O
pen
A
rchive
T
OULOUSE
A
rchive
O
uverte (
OATAO
)
OATAO is an open access repository that collects the work of Toulouse researchers and
makes it freely available over the web where possible.
This is an author-deposited version published in :
http://oatao.univ-toulouse.fr/
Eprints ID : 9981
To link to this article :
DOI:10.1016/j.bej.2012.05.001
URL :
http://dx.doi.org/10.1016/j.bej.2012.05.001
To cite this version :
Filali, Ahlem and Manas, Angela and Mercade, Myriam and
Bessiere, Yolaine and Biscans, Béatrice and Sperandio, Mathieu
Stability and performance of two GSBR operated in alternating
anoxic/aerobic or anaerobic/aerobic conditions for nutrient removal.
(2012) Biochemical Engineering Journal, vol. 67 . pp. 10-19. ISSN
1369-703X
Stability
and
performance
of
two
GSBR
operated
in
alternating
anoxic/aerobic
or
anaerobic/aerobic
conditions
for
nutrient
removal
Ahlem
Filali
a,b,c,1,
Angéla
Ma˜nas
a,b,c,d,1,
Myriam
Mercade
a,b,c,
Yolaine
Bessière
a,b,c,
Béatrice
Biscans
d,
Mathieu
Spérandio
a,b,c,∗aUniversitédeToulouse;INSA,UPS,INP;LISBP,135AvenuedeRangueil,F-31077Toulouse,France bINRA,UMR792IngénieriedesSystèmesBiologiquesetdesProcédés,F-31400Toulouse,France cCNRS,UMR5504,F-31400Toulouse,France
dCNRS,LaboratoiredeGénieChimique,UMR5503;4,alléeEmileMonsoBP84234-F-31432ToulouseCedex4,France
Keywords: Aerobicgranulation Nitrification Denitrification Phosphorus Filamentous Precipitation
a
b
s
t
r
a
c
t
Twogranularsludgesequencingbatchreactors(GSBR)withalternatinganoxic/aerobic(R1)and anaer-obic/aerobic(R2)conditionswereoperatedwitha4-carbon-sourcesyntheticinfluent.Thephysical propertiesofthegranularsludgewereverygood(SVI≈20mLg−1)andhighsolidconcentrations(up to35gL−1)wereobtainedinthebioreactoroperatedwithapre-anoxicphasewithadditionalnitrate (R1).Incontrast,performanceandgranulesettleabilitywerelowerinR2duetothedevelopmentof fil-amentousheterotrophicbacteriaonthesurfaceofgranules.Thesedisturbanceswerelinkedtothefact thatafractionofCODremainedduringtheaerobicphase,whichwasnotstoredduringtheanaerobic period.TostabilizeaGSBRwithamixtureoforganiccarbonsources,itisthusnecessarytomaximize theamountofsubstrateusedduringthenon-aerated,anaerobicoranoxic,phase.Comparable phos-phateremovalefficiencywasobservedinbothsystems;enhancedbiologicalPremovalbeinggreater inanaerobic/aerobicconditions,whilethecontributionofprecipitation(Ca–P)wasmoresignificantin anoxic/aerobicconditions.
1. Introduction
The aerobicgranular sludgeprocess hasbeen proposedas a promisingapproachtobiologicalwastewatertreatment[1].Thanks totheirdensestructure,aerobicgranuleshaveverygoodsettling abilitythatallowshighbiomassretentioninthebioreactor.This enablestheprocesstowithstandhigh-strengthwastewaterand resultsinthebiologicalreactorhavingasmallervolumethan con-ventionalactivated sludge systems[2].The sizeand density of thegranulesallowsimultaneousnitrification,denitrificationand phosphorusremoval,SNDPR[3,4]tobemaintained.However,the operatingconditionsthatimprovethestabilityofthereactor’s per-formanceandthephysicalpropertiesofaerobicgranularsludgestill needconsideration.Instabilityandpoorerpropertiesofgranular sludgehavebeenreportedforrealsewage,forexample,compared totheidealresultsreportedwithpurelyacetatefedgranules[5–7]. Variousoperatingparametershavebeenidentifiedthat influ-ence granule formation in aerobic systems. They include the
∗Correspondingauthorat:UniversitédeToulouse,INSA,UPS,INP,LISBP,135
AvenuedeRangueil,F-31077Toulouse,France.Tel.:+33561559755.
E-mailaddress:sperandio@insa-toulouse.fr(M.Spérandio).
1 Theseauthorscontributedequallytothiswork.
aerationrate,substratefeedingmode,organic loadingrate,and settlingtime[8–12].Ingranularsludgesequencingbatchreactors (GSBR),theaerationrateplaystwomajorroles:firstly,itimposes thehydrodynamicconditionsinthereactorand,secondly,it con-trolstheoxygen masstransferin theaggregates.High aeration ratehasbeenshowntoprovidehighshearforce,whicherodesthe surfaceofgranules;tostimulatebacterialstrainstosecretemore extracellular polymericsubstances (EPS),thus enhancing struc-turalintegrity;toreducesubstratetransferresistanceintheliquid boundarylayeratthegranulesurface;andtoprovidesufficient oxy-genfororganicsubstratedegradation[8,13].Variousstudieshave demonstratedthatahighaerationrate(expressedbythesuperficial airvelocitySAV)acceleratestheformationofstableaerobic gran-ules.Beunetal.[14]showedthatsmooth,stablegranulescouldbe obtainedonlywithanSAVabove2.0cms−1.Tayetal.[13]found
thatregular,rounder,compactaerobicgranulescouldbeformed onlyaboveaminimumaerationrate(SAV=1.2cms−1).Hence,the
developmentofstableaerobicgranulesinpure aerobicsystems islimitedbecauseofthehigh energydemandinvolved in aera-tion[15]andbecauseefficientnitrogenandphosphorusremoval requiresthepresenceofanaerobicoranoxicandaerobicconditions
[16].
Alternating anoxic/aerobic and anaerobic/aerobic conditions haveboth beenreportedtobehelpful for granulation.Possible
Nomenclature
ACP amorphouscalciumphosphate(Ca3(PO4)2·XH2O)
AOB ammoniumoxidizingbacteria COD chemicaloxygendemand(mgL−1)
DO dissolvedoxygenconcentration(mgL−1)
EBPR enhancedbiologicalphosphorusremoval EPS extracellularpolymericsubstance GSBR granularsequencingbatchreactor HAP hydroxy-apatiteCa5(PO4)3(OH) MLSS mixedliquorsuspendedsolid(gL−1)
MLVSS mixedliquorvolatilesuspendedsolid(gL−1)
NOB nitriteoxidizingbacteria
OLR organicloadingrate(kgCODm−3d−1)
PAO polyphosphateaccumulatingorganism SAV superficialairvelocity(cms−1)
SND simultaneousnitrificationdenitrification SVI sludgevolumeindex(mLg−1)
VFA volatilefattyacid
explanationsforthisarethatconversionofreadilybiodegradable CODtointernalstoredbiopolymerslimitsthesubstrateutilization rateduringtheaerobicphase[17]orthatanoxicgrowthinsidethe granuleimprovesaggregatedensity[18,19].Anotherreasonisthat alternatingnon-aeratedfeastperiodsandaeratedfamineperiods encouragestheselectionofslow-growingbacteria,whichare sup-posedtobepositiveforthedensificationofbio-aggregates[9].The workofWanetal.[18]showedthatthealternatinganoxicfeast andaerobicfamineregimesallowedtheformationofstable aero-bicgranulesandsimultaneousnitrificationdenitrification(SND)at areducedairflowrate(SAV=0.6cms−1).Ontheotherhand,the
alternationofanaerobicandaerobicconditionshasbeenwidely reported topromote internal biopolymer storage and enhance biological phosphorus removal by polyphosphate-accumulating organisms[20,21].
Therefore,theaimofthisstudywastocomparetheeffectof alternatinganoxic/aerobicandanaerobic/aerobicconditionsonthe performanceandstabilityofanaerobicgranularsludgeprocessfor simultaneouscarbon,nitrogenandphosphorusremoval.Forthis purpose,two reactorswererunin parallel.Theywerefed with amixtureoforganicsubstratesandoperatedwithasimilar aer-ation rate.Thefirst reactor(R1) wasoperatedwithalternating anoxic/aerobicconditions,thesecondone(R2)wasoperatedwith alternatinganaerobic/aerobicconditions.Processperformanceand themicroscalestructure of granuleswereinvestigatedfor both reactors.
2. Materialsandmethods
2.1. Reactoroperatingconditions
Theexperimentalset-upincludedtwogeometricallyidentical granularsequencingbatchreactors(GSBR),eachwithaworking volumeof 17L (internaldiameter=15cm, total height=105cm, H/D ratio=7). Both reactors were inoculated with the same concentrationofastabilizedhybridsludge(containingbothflocs andgranules)cultivatedinalternatinganoxic/aerobicconditions. TheinitialMLSSand MLVSSconcentrationswere19.5gL−1 and
13.1gL−1respectively.TheinitialSVIwas22mLg−1
MLSS.ReactorR1
wasoperatedwithalternatinganoxic/aerobicconditions,whereas R2wasoperated withalternating anaerobic/aerobicconditions. Eachreactorwasoperatedsequentiallywithacycletimeof 4h including15minoffeeding,20minofanoxicoranaerobicreaction (nitrogen gasinjection); 145minof aerobic reaction;30minof
Table1
Operatingconditionsofbothreactors.
Parameter R1 R2
Volumetricexchangeratio(%) 47
Hydraulicretentiontime(h) 8.5
Organicloadingrate(kgCODm−3d−1) 2.8
Ammonialoadingrate(kgN–NH4m−3d−1) 0.14
Nitrateloadingrate(kgN–NO3m−3d−1) 0.28 0
Phosphorusloadingrate(kgP–PO4m−3d−1) 0.08
SuperficialupflowvelocityofN2(cms−1) 1.1±0.1 0.6±0.1
Superficialupflowvelocityofair(cms−1) 1.0±0.1
Temperature(◦C) 20±2
pH(notregulated) 7.5–9.2 7.2–8.5
settling and 30min of discharge (with a volumetric exchange ratioof47%).Theaerationratewassimilarinbothreactors,with a superficialair upflow velocity (SAV) of 1.0±0.1cms−1. Both reactorswerefedatthebottomofthecolumnwhenaerationwas stopped (static fill).Thefeedconsisted ofa synthetic substrate
[18]havingthefollowingcomposition:CODof1000mgL−1(25%
contributioneachofglucose,acetate,propionicacidandethanol), [PO43−]=30mgPL−1, [Ca2+]=46mgL−1, [HCO3−]=100mgL−1,
[MgSO4·7H2O]=12mgL−1, [NH4+]=50mgNL−1. A COD/N–NH4+
ratioof20wasmaintained.NitratewasdosedinR1inorderto maintainanoxicconditionsafterfeeding([NO3−]=100mgNL−1).
ThepHprobeandDOprobewereinstalledonlineandthedata wereacquiredbythecomputerevery30s.pHfluctuatednaturally duringareactorcycle,from7.5to9.2inR1andfrom7.2to8.5 inR2.Thetemperaturewasmaintainedat20±2◦Cwithawater jacket.Thereactorperformancewasmonitored throughweekly cyclestudies,inwhichsampleswereanalysedatregularintervals during an SBARcycle. Table 1 summarizes themain operating conditionsofthetworeactors.
Duetoannualclosure,thesupplyofinfluenttothereactorswas interruptedfortwoconsecutiveweeksandthecycleofoperation wasmodified:thenew2-hcycleconsistedof15minaerationand 105minsettling.Thisperiod(fromday105today120)isreferred toasthe“starvationperiod”.
2.2. Analyticalcharacterizationoftheliquidandsolidphases
Analyses were conducted according to standard methods (AFNOR)[22]:COD(NFT90-101),MLSS(NFT90-105)andMLVSS (NFT90-106).NO2−,NO3−,PO43−,NH4+,Ca2+,K+,Mg2+
concentra-tionswereanalysedbyionchromatography(IC25,2003,DIONEX, USA)withpriorfilteringofthesamplesthrougha0.2mm pore-sizeacetatefilter.Thesludgevolumeindex(SVI)wasmeasured inthereactorafter30minofsettlingandshowedlessthan±10% ofdifferencerelativetothestandardprocedureinagraduatedtest tube.Microscopicobservationsofsludgesamplesweremadewitha Biomed-Leitz®binocularphotonicmicroscope.Particlesize
distri-butionwasmeasuredwithaMalvern2000Mastersizer®analyzer
andwithstatisticalimageprocessing.
Theproportionofgranulesbymassandbyvolumewas esti-matedusingEqs.(1)and(2)respectively:
Percentageofgranulesbymass =MLSSMLSSgranule×100
hybridsludge
(1)
Percentageofgranulesbyvolume =VVgranule×100
hybridsludge (2)
whereMLSSgranuleandMLSShybridsludgearethemixedliquor
sus-pendsolidsingranulesandhybridsludgerespectively.Inorderto assesstheMLSSofgranules,sievingat315mmwasperformedas describedinFilalietal.[23].VgranuleandVhybridsludgerepresentthe
apparentvolumeofgranulesandhybridsludge,respectively,inthe reactorafter30minofsettling.
0 5 10 15 20 25 30 35 40 0 100 200 300 time (day) M L S S a n d M L V S S ( g L -1) 0 20 40 60 80 100 S V I (m L g -1 ) 0 5 10 15 20 25 30 35 40 0 100 200 300 time (day) MLSS and MLVSS (g L -1) 0 20 40 60 80 100 S V I (m L g -1 )
2
R
)
(b
1
R
)
(a
Fig.1.Evolutionofsuspendedsolidsduringtherunningperiodofthereactors()MLSS,()MLVSSand( )SVIinR1(a)andR2(b).
2.3. Microbialcharacterization
2.3.1. FISHprobing
Floc and granule samples were fixed as described in Filali et al. [23]. Filamentous bacteria that had developed at the surfaceofgranulesweredetachedwithasterilescalpeland sub-jectedtothesameprocedure astheflocs. Insitu hybridization wasperformed according to the standard hybridization proto-col[24].The fluorescently labelled oligonucleotideprobesused wereas follows: Nso190 and Nso1225 (labelledwith FITC) for AOB[25],Nit3(labelledwithCy3) forNitrobacter spp.[26] and Ntspa662(labelledwithCy3)forNitrospiraspp.[27].DAPI(4′
,6-diamidino-2-phenylindole,dihydrochloride)wasusedtostainall the DNA-containing organisms. To avoid non-specific staining, unlabelledcompetitorprobesCNit3andCNtspa662wereadded withequimolaramountsofNit3andNtspa662,respectively.
Fluorescent insitu hybridizationimageswerecollected with a confocal laser scanning microscope (LEICA SP2, DMRXA2, Germany) using an argon laser (488nm) for FITC excitation, a helium–neon laser for Cy3 (543nm) and a diode laser for DAPI(405nm).Theirfluorescencewasdetectedat498–550nm, 571–630nm or 415–450nm respectively. To obtain images of half-granulesections,10–20(dependingonthesizeofthe gran-ule)overlapping,consecutiveimagesof1024×1024pixelswere acquiredusinga16×oilobjective.Thefinalcompositeimageof thegranulesectionwasthenreconstructedfromallthe individ-ualimagescollectedusingINKSCAPEopensourcescalablevector graphics.Imagesofbacterialclusterswereacquiredusinga100× oilobjective.
2.3.2. PHBstaining
Poly-hydroxybutyrate(PHB)stainingwascarriedoutusinga protocoladaptedfromPandolfietal.[28]. Sampleswerecutto 100-mmthicknessandspreadoveraglassplate.Onceair-dried, thesampleswereplacedinasolutionofSudanblackfor5min,then rinsedwithethanolat70vol.%anddried.Safraninwasdroppedon tothedrysamples,coveringthemcompletely,andleftincontact for5s.Thesamplesweredriedagain,thenrinsedwithdistilled waterandobservedwithanopticalmicroscope.Safraninwasused toshowupthe cytoplasmicmembranesin red(bacterial cells) whereasSudanblackstainedthePHBinclusionsblue.
3. Resultsanddiscussion
3.1. Performancestability
Figs.1and2showtheevolutionofsolidconcentrationandthe removalefficienciesofreactorsR1andR2,respectively,duringthe wholeoperation.Bothreactorswerestartedwithsimilarseedsand showedthesamesuspendedsolidsconcentration,i.e.19.5gTSSL−1
and13.1gVSSL−1,togetherwithverygoodsettlingproperties(SVI
wasinitiallycloseto22mLg−1).
Afterabout50daysofrunninganduptotheendofthestudy (300days),performancewasverystableintheanoxic/aerobic reac-tor(R1).TheremovalofsolubleCOD,ammoniaandtotalnitrogen were96%,100%and89%,respectively.Theremovalof phospho-rousgraduallyincreasedandstabilizedat45%.MLSSandMLVSS firststabilizedaround21.0±1.5gTSSL−1and13.8±1.2gVSSL−1,
respectively,and then graduallyaccumulated in thereactor up to35gTSSL−1 and25gVSSL−1,witha VSS/TSSratioof65–70%.
Theaccumulationofsuspendedsolidswasalsoassociatedwitha decreaseofSVI(16±2mLg−1).Performanceswerestableandit
wasinterestingtoobservethat,afterastarvationperiodof15days (startingonday105)duringwhichitwasnotfed,thereactorwas easilystartedagainandittooklessthanoneweekforremoval per-formancestobefullyrecovered,underliningtherobustnessofthe granulesfromR1.
Inthesecondreactor(R2),performanceinitiallydeclinedand neitherthe sludge concentration northe reactor performances werefullystabilizedatanytimeduringthestudy.TheMLSS concen-trationinitiallyincreased,reaching32gL−1,butthenprogressively
decreased toless than 10gL−1.The sludge volume index (SVI)
fluctuated around70±15mLg−1 during thefirst 100 days and
thenincreasedto90mLg−1.TheCODremoval efficiencyvaried
between90and95%,whereasammoniaandtotalnitrogenremoval fluctuatedbetween60%and100%(meanvaluesof86%and84% respectively).Phosphorusremovaloscillatedbetween10and80% withameanefficiencyof42%.
3.2. Evolutionofsludgeproperties
Asbothreactorswereinitiallyseededwithamixtureof gran-ulesandflocs,theevolutionofthepercentageofgranulesinthetwo reactorswasaninterestingindicatorofsludgeproperties(Fig.3). AslightenrichmentingranuleswasobservedinR1,thegranule percentageincreasingfrom60to80%ofthetotalvolumeand con-stitutingmorethan90%ofthemass.Granulesizealsoprogressively increased(from1to3mm).Concomitantly,adecreasingproportion offlocswasobservedthroughoutthestudy.
Incontrast,inR2,sludgepropertiesevolveddifferently.From day33, an increase of granule fraction bymass wasobserved, togetherwithadecreaseoftheirfractionbyvolume,whichclearly indicatesastrongmodificationoftheaggregatepropertiesandfloc washout.Thiscanbeexplainedbythefactthatgranulesrapidly becamebigger(from1to4mmindiameter)duetoafilamentous (fluffy)growthonthesurface.Asgranularsludgeoccupiedalarger volumeinthereactoraftersettling,flocswerewashedoutandthe granulepercentageinitiallyincreased.Concomitantly,filamentous suspendedgrowthwasalsoobservedandtheflocsthusgenerated, eventhoughtheycontributedlittletothetotalmassofsludge,were
Fig.2.Evolutionofremovalefficienciesof( )COD,( )N–NH4,( )TN,()P–PO4and( )Ca2+inR1(a)andR2(b).
likelytooccupyalargevolumeandtoconsiderablyaffectsludge settlingastheSVIroseto70±15mLg−1.
Fig.4showsthecorrelationbetweentheMLSSconcentration andthepercentageofgranulesinbothreactors.Itindicatesthat theMLSSgloballyincreasedwiththeproportionofthevolume occupiedbythegranulesandthediminutionofsuspendedbiomass growth.
Fromthemicroscopeobservations(Fig.5),itcanbeseenthat granulesfromR1presentedasmoothroundsurfacesimilartothe granulesinitiallyseeded(Fig.5a).Incontrast,asalreadymentioned, filamentous bacteriagrew out fromthe surface of R2granules
(Fig.5b),andthisfilamentouslayerseemedtocauseoxygen trans-ferlimitationasindicatedbythereleaseofgasbubbles(Fig.5e). AssuggestedbyMosquera-Corraletal.[15],long-termanaerobic conditionsinsidethegranulecouldleadtogrowthofmethanogenic bacteria.
Withtime,thesurfaceofsomeofthematuregranulesfrom bothreactorsbecamecrackedandbroken,revealingastrongercore (Figs.5candd).Thestrengthoftheinternalpartofthegranules wasprobablyexplainedbymineralprecipitationasdemonstrated byMa˜nasetal.[29].Whereas,inR1,thelossofefficiencydueto maturegranuledisintegrationseemedtobecompensatedbythe
0% 20% 40% 60% 80% 100% 300 250 200 150 100 50 0
me (day) me (day)
MLSS Volume 0% 20% 40% 60% 80% 100% 300 250 200 150 100 50 0 Granule proporon (%) Granule proporon (%) MLSS Volume
(b)R2
(a)R1
Fig.3.EvolutionoftheproportionofgranulesinMLSSandinvolumeofthehybridsludgeinbothreactors.
0 10 20 30 40 100% 80% 60% 40% 20% 0%
Granule proportion in volume (%)
M L SS o f th e hyb rid slud ge (g L -1 ) R2 R1
Fig.4.RelationshipbetweentheMLSSconcentrationandthegranulepercentage
byvolume.
birthofsmallnewones,granulesfromR2progressivelylosttheir densitybutnonewsmallgranulesseemedtobegeneratedinthis reactor.
3.3. Nitrogenremoval
3.3.1. Kinetics
Kineticanalyseswereperformedduringthebatchcycletoassess COD,nitrogenandphosphorusremoval(Fig.6).
During the anoxic phase in reactor R1, COD and nitrate were rapidly depleted. Denitrification occurred at a rate of 350mgNL−1h−1,andnitriteconcentrationincreased,transiently
reaching5mgNL−1.Ammoniumwasdepletedafter1.5hduring
theaerobic phase, 46%beingremoved byheterotrophic assim-ilation during the anoxic period and 54% aerobically at a rate (AUR)of8.9mgNL−1h−1.Nitriteandnitrateaccumulated,
reach-ing6mgNL−1and14mgNL−1respectively.DOwasfirststabilized
at5mgL−1 duringnitrificationand thenincreasedto6.3mgL−1
afterammoniadepletion.ThestableprofileofTNduringthe aero-bicperiodindicatesthatsimultaneousnitrificationdenitrification didnotoccursignificantlyinthisreactor.Thiscouldbeexplained byDOnotbeinglowenoughandinsufficientorganiccarbonbeing storedduringtheanoxic(feast)phasetoallowdenitrificationinthe granulesduringtheaeratedphase.
InreactorR2,duringthenon-aeratedphase,theCODdecreased by134mgL−1,whichcorrespondsto25%oftheCODfedtothe
system,andreachedaplateauatabout400mgL−1.Duringthe
aer-atedphase,theoxygenprofilereachedafirstplateauat2.5mgL−1,
asecondplateauat4mgL−1,andthenincreasedto6.3mgL−1after
85%oftheCODhadbeenremoved.ThepHincreasefrom7.2to8.2 duringthefirstoxygenplateauindicatedprobableconsumptionof VFAandstrippingofCO2.Duringthesecondplateau,ammonium wasprogressively consumed but nitrate and nitrite concentra-tionremainednegligible.Concomitantly,thepresenceofAOBwas demonstratedbyFISHanalysisingranulesfromR2whereasNOB werenotdetected(seeSection3.3.2).Inaddition,thenitrification
Fig.6. Typicalprofileofsolublecompoundsinbothreactorsduringacyclestudyperformedonday167.Theseparationbetweentheanoxic(R1)oranaerobic(R2)/aerobic
phasesisdepictedbyaverticalline.
ratewasalsomeasuredinbatchrespirometrictestswithgranular sludgecollectedfromR2attheendoftheaerobicphase(atDOof 5mgL−1,withammoniaadditionof10mgNL−1,withoutaddition
oforganiccarbon).AURwas0.99mgNg−1VSSh−1,nonitratewas
observedandnitriteaccumulatedatarateof0.57mgNg−1VSSh−1.
Theseobservations encourageus tothink thatammonium was simultaneouslyremovedbyheterotrophicassimilationand SND duringtheaerobicphaseinthereactorR2.Duetoheterotrophic bacteriarespirationandgrowthonstoragecompounds,thenitrite producedbyAOBwasfullyconsumedandsomeoftheammonia wasalsoassimilated.Thisisinaccordancewiththepresenceof AOBandtheabsenceofNOBinthesegranules.Despitethefactthat theAURinR2wasclosetothatobservedinR1(8.3mgNL−1h−1),
theammoniumwasnotdepletedattheendofthecyclebecause ammoniumstarted tobe nitrified afterthe COD wasdepleted. Beforethat,theoxygenlevelinthegranuleswasprobablytoolow becauseofhighheterotrophicactivity.Thefinalconcentrationof ammoniumintheeffluentwas11mgNL−1butvariedgreatlyfrom
onecycletoanother,leadingtounstableammoniumremoval effi-ciency(Fig.2b).Thiscriticalinstabilityofnitrificationwasduetothe smallproportionofammoniaassimilatedduringthenon-aerated
phaseandtothestrongcompetitionbetweenheterotrophicand autotrophicbacteriaforoxygenduringtheaerobicphase.
3.3.2. Spatialdistributionofnitrifiers
FISHanalysiswasperformedtoassessthemicroscalestructure ofgranulesandthelocalizationofnitrifyingbacteria.Fig.7shows theimagesofgranulesfromR1(a,candd)andR2(bandg),together withsuspendedbiomassfromR1(e)andfilamentousbacteriafrom R2(f).DAPIstaining(blue)indicatedthepresenceofheterotrophic bacteriainthegranules.Fig.7bconfirmsthatgranulesfromR2were muchmoreirregularthanthoseofR1(a).
In thefirst reactor,thelocalization of nitrifiersinthe gran-ules did not change significantly during therun. The imageof ahalf-granule section(Fig.7a)showsthatAOB(magenta)were distributedthroughoutthegranuleand,inparticular,nearlarge channelsandinternalvoids.TheNOB(yellow)werefoundtobe locatedindeeperlayersofthegranule (about250mmfromthe surface)andprofuselyaroundtheinternalcoreofthegranule.As observedinpreviousstudies[30,31],channelsandinternalvoids mayplayakeyroleinthetransportofoxygenandsubstrate,which would explainthis localization.AOBwerefoundtoformdense
Fig.7. ConfocallaserscanningmicroscopyimagesofFISHmicrographsofnitrifiers.HalfgranulesectionofR1(a)andpartofthesection(candd).Halfgranulesectionof
R2(b)andpartofthesection(g).ConfocallaserscanningmicrographsofFISHperformedonaflocsofR1(e)andfilamentousbacteriaofR2(f).AOBappearinmagenta
(hybridizedwithFITC-labelledNso190+Nso1225),NOBappearinyellow(hybridizedwithCy3-labelledNit3+Ntspa662),otherbacteriaappearinblue(DNAstainingwith
DAPI).(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthearticle.)
bacterialclusterswithsizesrangingfrom10to50mm(Fig.7c). NOBclusters weredense and small,their sizerarelyexceeding 10mm(Fig.7d).HybridizationofsamplesofflocsfromR1wasalso performed(Fig.7e).AOBwerefrequentlyobservedinfloc sam-ples,forminglarge,dense clusters,whereasveryfewNOBwere observed.
The spatialdistributionofnitrifiersin ahalf-granule section ofR2(observedonday227)isreportedinFig.7b.The distribu-tionofAOBinR2granuleswaslessextensivethanthedistribution observedinR1.AOBwerelocatedonlyintheoutermost250mmof theR2granules.Theirdistributionwasnothomogeneousthrough thatlayer.AOBclusterscolonizedsomelargepartsoftheaggregate andweretotallyabsentfromothers.Inaddition,manyofthemwere foundtodevelopatthesurfaceofthegranule.Fig.7gshowstypical denseclustersofAOBthathavedevelopednearthesurfaceofthe granule.IncontrasttoR1clusters,AOBclustersinR2wererelatively smallandrarelyexceededa sizeof20mm.Thehybridizationof NOBwiththeoligonucleotidesNit3andNtspa662gavealowsignal similartobackgroundnoiseandnotypicalbacterialclusterscould beidentified(witha magnification100×).Itwasthusobserved thatNOBwerenotsignificantlypresentingranulesfromR2.The hybridizationofsamplesofflocsandfilamentousbacteriathathad developedatthesurfaceofthegranules(Fig.7f)fromR2indicated
thatsuspendedbiomassdidnotcontainnitrifyingbacteria,bethey AOBorNOB(resultnotshown).
TheFISHresultscorroboratethekineticassessmentthatneither nitritenornitratewasaccumulated(Section3.3.1),indicatingthat nitrifiersaredifficulttomaintaininananaerobic/aerobicsystem becauseofstrongcompetitionwithheterotrophsforoxygenand space.Furthermore,theabsenceofNOBinthereactorconfirms thatSNDfavoursthedirectdenitrificationofnitriteinthecoreof granules,whichprogressivelylimitsthedevelopmentofNOB.
3.4. Phosphorusremoval
3.4.1. Kineticassessment
Kineticsassessedoverareactorcycle(Fig.6)showedthe dif-ferentprocessesinvolvedinphosphorusremoval.InR1(Fig.6a), a rapiddecrease ofPO4 wasobservedduring theanoxicphase
concomitantwithapHincreaseduetodenitrification.Ithasbeen demonstratedpreviously[29]thatcalciumphosphateprecipitation isresponsibleforthisphenomenon.Therewasnoapparent phos-phatereleaseinR1duringtheunaeratedperiod.Thisobservation seemslogicalbecause,usually,apre-anaerobic(ratherthananoxic) periodisusedtoinducebiologicalphosphateremoval.Here,nitrate was preferentially used by ordinary heterotrophic biomass for
Fig.8.Microscopicimagesofthestained10mm-cutgranules:(a)R1;(b)R2;(c)R1filamentousbacteria;(d)R2filamentousbacteria.Thescalebar=100mmfor(a)and(b)
and5mmfor(c)and(d).PHBgranulesarestainedinblue/blackwhereascytoplasmicmembrane(cells)arestainedinpink/red.(Forinterpretationofthereferencestocolour
inthisfigurelegend,thereaderisreferredtothewebversionofthearticle.)
substrateoxidation.However,alittlepolyphosphate-accumulating organism(PAO)activity cannevertheless besuspectedin reac-torR1fortworeasons:first,thepotassiumconcentrationshowed a profile (release and uptake) typically explained by its inclu-sioninapolyphosphatestructureand,second,specificanaerobic batchtestsperformedwithgranularsludgefromR1(withacetate asthesolecarbonsource,withoutadditionalcalciumor magne-sium,atapHthatavoidedprecipitation)showedaPreleaserate of2.18mgPg−1MLVSSh−1,whichiscomparabletodataobtained
withsludgefromtheEBPRprocess[32].ThepresenceofPAOin reactorR1couldappearsurprisingasthisreactorwasoperated mainlyinanoxic/aerobicconditions.Howeveritshouldbenoted thatnitrateandnitriteweredepletedafter10–15minandtheend oftheperiodwasanaerobic.Moreover,duetothedensityof gran-ules,ananaerobiczonecouldbefoundinsidethem,eventhough nitrateandnitriteweremeasuredinthebulk.Thus,anaerobicP releaseanduptakewaspossiblymaskedbyPprecipitationinthe GSBRreactorR1.Finally,despitethepossiblepresenceofPAO,the contributionofbiologicalreactionstoPremovalwasprobablylow comparedtoprecipitation(seeSection3.4.3).
AsshownbyFig.6b,significantanaerobicreleaseofPO4 took
placein thesecondreactor (R2).Thisrelease continuedduring thefirstoxygenstepintheaerobicphase(DO=2.5mgL−1),which
meansthatsomeofthePAOwerestillconvertingsubstrate(VFA)to internalpolymers(assuggestedbytheconcomitantpHincrease). Thiswaspossiblyduetothepresenceofanaerobiczonesinthe granuleduringthisphase asoxygenuptakeratewashighuntil CODwasdepletedinthereactor.Inthekineticspresented,thefinal phosphateconcentrationissimilartothatobservedinreactorR1. However,dependingonthecycle,phosphateremovalwasunstable inR2.Thismayhavebeenduetosignificantvariabilityinthe het-erotrophicactivityatthebeginningoftheaerobicphase(Fig.2b), whichwouldinfluencePuptakerate.Anotherexplanationcouldbe
theheterogeneoussludgewastageimposedinthegranularsludge process.Sludgewasnaturallywastedvialossofsuspendedsolids afterbiomassdetachmentfromgranules.Thismeansthatitwas difficulttosetagivenretentiontimeforbacteriaasinthe conven-tionalEBPRprocesses.Here,apparentmeanSRTvariedfrom5to15 daysbasedonthesuspendedsolidsmassbalance.However,local SRTsinthegranuleswereprobablyhigher.ThebiologicalPremoval wouldprobablyhavebeenmorestableifaregularwastageof gran-uleshadbeenperformed.Thisassumptionneedstobeconfirmed infuturework.
3.4.2. PHBstaining
BacterialcarbonstoragewasanalysedusingthePHBstaining techniqueinbothreactors.Fig.8aandbshowsgranulesections,one fromR1andonefromR2,obtainedfromsamplescollectedatthe beginningoftheaerobicphase(PHBgranulesarestainedblue/black whereascytoplasmicmembranes(cells)arestainedpink/red).
PHBwasdistributedinathinlayerintheperipheralzoneofthe granulesandclosetointernalvoidsandchannelsintheinternal partofthegranules.Comparisonofsamplesfromthetworeactors indicatedthat,inthegranulesfromR2,PHBwasmorepresentin theperipheralzone,includingthefilamentousbacterialoutgrowth (Fig.8d).Thisconfirmsthathighheterotrophicactivityandcarbon storageoccurredatthesurfacewhereasgrowth/storageinthe cen-trebecamedifficult,probablybecauseofdiffusionlimitationinthe granulesfromR2.
3.4.3. Phosphateprecipitationcontribution
The contribution of precipitation to overall P removal can be estimated from calcium behaviour. Calcium is not signifi-cantly implicated in the biological formation of polyphosphate
[33]butprecipitatesmainlyintheformofcalciumphosphateas demonstratedbyMa˜nasetal.[29].Fromthermodynamicanalysis,
amorphouscalciumphosphate(ACP)andhydroxy-apatite(HAP) werethemostprobableminerals.X-raydiffractionpatterns(not shown)werecomparedforgranulesfromR1andR2,andshowed thatthemostsignificantmineralwasHAPCa5(PO4)3(OH)although,
inthe case of theR1 spectrum,a small amountof whitlockite (Ca18Mg2H2(PO4)14)wasalsoindicated.
PrecipitationwasmainlycontrolledbypHinthebioreactors. Calciumconcentrationdecreased inR1duringtheanoxicphase whenpHreachedthehighestvalues,closeto9,andcalciumwas progressivelyreleasedduringtheaerobicperiodaspHdecreased withnitrification (Fig. 6a). The calcium profile in R2 (Fig. 6b) remainedstableduringtheanaerobicperiodwithstablepHaround 7.5butdecreasedduringtheaerobicphaseasaconsequenceof pHincreasingto8.ThetotalCa2+ removalwas40%inreactor1
(seeFig.2)whereasitprogressivelydecreasedto20%inreactorR2 (moreunstable).Thistendstoshowthatcalciumphosphate pre-cipitationwasgreaterinthefirstreactorthaninthesecond,which isconsistentwiththefactthatmeanpHwashigherinthefirst reac-torduetomoreintensedenitrification.Moredetailsontherelation betweenpH,calciumconcentrationandphosphateprecipitationin granulescanbefoundinarecentarticlebyMa˜nasetal.[34].
4. Discussion
In this study,significant differences wereobservedbetween GSBRsin termsofperformancesandstability.Thesedifferences canfirstlybeexplainedbythefactthatanoxic/aerobicconditions encouraged pre-denitrification (R1) whereas anaerobic/aerobic conditionsfavouredbiologicalphosphorusremoval(R2).Secondly, anotherimportantconsequenceofworkingwithamixtureof car-bonsourcesandarelativelyshortanaerobicperiodinR2wasthe factthatafractionoftheeasilybiodegradableorganicsubstrate wasnotremovedanaerobicallybutwasdegradedintheaerobic period,whichprobablyexplainsthegrowthoffilamentous bacte-ria.Despitethefactthatalternatinganaerobic/aerobicconditions theoreticallyfavourgranulationbypromotingcarbonstorage, fil-amentousbacteriathatdevelopedatthesurfaceofgranuleshada negativeimpactonsettleabilityandnitrification.
The stability of physicalproperties is a key issuefor ensur-ingthelong-termoperationofagranularsludgereactoranditis characterizedbypropersettleability,whichallowshighbiomass retention.Here,averylowsludgevolumeindex(15mLg−1)and
astable,highMLSSconcentration(upto35gL−1)wereobtained
intheGSBRoperatedwithalternatinganoxic/aerobicconditions (R1).Incontrast,filamentous growthatthesurface ofgranules (andfinally in thebulk)progressively deteriorated the proper-tiesofgranularsludgeanddegradedtheperformancereactorR2, which was operated with alternating anaerobic/aerobic phases (SVI=70±15mLg−1, MLVSS dropped to 10gL−1). Filamentous
growthhaspreviouslybeenobservedinaerobicgranularsludge SBRs, resulting from a changein the wastewater composition, organicloadingrate(OLR)orDOconcentration[13,15,35,36].It isgenerallyacceptedthatfilamentousgrowthisfavouredatlow oxygenavailability.ForinstanceMosquera-Corraletal.[15]noted thataerobicgranuleslosttheirstabilityduetotheoutgrowthof filamentousbacteriawhentheDOwasreducedtolessthan40%of saturation.Inthepresentstudy,despitesimilaraerationratesbeing imposedinbothreactors,lowerDOconcentrationwasobtained atthebeginningoftheaerobic phasein R2(anaerobic/aerobic) duetoresidualCODthatwasnotremovedduringthenon-aerated phase.Thisleadsustothinkthat,here,filamentousgrowthwasdue tooxygen-limitedgrowthofheterotrophicbacteriawithaneasily availablecarbonsource(glucose,ethanolorresidualacetateor pro-pionate)atthebeginningoftheaerobicphase.ThehighCOD:DO
ratiomaintainedinthebulkatthisperiodmadeitdifficultfora densebiofilmtogrowonthesurfaceofthegranules.
Traditionally, physiological studies have revealed that most of the filamentous bacteria have strictly aerobic metabolisms, althoughasmallnumberofmorphotypeshavebeenclaimedto havefermentativemetabolisms[37],givingthemselection advan-tagesin anaerobic/aerobic systems. A plug-flowregime reactor liketheSBRexertsaselectionpressureonfloc-formingbacteria throughthehighgradientsofsubstratepreserved[38]andthe con-versionofeasilybiodegradablesubstratetointernalpolymers[17]. Inthisworkwithacarbonsourcecomposedofglucose,propionate, acetateandethanol,onlyafractionofCODwasremoved anaerobi-callyandconvertedtoPHB.Astheanaerobicperiodwasrelatively short,anaerobicfermentationdidnotplayanimportantroleinthe reactorstudied.TheconsequencewasthatafractionofCODwas notstoredbyPAOandwasstillpresentintheaerobicphase.This wasdetrimentaltogranulepropertiesduetofilamentousgrowth. Areactorwithalongerpre-anaerobicphaseorapre-fermentation phasewouldprobablybebeneficialtogranularsludgeformation. Analternativesolutionwouldbetoincreasetheaerationrateas shownbyMosquera-Corraletal.[15]butthiswouldinducehigher energyconsumption.
Thepresenceof nitrateduringthefeastperiod seemed crit-icalfor the stability ofthe granulesin the firstreactor (R1).A firstexplanationisthatCODwasfullyremovedbydenitrification, thusensuringtheabsenceofCODintheaerobicphase(unlikein R2).Moreover,asshownbyWanetal.[18],theanoxicgrowthof heterotrophicbacteriaintheinnerlayerscouldencourage aggre-gatedensification.Boththesesituationsarelikelytoenhancethe strengtheningofthegranulestructureandimprovegranule sta-bility.Inrealwastewater,nitrateandnitritearerarelypresentand cannotreasonablybeinjectedcontinuously,exceptperhapsduring startuptoacceleratetheformationofgranules.Thismeansthat, forgranularsludgestability,itiscriticaltomaximizetheamountof substrateusedduringthenon-aerated(eitheranaerobicoranoxic) phase.Forthatreason,anadequatecombinationofanoxic, anaer-obicandaerobicphasesshouldbechosen,accordingtotheratio betweenCOD,nitrogenandphosphorusintheinfluent[5].
Thedifferences in termsof nutrient removalshouldalso be pointedout. First,stablefull nitrificationwasmaintained in R1 andammoniumnitrifyingbacteria(AOB)werefoundtobe dis-tributedevenlythroughoutthegranules.Incontrast,filamentous heterotrophicbacteriagrowingattheperipheryofgranulesinR2 madenitrificationdifficultand unstable,andFISHanalysis con-firmedthat nitrifyingbacteriawerelocatedbehindthelayerof filaments.Ontheotherhand,SND(duringtheaerobicphase)was notobservedinR1butwasprobableinR2.Thiscanbeexplained bythefactthatlessstoredCODwasavailablefordenitrification duringtheaerobicperiodinasystemworkingwithapre-anoxic phase(R1).Regardingphosphorus,therecentworkofMa˜nasetal.
[34]demonstrates thatboth biologicaland physical–chemicalP removalshouldbeconsideredinEBPRgranularsludge.Here, bio-logicalPreleasewaslowerinR1thaninR2,whichwaslogically explainedbyless VFAstorage byPAOinanoxicconditions.But globalPremovalefficiencywassimilarandmorestableinR1.This wasbecausemorecalciumphosphateprecipitatedinR1,thanks toahigherpHduetodenitrification.Althoughsomeauthorshave questionedtheactiveroleofcalciuminEBPR[39,40],theeffectof calciumprecipitationongranularsludgeperformanceand prop-ertiesstillneedsmoreinvestigation.In thisstudy,basedonthe calciummassbalanceandconsideringthatmostofthecalcium wasremovedviaprecipitation,thecontribution ofprecipitation was estimated as in Ma˜nas et al. [34]. Depending on the pH, 60–70%ofphosphorusremovalwasduetotheprecipitation mech-anismin R1 whereas precipitationwasresponsible for 40–45% ofPremovalinR2.Hencethesignificantcontributionofcalcium
phosphateprecipitationshouldbeconsideredwhenevaluatingthe fateofremovedphosphorusinGSBR.Onlypartofthe phospho-rusremovedis extracted aspolyphosphate inthebiomass and mostofitcanbeaccumulatedinthegranules,mainlyascalcium phosphate(dependingonthepH,andthewastewatercalciumand phosphateconcentrations).Consequently,adedicatedworkwould benecessarytofindtheoptimalextractionstrategyforgranules, consideringtheeffectofprecipitationongranulestrengthand sta-bility.
5. Conclusions
Inthisstudy,significantdifferenceswereobservedbetweentwo GSBRsworkinginanoxic/aerobicandanaerobic/aerobicconditions. In the anoxic/aerobic reactor, very good settling properties forgranular sludge (SVI≈20mLg−1)and high MLSS
concentra-tionwere stabilized. Full and stable nitrificationwas obtained. Apre-anoxicperiodresultedintheCODbeinguseddirectlyfor denitrification,thuslimiting biologicalphosphorusremovaland SNDbutinducingmorecalcium–phosphateprecipitationdue to pHincrease.
Filamentousbacteriawereobservedintheanaerobic/aerobic reactorand weredetrimentaltogranularsludge propertiesand performances.Apossibleexplanationisthepresenceofresidual, easilybiodegradableCODattheendoftheanaerobicphase,which wasthendegradedaerobically.Acriticalpointwhenusinga mix-tureoforganiccarbonsourcesisthustomaximizetheamountof substrateusedduringthenon-aerated,eitheranaerobicoranoxic, phase.
Acknowledgements
Theauthorswould liketoacknowledge P.Boe,C.Caudan,S. Julien,E.Mengelle,M.Bounouba,andD.Delagnesfortheir help-fulcontributionstotheexecutionofthisworkandC.Pouzet(T.R.I. imagingunit,FR3450,CastanetTolosan)forherhelpwiththe con-focalmicroscopy.
References
[1]M.K.deKreuk,N.Kishida,M.C.M.vanLoosdrecht,Aerobicgranularsludge–
stateoftheart,WaterSci.Technol.9(2007)75–81.
[2]E.Morgenroth,T.Sherden,M.C.M.VanLoosdrecht,J.J.Heijnen,P.A.Wilderer,
Aerobicgranularsludgeinasequencingbatchreactor,WaterRes.31(12)
(1997)3191.
[3]M.deKreuk,J.J.Heijnen,M.C.M.vanLoosdrecht,SimultaneousCOD,nitrogen,
andphosphateremovalbyaerobicgranularsludge,Biotechnol.Bioeng.6(2005)
761–769.
[4]G.Yilmaz,R.Lemaire,J.Keller,Z.Yuan,Simultaneousnitrification,
denitrifica-tion,andphosphorusremovalfromnutrient-richindustrialwastewaterusing
granularsludge,Biotechnol.Bioeng.3(2008)529–541.
[5]Coma,S.Puig,M.D.Balaguer,J.Colprim,Theroleofnitrateandnitriteina
gran-ularsludgeprocesstreatinglow-strengthwastewater,Chem.Eng.J.1(2010)
208–213.
[6]Y.-Q.Liu,B.Moy,Y.-H.Kong,J.-H.Tay,Formation,physicalcharacteristicsand
microbialcommunitystructureofaerobicgranulesinapilot-scalesequencing
batchreactorforrealwastewatertreatment,EnzymeMicrob.Technol.6(2010)
520–525.
[7]A.Nor-Anuar,Z.Ujang,M.C.M.vanLoosdrecht,M.K.deKreuk,G.Olsson,
Strengthcharacteristicsofaerobicgranularsludge,WaterSci.Technol.65(2)
(2012)309–316.
[8]Y.Liu,J.H.Tay,Theessentialroleofhydrodynamicshearforceinthe
for-mation of biofilm and granularsludge, Water Res. 36 (avr.(7)) (2002)
1653–1665.
[9] Y. Liu, S.-F. Yang, J.-H. Tay, Improved stability of aerobic granules by
selecting slow-growing nitrifying bacteria, J. Biotechnol. 108 (2) (2004)
161.
[10]B.S.McSwain,R.L.Irvine,P.A.Wilderer,Theeffectofintermittentfeedingon
aerobicgranulestructure,WaterSci.Technol.12(2004)19–25.
[11]B.Y.P.Moy,J.H.Tay,S.K.Toh,Y.Liu,S.T.L.Tay,Highorganicloadinginfluences
thephysicalcharacteristicsofaerobicsludgegranules,Lett.Appl.Microbiol.6
(2002)407–412.
[12]L.Qin,Y.Liu,J.-H.Tay,Effectofsettlingtimeonaerobicgranulationin
sequenc-ingbatchreactor,Biochem.Eng.J.21(1)(2004)47.
[13]J.H.Tay,Q.S.Liu,Y.Liu,Theeffectsofshearforceontheformation,structure
andmetabolismofaerobicgranules,Appl.Microbiol.Biotechnol.57(Oct.(1–2))
(2001)227–233.
[14]J.J.Beun,A.Hendriks,M.C.M.VanLoosdrecht,E.Morgenroth,P.A.Wilderer,J.J.
Heijnen,Aerobicgranulationinasequencingbatchreactor,WaterRes.33(juill.
(10))(1999)2283–2290.
[15] A.Mosquera-Corral,M.K.deKreuk,J.J.Heijnen,M.C.M.vanLoosdrecht,Effects
ofoxygenconcentrationonN-removalinanaerobicgranularsludgereactor,
WaterRes.12(2005)2676–2686.
[16] S.Tsuneda,T.Ohno,K.Soejima,A.Hirata,Simultaneousnitrogenand
phos-phorusremovalusingdenitrifyingphosphate-accumulatingorganismsina
sequencingbatchreactor,Biochem.Eng.J.27(janv.(3))(2006)191–196.
[17]M.C.M.VanLoosdrecht,M.A.Pot,J.J.Heijnen,Importanceofbacterialstorage
polymersinbioprocesses,WaterSci.Technol.1(1997)41–47.
[18] J.Wan,Y.Bessiere,M.Spérandio,Alternatinganoxicfeast/aerobicfamine
condi-tionforimprovinggranularsludgeformationinsequencingbatchairliftreactor
atreducedaerationrate,WaterRes.20(2009)5097–5108.
[19]J.Wan,M.Sperandio,Possibleroleofdenitrificationonaerobicgranularsludge
formationinsequencingbatchreactor,Chemosphere2(2009)220–227.
[20]P.M.J.Janssen,K.Meinema,H.F.vanderRoest,BiologicalPhosphorusRemoval:
ManualforDesignandOperation,IWAPublishing,2002.
[21]T.Mino,W.Liu,F.Kurisu,T.Matsuo,Modellingglycogenstorageand
denitrifi-cationcapabilityofmicroorganismsinenhancedbiologicalphosphateremoval
processes,WaterSci.Technol.31(janv.(2))(1995)25–34.
[22]AfnorÉd.,Qualitédel’eau,in:Recueildenormes,AfnorEditions,1994.
[23]A.Filali,Y.Bessiere,M.Sperandio,Effectsofoxygenconcentrationonthe
nitri-fyingactivityofanaerobichybridgranularsludgereactorRIDA-1878-2012,
WaterSci.Technol.65(2)(2012)289–295.
[24]R.Amann,B.M.Fuchs,S.Behrens,Theidentificationofmicroorganismsby
flu-orescenceinsituhybridisation,Curr.Opin.Biotechnol.12(juin(3))(2001)
231–236.
[25]B.K.Mobarry,M.Wagner,V.Urbain,B.E.Rittmann,D.A.Stahl,Phylogenetic
probesforanalyzingabundanceandspatialorganizationofnitrifyingbacteria,
Appl.Environ.Microbiol.62(janv.(6))(1996)2156–2162.
[26] M.Wagner,G.Rath,H.-P.Koops,J.Flood,R.Amann,Insituanalysisofnitrifying
bacteriainsewagetreatmentplants,WaterSci.Technol.2(1996)237–244.
[27]H.Daims,P.H.Nielsen,J.L.Nielsen,S.Juretschko,M.Wagner,Novel
Nitrospira-like bacteria as dominant nitrite-oxidizers in biofilmsfrom wastewater
treatmentplants:diversityandinsituphysiology,WaterSci.Technol.41(juin
(4–5))(2000)85–90.
[28]D.Pandolfi,M.-N. Pons,M.daMotta,Characterization ofPHBstoragein
activatedsludgeextendedfilamentousbacteriabyautomatedcolourimage
analysis,Biotechnol.Lett.29(août(8))(2007)1263–1269.
[29] A.Ma˜nas,B.Biscans,M.Spérandio,Biologicallyinducedphosphorus
precipita-tioninaerobicgranularsludgeprocess,WaterRes.12(2011)3776–3786.
[30]V.Ivanov,S.T.-L.Tay,Q.-S.Liu,X.-H.Wang,Z.-W.Wang,J.-H.Tay,Formation
andstructureofgranulatedmicrobialaggregatesusedinaerobicwastewater
treatment,WaterSci.Technol.7(2005)13–19.
[31]Y.-M.Zheng,H.-Q.Yu,G.-P.Sheng,Physicalandchemicalcharacteristicsof
granularactivatedsludgefromasequencingbatchairlift reactor,Process
Biochem.40(févr.(2))(2005)645–650.
[32]J.K.Park,L.M.Whang,J.C.Wang,G.Novotny,Abiologicalphosphorusremoval
potential test forwastewaters, Water Environ. Res. 73 (mai (3))(2001)
374–382.
[33]N.Jardin,H.J.Pöpel,Behaviorofwasteactivatedsludgefromenhanced
biolog-icalphosphorusremovalduringsludgetreatment,WaterEnviron.Res.68(6)
(1996)965–973.
[34]A.Ma˜nas,M.Pocquet,B.Biscans,M.Sperandio,Parametersinfluencingcalcium
phosphateprecipitationingranularsludgesequencingbatchreactor,Chemical
EngineeringScience,(2012)http://dx.doi.org/10.1016/j.ces.2012.01.009.
[35]N.Schwarzenbeck,R.Erlay,P.A.Wilderer,Aerobicgranularsludgeinan
SBR-systemtreatingwastewaterrichinparticulatematter,WaterSci.Technol.12
(2004)41–46.
[36]Y.-M.Zheng,H.-Q.Yu,S.-J.Liu,X.-Z.Liu,Formationandinstabilityofaerobic
granulesunderhighorganicloadingconditions,Chemosphere63(10)(2006)
1791.
[37]G.Nowak,G.D.Brown,CharacteristicsofNostocoidalimicolaanditsactivityin
activatedsludgesuspension,Res.J.WaterPollut.ControlFed.62(mars(2))
(1990)137–142.
[38]A.M.P.Martins,K.Pagilla,J.J.Heijnen,M.C.M.vanLoosdrecht,Filamentous
bulk-ingsludge–acriticalreview,WaterRes.38(févr.(4))(2004)793–817.
[39] C.F.C.Bonting,G.J.J.Kortstee,A.Boekestein,A.J.B.Zehnder,Theelemental
com-positiondynamicsoflargepolyphosphategranulesinAcinetobacterstrain
210A,Arch.Microbiol.159(mai(5))(1993)428–434.
[40]C.Schönborn,H.-D.Bauer,I.Röske,Stabilityofenhancedbiologicalphosphorus
removalandcompositionofpolyphosphategranules,WaterRes.35(Sept.(13))
