Wash water in waterworks: contaminants and process options for reclamation

Available online at www. sciencedirect.comJO。JOURNALOF,SCIENCESScienceDirectJES夕LELEJoural of Enviroumental Sciences 202008) 1300-1305ww.jesc.ac.cnWash water in waterworks: contaminants and process options for reclamationC B Chidambara Rajl", Tan Ee Kwong', Wong Wai Cheng2, Lee Mun Fong2,Soh Hoo Tiong, Paul Stefan Klose'1. Centre for Advanced Water Technology, Singapore Uilities Intermational Pte Ltd, 82 Toh Guan Road East#C4-03,Singapore 608575. E-mail: cbcraj@ cawt.suicom.sg2. Public Utilities Board, Water Supply (Plants) Department, 40 Scots Road #14-00, Singapore 228231Received 15 October 2007; revised 11 March 2008; accped 5July 2008AbstractReclamation of clean water from filter backwash water was studied through pilot scale experiments. The pilot plant consistedof clarification, sand- fltration, and ultrafiltration modules in sequence, with a provision to bypass the sand filter. Clean water thatconformed to World Health Organization (WHO) guidelines on Potable Quality was reclaimed. Turbidity, aluminum and iron werefound to be critical contaminants in process selection and design. Clarification, fllowed by sand fitration, was found to be the minimumrequirement for recycling filter backwash. However, membrane filtration would enhance reclaimed water quality as the membrane actsas an additional barrier against Giardia and Cryptosporidium.Key words: filter backwash; clarification; water treatment; aluminium; GiardiaIntroductionterm experiments extended for about 20 h. A bench-scalestudy with MF membranes was reported by Taylor et al.Potable water is typically produced by clarification of (2000). Samples from five water treatment plants in thesurface water. Often, alum is used as the primary coagulant United States were tested for removal of turbidity, totalfor clarifcation, which results in the production of“"alum suspended solid, color, particle count, UV absorbance andsludge"; this is also called as“clarifier sludge". Aftermicrobial counts, which concluded membrane filtration asclarification, the water is fitered through sand beds. The an economically-feasible recycling technique. Furrey et al.sand beds are backwashed after several hours of service, (2000) reported on a pilot trial of reclamation from filterwhich results in filter backwash water. Alum sludge and backwash and liquid supematant from the holding basinfilter backwash are waste streams produced during water of a waterworks in New Jersey, USA using pall cross-flowtreatment. Typically, in a waterworks, the flow rate of membranes. The performance of the process was said tothese two streams is of the order of several hundred cubic be consistent and the product water was of potable quality.meters per day. In an effort to reduce the volume 0Noticeably, most of the above reports focused on a singlethese waste streams and to conserve as much water as purification process, a membrane filtration.possible, a pilot-scale study was undertaken to identifyArora et al. (2001) focused on the presence of protozoa,the major contaminants and to investigate the feasbility such as, Cryptosporidium and Giardia, in filter backwash,of reclamation of clean water.examined the use of backwashable depth flter technologyCertain cases of recycling filter backwash water and in the place of conventional filtration and the use of poly-recycling filtrate from clarifier sludge were reported. For mers resulted in excellent removal of turbidity, particlesexample, Thompson et al. (1995) reported on pilot testing and microorganisms. In a series of articles, Edzwald etof micofilration (MF) membranes for filter backwash al. (2003) and Tobiason et al. (2003a, 2003b) noted thattreatment, where turbidity of 500 NTU were reduced to the quality of water recycled from filter backwash wasless than 5 NTU. In another trial, a mixture of filter back- temporally variable for plants with only fow equalizationwash and clarfer sludge was processed by MF to produce (without solids removal), exhibiting significant peaks inwater of turbidity 0.1 NTU. In a laboratory-scale study, solid levels that led to a short-term increases in influentVigneswaran et al. (1996) treated filter backwash froma turbidf two diferent clar-water treatment plant in Thailand using cross-fow micro- ificati中国煤化工and neere palesedimfiltration ceramic tubular mermbranes. Turbidity, bacteria, sedim:Y片C N M H Gn with dual-mediaand aluminium were removed significantly and their long- (anthracite/sand) filtration. Further, full-scale backwashrecycling practces at six treatment plants in the USA were* Corresponding author. E-mail: cbcraj@cawt.sui.com.sgNo.11Wash water in waterworks: contaminants and process options for reclamation1301also reported.1.2 Filter backwash waterWhile the aforementioned studies highlight some ofThe waterworks, where this pilot study was conducted,the issues associated with recycling filter backwash, itis recognized that a proper design of recycling wouldemployed conventional coagulation-filtration process toinvolve site-specific details on water contaminants andproduce potable water from surface water. There were 12reclaiming water that is of aceptable quality. A treatmentclarifiers and 16 sand filters in the waterworks. Alum wasplant operator is primarily interested in a process thatused as the primary coagulant. Filters were backwashed af-could withstand the variations in loading while producingter approximately 25- -28 h of service. Sedimentation tanksclean water of consistent quality which would conformwere emptied after 10-12 d of service. These two wasteto the Guidelines for Drinking Water established by thestreams entered the sludge pond, from which the effluentWorld Health Organization (WHO, 2006). Consequently,was discharged to the public sewer. During backwashing athe primary goal of this research was to understand thefilter, typical flow rate of water was about 45- -50 m/minlevel of contaminants in the waste stream and to investigateand about 150 -200 m' of wastewater was generated perthe feasibility of reclamation of back wash water throughfiter. Filter backwash water was the largest fraction ofpilot-scale trials. Most studies on filter backwash werewastewater and it accounted for about 2% of the raw waterreported from the United States. The nature and constraintsentering the waterworks.associated with filter backwash were largely unknown in1.3 Analytical methodsSingapore and also, in this part of Asia, prior to this study.For water quality analysis, methods outined by APHASingapore has equatrial climate with annual rainfall(1998) and USEPA were performed. Color was analyzedclose to 2,800 mm. The city-state is highly urbanizedby APHA 2120B Visual Comparison Method. Silica wasand the central region of the island is designated as aanalyzed by APHA 4500-SiO2 F Flow Injection Analysis.“protected catchment", which means the runoff is freeTOC was analyzed by APHA 5310B High-Temperatureof industrial pollutants. Agricultural activity is limited inCombustion Method. Iron, manganese, copper, and alu-Singapore to certain peripheral areas. Hence, the use ofminum were analyzed by EPA Method 6010B withorganic pesticides and impact of nutrients derived (N andPlasma-Atomic Emission Spectrometry (Optima 5300DV,P) from fertilizers are minimal. Furthermore, the city'sPerkin Elmer, USA). Fluoride, nitrate, chloride, sulphate,sewer system reaches hundred percentage of the public,and phosphate were analyzed by EPA Method 300.0consequently, untreated sewage does not enter pristine sur-Ion Chromatography (DX120, Dionex Corp, USA). Totalface water sources. In addition, treated wastewater does notColiform Count was done by membrane filter techniqueflow into reservoirs, which are drawn for the production(APHA 9222B). Giardia and Cryptosporidium were quan-of potable water. While acknowledging that Giardia andtifed by Real-Time PCR (SmartCycler II, Cepheid, USA)Cryptosporidium levels impact the treatment strategies forwith two sets of primers (β-Giardin P241f and β-Giardinrecycle (Arora et al, 2001; Nieminski and Ongerth, 1995)P241r) for Giardia and two sets (COWP P702f and COWPand that mammals in the watershed often serve as sourcesP 702r) for Cryptosporidium (Guy et al., 2003). Theof such protozoa (Rose, 1988), very ltte information wasprimers were purchased from 1st Base Pte. Ltd, Singa-available on the presence of such parasites, since Singaporepore.has lttle or no operations relating to animal husbandry.Therefore, this study was undertaken to assess the con-taminants in backwash water in the urban environment 2 Resultsand treatment strategy for recycle on other parameters,such as, turbidity, color, total dissolved solids (TDS), total2.1 Preliminary assessmentorganic carbon (TOC), nitrate, fuoride, chloride, sulphate,A list of about 17 parameters, as shown in Table 1, wasphosphate, metals, and coliforms.analyzed during the first three weeks of the study. The1 Materials and methodsobjective was then to select and to focus on the parametersthat would critically influence the treatment strategy andthe ultimate water quality.1.1 Pilot plantA custom-built pilot plant with a capacity of processing2.1.1 Color, iron, and manganese70 m/d of raw water was used in this study. It consisted ofColor of the raw water was in the range of 5- -25modules for clarification, sand-iltration, and ultrafiltration Hazen and it could be reduced to < 5 Hazen in productin sequence. Raw (backwash) water was fed to the lamella water by UF permeate. Since color of the waste streamsclarifier, which had about 20 inclined plates. Clarified fuctuated from time to time, continual monitoring waswater was then filtered through a sand bed (about 400 mm considered to be necessary in the long-term. The ironin diameter and 500 mm in height) and the ultrafiltrationoncemg/L during thesystem subsequently. The UF membrane modules were first中国煤化工: 0.01 mg/L duringpurchased from Koch Membrane Systems Inc, USA. The subsedVHO guideline, ironUF permeate was dosed with sodium hypochlorite for conceYHCNM H Guid led to an issedisinfection. After passing through the sand filter the on color in potable water and hence, it is considered to beclarified water could be fed directly to the UF system.monitor. Presence of manganese could also lead to issues1302C B Chidambara Raj etal.Vol. 20Table 1 Preliminary ascesment of water qualityWeck2Week3WHO guidelineWeck 1.ParumeterRawwater UF permeata Rawwater UPpermeate Rawaer P pemeatse<15Color (Hazen)22≤50.2Turtidity NTU)1.0696.75.9Neutral7.6.9591.73991.243Conductivity (uS/cm)721201445546< 10000TDS(mgL);12002.582.22TOC (mg/L)330390.220.310.270.28Nitrate (mg/L)0050.070.180.130.1923.78.899.311.331.0Chloride (mg/L)14020.712.63.213.5< 2500.08< 0.08<0.08Phosphate (mg/L)Silica (mg/L)0.68.790.770.750.79lIron (mg/L)0850.04< 0.0030.010.004<03Manganese (mg/L)0.013<0.003<0.5< 0.0020.006.01Copper (mg/L)11.90.703.270.090.23<0.2Aluminium (mg/L)CFU/100ml > 1,0001> 1,000Total Coliforms,chlonde, conductivity and TDS after treatment. There were naturalSodium hpclorole ws dosed for dsnfncien which cobmr ine w ; Wonr H ea oqxg) ato; TDS: wotal dsved solid; ToC: wotal(minon) fucuaiois in the level of sulphate and fuoide from time 00 time. WHO: Word Healhl ororganic carbon.on color. Manganese in raw water was less than 0.05 mg/L2.2 Monitoring programduring the first three: weeks and it reduced as the water wasThe performance of the pilot plant was monitored forpocessed throughclarfication and uF. This was one orderabout twenty weeks to gain a better understanding of theof magnitude lower than the guideline value (0.5 mg/)forlevel of contaminants, rejection of the contaminants atpotable quality. Hence, manganese was not a contaminantvarious stages in the reclamation process, the implcationsof significance in this location.of the contaminants on the treatment strategy and the2.1.2 Nitrate, phosphate and fuoridequality of reclaimed water.Nitrate was normally lower than 1 mg/L in raw water 2.2.1 Turbidityand was lower than the guideline (50 mg/L) of WHO.Turbidity of raw water vared between 1 and 300 NTUConsequently, nitrate was not expected to pose a challengeas shown in Fig.1. Rapid fuctuations in turbidity wereto water quality here.'Pluoride (0.5 mg/L) was also lowerobserved whenever backwashing was initiated for a filter.than WHO guideline of 1.5 mg/L. Phosphate was lowerMost of the turbidity was removed at the lamella carfier,that detection limit. Similarly, the very low levels of chlo- where the efuent was generally below 5 NTU (WHOride, sulphate, silica and copper lead these contaminantsguideline). Further, upon fitration, turbidity reduced towere not of signifcance. Usually the presence of nitratei NTU consistetly. The UF system further reduced theand phosphate in surface/poable waters were ascribed toturbidity below 1 NTU. The bulk of turbidity was removednon-point source pollution from agricultural runof. Sinceat the clanfer, particulate load on the filter and UF wereagricultural activity was quite limited in the central part of minimal, This suggested the possibility that either one ofSingapore and the raw water was primarily from the centralthe steps-sand fiter or UF module could besidered tolecintheregion, N and P levels were found to be negligible in thebe redundant and be eliminated in fll-scalc design. Thiscurrent context of recyeling.aspect was explored by passing through the sand fiter forfour weeks of experimental trials.2.1.3 AluminumAluminum was high in raw water, although it decreasedwhile after passing throughinothroughclarification and UF, it was日Raw water-古- Clarifed waterx Filtered waterUF permeatestill high enough that close monitoring was warranted. The100.00 rpresence of aluminum was primarily atributed to the useWHO gileieof alum as the coagulant.100.002.1.4 Total Coliform Count10.00 .rotal Coliform Count was of the order of severalthousands (CFU/100 m) in raw water. The product wa-ter, which was dosed with sodium hypochlorite (4 mg/Ldosage) as the disinfectant, showeda countof 1 or, <中国煤化工十20 241 CFU/100 ml consistently.Therefore, it was concludedthat dosing of sodium hypochlorite was adequate to reduceMHCNMHGock)Coliform Count,Fe 1 Turbidity removal t various stages.Wash water in waterworks: contaminants and process options for reclamation13032.2.2 Color ;reduced to the range of 0.003- -0.02 mg/L in UF permeate.Color of raw water fuctuated in the range 15- -500 HazenManganese concentration in raw water did not exceed(Fig.2). Obviously, most of the color was removed at the0.5 mg/L. Coagulation Sedimentation, followed by sand-clarification-stage. Clarifier efuent was consistently offiltration, removed about 70% of the metal so that the level10 Hazen or less, which conformed to WHO guidelines.of manganese in fitered water did not exceed 0.1 mg/L.Further, this set of evidence led to the conclusion that mostReduction below this level was infrequent and UF perme-of the color-causing contaminants were of in particulateate reached about 0.02 mg/L in certain weeks. The datapattem, which settled in the clarifier as flocs. The focsof manganese indicated that coagulation-sedimentation-were often pale brown in color. In comparison to thesefltration were adequate to meet WHO guideline (0.5values, the color of filter backwash exceeded 1,000 Hazenmg/L) for manganese. Filtered water and UF permeate hadin a surface water treatment plant in New Jersey (Furreyalmost the same level of manganese, suggesting that theet al, 2000) and often exceeded 15 Hazen even afterinfluence of UF on the removal of manganese was minor.filtration. Color was usually related to the presence of ironCopper was lower than 0.05 mg/L in raw water aland manganese in raw water (Cleasby, 1975; O'Connor,through the monitoring period (WHO guideline: 2 mg/L)1971). Therefore, it was imperative to examine thoseand hence, it did not pose a threat to the quality ofparameters in conjunction with color.reclaimed water. Almost always, copper was reduced to0.002 mg/L in filtered water, After a filtration, copper2.2.3 Metal concentrationstayed at about 0.002 mg/L in UF permeate suggestingSince iron in raw water was in the range of 0.05 9.0that the fraction of copper removed by UF was negligible.mg/L (Fig.3) exceeding the WHO guideline 0.3 mg/L,Thus, coagulation-sedimentation-filration was adequateremoval of iron was important. lron decreased to the rangefor removing most of the copper.of 0.05- 0.37 mg/L in clarified water, suggesting that aAluminium was high in filter backwash (1 -80 mg/L)portion of the iron was precipitated in the clarifier. Further,ls shown in Figs.3 and 4. During the first four weeks,it reduced to the range of 0.003- -0.05 mg/L in filteredreduction in aluminium was not signifcant and it exceed-water, indicating that certain precipitates of iron could beed WHO guideline (0.2 mg/L). Interestingly, aluminiumsmall, and that escaped the clarifer were caught in thewas not removed by clarification, filtration, and UF, thusfiter. Coagulation-sedimentation, followed by filtration,indicating that aluminium was present in soluble formwas adequate to remove iron well. Furthermore, iron waswhich could passed through the UF membrane. Duringthis period, pH of clarification was not controlled. Startingthe 4th week, pH of clarification was adjusted to 5.9-x Filteredwater- UF permcate6.0, which was the level of lowest solubility of aluminium1000 rin water (Amirtharajah and Mills 1982). After a while,WHO guidelinealuminium in filtered water decreased to 0.019 mg/L.15 Hazcn100Beyond a filtration, aluminium removal was negligibleand its concentration in UF permeate was often close to0.019 mg/L. Raw water fed to the pilot facility (ilter号1backwash) was close to pH 7.0. Hence, acid dosing wasnecessary to effect the removal of aluminium. On anaverage, consumption of acid (sulphuric) was 0.1 mg/L.2024The data on aluminium confirmed that the coagulation-Monitoring time (week)flocculation efect and the solubility of aluminium in waterFg2 Colour removal.around pH 5.9 -6.0 played an important role in its removalat the clarifier. Therefore, pH-control at the clarifier was100.000the proper strategy for aluminium removal.Raw10.000 -water| Clarifed。Raw watera Clarified waterwaterX Filtered water0 UF permeate1.000Clarified100.00 n。。。o°0。Filtered营10.0000.100water.IF|water permcaRav1.00 t0.010Fitered UFwater permeateX四a 8r8008088 000oFA1中国煤化工_↓WHaudeieFig3_ Removal of iron and aluminium. Horizontal lines representWHO Guidelines for each metal. The bars represent 10% -90% range ofHCN M H Gwek)concentration and the sticks (above and below) indicate the maximum andthe minimum in the data.Fig.4 Removal of aluminium and signifcance of pH.1304C B Chidambara Raj et al.Vol. 202.2.4 TOCpores. Such fouling was not observed during the pilotTOC in raw water was 2 -8 mg/L. However, TOC ofstudy, which suggested that aluminium flocs were removedfiltered water was mostly in the narrow range of 1-3 at the sand filter and thus, the tendency for membranemg/L, indicating that the balance of organic content was infouling was minimal. Data in Figs.3 and 4 confirmed thatparticulate form and was removed by the carifir and fiter. aluminium conent of ftered water was ully 0.02 mg/LWhile WHO did not have a guideline on TOC, USEPAand the maximum was 0.1 mg/L, while the feed couldhad recommended enhanced-coagulation if TOC in treatedhave aluminium as high as 60 mg/L. Therefore, most ofwater were to exeed 2 mg/L. In the present case, the the aluminium was rermoved at the caifier and at thesurface water that reached the waterworks had TOCinthe sand filter. Due to those reason, it could be concludedrange of 1-2 mg/L. Further, if rcycling of fiter backwash that sand filter acted as a major brier against cllidalwere implemented on fl scale, it would contribute tonly aluminium and membrane fouling did not occur in this1% of the flow through the waterworks. Consequenly, study. Moreover, if membrane flration were to be adapedthe recycled water would not significantly alter the TOCas a part of the reclamation process, it would be good tolevel when mixed with the total flow. Hence, removal of keep the sand fler in the front to minimize membraneTOC from filter backwash water was not important at thisfouling.location.3.3 Occurrence of protozoa3 DiscussionFilter Backwash Recycling Rule (FBRR), publishedby the United States Environmental Protection Agency3.1 Bypassing the filter(USEPA, 2002), was established to regulate managementof the recycle streams in water treatment plants. TheIt was observed that filter backwash water could form objective of FBRR was to ensure adequate level of publicgood flocs without the addition of any coagulant and health protection by minimizing the risk associated withthe focs settled readily leaving clear water on the top. Cryptosporidium and Giardia in recycle flows. FBRR at-This indicated the possibility that more than 99% of filtertempeted to regulate three recycle streams in conventionalbackwash water could be reclaimed by simple flocculation treatment plants, viz, Filter backwash water, Thicken-and sedimentation. Clarifed water was largly clear and er supermatant, and Liquids fron dewatering processes.the clarity of the water enhanced on passing through thePathogenic protozoa, such as, Cryptosporidium and Gi-sand filter and membrane filer. This led to that either ardia, were known to present at elevated levels in theseone of the filters-sand filter or membrane filter could berecycle streams and these were not easily inactivateddispensed with.by commonly used disinfectants, such as chlorine andDuring the last four weeks of the study (since 20th chloramines. Sedimentaion and fltration were knownweek), the sand filter was by-passed and clarified water (USEPA, 2002) to be the main barriers for the removalwas directly pumped to the ultrafiltration system. This wasof these pathogens. USEPA methods 1622 and 1623 weredone to investigate the eficiency of such a process configu-employed for detection for past several years. Giardia andration. The relevant data points were included in Figs.1,2, Cryptosporidium were not detected in samples of freshand 4. Under this scenario, as there was no“filtered water",water entering the waterworks and also, in the finishedthe points (x) are absent. Product water (UF permeate)water exiting the waterworks. Incidentally, the raw waterconformed to WHO guidelines. This confirmed that thefor the waterworks came from a protected catchmentsand filter could be dispensed with. Clarification followedand hence, the threat associated with these protozoa wasby membrane filtration was adequate for reclaiming filterconsidered to be minor.backwash.Since the beginning of this pilot study, data on GiardiaIn addition, it could be asessed whether the membrane and Cryptosporidium were actively sought and there was afilter could be dispensed with. Under such a scenario, renewed atempt to employ methods that were more sensi-“clarified water" and“fitered water" would be present intive than USEPA methods 1622 and 1623. Consequently,Figs.1, 2, and 4 and the data points (rectangles) on“UFthe technique based on real-time PCR was chosen (Guypermeate" would be absent.“Filtered water" conformedet al, 2003). Again, Giardia and Cryptosporidium wereto the guidelines of WHO. Therefore, sand filter couldnot detected in the surface water entering the waterworksbe retained and membrane filter could be dispensed with. .and in the finished water leaving the waterworks. WithinThese heuristically indicate that either one of the filters was the waterworks, filter backwash and clarifier sludge wereadequate for reclaiming filter backwash.discharged into a Sludge Pond. In other words, Sludge3.2 Membrane foulingPond was the place where the cysts could be present inconcentrated form. Three samples (originating on diferentBecause the backwash water was high in aluminium dates) from the Sludge Pond were then analyzed. Giardiacontent and aluminium remained largely in colloidal formlambli中国煤化工-3-4 cysts/L in allduring the reclamation process, it could be likely thatthe thrn parvum was notsuch colloids could lead to fouling of membrane surfaces.detectefY片CNMHGescence of Cryp-Specifically, flocs formed from alum were sticky in naturetosporidium wal protein gene did not increase throughand thus, were likely to be trapped within the mermbraneforty cycles of amplification.No.11Wash water in waterworks: contaminants and process opions for reclamation1305Interestingly, based on the total flow through the wa- steps for water reclamation in this case. Utrailtration wasterworks, only about 1% was discharged into the Sludge not a key requirement. However, UF was a better barrierPond, which amounted to 10 L/m3 of fresh water en- against protozoa.tering the waterworks. Even within the Sludge Pond,Acknowledgmentswhile sampling, we ignored the clear supematant (whichwas about 70%) standing at the top and sarnpled theThis project was funded by the Public Utilities Board“concentrated sludge" at the bottom (which made up of (PUB), Singapore. The content and conclusions are the30%). Consequently, the detected level of 4 cyst/L wouldviews of the authors and do not necessarily reflect theapproximately correspond to 10 cystm' of fresh water views and policies of PUB. Mention of trade names oror 0.01 cystL, which was an extremely small number.commercial names does not constitute endorsement orFuther, Ongerth (1989) noted that the recovery eficiency recommendation for use.of membrane filtration averaged about(21.8土6)% in theanalytical protocol. Hence, out of the 0.01 cyst/L, the Referencesactual number that could possibly be recovered would be0.0025 cyst/L or less, which is a significant tiny number(