o) monitoring in full-scale wastewater treatment processes
TRANSCRIPT
AbbreviationsamoA
Review
Adecadeofnitrousoxide(N2O)monitoringinfull-scalewastewatertreatmentprocesses:Acriticalreview
V. Vasilakia
T.M.Massarab
P. Stancheva
F. Fatonec
E. Katsoua,b,∗
aDepartmentofCivil&EnvironmentalEngineering,BrunelUniversityLondon,UxbridgeCampus,Middlesex,UB83PH,Uxbridge,UK
bInstituteofEnvironment,HealthandSocieties,BrunelUniversityLondon,UxbridgeCampus,Middlesex,UB83PH,Uxbridge,UK
cDepartmentofScienceandEngineeringofMaterials,EnvironmentandCityPlanning,FacultyofEngineering,PolytechnicUniversityofMarche,Ancona,Italy
∗Correspondingauthor.DepartmentofCivil&EnvironmentalEngineering,BrunelUniversityLondon,UxbridgeCampus,Middlesex,UB83PH,Uxbridge,UK.
Abstract
Direct nitrous oxide (N2O) emissions during the biological nitrogen removal (BNR) processes can significantly increase the carbon footprint of wastewater treatment plant (WWTP) operations. Recent onsite
measurementofN2OemissionsatWWTPshavebeenusedasanalternativetothecontroversialtheoreticalmethodsfortheN2Ocalculation.Thefull-scaleN2OmonitoringcampaignshelptoexpandourknowledgeontheN2O
productionpathwaysand the triggeringoperationalconditionsofprocesses.TheaccurateN2Omonitoringcouldhelp to findbetterprocesscontrolsolutions tomitigateN2Oemissionsofwastewater treatmentsystems.
However,quantifyingtheemissionsandunderstandingthelong-termbehaviourofN2OfluxesinWWTPsremainschallengingandcostly.
Areviewoftherecentfull-scaleN2Omonitoringcampaignsisconducted.Theanalysiscoversthequantificationandmitigationofemissionsfordifferentprocessgroups,focusingontechniquesthathavebeenapplied
forthe identificationofdominantN2Opathwaysandtriggeringoperationalconditions, techniquesusingoperationaldataandN2Odata to identifymitigationmeasuresandmechanisticmodelling.Theanalysisofvarious
studiesshowedthattherearestilldifficultiesinthecomparisonofN2Oemissionsandthedevelopmentofemissionfactor(EF)databases;theN2Ofluxesreportedinliteraturevarysignificantlyevenamonggroupsofsimilar
processes.TheresultsindicatedthatthedurationofthemonitoringcampaignscanimpacttheEFrange.MostN2Omonitoringcampaignslastinglessthanonemonth,havereportedN2OEFslessthan0.3%oftheN-load,
whereasstudieslastingoverayearhaveamedianEFequalto1.7%oftheN-load.Thefindingsofthecurrentstudyindicatethatcomplexfeatureextractionandmultivariatedataminingmethodscanefficientlyconvert
wastewateroperationalandN2Odataintoinformation,determinecomplexrelationshipswithintheavailabledatasetsandboostthelong-termunderstandingoftheN2Ofluxesbehaviour.Theacquisitionofreliablefull-scale
N2Omonitoringdataissignificantforthecalibrationandvalidationofthemechanisticmodelsof-describingtheN2OemissiongenerationinWWTPs.Theycanbecombinedwiththemultivariatetoolstofurtherenhancethe
interpretationofthecomplicatedfull-scaleN2Oemissionpatterns.Finally,agapbetweentheidentificationofeffectiveN2OmitigationstrategiesandtheiractualimplementationwithinthemonitoringandcontrolofWWTPs
hasbeenidentified.Thisstudyconcludesthatthereisafurtherneedfori)long-termN2Omonitoringstudies,ii)developmentofdata-drivenmethodologicalapproachesfortheanalysisofWWTPoperationalandN2Odata,and
iii) betterunderstandingofthetrade-offsamongN2Oemissions,energyconsumptionandsystemperformancetosupporttheoptimizationoftheWWTPsoperation.
Keywords:Nitrousoxide;Full-scalemonitoringcampaigns;Emissionfactors;Productionpathways;Triggeringoperationalconditions;Mitigationmeasures;Multivariatedataminingmethods;Mechanisticmodels
Ammoniamonooxygenase
Anammox
AnoxicAmmoniaOxidation
ANN
ArtificialNeuralNetworks
A/O
Anaerobic-Oxic
A2/O
Anaerobic-Anoxic-Oxic
AOA
AmmoniaOxidizingAchaea
AOB
AmmoniaOxidizingBacteria
AOR
AmmoniaOxidationRate
AS
ActivatedSludge
ASM
ActivatedSludgeModels
BNR
BiologicalNitrogenRemoval
BSM2
BenchmarkSimulationModel2
CAS
ConventionalActivatedSludge
CO2
CarbonDioxide
COD
ChemicalOxygenDemand
DO
DissolvedOxygen
EF
EmissionFactor
GC
GasChromatography
GHG
GreenhouseGas
HAO
HydroxylamineOxidoreductase
HRT
HydraulicRetentionTime
ICA
IndependentComponentAnalysis
LCA
Lifecycleassessment
MDS
MultidimensionalScaling
MLE
ModifiedLudzack-Ettinger
MLSS
MixedLiquorSuspendedSolids
N
Nitrogen
N2O
NitrousOxide
N2OR
NitrousOxideReductase
NH2OH
Hydroxylamine
NH3
Ammonia
NH4+-N
AmmoniumNitrogen
NO
NitricOxide
NO2−-N
NitriteNitrogen
NO3−-N
NitrateNitrogen
NOB
NitriteOxidizingBacteria
NOR
NitricOxideReductase
nosZ
NitrousOxideReductase
OD
OxidationDitch
PCA
PrincipalComponentAnalysis
PE
PopulationEquivalent
PF
PlugFlow
PLS
PartialLeastSquares
r-A2/O
ReversedA2/O
qPCR
Real-timePolymeraseChainReaction
RAS
ReturnActivatedSludge
Redox
Reduction-Oxidation
RT-qPCR
ReverseTranscriptionPolymeraseChainReaction
SBR
SequencingBatchReactor
SP
SitePreference
SRT
SludgeRetentionTime
SVM
SupportVectorMachine
TKN
TotalKjeldahlNitrogen
TN
TotalNitrogen
WWTP
WastewaterTreatmentPlant
1IntroductionWastewatertreatmentplants(WWTPs)producesubstantialanthropogenicnitrousoxide(N2O)amounts(Lawetal.,2012).N2Oemissionsfromwastewatertreatmentprocesseshaveincreasedby44%from1990to2014(US
EPA,2016).Moreover,theyareresponsibleforupto26%ofthegreenhousegas(GHG)emissionsofthewholewatersupplychain(includingdrinkingwatersupply,wastewatercollectionandtreatment,effluentdischarge,sludge
processinganddisposal) (Laneetal.,2015).Therefore, theevaluationof theenvironmental impactofWWTPs isgaining increasingattentionworldwide (Guoetal.,2017;Harris et al., 2015a,bHarrisetal.,2015;Law et al., 2012;
Manninaetal.,2016;Massaraetal.,2017).N2Ohasanapproximately300timeshigherGHGeffectthanCO2(IPCC,2013).Comparedtootherunregulatedcombinedhalocarbonemissions,reducingtheanthropogenicN2Oemissionsin
theatmosphereplaysthemostsignificantroleinpreventingtheozonelayerdepletion(Danieletal.,2010;Portmannetal.,2012).
StricterenvironmentalregulationsarebeingimposedonWWTPs,whiletheneedtocontroltheenergyconsumptioncosts,pushestheplantmanagerstowardsthedeploymentofmoreenergyandcostefficientoperational
strategies(Batstoneetal.,2015;Ghoneimetal.,2016).Theintegrationofadditionalsustainabilitymetrics(e.g.GHGemissions,resourcerecovery)fortheevaluationoftheWWTPperformanceisalsogainingattention(Flores-Alsinaet
al.,2014;Cornejoetal.,2016).RecentstudieshaverevealedthatthedirectN2OemissionsofbiologicalprocessesinWWTPscanincreasetheoperationalcarbonfootprintby∼78%(Daelmanetal.,2013a)oreven∼83%(Deslooveret
al.,2011).Moreover,thenewtechnologyadoptioninWWTPsrequirestheconsiderationoftrade-offsbetweendirectandindirectGHGemissionstoensurethatitwillnotresultinincreaseoftheoverallcarbonfootprint.SincetheGHG
emissionshaveamajorcontributiontotheoveralltotalenvironmentalimpactofWWTPs,theyshouldbeconsideredaspartofthedecision-makingprocessfortheimprovementofthetechnologicalandoperationalplantperformance
(Sunetal.,2017bSunetal.,2017a,b;Frutosetal.,2018;Contheetal.,2018).Therefore,researchhasbeenconductedinthepastyearstoidentifyandsuggeststrategiesleadingtoN2Omitigationduringthebiologicalnutrientremoval
(BNR) processes (Desloover et al., 2012;Duan et al., 2017; Law et al., 2012). Different operating parameters, such as dissolved oxygen (DO), pH, ammonium (NH4+) and nitrite (NO2−) concentration, configuration types and
environmentalconditions(e.g.temperature),canaffecttheN2OproductioninWWTPs(Deslooveretal.,2012;Kampschreuretal.,2009b;Massaraetal.,2017).Theexact triggeringoperationalandenvironmentalconditions that
governtheN2Ogenerationarestillunderinvestigationbyresearchersandoperators(Wanetal.,2019).Additionally,theexactmechanismsthatdeterminetheN2Ogenerationarenotfullyunderstood,thushinderingtheestablishment
ofmitigationmeasures(Duanetal.,2017).Consequently,thelong-termdynamicsofN2Oemissionsinfull-scaleWWTPscannotbefullyexplained,evenforconventionalprocesses(e.g.plug-flow(PF)reactors)(Daelmanetal.,2015).
Differentdata,samplingtechniquesandanalyticaltoolshavebeenemployedinexistingfull-scaleN2Ostudies.Theobjectivesofthecurrentreviewpaperareto:i)evaluaterecentfindingsfromN2Omonitoringcampaignsin
termsofN2OEFs,dominantpathwaysandmitigationmeasuresfordifferentgroupsoffull-scaleconfigurationsandreactortypes,ii)examinethediscrepanciesamongthemethodsappliedindifferentmonitoringcampaigns,andiii)
evaluatethedataprovidedbystudiesinvestigatingN2Oemissions inWWTPs.Thecurrentstudycriticallyassesseshowthe findingsof theresearchonN2Oemissionshavebeenextrapolated indifferent full-scaleprocessgroups.
Moreover,thisworkevaluatesthemethodsthathavebeenappliedatfull-scalesystemsfortheN2Oquantification,controlandmitigation.Finally,theresearchgapsthatneedtobeaddressedinfuturestudiesarehighlighted.
2TheimpetusforquantifyingtheN2OemissionsinWWTPs2.1EFbenchmarks
EFquantificationinconventionalandadvancedwastewatertreatmentprocessesisessentialtoassessandreducetheresultingenvironmentalimpact(Foleyetal.,2015).CurrentmethodsforthetheoreticalcalculationofN2O
EFs atWWTPs rely on fixed (Palut andCanziani, 2007) or country-specificEFs and similar over-simplified approaches (SinghandMaurya, 2016). Thesemethods underestimate the actual emissions and are considered unreliable
(CadwalladerandVanBriesen,2017),sincetheyarenotrepresentativefordifferentprocessconfigurations,operationalandenvironmentalconditions.Therefore,amaintargetofthestudiesanalysedinthispaperwasto:i)developana
N2Oemissiondatabasefromdifferentmainstream(Ahnetal.,2010b,2010a)/sidestream(Kampschreuretal.,2008;Weissenbacheretal.,2010)processesandinnovativeconfigurations(Deslooveretal.,2011),andii)identifyN2OEFsfrom
processeslocatedinpreviouslyunreportedregionswithdifferentenvironmentalconditions(Wangetal.,2011).
2.2N2OpathwaysandtriggeringoperationalconditionsThemajorityofthefull-scaleN2OmonitoringcampaignsweredrivenbytheneedtoexpandtheknowledgeontheN2Ogeneration inwastewatertreatmentsystems.Severalstudieshavehighlightedthat it is importantto
understandthedominantN2Ogenerationpathwaysinfull-scaleprocessesandinvestigatetheeffectofspecificoperationalconditionstriggeringspecificenzymaticreactionslinkedwithelevatedEFsinfull-scalebiologicalsystems
(Bollonetal.,2016a;Panetal.,2016;Wangetal.,2016b).
TheN2O production during wastewater treatment involves several microbiological reactions during both autotrophic and heterotrophic processes that require either aerobic or anoxic conditions. Threemain biological
pathways have been identified in BNR systems; hydroxylamine (NH2OH) oxidation, nitrifier denitrification and heterotrophic denitrification (Kampschreur et al., 2009b). The NH2OH oxidation pathway is mainly catalysed by the
autotrophicammoniaoxidizingbacteria(AOB)andammonia-oxidizingarchaea(AOA).However,severalstudieshaveshownthatthecontributionofAOAtoN2Oemissionsinwastewaterisexpectedtobelow(Hooper,1968;Ritchieand
Nicholas,1972;Lawetal.,2012).Carantoetal.(2016)haverecentlyshowedthatN2OcanalsobethemainproductofanaerobicNH2OHoxidationcatalysedbythecytochromeP460inN.europaea.Thelattercanbeconsideredasevidence
ofbiologicalN2OgenerationunderlimitedDOandhighNH3concentrations(i.e.Lawetal.,2012),oraspotentialexplanationforthehighN2Oemissionsobservedduringthetransitionfromaerobictoanoxicconditions.Whiteand
Lehnert(2016)havealsosuggestedthatN2OcanbedirectlyproducedduringtheNH2OHoxidation(mediatedbytheNH2OHoxidoreductase(HAO)enzyme)underaerobicconditions,whereastheNO2−detectedcanresultasby-product
ofthenitricoxide(NO)oxidation.TheAOBcanalsoreduceNO2−toNO(bytheaidofnirk)and,subsequently,NOtoN2O(catalysedbynorB)mainlyunderoxygen-limitingconditionsviatheothernitrification-relatedpathway(nitrifier
denitrification)(PothandFocht,1985).Duringdenitrification,theheterotrophicdenitrifiersareresponsibleforthereductionofNO3−/NO2−tonitrogengas(N2).N2Oisanintermediateofdenitrification(SchulthessandGujer,1996).With
theNOreductase(NOR)ascatalyst,NOisreducedtoN2O(HochsteinandTomlinson,1988).NOcanalsoresultasby-productoftheincompleteNH2OHoxidationandthenserveasasubstrateduringdenitrification(HooperandTerry,
1979).Ifthedenitrificationprocesscontinuesundisturbedly,N2OisreducedtoN2inthefinaldenitrificationstep(catalyzedbytheN2Oreductase(N2OR)).Consequently,theheterotrophicdenitrificationprocesscanacteitherasasink
orasasourceofN2O(RobertsonandTiedje,1987).UnderelevatedNH2OHandNO2−concentrations,abioticyetbiologically-drivenN2OpathwayscanalsoconstituteimportantcontributorstotheN2Oemissions(Soler-Jofraetal.,2016;
Teradaetal.,2017;Harperetal.,2015).Insidenitritationreactorsforexample,theabiotic-bioticpathwayofnitrosationispossible;NO2−canreactwiththebiologicallyproducedNH2OHandformN2Oasend-product(Zhu-Barkeretal.,
2015).
BasedontheexistingknowledgeontheN2Oproductionpathways,recentreviewsonN2OemissionsfromwastewatertreatmentprocesseshaveconcludedthatthekeyoperationalvariablesresponsiblefortheN2Ogeneration
includebutarenotlimitedtothefollowing:i)lowDO,NO2−orfreenitrousacid(HNO2)accumulationandchangesintheNH4+concentrationinthenitrifyingzones,ii)limitationoforganicsubstrate(i.e.lowchemicaloxygendemand
toN(COD:N)ratio),aswellas,NO2−accumulationinthedenitrifyingzones,iii)alternationofanoxic/aerobicconditions,andiv)abruptchangesintheprocessesandsystemshocks(Duanetal.,2017;Guoetal.,2017;Lawetal.,2012;
Massaraetal.,2017).
Therefore,N2Oemissionscanoccurbecauseofdiversecontributingfactorsandenzymaticreactions.However,theseparametersandreactionscanoccursimultaneously,dynamicallyandbeyondoperators’controlinfull-scale
systems,whereassmallchanges(e.g.DOchanges)cansignificantlyaffecttheN2Oformation.Previousstudiesonfull-scalemonitoringcampaignsintendedto:i)identifythemostimportantoperatingconditions(e.g.aerationrate,DO,
NO2−concentration,pH,etc.)andcorrelatethemwiththeN2Ogeneration(Brottoetal.,2015;Rodriguez-Caballeroetal.,2014),ii)revealtheeffectsofseasonalvariationsontheN2Oformation(Yanetal.,2014),andiii)identifythekey
pathwaysfortheN2Oproduction(Wangetal.,2016b).
2.3N2OmitigationstrategiesAnotherkeyobjectiveof theN2Omonitoring campaignsperformed in thepast years, particularly significant for theWWTPoperators,was thedevelopmentof operational strategies for theminimizationof the emissions
(Deslooveretal.,2012).Therefore,severalauthorshavesuggestedN2Omitigatingmeasuresbasedonthefindingsoffull-scaleN2Omonitoringcampaigns(i.e.Chenetal.,2016;Mampaeyetal.,2016;Panetal.,2016;Wangetal.,2016b).The
proposedstrategiestocontrolN2Oemissions,areanalysedinthefollowingsections.
3EFestimationusingfull-scaleN2OmonitoringdataThis section emphasizes the need to increase the comparability amongst different studies that report N2O emissions. Moreover, it investigates potential trends in the EFs for certain groups of processes and
summarizessummarises thedata requirements for theEFassessment inWWTPs. Incaseswhere theEFswere reportedwith respect tounitsother than the influent totalnitrogen (TN)or the influentNH4+ content, appropriate
conversionsweremade(wherepossible)(seethesupplementarymaterialA–TableA2).
Fig.1showsthepercentageofpastfull-scaleN2Omonitoringcampaignswithreferencetothetreatmentconfigurationsappliedeachtime.AllprocessesconsideredinFig.1aregiveninthesupplementarymaterialB(Tables
B1andB2).Distinctmainstreamprocessconfigurations include,modifiedLudzack-Ettinger (MLE)reactors,conventionalactivatedsludge (CAS)systems (onlyaerobic reactors),A2/O (anaerobic/anoxic/aerobic)processesandA/O
(anoxic/aerobic)reactors.Additionally,oxidationditch(OD)reactortypesandsequencebatchreactor(SBR)typeshavebeenconsideredasdistinctprocessgroups.Sidestreamprocessesthatincludepartial-nitritationreactors,1-step
and2-steppartial-nitritation-anammoxconfigurations)areconsideredadistinctprocessgroupinFig.1.Theprocessesthatdonotbelongtotheaforementionedprocessgroupsarecategorizedseparately(i.e.Bareseletal.,2016;Mello
etal.,2013;Wangetal.,2016a).Detailsforalltheprocessesareprovidedinthesupplementarymaterial.
Most of the process-focusedmonitoring campaigns include, CAS systems (∼12%), (MLE) configurations (∼12%), A2/O configurations (∼10%), oxidation ditches (ODs) (∼8%) and sidestream partial-nitritation reactors or
anammoxsystems(∼11%)(Fig.1).Overall,thesestudiesrefertoawiderangeofconfigurations(i.e.Aboobakaretal.,2013;Brottoetal.,2015;Castro-Barrosetal.,2015;Sunetal.,2017a,bSunetal.,2017a)thathavebeenmonitored
mainlyforshortperiods(i.e.Ahnetal.,2010b;Bellandietal.,2018;Foleyetal.,2010;Panetal.,2016)withvaryingmethodology(e.g.differentgaseoussamplingandanalyticalmeasurementprotocols)(Daelmanetal.,2015;Renet
al.,2013).
Fig.1Thenumberofmonitoringcampaignsreportedinliteratureperprocessgroup;A/O:Anoxic/oxicreactor,A2/O:anaerobic-anoxic-oxicreactor,CAS:conventionalactivatedsludge,MLE:ModifiedLudzack-Ettingerreactor,OD:oxidationditch,SBR:
sequencingbatchreactor,PNandPN/A:partial-nitritationandpartial-nitritation-anammox(1and2-stage).
alt-text:Fig.1
EFcomparabilitylimitationsandbenchmarkingamongstthevariousprocesseswillbediscussedinthefollowingsections.However,asageneralremark,theidentificationofpotentialemissionpatternsandtheEFclassification
forspecificgroupsofprocessesisstillchallenging,mainlyduetodifferencesinmonitoringstrategies,operationalconditionsandlengthofmonitoringperiodsamongtheexistingstudies.Additionally,thereisstilllittlereal-fielddata
regardingN2Oemissionsforseveralconventionalandadvancedbiologicalprocesses(e.g.tricklingfilters,denitrifyingpackedbedreactors,biofilmorhybridpartial-nitritationanammoxsystems,etc.).
3.1EFofsecondaryandsidestreamtreatmentprocessesTheN2Oemissionsofthefull-scalewastewatertreatmentprocessesreportedinpaststudiesvarysignificantly;e.g.rangingfrom0.0025%oftheTN-loadforamainstreamMLEreactor(Spinellietal.,2018)to5.6%oftheTN-
loadforamainstreamaerobic/anoxicsettlingSBRreactor(Sunetal.,2013).Overall,thepotentialofN2Oemissionsfromsidestreamreactors(rangingfrom0.17%to5.1%oftheinfluentN-load–supplementarymaterialB,TableB1)is
considered higher compared to the mainstream BNR processes. The latter is mainly because the nitritation/nitrification occurring during sidestream treatment is linked with higher ammonia oxidation rate (AOR) and NO2−
accumulation(Deslooveretal.,2011;GustavssonandlaCourJansen,2011;Kampschreuretal.,2008).Fig.2showsboxplotsoftheobservedEFs(withrespecttotheinfluentN-load)basedonthetreatmentstep.Thewidthoftheshadedarea
surroundingtheboxplotsrepresentsthedatakerneldensitydistributionoftheEFs.SpecificinformationforthemainstreamandsidestreamtechnologiesincludedinFig.2canbefoundininthesupplementarymaterial(TablesA1,B1
andB2).AverageN2Oemissionsforthestudiedmainstreamprocessesisequalto∼0.87%oftheN-load,whereasthemajorityofthequantifiedEFsarebelow0.27%oftheinfluentN-loadaccordingtoFig.2.Ontheotherhand,Fig.2
showsthatN2OEFsresultingfromthetreatmentoftheanaerobicdigestionsupernatant(sidestreamprocess)arehighlyconcentratedjustbelowthemedian(2%oftheN-load).Onaverage,∼2.1%oftheN-loadisemittedasN2Oin
sidestreamprocesses(supplementarymaterialB,TableB2).Accordingtoalifecycleassessment(LCA)studyquantifyingthedirectGHGemissionsforaWWTPinAustria,thesidestreamDEMONprocesscontributedbyover90%tothe
total directN2O emissions compared to themainstreamBNR (Schaubroeck et al., 2015). However, examples of full-scale sidestream Anammox processes with EFs lower than 1% exist in the literature and demonstrate that the
configurationandefficientoperationalstrategiescanmitigateasignificantpercentageamountoftheN2Oproduced(Jossetal.,2009;Weissenbacheretal.,2010).
Fig.3showstheEFboxplotsofvariousprocessesappliedinWWTPs.Intotal51systemswereconsideredinFig.3(supplementarymaterial,TablesA1andB2).AgeneralremarkisthattheN2OEFforthemajorityofthedifferent
processgroups shown inFig.3 varies from0.01 to2%of theN-load.Discrepancies in the emission loads are observed in themajority of thedifferent processgroups and canbepartially attributed to thedifferent site-specific
operationalcharacteristicsandcontrolparameters.Thisindicatesthatthatapartfromthereactorconfiguration,emissionfluxesdependalsoontheoperational/environmentalconditionsandpreferredenzymaticpathways(Wanetal.,
2019).
Fig.2BoxplotsoftheofthereportedEFswithrespecttothestageofthetreatmentprocesses(i.e.mainstreamorsidestream)usingviolinplotoutlines.Therectanglesrepresenttheinterquartilerange.Themedianisdenotedbytheblackhorizontallinedividingtheboxintwoparts.
Thedotsrepresentthevaluesexceeding1.5timestheinterquartilerange.Theupperandlowerwhiskersstandforvalueshigherorlowertheinterquartilerange,respectively(within1.5timestheinterquartilerangeaboveandbelowthe75thand25thpercentile,respectively).The
violinplotoutlinesshowthekernelprobabilitydensityoftheEFinmainstreamandsidestreamprocesses;thewidthoftheshadedarearepresentstheproportionofthedatalocatedthere.
alt-text:Fig.2
Fig.3BoxplotsvisualizingtheEFrangeforthedifferentgroupsofmainstreamprocesses.Therectanglesrepresenttheinterquartilerange.Themedianisdenotedbytheblackhorizontallinedividingtheboxintwoparts.Thedotsrepresentvaluesexceeding1.5timesthe
MainstreamSBRsaregenerallyassociatedwithhigherN2Oemissionscomparedtotheotherprocessgroups.EFsrangebetween2%oftheinfluentTKN-loadforanSBRoperatingunderaeratedfeeding,aerobic,settlingand
decantingsequences(Foleyetal.,2010)and5.6%oftheinfluentTN-loadforanSBRoperatingunderaeratedfeeding,aerobicandanoxicsettlinganddecantingsequences(1 heach).HighN2OfluxesinSBRsareattributedtosudden
changesintheconcentrationsofNH4+andNO2−withinthecycle(comparedtootherconfigurations)ortoaccumulateddissolvedN2Oduringanoxicsettlinganddecantinginthesubsequentaerobicphase(Pijuanetal.,2014).
ODreactortypeshavebeenlinkedwithrelativelylowN2Oemissions(averageequalto0.14%oftheN-load),probablyduetothestrongdilutionofthereactorconcentrations(veryhighrecyclingrates)andlesssensitivityto
systemshocks.OneexceptionisthestudyofDaelmanetal.(2015)whomonitoredacoveredanaerobic/anoxic/oxicplug-flowreactorfollowedbytwoparallelCarrouselreactorsfor1yearandfoundthatthesystemhadEFequalto2.8%
oftheN-load.TheauthorsarguedthatintheCarrouselreactor,thesurfaceaeratorsledtozoneswithlimitedoxygenconcentrationtoallowforcompletenitrification(leadingtonitriteNO2−accumulation),whereastheanoxiczones
werealsolimitedtoallowforcompletedenitrification.
CASsystemsshowninFig.3consistofaerobicreactors(1-stepfeedormultiplestep-feed)withoutdedicatedanoxiczonesfordenitrification.TheyarecharacterisedbyaverageEFequalto0.27%oftheN-load,whereasthe
NH4+removal,rangesbetween38%and53%.PeakloadsandrecirculationoftheanaerobicsupernatantcanberesponsiblefortheN2OfluxesobservedinCASsystems,whereashighaerationrateshavebeenreported,enhancingN2O
stripping(Chenetal.,2016).
MLEconfigurationshaveamedianEFequalto0.857%oftheN-load.MLEprocesseswithhighEFs(upto4%oftheN-load)havebeenreportedbyFoleyetal.(2010).LowEFsinMLEconfigurations(i.e.0.003%,0.065%)have
beenobservedinreactorswithdilutedinfluentconcentrationsduetogroundwaterinfiltration(Bellandietal.,2018),nitrificationefficiencylessthan73%(Ahnetal.,2010b)andlowTNremoval(∼59%)duetoCOD/TN < 1.9(Spinellietal.,
2018).LowEFinMLEreactorsrangingfrom0.003%to0.065%oftheNH4+loadhavebeenalsoreportedinthestudiesofCaivanoetal.(2017)andBellandietal.(2018);however,conversionoftheseEFto%N-loadwasnotpossibleand
werenotincludedinFig.3.
N2OemissionfluxesinA2/Oconfigurations,arerelativelylow,inthemajorityofthestudies,withmedianequalto0.1%oftheN-load.OneexceptionisthestudyofWangetal.(2016b);theymonitoredanA2/Oreactoronceper
monthfor1yearandshowedthattheEFvariedfrom0.1to3.4%oftheN-loadbetweendifferentmonths.TheDOconcentrationandoperatingconditionsvariedsignificantlyinthereactor(i.e.DOrangedfrom0.6to6.8 mg L−1).
LimitedN2Omonitoringstudiesexistinfull-scalesidestreamprocesses.One-stagegranularanammoxreactorshaveanaverageEFof1.1%oftheN-load.Thesametwo-stagesuspendedbiomasspartial-nitritationandanammox
processhasbeenmonitoredintwostudies(Kampschreuretal.,2008;Mampaeyetal.,2016).Inthesestudies,theaverageEFinthepartial-nitritationSHARONreactorwas∼2.8%oftheN-loadandwaselevatedcomparedtofull-scale
one-stageanammoxreactors.N2Ofluxesquantification,inlabandpilot-scalesingle-stagegranularanammoxreactorshaveshownEFsrangingfrom0.1to12.19%ofN-load(Wanetal.,2019).Therefore,morestudiesarerequiredto
establishreliablerangesofEFsinsidestreamprocesses.
DifferencesinthereportedN2Ofluxesarealsoobservedinstudiesthatapplysimilarconfigurationsandoperationalconditions(supplementarymaterialB,TablesB1andB2).Forexample,Kampschreuretal.(2008)andMampaey
etal.(2016)monitoredtheN2Oemissionsinthesametwo-stepSHARON-AnammoxreactorsystemandobservedEFsthatwereequalto1.7%and3.8%oftheN-load,respectively.ApartfromslightlydifferentDOsetpoints(2 mg L−1in
theworkofMampaeyetal.(2016)and2.5 mg L−1intheworkofKampschreuretal.(2008),theoperationalconditions(i.e.temperature,influentN-load,systemtreatmentefficiency,hydraulicretentiontime(HRT),etc.;TableB1)during
thetwomonitoringperiodswerequitesimilar.ThemainidentifieddifferencewastheincreasedanoxicliquidN2Oformationduringtheanaerobicperiodofthepartial-nitritationreactorinthestudyofMampaeyetal.(2016).Intermsof
monitoringprotocols,Kampschreuretal.(2008)collectedgrab-samplesforapproximately3days,whereascontinuousgasmonitoringfor21dayswasconductedbyMampaeyetal. (2016).Moreover,theair infiltrationinthecovered
reactorduetonegativepressurewasnotconsideredinthestudyofKampschreuretal.(2008).
TheaforementionedexamplesemphasizethedifficultyinthecomparisonofGHGemissionsamongvariousstudiesanddevelopmentofEFdatabasesforprocessgroups.ThebenchmarkingofGHGemissionsofdifferentplants
canbehamperedevenifthesamemonitoringprotocolisappliedtomonitorprocessesbelongingtothesamegroup.Therearealsocaseswhereemissionshavebeenmeasuredaccordingtodifferentanalyticalprocedureswithinthe
samesystem;factthataddsfurtherdifficultyinthecomparisons.TherelativelyshortmonitoringperiodsandvaryingmonitoringstrategiescanalsoinfluencethecomparabilityandaccuracyofthereportedEFs.Thiswillbediscussed
indetailinthefollowingsections.
3.2DurationofmonitoringcampaignsandseasonalitySeasonalenvironmentalvariabilities,suchastemperature,caninfluencethebacterialcommunitystructureinWWTPs(Flowersetal.,2013).TheN2OformationandemissionsduringtheBNRareexpectedtohavetemporal
interquartilerange.Theupperandlowerwhiskersrepresentvalueshigherorlowertheinterquartilerange,respectively(within1.5timestheinterquartilerangeaboveandbelowthe75thand25thpercentile,respectively).
alt-text:Fig.3
variations.TemperaturecansignificantlyaffecttheAOBspecificgrowthrateduringnitrification(VanHulleetal.,2010).ThehighertemperaturealsodecreasestheN2Osolubility,thusintensifyingtheN2Ostrippingtotheatmosphere
(Reinoetal.,2017).Ontheotherhand,Adouanietal.(2015)observedthattheN2Oemissionsincreasedupto13%,40%and82%oftheTN-removedattemperaturesequalto20 °C,10 °Cand5 °C,respectively,inabatchreactorfedwith
syntheticwastewater.ThelatterwasattributedtotheincreasedsensitivityoftheN2Oreductaseactivitiesatlowertemperaturescomparedtootherdenitrificationenzymesand,therefore,toincompletedenitrification.Otherseasonal
variations(e.g.influentloading,wetanddryseason)canalsoimpactontheenzymaticreactionsandaffecttheemissions.Vasilakietal.(2018)observedpeaksofN2Oemissionscoincidingwithprecipitationevents,atlowtemperatures,
inanODduringa15-monthmonitoringcampaign.However,furtherinvestigationisrequiredtounderstandpotentialseasonaleffectsontheN2Oemissions.
Themonitoringperiodsofthefull-scalecampaignsforalltheprocessesconsideredinthisanalysisaresummarizedinsupplementarymaterialB,TablesB1andB2.Fig.4showstheEFformainstreamtechnologiesbasedonthe
lengthofthemonitoringperiod.OnlystudiesthathavereportedEFsintermsoftheinfluentN-loadhavebeenconsidered(supplementarymaterialA-TableA1).Themonitoringcampaignshavebeencategorizedinto3distinctgroups,
basedontheirduration,i)short-termcampaignsperformedinalimitedperiodoftime(lessthan1month),ii)medium-termmonitoringcampaignsthatlastmorelongerthanonemonthbuthavenotcapturedallthetemperatureranges
observedinthesystem,andii) long-termmonitoringcampaignsthat lastat least1year.Bothcontinuousanddiscontinuousmonitoringstudieshavebeenincludedintheanalysis.InthediscontinuousN2Omonitoringstudies, the
gaseousN2Ofluxeshavebeengrab-sampled(i.e.viagas-bags,closedchambersetc.)andsubsequentlyquantifiedusinganalyticalmethodsinthelab(offlinemonitoring)orintermittentlysampled(i.e.withfloatingchambersfor1–2
days/month)butquantifiedcontinuouslyon-site(i.e.viaGHGanalyzers) (onlinemonitoring).Mostdiscontinuousmonitoringcampaignshadmonthlyorbi-monthlysamplingfrequency.Theonlyexception is thestudyof (Ahnetal.,
2010b)Ahnetal.,(2016b)whomonitoredseveralsystemsinonlytwodistinctseasons(warmestandcoldesttemperatures).Discontinuousstudieswithsamplingextendingover1-monthperiod,havebeencategorizedasmedium-termor
long-termbasedonthedurationofthestudy(lessormorethan1year).IncontinuousmonitoringcampaignsN2Ofluxeshavebeencollectedcontinuously(i.e.viachambers)andquantifiedonline,on-site(i.e.viaGHGanalysers).
About30%oftheEFsshowninFig.4arebasedonmonitoringperiodslastinglessthantwodays(Foleyetal.,2010;Filalietal.,2013;Samuelssonetal.,2018).AnnualEFvariationhasbeeninvestigateddiscontinuously(monthlyor
bimonthlysamplingfrequency)inapproximately10%ofthesesystems(i.e.Sunetal.,2015;Wangetal.,2016b),whereaslong-term(≥1year)continuousmonitoringcampaignshavebeenperformedonlyintwooftheexaminedworks
(Daelmanetal.,2015;Kosonenetal.,2016).SBRshavebeenexcludedinordertoavoidfurtherbiasesintheresults,sincethereportedaverageN2OemissionsaresignificantlyhigherthantheaverageEFofothermainstreamN-removal
configurations.Sidestreamtechnologieshavebeenalsoexcludedbecausetheyhavenotbeenmonitoredlong-termtoexamineseasonaleffects.TheaverageEFisequalto0.8%(median0.2%)and0.3%(median0.1%)oftheN-loadfor
thesystemsmonitoredshort-termandmedium-term(butwithoutcapturingthewholespectrumofseasonalityeffects),respectively.ThestudiesinvestigatingseasonaltrendsofN2OemissionsreportedanaverageEFof1.5%(median
1.7%)oftheN-load.
Daelmanetal.(2013b)demonstratedthatshort-termcampaigns,inthesysteminvestigated,arelikelytoproduceunreliableEFestimatesindependentlyofthemonitoringapproach.Additionally,theauthorsfoundthatshort-term
campaignshaveahighprobabilitytounderestimateactualemissions.AccordingtoFig.4,thehighestgaseousN2Oloadsbelongtolong-termcontinuousordiscontinuousmonitoringcampaigns.
Long-termandmedium-termcampaignshavealsoshownahighvariabilityof the reportedN2Oemissions.Amongst theexaminedstudies,Daelmanet al. (2015) implemented the longest continuous real-field campaign that
reinforcedtheexistenceofseasonalemissionvariability.Seasonalityisalsosupportedbythefindingsofseveralotherstudies(Brottoetal.,2015;Sunetal.,2013;Yanetal.,2014).Bollonetal.(2016a)Bollonetal.,(2016b)andBollonetal.
(2016b)Bollonetal.,2016a)studiedasidestreamnitrifyingandpost-denitrifyingbiofiltrationsystem,respectively,byperformingtwomonitoringcampaigns;oneinsummerat22.5 °Candoneinwinterattemperatureslowerthan14 °C.
TheirresultsindicatedasignificantseasonalvariationoftheN2Oformation.TheEFofthenitrifyingfilterswasequalto2.3%oftheNH4+removedduringthesummercampaign,and4.9%oftheNH4+removedduringthewinter
campaign.ThedissolvedN2Oconcentrationinthepost-denitrifyingbiofiltereffluentwasequalto1.3%insummerand0.2%inwinterwithrespecttotheNO3−uptake.
Short-termmonitoringperiodsarelikelytomissunderlyingseasonalvariationsintheN2Oformation(orbeaffectedbyshort-termprocessperturbations),and,consequently,complicatethedirectcross-comparisonsbetween
differentstudiesandtheirfindings.Forinstance,themonitoringofanA2/Oprocessforninemonths(grab-samplestakenoncepermonth)ledtoanaverageEFequalto0.08%oftheinfluentTN(Yanetal.,2014),whereasasimilarA2/O
processmonitoredfortwodays(bytakinggrab-samples)presentedanEFof0.85%oftheinfluentTKN(Foleyetal.,2010).Wangetal.(2016b)showedthattheEFfromaPFA2/Oreactorwascharacterizedbysignificantseasonalityand
Fig.4EFvalueswithrespecttothelengthofthemonitoringperiodformainstreamtreatmenttechnologies.
alt-text:Fig.4
variedfrom0.01%to3.5%oftheinfluentTN;withintherangeofEFsreportedinthestudiesbyYanetal.(2014)andFoleyetal.(2010).ItcanbeconcludedthattheEFdifferencesbetweensimilarconfigurationsshowninFigs.2and3,
arestronglyaffectedfrombytheseasonalityoftheemissions.
3.3MonitoringandsamplingmethodsSeveralauthorshavehighlightedthatthesamplingmethodologycaninfluencetheEFquantification(Daelmanetal.,2013b;Aboobakaretal.,2013;Wangetal.,2016a;Kosonenetal.,2016;Hwangetal.,2016).Forinstance,several
sourcesofuncertaintyinchambertechniquessimilartothetechniquesappliedinthewastewatersector,havebeenidentifiedinGHGmonitoringcampaignsofrunningwaters(Ducheminetal.,1999;Lorkeetal.,2015;Matthewsetal.,
2003;Vachonetal.,2010).Adetailedanalysisofthepotentialuncertaintiesrelatedtothedifferentsamplingmethodswasoutofthescopeofthisreview.However,itisimportanttounderlinethatfurtherresearchisneededtorevealthe
influenceofdifferentsamplingmethodsandfacilitatethebenchmarkingofEFvaluesfordifferentgroupsofprocesses.
Thissectionfocusesonthesamplingmethod(continuousanddiscontinuous).ThemaincharacteristicsofthesamplingstrategiesareshowninsupplementarymaterialB,TableB1andB2.
Fig.5 illustrates theboxplotof theaverageEFofmainstreamprocesses forcasesof continuousgaseousmonitoringusingagasanalyzeranalyser versus theboxplot for studieswith intermittent samplingcampaigns.Only
medium-termandlong-termstudieshavebeenconsideredintheanalysis(supplementarymaterialA–TableA1).MainstreamprocessesmonitoreddiscontinuouslyexhibitedanaverageEFof0.44%oftheN-load(medianEFwas0.2%of
theN-load),whereasprocessesmonitoredcontinuouslywithgasanalyzersanalysershadanaverageEFequalto1.2%oftheN-load(medianEFis1.1%).Themajorityoftheprocessmonitoredintermittently(onceortwicepermonth)
havecollectedgrab-samplesofN2O fluxes (supplementarymaterialB,TableB2).Offlinegrab sampling is often characterisedby time limitations; usually the sampling occursduringWWTPoperating times andprovidesdiscrete
measurements(e.g.Wangetal.,2011)thatareunabletocapturethewholespectrumofdiurnalvariabilities(Daelmanetal.,2015;Wangetal.,2016a).Additionally,thetemporalvariabilityofN2Oemissionsishighlydynamic(i.e.Daelman
etal.,2015)andstronglyaffectedbyoperationalconditions.Therefore,low-frequency,long-termsamplingmightnotcaptureadequatelythewholerangeofN2Oemissionsinducedbyshort-termchangesinoperationalconditionsand
pollutant concentrations. Overall, the differences can be attributed to the case-specific nature of the EF, as well as to the restrictions concerning the duration and frequency of the discontinuous campaigns (difficulty in
performingcollectinggrab-samplesforlongerperiods-wholedays,nighttime,weekends,etc.).
3.4TowardsbenchmarkingofEFs:progressandlimitationsTheamountofquantifiedemissionsishighlyaffectedbyavarietyofparameters(e.g.processtype,WWTPcharacteristics,monitoringstrategy,durationofmonitoringcampaign,etc.).Therefore,estimatingtheN2OEFs in
WWTPswitheitherofflineoronlinemonitoringcampaignsremainschallenging.Inthissection,effortshavebeenfocusedontheclassificationofEFsfordifferentgroupsofprocessesbasedontheNH4+removalefficiencyandthe
influentflow-rate.Additionally,anan-N2OmonitoringframeworkforthedevelopmentofcomparableEFsforthewastewatersectorisalsodiscussed.
Fig.6showstheEF(withrespecttotheinfluentN-content)formainstreamprocessesandtheachievedNH4+removalefficiency(%).Thedifferentprocessesarerepresentedbydifferentcolours.Adetailedlistoftheexamined
studiesisprovidedinthesupplementarymaterialB(TableB2).Trianglesshowthatseasonaleffectshavebeeninvestigatedintherespectiveprocess,whereascirclesrepresentshort-termstudiesthathavenotinvestigatedseasonality.
Fig.5BoxplotsoftheaverageEFwithrespecttothemethodofgaseoussamplingformedium-termandlong-termstudies(D:discontinuousmonitoring,C:continuousmonitoring)formainstreamtreatmentprocesses.Therectanglesrepresenttheinterquartilerange.Themedianis
representedbytheblackhorizontallinedividingtheboxintwoparts.Thedotsrepresentvaluesexceeding1.5timestheinterquartilerange.Theupperandlowerwhiskersrepresentvalueshigherorlowertheinterquartilerange,respectively(within1.5timestheinterquartilerange
aboveandbelowthe75thand25thpercentile,respectively).
alt-text:Fig.5
ThesizeofthedatapointsrepresentsthesizeoftheWWTP.NospecifictrendwasidentifiedbetweentheobservedEFsandNH4+removalforthedifferentmainstreamprocesses.TheN2Oemissionsformostoftheprocessesinsmaller
WWTPs(influentflow-rate<200,000 m3 d−1)werelessthan0.5%oftheN-load,independentlyontheprocesstypeandnitrificationefficiency.Inaddition,mostoftheprocesseswithN2Oemissionslessthan0.1%oftheN-loadreferred
tostudiesperformingshort-termandmedium-termN2Omonitoringcampaigns.
In-depthcomparisonsrequiremoredetailsonconfigurations,controlstrategiesandoperationalconditions.Currently,therearestillnospecificguidelinestostandardizethereportingofoperational,processandmonitoring
strategyinformationfromexistingstudies.ThesupplementarymaterialBsummarizesprocess-basedinformationfrompastfull-scaleN2Omonitoringcampaigns.Overall,∼70%ofthemainstreamstudieshavereportedtheEFsinterms
ofN-load.Additionally,influentandeffluentNH4+concentrationsareavailablefor∼35%ofthesystemsanalysed.Informationonthewatertemperatureduringthemonitoringcampaignhasnotbeenprovidedforhalfofthestudies.
Limitedstudiesprovidedinformationonthecontrolstrategyofthesystem(i.e.DOset-point)andotheroperationalparametersoftheprocesses(i.e.HRT,SRT).GiventhevariabilityofEFsamongstsimilarprocessgroups(Figs.2–6),
theidentificationofEFpatternsneedstoconsiderprocess-specificoperationalandenvironmentalinformation.
Themonitoring strategies require sampling protocols that are case-specific (e.g. the choice of appropriate sampling locations). Specific protocols for the design ofmonitoring strategies can be found in the study of van
Loosdrecht et al. (2016). Elemental mass balances can also be used to confirm the validity of the measurements (Castro-Barros et al., 2015;Mampaey et al., 2016) independently of the applied monitoring protocol. As shown in
supplementarymaterialB,TableB2,grab-samplingthegaseousfluxesoftenlackstheacquisitionofweekendornight-timesamples,thusfailingtodepictthediurnalvariabilityoftheemissions.Moreover,short-termmonitoringstudies
arefrequentlyunabletoaccuratelycapturethetemporalN2Odynamics(Daelmanetal.,2015;Kosonenetal.,2016).Daelmanetal.(2013b)concludedthattheaccuratequantificationoftheaverageN2Oemissionsrequireslong-termonline
orgrab-samplingmonitoringcampaignsthatconsidertheseasonalvariationsoftemperature.
Theanalysisofhistoricalprocessandplantdatacanbealsouseful,linkingtheemissionswithspecificandreoccurringoperationalandenvironmentalconditions(i.e.e.g.dryvswetweather,temperature)forshort-termand
long-termmonitoringcampaigns.Relationshipsestablishedduringshort-termmonitoringcampaignscanbelinkedwiththeperiodicoperationalandenvironmentalprocessconditionsandcannotbegeneralizedtounderstandthelong-
termN2Odynamics(Vasilakietal.,2018).Thelatterisimportantduetotherestrictionsonthedurationofthemonitoringcampaignsduetoresultingfromtheentailedcosts.Itisalsoessentialtoidentifyandreportprocessperturbations
thatcanaffecttheN2Oemissionsevenonalong-termbasis(Vasilakietal.,2018).
ThecurrentanalysisshowsthatN2Ofluxesmustbereportedtogetherwiththeconfigurationtype,theseasonaloperatingconditions(e.g.pH,temperature,influentTN,effluentTN,PE,wastewatervolume,COD,mixedliquor
suspendedsolids(MLSS),SRT,HRT,recycleratios,etc.).
4N2OmonitoringcampaignsandN2OdynamicsSeveralmethodsthathavebeenappliedinN2OmonitoringcampaignscanincreasetheunderstandingoftheN2Ogenerationinfull-scaleprocesses.Theseinclude:i)techniquesforthetranslationofWWTPoperationalandN2O
dataintoinformation(e.g.graphicalrepresentationofvariables,featureextractiontechniques,multivariateanalysis,etc.),ii)mechanisticmodelssimulatingtheN2Odynamics,andiii)techniquesforunveilingtherelativecontribution
Fig.6EFofthemainstreamtechnologieswithrespecttotheachievedNH4+removal.Thedifferentcoloursrepresentdifferentprocesses,whereasthedifferentshapesdifferentiateshort-term,medium-termandlong-termstudies.Thesizeofthedatapointsdepictsthesizeofthe
WWTPintermsofinfluentflow-rate.(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtotheWebversionofthisarticle.)
alt-text:Fig.6
ofdifferentpathwaystotheemissions(e.g.isotopicanalysis,real-timepolymerasechainreaction(qPCR)etc.).
TheeffectofseveralparametersthataresignificantforN2Ogeneration(i.e.DO,COD/N,pH,temperature)hasbeenextensivelyinvestigated(Kampschreuretal.,2009b;Lawetal.,2012;Massaraetal.,2017)basedonlab-
scale, pilot scale and full-scalemainstreamand sidestreamprocesses. This section aims to complement these studies; themain findings of the full-scale technologies are categorized for different processgroups focusingon the
techniquesthathavebeenappliedandtheircontributiontotheunderstandingofthebehaviourofN2Oemissions.
4.1Overviewofthetechniques4.1.1TechniquestranslatingWWTPoperationalandN2Odataintoinformation
Morethan40physical,physico-chemicalandbiochemicalvariables(e.g.temperature, flow-rates,reduction-oxidation(redox)potential,DO,N-compoundsandorganicmatterconcentrations,alkalinity,etc.)canbemonitoredonlinetoevaluate
processperformance(VanrolleghemandLee,2003).Onlinemonitoredvariableswhencombinedwithlaboratoryanalysescanprovideuseful insightintotheN2Opatternsandbehaviour.TechniquesthathavebeenappliedtotranslateWWTPdatainto
informationinfull-scaleN2Omonitoringcampaignsinclude:i)graphicalrepresentationsandsimplefeatureextractionmethodsandii)statisticalanalysisanddataminingmethods.
Inmostofthestudies,theonlineandlaboratorydatautilizationhasbeenlimitedtotheinvestigationandgraphicalrepresentationofthesignificantparameters’profiles(i.e.DO,NO2−,NH4+,aerationflow-rate)incombinationwiththeresponseof
theN2Oemissionbehaviour.Additionally,descriptivestatistics(i.e.centraltendency,dispersion,position,etc.)oftheprocessvariablesandN2Oemissionsiscommonlyanalysedandreported.
CorrelationanalysisandlinearmultivariateregressionmodelsarethemainstatisticaltechniquesthathavebeenusedtorevealtheN2Oemissions’dependencieswithoperationalvariablesinfull-scalesystems(i.e.Brottoetal.,2015;Bollonetal.,
2016a;Aboobakaretal.,2013).Dimensionalityreductiontechniques(e.g.principalcomponentanalysis(PCA),independentcomponentanalysis(ICA)),clustering(e.g.hierarchical,k-means),linearandnon-linearsupervisedlearningtechniques(e.g.partial
leastsquares(PLS),artificialneuralnetworks(ANN))andsupportvectormachines(SVM))arealsopowerfultoolsutilisedtotransformtheWWTPdataintoknowledge(Haimietal.,2013;Corominasetal.,2018).However,advancedinformationextraction
methodshaverarelybeenusedtoanalysedatafromN2Omonitoringcampaigns.Recently,Sunetal.(2017a,b)constructedaback-propagationANNtosimulateN2Oemissionsinananaerobic-oxic(A/O)process;thusdemonstratingthefeasibilityand
simplicityofpredictingN2Oemissionswithdata-drivenmodels.
4.1.2Experimentalstudiesinfull-scalesystems–modificationofoperationalconditionsSeveralN2Omonitoringcampaignsinfull-scalesidestreamprocesseshavetesteddifferentoperationalconditionstoinvestigatetheirimpactontheemissions(Bollonetal.,2016b;Castro-Barrosetal.,2015;Mampaeyetal.,2016).Themajorityof
thestudieshavefocusedoninducingchangesinthedurationandflow-rateofaerationcomparedtothebaselinecontrolstrategyoftheexaminedreactors.
4.1.3TechniquesunveilingtherelativecontributionofdifferentpathwaystotheemissionsIsotopicandmolecularbiologyanalysisareemergingtechniquesthatcanprovideinsightsintotheN2Ogenerationpathways.Molecularbiologymethods(e.g.quantitativereversetranscriptionpolymerasechainreactionRT-qPCR,FISH,etc.)can
quantifythemicrobiologicalstructuredrivingtheN-cycleandthebacterialpopulationabletoreduceN2OatWWTPsundervariousenvironmentalandoperationalconditions(Castellano-Hinojosaetal.,2018;Songetal.,2014).Isotopetechniqueshaveonly
recentlybeenimplementedatfull-scalesystemstodistinguishtherespectivecontributionamongtheN2OpathwaysandincreasetheunderstandingofthepathwaysthatareresponsiblefortheN2Oformation(Townsend-Smalletal.,2011;Tumendelgeret
al.,2014).ArecentcriticalevaluationofnaturalabundanceandlabelledisotopesforN2OstudiescanbefoundinthestudyofDuanetal.(2017).
4.1.4MechanisticmodelsThemechanisticmodelsareapopulartoolforthepredictionoftheN2OgenerationandemissionduringtheBNRinWWTPs.Basedondifferentassumptions,avarietyofone-andmultiple-pathwaymodelshavebeensuggested.Theirstructureis
basedeitheronthewidelyacceptedASMlayout(assuggestedbyHenzeetal.,1987,2000),oronthemorerecentelectroncarrierconceptthatdescribestheN2Oproductionviathemechanismoftherelevantcomplexoxidation-reductionreactionstaking
placeduringwastewatertreatment(e.g.Nietal.,2014).Allmodels,though,considertheeffectofchangingoperationalparameters(e.g.DO,NO2−levels,aerationregime,etc.)ontheN2Ogeneration.Acomprehensiveevaluationofthedifferentmodelling
approaches,underlyingassumptions,kinetics,stoichiometricparameters,calibrationandvalidationproceduresofseveralsingle-pathwayandtwo-pathwayAOBmodels,heterotrophicdenitrificationpathwaymodels,andintegratedN2Omodelsdescribing
theall threemajormicrobiologicalpathways isprovidedbyNiandYuan(2015).Theauthorshaveprovidedguidelinesfortheselectionof themostappropriatemodellingapproachunderdifferentDOandNO2−concentrationsbasedon thestructural
assumptionsofthemodels.ThedebateonthemodelthatbestdescribesanddecouplesthemajorN2Oformationpathwaysisstillongoingwithseveralextensionsandvariationsoftheoriginalapproachesdevelopedrecently(Dingetal.,2017;Domingo-
FélezandSmets,2016;Massaraetal.,2018).
4.2Process-basedinsightsbasedontheappliedtechniques
Robustdocumentationofthedominantpathwaysamongthedifferentprocessconfigurationsisstillmissing(Maetal.,2017).ThissectiondiscussescorrelationsbetweenN2Oemissionsandoperationalvariablesanddominant
N2Opathwaysthathavebeenidentifiedfordifferentfull-scaleprocessgroups.Table1providesasummaryofthedominantN2Opathwaysthathavebeenreportedfordifferentwastewatertreatmentprocessesbasedonthetechniques
thathavebeenappliedinthemonitoringcampaigns.StudiesthathavenotdiscussedpossibleN2Opathwayshavenotbeenconsidered.
Table1MainfindingsofpaststudiesthatresultintheidentificationofthemostcontributiveN2Oproductionpathway(wherepossible).a:Visualizationofsignificantprofiles&descriptiveanalysis,b:Modified
operationmode,c:Statisticalanalysisanddatamining,d:Mechanisticmodeldevelopment,e:Isotopicanalysis,f:real-timeqPCR.
alt-text:Table1
Source Process NH2OHoxidation Nitrifierdenitrification Heterotrophicdenitrification
Castro-Barrosetal.,2015a,b
One-stagePNAgranularSidestream N2Oemissionselevatedduringshiftsfromlowtohighaeration(NH4+accumulation,highAOR)→mainpathway
NotamainpathwayN2Oemissionselevatedduringshiftsfromlowtohighaeration(NH4+accumulation,highAOR)→potentialcontributor
Mampaeyetal.,2016a,b
One-stageSHARONgranularReactorSidestream Notamainpathway PresenceofNH2OHinanoxicperiods;lowerDO
resultinginincreasedN2OemissionsN2OformationduringanoxicperiodsunderthepresenceofNO2−&smallamountsoforganicsubstrate
Kampschreuretal.,2008a
Two-reactorpartial-nitritation-anammoxprocessSidestream
Excludedduringanoxicconditionsinthenitritationreactor(despitesignificantN2Oformation)
• Anammoxreactor:absenceofO2• Nitritationreactor:emissionsnotaffectedbytheinfluentcomposition(thereforeC/N
ratio)
Notamainpathway
Stenströmetal.,2014a
Nitrification-denitrificationSBR,Sidestream Notdiscussed ConsiderableN2OformationunderDO = 0.5 mg L−1&
NO2−accumulation(>20 mg L−1)
N2Oaccumulationduringdenitrification(underconditionsoflowCOD:NandhighNO2−);quicklystrippedofftotheatmosphereassoonasaerationresumed
Wangetal.,2016ba,c,f A2/Owithplug-flowpattern Notadominantpathway(lowNH4+concentrations)
• CoexistenceofNO2−,NH4+&O2-limitingconditions
• CorrelationbetweenNO2−andN2Oemissions
• StrongresponsesbetweenNOandN2Oemissionsandtherelativeabundanceof
AOB.
Notadominantpathway(nopeaksobservedafteranoxiczones)
Wangetal.,2011a A2/O Notamainpathway
• RapidlyincreasedN2OemissionsduetoDOlimitation(DO < 2.5 mg L−1);maximumN2OemissionatDO = 0.75 mg L−1
• IncreaseinNO2−concentration(from0.2toNotamainpathway
0.6 mg L−1)duringnitrificationleadingtoincreaseinN2Ofluxes
Toyodaetal.,2011a,f A2/O
SPIsotopicanalysis:
• ∼50%contributioninthebeginningofaerobictank
SPIsotopicanalysis:
• ∼50%contributioninthebeginningofaerobictank
• Dominantpathwayfrommiddletotheendofaerobictank
N2Owasproducedduringdenitrification
Aboobakaretal.,2013a,c A/Oplug-flowreactor Notdominantpathway ConsidereddominantinzoneswithDO < 1.5 mg L−1
ConsidereddominantinzoneswithdepletedNH4+,DOfluctuations,NO3−availability&lackofNO2−
Sunetal.,2017a,bSunetal.,2017aa
A/OHigherDO:certainNO2−amountpotentiallyutilizedtooxidizeNH3toNH2OH,thusleadingtotheN2Oproduction
Low-DOcondition(i.e.<1 mg L−1)usuallyobservedatthebeginningoftheoxiczone Notamainpathway
Kosonenetal.,2016a,c&Blombergetal.,2018d
A/ObioreactorOnlythisAOBpathwaymodelledduetotheexistingDO&NO2−conditions,N2Oproductionmainlyintheaeratedzones,N2Oconsumptionintheanoxiczones→mainpathway
• IncreasingthenumberofnitrifyingzonesresultinginhigheroverallN2Oemissions
(N2Oproductionpossiblyvianitrifier
denitrification)
• Giventhatanoxic-aerobicvolumecontrolledbytheNH4+concentration,unclearif
increasedemissionscausedbyincreased
NH4+concentrationorincreasednumberof
nitrifyingzones
Notmainpathway
Panetal.(2016)&Nietal.,(2015)a,d
2step-feeding,anoxic/oxic/anoxic/oxicplug-flowreactor
N2OemissionsincreasingwiththeAORincrease(2ndstepoftheplug-flowreactor)
N2OemissionsincreasingwiththeAORincrease(2ndstepoftheplug-flowreactor) Notamainpathway
Rodriguez-Caballeroetal.2014a,b
Anoxic/oxic/shortanoxic/oxicplug-flowreactor Notamainpathway
• N2Opeakswhentransitioningfromanoxictoaerobicconditions;O2limitationconsidered
asenhancingtheactivationofthenitrifier
denitrificationpathway
• N2Oemissionsincreasingwithpotentialshockloads;theAOBlikelytoactivatetheir
denitrificationpathwayaftershockloadsof
toxiccompounds
Notamainpathway
Castellano-Hinojosaetal.,2018c,e
Twosequentialbioreactors(anoxicandaerated) Notamainpathway
• StrongpositivecorrelationbetweenAOBabundance&N2Oemission;hence,more
possiblepathwayunderanoxicconditions
• 0.5 < DO < 1 mg L−1:enoughO2providedtotheAOBfortheoxygenationofNH3to
NH2OHbutnotforaerobicrespiration;NO2−
potentiallyusedasalternativeelectron
acceptortocompletenitrification
Notamainpathway
Tumendelgeretal.,2014a,f CAS
SPIsotopicanalysis:
• Upto90%contributionatDO∼2.5 mg L−1
• ∼50%contributionatDO∼1.5 mg L−1
SPIsotopicanalysis:
• DominatedatDO < 1.5 mg L−1
• ∼50%contributionatDO∼1.5 mg L−1Notdiscussed
Daelmanetal.,2015a,c Carrouselreactor
Carrousel:emissionscoincidingwithaeratedperiods(AORgovernedbyDO);therelationshipbetweentheAOR&theN2OproductionusuallyexplainedbyreferringtotheNH2OHpathway;however,notconsidereddominant
• Carrousel:emissionscorrelatedwiththeNO2-concentrationpeaks
• Prevalenceoflow-DOzones
Carrousel:reactorlackingsufficientanoxicspacetoallowthecompletionofdenitrification
Nietal.,2013d ODwithsurfaceaerators
MainpathwaysincehighNH4+concentrationswereobservedwithoutsimultaneousNO2−increaseintheaeratedzones/phases
Notamainpathway Notamainpathway
Nietal.,2013d
Feedingandaeration(90 min)/settling(35 min)/decanting(55 min)SBR
MainpathwaysincehighNH4+concentrationswereobservedwithoutsimultaneousNO2−increaseintheaeratedzones/phases
Notamainpathway Notamainpathway
Sunetal.,2013a,c
Feeding(synchronousaeration)/aeration/settling/decantingSBR(1 heach)
Notamainpathway
• LowDOduringnitrificationsignificantlyaffectingtheN2Oproduction→mainpathway
CorrelationbetweenN2OemissionandinfluentCOD/N→contributor
Rodriguez-Caballeroetal.,2015a,b
Reactionphase(∼130 min)/settling(∼65min)anddecanting(∼65 min)SBR(anoxic/aerobicalternations–3cycletypes)
Notamainpathway
• CertainNO2−accumulationunderaerobicconditions
• N2Ogenerationcontinuingafteraerationstop
N2Ogenerationcontinuingafteraerationstop
• SignificantlinearcorrelationbetweenN2O&• SignificantlinearcorrelationbetweenN2O&NOEFsin
Wangetal.,2016aa,c
Full-scalebiologicalaeratedfilter(BAF)forsecondarynitrification
LowinfluentNH3concentration(<6 mg L−1)→notamainpathway
NOEFsindifferentseasons
• Nitrifierdenitrificationsuggestedaspossiblepathwayinaccordancewiththefactthatthe
influentNO2−foundaskeyfactorregarding
theN2O&NOproduction
differentseasons
• Minorpossibilityofheterotrophicdenitrification
contributionduringthe
denitrificationofNOtoN2O
Bollonetal.,2016a,bBollonetal.,2016ba
Nitrifyingbiofiltration(Biostyr®filters)
Notdiscussed Notdiscussed
• RapidincreaseinthenetN2Oproductionrateat
BOD:N < 3
• Intensityincreasedwiththedurationofcarbon-limiting
conditions
Overall,themajorityofthestudiesinvestigatingtheN2Odominantpathwaysinmainstreamfull-scalesystemswithdescriptivestatisticsandvisualinspection(i.e.viaunivariate/bivariategraphs)ofsignificantvariables(e.g.
NH4+,DOconcentrations,influentflow-rateandN2Oemissions)havenotconsideredNH2OHtobeasignificantpathwayforN2Ogeneration,regardlesstheconfiguration(Table1).
Specifically,in1-stepfeedA/OandA2/Oconfigurationsandprocesseswithanoxic/aerobicalternationsandplug-flowpattern,studieshaveobserved:i)spatialN2Opeaksinthetransitionsfromanoxictoaerobiczones(i.e.
Rodriguez-Caballeroetal.,2014;Sunetal.,2017a,bSunetal.,2017a)underlowDOconcentrations(<1 mg L−1),ii)temporalincreaseintheN2OemissionsthatcoincidewithelevationofincreasesintheNO2−andNH4+concentrations(i.e.
peak loads)underO2oxygen-limitingconditions (Wanget al., 2011,2016b), and iii)N2Oemission peaks coincidingwith elevatedNO2−concentrations in theaerobic zones (Sun et al., 2017a,bSun et al., 2017a). Therefore, the nitrifier
denitrificationpathwayhasbeensuggestedtobedominant.ThisisalsosupportedbyWangetal.(2016b)whostudiedtherelativeabundancyoftheAOBandthedenitrifyingbacteriaunderdifferentseasonalconditionsforanA2/O
processwith a plug-flow pattern (DO 0.6–6.8 mg L−1 andN-concentration up to 30 mg L−1) to provide insights on the N2O generation pathways. The authors quantified the expression of functional genes harboring the NH3
monooxygenase(amoA)(i.e.enzymecatalyzingthefirststepofnitrification)fortheAOB,aswellasofthenosZharboringtheN2Oreductase(i.e.enzymecatalyzingthereductionofN2OtoN2)forthedenitrifiersbyRT-qPCR.Wangetal.
(2016b),alsoappliedalsocorrelationanalysis;theN2Oemissions(rangingfrom0.01to3.4%oftheTN-load)weremainlydependentbothontheNO2−concentrationsandaswellasontherelativeAOBabundances,henceindicatingthat
nitrifier-denitrificationwasthedominantpathwayintheaerobiczonesofthereactor.Onthecontrary,theobtainedresultsrevealedthattheemissionswerenotaffectedbytherelativeabundanciesofthedenitrifiers.However,itmust
benotedthatthefunctionalgenelevelsarenotalwaysrepresentativeoftheactivityofthecorrespondingenzyme(Hendersonetal.,2010).Similarly,Castellano-Hinojosaetal.(2018)quantifiedthe16SrRNA,amoAandnosZgenesofthe
totalbacterialandarchaealpopulationinthreefull-scalepredenitrification-nitrificationsystems.Theyemployedmultivariateanalysis(i.e.non-metricmultidimensionalscaling(MDS)andsimilarityanalysisbasedonEuclideandistance,
toidentifytheenvironmentalvariablesthatarebestlinkedtothepatternsofcommunitystructureviaBIO-ENVprocedure)tolinkthebacterialstructurewiththeN2Oemissionsandenvironmental/operationalvariables.Theyfounda
strongpositivecorrelationbetweentheAOBandtheemissionsintheanoxiccompartments,wheretheN2Oreleasewashigher.Therefore,theauthorssuggestednitrifierdenitrificationasthedominantpathway.Ontheotherhand,the
emissionswerenegativelycorrelatedwiththeAOAabundanceandtheN2Oreducers.ItwasconcludedthattheelevatedNO2−concentrations,thelowtemperaturesandshortSRTsmainlyinfluencedtheabundanceofthebacterial
communityencodingthenosZgeneandcontributedtotheN2Oaccumulation.
NH2OHoxidationhasnotbeenconsideredasadominantN2OpathwayinthemajorityofA2/OandA/Oprocessgroups.However,Toyodaetal.(2011),appliedsite-preference(SP)isotopicanalysisinaA2/Oconfigurationand
foundthattheNH2OHoxidationandnitrifierdenitrificationpathwayscontributedalmostequallytotheN2Oformationinthebeginningoftheaerobictank.TheDOconcentrationsinthereactor,though,werenotreported.Additionally,
Blombergetal.(2018)developedanASM3-typeNH2OH-heterotrophicdenitrificationN2OmodelwithaKkLa-basedapproachfortheN2Ostripping(KkLaN2O:masstransfercoefficientforN2O).Themodelwasdevelopedforthefull-scale
undergroundWWTPofViikinmäki(Kosonenetal.,2016)thatisdividedintosixzones(i.e.oneanoxicpre-denitrifyingzone,twoalternatingswitchzones,threeaeratednitrifyingzones).HighDOconcentrationsintheaeratedzones(i.e.
1.5–3.8 mg L−1) and low NO2− concentrations (i.e. 0.1 and 0.7 mg L−1), were observednoted and the model adequately fitted the observed dissolved N2O profiles. However, under the applied stripping modelling, the model
overestimatedtheEF;henceshowingthatthestrippingmodellingapproachmustbeimproved.
Nietal.(2015)consideredallN2Oproductionpathwaysinatwo-stepplug-flowtypereactor(anoxic/aerobic/anoxic/aerobic),inanattempttoexplainthedifferencebetweentheEFsofeachstep(1ststep:0.7%ofinfluent-N,2nd
step:3.5%ofinfluent-N)usingrealdataobtainedfromthestudyofPanetal.(2016),forafull-scalestep-feedplug-flowreactor.TheN2Oproductionwasmainlyattributedtotheheterotrophicdenitrificationtakingplaceintheanoxic
zoneofthe2ndstepthatwasreceiving70%lessbiomasscomparedtothe1ststep(Table1).ThismodelhasbeensuccessfullyappliedfortheexplanationoftheobservedEFdifferenceandidentificationofdominantpathwayssinceit(i)
includesallthepossibleproductionpathways,(ii)hasconsideredthedesignandoperatingfeaturesoftheWWTP,and(iii)wascalibrated/validatedusingdatafromtheplantoperation.
Itmustbenotedthat, inananoxic-aerobicplug-flowreactor, theapplicationofzone-basedstepwisemultipleregressionshowedthat theeffectof theN-load,DOandtemperature inontheN2Oemissions variedwithin the
reactor(Abookabaretal.,2013).Therefore,thedominantpathwayscanpotentiallyvaryintheaforementionedstudiesbasedonthelocationofthesamplingbetweenamongthedifferentstudies.
IntheODreactorsN2Ofluxeshavebeenmainly linkedwith: i)strippingofthedissolvedN2Othat isgenerated in theanoxiczones (Sunetal.,2015;Yanetal.,2014)andii)NO2− accumulation in thelow DO -DO zones, thus
indicatingnitrifier denitrification andheterotrophicdenitrification (Daelmanet al., 2015) as dominant pathways. For instance, in anOD reactor, strong positive correlation (Pearson's coefficient)was identified between dailyN2O
emissionsanddailyNO2−peaks(0.7)(487daysmonitoringcampaign)(Daelmanetal.,2015).TheauthorsproposedtouseNO2−peaksasadiagnosticmethodforthepredictionofN2Opeaks.Inthesamesystem,Vasilakietal.(2018)
appliedchangepointdetectiontechniquescombinedwithhierarchicalk-meansclusteringandPCA,torevealtheN2Oemissionpatterns,andgenerationpathwaysandidentifiedchangesintheN2Ofluxes.Thestudyconcludedthatthe
N2Odependencieswithotheroperationalvariables(i.e.NH4+,NO3−,DO)aredynamicandaffectedbytheseasonalvariations.ThepreferredN2Opathwayswerealsofoundtobedependentontimeandoperationalconditions.
Additionally,afull-scaleODwasmodelledbyNietal.(2013)consideringtheNH2OHoxidationpathway(viatheelectroncarriersapproach,andassumingtheNH2OH/NOmodelandnoinhibitionofAOBNOreductionbyDO)and
theheterotrophicdenitrificationpathway(basedonHiattandGrady(2008)andtheelectroncompetitionbetweendenitrificationsteps).EventhoughAlthough,theoperationalcontroloftheODisnotexplicitlydescribed(i.e.aeration/DO
set-points),morethan90%oftheN2Oemissionswereobserved inaeratedzoneswithDO > 2 mg L−1.Themodelwascalibratedusing3-daysdata froman intensivemonitoringcampaignandvalidatedbasedon1-daydatawith
differentinfluentconditions.ThedevelopedmodellinkedthehigherN2OgenerationwiththeNH4-N+concentrationpeaks(upto∼9 mg L−1inthecalibrationand∼4 mg L−1inthevalidationdataset)withintheaeratedzones(OD)
andsuggestingtheNH2OHoxidationpathwayasdominant.However,inthestudyofNietal.(2013),therespectivecontributionamongthetwoAOBpathwayswasnotexploredwhereasshort-termdatawereusedtovalidatethemodel.
TheAOBdenitrificationmodeldevelopedbyMampaeyetal.(2013)wasappliedusingthesamedatasetfromtheOD,forcalibrationandvalidationpurposes(Spérandioetal.,2016).ThemodelcouldadequatelyfollowthetrendsoftheN2O
behaviouraftercalibrationoftheanoxicreductionfunction(highvalueof0.63wasrequired).NO2−accumulationandtheresultingnitrifierdentificationcontributiontotheemissionscannotbeexcludedinfull-scaleWWTPs.Future
enhancedmodelversionsofmodelsmustconsiderthisfact.
Overall,asshowninsection3,N2OfluxesinCASsystemsaregenerallylow.Tumendelgeretal.(2014)appliedSPisotopicanalysisandobservedthattheNH2OHoxidationpathwaywasresponsibleforupto90%oftheN2O
formationunderhighDO(∼2.5 mg L−1atthemiddleandendoftheaerobictank)inaconventionalASsystem(>44.2%ofNH4+wasnitrifiedandremovedingaseousformprobablyduetounintentionalzoneswithlowDO).Nitrifier
denitrificationandNH2OHoxidationwerealmostequallycontributingtoDOlevelsaround1.5 mgL1,whereasnitrifierdenitrificationdominatedatDOsbelow1.5 mg L−1.
Insidestreamreactors,elevatedN2Oemissionsduringshiftsfromlowtohighaeration(NH4+accumulation,highAOR),havebeenattributedtotheNH2OHpathway(Castro-Barrosetal.,2015).Aerationintensityandprofiles
havebeendeterminedassignificantcontrolparametersfortheN2Ogeneration(Harrisetal.,2015a,bHarrisetal.,2015;Rathnayakeetal.,2015).TheN2Odynamicsunderdifferentaerationintensitiesislikelytodependonthereactor
configuration.Forexample,Mampaeyetal.(2016)andStenströmetal.(2014)observedhigheremissionswhenlowerDOwasappliedinapartialnitritation-anammoxsystemandasidestreamnitrification-denitrificationSBR,respectively.
Ontheotherhand,(Kampschreuretal.,2009a),)couldnotidentifyarelationshipbetweentheN2Oincreaseandthehigheraerationflow-rateduringaprolongedaerationexperimentinasingle-stagenitritation-anammoxreactor.Hence,
theinfluenceoftheaerationregimeontheN2Ogenerationisvariableanddependson,dependingon thereactorconfiguration. InAnammoxreactors,N2Oformationduringanoxicperiodshasbeenmainlyattributedto thenitrifier
denitrification pathway and partially to heterotrophic denitrification (e.g. under conditions of limited substrate provision; Castro-Barros et al., 2015;Mampaey et al., 2016). However, a recent study performed byMa et al. (2017)
demonstratedthatN2OformationviaNH2OHoxidationcanalsooccuratlowDO(∼1 mg L−1)probablycatalysedbycytochromeP460.Thisfindingcontradictspreviousexperimentalandmodel-basedworksaccordingtowhichthe
NH2OHoxidationpathwaydominatessolelyathigherDOconcentrations(e.g.Brottoetal.,2015)andislinkedwiththeAOR(Pengetal.,2014).
StudiesinvestigatingdominantN2Opathwaysforseveralgroupsoffull-scaleprocessesarestillmissing(Table1)andfurtherresearchisrequiredforarobustmappingofthedominantpathwaysindifferentprocessgroups.Asa
generalremark,inmostprocesseswithelevatedN2Oemissions(>0.85%oftheN-load),independentlyoftheconfiguration,elevatedNO2−concentrationswerealsoobserved(Daelmanetal.,2015;Foleyetal.,2010;Rodriguez-Caballeroet
al.,2015;Wangetal.,2016b).
4.3LimitationsandfutureresearchSeveral authorshaveunderlined thedifficulties in determining the respective contribution of eachN2Ogenerationpathwayduring full-scalemonitoring campaigns (Aboobakar et al., 2013;Wang et al., 2016b).None of the
techniquesanalyzsed,intheprevioussectionscanbeappliedstandalonetoexplaintheambiguitiessurroundingthemechanismsandoperationalconditionsthatenhancetheN2Oformationduringwastewatertreatment.Thissection,
summarizesthemainlimitationsofthetechniquesappliedtoexplainN2Ofluxesbehaviourinwastewatersystemsanddescribeshowthesetechniques(combined,)canmaximizetheoutcomeoffuturemonitoringcampaigns.
TwomajordrawbackshavebeenidentifiedwhenstudiesrelyexclusivelyonsimpledescriptivestatisticsandgraphicalrepresentationsofoperationalvariablestoextractprovideinsightsontheN2Oemissionsbehaviour.Firstly,
thisapproach,considersindependentlyseveralparametersthataffecttheN2Ogeneration.Thus,itbecomesdifficulttoquantifythecombinedeffectofseveralvariablesinfull-scalesystemsviasimpleunivariateorbi-variategraphical
representations.Forinstance,Castro-Barrosetal.(2015)observedhigherN2Oemissionsandformationinthetransitionfromtheanoxictotheaeratedperiodsinaone-stagegranularpartialnitritation-anammoxreactor.However,this
increasecannotbesolelyattributedtotheDOchange,sincetheAOR,theNH4+andtheNO2−concentrationswerealsoelevatedduringthetransition.Secondly,graphsrepresentingthebehaviourofprocessvariablesinrelationtothe
N2Oemissions usually cover only a limited period (often shorter than themonitoring campaign duration) or visualize average data in themajority of the reported studies. There are limitations in the effective visualization and
dependenciesintheextractioninofinformationfromlong-termtemporalmultivariatedatasets(i.e.overcrowdedandclutteredvisualizationvisualisation)(Shurkhovetskyyetal.,2018).Forexample,Aboobakaretal.(2013)monitoredtheN2O
emissionsofaPFreactorfor56days;theyreportedthediurnalprofileofthenormalizednormalisedaverageemissions,theaverageammoniaNH4+diurnalprofileandtheaveragedailyDOconcentration,-overthroughoutthedurationof
themonitoringcampaign.
The N2O emission behaviour is characterised by temporal variations. Additionally, for a specific process, the dependencies of the emissions with operational variables are also expected to fluctuate under different
environmental/operationalconditionsbasedonthepreferredN2Oproductionpathway.Vasilakietal.(2018)showedthatthedependenciesbetweenN2OemissionsandoperationalvariablesfluctuatedinaCarrouselreactorthatwas
monitored for 15 months. Therefore, employment of advanced visualizationvisualisation techniques capturing the dynamic behaviour of the operational variables from the whole duration of the monitoring campaigns (e.g. data
abstraction,principal component-basedanalysis, clustering-;Shurkhovetskyy et al., 2018;Aigner et al., 2008) can facilitate an accurate and deep understanding of the long-termN2Obehaviour in both sidestream andmainstream
treatmentprocesses.Thereisaninterchangeablelinkamongoperational/environmentalconditions,N2Oproductionpathwaysandemissionrates.Thecombinationofdataminingmethodscanalsobeappliedtoidentifydistinct,and
differentpatternsandoperationalconditions inordertodeterminetherespectivecontributionamongtheN2Oproductionpathways.Itcanrevealthe linksandrelationshipsamongtheconditionstriggeringtheN2Oemissions, the
dominantpathwaysand theemission rates.However, fewdata-drivenmonitoringandcontrol approacheshavebeenvalidated in full-scale applications in thewastewater sector (Haimi et al., 2013). Additionally, there is still little
guidancefortheselectionofthemostappropriatetechniquesforparticularwastewaterapplications(Hadjimichaeletal.,2016).Multivariatestatisticalanalysis techniqueshaveonlyrecentlybeenappliedtotranslatedata fromthe
monitoringcampaignsintousefulinformationregardingtheN2Oproduction(Vasilakietal.,2018).Hence,structuredapproachesanddata-drivenextractiontechniquesneedtobedevelopedtoprocesstheincomingdatafromWWTPs
(Corominasetal.,2018)andacquireinformationconcerningtheN2Oemissionpatterns.
Given thatpreviouslyunreportedpathways (Harrisetal.,2015a,bHarris et al., 2015), oralternativeconditionsunderwhich thealreadyknownpathwaysareactivatedhavebeen recently identified in literature, isotopicand
molecularbiologyanalysiscanbeappliedtosupportprovideinsightsintotheN2Ogenerationpathways.InarecentreviewontheisotopicmethodsfortheidentificationoftherespectivecontributionsofthedifferentN2Opathways,Duan
etal.(2017)concluded,though,thattherearestilluncertaintiesregardingtheaccuracyoftheSPmethods(i.e.notstandardSPsignaturevalues,unknownN2Oproductionpathways,etc.).Therefore,theauthorssuggesttocomplement
theisotopicmethodswithotherapproaches,suchasthemRNA-basedtranscriptionanalysis(Ishiietal.,2014).
MechanisticmodelsconsideringallthepossibleN2Oproductionpathwaysarepowerfultoolstodescribetheoperationoffull-scaleWWTPs,unveilthemostcontributiveN2Oemissionsgenerationpathwaysandguidetowards
mitigationmeasures.However,therearestill,severalchallengesinthepracticalapplication,calibrationandvalidationofmechanisticN2Omodelsinfull-scalesystems.Parameteruncertaintystillplaysasignificantroleinexplicitly
differentiatingthecontributionofdifferentN2Opathwaysviamodellingstudies;forinstance,differentAOBpathwaymodels(i.e.Spérandioetal.,2016)andmodelswithdifferentcontributionsofdenitrificationN2O-producingpathways
(i.e.Domingo-Félezetal.,2017)havebeenadequatelyfittedtodescribetheN2Oemissionsbehaviourinthesamesystems.InclusionofallmajorN2Oproductionpathwaysresultsincomplexandoverparameterizedmodelsimpairing
reliablecalibrationandvalidation.Additionally,short-termcalibrationandvalidationofmodelsunderspecificoperationalconditions(i.e.dryweather)limitstheiraccuracywhenthesystemvariessignificantly(GuoandVanrolleghem,
2014).
Futureresearchcanalsoexplorethepossibilityofcouplingsophisticatedstatisticaltools(e.g.multivariatestatistics,machinelearningalgorithms)withmultiple-pathwaymechanisticmodelsforfull-scaleapplications,and,thus,
tofacilitatethefastandadaptableonline implementationofmodelpredictivecontrolandforecastingdecisionsupporttools.Theco-applicationofmachine learningandmechanisticmodels inserialorparallelconfigurationshave
already been demonstrated in thewastewater sector. For instance, studies have shown that artificial neural network (ANN)models trainedwith the residuals of ASM-typemodels can enhance the prediction and generalization
capabilitiesoftheASMmodelsforhighlydailyvariableinfluentpollutantconcentrationsandflow-ratesthatareusuallyobservedatWWTPs(Fangetal.,2010;KeskitaloandLeiviskä,2015).Additionally,detailedsimulationsofvariables
thatarenotconventionallymonitoredinWWTPs(i.e.NO2−,NO)canbederivedfromthecomputationallyuniversalmechanisticmodelsandusedtogetherwithrawprocessdatatotrainthemachinelearningmodels.Trainedmachine
learningmodelsareveryefficientforonlinemonitoringandprovisionofdecisionsupport,andarewidelyappliedintheindustry(Wangetal.,2018).Forexample,asupportvectormachineregression(SVM)modelwastrainedwiththe
simulationresultsofanASM2dmodelforafull-scaleA2/OreactorinthestudyofZhangetal.(2014),todevelopaflexiblemulti-objectiveoptimizationtool.TheSVMmodellinkedtheoperatingparameterdatasetsfromtheASM2d
simulationswithperformanceindexesconsideringcostandpollutantconcentrationaspects.ThedevelopmentoflayerbasedonmachinelearningtechniquescanbeintegratedinauniversalcomplexN2Omodeltofacilitateindividual
processparameterscalibration.MultivariatestatisticsandpatternrecognitionalgorithmscanbeappliedtothevariablesmonitoredonlineinWWTPstodifferentiateoperationalconditions(i.e.basedonseasonalinfluentcomposition
variations,differentprocessratesaffectedbyenvironmentalconditions,systemshocks,etc.)andguidetowardsdifferentcalibrationrequirementswithinthesameprocess.Finally,multivariatestatisticscanbeappliedtoidentifyand
isolatecomplexrelationshipsbetweensystemvariablesandguidetowardsprocess-specificsimplifiedmodellingapproaches.Suchintegrated,practicaltoolscanhelpplantoperatorstodesigneffectivemitigationstrategies.
QuantifyingthecontributionoftheN2Oproductionpathwaysinadditiontothetriggeringmechanismsinbiologicalprocessesremainsachallengeandstillrequiresextensiveresearch(Guoetal.,2017).Asageneralremark,
standardizationofmonitoringandreporting,oflong-termN2Omonitoringcampaigns,alongwithcombinedmultivariateanalysesoftheprovideddataandmechanisticmodeldevelopmentarerequiredtoincreasetheunderstandingand
effectivelycontroltheN2OemissionsatWWTPs.
5MonitoringcampaignsandmitigationstrategiesMitigationmeasureshavebeendevelopedandproposedmainlyasoutcomeofstudiesthattargetedatthe:i)testingofdifferentaeration/feedingcontrolstrategiesinfull-scalesidestreamtechnologies(e.g.Castro-Barrosetal.,
2015;Mampaeyetal.,2016;Rodriguez-Caballeroetal.,2015),ii)developmentofmechanisticmodelssimulatingchangingoperationalconditions(e.g.Nietal.,2015),andiii)establishmentofnon-linearregressionmodels(e.g.ANNs)
topredictthebehaviourofN2Oemissions(e.g.Sunetal.,2017a,bSunetal.,2017a).
5.1Mitigationmeasuresandfull-scalemonitoringcampaignsTable2summarizessummarisesthemainN2Omitigationstrategiesthathavebeenproposedforfull-scalesystemsalongwiththemethodologicalapproachesthatfacilitatedtheidentificationofthesemeasures.AsshowninTable
2,thereisnostandardizedstandardisedmethodologyfortheestablishmentofN2Omitigationstrategiesinfull-scalesystems.
Table2Methodsandmainfindingsofstudiesresultinginmitigationmeasures.
alt-text:Table2
Source Process Method Mainfindings MitigationMeasures
Castro-Barrosetal.(2015) One-stagePNAgranularSidestream
• CalculationofdissolvedN2ObasedonMampaey
etal.(2016)
• Modifiedoperationmode:Prolongedanoxic&
aerationperiods
• Visualizationofsignificantprofiles(i.e.DO)&
descriptiveanalysis
• SmootheraerationtransitionsduringnormalreactoroperationconnectedwithlowerN2O
emissions;comparisonwithexperiments
• ProlongedanoxicperiodsleadingtoincreasedN2Oemissions
• Optimizetheaerationregime
• Ensuresmoothshiftsintheaerationpattern
• Optforshortaerationintervals
• Modifiedoperationmode:prolongedanoxic&
aerationperiods,lower
DOexperiments,shorter
SBRcycles
• Nitritationreactor:N2Oformationhigherduringanoxicperiods
• Splittingtheanoxicperiod:averageanoxicN2O
• Preferablyoperateundershortercycles
• Applycontinuousaerationinnitritation
Mampaeyetal.(2016)
One-stageSHARONgranularReactorSidestream
• CalculationofdissolvedN2ObasedonMampaey
etal.(2016)
• Visualizationofsignificantprofiles(i.e.DO)&
descriptiveanalysis
formationratedecreased
• ShortercyclesreducingtheN2OEFby56%attheexpenseofhigherNO3−concentrations
reactor;thisrequiringoptimization
• AlternativelyoperateunderlowerDOsetpoint
Kampschreuretal.,2009a,bKampschreuret.al.,2009a
One-stagePNAgranular
• Modifyoperation:varyingaerationrate
• Visualizationofsignificantprofiles&descriptive
analysis
• Over-aerationsignificantlyimpactingonN2Oemissions
• Ensuresufficientaerationcontrol
Kampschreuretal.(2008)
Two-reactorpartial-nitritation-anammoxprocessSidestream
• Visualizationofsignificantprofiles(N-compounds)&
descriptiveanalysis
• Nitritationreactor:N2Oaccumulationduringthenon-aeratedphase
• Anammoxreactor:NO2−accumulationpotentiallyincreasingN2Oemissions
• Avoidanoxicphasesinnitritationreactor(i.e.smallerreactorstoensuresufficient
HRT)
• Controltheaerationinthenitritationreactor
• Operateaone-reactornitritation-anammoxsystem;potentiallyemittinglessN2Odue
tolimitedNO2−accumulation
Ahnetal.,2010a,b Multipleprocesses(i.e.MLE,step-feedBNR,OD)
• Multiplelinearregressionforseveralprocesses
• InvestigatepossiblelinksbetweenWWTPoperatingconditions&N2Oemissionfluxes
• Aerobiczones:N2Ofluxescorrelatedwithlocation-specificpH,ASmixedliquor
temperature,DO,NH4+&NO2−
concentrations&interactivecombinations
• Anoxiczones:N2Ofluxescorrelatedwithlocation-specificsCOD,pH,ASmixed-liquor
temperature,DO,NO2−&NO3−
concentrations&interactivecombinations
• BNRprocesses:AvoidhighNH4+&NO2−
concentrations,DO&transients
• Aerobicprocesses:avoidincomplete/intermittentnitrification&
over-aeration
• RelyonmoreuniformspatialDOprofilestopromoteSND
• MinimizepeakN-flow(flowequalization)
Nietal.(2013)
ODwithsurfaceaerators&Feedingandaeration(90 min)/settling(35 min)/decanting(55 min)SBR
• Mechanisticmodeldevelopment
• Modellingoftwofull-scalemunicipalWWTPs(i.e.anOD&anSBR)
• OD:decreaseintheNH4+concentrationwithoutsimultaneousNO2−increaseinthe
aeratedzones
• SBR:NH4+accumulationleadingtoahighAORduringtheaerobicSBRphases&,finally,tothe
increasedproductionofintermediates(e.g.
NH2OH)
• Usethedevelopedmodeltoaccuratelysimulatetheemissionsfromthesurface
aeratorzoneinODWWTPs,thus
potentiallycorrectingtheN2Oemission
underestimationinfull-scaleWWTPs
wherethefloatingchambermethodisnot
valid
Lietal.,2016 ReversedA2/OandOD• Observations&literature
• N2Ogenerated&emittedmoreinsummerthaninwinter
• Microbialpopulation&aerationstrategyaskeyfactorsofN2Ogeneration&emission
• Avoidincomplete/intermittentnitrification&over-aerationduringtheaerobic
processestoachievelowerN2O
emissions
• ApplyuniformspatialDOprofilestopromoteSNDthatprobablyleadstoless
N2Oemissions
• PerformflowequalizationtocontrolthepeakingfactoroftheinfluentN-loadingto
theAS
• EnsureasufficientlylongSRTtopreventNO2−accumulationduringnitrification
• AvoidtheCODlimitationofthedenitrificationprocessbyminimizingthe
pre-sedimentationoforganiccarboninthe
influent&dosingadditionalorganic
carbon
Panetal.(2016)&Nietal.(2015)
2step-feeding,anoxic/aerobic/anoxic/aerobicplug-flowreactor
• Mechanisticmodeldevelopment
• Step-feedingresultinginincompletedenitrification&affectingtheAORin
nitrification,henceincreasingthetotalN2O
emission
• DecreasetheN2OEFtothelowestvalueof<1%if30%ofthetotalRASreturnsto
the2ndstep
• IncreaseDOavailabilityforbothAOB&
Wangetal.(2016b) A2/Owithplug-flowpattern• InvestigationofAOBabundances
• N2Oemittedmainlyfromtheoxiczone,withtheemittinglevelsincreasinggreatlyfromthe
beginningoftheoxiczonetowardsthezone
end
• NO2−accumulationdirectlytriggeringN2Oproduction
• Bothdiurnal&seasonalN2Oemissionlevelsfluctuatingstrongly
• OtherfactorsinfluencingtheN2Oemission:lowDO/temperature
NOB
• ImprovetheAOBlivingconditions
• Applyastep-stageaerationmodewithvaryingaerationintensities(location-
specificemissionpatternsforaplug-flow
process)
• Ensureabettermixingviaahigherhorizontalplug-flowratecombinedwithan
appropriateverticalairflowflux;thelarge
cross-sectionwidthsreducedusing
partitionwallstoelevateflowvelocities
underaconstantA2/Otankworking
volume
Sunetal.(2013)Feeding(synchronousaeration)/aeration/settling/decanting(1 heach)SBR
• Visualizationofsignificantprofiles(DO&N2O)&
descriptiveanalysis
• Multiplelinearregressionanalysistoinvestigate
relationshipofN2O
emissions&
environmentalfactors
• Bimonthlysamplingtoexaminechangesinthe
relationshipbetweenN2O
emissions&environmental
factors(long-term:12
months)
• N2Ofluxfromdifferenttreatmentunits/periodsfollowingadescendingorder:feedingperiod,
aerationperiod,settlingperiod,swirlgrittank,
decantingperiod&wastewaterdistribution
tank
• Feeding&aerationperiodsaccountingfor>99%ofN2Oemissions
• LowDOduringnitrificationmajorlyinfluencingN2Oproduction
• Increasetheaerationrateduringthefeedingperiod&decreaseittoaproper
levelfornitrificationintheaerobicstage
• Supplyexternalcarbonsourceduringdenitrification/changetheoperationalSBR
mode(fromfeedingundersynchronous
aerationtofeedingwithanoxicstirring)to
ensureenoughCODprovision/better
utilizationofinfluentCODfor
denitrification
Rodriguez-CaballeroReactionphase(∼130 min)/settling(∼65min)anddecanting(∼65 min)
• Duringtheexperimentalcampaign,3different
cycleconfigurations
implementedaspartof
thenormalSBR
operation
• N2Oemissionsaccountingfor>60%ofthetotalcarbonfootprintoftheWWTP
• CycleswithlongaeratedphasesshowingthelargestN2Oemissions,withaconsequent
• ApplyintermittentaerationtoreducetheNO2−accumulation
• Adoptacycleconfigurationwithshort
etal.(2015) SBR(anoxic/aerobicalternations–3cycletypes)SBR
• Testingofamodifiedcycleconfigurationforthe
N2Omitigation
• Observedissolved&gaseousN2Oprofilesvs
time
increaseinthecarbonfootprint
• TransientNH4+&NO2−concentrations&transitionfromanoxictoaerobicpossibly
involvedintheincreasedN2Oproduction
aeratedperiods
• AllowthesystemtoconsumeN2Othroughdenitrification
Spinellietal.(2018) MLE
• Event-basedsensitivityanalysis
• Box-plotsofdiurnalbehaviourofsignificant
variables&N2O
• LowerCOD:NresultinginhigherN2Oemissionsduetodisturbeddenitrification
• DailyN2Opeaksoccurringunderconditionsofhigheraerationflow-rate(moreintense
stripping)
• Equalizationoftheinfluentflow-rate
Townsend-Smalletal.(2011) MLE
• IsotopiccompositionofN2O
• Bothnitrification&denitrificationcontributingtotheN2OemissionswithinthesameWWTP
• BNRsignificantlyincreasingurbanN2Oemissions
• Applyengineeringprocessesfortheselectionofbacteriacapableofreducing
NO3−withoutreleasingsignificantN2O
amounts
Castellano-Hinojosaetal.(2018)
Twosequentialbioreactors(anoxicandoxic)
• SimultaneouslylinktheabundanceofAOB,AOA&
N2O-reducerswiththe
changesofthe
operational/environmental
variables
• N2OemissionsstronglycorrelatedwithincreasedabundancesofAOB&lowercountsof
N2O-reducers
• UnlikelysignificantcontributionofAOAtoN2Ogenerationsincetheirabundancecorrelated
negativelytoN2Oemissions
• AOBabundancefavouredbyhigherNO3−&NO2−concentrationsintheAS
• AvoidNO2−accumulation,lowtemperatures&excessDOintheanoxic
bioreactorstoenablecomplete
heterotrophicdenitrification&hinder
nitrifierdenitrification
Sunetal.(2017a,b) Biologicaltankwithananoxic&anoxiczoneA/O
• Construction&performanceevaluationof
BP-ANNmodel
• DOhavingasignificantinfluenceontheN2Oproduction
• BP-ANNmodelsuitableforthepredictionofN2OemissionsinotherWWTPswithdifferent
configurations(e.g.A2/O,SBR&nitrification-
anammox),ifinfluent/environmentalparameters
• ApplypropercontrolofDOduringbothnitrification&denitrification
• ApplytheBP-ANNmodelasaconvenient&effectivemethodforthepredictionof
N2OemissionsinanA/OWWTP
&N2Oemissiondatacanbeinvestigated
throughfull-,pilot-orlab-scaleexperiments
Blombergetal.(2018)basedonKosonenetal.(2016)
A/Obioreactor• Mechanisticmodeldevelopment
• Modeldescribingthefull-scaleundergroundWWTPofViikinmäki
• AOBpathways:onlyNH2OHoxidationincludedduetothedynamic&relativelyhighDO
concentrations(1.5–3.8 mg L−1)intheaeratedzones&thelowNO2−concentrations(0.1&
0.7 mg L−1)
• N2Oproductionmainlyintheaeratedzones,minorN2Oconsumption&minorstripping
effectintheanoxiczones
• Appliedstrippingmodel:EFoverestimation
• Improvethestrippingmodellingapproach
• Considerthenitrifierdenitrificationcontributioninfuturemodelversions
Chenetal.(2016) CAS
• Fugacitymodel
• Lab-scaleinsituexperiments
• Comparedtootherparameters(e.g.sludgeconcentration/retentiontime),theadjustmentof
theaerationrateeffectivelymitigatedtheGHG
emissionintheASwithoutsignificantly
affectingthetreatedwaterquality
• N2OasmaincontributortothetotalGHGemission(i.e.57–91%oftotalGHGemission)
• LoweringtheaerationrateintheASby75%enableddecreasingthemassfluxofN2Obyup
to53%
• Mostimportantbenefitofchangingtheaerationrate:lowerenergyconsumptionduringthe
WWTPoperation(fractionalcontributionof
pumpingtothetotalemissionfromthe
WWTP = 46–93%withintherangeoftheaerationratetested)
• Reducetheaerationrate
• Addananoxiczone&recirculationtoanon-BNRsystemfornitrification;
Ribeiroetal.,2017 ExtendedaerationCAS• Differentaerationratestested
• NitrificationasthemaindrivingforcebehindN2Oemissionpeaks
• Airflow-ratevariationspossiblyinfluencingtheN2Oemissions;highN2Oemissionsunder
conditionsofover-aerationorincomplete
nitrificationalongwithNO2−accumulation
otherwise,highN2Oemissionsexpectedin
caseofincreasedDO
• ControltheDO;dynamicchangesinDOconcentrationsreportedasbeing
responsibleforN2Oemissionpeaksin
SNDBNRsystems
• ControltheTN(denitrification)
• AvoidtheconcurrenceofdecreasedDO&NO2−accumulation
Severalstudieshavemodifiedtheaerationintensity,DOandcycledurationtoinvestigatetheeffectonN2Oemissionswithinfull-scalesystems(Castro-Barrosetal.,2015;Kampschreuretal.,2009a;Mampaeyetal.,2016;Rodriguez-
Caballeroetal.,2015).For instance,Mampaeyetal. (2016)achievedareduction in theN2Oemissionsby56%whenthecycles inaone-stagegranularSHARONreactorwereshortenedby1 h.Rodriguez-Caballero et al. (2015)tested
differentoperationalconditionsinafull-scaleSBR.TheyhavesuggestedanoptimumcontrolstrategyfortheminimizationminimisationofN2Oemissionsbasedontheapplicationofshortaerobic-anoxiccycles(20-minaerobicphaseand
shortdurationofanoxicstage).Therefore,testingdifferentoperationalmodesisregardedasoneofthemosteffectivewaystoidentifymeasuresfortheemissionmitigation(Table2).
Nietal.(2015)developedamechanisticmodelutilizingutilisingthedatafromatwo-stepPFreactor(Panetal.,2016)showingthatthebiomassspecificN-loadingratewasresponsiblefortheelevatedN2Oemissionsobservedin
the2nd stepof theprocess.Differentoperational conditionswere testedwith themodeldemonstrating that lowerN2Oemissions (<1%of theN-load) canbeachieved if 30%of the total returnactivated sludge (RAS) stream is
recirculated to the2nd stepof thePF reactor (Table2).However, it is unknownwhether the suggestedmitigation strategywasactuallydemonstrated in the system.Ahnet al. (2010b) identifieddependenciesbetween theWWTP
operatingconditionsandtheN2Oemissionsviamultiplelinearregression.Accordingtotheirfindings,intermittentaerationorover-aerationmustbeavoidedinaerobicreactors.Castellano-Hinojosaetal.(2018)linkedthepopulationof
AOB,AOAandN2O-reducerswiththechanges intheoperationalandenvironmentalvariables in four4conventionalASsystemswithpre-denitrificationzones.TheyobservedthatN2Oemissionsmainlyoccurreddueto incomplete
denitrification,thusunderliningtheimportanceofensuringthecompletionoftheprocessfortheemissionsmitigation.Overall,themaintechniquesformitigatingtheN2Oemissionsinwastewatertreatmentprocessesinclude:i)the
applicationoftheoptimalaerationintensityandDOconcentration,ii)preventingNH4+concentrationpeaks(e.g.viaequalizationequalisationtanks),iii)theavoidanceofNO2−accumulationthroughpropercontrol,andiv)thesupplyof
additionalcarbonsource(whenrequired)toensurecompletedenitrificationintheanoxicreactors(Table2).
However,studiesapplyingandevaluatingmitigationmeasuresforlong-termapplicationsarestillmissing.Additionally,thereisagapbetweenthedatacomingfromthemonitoringcampaignsandtheirprocessinginorderto
establishamitigationstrategy.Moreover,severalmonitoringcampaignsdonotconcludeonthedevelopmentofmitigationstrategies(Filalietal.,2013;Stenströmetal.,2014;Yanetal.,2014).Thelong-termimplementationandevaluation
oftheproposedmitigationstrategiesisstillanissue.Therefore,theestablishmentofstandardizedstandardisedmethodologicalapproachesfortheidentificationofN2Omitigationstrategiesisrequired.
5.2N2Omitigationstrategies:progressandlimitationsAsshowninsection5.1,statisticaland/ormechanisticmodellingaswellasobservatoryanalysisoftheN2Oemissions’behaviourhavebeencommonlyusedeitherasstandaloneorincombinedanalysesforthedevelopmentof
mitigationstrategies.However,thesuggestedmitigationschemeshavenotyetreachedcommercialapplicationsatfull-scalewastewatertreatmentprocesses.ThereisagapbetweentheidentificationofappropriateN2Omitigation
measures and their integration into the control ofWWTPs. Future studies shall focus on the development, implementation and integration of themitigation strategies into the existing control strategy ofwastewater treatment
processes. Special attention must be paid to trade-offs between GHG emissions, energy consumption, system performance and compliance with the legislative requirements in order to support evidence-based multi-objective
optimizationoptimisationoftheWWTPsoperation.
The investigation of direct GHG emissions at full-scalewastewater systems is important for theminimizationminimisation of the environmental footprint ofWWTPs and the integration of the sustainability dimension into
wastewatertreatmentprocesscontrol.
5.3IntegratingtheN2OemissionmonitoringintotheWWTPoperation
Multivariableandmulti-objectiveapproacheshavebeenproposedfortheoptimizationoptimisationoftheWWTPsperformancetoenablethecorrelationofenergyconsumptionoroperationalcostswithsystemperformance(Qiao
andZhang,2018;Zhangetal.,2014).Sweetappleetal.(2014)usedamodifiedversionofthebenchmarksimulationmodel2(BSM2)(Jeppssonetal.,2006)toidentifycontrolstrategiesforthesimultaneousminimizationminimisationofGHG
emissions, operational costsandpollutant loads.TheBSM2modelshavealsobeencombinedwithLCA toevaluate the sustainabilityofdifferentoperating strategies (Flores-Alsinaet al., 2010;Arnell et al., 2017).However, studies
utilizingutilisinglong-termreal-fieldWWTPdataarestillscarce.
Fig.7summarizessummarisestheresearchprioritiesintermsofreal-fieldN2OmonitoringcampaignsthatcanactasfoundationfortheintegrationofN2OemissionsintoWWTPmonitoringandcontrol.Long-termmonitoring
campaignsthatcapturetheseasonalvariability,studiesontheuncertaintiesofthesamplingstrategiesandreportingofoperational,environmentalandsamplingdatacanensuretherobustnessandcomparabilityofthemonitoring
results. The development ofmethodological approaches for the translation ofWWTP data into information can facilitate the understanding of theN2Oemission behaviour and relationship with different operational conditions.
Combinationofmechanisticmodelsand/ordataminingtechniqueswithmethodsfortheN2Oquantificationcanvalidatethemodelsandprovide insights intothedominantpathwaysunderchangingoperationalconditions.Finally,
researchstudiesimplementingandtranslatingtheN2Omitigationmeasuresintocontrolstrategiesareessential.
6ConclusionsAnumberoffull-scaleN2OmonitoringcampaignsinWWTPwerestudied.TheprocesseswereclassifiedsothattherangesofN2OemissionfactorsEFs,dominantN2Opathwaysandtriggeringoperationalconditions,andthe
mitigationmeasuresfordifferentprocessgroupscouldbeset.Thekeyconclusionsofthecurrentrevieware:
• ThereisawiderangeofEFswithinsimilargroupsofwastewatertreatmentprocesses.Theemissionfactorrangesbetween2%and5.6%oftheinfluentN-loadinmainstreamSBR,whileODreactortypes,exhibithavealowEF,∼0.14%oftheN-
load.Long-termcontinuousordiscontinuousmonitoringcampaignsarecharacterisedbyhigherEFscomparedtoshort-termcampaigns.ThestudiesinvestigatingtheseasonalbehaviourofN2Oemissions,haveanaverageEFequalto1.7%of
theN-load,whereasmonitoringcampaignslastinglessthanamonthhaveanaverageEFequaltoof0.7%.Long-termcampaignsshowahighvariabilityofintheN2Oemissions.ThereisnospecificEFcorrelationwiththeNH4+removalinthe
mainstreamprocessesorinspecificgroupsofprocesses.MostoftheprocessesinsmallerWWTPs(i.e.flow-rate<200,000 m3 d−1)hadEFslessthan0.5%oftheN-load,independentlyoftheprocesstypeandnitrificationefficiency.Thisstudy
concludedthatefficientoperationalstrategiescanmitigatethegeneratedN2Ofordifferentconfigurationsandgroupsofprocesses.
• ItisdifficulttocomparetheresultsoftheN2Omonitoringcampaignsbecauseof:i)thedifferencesinthedurationofthemonitoringcampaigns(e.g.short-termcampaignsignoringseasonalvariations),ii)theuncertaintyinthegaseoussampling
methodsandanalyticalmeasurements,andiii)theinsufficientreportingregardingthereactorcontrolstrategy,operationalandenvironmentalconditions,etc.
• SimplefeatureextractionandgraphicalrepresentationsofselectedprocessvariablesandN2OemissionsareemployedtoexplaintheN2Otriggeringmechanisms.GiventhatacombinationofseveralparametersaffectstheN2O generation,
multivariatestatisticalanalysistechniquescanbeausefulalternativeforanalyzinganalysingdataandunderstandingtheN2Oemissionbehaviour.Thecombinedapplicationofmechanisticmodelsandstatisticaltechniquescanleadtobetter
Fig.7MonitoringN2Oemissionsinfull-scalewastewatertreatmentsystems-researchpriorities.
alt-text:Fig.7
designofmitigationstrategies.
• Isotopicandmolecularbiologyanalysesareemerging techniques thatcanqualitativelyandquantitativelyassess theN2Ogenerationpathways.Dataminingmethodscanbedeployed to identifypatternsofoperationalconditionsandN2O
emissions.ThiscanbecomplementedwithtechniquesforthedeterminationoftheN2Oproductionpathways.
• Studiestestingandvalidatingthelong-termfull-scaleN2Omitigationmeasuresarestillmissing.
• Futureresearchshouldfocuson:i)long-termN2Omonitoringcampaigns, ii)theuncertaintiesofdifferentsamplingprotocols, iii)theapplicationofdataminingapproaches,machinelearningandmechanisticmodelsforthedevelopmentof
effectiveandadaptivemodelstothatwillbeintegratedintoWWTPoperationandcontrol,andiv)thedevelopment,implementationandintegrationofthemitigationstrategiesintotheexistingcontrolstrategiesofWWTPs.
Declarationofinterests☒Theauthorsdeclarethattheyhavenoknowncompetingfinancialinterestsorpersonalrelationshipsthatcouldhaveappearedtoinfluencetheworkreportedinthispaper.
AcknowledgementsThisresearchwassupportedbytheHorizon2020researchandinnovationprogramSMART-Plant(grantagreementNo690323).
AppendixA.SupplementarydataSupplementarydatatothisarticlecanbefoundonlineathttps://doi.org/10.1016/j.watres.2019.04.022.
References(Pleaseintegrateinthereferencelistthefollowingpapers:Ni,B.-J.,Yuan,Z.,2015.Recentadvancesinmathematicalmodelingofnitrousoxidesemissionsfromwastewatertreatmentprocesses.WaterRes.87,336–346.https://doi.org/10.1016/j.watres.2015.09.049Palut,M.P.J.,Canziani,O.F.,2007.ContributionofworkinggroupIItothefourthassessmentreportoftheintergovernmentalpanelonclimatechange.CambridgeUniversityPress.Li,H.,Peng,D.,Liu,W.,Wei,J.,Wang,Z.,Wang,B.,2016.N2OgenerationandemissionfromtwobiologicalnitrogenremovalplantsinChina.DesalinationWaterTreat.57,11800–11806.https://doi.org/10.1080/19443994.2015.1046145Ribeiro,R.P.,Bueno,R.F.,Piveli,R.P.,Kligerman,D.C.,deMello,W.Z.,Oliveira,J.L.M.,2017.TheresponseofnitrousoxideemissionstodifferentoperatingconditionsinactivatedsludgewastewatertreatmentplantsinSoutheasternBrazil.WaterSci.Technol.76,2337–2349.https://doi.org/10.2166/wst.2017.399)AboobakarA.,CartmellE.,StephensonT.,JonesM.,ValeP.andDotroG.,Nitrousoxideemissionsanddissolvedoxygenprofilinginafull-scalenitrifyingactivatedsludgetreatmentplant,WaterRes.47,2013,524–534
https://doi.org/10.1016/j.watres.2012.10.004.
AdouaniN.,LimousyL.,LendormiT.andSireO.,N2OandNOemissionsduringwastewaterdenitrificationstep:influenceoftemperatureonthebiologicalprocess,In:ComptesRendusChim.,InternationalChemicalEngineering
Congress(ICEC)2013:fromFundamentalstoAppliedChemistryandBiochemistryvol.18,2015,15–22https://doi.org/10.1016/j.crci.2014.11.005.
AhnJ.H.,KimS.,ParkH.,KatehisD.,PagillaK.andChandranK.,Spatialandtemporalvariabilityinatmosphericnitrousoxidegenerationandemissionfromfull-scalebiologicalnitrogenremovalandnon-BNRprocesses,
WaterEnviron.Res.82,2010a,2362–2372https://doi.org/10.2175/106143010X12681059116897.
AhnJ.H.,KimS.,ParkH.,RahmB.,PagillaK.andChandranK.,N2Oemissionsfromactivatedsludgeprocesses,2008−2009:resultsofanationalmonitoringsurveyintheUnitedStates,Environ.Sci.Technol.44,2010b,
4505–4511https://doi.org/10.1021/es903845y.
AignerW.,MikschS.,MüllerW.,SchumannH.andTominskiC.,Visualmethodsforanalyzingtime-orienteddata,IEEETrans.Vis.Comput.Graph.14,2008,47–60https://doi.org/10.1109/TVCG.2007.70415.
ArnellM.,RahmbergM.,OliveiraF.andJeppssonU.,Multi-objectiveperformanceassessmentofwastewatertreatmentplantscombiningplant-wideprocessmodelsandlifecycleassessment,J.WaterClim.Change8,2017,
715–729https://doi.org/10.2166/wcc.2017.179.
BareselC.,AnderssonS.,YangJ.andAndersenM.H.,Comparisonofnitrousoxide(N2O)emissionscalculationsataSwedishwastewatertreatmentplantbasedonwaterconcentrationsversusoff-gasconcentrations,Adv.
Clim.ChangeRes.7,2016,185–191,Includingspecialtopiconatmosphericblackcarbonanditseffectsoncryospherehttps://doi.org/10.1016/j.accre.2016.09.001.
BatstoneD.J.,HülsenT.,MehtaC.M.andKellerJ.,Platformsforenergyandnutrientrecoveryfromdomesticwastewater:areview,Chemosphere140,2015,2–11.
BellandiG.,PorroJ.,SenesiE.,CarettiC.,CaffazS.,WeijersS.,NopensI.andGoriR.,Multi-pointmonitoringofnitrousoxideemissionsinthreefull-scaleconventionalactivatedsludgetanksinEurope,WaterSci.Technol.77,
2018,880–890https://doi.org/10.2166/wst.2017.560.
BlombergK.,KosseP.,MikolaA.,KuokkanenA.,FredT.,HeinonenM.,MulasM.,LübkenM.,WichernM.andVahalaR.,DevelopmentofanextendedASM3modelforpredictingthenitrousoxideemissionsinafull-scale
wastewatertreatmentplant,Environ.Sci.Technol.52,2018,5803–5811https://doi.org/10.1021/acs.est.8b00386.
BollonJ.,FilaliA.,FayolleY.,GuerinS.,RocherV.andGillotS.,Full-scalepostdenitrifyingbiofilters:sinksofdissolvedN2O?,Sci.TotalEnviron.563–564,2016a,320–328https://doi.org/10.1016/j.scitotenv.2016.03.237.
BollonJ.,FilaliA.,FayolleY.,GuerinS.,RocherV.andGillotS.,N2Oemissionsfromfull-scalenitrifyingbiofilters,WaterRes.102,2016b,41–51https://doi.org/10.1016/j.watres.2016.05.091.
BrottoA.C.,KligermanD.C.,AndradeS.A.,RibeiroR.P.,OliveiraJ.L.M.,ChandranK.andMelloW.Z.de,Factorscontrollingnitrousoxideemissionsfromafull-scaleactivatedsludgesysteminthetropics,Environ.Sci.Pollut.
Res.22,2015,11840–11849https://doi.org/10.1007/s11356-015-4467-x.
CadwalladerA.andVanBriesenJ.M.,Incorporatinguncertaintyintofutureestimatesofnitrousoxideemissionsfromwastewatertreatment,J.Environ.Eng.143,2017,04017029https://doi.org/10.1061/(ASCE)EE.1943-
7870.0001231.
CaivanoM.,BellandiG.,ManciniI.M.,MasiS.,BrienzaR.,PanarielloS.,GoriR.andCanianiD.,Monitoringtheaerationefficiencyandcarbonfootprintofamedium-sizedWWTP:experimentalresultsonoxidationtankand
aerobicdigester,Environ.Technol.38,2017,629–638https://doi.org/10.1080/09593330.2016.1205150.
CarantoJ.D.,VilbertA.C.andLancasterK.M.,NitrosomonaseuropaeacytochromeP460isadirectlinkbetweennitrificationandnitrousoxideemission,Proc.Natl.Acad.Sci.113,2016,14704–14709
https://doi.org/10.1073/pnas.1611051113.
Castellano-HinojosaA.,Maza-MárquezP.,Melero-RubioY.,González-LópezJ.andRodelasB.,Linkingnitrousoxideemissionstopopulationdynamicsofnitrifyinganddenitrifyingprokaryotesinfourfull-scalewastewater
treatmentplants,Chemosphere200,2018,57–66https://doi.org/10.1016/j.chemosphere.2018.02.102.
Castro-BarrosC.M.,DaelmanM.R.J.,MampaeyK.E.,vanLoosdrechtM.C.M.andVolckeE.I.P.,EffectofaerationregimeonN2Oemissionfrompartialnitritation-anammoxinafull-scalegranularsludgereactor,WaterRes.68,
2015,793–803https://doi.org/10.1016/j.watres.2014.10.056.
ChenW.-H.,YangJ.-H.,YuanC.-S.andYangY.-H.,Towardbetterunderstandingandfeasibilityofcontrollinggreenhousegasemissionsfromtreatmentofindustrialwastewaterwithactivatedsludge,Environ.Sci.Pollut.Res.
23,2016,20449–20461https://doi.org/10.1007/s11356-016-7183-2.
ContheM.,ParchenC.,StoutenG.,KleerebezemR.andvanLoosdrechtM.C.M.,O2versusN2Orespirationinacontinuousmicrobialenrichment,Appl.Microbiol.Biotechnol.102,2018,8943–8950
https://doi.org/10.1007/s00253-018-9247-3.
CornejoP.K.,ZhangQ.andMihelcicJ.R.,Howdoesscaleofimplementationimpacttheenvironmentalsustainabilityofwastewatertreatmentintegratedwithresourcerecovery?,Environ.Sci.Technol.50,2016,6680–6689
https://doi.org/10.1021/acs.est.5b05055.
CorominasL.,Garrido-BaserbaM.,VillezK.,OlssonG.,CortésU.andPochM.,Transformingdataintoknowledgeforimprovedwastewatertreatmentoperation:acriticalreviewoftechniques,Environ.Model.Softw.106,
2018,89–103,SpecialIssueonEnvironmentalDataScience.ApplicationstoAirqualityandWatercyclehttps://doi.org/10.1016/j.envsoft.2017.11.023.
DaelmanM.R.,DeBaetsB.,vanLoosdrechtM.C.andVolckeE.I.,Influenceofsamplingstrategiesontheestimatednitrousoxideemissionfromwastewatertreatmentplants,WaterRes.47,2013a,3120–3130.
DaelmanM.R.J.,vanVoorthuizenE.M.,VanDongenL.,VolckeE.I.P.andVanLoosdrechtM.C.M.,Methaneandnitrousoxideemissionsfrommunicipalwastewatertreatment–resultsfromalong-termstudy,WaterSci.Technol.
67,2013b,2350–2355.
DaelmanM.R.J.,vanVoorthuizenE.M.,vanDongenU.G.J.M.,VolckeE.I.P.andvanLoosdrechtM.C.M.,SeasonalanddiurnalvariabilityofN2Oemissionsfromafull-scalemunicipalwastewatertreatmentplant,Sci.Total
Environ.536,2015,1–11https://doi.org/10.1016/j.scitotenv.2015.06.122.
DanielJ.S.,FlemingE.L.,PortmannR.W.,VeldersG.J.M.,JackmanC.H.andRavishankaraA.R.,Optionstoaccelerateozonerecovery:ozoneandclimatebenefits,Atmos.Chem.Phys.10,2010,7697–7707
https://doi.org/10.5194/acp-10-7697-2010.
DeslooverJ.,DeClippeleirH.,BoeckxP.,DuLaingG.,ColsenJ.,VerstraeteW.andVlaeminckS.E.,Floc-basedsequentialpartialnitritationandanammoxatfullscalewithcontrastingN2Oemissions,WaterRes.45,2011,
2811–2821https://doi.org/10.1016/j.watres.2011.02.028.
DeslooverJ.,VlaeminckS.E.,ClauwaertP.,VerstraeteW.andBoonN.,StrategiestomitigateN2Oemissionsfrombiologicalnitrogenremovalsystems,Curr.Opin.Biotechnol.23,2012,474–482
https://doi.org/10.1016/j.copbio.2011.12.030.
DingX.,ZhaoJ.,HuB.,LiX.,GeG.,GaoK.andChenY.,Mathematicalmodelingofnitrousoxide(N2O)productioninanaerobic/anoxic/oxicprocesses:improvementstopublishedN2Omodels,Chem.Eng.J.325,2017,
386–395https://doi.org/10.1016/j.cej.2017.05.082.
Domingo-FélezC.,Pellicer-NàcherC.,PetersenM.S.,JensenM.M.,PlószB.G.andSmetsB.F.,HeterotrophsarekeycontributorstonitrousoxideproductioninactivatedsludgeunderlowC-to-Nratiosduringnitrification-
Batchexperimentsandmodeling,Biotechnol.Bioeng.114,2017,132–140https://doi.org/10.1002/bit.26062.
Domingo-FélezC.andSmetsF.B.,AconsiliencemodeltodescribeN2OproductionduringbiologicalNremoval,Environ.Sci.WaterRes.Technol.2,2016,923–930https://doi.org/10.1039/C6EW00179C.
DuanH.,YeL.,ErlerD.,NiB.-J.andYuanZ.,Quantifyingnitrousoxideproductionpathwaysinwastewatertreatmentsystemsusingisotopetechnology–acriticalreview,WaterRes.122,2017,96–113
https://doi.org/10.1016/j.watres.2017.05.054.
DucheminE.,LucotteM.andCanuelR.,Comparisonofstaticchamberandthinboundarylayerequationmethodsformeasuringgreenhousegasemissionsfromlargewaterbodies§,Environ.Sci.Technol.33,1999,350–357
https://doi.org/10.1021/es9800840.
FangF.,NiB.-J.,XieW.-M.,ShengG.-P.,LiuS.-G.,TongZ.-H.andYuH.-Q.,Anintegrateddynamicmodelforsimulatingafull-scalemunicipalwastewatertreatmentplantunderfluctuatingconditions,Chem.Eng.J.160,2010,
522–529https://doi.org/10.1016/j.cej.2010.03.063.
FilaliA.,FayolleY.,PeuP.,PhilippeL.,NauleauF.andGillotS.,Aerationcontrolinafull-scaleactivatedsludgewastewatertreatmentplant:impactonperformances,energyconsumptionandN2Oemission,In:Presentedat
the11èmeConférenceIWAsurl’instrumentation,lecontrôleetl’automatisation.ICA2013,2013,p.4pp..
Flores-AlsinaX.,ArnellM.,AmerlinckY.,CorominasL.,GernaeyK.V.,GuoL.,LindblomE.,NopensI.,PorroJ.,ShawA.,SnipL.,VanrolleghemP.A.andJeppssonU.,Balancingeffluentquality,economiccostandgreenhousegas
emissionsduringtheevaluationof(plant-wide)control/operationalstrategiesinWWTPs,Sci.TotalEnviron.466–467,2014,616–624https://doi.org/10.1016/j.scitotenv.2013.07.046.
Flores-AlsinaX.,GallegoA.,FeijooG.andRodriguez-RodaI.,Multiple-objectiveevaluationofwastewatertreatmentplantcontrolalternatives,J.Environ.Manag.91,2010,1193–1201
https://doi.org/10.1016/j.jenvman.2010.01.009.
FlowersJ.J.,CadkinT.A.andMcMahonK.D.,Seasonalbacterialcommunitydynamicsinafull-scaleenhancedbiologicalphosphorusremovalplant,WaterRes.47,2013,7019–7031
https://doi.org/10.1016/j.watres.2013.07.054.
FoleyJ.,deHaasD.,YuanZ.andLantP.,Nitrousoxidegenerationinfull-scalebiologicalnutrientremovalwastewatertreatmentplants,WaterRes.44,2010,831–844https://doi.org/10.1016/j.watres.2009.10.033.
FoleyJ.,YuanZ.,KellerJ.,SenanteE.,ChandranK.,WillisJ.,ShahA.,vanLoosdrechtM.C.M.andvanVoorthuizenE.,N2OandCH4EmissionfromWastewaterCollectionandTreatmentSystems:StateoftheScienceReport
andTechnicalReport,2015,IWAPublishing.
FrutosO.D.,QuijanoG.,AizpuruA.andMuñozR.,Astate-of-the-artreviewonnitrousoxidecontrolfromwastetreatmentandindustrialsources,Biotechnol.Adv.36,2018,1025–1037
https://doi.org/10.1016/j.biotechadv.2018.03.004.
GhoneimW.A.M.,HelalA.A.andWahabM.A.,Minimizingenergyconsumptioninwastewatertreatmentplants,In:RenewableEnergiesforDevelopingCountries(REDEC),20163rdInternationalConferenceon,2016,IEEE,1–8.
GuoG.,WangY.,HaoT.,WuD.andChenG.-H.,Enzymaticnitrousoxideemissionsfromwastewatertreatment,Front.Environ.Sci.Eng.12,2017,10https://doi.org/10.1007/s11783-018-1021-3.
GuoL.andVanrolleghemP.A.,Calibrationandvalidationofanactivatedsludgemodelforgreenhousegasesno.1(ASMG1):predictionoftemperature-dependentN2Oemissiondynamics,Bioproc.Biosyst.Eng.37,2014,
151–163https://doi.org/10.1007/s00449-013-0978-3.
GustavssonD.J.I.andlaCourJansenJ.,Dynamicsofnitrogenoxidesemissionfromafull-scalesludgeliquortreatmentplantwithnitritation,WaterSci.Technol.J.Int.Assoc.WaterPollut.Res.63,2011,2838–2845.
HadjimichaelA.,ComasJ.andCorominasL.,Domachinelearningmethodsusedindataminingenhancethepotentialofdecisionsupportsystems?Areviewfortheurbanwatersector,AICommun.29,2016,747–756
https://doi.org/10.3233/AIC-160714.
HaimiH.,MulasM.,CoronaF.andVahalaR.,Data-derivedsoft-sensorsforbiologicalwastewatertreatmentplants:anoverview,Environ.Model.Softw47,2013,88–107https://doi.org/10.1016/j.envsoft.2013.05.009.
HarperW.F.,TakeuchiY.,RiyaS.,HosomiM.andTeradaA.,Novelabioticreactionsincreasenitrousoxideproductionduringpartialnitrification:modelingandexperiments,Chem.Eng.J.281,2015,1017–1023
https://doi.org/10.1016/j.cej.2015.06.109.
HarrisE.,JossA.,EmmeneggerL.,KipfM.,WolfB.,MohnJ.andWunderlinP.,Isotopicevidencefornitrousoxideproductionpathwaysinapartialnitritation-anammoxreactor,WaterRes.83,2015a,258–270
https://doi.org/10.1016/j.watres.2015.06.040.
HarrisE.,JossA.,EmmeneggerL.,KipfM.,WolfB.,MohnJ.andWunderlinP.,Isotopicevidencefornitrousoxideproductionpathwaysinapartialnitritation-anammoxreactor,WaterRes.83,2015b,258–270. (Pleaseremovethisreference;itisthesamewith:Harris,E.,Joss,A.,Emmenegger,L.,Kipf,M.,Wolf,B.,Mohn,J.,Wunderlin,P.,2015a.Isotopicevidencefornitrousoxideproductionpathwaysinapartialnitritation-anammoxreactor. Water Res. 83, 258–270. https://doi.org/10.1016/j.watres.2015.06.040)
HendersonS.L.,DandieC.E.,PattenC.L.,ZebarthB.J.,BurtonD.L.,TrevorsJ.T.andGoyerC.,Changesindenitrifierabundance,denitrificationgenemRNAlevels,nitrousoxideemissions,anddenitrificationinanoxicsoil
microcosmsamendedwithglucoseandplantresidues,Appl.Environ.Microbiol.76,2010,2155–2164https://doi.org/10.1128/AEM.02993-09.
HenzeM.,GradyC.L.,Jr.,GujerW.,MaraisG.andMatsuoT.,ActivatedSludgeModelNo.1.IAWQScientificandTechnicalReportNo.1IAWQ.Lond.UK,1987.
HenzeM.,GujerW.,MinoT.andVanLoosdrechtM.C.M.,ActivatedSludgeModelsASM1,ASM2,ASM2dandASM3,2000,IWApublishing.
HiattW.C.andGradyC.P.L.,AnUpdatedProcessModelforCarbonOxidation,Nitrification,andDenitrification[WWWDocument],2008https://doi.org/info:doi/10.2175/106143008X304776.
HochsteinL.I.andTomlinsonG.A.,Theenzymesassociatedwithdenitrification,Annu.Rev.Microbiol.42,1988,231–261https://doi.org/10.1146/annurev.mi.42.100188.001311.
HooperA.B.,Anitrite-reducingenzymefromNitrosomonaseuropaeaPreliminarycharacterizationwithhydroxylamineaselectrondonor,Biochim.Biophys.ActaBBA-Bioenerg.162,1968,49–65
https://doi.org/10.1016/0005-2728(68)90213-2.
HooperA.B.andTerryK.R.,HydroxylamineoxidoreductaseofNitrosomonas:productionofnitricoxidefromhydroxylamine,Biochim.Biophys.ActaEnzymol.571,1979,12–20https://doi.org/10.1016/0005-2744(79)90220-
1.
IPCC,ThePhysicalScienceBasis.ContributionofWorkingGroupItotheFifthAssessmentReportoftheIntergovernmentalPanelonClimateChange,2013,CambridgeUniversityPress;USA.
IshiiS.,SongY.,RathnayakeL.,TumendelgerA.,SatohH.,ToyodaS.,YoshidaN.andOkabeS.,Identificationofkeynitrousoxideproductionpathwaysinaerobicpartialnitrifyinggranules,Environ.Microbiol.16,2014,
3168–3180https://doi.org/10.1111/1462-2920.12458.
JeppssonU.,RosenC.,AlexJ.,CoppJ.,GernaeyK.V.,PonsM.N.andVanrolleghemP.A.,Towardsabenchmarksimulationmodelforplant-widecontrolstrategyperformanceevaluationofWWTPs,WaterSci.Technol.J.Int.Assoc.
WaterPollut.Res.53,2006,287–295.
JossA.,SalzgeberD.,EugsterJ.,KönigR.,RottermannK.,BurgerS.,FabijanP.,LeumannS.,MohnJ.andSiegristH.,Full-scalenitrogenremovalfromdigesterliquidwithpartialnitritationandanammoxinoneSBR,Environ.
Sci.Technol.43,2009,5301–5306https://doi.org/10.1021/es900107w.
KampschreurM.J.,PoldermansR.,KleerebezemR.,vanderStarW.R.L.,HaarhuisR.,AbmaW.R.,JettenM.S.M.andvanLoosdrechtM.C.M.,Emissionofnitrousoxideandnitricoxidefromafull-scalesingle-stagenitritation-
anammoxreactor,WaterSci.Technol.60,2009a,3211–3217https://doi.org/10.2166/wst.2009.608.
KampschreurM.J.,TemminkH.,KleerebezemR.,JettenM.S.M.andvanLoosdrechtM.C.M.,Nitrousoxideemissionduringwastewatertreatment,WaterRes.43,2009b,4093–4103
https://doi.org/10.1016/j.watres.2009.03.001.
KampschreurM.J.,vanderStarW.R.L.,WieldersH.A.,MulderJ.W.,JettenM.S.M.andvanLoosdrechtM.C.M.,Dynamicsofnitricoxideandnitrousoxideemissionduringfull-scalerejectwatertreatment,WaterRes.42,2008,
812–826https://doi.org/10.1016/j.watres.2007.08.022.
KeskitaloJ.andLeiviskäK.,Artificialneuralnetworkensemblesinhybridmodellingofactivatedsludgeplant,In:AngelovP.,AtanassovK.T.,DoukovskaL.,HadjiskiM.,JotsovV.,KacprzykJ.,KasabovN.,SotirovS.,SzmidtE.and
ZadrożnyS.,(Eds.),IntelligentSystems'2014,AdvancesinIntelligentSystemsandComputing,2015,SpringerInternationalPublishing,683–694.
KosonenH.,HeinonenM.,MikolaA.,HaimiH.,MulasM.,CoronaF.andVahalaR.,Nitrousoxideproductionatafullycoveredwastewatertreatmentplant:resultsofalong-termonlinemonitoringcampaign,Environ.Sci.
Technol.50,2016,5547–5554https://doi.org/10.1021/acs.est.5b04466.
LaneJ.L.,deHaasD.W.andLantP.A.,Thediverseenvironmentalburdenofcity-scaleurbanwatersystems,WaterRes.81,2015,398–415https://doi.org/10.1016/j.watres.2015.03.005.
LawY.,YeL.,PanY.andYuanZ.,Nitrousoxideemissionsfromwastewatertreatmentprocesses,Phil.Trans.R.Soc.B367,2012,1265–1277https://doi.org/10.1098/rstb.2011.0317.
LorkeA.,BodmerP.,NossC.,AlshboulZ.,KoschorreckM.,Somlai-HaaseC.,BastvikenD.,FluryS.,McginnisD.F.,MaeckA.,MüllerD.andPremkeK.,Technicalnote:driftingversusanchoredfluxchambersformeasuring
greenhousegasemissionsfromrunningwaters,Biogeosciences12,2015,7013–7024.
MaC.,JensenM.M.,SmetsB.F.andThamdrupB.,PathwaysandcontrolsofN2Oproductioninnitritation–anammoxbiomass,Environ.Sci.Technol.51,2017,8981–8991https://doi.org/10.1021/acs.est.7b01225.
MampaeyK.E.,BeuckelsB.,KampschreurM.J.,KleerebezemR.,vanLoosdrechtM.C.m.andVolckeE.I.p.,Modellingnitrousandnitricoxideemissionsbyautotrophicammonia-oxidizingbacteria,Environ.Technol.34,2013,
1555–1566https://doi.org/10.1080/09593330.2012.758666.
MampaeyK.E.,DeKreukM.K.,vanDongenU.G.J.M.,vanLoosdrechtM.C.M.andVolckeE.I.P.,IdentifyingN2Oformationandemissionsfromafull-scalepartialnitritationreactor,WaterRes.88,2016,575–585
https://doi.org/10.1016/j.watres.2015.10.047.
ManninaG.,EkamaG.,CanianiD.,CosenzaA.,EspositoG.,GoriR.,Garrido-BaserbaM.,RossoD.andOlssonG.,Greenhousegasesfromwastewatertreatment—areviewofmodellingtools,Sci.TotalEnviron.551–552,2016,
254–270https://doi.org/10.1016/j.scitotenv.2016.01.163.
MassaraT.M.,MalamisS.,GuisasolaA.,BaezaJ.A.,NoutsopoulosC.andKatsouE.,Areviewonnitrousoxide(N2O)emissionsduringbiologicalnutrientremovalfrommunicipalwastewaterandsludgerejectwater,Sci.
TotalEnviron.596,2017,106–123.
MassaraT.M.,SolísB.,GuisasolaA.,KatsouE.andBaezaJ.A.,DevelopmentofanASM2d-N2OmodeltodescribenitrousoxideemissionsinmunicipalWWTPsunderdynamicconditions,Chem.Eng.J.335,2018,185–196.
MatthewsC.J.D.,StLouisV.L.andHessleinR.H.,Comparisonofthreetechniquesusedtomeasurediffusivegasexchangefromshelteredaquaticsurfaces,Environ.Sci.Technol.37,2003,772–780
https://doi.org/10.1021/es0205838.
MelloW.Z.de,RibeiroR.P.,BrottoA.C.,KligermanD.C.,PiccoliA.deS.andOliveiraJ.L.M.,Nitrousoxideemissionsfromanintermittentaerationactivatedsludgesystemofanurbanwastewatertreatmentplant,Quím.Nova
36,2013,16–20https://doi.org/10.1590/S0100-40422013000100004.
NiB.-J.,PanY.,vandenAkkerB.,YeL.andYuanZ.,Full-scalemodelingexplaininglargespatialvariationsofnitrousoxidefluxesinastep-feedplug-flowwastewatertreatmentreactor,Environ.Sci.Technol.49,2015,
9176–9184https://doi.org/10.1021/acs.est.5b02038.
NiB.-J.,PengL.,LawY.,GuoJ.andYuanZ.,Modelingofnitrousoxideproductionbyautotrophicammonia-oxidizingbacteriawithmultipleproductionpathways,Environ.Sci.Technol.48,2014,3916–3924
https://doi.org/10.1021/es405592h.
NiB.-J.,YeL.,LawY.,ByersC.andYuanZ.,Mathematicalmodelingofnitrousoxide(N2O)emissionsfromfull-scalewastewatertreatmentplants,Environ.Sci.Technol.47,2013,7795–7803
https://doi.org/10.1021/es4005398.
PanY.,vandenAkkerB.,YeL.,NiB.-J.,WattsS.,ReidK.andYuanZ.,Unravellingthespatialvariationofnitrousoxideemissionsfromastep-feedplug-flowfullscalewastewatertreatmentplant,Sci.Rep.6,2016
https://doi.org/10.1038/srep20792.
PengL.,NiB.-J.,ErlerD.,YeL.andYuanZ.,TheeffectofdissolvedoxygenonN2Oproductionbyammonia-oxidizingbacteriainanenrichednitrifyingsludge,WaterRes.66,2014,12–21
https://doi.org/10.1016/j.watres.2014.08.009.
PijuanM.,ToràJ.,Rodríguez-CaballeroA.,CésarE.,CarreraJ.andPérezJ.,Effectofprocessparametersandoperationalmodeonnitrousoxideemissionsfromanitritationreactortreatingrejectwastewater,WaterRes.49,
2014,23–33https://doi.org/10.1016/j.watres.2013.11.009.
PortmannR.W.,DanielJ.S.andRavishankaraA.R.,Stratosphericozonedepletionduetonitrousoxide:influencesofothergases,Phil.Trans.R.Soc.B367,2012,1256–1264https://doi.org/10.1098/rstb.2011.0377.
PothM.andFochtD.D.,15NkineticanalysisofN2Oproductionbynitrosomonaseuropaea:anexaminationofnitrifierdenitrification,Appl.Environ.Microbiol.49,1985,1134–1141.
QiaoJ.andZhangW.,Dynamicmulti-objectiveoptimizationcontrolforwastewatertreatmentprocess,NeuralComput.Appl.29,2018,1261–1271https://doi.org/10.1007/s00521-016-2642-8.
RathnayakeR.M.,OshikiM.,IshiiS.,SegawaT.,SatohH.andOkabeS.,EffectsofdissolvedoxygenandpHonnitrousoxideproductionratesinautotrophicpartialnitrificationgranules,Bioresour.Technol.197,2015,15–22.
ReinoC.,vanLoosdrechtM.C.M.,CarreraJ.andPérezJ.,EffectoftemperatureonN2Oemissionsfromahighlyenrichednitrifyinggranularsludgeperformingpartialnitritationofalow-strengthwastewater,Chemosphere
185,2017,336–343https://doi.org/10.1016/j.chemosphere.2017.07.017.
RenY.g.,WangJ.h.,LiH.f.,ZhangJ.,QiP.y.andHuZ.,Nitrousoxideandmethaneemissionsfromdifferenttreatmentprocessesinfull-scalemunicipalwastewatertreatmentplants,Environ.Technol.34,2013,2917–2927
https://doi.org/10.1080/09593330.2012.696717.
RitchieG.a.F.andNicholasD.J.D.,IdentificationofthesourcesofnitrousoxideproducedbyoxidativeandreductiveprocessesinNitrosomonaseuropaea,Biochem.J.126,1972,1181–1191
https://doi.org/10.1042/bj1261181.
RobertsonG.P.andTiedjeJ.M.,Nitrousoxidesourcesinaerobicsoils:nitrification,denitrificationandotherbiologicalprocesses,SoilBiol.Biochem.19,1987,187–193https://doi.org/10.1016/0038-0717(87)90080-0.
Rodriguez-CaballeroA.,AymerichI.,MarquesR.,PochM.andPijuanM.,MinimizingN2Oemissionsandcarbonfootprintonafull-scaleactivatedsludgesequencingbatchreactor,WaterRes.71,2015,1–10
https://doi.org/10.1016/j.watres.2014.12.032.
Rodriguez-CaballeroA.,AymerichI.,PochM.andPijuanM.,Evaluationofprocessconditionstriggeringemissionsofgreen-housegasesfromabiologicalwastewatertreatmentsystem,Sci.TotalEnviron.493,2014,384–391
https://doi.org/10.1016/j.scitotenv.2014.06.015.
SamuelssonJ.,DelreA.,TumlinS.,HadiS.,OfferleB.andScheutzC.,Opticaltechnologiesappliedalongsideon-siteandremoteapproachesforclimategasemissionquantificationatawastewatertreatmentplant,Water
Res.131,2018,299–309https://doi.org/10.1016/j.watres.2017.12.018.
SchaubroeckT.,DeClippeleirH.,WeissenbacherN.,DewulfJ.,BoeckxP.,VlaeminckS.E.andWettB.,Environmentalsustainabilityofanenergyself-sufficientsewagetreatmentplant:improvementsthroughDEMONandco-
digestion,WaterRes.74,2015,166–179https://doi.org/10.1016/j.watres.2015.02.013.
SchulthessR.V.andGujerW.,Releaseofnitrousoxide(N2O)fromdenitrifyingactivatedsludge:verificationandapplicationofamathematicalmodel,WaterRes.30,1996,521–530https://doi.org/10.1016/0043-
1354(95)00204-9.
ShurkhovetskyyG.,AndrienkoN.,AndrienkoG.andFuchsG.,Dataabstractionforvisualizinglargetimeseries,Comput.Graph.Forum37,2018,125–144https://doi.org/10.1111/cgf.13237.
SinghD.P.andMauryaN.S.,Acomparativestudyofthevariousmethodologiesforestimationofgreenhousegasemissionfromwastewatertreatmentsystems(Areview),Orient.J.Chem.32,2016,1373–1380.
Soler-JofraA.,StevensB.,HoekstraM.,PicioreanuC.,SorokinD.,vanLoosdrechtM.C.M.andPérezJ.,Importanceofabiotichydroxylamineconversiononnitrousoxideemissionsduringnitritationofrejectwater,Chem.Eng.
J.287,2016,720–726https://doi.org/10.1016/j.cej.2015.11.073.
SongK.,SuenagaT.,HamamotoA.,SatouK.,RiyaS.,HosomiM.andTeradaA.,Abundance,transcriptionlevelsandphylogenyofbacteriacapableofnitrousoxidereductioninamunicipalwastewatertreatmentplant,J.
Biosci.Bioeng.118,2014,289–297https://doi.org/10.1016/j.jbiosc.2014.02.028.
SpérandioM.,PocquetM.,GuoL.,NiB.-J.,VanrolleghemP.A.andYuanZ.,Evaluationofdifferentnitrousoxideproductionmodelswithfourcontinuouslong-termwastewatertreatmentprocessdataseries,Bioproc.Biosyst.
Eng.39,2016,493–510https://doi.org/10.1007/s00449-015-1532-2.
SpinelliM.,EusebiA.L.,VasilakiV.,KatsouE.,FrisonN.,CingolaniD.andFatoneF.,Criticalanalysesofnitrousoxideemissionsinafullscaleactivatedsludgesystemtreatinglowcarbon-to-nitrogenratiowastewater,J.
Clean.Prod.190,2018,517–524https://doi.org/10.1016/j.jclepro.2018.04.178.
StenströmF.,TjusK.andJansenJ.laC.,Oxygen-induceddynamicsofnitrousoxideinwaterandoff-gasduringthetreatmentofdigestersupernatant,WaterSci.Technol.69,2014,84–91
https://doi.org/10.2166/wst.2013.558.
SunS.,BaoZ.,LiR.,SunD.,GengH.,HuangX.,LinJ.,ZhangP.,MaR.,FangL.,ZhangX.andZhaoX.,ReductionandpredictionofN2OemissionfromanAnoxic/OxicwastewatertreatmentplantuponDOcontrolandmodel
simulation,Bioresour.Technol.244,2017a,800–809https://doi.org/10.1016/j.biortech.2017.08.054.
SunS.,BaoZ.andSunD.,Studyonemissioncharacteristicsandreductionstrategyofnitrousoxideduringwastewatertreatmentbydifferentprocesses,Environ.Sci.Pollut.Res.22,2015,4222–4229
https://doi.org/10.1007/s11356-014-3654-5.
SunS.,ChengX.andSunD.,EmissionofN2Ofromafull-scalesequencingbatchreactorwastewatertreatmentplant:characteristicsandinfluencingfactors,Int.Biodeterior.Biodegrad.85,2013,545–549
https://doi.org/10.1016/j.ibiod.2013.03.034.
SunY.,GuanY.,PanM.,ZhanX.,HuZ.andWuG.,EnhancedbiologicalnitrogenremovalandN2Oemissioncharacteristicsoftheintermittentaerationactivatedsludgeprocess,Rev.Environ.Sci.Biotechnol.16,2017b,
761–780https://doi.org/10.1007/s11157-017-9444-z.
SweetappleC.,FuG.andButlerD.,Multi-objectiveoptimisationofwastewatertreatmentplantcontroltoreducegreenhousegasemissions,WaterRes.55,2014,52–62https://doi.org/10.1016/j.watres.2014.02.018.
TeradaA.,SugawaraS.,HojoK.,TakeuchiY.,RiyaS.,HarperW.F.,YamamotoT.,KuroiwaM.,IsobeK.,KatsuyamaC.,SuwaY.,KobaK.andHosomiM.,Hybridnitrousoxideproductionfromapartialnitrifyingbioreactor:
hydroxylamineinteractionswithnitrite,Environ.Sci.Technol.51,2017,2748–2756https://doi.org/10.1021/acs.est.6b05521.
Townsend-SmallA.,PatakiD.E.,TsengL.Y.,TsaiC.-Y. andRossoD.,NitrousoxideemissionsfromwastewatertreatmentandwaterreclamationplantsinsouthernCalifornia,J.Environ.Qual.40,2011,1542–1550
https://doi.org/10.2134/jeq2011.0059.
ToyodaS.,SuzukiY.,HattoriS.,YamadaK.,FujiiA.,YoshidaN.,KounoR.,MurayamaK.andShiomiH.,IsotopomeranalysisofproductionandconsumptionmechanismsofN2OandCH4inanadvancedwastewatertreatment
system,Environ.Sci.Technol.45,2011,917–922https://doi.org/10.1021/es102985u.
TumendelgerA.,ToyodaS.andYoshidaN.,IsotopicanalysisofN2Oproducedinaconventionalwastewatertreatmentsystemoperatedunderdifferentaerationconditions,RapidCommun.MassSpectrom.28,2014,
1883–1892https://doi.org/10.1002/rcm.6973.
USEPAO.,U.S.GreenhouseGasInventoryReport:1990-2014[WWWDocument],2016,accessed1.25.17https://www.epa.gov/ghgemissions/us-greenhouse-gas-inventory-report-1990-2014.
VachonD.,PrairieY.T.andColeJ.J.,Therelationshipbetweennear-surfaceturbulenceandgastransfervelocityinfreshwatersystemsanditsimplicationsforfloatingchambermeasurementsofgasexchange,Limnol.
Oceanogr.55,2010,1723–1732https://doi.org/10.4319/lo.2010.55.4.1723.
VanHulleS.W.H.,VandeweyerH.J.P.,MeesschaertB.D.,VanrolleghemP.A.,DejansP.andDumoulinA.,Engineeringaspectsandpracticalapplicationofautotrophicnitrogenremovalfromnitrogenrichstreams,Chem.Eng.J.
162,2010,1–20https://doi.org/10.1016/j.cej.2010.05.037.
vanLoosdrechtM.C.,NielsenP.H.,Lopez-VazquezC.M.andBrdjanovicD.,ExperimentalMethodsinWastewaterTreatment,2016,IWAPublishing.
VanrolleghemP.A.andLeeD.S.,On-linemonitoringequipmentforwastewatertreatmentprocesses:stateoftheart,WaterSci.Technol.J.Int.Assoc.WaterPollut.Res.47,2003,1–34.
VasilakiV.,VolckeE.I.P.,NandiA.K.,vanLoosdrechtM.C.M.andKatsouE.,RelatingN2Oemissionsduringbiologicalnitrogenremovalwithoperatingconditionsusingmultivariatestatisticaltechniques,WaterRes.140,
2018,387–402https://doi.org/10.1016/j.watres.2018.04.052.
WanX.,BaetenJ.E.andVolckeE.I.P.,EffectofoperatingconditionsonN2Oemissionsfromone-stagepartialnitritation-anammoxreactors,Biochem.Eng.J.143,2019,24–33https://doi.org/10.1016/j.bej.2018.12.004.
WangJ.,MaY.,ZhangL.,GaoR.X.andWuD.,Deeplearningforsmartmanufacturing:methodsandapplications,J.Manuf.Syst.SpecialIssueSmartManuf.48,2018,144–156https://doi.org/10.1016/j.jmsy.2018.01.003.
WangJinhe,ZhangJ.,WangJing,QiP.,RenY.andHuZ.,NitrousoxideemissionsfromatypicalnorthernChinesemunicipalwastewatertreatmentplant,DesalinationWaterTreat.32,2011,145–152
https://doi.org/10.5004/dwt.2011.2691.
WangY.,FangH.,ZhouD.,HanH.andChenJ.,Characterizationofnitrousoxideandnitricoxideemissionsfromafull-scalebiologicalaeratedfilterforsecondarynitrification,Chem.Eng.J.299,2016a,304–313
https://doi.org/10.1016/j.cej.2016.04.050.
WangY.,LinX.,ZhouD.,YeL.,HanH.andSongC.,Nitricoxideandnitrousoxideemissionsfromafull-scaleactivatedsludgeanaerobic/anoxic/oxicprocess,Chem.Eng.J.289,2016b,330–340
https://doi.org/10.1016/j.cej.2015.12.074.
WeissenbacherN.,TakacsI.,MurthyS.,FuerhackerM.andWettB.,Gaseousnitrogenandcarbonemissionsfromafull-scaledeammonificationplant,WaterEnviron.Res.82,2010,169–175
https://doi.org/10.2175/106143009X447867.
WhiteC.J.andLehnertN.,IsthereapathwayforN2Oproductionfromhydroxylamineoxidoreductaseinammonia-oxidizingbacteria?,Proc.Natl.Acad.Sci.113,2016,14474–14476
https://doi.org/10.1073/pnas.1617953114.
YanX.,LiL.andLiuJ.,Characteristicsofgreenhousegasemissioninthreefull-scalewastewatertreatmentprocesses,J.Environ.Sci.26,2014,256–263https://doi.org/10.1016/S1001-0742(13)60429-5.
ZhangR.,XieW.-M.,YuH.-Q.andLiW.-W.,Optimizingmunicipalwastewatertreatmentplantsusinganimprovedmulti-objectiveoptimizationmethod,Bioresour.Technol.157,2014,161–165
https://doi.org/10.1016/j.biortech.2014.01.103.
Zhu-BarkerX.,CavazosA.R.,OstromN.E.,HorwathW.R.andGlassJ.B.,Theimportanceofabioticreactionsfornitrousoxideproduction,Biogeochemistry126,2015,251–267https://doi.org/10.1007/s10533-015-0166-4.
AppendixA.SupplementarydataThefollowingaretheSupplementarydatatothisarticle:
MultimediaComponent1
Multimediacomponent1
alt-text:Multimediacomponent1
MultimediaComponent2
Multimediacomponent2
alt-text:Multimediacomponent2
Highlights
• N2Oemissionsbenchmarkinginwastewaterishinderedbynon-standardreporting.
• StructuredmethodologicalapproachesforN2Odataanalysisarestillmissing.
• Mechanisticmodelsmustbecalibratedwithlong-termfull-scaledata.
• Isotopicanalysesshouldbecoupledwithmechanistic/data-drivenmodels.
• StudiesvalidatingN2Omitigationmeasuresarestillmissing.
Graphicalabstract
alt-text:Image1