A pilot-scale study was conducted to evaluate the impact of several biofiltration enhancement strategies in terms of organic removal to reduce disinfection by-product (DBP) formation potential and mitigate ultrafiltration (UF) fouling. Strategies included nutrient addition (nitrogen and phosphorus) to optimize metabolic degradation of organics, use of hydrogen peroxide (H2O2, peroxide) to improve filter run times, and the application of in-line aluminum sulphate (alum) for biopolymer removal. The impact of media type on performance was also examined (anthracite versus granular activated carbon (GAC)). Passive biofiltration (without enhancement) reduced dissolved organic carbon (∼5%), biopolymers (∼20%), and trihalomethane and haloacetic acid precursors (∼20% and ∼12%, respectively) while mitigating UF irreversible fouling (∼60%). Nutrient addition was not observed to enhance biological performance. Addition of 0.5 mg/L hydrogen peroxide decreased head loss by up to 45% without affecting organic removal; however at a dosage of 1 mg/L, it negatively impacted both UF fouling and DBP precursor removal. In-line alum addition prior to biofiltration (<0.5 mg/L) improved UF fouling control by up to 40%, without sacrificing head loss. Overall, GAC provided superior performance when compared to anthracite.
Natural organic matter (NOM) found in drinking water sources represents a complex mixture of organic compounds produced by aquatic or vegetational processes (Matilainen et al. 2011). NOM itself does not present a direct health concern; however, it can impact drinking water quality by reacting with chlorine and other disinfectants to form disinfection by-products (DBPs) (Richardson et al. 2007). NOM can also hinder treatment processes, for example foul membranes and increase chlorine demand (Howe & Clark 2002). Liquid chromatography-organic carbon detection (LC-OCD) may be used to characterize NOM into five fractions including biopolymers such as proteins and polysaccharides and humic substances (HS) as well as other smaller organics (Huber et al. 2011). In particular, biopolymers have been identified as the major foulant for ultrafiltration (UF) membranes (Hallé et al. 2009; Neubrand et al. 2010).
When granular filters are operated without exposure to chemical disinfection, such as chlorinated backwash, microbial communities can colonize the media and consume organic carbon (Zhu et al. 2010). Biologically active filtration known as biofiltration represents a cost-effective, chemical-free treatment process for organic carbon removal (Huck & Sozański 2008). In terms of water quality improvement, biofiltration has been shown to biodegrade DBP precursors (Chaiket et al. 2002) and taste and odor compounds (Elhadi et al. 2006). In addition, the removal of biodegradable carbon can limit microbial growth in distribution systems (Van der Kooij 1992).
Recent studies aimed at enhancing biological performance have supplemented key metabolic nutrients to achieve an optimal ratio between carbon, nitrogen and phosphorus. In one study, adjustment of the carbon to phosphorus ratio (C:P) to 100:2 was observed to increase dissolved organic carbon (DOC) removal by 75% and decrease head loss by 15% (Lauderdale et al. 2012). The addition of hydrogen peroxide (H2O2) was also shown to be effective for improving operational performance by decreasing head loss by up to 75% without negatively impacting DOC removal (Urfer & Huck 2000; Lauderdale et al. 2012). The impact of nutrient enhancement and peroxide addition prior to biological filtration on specific NOM fraction and DBP precursor removal has not been examined in detail. Studies reported in the literature suggested that using exhausted granular activated carbon (GAC) versus anthracite as biofilter media provides up to 10% additional organic carbon removal at temperatures <12 °C (Dussert & Tramposch 1997; Emelko et al. 2006). Due to its high surface area which can be >1,000 m2/g and porous nature with a pore volume >0.15 cm3/g, GAC could favor particle attachment and microbial growth (Dussert & Tramposch 1997). Others have attributed the superior performance of GAC to bioregeneration, whereby microorganisms consume adsorbed organic matter and regenerate adsorption sites (Zhu et al. 2010; Velten et al. 2011a).
UF has emerged as a cost-effective treatment process to achieve water quality regulations, while reducing the footprint of a treatment facility (Singh 2006; Rana et al. 2012). Fouling mitigation is an important issue when implementing UF membranes, since this can greatly impact treatment efficiency and reduce the maintenance and operational costs (Singh 2006). Introduction of coagulation has been shown to reduce membrane fouling (Crittenden et al. 2012). Wray et al. (2014) reported effective biopolymer removal using in-line alum doses of 1.0 mg/L or less prior to UF. Biofiltration has also been applied as a pretreatment strategy for UF fouling control (Hallé et al. 2009; Peldszus et al. 2012). Therefore, a combination of pretreatment strategies has the potential to improve UF performance, whereby low alum dosages (<0.5 mg/L) could be applied prior to biofiltration thereby improving biopolymer removal without adversely impacting head loss.
The objective of this pilot-scale study was to investigate the effect of biofiltration enhancement strategies and filter media type on NOM removal, DBP precursor reduction, and mitigation of UF fouling. Six pilot-scale biofiltration columns were used to examine the impact of nutrients, namely phosphorus and nitrogen, peroxide, in-line alum, as well as the selection of media type (GAC versus anthracite).
MATERIALS AND METHODS
Pilot studies were conducted at the Peterborough Water Treatment Plant, Ontario, Canada, where the Otonabee River serves as the source water. Raw water DOC and turbidity values ranged between 5.8 and 7.3 mg/L, and 0.3 and 1.5 NTU, respectively. In addition, temperatures ranged from 5 to 29 °C during the study. The raw water was prechlorinated at the intake for zebra mussel control when water temperatures exceeded 12 °C, which is standard practice for drinking water treatment plants across Southern Ontario. Prior to biofiltration, any remaining chlorine (typically <0.5 mg/L) was quenched by the addition of 3.5 mg/L of sodium thiosulfate per 1 mg/L of free chlorine.
The biofiltration pilot consisted of six parallel treatment trains incorporating two large glass biofilters with diameter of 15.24 cm and four small acrylic biofilters with a diameter of 7.62 cm (Figure S.1, available online at http://www.iwaponline.com/ws/015/091.pdf). The large and small biofilters were operated at empty bed contact times (EBCTs) of 11 and 10 min, respectively. Each column consisted of 50 cm of anthracite over 50 cm of sand, except for one which consisted of 50 cm of exhausted GAC (F300 Calgon Carbon) over 50 cm of sand. The exhausted GAC was harvested from a full-scale biofilter at a drinking water treatment plant in Ontario, Canada. The filters were operated in a constant head mode and backwashed using their own effluent three times per week, or as needed to maintain the desired EBCT. The backwash procedure consisted of a slow wash with air scour at collapse pulsing conditions for 2 min which ensured 30% bed fluidization, followed by 8 min of fast wash that facilitated 50% bed fluidization, and a slow wash for 4 min.
Ultrafiltration and fouling quantification
Two automated bench-scale ultrafiltration membrane systems incorporated polyvinylidene fluoride hollow fiber UF membranes, Zeeweed 1 Series 500 (ZW-1) (GE, Oakville, ON) with a nominal surface area of 0.047 m2 were used in this study. The units were operated at a flux of 30 L/(m2 × hour) (LMH) to mimic typical full-scale operation in filtration facilities located in the same regions. Each UF run consisted of 30 min permeation followed by backpulsing with air scour for 15 min; membrane tanks were completely drained and refilled prior to each run (Wray et al. 2014). Chemical cleaning was performed after each UF run (60 cycles over 48 hours) and consisted of a sodium hypochlorite (NaOCl) soak (750 mg/L) for 24 hours, followed by permeation with fresh 750 mg/L NaOCl for 30 min. Finally, the membranes were permeated with distilled water for 30 min. Membranes were stored in a 50 mg/L NaOCl solution to prevent microbial regrowth between experimental runs. UF runs were conducted in parallel or back-to-back within 48 hours.
Membrane fouling was quantified in terms of membrane resistance which is calculated by dividing the measured transmembrane pressure by flux (Wray et al. 2014). Reversible fouling for each 30 min cycle was calculated as the difference between the final resistance of a given cycle and the starting resistance of the following cycle. Irreversible fouling was calculated as the difference between starting resistances.
Chemical additions of phosphorus, nitrogen, peroxide, and alum were performed in parallel. Phosphorus and nitrogen were added in the form of phosphoric acid and ammonium chloride, respectively. Prior to chemical enhancement, the filters were fed raw water and backwashed with non-chlorinated backwash allowing for the biological community to acclimate. Biological activity was determined by measuring adenosine triphosphate (ATP) concentration >500 ng/g and monitoring DOC removals until steady state was observed (7 ± 4%) (Figure S.2, available online at http://www.iwaponline.com/ws/015/091.pdf). Acclimation period extended for 9 and 6 months, for the large and small filters, respectively. The first phase of chemical enhancement extended over a period of 69 days followed by the second and third phases for 24 and 28 days, respectively (Table 1). Each phase was extended until steady-state DOC levels were observed. Chemical doses were selected based on values reported in the literature (Urfer & Huck 2000; Lauderdale et al. 2012; Wray et al. 2014) and adjusted on the basis of head loss and DOC removal during the study. Filters were operated for at least three weeks following a change in chemical dose prior to sampling. Phosphorus and nitrogen were added to achieve carbon: nitrogen: phosphorus (C:N:P) ratios of 100:30:2, 100:40:2 and 100:40:20. Peroxide and alum were added in-line to the filter influent at doses of 0.1–1 mg/L and 0.1–0.5 mg/L, respectively (Table 1). The large and small biofilters were operated without chemical addition and served as controls.
|Filter .||Acclimation .||Phase 1 .||Phase 2 .||Phase 3 .|
|Nutrient enhancement||C:N:P 100:30:0||C:N:P 100:30:2||C:N:P 100:40:2||C:N:P 100:40:20|
|Peroxide addition||0 mg/L||0.1 mg/L||0.5 mg/L||1 mg/L|
|Alum addition||0 mg/L||0.1 mg/L||0.5 mg/L||0.25 mg/L|
|Filter .||Acclimation .||Phase 1 .||Phase 2 .||Phase 3 .|
|Nutrient enhancement||C:N:P 100:30:0||C:N:P 100:30:2||C:N:P 100:40:2||C:N:P 100:40:20|
|Peroxide addition||0 mg/L||0.1 mg/L||0.5 mg/L||1 mg/L|
|Alum addition||0 mg/L||0.1 mg/L||0.5 mg/L||0.25 mg/L|
DOC was measured using a wet oxidation method as described in Standard Method 5310 D (APHA 2012) with an O-I Corporation Model 1010 TOC Analyzer (College Station, Texas, USA). LC-OCD analyses were conducted at the University of Waterloo (Waterloo, ON) according to the method described by Huber et al. (2011).
ATP concentrations were assayed with a LuminUltra Deposit Surface Analysis kit (DSA-100C, Fredericton, NB) following the manufacturer's instructions. Extracellular polymeric substance (EPS) was extracted in Tris-EDTA buffer (10 mM Tris, 10 mM EDTA), where 10mL of sample was added to 2 g media, shaken at 300 rpm for 4 h at 4 °C and centrifuged (Liu & Fang 2002). The supernatant was filtered through a 0.45 μM and stored at −20 °C. Protein and polysaccharide components of EPS were quantified according to Pierce™ BCA (Thermo Scientific) and DuBois et al. (1956) methods, respectively. For protein quantification, a CE 3055 Single Beam Cecil UV/Visible Spectrophotometer (Cambridge, UK) was used; a Hach Odyssey DR/2500 Scanning Spectrophotometer (Mississauga, ON) was employed for polysaccharide measurements.
For DBP analyses, water samples were chlorinated to provide a residual of 1.5 ± 0.5 mg/L after 24 hours at room temperature (22 °C). Free chlorine residuals were measured as described in Standard Method 4500-Cl G (APHA 2012) and quenched with ascorbic acid (100 mg/L). The pH was not controlled (7.5–8.2) in order to mimic actual water quality conditions that may be observed at full scale.
Trihalomethanes (THMs) including chloroform, bromodichloromethane, dibromochloromethane, and bromoform and haloacetic acids (HAAs) such as monochloroacetic acid, monobromoacetic acid dichloroacetic acid, trichloroacetic acid, bromochloroacetic acid, dibromoacetic acid, bromodichloroacetic acid, dibromochloroacetic acid, and tribromoacetic acid were analyzed using a liquid–liquid extraction gas chromatographic method based on Standard Method 6232 B and 6251 (APHA 2012), respectively. A Hewlett Packard 5890 Series II Plus Gas Chromatograph (Mississauga, ON) equipped with an electron capture detector (GC-ECD) and a DB 5.625 capillary column (Agilent Technologies Canada Inc., Mississauga, ON) was used for both THM and HAA analysis. Absorbable organic halogens (AOX) were analyzed using a titration method based on Standard Method 5320 (APHA 2012) with a Trace Element Instruments Xplorer Organic Halogens Analyzer (Delft, The Netherlands).
A paired t-test (Walpole et al. 2011) was performed to evaluate the impact of the nutrient either nitrogen and/or phosphorus, peroxide, and alum addition on DOC removal, turbidity reduction and head loss. Student's t-test (Walpole et al. 2011) was used to determine impacts on DBP precursor removal. All statistical analyses were performed at the 95% confidence level.
RESULTS AND DISCUSSION
Biological acclimation and activity
Biological acclimation of the filters was determined by monitoring ATP concentrations and DOC removal across the filters (Figure S.2). Following 2 months of operation, the large filters had an ATP level of approximately 800 ng ATP/g media while DOC removal remained at 2%. As raw water temperatures increased, DOC removal increased and reached steady state (7 ± 4%) once temperatures exceeded 20 °C. By the end of the acclimation periods, ATP levels in both the small (541 ± 78 ng ATP/g media) and large (800 ± 107 ng ATP/g media) filters were similar; the average DOC removal for the small filters was 4 ± 3%. The GAC filter had an initial ATP concentration of 1,350 ATP ng/g media and DOC removal of 5 ± 4%. By the end of the acclimation period, DOC removal and ATP concentration for all filters was statistically similar when using a paired t-test (α = 0.05). The two control filters (small and large) statistically matched one another in terms of DOC, LC-OCD fraction, and DBP precursor removals (α = 0.05).
Passive biofiltration performance
DOC removal across the control biofilters was consistent (5 ± 2%, n = 18) throughout the study (Figure 1). Biofiltration studies conducted by Hallé et al. (2009) reported slightly higher removal rates of 11%, whereas Peldszus et al. (2012) reported <15% removals. In this study, it was observed that DOC removal was heavily influenced by water temperature (Figure S.3, available online at http://www.iwaponline.com/ws/015/091.pdf), where temperatures <12 °C resulted in <3% removal, consistent with trends reported by Emelko et al. (2006) and Moll et al. (1999) temperatures <10 °C.
The control biofilters readily removed 17 ± 6% of raw water biopolymers (initial concentration (C0) = 0.41–0.53 mg/L) (Figure 1). This was not reflected in the DOC data and demonstrated that gross DOC measurements are not an effective means of assessing biopolymer removal. Biofiltration was ineffective at removing HS, where the control biofilters removed 1 ± 2% (n = 8) of the raw water values (C0 = 3.0–3.3 mg/L). Preferential biopolymer removal highlighted the potential for biofiltration to control UF membrane fouling. Also, since biopolymer removal remained >15% at temperatures <12 °C, biofiltration could potentially be applied as a membrane pretreatment on a year-round basis, even when source water temperatures may be low (below 4 °C) during cold seasons.
Biofiltration provided an effective means of DBP precursor removal in terms of THMs, HAAs, and AOX (Figure 1). THMs observed above detection limits consisted of chloroform (∼82%) and bromodichloromethane (∼18%), whereas HAAs consisted only of dichloroacetic acid (∼37%) and trichloroacetic acid (∼63%). THMs and HAAs accounted for 55 ± 7% of the total AOX by mass, leaving 45% of the halogenated compounds unidentified. THM formation potentials (C0 = 145–202 μg/L in raw water) were significantly reduced by 19 ± 9% across the biofilter (α = 0.05). Similarly, HAAs (C0 = 60–132 μg/L in raw water) decreased by 13 ± 9% (α = 0.05). AOX precursors (C0 = 475–557 μg/L) were reduced by 8 ± 10%, but this was not statistically significant. Similar to the DOC trends, DBP precursor removal was adversely impacted at low temperatures. However, the ability of biofiltration to reduce DBP precursors at temperatures <12 °C highlights its potential application for locations with high seasonal variability.
Impact of nutrient enhancement
Nutrient enhancement was not observed to improve biofilter performance in terms of DOC, biopolymer or DBP precursor removal. When the C:N:P was increased to 100:40:20, biopolymer removal decreased by 25% when compared to the control (Figure 1). This demonstrates that a nutrient surplus beyond 100:10:1 as reported by LeChevallier et al. (1991) for optimal microbial growth may potentially hinder biological performance. The ineffectiveness of nutrient enhancement is in contrast with results presented by Lauderdale et al. (2012). Neither phosphorus nor nitrogen was utilized across the filter. It is possible that the biodegradable carbon fraction was limited in this study such that biofilters may have degraded all of the assimilable organic material. Therefore, any addition of nitrogen or phosphorus would not be beneficial.
Impact of peroxide addition
The addition of 0.1, 0.5 and 1 mg/L H2O2 had a positive effect on biofilter performance by decreasing head loss by 9, 48, and 40%, respectively (Table S.1, available online at http://www.iwaponline.com/ws/015/091.pdf). The improvement was statistically significant (α = 0.05) at both 0.5 and 1 mg/L levels. Peroxide demand was 0.1–0.3 mg/L prior to the filter in the effluent and <0.1 mg/L was observed in the filter effluent. Head loss reduction may be a direct result of biofilm degradation. The biofilm, or EPS, is composed mainly of proteins and polysaccharides. Both constituents were measured to be 6 and 52% lower than the control when 0.5 and 1 mg/L peroxide was applied, respectively (Figure S.4, available online at http://www.iwaponline.com/ws/015/091.pdf). Similarly, Lauderdale et al. (2012) reported lower EPS concentrations in a peroxide-enhanced biofilter. In the current study, EPS concentrations were 30% higher in comparison to the control when 0.1 mg/L of H2O2 was applied. Since the peroxide dose was lower than the raw water demand, NOM oxidization could have increased the biodegradable carbon resulting in biofilm growth. At a dose of 1 mg/L, biopolymer removal decreased by 6%. This could reflect biofilm oxidation and the release of biopolymers. This is supported by peroxide-induced degradation of biofilm, where lower EPS values were measured on the media in comparison to the control biofilter. Also at 1 mg/L, THM, HAA, and AOX precursor removals decreased by 2–12% when compared to the control while no adverse impacts on DBP formation potentials were observed at 0.1 and 0.5 mg/L (α = 0.05). These findings show that the optimal peroxide dose for maximizing head loss while maintaining biofilter performance for the water studied was approximately 0.5 mg/L.
Impact of alum addition
An alum dose of 0.1–0.5 mg/L has been reported by Wray et al. (2014) to effectively remove 0.1 mg/L biopolymers. In the present study, application of 0.1 and 0.25 mg/L of alum removed an additional 3% and 4% of biopolymers, respectively, when compared to the control. This translated into an overall removal of 0.01 and 0.02 mg/L of biopolymers across the filter. With respect to head loss, the effect of in-line alum addition was found to be dose-dependent, where 0.1 mg/L resulted in a 40% reduction. When the dose was increased to 0.25 mg/L, the filter performed similarly to the control (α = 0.05). However, a significant increase of 52% in the head loss was observed when a dose of 0.5 mg/L (α = 0.05) was applied. Typical direct filtration coagulant dosages for surface waters are <20 mg/L and are expected to increase the head loss as result of aluminum hydroxide precipitation (Edzwald et al. 1987). In this study, 0.1 and 0.25 mg/L of alum did not have an adverse head loss effect. As such, low coagulant dosages (<0.5 mg/L) have the potential for use in high DOC (5–7 mg/L) waters prior to UF without significantly impacting head loss. As was the case for biopolymers, an alum dose of 0.1 mg/L resulted in the removal of an additional 5% of THM (11 μg/L) and 3% of HAA (6 μg/L) precursors when compared to the control. Addition of 0.25 mg/L alum did not impact THM or HAA removal but increased AOX concentrations by 51 μg/L (α = 0.05). An alum concentration of 0.5 mg/L had a detrimental impact on DBP precursor removals, where THM removal was 7% lower when compared to the control; no HAA precursor removal was observed. These results highlight the importance of dosing at an optimal alum concentration (<0.5 mg/L).
Impact of filter media type
GAC removed on average an additional 0.1 mg/L of DOC when compared to anthracite (α = 0.05). As previously discussed, the ability of GAC to remove DOC may be attributed to the higher surface area and/or the bioregeneration at adsorption sites. However, GAC did not provide any additional advantage for biopolymer reduction as removals were similar to the anthracite control (α = 0.05) at 15 ± 10%. It has been documented that biopolymers are relatively large in size, which may prevent them from accessing GAC adsorptive sites (Velten et al. 2011b; Gibert et al. 2013); thus, biopolymer removal across the GAC was attributed to physical removal and biodegradation by the microorganisms. Bioregeneration may account for the increase in DOC removal. When compared to anthracite, GAC removed an additional 4% of THM and 6% of AOX precursors, but the trend did not extend to HAAs where anthracite outperformed GAC by removing on average an additional 4% of HAA precursors (n = 4).
Engineered biofiltration for UF fouling mitigation
To evaluate whether biofiltration could reduce UF fouling, two ultrafiltration units were operated in parallel; one receiving raw water and the other receiving control biofilter effluent for 30 hours of permeation. Biofiltration reduced the irreversible fouling resistance rate from (7.4 ± 29) × 1010 m−1/cycle to (2.7 ± 25) × 1010 m−1/cycle (Figure 2(a)) representing a 60% improvement. All data were corrected for temperature using Equation S.1 (available online at http://www.iwaponline.com/ws/015/091.pdf). No impact was observed for the reversible fouling as it remained at 1.1 × 1012 m−1/cycle for both UF units. Irreversible fouling mitigation is essential to improve membrane operations since this can reduce the backwash frequency. Reduction in irreversible UF fouling was attributed to biopolymer removal (17 ± 6%) and particulate (turbidity) removal (59 ± 14%) rates.
Addition of phosphorus and nitrogen (C:N:P = 100:40:20) did not improve irreversible fouling, where resistance of (7.8 ± 22) × 1010 m−1/cycle and (7.7 ± 57) × 1010 m−1/cycle was calculated for the nutrient enhanced and control biofilters, respectively (Figure 2(b)). Furthermore, reversible fouling increased by approximately 50%, attributable to inefficient biopolymer removal by the nutrient enhanced biofilter.
In-line alum coagulation prior to biofiltration at dosages of 0.1, 0.25, and 0.5 mg/L was evaluated as pretreatment strategy for UF fouling control (Figures 2(c)–2(e)). Application of 0.1 and 0.25 mg/L alum decreased the irreversible fouling by 60% and 35%, respectively. This improvement may be attributed to an additional 3–4% biopolymer removal (when compared to the control biofilter) and underscores the importance of this specific organic fraction when considering UF fouling strategies. It is well established that the alum is less efficient at low temperatures (Van Benschoten & Edzwald 1990). Temperatures steadily decreased while the alum dose increased over the duration of the study. The impact of temperature on alum performance could have resulted in poor biopolymer removal and UF fouling control, despite the increase in dosage. The increase in alum dose may also result in higher aluminum residuals, which could foul the UF membrane (Zheng et al. 2012). Conversely, an alum dosage of 0.5 mg/L resulted in lower reversible fouling (35%) when compared to the control. This improvement can be attributed to the formation of a cake layer, which prevented further attachment of foulants, as discussed by Paar et al. (2011) and Zheng et al. (2012). LC-OCD data were not available for the 0.5 mg/L alum condition; therefore biopolymer removal could not be analytically quantified or correlated.
Peroxide addition of 1 mg/L increased reversible fouling by 45% and had little impact on irreversible fouling, where only a nominal increase of 5% was observed (Figure 2(f)). Peroxide can oxidize the EPS and cause the release of proteins and polysaccharides. This is supported by the 52% reduction of EPS constituents on the biofilter media when 1 mg/L peroxide was added, as discussed earlier. The improvement in head loss with peroxide addition did not translate into improvement in UF fouling. In fact, peroxide adversely impacted UF performance. These findings merit further investigation of peroxide doses <1 mg/L on biofiltration as a pretreatment for UF; however, they do show promise for improving biofiltration run times.
GAC provided similar irreversible fouling to that observed for anthracite, but decreased reversible fouling by 29% (Figure S.5, available online at http://www.iwaponline.com/ws/015/091.pdf).
Biofiltration effectively removed turbidity, biopolymer, and DBP precursors as well as reduced irreversible UF fouling. Increasing the C:N:P ratio in the filter influent to 100:30:2, 100:40:2, and 100:40:20 did not improve the biological performance of the microbial community. Addition of hydrogen peroxide at dosages <1 mg/L decreased the head loss. However at 1 mg/L, the gains in filter run time did not translate into UF fouling mitigation and DBP precursor removals were negatively impacted. In-line alum coagulation (<0.5 mg/L) prior to biofiltration reduced UF fouling and DBP precursors by removing biopolymer fraction without adversely impacting head loss. GAC outperformed anthracite in NOM and DBP precursor removal and reduced UF reversible fouling.
This work was funded in part by the Natural Sciences and Engineering Research Council of Canada (NSERC) Chair in Drinking Water Research at the University of Toronto and the Ontario Research Fund (ORF). The authors would like to thank the personnel of Peterborough Utilities Services for their support as well as Dr Monica Tudorancea and Dr Sigrid Peldzsus at the University of Waterloo for performing LC-OCD analyses.