Recently, treatment plants have adopted biofiltration to compliment conventional treatment and ozonation. Previous literature has focused on passive applications of biofiltration. In this study, several pilot-scale biofiltration trains were designed in parallel to conventional treatment to investigate the impact of nutrient addition (nitrogen and phosphorus), use of hydrogen peroxide, aluminum sulphate (alum), and different filtration media (anthracite vs. granular activated carbon (GAC)) on biofiltration performance. Parameters measured included organic removal, reduction of DBP precursor, improvements in filter runtimes and ultrafiltration (UF) fouling control. Nutrient addition did not improve biofiltration performance. Supplementing hydrogen peroxide (<1 mg/L) decreased headloss, DBP formation potentials while adversely affecting UF fouling. In-line alum addition (<0.5 mg/L) improved biofilter’s ability to control fouling and DBP precursor without adversely impacting headloss. GAC provided superior performance when compared to anthracite. Conventional treatment provided higher DOC, and DBP precursor removal, as well as better UF fouling control compared to biofiltration.

Recently, treatment plants have adopted biofiltration to compliment conventional treatment and ozonation. Previous literature has focused on passive applications of biofiltration. In this study, several pilot-scale biofiltration trains were designed in parallel to conventional treatment to investigate the impact of nutrient addition (nitrogen and phosphorus), use of hydrogen peroxide, aluminum sulphate (alum), and different filtration media (anthracite vs. granular activated carbon (GAC)) on biofiltration performance. Parameters measured included organic removal, reduction of DBP precursor, improvements in filter runtimes and ultrafiltration (UF) fouling control. Nutrient addition did not improve biofiltration performance. Supplementing hydrogen peroxide (<1 mg/L) decreased headloss, DBP formation potentials while adversely affecting UF fouling. In-line alum addition (<0.5 mg/L) improved biofilter’s ability to control fouling and DBP precursor without adversely impacting headloss. GAC provided superior performance when compared to anthracite. Conventional treatment provided higher DOC, and DBP precursor removal, as well as better UF fouling control compared to biofiltration.

A series of pilot and full-scale experiments were carried out at the Mannheim Water Treatment Plant in Kitchener, Ontario to examine the impact of backwash technique, filter media characteristics, and combinations thereof on single stage drinking water biological filter performance. The media characteristics investigated were effective size, uniformity coefficient, and media type (GAC and anthracite). Backwash techniques investigated were the collapsed pulse backwash, the extended terminal subfluidization wash (ETSW), and the presence of chlorine in the wash water. Single stage biological filters must serve the dual purpose of biologically mediated removal of biodegradable organic matter (BOM), as well as meeting traditional filter performance criteria such as turbidity removal with minimal head loss accumulation. Accordingly, dissolved organic carbon removal, biodegradable dissolved organic carbon removal, biological respiration potential, turbidity removal, filter ripening time, and head loss accumulation were all quantified as measures of biological filtration performance. The results of this study have several implications for optimized design and operation of biological filters during drinking water treatment. An increase in effective size of media grains from 1.0 mm to 1.3 mm was shown to significantly extend filter run time by minimizing head loss accumulation without compromising turbidity or BOM removal. Uniformity coefficient however, showed no significant effect on biological filter performance; indicating that the performance benefits associated with highly uniform media may not be commensurate with cost. GAC was found to be significantly more resilient to backwashing in collapsed pulse and chlorinated modes, which impaired BOM removal in anthracite filters. This resilience imparts a high degree of operational flexibility to backwashing GAC filters. The significant decrease in BOM removal by anthracite filters can be minimized; however, by using an optimized backwashing technique. Collapsed pulse backwashing was found to have a significant effect on biological filter performance. When chlorinated collapsed pulse was used, filter cycles were significantly shortened by approximately 30 – 50% due to a sudden surge in effluent turbidity. This effect is thought to be the result of biofilm, damaged during the course of backwashing sloughing from the media. Extended terminal subfluidization wash was found to significantly reduce, and often eliminate filter ripening entirely. Additionally, the extended contact time with chlorine associated with chlorinated ETSW did not appear to have a significant effect on filter BOM removal. By eliminating filter ripening without compromising biological performance, ETSW shows promise for significant water and production cost savings by minimizing the filter-to-waste period during filter ripening. The presence of chlorine however, was associated with decreased DOC, 24 hours in to the filter cycle. This factor, combined with the negative interaction between chlorine and collapsed pulse suggests chlorinated wash water should be avoided in biological filtration systems like the ones investigated.

A series of pilot and full-scale experiments were carried out at the Mannheim Water Treatment Plant in Kitchener, Ontario to examine the impact of backwash technique, filter media characteristics, and combinations thereof on single stage drinking water biological filter performance. The media characteristics investigated were effective size, uniformity coefficient, and media type (GAC and anthracite). Backwash techniques investigated were the collapsed pulse backwash, the extended terminal subfluidization wash (ETSW), and the presence of chlorine in the wash water. Single stage biological filters must serve the dual purpose of biologically mediated removal of biodegradable organic matter (BOM), as well as meeting traditional filter performance criteria such as turbidity removal with minimal head loss accumulation. Accordingly, dissolved organic carbon removal, biodegradable dissolved organic carbon removal, biological respiration potential, turbidity removal, filter ripening time, and head loss accumulation were all quantified as measures of biological filtration performance. The results of this study have several implications for optimized design and operation of biological filters during drinking water treatment. An increase in effective size of media grains from 1.0 mm to 1.3 mm was shown to significantly extend filter run time by minimizing head loss accumulation without compromising turbidity or BOM removal. Uniformity coefficient however, showed no significant effect on biological filter performance; indicating that the performance benefits associated with highly uniform media may not be commensurate with cost. GAC was found to be significantly more resilient to backwashing in collapsed pulse and chlorinated modes, which impaired BOM removal in anthracite filters. This resilience imparts a high degree of operational flexibility to backwashing GAC filters. The significant decrease in BOM removal by anthracite filters can be minimized; however, by using an optimized backwashing technique. Collapsed pulse backwashing was found to have a significant effect on biological filter performance. When chlorinated collapsed pulse was used, filter cycles were significantly shortened by approximately 30 – 50% due to a sudden surge in effluent turbidity. This effect is thought to be the result of biofilm, damaged during the course of backwashing sloughing from the media. Extended terminal subfluidization wash was found to significantly reduce, and often eliminate filter ripening entirely. Additionally, the extended contact time with chlorine associated with chlorinated ETSW did not appear to have a significant effect on filter BOM removal. By eliminating filter ripening without compromising biological performance, ETSW shows promise for significant water and production cost savings by minimizing the filter-to-waste period during filter ripening. The presence of chlorine however, was associated with decreased DOC, 24 hours in to the filter cycle. This factor, combined with the negative interaction between chlorine and collapsed pulse suggests chlorinated wash water should be avoided in biological filtration systems like the ones investigated.

This thesis outlines results of a series of studies investigating the removal of two common taste and odour compounds, 2-methylisoborneol (MIB) and geosmin, from drinking water using biologically active sand filtration. A combination of full-, pilot- and laboratory-scale studies were carried out. A review of long term water quality data from a South Australian water treatment plant indicated that the conventional plant was capable of removing MIB and geosmin to below detection limit without the need for additional treatment. A series of laboratory studies were carried out, validating the theory that the geosmin removal was occurring through biological activity in the rapid gravity filters of the water treatment plant. Microorganisms capable of geosmin removal were found to be present in the settled water of two South Australian water treatment plants, Morgan and Happy Valley. Laboratory sand column experiments were conducted with these waters and a range of sand media, investigating the effect of biofilm development on MIB and geosmin biodegradation. It was found that the process could produce effective removals, however long start-up periods were often required. A laboratory-scale column utilising new sand fed with Happy Valley settled water took in excess of 300 days before it was capable of removing MIB and geosmin by greater than 80%. Studies on sands with inactivated pre-existing biofilms required much shorter biofilm development periods, from 30 to 40 days. The results of the column studies indicated that a method to encourage sand filters to operate biologically for MIB and geosmin removal would be advantageous. Two methods were studied: preozonation and bacterial inoculation. Pre-ozonation was carried out at a pilot plant, constructed at the Happy Valley water treatment plant. Additional factors investigated during this study were the length of the biofilm development period and the impact of empty bed contact time (EBCT). Preozonation is often used in tandem with biological filtration to increase the fraction of biodegradable organic matter and in turn increase the biomass activity of the filter. The pilot plant consisted of two sand filters; one fed with settled water and one fed with preozonated settled water. Pre-ozonation did not enhance the biodegradation of MIB or geosmin. The pre-ozonated column was run for 550 days. Removals of MIB and geosmin were inconsistent throughout the trial. The maximum removal obtained during the study was 80% for MIB and geosmin, at an EBCT of 45 minutes, after 380 days of operation. The settled water column was run for over 650 days. By day 560, the column was able to remove 60% of the influent geosmin and 40% of the influent MIB at an EBCT of 10 minutes, which is close to that used in full-scale plants. Significant effects of empty bed contact time were not noted in the range of 10 to 30 minutes. Bacterial inoculation studies were carried out at laboratory-scale. The inoculum comprised of a geosmin-degrading consortium of three Gram-negative bacteria previously isolated from the biofilm of the Morgan water treatment plant filter sand. A sand column with a pre-existing biofilm was inoculated with the organisms, achieving 70% removal of geosmin. Inoculation of columns without biofilms gave lower geosmin removals, with an average of 41% removal. These were preliminary studies only, and further work is required. A biomass activity assay, based on the concentration of adenosine triphosphate (ATP), was developed over the course of the project. This assay was particularly helpful when studying the attachment of the inoculum in the laboratory columns. Other methods to study biomass were flow cytometry to enumerate the water-borne and biofilm associated bacteria, and scanning electron microscopy to obtain a visual observation of the biofilms on various sands. This work demonstrated the potential of biological sand filtration for MIB and geosmin control. It was shown that long biofilm development periods are evident before effective removal of the compounds can occur. The potential to minimise these long biofilm development periods by inoculation of filters with geosmin degrading organisms was demonstrated. / http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=1340100 / Thesis (Ph.D.) -- University of Adelaide, School of Chemical Engineering, 2008

In order to meet increasing water demands and more stringent regulations drinking water treatment plant managers must continually look to new treatment strategies and optimization techniques. One such strategy is to eliminate chlorine residual before filtration, allowing indigenous bacteria already present in the source water to grow on the filter media. These microorganisms help improve effluent water quality by increasing organic and inorganic contaminant removal. The process is known as biological filtration, or biofiltration. The implications of converting a conventional filtration plant (not specifically designed for biofiltration) to a biofiltration plant are still not well understood. Therefore, the purpose of this study was to evaluate water quality and operational trends of a side-by-side full-scale biofiltration conversion at the Quail Creek Water Treatment Plant (QCWTP), located in Hurricane, Utah, and to determine the impact of pre-chlorination elimination on filter performance. Four of twelve filters at the QCWTP were used to test the plant’s ability to operate in biological mode. One acted as a control and ran similar to the other eight filters in the treatment plant. The other three were converted to biofilters by quenching the influent chlorine residual with thiosulfate. The experiment lasted one year, so filter performance could be evaluated in each season. The results from the study indicated that the influent water was low in organic carbon (i.e. food for microorganisms), which resulted in small differences in biological activity between filters. Disinfection by-products (DBPs) (i.e. cancer causing agents created from the combination of chlorine and organic matter) were lower in the biofilters relative to the control. Biological conversion resulted in slightly higher and more variable final effluent turbidity values (though still within EPA drinking water standards and operational goals) compared to the non-biological filters; however, filter run times were unaffected.

This research aimed to address the knowledge gaps in the literature regarding the removal of waterborne pathogens (viruses and protozoa) by modified multistage slow sand filtration. In the current study, two pilot-scale multistage slow sand filtration systems were operated continuously for over two years. The pilot systems treated agricultural- and urban-impacted raw river water of variable quality with turbidity peaks over 300 NTU and seasonal cold temperatures <2??C. The first system (Pilot 1) consisted of two independent trains that included pre-ozonation, shallow-bed upflow gravel roughing filtration, and shallow-bed slow sand filtration. Pilot 1 was a pilot-scale version of an innovative, commercially available full-scale system. The second system (Pilot 2) included a full-depth upflow gravel roughing filter, a full-depth slow sand filter, and a second shallow-depth slow sand filter in series. The SSFs of both pilots were operated at high hydraulic loading rates (typically 0.4 m/h) at the upper limit of the literature recommended range (0.05 to 0.4 m/h). Both pilot systems provided excellent turbidity removal despite the high filtration rates. Effluent turbidity of all multistage SSF pilot systems were within the regulated effluent limits in Ontario for full-scale SSFs (below 1 NTU at least 95% of the time and never exceeded 3 NTU), despite raw water turbidity peaks over 100 NTU. The roughing filters contributed to approximately 60-80% of the full-train turbidity removal, compared to and 20-40% for the slow sand filters. On average, the second slow sand filter in pilot 2 provided almost no additional turbidity removal. The slow sand filter run lengths were short because of frequent high raw water turbidity, with about 50-80% of the runs in the range of 1-3 weeks. To prevent excessive SSF clogging and maintenance, filtration rates should be decreased during periods of high turbidity. Seven Cryptosporidium and Giardia challenge tests were conducted on the slow sand filters of both pilot systems at varying filtration rates (0.4 or 0.8 m/h), temperatures (2 to 25??C), and biological maturities (4 to 20 months). Removal of oocysts and cysts were good regardless of sand depth, hydraulic loading rate, and water temperature in the ranges tested. Average removals in the SSFs ranged from 2.6 to >4.4 logs for Cryptosporidium oocysts and ranged from >3.8 to >4.5 logs for Giardia cysts. This was consistent with findings in the literature, where oocyst and cyst removals of >4 logs have been reported. Cryptosporidium oocyst removals improved with increased biological maturity of the slow sand filters. At a water temperature of 2??C, average removal of oocysts and cysts were 3.9 and >4.5 logs, respectively, in a biologically mature SSF. Doubling the filtration rate from 0.4 to 0.8 m/h led to a marginal decrease in oocyst removals. Sand depths in the range tested (37-100 cm) had no major impact on oocyst and cyst removals, likely because they are removed primarily in the upper section of slow sand filter beds by straining. In general, good oocyst and cyst removals can be achieved using shallower slow sand filter bed depths and higher filtration rates than recommended in the literature. There are very few studies in the literature that quantify virus removal by slow sand filtration, especially at high filtration rates and shallow bed depths. There are no studies that report virus removal by slow sand filtration below 10??C. As such, 16 MS2 bacteriophage challenge tests were conducted at varying water temperatures (<2 to >20??C) and filtration rates (0.1 vs. 0.4 m/h) between February and June 2006 on biologically mature slow sand filters with varying bed depths (40 vs. 90 cm). Biologically mature roughing filters were also seeded with MS2. Average MS2 removals ranged from 0.2 to 2.2 logs in the SSFs and 0.1 to 0.2 logs in the RFs under all conditions tested. Virus removal by slow sand filtration was strongly dependant on hydraulic loading rate, sand depth, and water temperature. Virus removal was greater at a sand depth of 90 cm vs. 40 cm, at an HLR of 0.1 m/h vs. 0.4 m/h, and at warm (20-24??C) vs. cold (<2-10??C) water temperatures when sufficient warm water acclimation time was provided. Increased sand depth likely increased MS2 removal because of greater detention time for predation and greater contact opportunities for attachment to sand grains and biofilms. A lower HLR would also increase MS2 removal by increasing detention time, in addition to decreasing shear and promoting attachment to filter media and biofilms. Greater MS2 removal at warmer water temperatures was attributed to improved biological activity in the filters. Schmutzdecke scraping was found to have only a minor and short-term effect on MS2 removals. Virus removal can be optimized by providing deep SSF beds and operating at low filtration rates. Virus removal may be impaired in cold water, which could affect the viability of using SSF/MSF at northern climates if communities do not use disinfection or oxidation. As a stand-alone process, slow sand filtration (with or without roughing filtration) may not provide complete virus removal and should be combined with other treatment processes such as disinfection and oxidation to protect human health.

The main objective of this study was to improve our understanding of biological filtration (biofilm type) treatment for manganese (Mn) removal in drinking water. Biological filtration treatment involves biofilms of Mn(II)-oxidizing microorganisms attached to solid filter material that remove and immobilize dissolved Mn(II) in raw water by conversion to black MnO2(s) precipitates. Mn-biological filtration is an emerging green technology that can serve as an alternative to conventional physicochemical treatments but its full potential is hindered by various factors. These include lack of understanding the: (1) optimal removal conditions for Mn, (2) mechanisms for Mn releases of the accumulated Mn in the biofilter, and (3) effects of recalcitrant natural organic matter (NOM) on biofiltration. Confounding these issues is the unknown identity of the diverse microbial communities which occupy the biofilms attached to the filter media. To investigate these issues, biological Mn removal was studied in laboratory bench scale reactors using a new Mn(II)-oxidizing bacterium isolate, Pseudomonas Putida EC112. The main research hypothesis formulated that the transition metal catalyst, MnO2(s), can increase the bioavailable carbon and energy from recalcitrant NOM (e.g., humic acids (HA)) to biological filters. Mn and HA can be found in most natural waters, including groundwaters, lakes and streams. To test the hypothesis, the potential for strain EC112 growth and Mn(II) oxidation utilizing the organic substrate products from the oxidation reaction between HA and MnO2(s) was assessed. Biological Mn(II)-oxidation kinetics were investigated in batch (suspended cell) and continuous flow (biofilm) bioreactors at optimal pH and temperature conditions for strain EC112. Batch kinetics was successfully characterized with the Monod model. Continuous flow steady-state kinetics was modeled with a single, zero-order kinetic parameter. Enhanced Mn(II) removal capacity was observed for strain EC112 in batch and continuous flow reactors in the presence of HA and MnO2. The effect of MnO2(s) on HA biodegradability was studied and optimal conditions for biodegradation were identified. Biofilter Mn(II) releases were observed during the continuous flow bioreactor experiments. Release conditions were identified and releases modeled using pseudo first-order kinetics. Changes in HA structure induced by MnO2(s) oxidation were studied with Fourier transform infrared (FT-IR) and proton nuclear magnetic spectroscopy (1H-NMR).

This research aimed to address the knowledge gaps in the literature regarding the removal of waterborne pathogens (viruses and protozoa) by modified multistage slow sand filtration. In the current study, two pilot-scale multistage slow sand filtration systems were operated continuously for over two years. The pilot systems treated agricultural- and urban-impacted raw river water of variable quality with turbidity peaks over 300 NTU and seasonal cold temperatures <2°C. The first system (Pilot 1) consisted of two independent trains that included pre-ozonation, shallow-bed upflow gravel roughing filtration, and shallow-bed slow sand filtration. Pilot 1 was a pilot-scale version of an innovative, commercially available full-scale system. The second system (Pilot 2) included a full-depth upflow gravel roughing filter, a full-depth slow sand filter, and a second shallow-depth slow sand filter in series. The SSFs of both pilots were operated at high hydraulic loading rates (typically 0.4 m/h) at the upper limit of the literature recommended range (0.05 to 0.4 m/h). Both pilot systems provided excellent turbidity removal despite the high filtration rates. Effluent turbidity of all multistage SSF pilot systems were within the regulated effluent limits in Ontario for full-scale SSFs (below 1 NTU at least 95% of the time and never exceeded 3 NTU), despite raw water turbidity peaks over 100 NTU. The roughing filters contributed to approximately 60-80% of the full-train turbidity removal, compared to and 20-40% for the slow sand filters. On average, the second slow sand filter in pilot 2 provided almost no additional turbidity removal. The slow sand filter run lengths were short because of frequent high raw water turbidity, with about 50-80% of the runs in the range of 1-3 weeks. To prevent excessive SSF clogging and maintenance, filtration rates should be decreased during periods of high turbidity. Seven Cryptosporidium and Giardia challenge tests were conducted on the slow sand filters of both pilot systems at varying filtration rates (0.4 or 0.8 m/h), temperatures (2 to 25°C), and biological maturities (4 to 20 months). Removal of oocysts and cysts were good regardless of sand depth, hydraulic loading rate, and water temperature in the ranges tested. Average removals in the SSFs ranged from 2.6 to >4.4 logs for Cryptosporidium oocysts and ranged from >3.8 to >4.5 logs for Giardia cysts. This was consistent with findings in the literature, where oocyst and cyst removals of >4 logs have been reported. Cryptosporidium oocyst removals improved with increased biological maturity of the slow sand filters. At a water temperature of 2°C, average removal of oocysts and cysts were 3.9 and >4.5 logs, respectively, in a biologically mature SSF. Doubling the filtration rate from 0.4 to 0.8 m/h led to a marginal decrease in oocyst removals. Sand depths in the range tested (37-100 cm) had no major impact on oocyst and cyst removals, likely because they are removed primarily in the upper section of slow sand filter beds by straining. In general, good oocyst and cyst removals can be achieved using shallower slow sand filter bed depths and higher filtration rates than recommended in the literature. There are very few studies in the literature that quantify virus removal by slow sand filtration, especially at high filtration rates and shallow bed depths. There are no studies that report virus removal by slow sand filtration below 10°C. As such, 16 MS2 bacteriophage challenge tests were conducted at varying water temperatures (<2 to >20°C) and filtration rates (0.1 vs. 0.4 m/h) between February and June 2006 on biologically mature slow sand filters with varying bed depths (40 vs. 90 cm). Biologically mature roughing filters were also seeded with MS2. Average MS2 removals ranged from 0.2 to 2.2 logs in the SSFs and 0.1 to 0.2 logs in the RFs under all conditions tested. Virus removal by slow sand filtration was strongly dependant on hydraulic loading rate, sand depth, and water temperature. Virus removal was greater at a sand depth of 90 cm vs. 40 cm, at an HLR of 0.1 m/h vs. 0.4 m/h, and at warm (20-24°C) vs. cold (<2-10°C) water temperatures when sufficient warm water acclimation time was provided. Increased sand depth likely increased MS2 removal because of greater detention time for predation and greater contact opportunities for attachment to sand grains and biofilms. A lower HLR would also increase MS2 removal by increasing detention time, in addition to decreasing shear and promoting attachment to filter media and biofilms. Greater MS2 removal at warmer water temperatures was attributed to improved biological activity in the filters. Schmutzdecke scraping was found to have only a minor and short-term effect on MS2 removals. Virus removal can be optimized by providing deep SSF beds and operating at low filtration rates. Virus removal may be impaired in cold water, which could affect the viability of using SSF/MSF at northern climates if communities do not use disinfection or oxidation. As a stand-alone process, slow sand filtration (with or without roughing filtration) may not provide complete virus removal and should be combined with other treatment processes such as disinfection and oxidation to protect human health.

Research has shown biological filtration can be a successful treatment for manganese (Mn) removal from groundwater and surface water. In this study, bench-scale direct biofiltration was used to remove Mn and dissolved organic carbon (DOC) from a pH 6 surface water source in Halifax, Canada. The removal of Mn in pH 6 surface water was significantly (? = 0.05) removed with 200-300 µg/L phosphorus (P), and 500 µg/L hydrogen peroxide (H2O2). DOC removal was significantly (? = 0.05) improved with granular activated carbon (GAC) media, P enhancement at 200-300 µg/L, and H2O2 enhancement at 500 µg/L. Mn was likely removed by biological oxidation and physical adsorption to biogenic Mn and iron (Fe) oxides. These results show direct biofiltration of surface water at pH 6 can remove Mn below the 50 µg/L aesthetic guideline from a Mn loading of over 1 mg/L. Further research is required to verify the microbial mechanism.