthattwowells
(WellsG&HintheBZoneSeeFigure7-3)in
Woburn,Massachusettswerecontaminatedwithtoxic
chemicals. Subsequentanalysisshowedthatpartsofthe
city experienced elevated levels of childhood leukemia
andotherillnessesattributedtodrinkingwaterderived
fromthesewells. Thiseventresultedinlegalaction,a
diversesetofscientificstudiesthatarestillongoing,
andthepublicationofabookentitledA Civil Action
(Harr,1995).Earlysteady-statedistributionsystem
hydraulicandwaterqualitymodelswerealsousedasa
meanstotrackthemovementofthecontaminatedwater
inthedistributionsystemunderarangeofoperating
anddemandconditions(Murphy,1986).Theaccompa-
nyingfigureisoneexampleofaplotresultingfromthis
earlymodelingeffort. AsshowninFigure7-3,basedon
themodeling,thecitywasdividedintothreezonesfor
eachscenariotheAzonethatreceivednowaterfrom
thecontaminatedwells,theBzonethatreceivedallof
itswaterfromthecontaminatedwells,andtheCzone
thatreceivedsomeofitswaterfromthecontaminated
wells.
Figure 7-3. Distribution System Zones Woburn, MA
(May 1969).
AReferenceGuideforUtilities
7.2.2 Overall Modeling Approach
Becauseofthelackofhistoricalhydraulicandwater
qualityinformation,thewaterdistributionsystemwas
characterizedusingdatagatheredduringanextensive
fieldinvestigationin1998.The1998fieldinvestiga-
tionconsistedoftwocomponents:(1)determining
spatiallocationsofdistributionsystemfacilities
(wells,tanks,pump,andhydrants)and(2)equipping
hydrants with continuous-recording digital data
loggersandmonitoringsupplysources(wells,pumps,
andtanks)tomeasuresystemresponsesduringwinter
demand(March)andsummerdemand(August)
periods. Twenty-five hydrants located throughout the
distributionsystemwereequippedwithdataloggers
tosimultaneouslycollectinformationonsystem
response(Masliaetal.,2000). Thecollectedresponse
dataincludedon-offcyclingofgroundwaterwells,
highserviceandboosterpumpoperations,pressure
variations,storagetankwater-levelfluctuations,and
total production.
Adetailedall-pipehydraulicnetworkmodelwas
developed and calibrated to present-day conditions
(1998)usingthefieldinvestigationresults.The
reliabilityofthecalibratedmodelwassuccessfully
demonstratedthroughawaterqualitysimulationof
thetransportofanaturallyoccurringconservative
element(barium)andacomparisonoftheresultswith
datacollectedinMarchandApril1996at21schools
and6pointsofentrytothewaterdistributionsystem.
Thereafter,todescribethehistoricaldistribution
systemnetworksspecifictotheDoverTownshiparea,
databasesweredevelopedfromdiversesourcesof
information.Thesedatasourcesincludedwaterutility
pipelineinstallationrecords,quarterlybillingrecords,
NJDHSSgroundwaterwellrecords,andannualwater
utilityreportstothestateboardofpublicutilities.
ThesedatawereappliedtoEPANETandsimulations
wereconductedforeachmonthofthehistorical
periodJanuary1962throughDecember1996(420
simulations). Aftercompletingthose35-year/420-
monthanalyses,source-traceanalysissimulations
wereconductedtodeterminethepercentageofwater
contributedbyeachwellorwellfieldoperating
duringeachmonthforallstudysubjectlocations.
Areviewofthehistoricalnetworkconfiguration
revealedthatthewaterdistributionsystemcomplex-
ityincreasedsignificantlyduringthisperiod. The
modelinputswereappropriatelyadjustedtoaccount
forthesehistoricalchanges. Forexample,the1962
waterdistributionsystemwasrepresentedwithan
approximatepeakproductionof1.3million
gallonsperday(MGD)producedfromthreewellsthat
servednearly4,300customers(population~17,200).
Bycontrast,in1996,thewaterdistributionsystem
hadanapproximatepeakproductionof13.9MGD
producedfrom20wellsthatservednearly44,000
7-5
AReferenceGuideforUtilities
customers(population~89,300). Appropriate
adjustmentsweremadetomodeledpipesegments,
storagereservoirs,andoperationaldetails.Graymanet
al.(2004)presentamoredetailed
accountoftheEPANETmodel
inputadjustments. Produc-
tiondataforthe420-
month historical
periodisgraphi-
cally represented
inFigure7-4.
Figure 7-4. Three-Dimensional Representation of Monthly
Production of Water, Dover Township Area, NJ (from
Maslia et al., 2001).
Toperformanextendedperiodsimulation(EPS)of
thedistributionofwaterforeachofthe420monthsof
thehistoricalperiod,informationwasrequiredon
networkconfiguration,demand,andoperationaldata.
However,operationaldatapriorto1978wereunavail-
able,requiringthedevelopmentofsystemoperation
parametersdesignatedasmasteroperatingcriteria
(MOC).TheMOCisbasedonhydraulicengineering
principlesnecessarytosuccessfullyoperatedistribu-
tionsystemssimilartotheoneservingtheDover
Townshiparea(Table7-1).From1978forward,for
selectedyears,operatorsofthewaterutilityprovided
informationonthegeneralizedoperatingpracticesfor
atypicalpeak-demand(summer)andnon-peak
demand(fall)day.Theseguidelineswereusedin
conjunctionwiththeMOCtosimulateatypical24-
hourdailyoperationofthewaterdistributionsystem
foreachmonthofthehistoricalperiod.
Themodelparameterofinterestfromtheepidemio-
logicstudyperspectivewastheproportionate
contributionofwaterfromwellsandwellfieldsto
locations throughout the historical pipeline networks.
Thus,thedistributionofwaterdeliveredtopipeline
locationswastheitemofinterestratherthanthe
specificoperationsofthewells,storagetanks,and
pumps(WSTP)thatdeliveredthewater.Normally,
detailedWSTPoperationalinputswouldberequired
forEPANETsimulation.However,tosimplifythe
simulationmethodologyandreducedatarequire-
ments,asupply-node-link(SNL)methodof
idealizingtheWSTPcombinationwasdeveloped. In
theSNLsimulationmethod,anequivalentamountof
waterissuppliedtothedistributionsystem(basedon
estimatedmonthlydemandsandthetypicaldaily
operationofthesystems). Todemonstratethatthe
surrogateSNLsimulationmethodsuppliesthe
distributionsystemwithanequivalentamountof
waterwhencomparedtothereal-worldWSTP
simulationmethod,bothsimulationmethodswere
appliedtothepresent-day(1998)waterdistribution
systemforconditionsexistinginAugust1998.The
resultsobtainedfromthesesimulationsproduced
nearlyidenticalflowsinthemodeledsystem.
7.2.3 Simulation Techniques
UsingtheEPANETnetworkmodeldevelopedforthe
DoverTownshiparea,hydraulicmodelingwas
conductedwherebyaveragenetworkconditionswere
simulatedforeverymonthofthehistoricalperiod
Table 7-1. Master Operating Criteria Used to Develop
Operating Schedules for the Historical Water Distribution
System, Dover Township Area, NJ (from Maslia et al.,
2001)
Parameter Criteria
Pressure
1
Minimumof15psi;maximumof110
psiatpipelinelocations,including
networkendpoints
Waterlevel Minimumof3ftabovebottom
elevationoftank;maximumequalto
elevationoftopoftank;ending
waterlevelshouldequalthestarting
waterlevel
Hydraulicdevice
on-linedate
June1ofyearinstalledtomeet
maximum-demandconditions
On-and-offcycling:
Manualoperation
Wellsandhigh-serviceandbooster
pumpscannotbecycledon-and-off
from2200to0600hours
On-and-offcycling:
Automatic
operation
Wellsandhigh-serviceandbooster
pumpscanbecycledon-and-offat
anyhour
Operatinghours Wellsshouldbeoperated
continuouslyforthetotalnumberof
productionhours,basedon
productiondata
2
1
Generally,forresidentialdemand,minimumrecommended
pressureisabout20psi.However,forsomelocationsin
theDoverTownshiparea(mostlyinareasneartheendof
distributionlines),lowerpressuresweresimulated.
2
SeeMasliaetal.(2001)forproductiondata(AppendixB)
andhoursofoperation(AppendixD)
7-6
(420simulations).Thesesimulationswerecompleted
underbalancedflowconditionsthatutilizedhydrau-
licengineeringprinciplesandconformedtotheMOC
(Table7-1).Thereafter,usingtheresultsofthe
monthlynetworkhydraulicsimulations,waterquality
simulations(source-traceanalysis)wereconductedfor
eachwatersource(pointofentry)ofthenetworkin
ordertodeterminethemonthlyproportionate
contributionofsourcewateratalllocationsinthe
DoverTownshipareaservicedbythewaterdistribu-
tionsystem.
EPANETisadynamicwaterqualitymodelthathas
theabilitytocomputethepercentageofwater
reachinganypointinthedistributionsystemover
timefromaspecifiedlocation(source)inthe
network.Toestimatethisproportionatecontribu-
tionofwater,asourcelocationisassignedavalue
of100percent.Theresultingsolutionprovidedby
thewaterqualitysimulatorinEPANETthen
becomesthepercentageofflowatanylocationin
thedistributionsystemnetwork(forexample,a
demandnode)contributedbythesourcelocationof
interest. Forthepurposesofthisanalysis,asource-
traceanalysiswasconductedforeverymonthofthe
historicalperiod.Sourcenodeswereassigneda
valueof100percentinordertoestimatethe
proportionate contribution of water to locations in
thehistoricaldistributionsystemnetworks. These
initial conditions were fully propagated through
mostofthedistributionsystembeforeretrieving
theproportionatecontributionresults(Masliaetal.,
2000). Accordingly,foreachmonthlyhistorical
networkmodel,24-hourdemandandoperational
patternsweredefinedandthesepatternswere
repeatedforapproximately1,200hourstoreacha
stateofstationarywater-qualitydynamics(dynamic
equilibrium). Formostoftheanalyses,hydraulic
timestepsof1hourandwater-qualitytimestepsof
5minuteswereusedwithinEPANET.Forsome
monthlysimulations,thewater-qualitytimesteps
werereducedto1minutetoensurethatthemass
balancesummedto~100percent(rangeof98to
101percentduetonumericalapproximations).
Withrespecttotheschedulingofgroundwaterwell
operations,theEPANETmodelwassettoutilize
patternfactorscorrespondingtothehourlyoperations
ofsupplywells.Thesepatternfactorsalongwiththe
operationalextremesofstoragetankwaterlevelswere
manuallyadjustedduringeachofthe420monthly
networksimulationstoachievebalancedflow
conditions.Thisapproachtosimulationwasdesig-
natedasthemanualadjustmentprocess. Asecond
simulation approach designated as the genetic
algorithm(GA)approachwasalsoutilizedtoachieve
balancedflowconditionsforeachofthe420monthly
networksofthehistoricalperiod.Thisapproach
AReferenceGuideforUtilities
requiredthedevelopmentofaninnovativemethodol-
ogyknownastheprogressiveoptimalitygenetic
algorithm(POGA),whichisanautomatedobjective
simulationtechnique(Araletal.,2004a,b).TheGA
simulations utilized the balanced flow conditions
obtainedbythemanualadjustmentprocessasstarting
conditions. TheGAtechniquewasusedtoaddress
the following key questions:
Ifabalancedflowoperatingconditionwas
achievedusingthemanualadjustmentprocess,
wastheresultingoperatingconditiontheonly
waythesystemcouldhavebeensuccessfully
operated?
Could alternative or additional operating
conditionsbedefinedsuchthatsystem
operationswouldalsobesatisfactoryoreven
optimal?
Thus,thePOGAmethodologywasusedinconjunc-
tionwithEPANETtosimulatealternativeand
possiblyoptimalwaterdistributionsystemoperations
andtoassesstheeffectsofvariationsinsystem
operationsontheresultsoftheproportionatecontri-
bution simulations.
7.2.4 Simulation Results and Conclusions
Figure7-5showstheaerialdistributionofsimulated
proportionatecontributionresultsforallmodelnodes
(pipelinejunctions)forthemonthofJuly1988,using
theParkwaywellfieldasthepointofentry(source
point). The simulated proportionate contribution
resultsaredividedintosixintervals(1to10percent,
10to25percent,25to50percent,50to75percent,
75to90percent,and90to100percent)andacoloris
assignedtoallnodeswithineachinterval(resultsare
not shown for negligible proportionate contributions
oflessthan1percent).
Simulatedproportionatecontributionresultscanalso
beviewedintermsofselectedpipelinelocationsand
thecombinationofwellsorwellfieldsthatcontribute
watertothatlocation.Fivegeographicallydistinct
pipelinelocationsareselectedfromthehistorical
networkstorepresentthespatialdistributionof
proportionate contribution results. These locations
areidentifiedonFigure7-5aslocationsA,B,C,D,
andE.Thesimulatedproportionatecontributionof
waterforJuly1988correspondingtoeachpipeline
locationisshowngraphicallyonFigure7-6. The
simulation results demonstrated that the contribution
ofwaterfromwellsandwellfieldsvariedbytimeand
location.However,theresultsalsoshowedthatcertain
wellsprovidedthepredominantamountofwaterto
locationsthroughouttheDoverTownshiparea.
Additionally,althoughthepatternfactorsforsome
hoursofoperationsshowedmarkeddifferences,the
simulatedproportionatecontributionsofwaterusing
7-7
AReferenceGuideforUtilities
Figure 7-5. Areal Distribution of Simulated Proportionate
Contribution of Water from the Parkway Wells (22, 23, 24,
26, 28, 29) to Locations in the Dover Township Area, NJ,
July 1988 Conditions (from Maslia et al., 2001).
A
B
C
D
E
Figure 7-6. Simulated Proportionate Contribution of Water
from Wells and Well Fields to Selected Locations, Dover
Township Area, NJ, July 1988 Conditions (from Maslia et
al., 2001).
patternfactorsderivedfromtheapplicationofthe
POGAmethodologyshowedlittledifferencethrough-
outtheDoverTownshipareawhencomparedto
corresponding proportionate contribution of water
simulatedusingthemanualadjustmentprocess. The
resultsofsensitivityanalysesconductedusingthe
historical reconstruction process indicated the
following:
Therewasanarrowrangewithinwhichthe
historicalwaterdistributionsystemscould
havesuccessfullyoperatedandstillsatisfy
hydraulic engineering principles and the
MOC.
Dailyoperationalvariationsoveramonthdid
not appreciably change the proportionate
contributionofwaterfromspecificsources.
Therefore,thereconstructedhistoricalwater
distributionsystemsweredeterminedtobethemost
plausibleandrealisticscenariosunderwhichthe
19621996 historical water distribution systems
wereoperated. Thehealthscientistsconducting
the case-control epidemiologic study used the
resultingpercentageofwaterderivedfromthe
differentsourcestoderiveexposureindicesfor
each study subject.
Theresultsfromthecase-controlstudyshowedthat
therewasanassociationbetweenprenatalexposureto
contaminatedcommunitywaterandleukemiain
femalechildren(NJDHSS,2003).Forexample,female
leukemiacaseswere5timesmorelikelytohave
occurredwhenexposedduringtheprenatalperiodto
ahighpercentageofParkwaywellwaterthanwere
controlchildren. Thecontrolchildrenarethose
livinginthestudyarea,butwerenotexposedtothe
waterfromthecontaminatedwellfields. These
findingswouldnothavebeenpossiblewithoutthe
resultsderivedfromtheinnovativewaterdistribution
systemmodelingefforts.Theseeffortshaveledto
developingnewmethodsforevaluatingtheaccuracy
ofmodelingresultsandexposureclassification
techniquesthatarecriticalcomponentsofepidemio-
logicstudies.Someoftheinnovationsdocumented
bytheDoverTownshiphistoricalreconstruction
analysisare:
Anewapproach,proportionatecontribution
analysis,wasdevelopedthatutilizedwater
distributionsystemmodelingandsource
tracingtoquantifyexposureonamonthlybasis
foralllocationshistoricallyservedbythe
distribution system.
Through the use of an innovative genetic
algorithmapproach(POGA),historicalwater
distributionsystemoperatingscheduleswere
7-8
1
9
7
0
1
9
7
1
1
9
7
2
1
9
7
3
1
9
7
4
1
9
7
5
1
9
7
6
1
9
7
7
1
9
7
8
1
9
7
9
1
9
8
0
1
9
8
1
1
9
8
2
Individual3
1.60E+06
1.80E+06
2.00E+06
1.40E+06
I
n
t
a
k
e
,
g
1.20E+06
1.00E+06
8.00E+05
6.00E+05
4.00E+05
2.00E+05
0.00E+00
Year
Figure 7-7. Estimated Upper 97.5 Percent Credibility
Limit for Annual Perchlorate Intake by One Plaintiff
(Grayman, 2004).
AReferenceGuideforUtilities
ThecityofRedlandsliesintheSanBernardinovalleyofCalifornia,approximately60mileseastofcentralLos
Angeles. In1981,aroutineanalysisforchlorinationbyproductsrevealedthepresenceoftrichloroethylene
(TCE)inasampleofwaterfromtheRedlandswatersystem. Subsequentwaterqualityanalysesrevealedthata
numberofwellssupplyingthecitywerecontaminatedwithTCE. In1997,theperchlorateanion(ClO
4
)was
alsodetectedinseveralwells. In1996,thefirstofaseriesoflawsuitswasfiledinCaliforniaStateCourt
allegingthatthesourceofthesecontaminantswasamanufacturingfacilitylocatedup-gradientfromthemost
seriouslycontaminatedwells.
Oneoftheselawsuitsclaimedthatplaintiffswereharmedbyexposuretotoxicchemicalsthatwereimproperly
disposedofatthemanufacturingsiteandfoundtheirwayintogroundwaterthatwassubsequentlyextracted
throughthecityswellsanddeliveredtowatercustomers,includingtheplaintiffs.Theplaintiffsburdenof
proofrequiresthemtoestablish,amongotherthings,thattheywereactuallyexposedtocontaminatedwaterat
theirhomes,placesofwork,orotherlocationsandthattheamountsofcontaminantsthatenteredtheirbodies
astheresultoftheseexposuresweresufficienttocauseharmtothem.
Toestablishthisproof,expertsfortheplaintiffs
reconstructedthehistoricalconditionsinthewater
distributionsystemoftheCityofRedlandsovera
periodfromthemid1950stothelate1990susing
theEPANETmodel.Aspartoflitigation,several
forensicreconstructionsofwaterqualityinthe
Redlandswaterdistributionsystemwereperformed.
Thereconstructioninvolvedestimatesofboth
humanexposuretotoxiccontaminantsandwhole-
bodyintakesofthesechemicals. Estimatesof
exposuresandintakeswereexpressedascredibility
intervals,whichwerecalculatedusingMonteCarlo
simulationtechniques.Asanexampleoutputfrom
theanalysis,Figure7-7illustratestheestimated
upper97.5percentcredibilitylimitofone
plaintiffsexposuretoperchlorate.Similarinforma-
tionwasdevelopedtodescribethelower2.5
percentcredibilitylimitforeachofthetestplain-
tiffsinthecase(Graymanetal.,2004).
synthesized. Sensitivity analyses indicated
that operating system changes did not
appreciably change the proportionate
contributionofwatertoDoverTownship
locations.
Theassociationbetweenexposureanddisease
wouldnothavebeenpossiblewithout
developing the integrated approach using
environmental science, engineering
evaluations, and epidemiologic analyses.
Historicalreconstructionofenvironmentalexposureis
notaneasytask.Theproceduresandresultssumma-
rizedherein(andthedetailedanalysesinMasliaet
al.,[2001])representoneofthemostcomprehensive,
well-documented,andquality-controlledstudiesofits
kind.Anotherexampleofpublicexposureassessment
usingmodelingisrelatedtotheCityofRedlands,
California.
7.3ApplicationofWater
DistributionSystemModeling
inSupportofaRegulatory
Requirement
ThenewDBPR2regulationthatisproposedfor
promulgationinthenearfuturerequiresallwater
utilitiesthathaveadisinfectantresidualinthe
distributionsystemtoperformanIDSEunlessthey
obtainasmall-systemor40/30waiver(EPA,2003).
SystemsthatcancertifyTTHMandHAA5compli-
ancedatatobelessthanorequalto40g/Lfor
TTHMand30g/LforHAA5arenotrequiredto
performanIDSE. ThegoaloftheIDSEistoidentify
compliancemonitoringsitesthatmayhavehighDBP
levelsindistributionsystems.Utilitiesmaychooseto
performanSMPthatinvolvesextensivemonitoring.
Alternativelytheymaychoosetoperformasystem-
7-9
AReferenceGuideforUtilities
specificstudy(SSS)thatuseshistoricaldata,distribu-
tionsystemmodels,orotheranalysescombinedwith
minimalmonitoringtoevaluateTTHMandHAA5
levelsthroughoutthedistributionsystemasthebasis
toselectfuturecompliancemonitoringsites. This
casestudydemonstrateshowahydraulic/water
qualitydistributionsystemmodelcanbeappliedto
satisfytheIDSErequirementsofanSSS.
Key Phrases to Characterize Case Study:Regula-
torymodeling,IDSE,waterqualitymodeling.
7.3.1 IDSE Requirements Overview
TheIDSEguidancemanualspellsoutaseriesof
suggestedminimumrequirementsfortheuseofa
calibratedwaterdistributionsystemhydraulicmodel
toperformanSSS.Ingeneral,thewaterdistribution
systemmodelshouldbemorecomprehensiveforthe
purposeofanSSSthanmodelstypicallyusedfor
long-rangecapitalimprovementprogramanalysis
(e.g.,masterplanning).Acalibratedhydraulicmodel
intendedfordetaileddistributionsystemdesign(e.g.,
forsystemimprovements)oroperationalstudiesis
likelytobeadequate. Becausesystemsarealways
changing(e.g.,populationgrowth,industrydevelop-
mentinnetworkarea,agingofmains),itisimportant
thatthemodelgenerallyreflectsystemconditionsand
demandatthetimeoftheIDSESSS.Amodelthathas
notbeenupdatedorcalibratedinthelast5to10
yearsisunlikelytobeadequateforanSSS.The
guidelinesprovidedinthedraftguidancemanualare
summarizedbelow:
EPSmodelthathasbeenrecentlycalibrated
using generally accepted methods.
Anall-pipemodelorskeletonizedmodelthat
includes(a)atleast50percentoftotalpipe
lengthinthedistributionsystem,(b)atleast75
percentofthepipevolumeinthedistribution
system,(c)all12-inch-diameterandlarger
pipes,(d)all8-inchandlargerpipesthat
connectmajorfacilities,(e)all6-inchandlarger
pipesthatconnectremoteareasofadistribution
system,and(f)allactivecontrolvalvesorother
systemfeaturesthatcouldsignificantlyaffect
theflowofwaterthroughthedistribution
system.
Waterdemandshouldbeallocatedamongthe
nodesofthemodelinamannerthatreflectsthe
actualspatialdistributionofsuchdemand
throughoutthesystem.
Asystem-specific,diurnal(24-hour)demand
patternshouldbeappliedtotheoverallsystem
demand.
Themodelshouldaccuratelysimulateseasonal
systemconfigurationsandoperationalchanges.
Oncethemodelisestablished,itisthenruninEPS
modeuntilaconsistent,repeatingtemporalpatternof
waterageisestablishedatallnodesofthemodel.
Generally,themodelshouldberununderhighDBP
formationconditions(typicallysummermonths)to
estimateresidencetimes. Basedonthemodeledwater
ageresults,preliminarymonitoringsitesareidentified
nearlocationsthatsatisfythesamplingsiterequire-
ments.Samplingsitesareselectedtorepresent:
High-TTHMSites:HighTTHMvaluesare
expected at high-residence-time locations.
Theselocationscanbeidentifiedbyreviewing
themodeledwaterageateachnodeinthe
model. Thesesitesaregenerallydownstreamof
storagefacilitiesandinremotelocations.
However,theregulationdoesnotrequire
extremesornon-representativesitestobe
sampled.
High-HAA5Sites:Thecriteriaandprocedure
forselectinghighHAA5sitesusingahydraulic
modelaregenerallythesameasthosedescribed
aboveforselectinghigh-TTHMsiteswithone
importantdifference:thesiteschosento
representhighHAA5shouldhaveadisinfectant
residualsufficienttosuppressbacteriawhich
candegradeHAAs.
Average-Residence-TimeSites: Average-
residence-timesitescanbeselectedfromsites
withresidencetimesclosetotheflow-weighted
meanofallnodalresidencetimes.
Near-Entry-PointSites:Modeledwateragecan
beusedtoidentifylocationsinthenearvicinity
toentrypointsintothewatersystem.
Requirementsforthenumberofmonitoringsiteshave
notyetbeenfinalized. Asaresult,thiscasestudy
demonstratesthegeneralusageofmodelsforIDSE
andreliesuponthe2003draftguidanceissuedby
EPAtoillustratetheusageofmodels.
7.3.2 Example Application of Modeling in the
IDSE Process
Thefollowingexampleisahypotheticalcasestudy
basedinlargepartonanactualwaterdistribution
system. Thesystempurchasesdisinfectedgroundwa-
terandservesapproximately15,000people. Water
entersthedistributionsystemfromtwoseparate
interconnectionstoawholesaleutility.Theaverage
demandis2.2MGD.Anorthinterconnectionoperates
intermittently and provides approximately 80 percent
ofthedemand,whilethesouthinterconnection
operatesatalltimesandprovidestheremaining20
percentofthesystemdemand.Thereisa1.5-million-
gallonstoragetank.InordertocomplywiththeStage
2requirements,thedraftproposedDBPR2statesthat
atotalofsixsitesarerequiredforautilityofthissize
7-10
Figure 7-8. Average Water Age in the Distribution
System Over Last 24 Hours of a 2-Week Simulation.
AReferenceGuideforUtilities
usinggroundwater:onerepresentingasiteneartothe
predominantentry,onerepresentingasitewith
averageresidencetime,twositesrepresentinghigh
TTHMconditions,andtworepresentinghighHAA5
conditions.
Thewaterutilitywaterdistributionsystemmodelhas
beenusedextensivelyinthepastforbothhydraulic
andwater-qualitystudies.Itisaskeletonizedmodel
thatincludesall8-inch-diameterandlargerpipesand
allmajorfacilities.Thepipesinthemodelrepresent
74percentofthetotallengthofpipeinthesystem
and86percentofthetotalvolume.Themodelhas
beenpreviouslycalibratedbasedontwotracerstudies
andhasbeenshowntohaveexcellentagreementwith
observedfieldresults.Demandshavebeenassigned
tonodesbasedonactualmeterreadings,andinforma-
tionfromtheSCADAsystemhasbeenusedtoconstruct
atypicaldiurnalwaterusepattern.Themodelisbeing
operatedinanEPSmodetosimulatea12-dayperiod.
Themodelhasalsobeencalibratedforuseinsimulat-
ingchlorineresidualinthedistributionsystem.
Aseriesofrunsofthemodelwereperformedtohelp
understandthemovementofwaterandwaterquality
transformationsinthesystem.Specificsimulations
includedwaterageandchlorineresidual.Theresults
ofthewater-agemodelrunareshowninFigure7-8.
Thisplotshowstheaveragewateragethroughoutthe
distributionsystemoverthelast24hoursofthe2-
weeksimulation.Thisperiodwasselectedtoavoid
theuncertaintyassociatedwithassigninginitialwater
ageinthesystem.Theplotillustratesthenodesinthe
vicinity of the predominant northern interconnection
thatreceivewaterwithanaverageageoflessthan2
hours.Asalsoshown,theaveragewaterageincreases
significantlyforareasthatarefurtherfromthe
interconnections.Thedemandflow-weightedaverage
wateragefordeliveredwaterwascalculatedtobe27
hours.However,wateragecanalsovaryquite
significantlyoverthecourseofadayinwater
systems,largelyduetotheimpactsofstoragetanks.
ThisisillustratedinFigure7-9,whichdepictsthe
minimumwaterageateachnodeoverthesame24-
hourperiod.ThisisalsoshownintheplotinFigure
7-10forNodeJ-456inthevicinityofthetank.The
IDSEguidancedoesnotrequireutilitiestoexplicitly
considertheeffectsoftanksondiurnalvariationsin
waterage,andthusontheformationofDBPs.IfNode
J-456wasselectedasrepresentativeofhighDBP
becauseofitshighwaterage,itwouldbeexpected
thattheDBPswouldonlybehighduringthepartof
thedaywhenwaterisbeingdischargedfromthetank.
Themodelwasalsousedtodeterminethechlorine
residualthroughoutthesystem.Figure7-11contains
aplotoftheminimumchlorineresidualthroughout
thesystem.Itisimportanttonoteareaswithhigh
Figure 7-9. Minimum Water Age in the Distribution System
Over Last 24 Hours of a 2-Week Simulation.
7-11
Figure 7-11. Minimum Chlorine Residual in
Distribution System Over Last 24 Hours of a 2-Week
Simulation.
AReferenceGuideforUtilities
Figure 7-10. Diurnal Water Age at Node J-456.
Figure 7-12. Zones Representing Potential
Monitoring Locations for IDSE Based on Modeling.
residencetimeandlowresidualforhigh-TTHMsites
andhighresidencetimeandhighresidualforhigh-
HAA5sites.Thisinformationcanbeusedtoavoid
selectingmonitoringsitesthataretobeusedas
representativeofhighHAA5concentrations.The
currentabilitytoaccuratelymodelHAA5ina
distributionsystemislimited.However,researchhas
shownthatdepressedchlorineresidualcanresultin
biodegradationofHAA5,thusloweringtheHAA5
concentrationsevenforolderwater.Figure7-11also
showsasmallareainthesouthwesternportionofthe
systemthatthemodelpredictstopotentiallyexperi-
encechlorineresidualslessthan0.2mg/Lofchlorine.
Basedonthemodelingresults,variouszoneswere
definedinthedistributionsystemrepresentingareas
thatareappropriatefordifferenttypesofcompliance
monitoringrequirements(Figure7-12).Inactualuse,
informationgeneratedbythemodelwouldbe
supplementedbyhistoricalfielddata.Thezones
shownintheplotinclude:
1. Nodesinthevicinityofthepredominantnorth
connectionwithwateragelessthan2hours
representativeofnearentrylocations;
2. Nodeswithaveragewaterageintherangeof21
to33hoursthatrepresentlocationsthat
approximatetheaverageresidencetimeof27
hours;
7-12
3. Nodeswithresidencetimesthatexceedtwice
theaveragewaterage(>54hours)andhavea
minimumchlorineresidualexceeding0.20mg/L
representingpotentialhigh-HAA5sites;and
4. Othernodeswithresidencetimesthatexceed
twicetheaveragewaterage(>54hours)
representingpotentialhigh-TTHMsites.
Asillustratedbythecaseexample,ifadetailed,
calibratedEPSmodelisavailable,themodelrepre-
sentsanefficientmeansofdefiningthecompliance
monitorlocationsasrequiredundertheforthcoming
regulation.
7.4UseofWaterDistribution
SystemModelsinthe
PlacementofMonitorsto
Detect Intentional
Contamination
Theincreasingconcernoverthepotentialforinten-
tionalcontaminationofawaterdistributionsystem
hasledtointerestintheplacementofmonitorsto
detectcontaminationandtoserveaspartofarapid
detectionsystem. Designofsuchmonitoringsystems
mustincludedecisionsonthetype,numberand
locationforthemonitors. Waterdistributionsystem
modelscanplayasignificantroleinthedecision
makingbyprovidingaquantitativemechanismfor
determiningthemovementofacontaminantthrough
thedistributionsystemandtestingtheeffectiveness
ofamonitoringsystemdesign.
Toillustratethisapplication,aredteam-blueteam
conceptisused(Graymanetal.,2005). Thered
team-blueteamconceptispartofwargaming
thatiswidelyusedtodayasamechanismfor
training and development and testing of security
plans. Theredteamactsastheaggressorandthe
blueteamactsasthedefenders. Eachteamhas
differenttypesandamountsofinformationavail-
abletothemanddifferentrulesorconstraintsthat
theymustfollow.Inthiscasestudy,network
modelsareusedintwomodestoassistinevaluat-
ing monitoring networks:
1. Aspartofaredteam-blueteamexerciseto
demonstratetheeffectivenessofmanual
selectionoflocationofmonitorsaspartofa
CWS.
2. Aspartofanoptimizationmodeltoselectthe
bestlocationsformonitorsbasedonastated
metricformeasuringtheeffectivenessofthe
monitoringsystem.
AReferenceGuideforUtilities
Key Phrases to Characterize Case Study:Water
security, contamination, optimization, monitor
placement
7.4.1 Red Team-Blue Team Exercise
Inthissimulatedexercise,theredteamattacksawater
distributionsystembyaddingaharmfulchemicalto
thewater.Theredteamisprovidedwithlimited
informationonthedistributionsystem,anumberof
potentiallocationswheretheycaninjectacontami-
nant,andpredeterminedinformationonthecharacter-
isticsofthecontaminant(quantityandlethalityofthe
contaminant).Theblueteamrepresentsthewater
utilityandattemptstoprotectthewatersystemby
installingthreemonitorsaspartofaCWSthatdetects
contaminants. Itisassumedthattheyhaveextensive
informationonthedesignandoperationofthe
distributionsystembutnofirminformationonwhere
theattackersmaychoosetointroducethecontami-
nantorthenatureofthecontaminationscenario.
Thewaterdistributionsystemnetworkusedinthis
exerciseisaskeletonizedversionofamajorpressure
zoneofawaterdistributionsysteminCalifornia
approximating the conditions (design and operation)
inthemid1990s.Thesystemisfedbytwosources;
onethatoperatescontinuouslyandonethatoperates
onlyduringtheday.Therearethreestoragefacilities
locatedinthenetwork.Thisnetworkisoneofthe
examplenetworksprovidedaspartoftheEPANET
model.Thesimulationperformedintheexerciseisa
24-hourEPSstartingat7AM.Themodelrepresenta-
tionofthenetworkisshowninFigure7-13. This
figurealsoillustratestherelativenodaldemands,and
thetypicalflowdirectionsandmagnitudeduringthe
day.Thisplotisgivenonlytotheblueteamto
providethemwithinformationonthedesignand
operationofthesystem. Theredteamisprovided
Figure 7-15. Contaminant Concentration Just Downstream
of Contaminant Introduction Location (Node 121).
Figure 7-16. Contaminant Concentration Far Downstream
of Contaminant Introduction Location (Node 143).
AReferenceGuideforUtilities
nodelocated1500feetimmediatelysouthofthe
injectionpointisshown.Asexpected,thecontami-
nantmovedveryrapidlyandreachedthisnodeinless
thananhouraftertheinjection. Ifamonitorwas
locatedatthisnode,andrapidanalysisandre-
sponseoccurred,itcouldbeveryeffectiveasan
earlywarningformostofthedistributionsystem.
Figure 7-16 illustrates concentrations resulting
fromthesamecontaminantintroductionlocation.
Thisnodeislocatedneartheeasternedgeofthe
distribution system approximately 2.5 miles
downstream of the contaminant introduction
location.Asillustrated,theconcentrationofthe
contaminantremainedaboutthesamebutthetravel
timetothispointwasapproximately7hours.For
thisinjectionscenario,amonitorlocatedatthis
pointwouldberelativelyineffectiveasawarning
deviceformostofthedistributionsystembecause
ofthesignificanttimelag.
Figure 7-14. Allowable Contaminant Introduction Locations.
onlywithamapofthedistributionsystemshowing
eightpotentialsitesthatcanbeusedtointroducea
contaminant(Figure7-14).
Followingtheselectionofpointsofattackbythered
teamandselectionofmonitorlocationsbytheblue
team,thecontaminantintroductionissimulatedusing
theEPANETmodel.Themovementandconcentration
ofthecontaminantsarethenviewedgraphicallyby
animatingthemovementofthecontaminantinthe
distributionsystemintheEPANETmodel.Thetime
historyofcontaminantconcentrationsisalsoviewed
atselectednodesinthedistributionsystem.The
effectivenessofmonitorsisillustratedbygraphing
theconcentrationsofthecontaminantsatmonitoring
nodesandassessingwhether(andhowquickly)the
monitorswillservetheirpurposeofdetectingthe
contaminant. Figures7-15and7-16showthe
concentrationsresultingfrom8hoursofcontamina-
tionatnode123startingat11a.m.
InFigure7-15,theresultingcontaminationata
Intheexercise,mostredteammemberstendtoselect
contamination introduction locations that they
perceivewouldresultinthemostwidespreadimpacts.
Themostoftenselectedsiteswerethoseneartothe
watersources.Littleattentionisgenerallygivento
theamountofdilutionthatwouldresultataparticu-
larlocation.Blueteammemberstendtoselect
monitoringlocationsthatcoverawiderangeof
locations.Frequently,thethreeallowablemonitors
willbelocatedinthenorth,central,andsouthern
portionsofthedistributionsystem.
7.4.2 Application of Optimization Model
The optimization model used in this demonstration
isamethodologydevelopedbyOstfeldand
Salomons(2004).ThemodellinksEPANETanda
geneticalgorithminanoverallframeworkfor
optimally allocating monitoring stations, aimed at
detecting deliberate external contamination into
waterdistributionsystemnodes.Themodel
operatesunderextendedperiod(unsteady)hydrau-
lics and water quality conditions. The optimization
routinedeterminesthemonitorplacementtodetect
contaminantsinordertominimizetheexposureof
7-14
customersaboveanallowableminimumconcentra-
tion.Thealgorithmcanbeusedtostudycontami-
nationoffixedduration,quantity,orlocationor
can simulate contamination under stochastic
conditions.Thereareseveralmodelparametersthat
canbespecifiedtocontrolthenumberofmonitors,
the allowable contaminant introduction locations, the
characteristicsoftheevent,andwhethertheevent
characteristicsanddemandsaretobeconsideredas
stochastic variables.
Inoneapplicationofthemodel,itwasassumedthat
thepollutantcouldbeintroducedatanysinglenode
ofthedistributionsystematanytime,allwiththe
same injection probability. The following additional
assumptionsweremade:
Themaximumcontaminationexposurevolume
tothepublicabovewhichanalarmsignalofthe
monitoringstationsisrequiredis25gallons.
Thewaterisconsideredcontaminatedabove1
mg/L.
Thepollutantflowdischargeis2kg/minfor5
minutes(i.e.,atotalof10kgofasolutionof100
percentisintroducedwithinatotalof5minutes).
Thepollutantflowdischargeofthe
contaminantintroducedandtheconsumer
demandsaredeterministic.
Themonitoringstationsareprovidingreal-time
dataanddetectionalarms.
Allmonitoringstationshaveadetection
sensitivityof1mg/L.
3monitorsaretobeplaced.
Themodelsuggestsplacingmonitorsatnodes143,
181,and213withadetectionlikelihoodof0.4354
(i.e.,thereisaprobabilityofabout44percentthatthe
contaminantwillbedetectedpriortotheconsump-
tionofmorethan25gallonsataconcentrationhigher
than1mg/L).Thelocationofthemonitorsisshown
inFigure7-17.Asillustrated,theselectedmonitor
locationswererelativelyevenlyspacedaroundthe
network.
Otherevaluatedscenarioslookedatadifferent
numberofallowablemonitors,theallowablecontami-
nant introduction locations, the critical exposure
threshold, and representation of contaminant quantity
andnodaldemandsasstochasticvariables.Though
theexactoptimallocationsvariedslightlybetween
thedifferentruns,typicallythemonitorswereplaced
throughoutthenetwork.However,theeffectivenessof
themonitoringnetwork,asmeasuredbythedetection
likelihooddoesvaryconsiderablybetweenscenarios.
AReferenceGuideforUtilities
Figure 7-17. Monitoring Locations Selected by the
Optimization Model.
7.4.3 Case Summary
Theredteamblueteamexerciseservesasagood
mechanismfordemonstratingboththedynamicsof
contaminantmovementinthedistributionsystemand
thepotentialeffectivenessofmonitors.Applicationof
theoptimizationmodel,bothasademonstration
procedureandasapracticaltool,providesamethod
thatmovesthemonitorplacementfromapurely
intuitiveprocesstoaquantitativeprocedure.Boththe
exerciseandtheoptimizationtoolshowtheimpor-
tanceinminimizingdelaysinthedetection,notifica-
tion,andresponseprocess. Evenanaddeddelayofan
hourortwocanleadtoasignificantincreaseinthe
numberofcustomersthatwouldbeimpactedbya
contamination event.
7.5CaseStudyUseof
PipelineNetModel
Thiscasestudyfocusesontheapplicationofthe
PipelineNetmodel,whichincorporatesbothGISand
theEPANETmodeldiscussedinthepreviouschapters
ofthisreferenceguide. Thesupportinginvestigations
wereprimarilysponsoredbytheAwwaResearch
Foundation(AwwaRF)andEPAwithworkperformed
byaconsultingfirm(SAIC)andassistancefromwater
utilitypersonnel (RonHunsinger,BillKirkpatrick,
DaveRehnstrom)workingattheEastBayMunicipal
UtilityDistrict(EBMUD),Oakland,CA. Thetextand
figuresareadaptedfromAwwaRFreport2922
preparedbyBahaduretal.(2003).
Key Phrases to Characterize Case Study: hydraulic
andwaterqualitymodeling,placementofmonitors,
exposure modeling, contamination assessment,
contamination response tools, geospatial analysis.
7-15
AReferenceGuideforUtilities
7.5.1 Overview
PipelineNetisanEPANET/ArcView
1
-basedmodel,
andusesthesamehydraulicengineasEPANET.The
EPANETportionofthemodelcansimulatethefate
and transport of potentially introduced contaminants
inawaterdistributionsystem.TheArcView(orthe
GISlayer)portionofthemodelcanrelatethe
geospatialcomponentsofthesimulatedimpact. The
GISlayerallowsforgeo-featuresandmapdisplay
withanoverlayofmodeloutput. Thisfeatureis
particularlyusefulinapplicationssuchasemergency
response,determiningoptimalplacementofsampling,
and monitoring instruments.
AwwaRFandEPAjointlyfundedaprojectto
develop techniques to locate monitoring points in a
distribution system, determine appropriate timing
andfrequencyofmonitoring,andestablishmoni-
toring techniques and relevant water quality
parameters. Forthispurpose,afullycalibrated
extendedperiodsimulation(EPS)networkmodel
hypotheticallyrepresentingaportionofEBMUD
wasdevelopedusingPipelineNet. Thisstudyarea
networkmodelrepresents16ofthe123pressure
zonesintheoverallEBMUDdistributionsystem.
Thestudyareacontained27tanks,748milesof
pipes,62pumpsand17,997pipesegmentswith
diametersequaltoorgreaterthan2inches. Figure
7-18showsapartialviewofthehypothetical
networkofpipelines.
7.5.2 Model Calibration
Thenetworkmodelwascalibratedbycomparing
theobserved(SCADAdata)andsimulated
(PipelineNetmodel)waterlevelin25tankslocated
inthestudyarea. Theprimaryfocusofthecalibra-
Figure 7-18. Hypothetical Water Distribution System
Showing Pipelines.
1
RegisteredTrademarkofESRI
tionwastomatchtheshapeoftheobservedwater
levelinthetanks.Themodelcalibrationwas
performedfora24-hourtimeintervalusingdata
measuredonJuly1,2001. Tofurtherenhance
calibration,thepumpcharacteristiccurveswere
usedtoachieveagoodcomparisonbetween
simulatedandobservedtanklevels. Theflowvalue
ofthecharacteristiccurvewaschangedasneces-
sarytoreflectfieldconditions. Eachpumpwas
operatedwithtimecontrols.
7.5.3 Monitoring Site Location Methodology
Ahierarchicalselectionprocesswasdevelopedto
locatemonitoringstationsinthedistributionsystem.
Athree-stepapproachwasemployedbasedonmodel
inputs,outputs,andGISlayers(seeFigure7-19).