ContentslistsavailableatSciVerseScienceDirect
Biochemical
Engineering
Journal
jou rn a l h o m e pag e :w w w . e l s e v i e r . c o m / l o c a t e / b e j
The
investigation
of
dairy
industry
wastewater
treatment
in
a
biological
high
performance
membrane
system
Burhanettin
Farizoglu
∗,
Suleyman
Uzuner
BalikesirUniversity,EngineeringandArchitectureFaculty,DepartmentofEnvironmentalEngineering,Balikesir,Turkey
a
r
t
i
c
l
e
i
n
f
o
Articlehistory: Received17May2011
Receivedinrevisedform10August2011 Accepted14August2011
Available online 22 August 2011 Keywords:
Jetloopreactor Dairywastewater
Industrialwastewatertreatment Ultrafiltration
Membranebioreactor Membranefouling
a
b
s
t
r
a
c
t
Thedairyindustryisgenerallyconsideredtobethelargestsourceoffoodprocessingwastewaterin manycountries.Thehighlyvariablenatureofdairywastewatersintermsofvolumesandflowrates andintermsofhighorganicmaterialscontentssuchasCOD921–9004mgL−1,BOD483–6080mgL−1, TNof8–230mgL−1andSSof134–804mgL−1makesthechoiceofaneffectivewastewatertreatment regimedifficult.Ahighperformancebioreactor,anaerobicjetloopreactor,combinedwithaceramic membranefiltrationunit,wasusedtoinvestigateitssuitabilityforthetreatmentofthedairyprocessing wastewater.Theoxygentransferratesofthebioreactorwerefoundtobeveryhigh(100–285h−1)onthe operatingconditions.Aloadingrateof53kgCODm−3d−1resultedin97–98%CODremovalefficiencies
under3hhydraulicretentiontime.ThehighMLSSconcentrationscouldberetainedinthesystem(upto 38,000mgL−1)withthecontributionofUF(ultrafiltration)unit.Duringthefiltrationofactivatedsludge,
thefluxesdecreasedwithincreasingMLSS.Cakeformationfoulingwasdeterminedasdominantfouling mechanisms.Theresultsdemonstratethatjetloopmembranebioreactorsystemwasasuitableand effectivetreatmentchoicefortreatingdairyindustrywastewater.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
The dairy industry is generally considered tobe thelargest sourceoffoodprocessingwastewaterinmanycountries.Water isusedthroughoutallstepsofthedairyindustry,including clean-ing,sanitization,heating,cooling,andfloorwashing;naturallythe industry’sneedforwaterishuge[1].Ingeneral,wastesfromthe dairyprocessingindustrycontainahighconcentrationoforganic materialsuchasproteins,carbohydratesandlipids,highBODand COD,andhighconcentrationsofsuspendedsolidsandsuspended oil-grease.Alloftheserequirespecializedtreatmentstoprevent orminimizeenvironmentalproblems.Dairywastewaters(DWs) arealsocharacterizedbywidefluctuationsinflowrates,relatedto discontinuityintheproductioncyclesofdifferentproducts[2].The highlyvariablenatureofdairywastewatersintermsofvolumeand flowratesandalsointermsofthepHandsuspendedsolids(SS) con-tentmakesitdifficulttochooseaneffectivewastewatertreatment regime[3].Tocomplywithnewdischargestandards, thedairy industrieshaveadoptedanelaborateeffluenttreatmentprotocol thatisaffectingtheoveralleconomyoftheplantandincreasingthe costsofconventionaltreatmentsystems.
∗ Correspondingauthor.Tel.:+9002666121194;fax:+9002666121426. E-mailaddress:bfarizoglu@gmail.com(B.Farizoglu).
Recently, researchershaveshiftedtheirintereststothe pos-sibilities ofreuse orrecycling ofindustrial wastewaters.Earlier researchershaveinvestigatedthedairyindustryeffluenttreatment throughthemembraneprocessandthepossibilityofreuse[4,5]. Thereis agrowinginterest incombiningmembraneswith bio-logicalwastewatertreatment.Themembranebioreactors(MBR) offer distinct advantages over traditional biological processes: higherbiodegradationefficiency,smallerfootprint,betterqualityof treatedwater,theabsolutecontrolofsolidsandhydraulicretention time,retentionofallmicroorganismsandviruses,andeasycontrol ofoperatingconditions[6,7].Inparticular,theabsoluterejectionof sludgebythemembranemakesitpossibletoovercomethe prob-lemofdependence onsettleability[8].Furthermore,membrane separationenablesasignificantincreaseinthebiomass concentra-tioninthebioreactor,thusreducingitssize[9].Asthereactionrate isdirectlyproportionaltobiomassconcentration,ahigh concentra-tionisdesirable.Ontheotherhand,tooperatethebioreactorathigh biomassconcentrations,specialreactortopologiesshouldbe cho-sen.Inordertoselectanddesignthecorrectbioreactortopology,it isnecessarytoknowthecharacteristicsofeffluents,volumes,laws ofmicroorganisms’growthrate,andbiokineticparametersdefined bymathematical models.Jet-loop reactors(JLRs), efficientthird generationbioreactors,mightrepresentanidealreactortopology foraneconomicsolutiontoDWtreatment.JLRsareabletodealwith veryhighorganicloadingratesduetotheirhighefficiencyof oxy-gentransfer,highmixingandturbulenceachieved.Consequently,
1369-703X/$–seefrontmatter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.bej.2011.08.007
reducedreactorvolumesareneededfortreatment,andlessland isrequired;oxygen(air)isforceddirectlyintothefermentation medium,resultinginsignificantsavingsininstallationand main-tenancecosts[10].Thecombinationofamembranemodulewitha JLRiscalledajetloopmembranebioreactor(JLMBR).
Theobjectiveofthis studyis toexaminetheperformanceof theJLMBR(highperformancecompactmembranesystem)inthe treatment of DW and to compare the results with the litera-ture.AnextensivecharacterizationofthewastewaterandJLBwas performedundervariousoperatingparameters.Thefiltration per-formanceandback-washingefficiencywasalsoexamined.
2. Materialsandmethods
2.1. Materials
2.1.1. Bioreactorsystemset-up
Theexperimental setupof JLB(working volume of approxi-mately18L)wasdesignedatBalikesirUniversity,Environmental EngineeringDepartment,Balikesir,Turkey.Theschematic repre-sentationofthereactorsetupisgiveninFig.1.TheJLBconsisted ofadrafttubeopenatbothendsinsideacylindricalvessel(height 1400mm,innerdiameter100mm)andadegassingtank.Two-fluid nozzleconsistedof twoconcentrictubes. Theouternozzlewas madeofTeflonmaterial(innerdiameter14mm).Theinnernozzle wasastainless-steeltubeof8mmindiameterand1mm thick-ness.Theairtothereactorwasprovidedfromanairpumpthrough theinner stainless-steeltubeviaa gasflow-meter.Gasand liq-uidflowrateswerecontrolledbythevalvesandflow-meterson theirrespectivepipelines.Thetwo-phasejetlocatedatthetopof thereactorcreatesadownwarddirectedtwo-phaseflowinsidethe drafttubeandatthesametimedispersestheairsuckedin,through thegastubelocatedwithintheliquidjet.Duetothemomentum oftheliquidjet,theliquidandthegasinsidethedrafttubeflows
downwardsand,afterthereflectionatthebottomofthereactor riseswithintheannulusbetweenthewallofthereactorandthe drafttube.Attheupperendofthedrafttube,partofthefluidis recycledintothedrafttubethroughasuctionofthetwo-phasejet resultinginare-dispersionofthebubblesandthebiomass pro-ducedinthebiologicalreaction.Thetemperatureofthebioreactor contentwasmaintainedaround22±2◦C bycirculatingcoldgas throughastainlesssteelheatexchangerimmersedinthedegassing tank.DWwaspumpedwithaperistalticpumpfromthefeedtank intothedegassingtank.TherecycleflowfromofbothJLBandthe ultrafiltration(UF) unit weremeasuredbytwo electromagnetic flow-meters.
Theblockdiagramshowedtheflowdirectionofthesystemis giveninFig.2
2.1.2. Membranefiltrationunit
Theseparationof activatedsludge tookplaceintheceramic membraneUFunit(JIUWUHITECH,China),whichwasintegrated intothesystemthroughanexternalcircuittothejetloopbioreactor (Fig.1).Intheexternalcircuit,permeatewasextractedby circulat-ingthemixedliquorathighpressurealongthemembranesurface. Inthiscase,theconcentratedmixedliquoratthefeedsiderecycles backtothedegasificationtank.Thepumpusedforthecirculation ismadeoutofstainless-steel.Theexcesssludgewasremovedvia aperistalticpumpfromthedegasificationtank,oncethedesired biomassconcentrationwasreachedorexceeded.Thespecifications oftheUFmembrane(tubulartypeceramicmembrane)areshown inTable1.Permeatewasmeasuredviaaflowmeterplacedonthe permeateside.Theflowreadingsweretransmittedtoacomputer andrecordedatthedesiredtimeintervals.
Aftereachrun,thesystemwasstoppedandtheUFunit was studiedforbackwashingandcleaningperformancebyoneormore cleaning methods in sequence, for the recovery of membrane permeability.
Fig.1.SchematiclayoutoftheJLMBRreactorsystem[1– circulationpump,2–heat-exchanger,3–flowmeter(forthemembraneunit),4–electromagneticflowmeter(for JLB),5– flowmeter(forthepermeatestream),6–peristalticpump,7–manometer,8–aircompressor,and9–wastewaterfeedtank].
Fig.2. TheblockdiagramoftheJLMBRsystem.
2.1.3. Dairywastewater(DW)
During theexperimentalperiod,approximately 250L ofDW wereobtained3–4timesaweekfroma dairyfactory,OnurSut Co.(Balikesir,Turkey),neartheUniversityCampusandcollectedin thelaboratory.Allexperimentswerecarriedoutinmesophilic con-ditions.DWcollectedfromthefactorywascharacterizedaccording tothedischargeparameters.
2.2. Methods
2.2.1. Analyticalmethods
Samples of influent and effluent were taken daily fromthe JLMBRsystem.ParameterssuchasCOD,MLSS(mixedliquor sus-pended solids), MLVSS (volatile liquor suspended solids), and N-NO3−,andN-NO2−wereanalyzedasdefinedinStandard Meth-ods[11].Totalnitrogen(TN),N-NH4+andtotalphosphate(TP)were measuredbyusingcommercialtestkitsobtainedfromMerck Com-pany.Theorganicnitrogen(ON)wascalculatedbysubtractingthe sumsoftheN-NH4+,andN-NO3−,N-NO2−concentrationsfromTN. Thesoluble(filtered)CODwasdefinedasthefiltratethrough What-manGF/Cglass-fiberfilters,alsousedinthedeterminationofMLSS andMLVSS.Dissolvedoxygen(DO),temperature,pHand conduc-tivityweremeasuredwithamulti-parametermeasurementdevice (suppliedfromWTWCompany)placedinthebioreactor.TheDO dataobtainedthroughtheDOmeterweresenttoacomputerfor furtheranalysis.
2.2.2. Membranecleaning
Backwashing withcompressed air (CA)and chemical clean-ingmethodswereinvestigatedintheexperimentsonmembrane cleaning.Backwashingwasperformedinaflowdirectionopposite toUFbyforcingCAthroughtheceramicmembraneat4.0barfor
Table1
Specificationsoftheceramicmembranecoupledinthesystem.
Manufacturer JiuwuHitech,Chinese Type Ceramicmembrane Rawmembranematerial 99%␣-Al2O3/ZrO2
Poresize 50nm Outsidediameter 40mm Numbersofchannel 37 Diameterofchannel 3.6mm Totallength 1000mm Membranearea 0.24m2 Netweight 2.40kg Permeatedirection Insidetooutside
pHrange 0–14
Temperature <150◦C
3min.Chemicalcleaningwasperformedbyimmersingthemodule
ineachofthecleaningagentfor12h.Thesequenceofthe
chemi-calcleaningwasalkalitreatmentofthemodule,followedbyabrief
rinseofthemodulewithde-ionized(DI)water,thenacidtreatment.
Thealkalisolutioncontaining1–2%NaOHand1–2%HNO3solution
wereusedforthechemicaltreatment.Tobesurethatthe
condi-tionandperformanceofthemembranemodulewereassimilaras
possibleinalltheexperiments,post-cleaningwasperformedafter
everyexperimenttoremoveanyfoulingnotremovedbyaspecific
cleaningmethodintheexperiment.Thiswasaccomplishedbyfirst
soakingthemoduleinanalkalinesolutionfor12h.Themembrane
modulewasthenbackwashedwithDIwaterfor5mintoremove
thealkaliandanyuncloggedmaterialfromtheinteriorofthe
mem-brane.Themodulewasthenimmersedinanacidicsolutionfor12h,
andagainbackwashedwithDIwaterfor5min.Theeffectivenessof
thepost-cleaningoperationwasevaluatedbymeasuringtheclean
waterfluxtodeterminethedegreeofinitialfluxrecovery.The
pro-portionofirreversiblefoulingwasminimizedlessthan7%ofthe
fluxreduction.
2.2.3. Membranefoulinganalysis
Thepermeationfluxofparticle-freewateracrossaclean
mem-branecanbedescribedbyDarcy’sLawas:
J=RP
m (1)
whereJ(m3m−2s−1)isthepermeateflux,P(Pa)trans-membrane
pressure(TMP),(Pas)theabsoluteviscosityofthewater,andRm
(m−1)thehydraulicresistanceofthecleanmembrane (orclean
membraneresistance).Forsuspensionfiltration,thepermeation
fluxwillalwaysbelowerthanthatgivenbyEq.(1).Fluxdeclineis
aresultoftheincreaseofmembraneresistancetothepermeating flow,resultingfrommembranefoulingorparticledepositiononor inthemembrane[12].Thus,thepermeationfluxthroughaUFunit treatingsuspensions,likewastewaterincludingactivatedsludge canbegiven,bymodifyingEq.(1),as:
J= P
(Rm+Rp+Rc) (2)
whereRp(m−1)istheresistanceduetoporeblockingandRc(m−1) theresistancearisingfromcakeformation.
2.2.4. Masstransferanalysisofthejetloopbioreactor
Allthemasstransfertestswereperformedwithtapwaterwhile thesystemwasrunningunderbatchmode(brokenflowlinesand relevant equipment excluded). Before each test, the DOin the waterwasstrippeddownto0.5mgL−1bynitrogenpurging.After
switchingovertoanairsupplyline,theconcentrationofoxygen wasmeasuredasafunctionoftimeusingaDOmeter(WTW,350i) equippedwithanoxygenprobe.TemperatureandpHwerealso measuredwiththesamedevice.TheDOdataobtainedthroughthe DOmeterwereinputintoacomputerforfurtheranalysis.
Masstransferisgenerallyexpressedintermsofmasstransfer perunitvolumeofthereactorandthemasstransfercoefficient (KLa)isexpressedastheoverallvolumetricmasstransfer coeffi-cient.KLacanbecomputedbyusinganon-linearexpression: C=C∗
s−(Cs∗−Ci)×e−(KLa)t (3)
where,CistheDOconcentrationinthemediumatagiventimet, C∗
s issaturationandCiisinitial(t=0)oxygenconcentrationunder experimentalconditions.
2.2.5. Start-upandtreatmentconditions
Activatedsludgeofdifferentorigins(BalikesirUrban Wastew-ater Treatment Plant and Manisa Organized Industrial Region WastewaterTreatmentPlant)wasadaptedtotheDWand then usedasinoculafortheJLMBR.Inordertoincreasetheamountof activatedsludgeinthebioreactor,initially,theJLMBRwasoperated inarepeated-batchprocessof2–3dayseachforatotalperiodof20 days.Attheendofthisperiod,theJLMBRwasfedcontinuouslyand theconcentrationofactivatedsludgereachedwasapproximately 2300mgL−1.Duringboththebatchandcontinuousoperating con-ditions,DO levelsinthereactorweremaintainedata range of approximately1.0–3.0mgL−1.Duringthetestperiodofabout35 weeksthefollowingoperatingconditionsforthereactorandUF unitwerevaried:
√
Biomassconcentrationinthereactor:2312–38684mgL−1 √
Loadperunitreactorvolume:4.8–53.6kgm−3g−1 √
Hydraulicwastewaterresidencetime:1.9–7.7h √
Sludgeage:2.6–79.6h √
Velocitywithinthemembranemodule:1.32–2.65ms−1
√
Transmembranepressure:0.6–4.0bar
3. Resultsanddiscussion
3.1. CharacterizationoftheDW
Thevolume,concentrationandcompositionoftheeffluentfrom inadairyplantaredependentonthetypeofproductbeing pro-cessed, the production program, operating methods, design of processing plant,thedegree of wastewater management being applied,andsubsequently,theamountofwaterbeingconserved [13].Informationaboutthegeneralcharacteristicsofdairy wastew-atersfromfull-scaleoperationsisinfactscarceinliterature[3],and fewcomprehensivestudieshavebeencarriedoutthatmight pro-videextensiveinformationabouttheparticularcharacteristicsof dairywastewatersfromvariousfull-scaleoperations.Inthisstudy,
dairywastewaterssuppliedfromacheesefactory(OnurSutCo.) nearbytheUniversityCampusinwhichvarioustypesofcheese pro-ducedwereusedasthewastewater.Thefactoryhasbeentreating itseffluentsinabiologicalwastewatertreatmentplant(anaerobic sludgeblanketreactor+classicalactivatedsludge).Sincethewaste effluentsfromthedairyindustryareusuallygenerated intermit-tently,theflowratesoftheseeffluentschangesubstantially,sothe wastewaterwastakenfromtheequalizationtankoftheplant.The DWwasthencharacterizedindetailaccordingtodischarge param-eters.Thecharacterizationexperimentscontinuedovera2-year period and thedatawere thenanalyzed statistically. The sum-maryofdataobtainedfromthecharacterizationexperimentson thegeneralpropertiesofdairywasteeffluentsisgiveninTable2.
Dairyindustrywastewatersaregenerallyproducedinan inter-mittent way and thus differin concentration and volume over theproduction period. Thus, theDW concentrations fluctuated ina wideband. Significantfractions oftheorganiccomponents andnutrientsindairywastestreamsarederivedfrommilkand milkproducts.Inindustrialdairywastewaters,nitrogenoriginates mainlyfrommilkproteins,andispresentinvariousforms;eitheran organicnitrogen(proteins,urea,nucleicacids),orasionslikeNH4+, andNO2−andNO3−.Phosphorusisfoundmainlyininorganicforms likeorthophosphate(PO43−)andpolyphosphate(P2O74−),butcan alsobe foundin organic forms [14]. Suspended solids in dairy wastewatersoriginatefromcoagulatedmilk,cheesecurdfinesor flavouringingredients.ConcentrationsofSSandvolatileSSareused toevaluatewastewaterstrengthandtreatability.Thereisalsoa strongtendencytowardshighCODconcentrationsindairyindustry wastewatersalthoughtheseconcentrationstendtofluctuate.
Dairy wastewaters include easily degradable carbohydrates, mainlylactose,aswellaslessbiodegradableproteinsandlipids.In cheese-processingwastewater,97.7%oftotalCODwasaccounted forbylactose,lactate,proteinandfat[15].Lactoseisthemain car-bohydrateindairywastewaterandisreadilybiodegradablebythe bacteria.However,dairywastewater,becauseofitsproteinand lipidcontent,caneasilybedefinedasacomplexsubstratetype. Lipidsarepotentiallyinhibitorcompoundsthatarealways encoun-teredduringanaerobictreatmentofdairywastewaters.
3.2. MasstransfercapacityoftheJLB
Inthetreatmentofhighorganiccontentindustrialwastewaters, anaerobictreatmentprocessestendtobefavouredoveraerobic processesduetotheirwell-knownbenefitssuchasmethaneyields, lesssludgegeneration,andlowernutrientrequirements.However, thedisadvantagesassociatedwithanaerobictreatmentsarehigh capitalcost,longstart-upperiods,andtheneedforstrictcontrol ofoperatingconditions,greatersensitivitytovariableloadsand organicshocks,aswellastoxiccompounds[2].Aerobictreatment processesarethuscommonlyusedalongwithanaerobicprocesses
Table2
Thecharacteristicsofdairywasteeffluents.
Parameter Concentration(mgL−1)
Maximum Minimum Average StandardDeviation
TotalCOD(CODt) 9004 921 3445 1323
SolubleCOD(CODs) 8064 635 2445 1236
CODs/CODt 0.90 0.68 71 0.14
BOD 6080 483 1860 394.5
Suspendedsolids(SS) 804 134 398.31 143.8 Volatilesuspendedsolids 506 168 329.25 121.16
Totalnitrogen(TN) 230 8.00 108.84 51.50 AmmoniumN 91.00 2.5 23.42 29.38 NitrateN 8.2 1.8 6.7 5.40 Totalphosphor 111.5 9 35.7 18.32 Oil-Gress 142 400 288 77.86 pH 5.78 5.52 5.63 0.07
Fig.3. TheevolutionofKLaversusthejetvelocitiesatdifferentairflowrates.
indairywastewatertreatmentinordertoachievetheeffluent
dis-chargelimitsforagro-industrywastewaters[3].
Inthisstudy,weexaminedandcomparedtheperformanceand efficiencyofanaerobicsystemforDWs,whichareahighstrength industrialwastewater.JLB,whichisclassifiedasathird-generation compactreactor,waschosenastheaerobicsystem.Theaimof this choicewastooperatethesystem athighorganicloadings andlowerhydraulicretentiontimes.Intheaerobicsystems,the mostfavourablecontributionfromthebioreactorisitsabilityto producehighoxygentransfercapacityastheoxygensupplyplays anactivepartinthesuccessandeconomicsofthesystem.Inthis stageofthestudy,theoxygentransfercharacteristicsoftheJLB wereinvestigatedanddiscussedbecauseofthemasstransfer kinet-icsquantifiedbythevolumetricoxygentransfercoefficientKLain practice.Fig.3showstheevolutionofKLaversusthejetvelocities atdifferentairflowrates.
TheJLBachievedKLavaluesbetween101and280h−1under var-iousoperatingconditions.ItcanbeseenthattheKLavaluesinthe presentstudyareabout100timeshigherthanthoseofthe conven-tionalreactorsystem.KLavaluesincreasedwiththeincreasingjet velocities,asseeninFig.3.Theprincipleinthisreactortypeisthe utilizationofthekineticenergyofahighvelocityliquidjetto main-tainthegasphaseandcreateafinedispersionoftwophases.The highshearratesfromtheliquidjetproducedveryfinegasbubbles, thustheequipmentgeneratedveryhighinterfacialareasandhigh volumetricmasstransferrates.Also,themoreclear-cutincreasesin KLavalueswereobservedwithincreasingairflowrates.Gas hold-upandinterfacialareaforthemasstransferalsoincreasedwiththe increasingairflowrates.Intheexperiments,impactofjet veloc-itiesonmixedliquortemperaturewasnotinvestigatedbecause aheatexchangerwasusedtokeepthereactorcontentatafixed temperature.
3.3. ThetreatmentefficienciesandperformanceoftheJLMBR system
Theenergyconsumptionincreaseswiththeincreasingjet veloc-ityinJLB.Consequently,ithasbeenwantedtooperatethesystem atminimumjetvelocities.Butatlowervelocities,thereactormixed liquorwasnot recirculatedand loopedin theJLB.Therefore,at thefixedairflowrate,theminimumjetvelocitywasselectedfor theliquid(reactormixedliquor)loop.Afterthecharacterization experiments forthe JLB, thesystem wassetto anair flowrate of1000Lh−1 andajetvelocityof40ms−1.TheJLMBRwasthen seededwithmixedliquor fromactivatedsludgeplants treating domestic wastewater (Balikesir Municipal Wastewater Treat-mentPlant[tricklingfilters])andindustrialwastewater(Manisa
Organized Industry Region Wastewater Treatment [activated sludge]).TheDWwaschosenforthefeedbecauseithasavery highBODandeaseoftransportationtothelaboratoryinits con-centratedstate,whichwasessentialbecauseofthelargeamount ofBODrequiredtorunthejetloopreactorathighloadingrates. AlthoughtheJLMBRwascontinuouslyoperatedformorethan15 months,Fig.4showsresultstakenoveraperiodof250days.This periodcouldbeconsideredasrepresentativeofalltheexperiments. Theappliedloadingrates werekepthighover aperiod of sev-eraldaystoallowthebiomasstobecomeacclimatizedtotheDW. Sincethecollectedwastewaterfromtheplantwasfeddirectlyinto thesystemattheselectedinfluentflowrates,theorganicloads variedaccordingtotheconcentrationofDW.Duringthefirst7 days,CODremovalefficienciesweremeasuredasinstabledueto thechangefrombatchtocontinuousoperation.Afterthis acclima-tionperiod,efficienciesstartedtoimprove.JLMBRwasfedorganic loadswithloadingratesthatvariedfrom4.8to53.6kgm−3d−1and withhydraulicretentiontimes(HRT)thatvaried,correspondingly, from1.9to7.8h.Eachvolumetricloadingtothereactorwas con-tinueduntilanincomingwastewateramountofmorethan10–12 timesthereactorvolumehadpassedthroughtheJLMBR.Whenthe effluentCODvaluesremainedinanarrowband,thesystemwasin steady-state.
JLMBR was operated at a loading rate of 26.7kgm−3d−1 between10and13days,under32.4kgm−3d−1between56and62 days,andunder34.3kgm−3d−1between85and89days.Under these organic loading rates, the efficiencies were estimated at 95.5–97.7%between10and13days,95–97%between56and62 days,and96–98%between85and89days.Organicloadingratewas thenincreasedto53.6kgm−3d−1 onday92.Intheseconditions, 95.6–97.8% COD treatment removal efficiencies were achieved fromthesystem.JLMBRwasrunundervaryingorganicloading ratesuntilday220.Duringthistimeperiod,variousprocessand operatingparameterswereexamined.Theorganicloadingrates wereincreasedto32.6kgm−3d−1onday220and40.3kgm−3d−1 onday224.Inthemeantime, treatmentefficienciesof approxi-mately99%and98%,respectively,wereobtained.Inthelastperiod of the study, the loading rates went up to 45.0kgm−3d−1 on day233,53.1kgm−3d−1onday238,and50.5kgm−3d−1onday 246.Evenfortheseveryhighorganicloads,treatmentefficiencies between95and98%wereobtained.
Itisinterestingtonotethatveryhighfluctuationsintheinlet loadingsresultedinonlyminorreductionsinthesystem perfor-mance.Onday53theCODloadingwasincreasedfrom17.4to32.4, resultinginareductionofCODremovalefficiencyfrom99%to95%. Similarly,onday232,anincreaseinCODloadingfrom25to45 resultedina3%reductioninCODremovalefficiency.
TheJLMBRsystemcouldbeoperatedatveryhighF/Mratios (food/microorganism)comparedtoconventionalactivatedsludge systems.Whileonday146,F/Mratioswereincreasedfrom2.07to 14.9kgCODkgMLVSS−1d−1,theCODremovalefficiencydecreased slightly,from98%to96.5%.Similarly,onday184,anincreasein F/Mratiofrom1.8to11.4kgCODkgMLVSS−1d−1resultedina1% reductioninCODremovalefficiency.Fluctuationsintheapplied F/Mas observedduringdays87–94similarlyresultedin onlya smallreductioninCODremovalefficiency.Ingeneral,JLMBRhas demonstratedahightolerancetoshort-termchangesintheapplied highCODloadingrates[16].TheconditionsprevailingintheJLB, withmuchhigherF/Mvaluesthanconventionalactivatedsludge systemsandwithaveryhighgrowthrateofactivebacteria,may bebeyondthelimitswithinwhichfilamentousorganismscan com-pletesuccessfullywiththerestofthepopulation[17].
Foaminginthereactorwasfoundtobeacommonoccurrence whenchangingloadingratesandfeedingwithhighinfluentCOD concentrations.Inotherwords,excessivefoamingwasobserved in thebioreactorat highF/M ratios. Whenthesystemreached
0 20 40 60 80 100 120 140 160 180 200 220 240 Time, days 0 10 20 30 40 50 60 70 80 90 100 COD treatment efficiency, % 0 5 10 15 20 25 30 35 40 45 50 55 60 Applied F/M, kg COD kg MLVSS -1day -1 Organic
loading rate, kg COD m
-3
day
-1
COD treatment efficiency Applied F/M Organic loading rate
Fig.4.TheeffectsoforganicloadingratesandF/MratiosonCODremovalefficienciesinDWtreatmentusingJLMBRsystem.
steady-state conditions, the foaming decreased toa minimum. Fig.5showsthechangesofMLSSandHRTversustime.MLSS con-centrationsintheJLMBRsystemweremeasuredbetween1000 and 38,000mgL−1. The system was mainly operated at MLSS
concentrationsover5000mgL−1.HRTvariedbetween1.9and7.7h. Nevertheless,thesystemwasusuallyoperatedwithanHRTof4h.It wasobservedthattheeffluentconcentrationswerebadlyaffected underHRTof1.9h.Thesludgeageswerechangedfrom2.6to79.6h
duringthestudy.Undertheseoperatingconditions,theeffluent’s CODconcentrationsweremeasuredtobebelow300mgL−1(Fig.5). Nevertheless,effluentCODvalueswereevaluatedatapproximately 100mgL−1orlowerformostoftheloadingconditions.
TheactivatedsludgeintheJLBwashighlymotilewhenobserved underamicroscopeandappearedtoflocculatelessreadilythan ses-sileciliates/protozoa.Besides,microscopicexaminationofbiomass showedthatnofilamentousbacteriaorprotozoawerepresentin theflocks.Thesituationoftheactivatedsludgemaybeattributed tothehighshearforcesinthenozzlecombinedwithhighgrowth ratesoftheactivebacteria,togetherwiththehighappliedF/Mand thenatureofthewastewater.Allofthesefactorsachievedahigh degreeofmicrobialselectionintheactivatedsludgeofthe biore-actor.Otherwise,thestructurewiththesmallestflocksizeallowed highcross-sectionalsurfaceareaexposuretooxygenandsubstrate transferforthemicroorganismsintheflock.Thisalsocontributed tothestrongperformanceoftheJLB.
Asthereactionrateinwastewatertreatmentisdirectly propor-tionaltobiomassconcentration,theorganicremovalrateincreases withMLSSconcentration[18].Thestudywasachievedatahigh biomassconcentration.TheJLMBR,asahighmasstransfer perfor-mancereactor,allowedmuchhigherbiomassconcentrationsthan wouldapparatuswithaconventionalmasstransfer.Nonetheless, theincreaseinbiomassconcentrationislimitedbythephysical propertiesofthesludge-wastewater-suspension[18].Otherwise, a clearrelationship couldnot beseen betweenMSSS and COD removalefficiency,inFig.5.Thisresultcouldbeattributedtotwo main factors.First, since a highMLSS concentrationand active biomassweregenerallyobtainedinthesystem,highperformance wasachievedin all situations.Secondly, withthe increasingin biomassconcentrationincreasedtheparticlesizesoftheflock con-siderably[16],andthustheoxygenandsubstratetransfertothe microorganismsweredecreased.Inaddition,theincreasingMLSS concentrationincreasedtheviscosityinbioreactorcontent,which alsoincreasedbubblesizesandcausedbubblecoalescenceinthe system.Thus,masstransferdecreasedbecauseofdecreased cross-sectionalsurfaceareaandgashold-up.
3.4. Ultrafiltrationofthesludgeandbackwashingperformance Theeffluentqualityarisingfromabnormalbacterialactivities suchasbulkingandfoamingintheaerationtankmakesseparation ofsludgefromwaterdifficult.Thishasgivenspaceformembrane filtrationtobeincreasinglyusedwiththeactivatedsludgeprocess [12].Therateofsludgeseparationfromeffluentisnolongerlimited bythesettleabilityoftheactivatedsludge,withamembrane filtra-tionunitreplacingaclarifier.Inthisstudy,themembranefiltration unitwasadeterminantfactorforthesystem’sperformancesince thereactorcontentwasoperatedwithasmallanddispersedflock structureandthesludgehadpoorsettleability.Thetubularceramic membrane filters were chosen as ultrafiltration unit and oper-atedatcross-flowfiltrationmode.Ceramicmembraneshavemany uniqueadvantages;suchasexcellentresistancetoacid/alkaline andoxidationchemicals,solventstability,highthermalstability, excellentmechanicalandabrasiveresistance,extremelylongwork lifecomparedwithpolymericmembrane,andeasycleaningand sanitizingwithbackflushing[19].Fig.6presentsthepermeateflux oftheceramicmembraneasafunctionoftimeforthefiltrationof variousMLSSconcentrations.
AsseeninFig.6,afterasharpdecreaseinthepermeatefluxof theUFmembraneduetoadsorptionandporeblockingfollowed bythe formationof a fouling layerwithinthefirst 10min,the permeatefluxreachedapseudo-steadystate.Inthis period,the attachmentoffoulantsontothemembranesurfaceduetothedrag forceofpermeateflowwasalmostbalancedbythedetachmentof foulantsfromthemembranesurfaceduetoshearforcebycrossflow
0 10 20 30 40 50 60 Time, min 0 20 40 60 80 100 120 140 160 180 Permeate flux, L m -2 h -1 10000 13000 15000 19000 25000 MLSS, mg L-1
Fig.6.VariationoffluxeswithtimeatdifferentMLSSconcentrationsofthe mem-braneunit(P=2bar,Vc=2.87ms−1).
velocityandbackdiffusionbyconcentrationgradient.Membrane foulingisgenerallycharacterizedasareductionofpermeateflux through themembrane asa result of increasedflow resistance toporeblocking,concentrationpolarization,andcakeformation [12,20].Manysharpdecreaseswereobservedinthefluxesatlow MLSSconcentrations(inFig.6).Thiscouldbeexplained inthat thefoulingatlowMLSSconcentrationsresultedindifferent mech-anisms.Theliteraturesuggeststhat poreblockingisthefouling mechanismsatlowMLSSandcakeformationisthefouling mech-anismathighMLSSconcentrations[12,21,22].
Biofoulingisanothermajorproblemarisingfrombiofilm for-mationintheporesoronthesurfaceofthemembrane[23].During thestudy,biofilmformationwasobservedintheceramic mem-brane’sflowchannels.Atlongoperationtimes,itwasseenthatthe thicknessofthebiofilmlayerincreased.Biofoulingmaybeinitiated withthedepositionofindividualbacteriacellsonthemembrane surface;thecellsthenmultiplyandformabiofilm.
However,thedevelopmentofmembranebioreactorshasbeen limitedbyproblemsofmembranefoulingduringfiltrationofthe activatedsludge.Foulingofthemembranedecreasesthefiltration fluxesandthusthetreatedwaterflow,andincreasesthe operat-ingcosts.Sincethecloggedmembranesmustbecleanedand/or replaced,researchershavedevelopedvariousstrategiestoreduce membranefoulingandtoimprovemembranecleaningefficiency forfluxrecovery.Thesestrategiesincludethedevelopmentofnew membranematerials,newdesignofamembranemodule, modifica-tionofflowpatternsandincorporationofinsituorexsitucleaning regimesinthemembraneunit[20,24–26].Acombinationofthese strategiesmaybeusedinsomeprocesses[12,27].Inthisresearch, theefficiencyofbackwashingwithCA,immersingthemodulein DIwater,andchemicalcleaningwerealsoexamined.Fig.7shows theefficiencyofbackwashingwithair.
Theacceptablefluxrecoverywasobtainedbybackwashingwith CA.Insteadof;highfluxrecoverieswereachievedathighMLSS concentrationsbybackwashingwithCA.Namely,the backwash-ingprocesswithCAwasan effectiveproceduretoseparatethe looselyattachedcakefromthemembranealongwiththesoluble microbialproductentrappedinit.Thisresultexplainedthatthe effectivemechanisminmembranefoulingwasincakeformation. Duringthefirststages ofthefiltrationcycle,undertheelevated fluxapplied,strongpermeationdragwouldaccelerateMLSS accu-mulation.Acakelayerwouldquicklyformtoactasasecondaryor dynamicmembranetoentrapfurthersolublemicrobialproducts. Inthisstudy,backwashingwasappliedatone-hourperiodsand
0 50 100 150 200 250 300 350 Time, min 20 25 30 35 40 45 50 55 Membrane flux, L m -2 h-1 Back-washed Not back-washed
Fig.7. Effectofbackwashingwithaironfluxrecoveryatthemembraneunit (P=2bar,Vc=2.87ms−1).
thusfurtherfoulingwasdiminishedordecreasedtoaminimum. Nonetheless,all fouling mechanismswereeventuated together, especiallywithactivatedsludgefiltration[12,28].
Theexperimentsonfoulingmechanismsandbackwashingor cleaningthemembraneareongoing.
4. Conclusions
Inthisstudy,thetreatmentofdairyindustrywastewaterwith ahighperformanceJLMBRwasinvestigatedand foundtowork atveryhighefficiencies.Theresultsobtainedaresummarizedas follows:
• Fromthecharacterizationexperiments,theconcentrationswere determinedtobe6080–483mgBODL−1,9004–921CODTmgL−1, 8064–635mgCODSL−1,230–8mgTNL−1,112–9mgTPL−1,and 804–134mgSSL−1,givingaBOD/CODTratioof0.68–0.52anda CODS/CODTratioof0.90–0.68.Thesubstantialfluctuations espe-ciallyinCODandBODconcentrationsoriginatedfromtheratio ofcheesewheyintroduction.
• In the investigated experimental operating conditions, KLa rangedbetween100and285h−1.These valuesare100times higherthanthoseforaconventionalairdiffuser(supplier appa-ratus).
• Aloadingrateof45kgCODm3d−1wasachievedwitha98%COD removalefficiencyatanHRTof2.8days.Also,treatment efficien-ciesof97–98%wereachievedunder53kgCODm3d−1 organic loadingrateand3hHRT.Theperformancevalues(efficiencies accordingtoorganicloadingratesandHRT)achievedinthisstudy coincidewiththehighestvaluesintheliteratureforDW treat-ment.Experimentalresultsshowedthatthecombinationof a highratebiologicalreactorandmembranefiltrationwasan effi-cient,reliableandcompactprocessforbiologicalDWtreatment. Moreover,excellentpurificationresults(96–99%CODremoval) fromthecombinationJLBRandceramicmembraneUFsystem reducedthecostofadditionaltreatment.
• HighMLSSconcentrationscouldberetainedinthebioreactor(up to38,000mgL−1)withthecontributionoftheUFmembraneunit. Consequently,thehighMLSSappliedcontributedconsiderably tothehighremovalefficienciesand performance.Inaddition, theJLBRcouldbeusedfortheoxygentransferfortheactivated sludge.
• Assuminganexcesssludgeconcentrationof15,000mgMLSSL−1 for a conventional activated sludge plant [18] and
38,000mgMLSSL−1 for the JLMBR system, the excess sludge volumeoftheJLMBRcostswasreducedbyabout70%.Thisisa hugeeconomicadvantage.
• Duringthestudy,theactivatedsludgeinthebioreactorformed anon-flocculatingmotilebacteriastructurethatwasslimyand poorly settleable. It was observed that the fluxes decreased withincreasingMLSSconcentrations.Cakeformationfoulingwas determinedas thedominantfouling mechanismand resulted inslowerpermeationflux decayover time.MLSS inthecake layerwerebestremovedwithbackwashing.Compressedairwas usedtocleanthemembraneeffectivelyoffoulingcausedmainly bycakeformation.Acombinationcleaningmethodofchemical cleaning,DIwaterimmersionandcompressedairbackwashing wasthemosteffectiveinrecoveringpermeationflux.
Acknowledgement
The authors wish to thank The Scientific and Technological ResearchCouncilofTurkey(TÜB˙ITAK)fortheirfinancialsupport ofthisresearch(ProjectNo:107Y260).
References
[1]B.Sarkar,P.P.Chakrabarti,A.Vijaykumar,V.Kale,Wastewatertreatmentin dairyindustries—possibilityofreuse,Desalination195(1–3)(2006)1–306. [2]T.J.Britz,M.VanDerMerwe,K.H.J.Riedel,Influenceofphenoladditionsonthe
efficiencyofananaerobichybriddigestertreatinglandfillleachate,Biotechnol. Lett.14(1992)323–328.
[3]B.Demirel,O.Yenigun,T.T.Onay,Anaerobictreatmentofdairywastewaters:a review,ProcessBiochem.40(8)(2005)2583–2595.
[4] B.Balannec,G.Gesan-Guiziou,B.Chaufer,M.Rabiller-Baudry,G.Daufin, Treat-mentsofdairyprocesswatersbymembraneoperationsforwaterreuseand milkconstituent’sconcentration,Desalination147(2002)89–94.
[5]I.Koyuncu,M.Turan,D.Topacik,A.Ates,Applicationoflowpressure nanofiltra-tionmembranesfortherecoveryandreuseofdairyindustryeffluents,Water Sci.Technol.41(1)(2000)213–221.
[6]C.H.Xing,X.H.Wen,Y.Qian,D.Sun,P.S.Klose,X.Q.Zhang,Foulingandcleaning ofmicrofiltrationmembraneinmunicipalwastewaterreclamation,WaterSci. Technol.47(1)(2003)263–270.
[7] E.Bouhabila,R.Ben,AimandBuissonFoulingcharacterisationinmembrane bioreactors,Sep.Purif.22–23(2001)123–132.
[8]T.H.Bae,T.M.Tak,Interpretationoffoulingcharacteristicsofultrafiltration membranes duringthefiltrationofmembranebioreactormixedliquor,J. Membr.Sci.264(1–2)(2005)151–160.
[9]J.Manem,R.Sanderson,WaterTreatmentMembraneProcess,AWWAResearch Foundation,McGraw-Hill,1996(chapter17).
[10] C.J.Bloor,G.Anderson,A.R.Willey,Highrateaerobictreatmentofbrewery wastewaterusingthejetloopreactor,WaterRes.29(1995)1217.
[11]APHA,StandardMethodsfortheExaminationofWaterandWastewater, Amer-icanPublicHealthAssociation,Washington,DC,2005.
[12] A.L.Lim,R.Bai,Membranefoulingandcleaninginmicrofiltrationofactivated sludgewastewater,J.Membr.Sci.216(1–2)(2003)279–290.
[13]T.J.Britz,C.Schalkwyk,Y.T.Hung,Treatmentofdairyprocessingwastewater, in:L.K.Wang,Y.-T.Hung,H.H.Lo,C.Yapijakis(Eds.),WasteTreatmentinthe FoodProcessingIndustry,TaylorandFrancisGroup,NewYork,2006,pp.1– 29.
[14]E.Guillen-Jimenez,P.Alvarez-Mateos,F.Romero-Guzman,J.Pereda-Martin, Bio-mineralizationoforganicmatterasaffectedbypH,theevolutionof ammo-niumandphosphates,WaterRes.34(2000)1215–1224.
[15]S.Hwang,C.L.Hansen,Characterizationofandbioproductionofshort-chain organicacidsfrommixeddairy-processingwastewater,Trans.Am.Soc.Agric. Eng.41(1998)795–802.
[16]S.Lübbecke,A.Vogelpohl,W.Dewjanin,Wastewatertreatmentinabiological highperformancesystemwithhighbiomassconcentration,WaterRes.29(3) (1995)793–802.
[17]F.B.Dilek,G.K.Anderson,J.Bloor,Investigationintothemicrobiologyofahigh ratejet-loopactivatedsludgereactortreatingbrewerywastewater,WaterRes. 34(5–6)(1996)107.
[18]G.Tchobanoglous,F.L.Burton,WastewaterEngineering:Treatment,Disposal andReuse,thirded.,Metcalf&EddyInc.,McGraw-Hill,NewYork,1991. [19]S.Judd,B.Jefferson,MembranesforIndustrialWastewaterRecoveryand
Re-use,ElsevierScienceLtd.,NewYork,2003.
[20]R.B.Bai,H.F.Leow,Microfiltrationofactivatedsludgewastewater—theeffect systemoperationparameters,Sep.Purif.Technol.29(2002)189.
[21] B.Farizoglu,B.Keskinler,Sludgecharacteristicsandeffectofcrossflow mem-brane filtrationonmembranefoulingin ajetloopmembranebioreactor (JLMBR),J.Membr.Sci.279(1–2)(2006)578–587.
[22]J.S.Kim,C.H.Leeand,H.D.Chun,Comparisonofultrafiltrationcharacteristics betweenactivatedsludgeandBACsludge,WaterRes.32(1998)3443–3451.
[23]H.F.Ridgeway,H.C.Flemming,Membranebiofouling,in:J.Mallevialle,P.E. Odendal,M.R.Wiesner(Eds.),WaterTreatmentMembraneProcesses, McGraw-Hill,NewYork,1996,pp.6.1–6.62.
[24]P.Wang,K.L.Tan,E.T.Kang,K.G.Neoh,Plasma-inducedimmobilizationofpoly (ethyleneglycol)ontopoly(vinylidenefluoride)microporousmembrane,J. Membr.Sci.195(2003)103.
[25]K.Parameshwaran,A.G.Fane,B.D.Cho,K.J.Kim,Analysisofmicrofiltration per-formancewithconstantfluxprocessingofsecondaryeffluent,WaterRes.35 (2001)4349.
[26]K.Yamamoto,M.Hiasa,T.Mahmood,T.Matsuo,Directsolid–liquidseparation usinghollowfibremembraneinanactivatedsludgeaerationtank,WaterSci. Technol.21(1989)43.
[27]C.Chiemchaisri,Y.K.Wong,T.Urase,K.Yamamoto,Organicstabilizationand nitrogenremovalinmembraneseparationbioreactorfordomesticwastewater treatment,WaterSci.Technol.25(1992)231.
[28]J.Wu,P.Le-Clech,R.M.Stuetz, A.N.Fane,V.Chen,Novelfiltrationmode forfoulinglimitationinmembranebioreactors,WaterRes.42(14)(2008) 3677–3684.
