Desalination of domestic wastewater effluents: phosphate ...gwri-ic.technion.ac.il/pdf/gwri_abstracts/2008/62.pdfBatch desalination of both kinds of pretreated effluents showed also

Desalination 222 (2008) 230–242

Presented at the conference on Desalination and the Environment. Sponsored by the European Desalination Societyand Center for Research and Technology Hellas (CERTH), Sani Resort, Halkidiki, Greece, April 22–25, 2007.

Desalination of domestic wastewater effluents: phosphate removal as pretreatment

Ilan Katz, Carlos G. Dosoretz* Faculty of Civil and Environmental Engineering and Grand Water Research Institute,

Technion — Israel Institute of Technology, Haifa 32000, IsraelTel. +972-4 8294962; Fax +972-4 8228898; email: [email protected]

Received 20 January 2007; accepted 30 January 2007

Abstract

Water shortage has become a global issue and not only a problem relevant to arid zones. In the past decadedesalination of sea and brackish water has been found to be a technically and economically acceptable solution forwater shortage. More recently, desalination of wastewater effluents has also begun to be considered for sustainablewater reclamation and reuse.

Performance of membrane desalination processes, such as reverse osmosis (RO), is restricted by differentfouling and clogging constraints, including inorganic scaling as well as organic fouling and biofouling. Utilization ofeffluents for desalination introduced to the membrane systems new feed water characteristics with reduced salinitybut higher concentration of dissolved organic matter and predominant inorganic ions that are less common in seawater. Phosphate scaling is found to be a main obstacle for effluents desalination, mainly due to the lack ofappropriate antiscalants to cope with its precipitation, especially at high recovery rates.

This work was aimed to study the chemical coagulation of phosphate from wastewater effluents as a pretreat-ment for RO desalination. Based on preliminary jar test experiments, sodium aluminate (SAL) at concentrationsof 20–30 mg/L, was found as an efficient coagulant for phosphate and turbidity reduction. SAL coagulation wastested either as simultaneous phosphate removal during membrane bioreactor (MBR) treatment or as post-removal of phosphate from secondary effluents from activated sludge process, followed by microfiltration. Bothalternative treatments resulted in almost complete phosphate removal (≥99%).

Batch desalination of both kinds of pretreated effluents showed also that calcium carbonate was a secondscaling agent upon phosphate removal. The simultaneous MBR-coagulation, which in addition to completephosphate removal displayed a significant reduction of alkalinity (75%), produced higher quality effluents andreduced the effect of calcium carbonate scaling, as compared to the post-coagulation of secondary effluents.Furthermore, the simultaneous MBR-coagulation process showed no detrimental effects on the MBR membraneperformance or on the biological activity and avoided the need of an additional sedimentation or filtration step

*Corresponding author.

0011-9164/06/$– See front matter © 2006 Published by Elsevier B.V.doi:10.1016/j.desal.2007.01.160

I. Katz, C.G. Dosoretz / Desalination 222 (2008) 230–242 231

prior to RO. In conclusion, the MBR-chemical coagulation process seems to be a technically feasible pretreatmentof domestic effluents for further desalination.

Keywords: Phosphate scaling; Coagulation; Desalination; RO; MBR; Wastewater treatment

1. Introduction

Water scarcity is a fact of life in arid andsemi-arid regions where agricultural, domesticand industrial demands compete for limitedresources. Continuous population growth, risingstandards of living, industrialization and urban-ization limit the freshwater available for agricul-tural practices. In Israel, where total demand forfresh water already exceeds the average naturalsupply, the freshwater deficit in agriculture isalready being replaced in a great extent bytreated wastewater (30% of the total water usedin agriculture today and approx. 50 in 2010 [1]).However, the treatment of effluents prior to itsuse in irrigation, as is presently practiced, is notadequate for sustaining long-term agriculturalproduction. Irrigation with inadequately treatedeffluent generates a gradual deterioration of soiland groundwater properties, specifically soiland groundwater salinization, and soil structuredeterioration.

Water qualifying for unrestricted irrigationcan be safely reclaimed applying combined state-of-the-art pressure driven membrane separationtechnologies capable of generating effluentsfree of pathogens, colloids, dissolved solids andorganic contaminants. Although pressure-drivenmembranes are known technologies, some issuesregarding their application on a large practicalscale for wastewater reclamation must be resolved,especially for reverse osmosis (RO).

The major constraints during RO of treatedeffluents, promoting fouling and thus limitingseparation efficacy, are the presence of solubleorganic matter that tends to adsorb producingorganic fouling and to promote microbialfouling — biofouling — as well as the scalingforming salts such as phosphate, that cannot be

easily avoided by use of antiscalants, and carbon-ate [2,3]. Also when organic matter and nutrientsare almost completely removed and biofoulingis hindered, in practical terms, calcium phos-phate, and in part carbonate, scaling became thedominant factor for membrane fouling and fluxdecrease during desalination of secondary or ter-tiary effluents, limiting RO recovery rates [4].Calcium phosphate scaling during secondaryeffluents desalination by RO was reported aslimiting factor for RO recovery during desalina-tion of treated effluents [5–7]. Williams et al. [8]reported that the major obstacle to operating anRO process for desalination of marginal or riverwater at high recoveries (above 85%) is the pre-cipitation of sparingly soluble inorganic salts.Furthermore, at recoveries higher than 95%, aswould be needed for feasible large-scale RO ofmarginal water, antiscalants are not effectiveand pH control does not prevent precipitation ofminerals such as barium sulfate and calciumsulfate, which cannot be removed by chemicalcleaning [3,7,9]. High recovery rate of treatedeffluents, especially in inland sites, is notrequired only in terms of high water recovery butalso disposal of large brine volume is not feasiblefrom either an economic or resource-conservationstandpoint.

Chemical precipitation of phosphorous, whichis mainly present as inorganic phosphates such asorthophosphates and polyphosphates, is a com-monly used process in water treatment [10,11].Simultaneous removal as well as post-precipi-tation of P is also applied in secondary activatedsludge to supplement biological removal duringwastewater treatment [12–17].

Al and Fe salts are conventionally consid-ered as precipitants for P removal. At the pH of

232 I. Katz, C.G. Dosoretz / Desalination 222 (2008) 230–242

wastewater, the use of Al or Fe(III) salts forphosphate removal may be related more to thecoagulation mechanism rather than precipita-tion. In general, two major mechanisms havebeen considered to be linked to the coagulation ofphosphate ions with Al/Fe(III) salts: formation ofAl/Fe-hydroxo-phosphate complexes with gen-eral formula of the type Me(OH)3–x(PO4)x, inwhich M indicates Al or Fe, and adsorption ofphosphate ions with Al(OH)3/Fe(OH)3 flocs,which are the predominant hydrolysis species inwastewater treatment practices. The best floccu-lation effect for Al3+ salts is generally experi-enced in the pH 6.0–8.0 range. The isoelectricpoint of Al in its hydroxide form is 7.4 pH, cor-responding to the point of minimum solubility[10,14,15,18].

The flocs resulting from Al salts are lessdense and slower to form than those from ironsalts. However, the advantage of Al compoundsis shown in a higher efficiency in the neutraliza-tion of surface charges and hence in coagulation/flocculation processes (e.g. removal of turbidity)[14]. With exception of aluminate — NaAlO2

(SAL), all common Fe and Al coagulants areacid salts and, therefore, their addition consumesalkalinity and lowers the pH of the treated water/wastewater. This is important because pH affectsboth particle surface charge and floc precipitationduring coagulation [10,15]. Since pH correctionto a neutral pH would incur major additionalchemical costs it is seldom practical to operate awastewater treatment process with pH control.

The present work was aimed to study thechemical precipitation of phosphate with SALas a pretreatment of RO feedwater during waste-water desalination. SAL coagulation was testedeither as simultaneous P removal during MBRtreatment or as post-removal of P from second-ary effluents form activated sludge process.Although SAL coagulation significantly helpedto maintain acceptable low flux decrease, feed-water acidification was still required to suppressflux decline during RO.

2. Materials and methods

2.1. Experimental setup and operational conditions

The experimental setup for effluents desali-nation applied for the research is described inFig. 1, and included primary settling of rawwastewater followed by simultaneous productionof tertiary effluents either by activated sludge(secondary treatment) plus microfiltration (MF)or by membrane bioreactor (MBR), and finallyRO. Chemical precipitation was alternativelytested as simultaneous P removal during MBRtreatment and as post-removal of P from acti-vated sludge effluents, as indicated (see red cir-cles in Fig. 1).

Raw wastewater was obtained from the fromthe Technion campus sewage collection system.Primary effluents were continuously produced bysedimentation of the raw wastewater in a stain-less steal sedimentation tank of 1 m3 designed tooperate at 1 h retention time. Secondary effluents(approx. 3 m3/day) were continuously producedin a sequential batch reactor (SBR) of 1.8 m3

capacity with computerized control, operated in5 cycles per day at an influent flowrate of 0.6 m3

Fig. 1. Outline of the research scheme for effluentsdesalination. AS: activated sludge; MBR: membranebioreactor; MF: microfiltration; Pe: primary effluents;PS: primary settling; RO: reverse osmosis; Se: second-ary effluents; UF: ultrafiltration. Circles denote thepoints where alternative coagulation was performed.


and aeration step of 2 h followed by 30 minanoxic stage, per cycle. MBR effluents (approx.0.65 m3/day) were produced in a ZW-10 unit(Zenon, Canada), equipped with a 0.2 m3 feedtank and 0.9 m2 UF-hollow fibers membranes ofnominal MWCO of 0.04 μm. Filtration flux wasapprox. 28 L/m2/h and a backpulse cycle of 45 swas conducted every 15 min. The biologicalreactors were operated in continuous regime andthe treated effluents exceeding the storagecapacity were returned to the municipal sewagesystem.

Microfiltration was done on a Memtek(USA) unit equipped with four-1 m2 polypropy-lene hollow fiber filters of 0.2 µm nominal poresize, assembled in parallel. The unit was operatedin crossflow mode at an initial transmembranepressure (TMP) of 0.5 bar. Reverse osmosis wasperformed on a unit comprising a 2.5" spiralwound module of 0.7 m2 (FilmTec TW30-2514,Dow). The system was run at a TMP of 10 barand operation mode was as follows: when anexperiment was started, half of the permeatewas discharged to obtain a twofold (50% recov-ery) or fourfold (75% recovery) feed concentra-tion, then batch operation (full recycling ofpermeate and concentrate) was maintained for24 h till a new batch was introduced. Tempera-ture was maintained constant at 25 ± 2°C in allthe units by means of a heat exchanger and awater chiller system.

Coagulation experiments were carried out atroom temperature in a six paddle stirrer-jartest apparatus (RAE Motors Corp.) in jars of2 L each, at the following conditions: mixing at200 rpm for 2 min, then addition of coagulantfollowed by mixing at 200 rpm for 5 min andthen at 25 rpm for 25 min, then settling for30 min and finally decantation.

Coagulation of SBR effluents for in pilotscale for RO experiments was performed in a150 L conical tank for decantation containing100 L effluents, in a similar time frame asdescribed above for the jar test experiments, with

30 mg/L SAL as coagulant. Low mixing wasperformed by gentle air sparging and fast mixingwas performed by means of a centrifugal pump.Following decantation, the supernant was furtherfiltered by MF for RO experiments. Simulta-neous coagulation in MBR was performed bycontinuous dosing (80–90 mL per day) of SALto maintain a nominal concentration of approx.18 mg/L in the mixed liquor. Both coagulated Seand MBR-effluents were stored in the dark atroom temperature during their use in RO experi-ments (1–3 days per run).

Unless otherwise indicated all coagula-tion experiments were performed without pHcorrection.

2.2. Analytical techniques

Conductivity and pH were measured usinga Cyborscan (Eutech Instruments). Turbidity wasmeasured with a 2100P turbidimeter (Hach).Total organic carbon (TOC) and total nitrogen(TN) were measured with an N/C 2100 multianalyzer (Analytik Jena). P, Ca, Mg and Nawere assayed by ICP-emission spectrometry(Optima 3000 DV, Perkin Elmer). Alkalinity,expressed as mg/L CaCO3, and water content insludge were conducted according to StandardMethods [19].

2.3. Chemicals

Ferric chloride — FeCl3 (60% w/w) was pur-chased from Spectrum Laboratory Products Inc.and sodium aluminate (12.5% w/v) was kindlyreceived from SWDP Ltd. (Haifa, Israel).

3. Results and discussion

3.1. Jar test experiments

Ferric chloride and SAL were first testedfor their ability for chemical clarification ofeither primary or secondary effluents in jar testexperiments. Evaluation of coagulants dosage


was tested in the range of 5–50 mg/L for SALand 40–100 mg/L for ferric chloride (Figs. 2and 3). Although P reduction was chosen as thecoagulation goal, removal of other parametersincluding turbidity, TOC, Ca, Mg and alkalinitywere tested. SAL displayed a considerable betterefficacy for Pe clarification than ferric chloride.Indeed, a concentration of approximately 25 mg/LSAL was sufficient for almost complete Premoval whereas ferric chloride did removeonly ~80% P, even at 100 mg/L dosage, and50 mg/L SAL reduced 91% turbidity in contrastto only 78% reduction achieved at 100 mg/L forferric chloride. Coagulation of Se with ferricchloride at pH 5.5 and 6.5 did not change signif-icantly phosphate removal (data not shown).Addition of SAL resulted in increase of pH from

7.5 to 8.8 which was paralleled by a increase ofalkalinity (24%), attributed to the basic nature ofSAL, and a decrease in Ca and Mg content (20%and 18%, respectively), indicating further pre-cipitation of the respective phosphates and car-bonates at the alkaline pH. In contrast, additionof ferric chloride resulted in almost no pHchange, as well as Ca and Mg content, due to aconsumption of alkalinity (19%).

As presented in Fig. 3, ferric chloride exhibitedbetter coagulation characteristics while treatingSe compared to that with Pe and a similar or evenslightly higher P reduction (97%) than SAL(95%), even if an higher dose was still required(100 vs. 50 mg/L, respectively). For secondaryeffluents, addition of SAL resulted again inincrease of pH and alkalinity (22%) with a

Fig. 2. Coagulant dose response profile of coagulation of primary effluents with sodium aluminate (top panels, fulllines) and ferric chloride (bottom panels, dashed lines) in jar test experiments. Bars indicate standard deviation.


concomitant decrease of Ca (25%) and Mg(15%) concentrations whereas coagulation withferric chloride was denoted by a consumption ofalkalinity (21%) paralleled by a decrease of pHwith almost no change in Ca an Mg levels.Whereas TOC remained almost unchanged dur-ing coagulation of primary effluents with bothcoagulants, coagulation of secondary effluentswas reflected by a reduction, being slight for SAL(~20%) and much more pronounced for ferricchloride (~60%), clearly denoting the differentnature of the colloidal organic matter in bothcases (see right panels in Figs. 2 and 3).

Kang et al. [20] also reported a higher effi-ciency for P removal from Se to levels below0.1 mg/L applying Al3+ salts compared to Fe3+

salts. Lopez-Ramírez et al. [21] achieved a final

turbidity of about 1–2 NTU during coagulationof Se, using 25 mg/L ferric chloride combinedwith 0.5–1 mg/L anionic charge flocculant(PASAFLOC FI-35) and lime at pH 10.5. Adinet al. [22] reported a reduction of membrane foul-ing permeate characteristic compared to nanofil-tration while practicing precoagulation with ferricchloride of secondary effluents before UF. AtpH 5.5, the optimal dose was between 150 and200 mg/L, whereas, at pH 7.8 the efficiencyincreases with the dose up to 300 mg/L. Sofferet al. [23] compared ferric chloride and alum forcoagulation of secondary effluents prior and inconjunction to UF. Precoagulation significantlyreduced membrane fouling and increased separa-tion performance, particularly at an optimumdose/pH condition. In ferric chloride coagulation

Fig. 3. Coagulant dose response profile of coagulation of secondary effluents with sodium aluminate (top panels, fulllines) and ferric chloride (bottom panels, dashed lines) in jar test experiments. Bars indicate standard deviation.


negative turbidity removal at the lower dosesincreased with decrease in pH. Alum removedturbidity and colloidal particles at a relativelynarrow pH range of 6–9, best at pH 6–7 whereasferric chloride performs well is larger being 3–10.

3.2. SAL coagulation as pretreatment for RO desalination

Based on the jar test results, sodium alumi-nate was further applied to test simultaneousMBR-coagulation of primary effluents and post-coagulation of secondary effluents followed byMF, as pretreatment alternatives to generate highquality tertiary effluents as feedwater for ROdesalination, in accordance to the wastewatertreatment scheme outlined in Fig. 1. Accordingly,SAL concentrations applied were 18 mg/L forsimultaneous coagulation (continuous operation)and 30 mg/L for post-coagulation (batch opera-tion). The results are presented in Tables 1 and 2.Both alternative treatment resulted in almostcomplete removal of P, with a residual concen-tration of <0.1 mg/L, and turbidity. In bothcases an increase in sodium concentration of

15–17% was noticed, in line with its contentin SAL.

The simultaneous MBR coagulation operatedwith a dose of 18 mg/L SAL resulted in almostno change of pH but 74% reduction of alkalinity,and a slight reduction of Ca and especially Mg.The high decrease of alkalinity (correspondingto 5.6 meq) can be attributed to a high CO2 strip-ping due the intensive aeration in the MBRcombined with carbonate precipitation (Table 1).

Post-coagulation of SBR effluents with aSAL concentration of 30 mg/L SAL, followedby MF filtration, resulted in one unit pH increaseand 28% reduction of alkalinity (correspondingto 1.7 meq) and 11% Ca removal (Table 2).Among coagulation and filtration, MF resultedmostly in a slight further decrease of TOC(approx. 15%) and turbidity. The decrease ofalkalinity may be due to and carbonate precipi-tation at the alkaline pH and CO2 stripping byair bubbling used for slow rate agitation in thecoagulation process.

Mass balance in terms of molar equivalentsshown in Table 1 reveals that for the case ofsimultaneous MBR-coagulation the reduction of

Table 1Characteristics of simultaneous coagulation-MBR treatment of primary effluentsa

aValues represent mean ± SD. bNaAlO2 titer ≈ 0.7 meq as Al3+. cAs P-PO4

3−.

Parameter pH Turbidity (NTU)

TOC (mg/L)

Na (mg/L)

Ca(mg/L)

Mg (mg/L)

P (mg/L)

Alkalinity (mg/L as CaCO3)

Primary effluent

7.6 ± 0.3 119.0 ± 14.1 65.4 ± 6.2 166.3 ± 7.8 62.6 ± 2.6 29.0 ± 1.6 9.7 ± 2.4 378.0 ± 9.2

MBR effluent 7.5 ± 0.2 1.3 ± 0.5 6.5 ± 1.1 195.8 ± 10.7 62.2 ± 4.2 26.9 ± 2.6 0.1 ± 0.1 97.1 ± 25.8

Coagulation reduction (%)

99 90 –17 1 7 99 74

Reduction (meq)b Ca Mg Pc Alkalinity

0.1 ± 0.3 0.2 ± 0.1 0.9 ± 0.2 5.6 ± 0.5


P and consumption of alkalinity surpass theexpected stoichiometry of phosphate precipita-tion with aluminium according to the followingprimary and secondary reactions:

Al3+ + H2PO4– → AlPO4 + 2H+ (1)

Al3+ + 3HCO3– → Al(OH)3 + 3CO2 (2)

both consuming alkalinity. This is interestingsince the stoichiometric molar ratio of Al:P of1:1 in AlPO4 is never achieved in practice, andthe actual ratio between Al dosed and P removedvaries between 2:1 and 3:1, especially at phos-phate lower than 10 mg P/L [10,14,15,18]. Forthe case of post-coagulation a ratio of Al:P of2.75:1, coinciding with the theoretical expecta-tions, whereas an Al:alkalinity ratio of 2:1 wasobtained, in spite of the increase of pH (seeTable 2). Hence, excess alkalinity removal isexplained by CO2 stripping due to intense aera-tion for both membrane operation and biologicaloxygen consumption in the MBR and spargingfor slow rage agitation during post-coagulationcombined with carbonate precipitation, as dis-cussed before.

The effect of phosphorus co-precipitationwith ferric chloride dosing on biological phos-phorus along with carbon and nitrogen removalwas investigated in a SBR for slaughterhousewastewater treatment, however a residual P con-centration of 3 mg/L was attained [24]. Chapmanet al. [25] studied preflocculation of secondaryeffluents with ferric chloride in a static floatingmedium flocculator to enhance MF perfor-mance. Coagulant concentration for P and sus-pended solids removal were optimum at 90 mg/Lwhereas for turbidity removal at 20–30 NTU.They achieved 83% removal of turbidity, 97%of phosphorus 45% of TOC. As a result of thetreatment, the permeate flux in MF was improvedby 70% and a 30–70% higher removal of organ-ics was observed.

Biological P removal in MBRs, and espe-cially enhanced biological phosphate removal isgetting increased interest, however this processis less robust than chemical P coagulation, andrequires additional anaerobic and anoxic com-partments [26]. The simultaneous coagulationwithin the MBR saved the necessity of a sepa-rate filtration step that was added to the SBReffluents coagulation before the feed to the RO

Table 2Characteristics of post-coagulation of secondary effluents followed by microfiltrationa

aValues represent mean ± SD. bNaAlO2 titer ≈ 1.1 meq (as Al3+). cAs P-PO4

3−.

Parameter pH Turbidity(NTU)

TOC(mg/L)

Na (mg/L)

Ca(mg/L)

Mg(mg/L)

P (mg/L)

Alkalinity (mg/L as CaCO3)

SBR effluent 7.7 ± 0.1 2.5 ± 0.2 7.8 ± 0.8 156 ± 4.4 58.4 ± 1.3 26.7 ± 0.7 4.3 ± 0.1 300 ± 12.5 Coagulation + MF 8.7 ± 0.0 1.8 ± 0.7 6.5 ± 0.0 180 ± 1.7 51.9 ± 6.0 25.9 ± 2.0 0.1 ± 0.1 215 ± 2.5 Total reduction

(coagulation + MF) (%)

29 17 −15 11 3 99 28

Reduction (meq)b Ca Mg Pc Alkalinity

0.3 ± 0.2 0.1 ± 0.1 0.4 ± 0.0 1.7 ± 0.2


system. The coagulation with SAL within theMBR did not result detrimental to the biologicalprocess, and in addition a satisfactory P removalaccompanied by alkalinity consumption at asuitable pH for biological activity was self-maintained. In addition, some improvement inthe quality of the mixed liquor suspended solids(MLSS) was observed. In fact, a reduction insludge volume index (SVI) from 180–250 mL/gwithout coagulation to 110–150 mL/g withcoagulation was obtained. The dewatering char-acteristics of the biosolids was improved as wellas a result of the coagulation, as reflected by theincrease of the free water content at expenses ofthe decrease of interstitial water content, similarto the proportions found in activate sludge bio-solids (Fig. 4). Nevertheless, the fixed solidscontent of the MLSS increased from 15–20% to35–40% as result of the coagulation, which maybe detrimental for the disposal and further utili-zation of the biosolids generated. Finally, thehydraulic performance of the MBR membraneswas not significantly altered by the simulta-neous coagulation in this study. Chemical coag-ulation for immersed UF membrane reactors, asfor UF filtration of secondary effluents, may beadvantageous, however, optimization of shear

stresses close to the immersed membrane tominimize a deposit set up on the membrane sur-face is still required [23,27].

3.3. RO desalination of tertiary effluents

Effluents desalination was studied in a batchRO system at 50% recovery. The impact of thepretreatment on RO performance was assessedby the change in relative permeability and massbalance of precipitating species obtained bychemical analysis of the feedwater, concentrateand permeate.

The RO performance comparing simulta-neous MBR-coagulation and post-coagulationof Se followed by MF is displayed in Fig. 5.MF-filtered Se without coagulation served ascontrol. Membrane permeability declined rapidlywhen no coagulation was performed reaching41% of its initial value after 48 h. Post-coagula-tion/MF of Se improved the performance of theRO separation and after 72 h of operation still55% of the initial permeability was maintained.

Fig. 4. Influence of simultaneous MBR-coagulationwith SAL (MBR + SAL) on sludge volume and waterholding capacity. MBR and SBR stand for controlswithout coagulation for membrane bioreactor and acti-vated sludge, respectively. Sludge volumes were inthe range of 0.1–0.5% and are not seen. Bar representstandard deviation.

Fig. 5. Influence of pretreatment on the relative ROpermeability during desalination of tertiary effluents.MBR + SAL: simultaneous MBR-SAL coagulation;Se + SAL/MF: post-coagulation of SBR effluents andMF; control: SBR effluents after MF, without coagulation.RO was performed with full recycle of permeate andconcentrate. Bars denote standard deviation.


The best performance was attained by the simul-taneous MBR-coagulation, in which after 72 hof operation still 90% of the initial permeabilitywas maintained, clearly showing the advantage ofthis pretreatment in removing P but also consid-erably reducing alkalinity. Average rejectionvalues were 92% of monovalent ions (Na, K),97% rejection of divalent ions (Ca, Mg), 100%for P and 93% TOC.

The difference in permeability decline ratecould be explained by mass balance calculationsmade from chemical analysis data of the feed-water, concentrate and permeate, suggestinginorganic scaling as the main cause for this decline(Fig. 6). Since the runs lasted only 1–3 days andTOC change was only 10–20%, permeabilitychange cannot be attributed to organic fouling orbiofouling. Nevertheless, organic fouling and thedevelopment of biofouling cannot be completelydiscarded as partially affecting the permeability,in spite of the short term of the experiments [28].As seen in Fig. 6, the runs without coagulation(control) displayed complete precipitation of allthe phosphate fed to the system, clearly indicat-ing that phosphate scaling, and in part carbonate,was the most probably cause for the severedecrease of permeability. For both coagulationtreatments, since phosphates were eliminated to

almost completion, the mass balance indicatesthat calcium carbonate is the most likely secondaryscaling specie, as evidenced by the change incalcium and alkalinity values, inline with previousreports [6,29].

The lower levels of Ca and alkalinity pre-cipitated during RO fed with the simultaneousMBR-coagulation effluents in comparison to thepost-coagulation ones are in line with the relativedecrease of permeability (compare Figs. 5 and 6).However, since in both cases Ca disappearanceequivalents exceed carbonate equivalents disap-pearance, other calcium salts seem to being pre-cipitated, such as sulfate which was not assayed.Calcium sulfate dihydrate (gypsum) has beenreported to crystallize over RO membrane duringdesalination of Colorado River water [30]. Cal-culation of Langelier saturation index (LSI) ofthe both RO feeds displayed that post-coagulatedSe, having higher LSI (1.3) and higher pH (8.5)than simultaneous MBR-coagulation (LSI = 0.75and pH = 8.0), may have a higher potentialfor carbonate precipitation. These values are incomplete agreement with the mass balance val-ues presented in Fig. 6 and in line with thedecrease of permeability presented in Fig. 5as well as with the coagulation data presentedin Tables 1 and 2, all showing the advantages

Fig. 6. Influence of pretreatment on the Ca, P and alkalinity mass balance during RO desalination of tertiary effluents.Left panel: equivalents missing to close mass balance; Right panel: percentage missing to close mass balance. Barsdenote standard deviation. For notations refer to Fig. 5.


of the simultaneous MBR-coagulation (75%reduction of alkalinity in addition to completephosphate removal).

To further reduce carbonate, and probablysulfate, precipitation, additional RO runs fedwith simultaneous MBR-coagulation effluentswere performed with pH adjustment to 6.5. Theresults are presented in Fig. 7. Under these cir-cumstances, the slight decrease of permeability(<8%) observed for the acidified simultaneouscoagulation-MBR effluent can be attributed toother fouling form than scaling. Moreover, thesefindings indicate the need for further acidifica-tion or high antiscalant dosing to reduce signifi-cantly reduce scaling during RO desalination ofsecondary effluents. A similar flux improve-ment for each one of the different pretreatmentswas achieved in RO experiments performedat 75% recovery, although a much more pro-nounced flux decline was evident for the controlwithout coagulation and acidification (data notshown).

4. Conclusions

Utilization of effluents for desalination intro-duce to membrane systems feedwater of reducedsalinity but higher concentration of dissolvedorganic matter and predominant inorganic ionsthat are less common in sea water. Although car-bonate scaling is present, phosphate scalingappears to be a main obstacle for effluent desali-nation, mainly due to the lack of appropriateantiscalant to handle its precipitation, especiallyat high recovery rates. This work was aimed tostudy the chemical coagulation of phosphate as apretreatment for RO of wastewater effluents.Based on preliminary jar test experiments, SALat concentrations of 20–30 mg/L was found as anefficient coagulant for phosphate and turbidityreduction. SAL coagulation was tested either assimultaneous phosphate removal during MBRtreatment or as post-removal of phosphate fromactivated sludge effluents followed by microfil-tration. Both alternative treatments resulted inalmost complete phosphate removal (≥99%).As well as being able to remove phosphate andreduce alkalinity, chemical coagulation is able toremove dissolved natural organic matter andother colloids, and especially humic substances,in the form of a colloidal metal–humate complex[10,31], which may increase membrane perfor-mance both during pretreatment and desalinationstages.

Desalination of both kinds of pretreatedeffluents showed also that calcium carbonatewas a second scaling agent following phosphateremoval. The simultaneous MBR-coagulation,which in addition to complete phosphate removaldisplayed a significant reduction of alkalinity(75%), produced higher quality effluents andreduced the effect of calcium carbonate scaling,as compared to the post-coagulation of sec-ondary effluents. Furthermore, the simultaneousMBR-coagulation process showed no detrimentaleffects on the MBR membrane performance oron the biological activity and avoided the need

Fig. 7. Effect of acidification to pH 6.5 on permeabilityduring RO desalination of tertiary effluents. SAL +acid: simultaneous MBR-SAL coagulation effluents,acidified; SAL: simultaneous MBR-SAL coagulation;control + acid: SBR effluents after MF, without coagu-lation and acidified; control: SBR effluents after MFwithout coagulation.


of an additional sedimentation or filtration stepprior to RO.

In conclusion, the MBR-chemical coagulationprocess seems to be a technically feasible pre-treatment of domestic effluents for further ROdesalination, generating soft effluents with lowcontent of organic matter and nutrients.

Acknowledgements

This work was funded by the Fund for Promo-tion of Research at the Technion and the Minis-tries of Agriculture and Science and Technologyof Israel. We are grateful to Y. David, A. Ronenand G. Shalem for their technical assistance.

References

[1] Inbar Committee, Upgraded Effluent Standards,Ministry of Environmental Protection, Israel,2004.

[2] H.C. Flemming, Reverse osmosis membrane bio-fouling, Exp. Therm. Fluid Sci., 14 (1997) 382–391.

[3] C. Faurie and S.F.E. Boerlage, Scaling and Partic-ulate Fouling in Membrane Filtration Systems,Taylor Francis, London, UK, 2001.

[4] I. Katz, M. Green, R. Semiat and C.G. Dosoretz,Treated effluents desalination and pretreatmentalternatives, Water Irrigation (in Hebrew), 502007 (in press).

[5] M.M. Tlili, A.S. Manzola and M. Ben Amor,Optimization of the preliminary treatment in adesalination plant by reverse osmosis, Desalination,156 (2003) 69–78.

[6] C.R. Bartels, M. Wilf, K. Andes and J. Iong,Design considerations for wastewater treatmentby reverse osmosis, Water Sci. Technol., 51(2005) 473–482.

[7] G. Greenberg, D. Hasson and R. Semiat, Limits ofreverse osmosis recovery imposed by calciumphosphate precipitation, Desalination, 183 (2005)789–804.

[8] M. Williams, R. Evangelista and Y. Cohen, Non-thermal process for recovering reverse osmosisconcentrate: process chemistry and kinetics,Water Qual. Technology Conference, AmericanWater Works Association, Seattle, WA, 2002.

[9] S.F.E. Boerlage, M.D. Kennedy, I. Bremere,G.J. Witkamp, J.P. van der Hoek and J.C. Schippers,Stable barium sulphate supersaturation in reverseosmosis, J. Membr. Sci., 179 (2000) 53–68.

[10] J.-Q. Jiang and N.J.D. Graham, Pre-polymerisedinorganic coagulants and phosphorus removal bycoagulation — a review, Water SA, 24 (1998)237–244.

[11] G.K. Morse, S.W. Brett, J.A. Guy and J.N. Lester,Phosphorus removal and recovery technologies,Sci. Tot. Environ., 212 (1998) 69–81.

[12] P.R. Thomas, D. Allen and D.L. McGregor,Evaluation of combined chemical and biologicalnutrient removal, Water Sci. Technol., 34 (1996)285–292.

[13] T. Clark, T. Stephenson and P.A. Pearce, Phos-phorus removal by chemical precipitation in abiological aerated filter, Water Res., 31 (1997)2557–2563.

[14] D.W. de Haas, M.C. Wentzel and G.A. Ekama,The use of simultaneous chemical precipitation inmodified activated sludge systems exhibiting bio-logical excess phosphate removal. Part 1: Litera-ture review, Water SA, 26 (2000) 439–452.

[15] M. Henze, P. Harremoes, J. Cour Jansen andE. Arvin, Wastewater Treatment: Biological andChemical Processes, 3rd edn., Springer, Berlin,DE, 2002.

[16] J.M. Ebeling, S.R. Ogden, P.L. Sibrell andK.L. Rishel, Application of chemical coagula-tion aids for the removal of suspended solids(TSS) and phosphorus from the microscreeneffluent discharge of an intensive recirculatingaquaculture system, N. Am. J. Aquacul., 66 (2004)198–207.

[17] E. Tykesson, L.-E. Jönsson and J. Cour JansenExperience from 10 years of full-scale operationwith enhanced biological phosphorus removal atÖresundsverket. Water Sci. Technol., 52 (2005)151–159.

[18] M. Maurer and M. Boller, Modelling of phospho-rus precipitation in wastewater treatment plantswith enhanced biological phosphorus removal.Water Sci. Technol., 39 (1999) 147–163.

[19] A.D. Eaton, L.S. Clesceri and A.E. Greenberg(Eds.), Standard Methods for the Examination ofWater and Wastewater, 19th edn., American Pub-lic Health Association, Washington, DC, USA,1995.


[20] S.J. Kang, K. Hoversten and D.E. Lund, Thehighest level of phosphorus removal practicablefrom municipal wastewater treatment plants,WEFTEC Conference Proceedings, 2001.

[21] J.A. Lopez-Ramírez, S. Sahuquillo, D. Salesa andJ.M. Quiroga, Pre-treatment optimisation studiesfor secondary effluent reclamation with reverseosmosis, Water Res., 37 (2003) 1177–1184.

[22] A. Adin, Y. Soffer and R. Ben Aim, Effluent pre-treatment by iron coagulation applying variousdose–pH combinations for optimum particle sepa-ration, Water Sci. Technol., 38 (1998) 27–34.

[23] Y. Soffer, R. Ben Aim and A. Adin, Membranefouling and selectivity mechanisms in effluentultrafiltration coupled with flocculation, WaterSci. Technol., 51 (2005) 123–134.

[24] Y. Filali-Meknassi, M. Auriol, R.D. Tyagi,Y. Comeau and R.Y. Surampalli, Phosphorusco-precipitation in the biological treatment ofslaughterhouse wastewater in a sequencing batchreactor, Pract. Periodical Haz., Toxic RadioactiveWaste Mgmt., 9 (2005) 179–192.

[25] H. Chapman, S. Vigneswaran, H.H. Ngo, S. Dyerband R. Ben Aim, Pre-flocculation of secondarytreated wastewater in enhancing the performance ofmicrofiltration, Desalination, 146 (2002) 367–372.

[26] M. Kraume, U. Bracklow, M. Vocks andA. Drews, Nutrients removal in MBRs for munic-ipal wastewater treatment, Water Sci. Technol.,51 (2005) 391–402.

[27] P. Choksuchart, M. Heran and A. Grasmick,Ultrafiltration enhanced by coagulation in animmersed membrane system, Desalination, 145(2002) 265–272.

[28] H. Ivnitsky, I. Katz, D. Minz, E. Shimoni,Y. Chen, J. Tarchitzky, R. Semiat and C.G.Dosoretz, Characterization of membrane bio-fouling in nanofiltration processes of wastewatertreatment, Desalination, 185 (2005) 255–268.

[29] J. MacAdam and A.S. Parsons, Calcium carbonatescale formation and control, Rev. Environ. Sci.Bio/Technol., 3 (2004) 159–169.

[30] W.Y. Shih, A. Rahardianto, R.W. Lee andY. Cohen, Morphometric characterization of cal-cium sulfate dihydrate (gypsum) scale on reverseosmosis membranes, J. Membr. Sci., 252 (2005)253–263.

[31] R. Storhang and B. Rusten, Upgrading a primarytreatment plant for nutrient removal, in: H.H. Hahnand R. Klute (Eds.), Chemical Water and WasteWater Treatment, Springer-Verlag, Berlin, 1990,pp. 461–470.

Documents

Desalination of domestic wastewater effluents: phosphate ...gwri-ic.technion.ac.il/pdf/gwri_abstracts/2008/62.pdfBatch desalination of both kinds of pretreated effluents showed also