Removal of Natural Organic Matter Fractions by Two Potable Water Treatment Systems: Dual Membrane Filtration and Conventional Lime Soda Softening

Transcription

1 Removal of Natural Organic Matter Fractions by Two Potable Water Treatment Systems: Dual Membrane Filtration and Conventional Lime Soda Softening Charles D. Goss* and Beata Gorczyca Department of Civil Engineering, University of Manitoba 15 Gillson Street, Winnipeg, Manitoba, R3T 5V6 Canada Submitted to: 4 th Annual IWA Specialty Conference on Natural Organic Matter: From Source to Tap and Beyond July 27-29, 2011 Costa Mesa, California, USA

2 Acknowledgments This group would like to acknowledge the following for financial support and technical assistance on this project. Financial Support Manitoba Water Stewardship Manitoba Water Stewardship Fund Technical Support Pembina Valley Water Cooperative and the Town of Morris, Manitoba Portage la Prairie Water Treatment Plant and the City of Portage la Prairie, Manitoba Jeff O Driscoll and Ken Anderson (Genivar) Victor Wei (Manager-University of Manitoba Environmental Laboratory

3 Abstract The objective of this study was to investigate the removal of dissolved organic carbon (DOC) fractions by two water treatment plants, the Portage La Prairie Water Treatment Plant (PPWTP), which uses lime/soda softening with granular activated carbon (GAC) filtration, and the Morris Water Treatment Plant (MWTP), a dual micro/nano membrane facility, located in Manitoba, Canada. The study aimed to determine the cause of reportedly high trihalomethane (THM) concentrations in plant effluent. Both the PPWTP and MWTP use surface water sources, the Assiniboine and Red River, respectively, which are reportedly high in DOC, fluctuating from 7mg/L to 18mg/L throughout the year. As a result of the high DOC in the source water both plants have reported high THM concentrations that have, in the past, exceeded the 100ppb maximum limit set by the Province of Manitoba. Solid phase extraction (SPE) was used to fractionate DOC in water samples collected during this study. The SPE method fractionated DOC into six fractions: hydrophobic acid (HPOA), hydrophobic base (HPOB), hydrophobic neutral (HPON), hydrophilic acid (HPIA), hydrophilic base (HPIB), and hydrophilic neutral (HPIN). Samples were collected from the PPWTP and Assiniboine River on November 8, 2010, January 20, 2011 and April 2, 2011 to evaluate the DOC and DOC fraction removal throughout the plant. Results found that the GAC filter was ineffective at removing DOC, often with DOC concentrations increasing after the GAC filter. The HPOA fraction, largely believed to contain the greatest THM precursors, was unaffected by GAC filter showing the potential cause for reported high THM levels at the plant. All hydrophilic fractions increased after the GAC filter and only the HPOB fraction was reduced by GAC filtration at the PPWTP. The recommendation to the PPWTP from this group was to improve the coagulation process to reduce organic loads on the GAC filter. Samples were collected from the Red River on September 25, 2010, and fractionated, to establish the relative composition of the river. The results found that the late summer composition of the Red River was 45% hydrophobic and 55% hydrophilic, with 40% of the total organic component being HPIN. On November 25, 2010 samples from the Red River and MWTP were fractionated to establish membrane removal efficiency. The results were unexpected finding that DOC increased from 8.7mg/L to 10.2mg/L. The HPIA and HPIN fractions increased after the nano filter from mg/L and mg/L, respectively. The HPOA fraction was found to be unaffected by the nano filter while the HPON, HPOB, and HPIB faction had small decreases in concentrations. However, for samples collected in February, 2011 DOC concentrations were reduced to <0.5mg/L by the nano filter. The reason for the high DOC found after the nano for the November sampling period is unclear however it is believed that (1) the samples were taken just prior to a cleaning event where filter was not removing DOC effectively or (2) that the use of citric acid to clean the nano membrane could have added a carbon source to the nano effluent. It is recommended that a pre-treatment process be implemented prior to the micro/nano membranes to reduce the DOC load on the membranes preventing both high THM concentrations and membrane fouling.

4 Introduction Potable water treatment facilities in Manitoba often suffer from sources waters with high concentrations of natural organic matter (NOM). High NOM causes concern due to the formation of harmful disinfection byproducts (DBPs), such as trihalomethanes (THMs), which form when chlorine reacts with NOM, during disinfection treatment (Singer, 1999). In Manitoba water chlorine demands often exceed 5 milligrams per liter (mg/l), resulting in residuals of mg/l. If NOM is not removed prior to chlorination treatment the unreacted chlorine can react with the NOM and form THMs, and other halogenated byproducts, which are potential carcinogens (Krasner, 2009). As a result, the province of Manitoba had adopted standards set by The Drinking Water Quality Act which requires all public potable water suppliers meet a quarterly average of <0.100 mg/l for total THMs (TTHMs) (Manitoba Water Stewardship, 2007). Therefore, water treatment facilities are faced with improving the removal of the organic matter from the raw water prior to chlorination in order to reduce the concentrations of THMs and other halogenated byproducts. NOM in Potable Water Treatment Natural organic matter can be removed by both chemical and physical treatments. The use of chemical coagulants in combination with filtration (sand or membrane) and physical adsorption (activated carbon) are processes which treatment plants apply in order to remove organics prior to disinfection. Although all of these processes remove NOM to some degree each comes with their own set of drawbacks. Separating NOM laden coagulation flocs by physical filtration either through sand or membrane filtration, or adsorption of NOM to activated carbon filters, creates several challenges to treatment facilities. Mainly, all filtration systems require some backwash, flushing or cleaning process in order to remove organics and other materials built up on the surface. These cleaning procedures are often laborious and costly as it requires the use of chemicals, water for backwashing, or physical cleaning by operators. Also, the build-up of organics on the surface of membranes can cause reversible and irreversible fouling and reduce the overall life of the membrane resulting in costly replacement (Agenson, 2007). The applications of adsorptive filters in situations where the NOM concentrations in the source water are high have been used with limited success. Often these activated carbon filters quickly reach adsorptive capacity due to size exclusion of large molecules blocking pores thereby reducing their effectiveness of NOM removal (Amy, 1992). Therefore, these filters require increased cleaning or flushing events and more frequent media replacement. NOM largely consists of humic and fulvic acids which are largely believed to contain the greatest THM precursors. Fulvic acids with high charge density, due to high carboxylic acid moieties, are minimally affected by charge neutralization during coagulation compared to humic acids (Musikavong et al., 2005; Amy, 1992). Although lowering the ph (ph 4-6.5) will improve coagulation by reducing the charge density of fulvic acids as well as reduce the solubility of the coagulant, water treatment operation at low ph ranges are difficult and costly due to high volumes of acids required to overcome the buffering capacity of waters with high alkalinity (Sawyer, 2003; Amy, 1992). Therefore, NOM not removed in coagulation pre-treatment has potential to form THMs. Natural organic matter is unique to location and therefore it is important to establish the composition of organic matter as well as concentration (Chow, 2008). Organic carbon

5 fractionation studies are conducted to gain better understanding of the chemical and physical properties of local organic matter. Studies often focus on dissolved organic matter, or dissolved organic carbon (DOC), which is the organic carbon that is able to pass through a 0.45 micrometer (µm) filter paper, as this is typically harder to remove than particulate NOM. Fractionation of DOC can be based on physical size through the use of filtration and size exclusion chromatography. Likewise, chemical properties, such as hydrophobicity and charge, can also be used to fractionate DOC. Methods developed that separate DOC on hydrophobicity and charge were largely based upon methods originally developed by Leenheer and Aiken (Leenheer, 1981, Aiken, 1992). Recent methods separate DOC into hydrophobic acid (HPOA), hydrophobic base (HPOB), hydrophobic neutral (HPON), hydrophilic acid (HPIA), hydrophilic base (HPIB), and hydrophilic neutral fractions (HPIN). The original methods used ion exchange resins (eg. XAD-4 and XAD-8) which were often laborious to prepare and required long run times (~24h). New methods have been developed to reduce sample preparation and run times. One method developed by Ratpukdi et al. uses solid phase extraction (SPE) to fractionate DOC into these six fractions (Ratpukdi, 2009). This method reduces run times to roughly 12 hours as well as preparation times as the SPE cartridges are pre-packaged and the sample requires only filtration and ph adjustment. Dissolved Organic Carbon Fractions and THMs It is largely suggested that the HPOA fraction, which is largely humic matter, has the greatest potential to form THMs, and other by-products, due to its high aromaticity and reactivity (Singer, 1999; Leenheer, 2003; Chow 2005). However, a review study conducted by Chow in 2005 mentions several studies that found the HPI fractions had greater THMFP while others suggest that both HPO and HPI form THMs when chlorinated (Chow, 2005). This review, and others, suggests that the formation of THMs is dependent on local environment and that THMFP and organic composition are unique to that location. Therefore, caution should be taken when estimating the potential of organic fractions to form THMs. Objectives and Significance of Research This study focused on the characterization of DOC and its removal in two water treatment plants located in Manitoba that suffer from high DOC concentrations in the source water. The first plant is located in Portage la Prairie and is a conventional softening plant with ballasted flocculation pretreatment and granular activated carbon (GAC) filtration. The second plant is located in Morris and is a newly constructed dual membrane (micro/nano) facility that replaced a conventional lime soda softening plant. Both plants use surface water sources, the Assiniboine and Red Rivers, which experience high NOM concentrations between 8 and 18 mg/l. As a result both plants have, in the past, experienced elevated THM concentrations occasionally exceeding the Manitoba guideline of mg/l. This study utilized the Ratpukdi et al. solid phase extraction method to fractionate water DOC in the samples collected from the Red and Assiniboine Rivers as well as the Morris and Portage la Prairie water treatment plants in order to establish the removal efficiency of DOC fractions by each process. Understanding removal efficiency of DOC from each plant will provide operators and design engineers with greater knowledge of which processes are effective at removing DOC and which need optimizing in order to reduce the formation of THMs.

6 Water Treatment Plants Analyzed in this Study Portage la Prairie Water Treatment Plant (PPWTP) The City of Portage la Prairie has a population of roughly 13,000 people and has been growing at a steady rate for the past several years due to the development of food processing industries and agriculture in the region. To meet the long term water demands of the growing community, along with the construction of a new potato processing facility in the area, the PPWTP underwent several upgrades from Upgrades to increase water production from 18 mega liters per day (ML/d) to 34ML/d were implemented in order to meet growing demands, as well as, improving and adding treatment processes to overcome challenges such as high turbidity, hardness, and organic matter, along with occasional increased algal growth, often seen in the Assiniboine River. Turbidity levels at the PPWTP were found to exceed 1500 nephelometric turbidity units (NTU) and peaks of 6000 NTU have been recorded (Table 1). The high levels of DOC (15mg/L) found in the river resulted in THM levels that often exceeded the mg/l guideline set by the Province of Manitoba. Table 1: Water quality for the Assiniboine River and PPWTP for samples collected in May Guidelines presented here are according to the Canadian Drinking Water Quality Guidelines for 2001 (Table adapted from Anderson, 2003) Parameters Raw Water Treated Water Guideline Hardness (mg/l CaCO3) * Turbidity (NTU) TOC (mg/l) ** Chloroform (mg/l) Bromoform (mg/l) BDCM (mg/l) DBCM (mg/l) Total THMs (mg/l) Note: * = Operational objective **= Aesthetic guideline To improve the removal of turbidity a John Meunier ACTIFLO Ballasted Flocculation Clarification system was implemented as a preclarification step (Figure 1). Pilot studies found that raw waters with turbidity levels of 2500 NTU could be reduced to 3 NTU overcoming the occasional high spikes in turbidity seen in the Assiniboine River (Anderson, 2003 and 2004). Figure 1: ACTIFLO Ballasted Flocculation system (Anderson, 2003)

7 Along with the ACTIFLO system, an additional clarifier and new chemical dosing systems were installed to increase plant flow. Improvements to the backwash system for the four dual media sand filters were made to improve plant performance and promote organics removal. Ozone was applied to softened and clarified water to promote biologically stable water, to minimize chlorine demand, reduce the formation of DBPs, and to improve taste and odor. In addition to sand filtration granular activated carbon adsorption was introduced to reduce organics. Upgrades were made to the storage reservoirs to increase disinfectant contact time and plant capacity. Lastly, a state of the art control system replaced the original system in order to provide operators with complete control and monitoring of the system. Although preliminary plant performance tests indicate that the new system improved the removal of turbidity and hardness, tests in suggest the plant may not be effectively removing organics as THM levels were found to occasionally exceed guideline limits (Table 2). Table 2: THM results for Portage la Prairie Water Treatment Plant from Date THM concentration (µg/l) Chloroform BDCM DBCM Bromoform Total THM January 9, <1 48 May 9, <1 36 August 23, December 11, January 7, April 22, < 9.5 < 59.5 *BDCM = Bromodichloromethane DBCM = Dibromochloromethane < = less than detection limits Pembina Valley Water Cooperative (Morris, MB) The Pembina Valley Water Cooperative owns and operates the water treatment plant in Morris, Manitoba. The original Morris water treatment plant (MWTP) was constructed in 1998 and was a typical lime soda softening plant with a flow capacity of 32 liters per second (L/s) however it was determined that the plant would need to be expanded to meet growth in population and industry in the area. The source for the plant is the Red River and, like the Assiniboine, is prone to high DOC concentrations often exceeding 12 mg/l, as well as turbidity levels ranging from NTU. Due to the high DOC concentrations in the Red River THM levels were also found to, at times, exceed the Manitoba guidelines (Table 3). Table 3: THM concentrations for two storage reservoirs supplied by the Morris water treatment plant. Samples were collected on November 4, 2009 THM concentration (µg/l) Parameter Miami (Influent) Rosenort (Influent) Bromodichloromethane Bromoform < < Chloroform Dibromochloromethane 2 3.8

8 In 2008, construction began at the MWTP to renovate the existing lime soda softening plant into a state of the art dual membrane facility. The upgrade included Pall ARIA micro membrane filtration and Pall Ultipleat High Flow nano membrane filters. According to Pall the micro filters would remove turbidity to 0.1 NTU along with three log reduction of Giardia and Cryptosporidium. The nano filters would remove hardness and reduce organic matter concentrations to <0.5 mg/l. The implementation of the membrane system would expand capacity from 32 L/s to 66 L/s with room to increase flow to 100 L/s if required. A 1,000,000 m 3 retention pond was also constructed to provide the plant with more stable source water and to ensure availability during drought. Figure 2 outlines the flow for the membrane facility in Morris. Construction of the facility was completed in late 2009 and went online in early Figure 2: Plant flow diagram for Pembina Valley Water Cooperative in Morris, Manitoba (Figure supplied by Anderson, 2009) Table 4: Water quality parameters for the Morris water treatment plant after the installation of the membrane system. Samples were collected in March 26, Sample Identification Unit Raw Water Post Micro Post Nano Tap DOC mg/l < True Color TCU <5.0 <5.0 TDS Calculated mg/l < Turbidity NTU <0.10 <0.01 Alkalinity mg/l CaCO₃ The water quality parameters for the MWTP, taken after the installation of the micro and nano membrane (Table 4), suggest that the system is capable of meeting water quality guidelines. However, the presence of high organics found in the source water could cause significant problems to membranes over long term use. DOC characterization can provide insight into the potential for these organics to cause fouling preventing unnecessary replacement costs. Research Methodology Sample collection Portage la Prairie Water Treatment Plant and Assiniboine River 1 liter (L) water samples were collected from the Assiniboine River, via an intake in the plant, as well as throughout the water treatment plant three times during this study; November 8, 2010, January 20, 2011 and April 2, These three sampling dates represent river water conditions during fall (prior to snow fall), winter and spring. Figure 3 diagrams the sampling

9 locations at the PPWTP. November 8 th samples only tested the overall removal of DOC throughout the plant. Samples were collected from before and after the GAC filter on January 20, 2011 to establish the DOC fraction removal efficiency of the filter. All samples collected at Portage la Prairie and the Assiniboine on April 2, 2011 were fractionated using SPE method (described below). Figure 3: Sampling locations for Portage la Prairie water treatment plant. (1) Assiniboine River (2) after ACTIFLO ballasted flocculation (3) after lime softening (4) after recarbonation (5) after ozonation (6) after sand filtration (7) after sand filter reservoir (8) after GAC (9) Finished water Sample collection Morris Water Treatment Plant and Red River 1-4L water samples were collected at various times during the study from the Red River as well as from the retention pond, post micro filter, and post nano filter effluent, prior to blending. Samples were collected on August 11, 2010 to establish an estimate of the summer THM and THM formation potential (THMFP) from the Red River and treated MWTP effluent. THM concentration and THMFP analysis was conducted by ALS laboratories (ALS Laboratories Winnipeg, Manitoba). Samples collected from the Red River on September 25, 2010 were fractionated to establish the relative composition of the river DOC during late summer. On November 23, 2010 and February 28, 2011 samples were collected from the river as well as the retention pond, post micro filter and post nano filter for full fractionation study. Note that February sampling did not include the Red River due to ice cover. All DOC measurements were made using Standard Method 5310 with a Tekmar Dohrmann Phoenix 8000 total organic carbon analyzer (Tekmar Dohrmann, Ohio) which was calibrated according to the instrument manual. All DOC measurements were made in triplicate. All samples collected during this study were filtered through 0.45µm nitro cellulose filter paper prior to analysis except for THM and THMFP analysis conducted by ALS laboratories.

10 Fractionation using Solid Phase Extraction (SPE) Prior to fractionation all samples were filtered through 0.45µm nitrocellulose filter paper and were brought to room temperature. The fractionation method used in this study was developed by Ratpukdi et al. (2009). The method fractionates DOC into six fractions: hydrophobic acid (HPOA), hydrophobic base (HPOB), hydrophobic neutral (HPON), hydrophilic acid (HPIA), hydrophilic base (HPIB), and hydrophilic neutral fractions and is as follows. The fractionation procedure uses three nonpolar Bond Elute ENV cartridges (Varian Inc, Lake Forest California), one Phenomenex Strata XC strong cation exchange cartridge, and one Phenomenex Strata X-AW weak anion exchange cartridge (Phenomenex, Torrance, California). All SPE cartridges contained 1gram of sorbent. The fractionation procedure begins with all cartridges being conditioned with 10mL of HPLC grade methanol (MeOH). The Strata XC and X-AW cartridges are then conditioned with 10mL of 1.0M hydrochloric acid (HCl). All five cartridges are then rinsed with deionized water until the effluent DOC measured <0.100 mg/l. 1L of sample was adjusted to ph 7.0 using either concentrated sulfuric acid (H 2 SO 4 ) or 1M sodium hydroxide (NaOH) and was drawn through the first ENV cartridge labeled ENV-1. The fraction collected on the ENV-1 SPE cartridge is defined as the HPON fraction. The same sample water was then adjusted to ph of 10 using 1.0M NaOH and drawn through the second ENV cartridge (ENV-2). The fraction retained on this cartridge is the HPOB fraction. Next, the sample water was adjusted to ph of 2 with H 2 SO 4 and drawn through the third ENV cartridge (ENV-3) capturing the HPOA fraction. Following the ENV-3 cartridge the sample water was drawn through the Strata XC cartridge without ph adjustment. The HPIB fraction was retained on the Strata XC cartridge. Lastly, the sample water was adjusted to ph of 7 with NaOH and drawn through the Strata X-AW cartridge which captured the HPIA fraction. The fraction of organic matter that was not retained by any of the five cartridges is defined as the HPIN fraction. All samples were drawn though the SPE cartridges at mmhg vacuum pressure. After the sample was drawn through each SPE cartridge a 40mL sample was collected to measure the DOC concentration. Figure 4 outlines the Ratpukdi et al. method. Figure 4: DOC SPE fractionation setup (Ratpukdi et al., 2009)

11 DOC (mg/l) Results and Analysis Characterization of DOC in Assiniboine River and Portage la Prairie Water Treatment Plant On November 23, 2010 samples were collected form the PPWTP and the Assiniboine River to establish the general removal of organics during the treatment process. As shown in Figure 5 the incoming DOC of 16.1mg/L is significantly reduced to 5.0 mg/l after coagulation and clarification. The coagulant that is used at the PPWTP is Alufer S25. The DOC is further reduced to 3.8mg/L after the sand filtration; however the concentration increases after the GAC filter to 6.7mg/L Figure 5: DOC removal at the Portage la Prairie Water Treatment Plant for samples collected on November 23, 2010 (Personal communication Hooshyar, 2010) It was suspected that the increase in DOC may be due to the GAC media being over capacity as a result of constant high DOC source water. Organics could leach into the water from the filter increasing the concentration. This increase could potentially provide enough organic content to form levels of THMs that would exceed guideline limits. It must be noted that the effluent DOC was 1.5mg/L showing a removal after the GAC filter. However, there is no treatment process after the GAC filter that would reduce the DOC; it is unclear how the removal occurred. More data is currently being collected at this plant to evaluate DOC removal by the GAC filter. Water samples were collected on January 20, 2011, from before and after the GAC filter, to determine the dissolved organic fraction removal efficiency/ability of the filter (Table 5). It was found that there was no DOC removed by the GAC filter, on the contrary, the DOC was found to increase by mg/L or 3.2% following GAC filtration. The water samples DOC were also fractionated (Table 5) to establish which fractions were affected by the filter. HPIB and HPIA fractions showed the greatest increase in concentration from mg/L to mg/L and mg/L to mg/L, while HPOA and HPIN fractions were largely unaffected by the GAC filter. The HPON and HPOB fractions experienced the greatest reduction by the GAC filter from mg/L to mg/L and mg/L to mg/L, respectively. The mechanisms that caused the overall change in DOC composition occurring on/near the filter was not a focus of this study therefore reasoning for the increase/decrease of fractions cannot be fully evaluated. However, the ineffective DOC removal by the GAC filter suggests that the filter has exceeded adsorptive capacity.

12 DOC (mg/l) Table 5: DOC concentration changes occurring before and after the GAC filter at the Portage la Prairie Water Treatment Plant for samples collected on January 20, Fraction DOC concentration (mg/l) Before GAC After GAC HPON 0.51± * HPOB 0.288± * HPOA 2.3± * HPIB ± ±0.01 HPIA 0.04± ±0.03 HPIN ± * Total *No error value could be established due to instrument carryover Figure 6: Overall DOC removal at the Portage la Prairie Water Treatment plant. Samples were collected on April 2, 2011(Hooshyar, 2011) Figure 6 shows the overall removal of organics at the PPWTP for samples collected on April 2, The removal of organics is largely done by the lime softening and recarbonation processes removing 10mg/L, combined. Note that the ACTIFLO system was not active during this sampling period. An increase in DOC from 7.0mg/L to 9.5 mg/l is seen after ozonation. Sand filtration reduces the concentration of DOC down to 6.9mg/L however the GAC filter is again ineffective at removing any organics showing an increase in DOC concentration to 7.2mg/L. Fractionation of all samples collected on April 2, 2011 was conducted to determine how each process within the plant affects DOC fraction removal (Figures 6 and 7). The fractionation results for samples collected from the river and the PPWTP show that nearly all fractions were reduced by lime softening and clarification with the HPOB fraction having the greatest reduction of nearly 92%. After the ozonation process all fractions increased, except the HPIB fraction, by 15-50%. Similar results were seen in a study conducted by Śweitlik et al. (2004) where it was found that after the application of ozone there was an increase in HPIA, HPIN and HPON, with a small increase in HPIB (Śweitlik, 2004). Śweitlik et al. suggests that the application of ozone will also increase the biodegradable organic carbon (BDOC) fraction (Śweitlik, 2004). Higher

13 DOC (mg/l) concentrations of BDOC could cause an increase in microbial populations in the system potentially causing another source for the increase in organics seen in the PPWTP. Further research on the effect of ozone and the increase in microbial populations is currently being conducted to better understand if the presence of microbes influences the formation of THMs. Sand filtration reduced all fractions by 21-42% except for HPIB which increased 97% after sand filtration. The HPON fraction was unaffected by the sand filtration process. The GAC filter was found to have similar results to the January 20, 2011 results which found the HPIN fraction to be largely unaffected with only a small decrease in concentration of 0.5mg/L. The HPOA fraction increased by 24% from1.6mg/l to 2.1mg/L, likewise the HPIB and HPIA fractions increased by 54% and 45%, respectively. Only the HPON and HPOB fractions were reduced by the GAC filter with a removal of 64% for HPON and 100% for HPOB. These results clearly show that the GAC filter is ineffective at removing all organic fractions. The GAC filter can be seen as an unfavourable process as organics, especially the HPOA fraction, increase after the filter. In turn, there is a large presence of HPOA in the finished water for this sampling period, and throughout this study, suggesting there is potential to form THMs based on the high reactivity of the these compounds noted in literature. However, there is some controversy as to the fraction of DOC that contains the largest THMFP. Therefore, further analysis into the THMFP of organics found in the Assiniboine River and PPWTP effluent is required. Currently, THMFP analysis for the river and treatment plant is being conducted by this group. The results lastly show an increase in HPIN, HPOA, HPOB and HPON fractions in the finished water while the HPIB and HPIA fractions decreased. Although it is unclear why there is an increase in DOC in the system from the GAC to finished water, this increase poses an issue for THM control. Further study into the increase seen after the GAC to the finished water is ongoing HPON HPOB HPOA HPIB HPIA HPIN Figure 7: Removal of DOC fractions from the Assiniboine River and Portage la Prairie Water Treatment plant. Samples were collected on April 2, 2011.

14 One suggestion to address the problem with high THMs at the PPWTP would be to optimize the coagulation process prior to the GAC filter. If the coagulation process is optimized for DOC removal the GAC filter will experience low concentrations allowing greater time before the filter reaches capacity and no longer removes organics. The use of a more effective type of GAC media is another possible solution to improve the removal of DOC. Cheng et al. demonstrated that modified activated carbon such as iron impregnated or modification with helium or ammonia significantly improved the removal of DOC over virgin activated carbon (Cheng, 2005). Further investigation into a more effective type of carbon media is being conducted by this group. Characterization of DOC in Red River and Pembina Valley Water Cooperative (Morris, Manitoba) On August 11, 2010 samples were collected from the Red River and the Morris water treatment plant to establish the THM concentrations as well as THMFP of the Red River and the treated effluent (Tables 6 and 7). The results show that the THM levels for plant effluent are below the guideline limits of 100ppb (Table 6), however the THMFP of the pond is significantly higher than the guideline suggesting there is potential to for THM levels that exceed the 100ppb limit. Likewise, the THMFP of the nano effluent also shows potential to form THM levels that are greater than 100ppb. Although the Morris plant is seemingly controlling THMs there is evidence that high DOC feed water can negatively impact the performance and lifetime of membrane filters (Cho, 1999). Table 6: THM concentrations for samples collected from the Red River and Morris water treatment plant on August 11, Samples were analysed by ALS Laboratories (Winnipeg, MB). Sample Location THM concentrations (mg/l) Chloroform Bromoform Dibromochloromethane Bromodichloromethane Red River < < < < Retention Pond < < < < Post Micro < < < < Post Nano < Table 7: THMFP for samples collected on August 11, 2010 from the retention pond and post nano effluent at the Morris water treatment plant. Samples were analysed by ALS Laboratories (Winnipeg, MB). Sample Location THMFP concentrations (mg/l) Chloroform Bromoform Dibromochloromethane Bromodichloromethane Total Retention Pond < Post Nano < Samples collected from the Red River on September 25, 2010 were fractionated to establish the relative composition of the river during late summer (Table 8). The results show that the Red River is 45% hydrophobic and 55% hydrophilic, with 40% of the total organic

15 DOC (mg/l) composition being HPIN. Although the concentrations of THMs at Morris are low (Table 6) there is a large HPOA fraction of nearly 22% suggesting the potential to form THM upon chlorination. Table 8: Fractionation results for the Red River collected on September 25, 2010 Fraction DOC (mg/l) % DOC HPON HPOB 0.21 <2 HPOA HPIB 0.21 <2 HPIA HPIN Total Samples collected from the Red River and Morris water treatment plant on November 23, 2010 were fractionated to determine the removal efficiency of the micro and nano filter membranes (Figure 8). The results obtained from this sample set deviated from what was expected. It was suggested by the Pall Corporation that the nano membranes would be able to reduce DOC concentrations to <0.5mg/L, however the results from this sample set found that the DOC concentration increased after the nano filter from 8.7mg/L to 10.2mg/L. The HPIA and HPIN fractions increased after the nano filter from mg/L and mg/L, respectively. The HPOA fraction was unaffected by the nano filter while the HPON, HPOB, and HPIB faction had small decreases in concentrations. It must be noted that shortly after this sampling period the Morris plant reported unexpected levels of THMs in the distribution system ranging from 75-86ppb (Fehr, 2010). Although the levels are still below the required 100ppb guideline the increase is of concern since the maximum allowable THMs concentration are rumoured to be reduced to 80ppb in the near future. The increase in THMs may have resulted from the poor removal of organics seen during the November 23 rd sampling period. The reason for the increase in DOC is not fully understood however there it is suggested that the sampling event may have taken place just prior to a cleaning event where DOC rejection was not efficiently occurring. Another potential cause is the use of citric acid as a cleaning agent for the nano membranes. If the citric acid was not fully rinsed after the cleaning cycle it could have caused the increase in HPIA. Further investigation into the relationship between membrane cleaning cycles, citric acid as a possible organic source, and organics removal is suggested Red River Pond Post Micro Post Nano HPON HPOB HPOA HPIB HPIA HPIN Figure 8: Fractionation results for Red River and throughout the Morris water treatment plant. Samples collected on November 23, 2010

16 DOC (mg/l) DOC (mg/l) Samples were collected on February 28, 2011 from the retention pond and at the Morris water treatment plant for DOC and fractionation analysis. Note samples could not be collected from the Red River due to ice cover. The DOC results found during this sampling period resembled the expected results with the overall DOC being reduced from 9.00mg/L to 0.42mg/L after nano filtration (Figure 9). Due to the low DOC in the nano filter effluent the sample could not be fractionated. Pond water and post micro filter water fractionation results (Figure 10) found that the change in overall DOC concentration was not significant after micro filter, however there were noticeable changes in the overall composition of DOC following micro filtration. There was a roughly 50% decrease in HPON (1.1mg/L to 0.53mg/L) and about a 50% increase in HPIA (1.37mg/L to 2.46mg/L). This could be related to microbial growth on the clean side of the nano-membrane surface or membrane surface-doc interactions changing the chemical properties of the compounds. The fractionation results also show there is a large HPOA component in the post micro membrane filter effluent. This effluent is blended with nano treated water at a 20-30% blend rate. This is done to increase hardness and alkalinity in the final effluent that is removed by the nano filter. However, with the HPOA fraction being 35% of the total DOC in the micro effluent caution should be taken when increasing this blend rate to ensure organic concentrations do not increase to where THM levels are exceeding guidelines. It is suggested by Fan et al. that hydrophobic polyvinylidene fluoride (PVDF) membranes, such as those at the Morris plant, are fouled largely by HPIN and HPOA fractions (Fan, 2001). These two fractions constitute nearly 52% of the total DOC entering the plant suggesting there is a potential for fouling of the membranes Pond Post Micro Post Nano Figure 9: Overall DOC removal at Morris water treatment plant. Samples collected February 28, HPON HPOB Pond Post Micro HPOA HPIB HPIA HPIN Figure 10: DOC fraction removal at Morris water treatment plant. Samples collected February 28, 2011

17 The evaluation of the organics removal at the Morris water treatment plant found that although the membranes were effective at reducing the DOC concentrations and in turn control the formation of THMs, there are points in time that the membrane is not effectively removing DOC. The reasons for the increase in organics seen in the November samples is inconclusive although there is evidence to suggest there may be a relation to cleaning events and/or the use of citric acid in the cleaning of the membranes. It is recommended that the relationship between cleaning events and increases in DOC should be investigated. High concentrations of HPIN and HPOA, suggested by Fan et al., could cause organic fouling to PVDF membranes. Therefore it is recommended that Morris re-implement a pretreatment that will remove organics before entering the membranes. Initially the engineers that designed the membrane upgrade used an existing clarifier from the original lime softening plant as a coagulation tank to reduce the organic load seen by the membranes. However, this process was not optimized and in turn unreacted coagulant (alum) was able to pass the micro filters causing the nano filters pressure to dramatically increase to risky levels in a short time (<24h). It was recommended by the engineers that the pretreatment be stopped due to the potential damage of the nano membranes. If this pretreatment step were optimized (mixing times, coagulant types, coagulant dose) the overall organic load would be reduced without the risk of damaging the membranes. This could extend the life of the membranes, as well as reduce the number of cleaning cycles, reducing the operation costs for the plant due to costly membrane replacements. Optimization of the coagulation process for removal of targeted DOC fractions is being currently investigated by Water Research Group at the University of Manitoba. Conclusions and Recommendations The objective of this research was to characterize the dissolved organic carbon and its removal efficiency in two potable water treatment plants located in Portage La Prairie and Morris (Manitoba, Canada). This study also aimed to establish the organic composition of the two river sources for each plant, the Assiniboine and Red Rivers, and to evaluate the concentration of fractions suspected to form THMs and foul membranes. The first plant located in Portage la Prairie, which uses the Assiniboine River as a source, is a lime softening plant with ballasted flocculation and granular activated carbon filtration. DOC removal results from all sampling at the Portage la Prairie water treatment plant during this study found that the granular activated carbon filter was ineffective at removing DOC, often with concentrations increasing post GAC filter. The HPOA fraction, suspected to largely contain THM precursors, was unaffected by the GAC filter and nearly all fractions, especially hydrophilic compounds, increased after the GAC filtration. The Morris plant is a newly designed dual membrane (micro/nano) facility that, like the Portage plant, experiences source water high in DOC (Red River). Two sampling periods were conducted on the Red River and at the Morris plant to evaluate DOC faction removal: November, 2010 and February Results from the November, 2010 sampling found that the nano membrane was not removing DOC effectively. Specifically the HPOA fraction and all hydrophilic fractions were not removed by the nano filter. The results obtained in February, 2011 were very different than the November results as the nano membranes were found to reduce DOC levels to <0.5mg/L. The reason for the high DOC found after the nano for the November sampling period is unclear however it is believed that (1) the samples were taken just

18 prior to a cleaning event where filter was not removing DOC effectively or (2) that the use of citric acid to clean the nano membrane could have added a carbon source to the nano effluent. Further analyses are currently being conducted at the Morris plant. Results obtained at both plants indicate that the major DOC fractions present in both raw water supplies (Red and Assiniboine Rivers), ie. HPOA, HPIN, HPIA and HPIB are not effectively removed by the treatment processes utilized. It is believed that the high concentration of the HPOA fraction could lead to increased THM concentrations after chlorination. The following recommendations can be made from this study: Due to the uncertainty as to which fraction contains the greatest potential to form THMs, a THMFP study of fractions collected from the Assiniboine River water intake be conducted to establish which fractions in the local environment from most THMs. Optimize coagulation process and GAC filtration for removal of the targeted DOC fractions. Identify DOC fraction with the highest nano-filter fouling potential. References Agenson, K.O., Urase, T., (2007) Change in membrane performance due to organic fouling in nanofiltration (NO)/ reverse osmosis (RO) applications. Separation and Purification Technology 55 pp Aiken, G. (1992) Isolation of hydrophilic organic acids from water using nonionic macroporous resins. Organic Geochemistry 18:4 pp.567 Amy, G.L., Sierka, R.A., Bedesse, J. Price, D. Tan, L. (1992) Molecular Size Distribution of Natural Organic Matter. American Water Works Association. June issue pp Anderson, K. (2003) Taming the turbid Assiniboine River. Presentation at 55 th Annual WCWWA Conference and Trade Show. Winnipeg, Manitoba, Canada Anderson, K. (2009) Morris WTP upgrades- Replacing cold lime soda softening with an integrated membrane system. Power point presentation presented on September 22, 2009 Anderson, K., Brant, B. O`Driscoll, J. (2004) Taming the turbid Assiniboine River: The upgraded Portage la Prairie WTP. Western Canada Water (Fall issue) pp Cheng,W., Dastgheib, A., Karanfil, T. (2005) Adsorption of dissolved organic matter by modified activated carbons. Water Research 39 pp Cho, J., Amy, G. Pellegrino, J. Yoon, Y. (1999) Characterization of clean and natural organic matter (NOM) fouled NF and UF membranes and foulants characterization. Water Supply 17:1 pp

19 Fan, L., Harris, J.L., Roddick, F.A., Booker, N.A. (2001) Influence of the characteristics of natural organic matter on the fouling of microfiltration membranes. Water Research 35:18 pp Fehr, J. ( ) Personal communication with Mr. Jake Fehr Regional Manager Pembina Valley Water Cooperative Chow, A.T., Dahlgren R.A., Zhang, Q., Wong, P.K. (2008) Relationship between specific ultraviolet absorbance and trrihalomethane precursors of different carbon sources. Journal of Water Supply: Research and Technology AQUA 57:7 pp Chow, A.T., Gao, S., Dahlgren R.A. (2005) Review paper: Physical and chemical fractionation of dissolved organic matter and trihalomethane precursors. Journal of Water Supply: Research and Technology AQUA 54:8 pp Fehr, J. ( ) Personal communication with Mr. Jake Fehr Regional Manager Pembina Valley Water Cooperative Hooshyar, M. ( ) Personal communication with Ms. Hooshyar (Graduate Student University of Manitoba) Krasner, S.W. (2009) The formation and control of emerging disinfection by-products of health concern. Phil. Trans.R. Soc A. 367 pp Leenheer, J. (1981) Comprehensive approach to preparative isolation and fractionation of dissolved organic carbon from natural waters and wastewaters. Environmental Science and Technology 15:5 pp Manitoba Water Stewardship (2011) Regulatory Information Acts and Regulations Retrieved April 2011 from Muskiavong, C., Wattanachira,S.W., Marhaba, T.F., Pavasant,P. (2005) Reduction of organic matter and trihalomethane formation potential in reclaimed water from treated industrial estate wastewater by coagulation. Journal of Hazardous Materials B127 pp Ratpukdi, T., Rice, J.A., Chilom, G., Bezbaruah, A. Kahn, E. (2009) Rapid Fractionation of natural organic matter in water using a novel solid phase extraction technique. Water Environment Research 81:11 pp Sawyer, C.N., McCarty, P.L., Parkin, G.F. (2003). Chemistry for Environmental Engineering and Science 5 th Ed. McGraw-Hill.New York, NY Singer, P. C.; (1999) Humic substances as precursors for potentially harmful disinfection byproducts. Water Science and Technology 40:9 pp Świetlik, J., Dabrowska, A. Raczyk-Stanislawiak, U., Nawrocki, J., (2004) Reactivity of natural organic matter fractions with chlorine dioxide and ozone. Water Research 38 pp