Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The efficiency of ultrafiltration membranes at removing TOC and THMFP in a British Columbia surface water Kenway, Samantha Elizabeth 2001

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2002-0129.pdf [ 6.27MB ]
Metadata
JSON: 831-1.0063725.json
JSON-LD: 831-1.0063725-ld.json
RDF/XML (Pretty): 831-1.0063725-rdf.xml
RDF/JSON: 831-1.0063725-rdf.json
Turtle: 831-1.0063725-turtle.txt
N-Triples: 831-1.0063725-rdf-ntriples.txt
Original Record: 831-1.0063725-source.json
Full Text
831-1.0063725-fulltext.txt
Citation
831-1.0063725.ris

Full Text

THE EFFICIENCY OF ULTRAFILTRATION M E M B R A N E S AT R E M O V I N G TOC A N D THMFP IN A BRITISH C O L U M B I A SURFACE W A T E R by S A M A N T H A E L I Z A B E T H K E N W A Y B.Sc.E, Queen's University, 1996 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE D E G R E E OF M A S T E R OF APPLIED SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Civil Engineering) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A November 2001 © Samantha E. Kenway, 2001 In presenting this thesis in partial fulfillment of the requirements for an advanced degree at the University of British Columbia, I agree that the library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the Head of the Department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Civil Engineering University of British Columbia 2324 Main Mall Vancouver, B C V6T 1W5 Date: November 2001 Abstract This research investigated the efficiency df membrane filtration technology at reducing the total trihalomethane formation potential in a British Columbia drinking water. Trihalomethanes are formed in drinking water as a result of chlorination of natural organic matter present in the source water. They are halogen substituted single-carbon compounds which are suspected human carcinogens. Chloroform, most frequently detected and at the highest concentrations in drinking water, often serves as an estimate for total trihalomethanes. In a national survey of 70 water supplies serving 38% of Canada's population, conducted in the winter of 1976/1977, chloroform concentrations 800 m downstream from point of chlorination, averaged 22.7 ug/L (ranging from 0 to 121 pg/L). The current Canadian guideline for total trihalomethanes is 100 pg/L. The United States' guideline (to be implemented January 2002) is 80 pg/L. Many water suppliers will not be able to meet this guideline. Both Canada and the US, state that the preferred method of controlling trihalomethanes is precursor removal (removal of naturally occurring organic matter), and the best method of precursor removal is organics removal. The present research evaluated the removal patterns of TOC and THMFP under different treatment conditions for Seymour Reservoir water in Vancouver, British Columbia. The TOC and THMFP removal patterns as measured for raw water, microfiltration (MF) filtered water, coagulated and M F filtered water, ultrafiltration (UF) filtered water, and powdered activated carbon (PAC) treated and UF filtered water were evaluated with respect to their removal efficacy. Additionally, the removal patterns were evaluated to determine if DBP production was primarily humic acid- or fulvic acid-controlled, and to determine the usefulness of TOC ii as a surrogate parameter. Other surrogate parameters, UV254, specific U V A (SUVA) and differential U V were also evaluated for the same waters. This research demonstrated that organic removal does not equate to a trihalomethane production reduction (estimated using chloroform production measurements). And the technologies effective at organics removal (membrane filtration technology with and with out applied pretreatments) are not always effective at removing trihalomethane formation potential. TOC removal did not equal THMFP removal. Tests demonstrated that coagulation effectively removed organics of above molecular above 10 000 Daltons but the greatest THMFP remained with organics of molecular weight <3000 Daltons. The pilot plant study also showed that TOC removal and THMFP removal are not equated. The M F membrane removed an average of 12.9% of the raw water TOC and an average of 27.2% of the raw water THMFP, while the UF membrane removed 30.9% of the raw water TOC but only 18.5% of the raw water THMFP. The water appears to be controlled by humic acids. The treatment process which removes organic material in the humic acid range (coagulation) showed the largest chloroform formation potential reductions. iii Table of Contents A B S T R A C T ii T A B L E O F C O N T E N T S iv LIST O F T A B L E S viii LIST O F FIGURES x G L O S S A R Y O F T E R M S & ABBREVIATIONS xii A C K N O W L E D G E M E N T S xiii 1.0 INTRODUCTION 1 2.0 L I T E R A T U R E R E V I E W 4 2.1 Chlorination 4 2.2 Disinfection By-Products 5 2.2.1 Types of Disinfection By-Products and Their Health Effects 5 2.2.2 Factors Controlling T H M Formation 9 2.3 Natural Organic Matter 13 2.4 Surrogate Parameters 15 2.4.1 TOC 16 iv TABLE OF CONTENTS 2.4.2 U V A and S U V A 16 2.4.3 Differential U V 17 2.5 NOM Removal 17 2.5.1 Membrane Processes 17 2.5.2 Coagulation 21 2.5.3 PAC 24 2.6 Summary 26 3.0 RESEARCH OBJECTIVES 29 3.1 General 29 3.2 Specific Tasks 30 4.0 GENERAL METHODS AND ANALYTICAL PROCEDURES 32 4.1 Sample Source and Background Information 32 4.1.1 Source Water 32 4.1.2 Pilot Scale Membrane Filtration and Pretreatment 34 4.2 Experimental Design 37 4.2.1 Characterization of Samples 37 4.2.2 Apparent Molecular Weight Distribution 38 4.2.3 Simulated and Material Specific Simulated Distribution System Tests 38 4.3 Analytical Methods 39 v TABLE OF CONTENTS 4.3.1 Chlorine Residual 39 4.3.2 Total Organic Carbon 39 4.3.3 Ultraviolet Absorbance 40 4.3.4 Trihalomethane Formation Potential 41 4.3.5 Trihalomethanes 42 4.3.6 Glassware and Reagent Preparation 43 4.4 Statistical Evaluation 44 4.4.1 Correlation Coefficients 44 4.4.2 Student's 7-tests 45 5.0 R E S U L T S AND DISCUSSION 47 5.1 Apparent Molecular Weight Characterization 47 5.1.1 Factors Affecting M W C O 47 5.1.2 M W C O Determination 49 5.1.3 A M W Characterization Results 54 5.2 SDS and MS-SDS Tests 58 5.2.1 The MS-SDS Test and Theory 58 5.2.2 Results of the SDS and MS-SDS Tests 59 5.3 T O C Removal 61 5.3.1 Stage 1 Disinfectants and Disinfectant By-Products Rule 61 5.3.2 Expected TOC Removal 61 5.3.3 Results 63 vi TABLE OF CONTENTS 5.4 Chloroform Formation Potential Removal 66 5.4.1 Stage 1 Disinfectants and Disinfectant By-Products Rule 66 5.4.2 Expected THMFP Removal 66 5.4.3 Results: THMFP Removal 67 5.4.4 Results: Chloroform Yield Reduction 69 5.5 Analysis of T H M F P Surrogates 73 5.5.1 Analysis of Surrogates: TOC 73 5.5.2 Analysis of Surrogates: UV254 and S U V A 73 5.5.3 Analysis of Surrogates: Differential U V 75 6.0 CONCLUSIONS 78 R E F E R E N C E S 80 vn List of Tables Table 2.1. List of Disinfection By-Products and Disinfection Residuals Suspected to Cause Adverse Health Effects 6 Table 2.2. Cancer Risk Classifications for Disinfectant Residuals and Disinfection By-Products 9 Table 2.3. Reduction of TOC and THMFP by Coagulation on Daytona Beach Aquifer Water 23 Table 2.4. Reduction of TOC and THMFP by Coagulation on Ilwaco Reservoir Water 23 Table 2.5. Percent Reduction of TOC and THMFP by Adsorbtion on 50 mg/L of P A C 26 Table 2.6. Comparison of the Efficiency of Alum and P A C for THMFP and DOC Removal 28 Table 4.1. Seymour Water Quality Parameters (Annual Averages 2000) 33 Table 4.2. Flux Optimization Results and Overall Water Quality Changes11' 34 Table 4.3. Membrane Manufacturer Details and Operating Conditions 37 Table 4.4. Gas Chromatograph Operating Conditions for Chloroform Analysis 43 Table 5.1. Percent Retained Results from Filtering Solution of Known Molecular Weight and Known Concentration Through the UF Membranes 53 Table 5.2. Results of a 3 ppm MW=4950 Solution Progressively Filtered Through a 10 000 M W C O UF Membrane 54 Table 5.3. Measured concentrations of TOC and THMFP for A M W samples 57 Table 5.4. Replicate Series and Statistical Analysis for Sample MJ29 §(5 mg/L coagulant and MF) 58 viii LIST OF T A B L E S Table 5.5. SDS and MS-SDS Results Showing Increased Chloroform Production for the MS-SDS Tests 59 Table 5.6. Average TOC Concentrations for Each Treatment Option 64 Table 5.7. T-test Comparison Between all M F Treatments for TOC Removal 65 Table 5.8. T-test Comparison Between all UF Treatments for TOC Removal 65 Table 5.9. Average THMFP Concentrations for Each Treatment Option 68 Table 5.10. T-test Comparison Between all M F Treatments for THMFP Removal 69 Table 5.11. T-test Comparison Between all UF Treatments for THMFP Removal 69 Table 5.12. Average Yield for Each Treatment Option 72 Table 5.13. T-test Comparison Between all M F Treatments for Chloroform Yield Reduction 72 Table 5.14. T-test Comparison Between all UF Treatments for Chloroform Yield Reductions : 72 Table 5.15. S U V A and Chloroform Formation Potential Results 75 ix List of Figures Figure 2.1. Effect of pH on T H M Formation from 1 mg/L Humic Acid (Adapted from (Brereton 1998)) 10 Figure 2.2. Effect of Chlorine Dose on T H M Formation 12 (Adapted from (Brereton 1998)) 12 Figure 2.3. Distribution of Surface Water TOC in Rivers of the United States 14 (Malcolm 1990) 14 Figure 2.4. Shape of Molecule has an Effect on Molecule Rejection (Adapted from (Baker 2000)) 19 Figure 2.5. M W C O of the Different Membrane Processes Compared to Common Particles (Courtesy of Aquasource/Denard) 20 Figure 4.1. Process flow diagrams for a) Membrane A: US Filter/Memcor, and b) Membrane B: DENARD/Aquasource 36 Figure 4.2. Examples of scatterplots showing the corresponding correlation coefficients ((Dallal 1999)) 45 Figure 5.1. Schematic Representing Membrane, Surface and Internal Fouling 48 Figure 5.2a. Molecular Weight Standards Calibration Curves (R2=1.00 for both curves) 50 Figure 5.2b. Molecular Weight Standards Calibration Curves (R2=1.00 for both curves) 51 Figure 5.3. UF Filter M W C O Tests 52 Figure 5.4. A M W Distributions of TOC and Chloroform Formation Potential for Un-pretreated (0 mg/L PAC1) and Pretreated (5 mg/L PAC1) M F Water 56 Figure 5.5. Percent TOC Removals for Each Treatment Option 64 x LIST OF FIGURES Figure 5.6. Percent THMFP Removals for Each Treatment Option 68 Figure 5.7. Percent Chloroform Yield Reductions for Each Treatment Option 71 Figure 5.8. Chloroform Formation Potential vs. UV254 76 Figure 5.9. Chloroform Formation Potential vs. S U V A 76 Figure 5.10. Chloroform Formation Potential vs. Differential U V at 254 77 Figure 5.11. Chloroform Formation Potential vs. Differential U V at 272 77 Glossary of Terms & Abbreviations AHS aquatic humic substances D-DBP Stage 1 Disinfectants/Disinfection By-Product Rule Differential U V U V A unchlorinated sample - U V A chlorinated sample DOC dissolved organic carbon G A C granular activated carbon GC gas chromatograph HAA5 five haloacetic acids M F microfiltration M W molecular weight M W C O molecular weight cut-off NF nanofiltration N O M natural organic matter P A C powdered activated carbon RO reverse osmosis S U V A specific ultraviolet absorbance (UV254/DOC) T H M trihalomethane THMFP trihalomethane formation potential TOC total organic carbon UF ultrafiltration USEPA United States Environmental Protection Agency U V A ultraviolet absorbance UV254 ultraviolet absorbance at 254 nm AUV254 differential absorbance at 254 nm AUV272 differential absorbance at 272 nm Yield chloroform produced/chlorine consumed Acknowledgements My sincere gratitude to my advisors, Dr. Eric Hall and Dr. Don Mavinic, for their knowledge, wisdom and patience. Thank you to Susan and Paula for the endless tutorials, and to my family and friends for their endless support. 1.0 INTRODUCTION As part of the Disinfection-Disinfection By-Products Rule (D-DBP), the United States Environmental Protection Agency (USEPA) is developing new regulations for disinfection by-products (DBPs). Removal of natural organic matter (NOM) to reduce trihalomethane (THM) and other DBP precursors and ultimately the total DBPs concentration, is the primary objective of this rule. Stage 1 of the rule was initiated in 1996 and established seven new standards and a treatment technique for further reduction of DBP formation by N O M removal. Included in the seven new standards are stricter guidelines for total T H M and haloacetic acid (HAA5s) concentrations, which are 0.08 mg/L and 0.06 mg/L, respectively. The total T H M guideline also contains maximum contaminant level goals of zero for bromodichloromethane and bromoform (EPA 1998). The current Canadian guideline for total trihalomethanes in drinking water is 0.1 mg/L, expressed as a running annual average of quarterly samples. The first critical compliance deadline for D-DBP is fast approaching. By January 1, 2002, all drinking water systems servicing a population > 10 000 people must be in compliance with the D-DBP Stage 1 requirements. As this deadline has been approaching, T H M control has become a very important issue. There are many effective methods of control which have been investigated, ranging from changing the point of chlorination to applying membrane filtration processes. Whatever the control method, an efficient, fast, inexpensive measurement of the T H M formation potential (THMFP), or the capacity of the organics in the system to produce T H M , needs to be available. There have been many surrogate measurements suggested and researched. The common surrogates include total organic carbon (TOC) and ultraviolet absorbance (UVA) at 1 1.0 INTRODUCTION 254 nm (UV254) (Batchelor et al. 1987, Korshin et al. 1997, Najm et al. 1994). The relationship between UV254 and THMFP has been reported to be stronger than the relationship between TOC and THMFP (Reckhow et al. 1990). As the Stage 1 rule requires TOC removal as an aide in reducing THMFP, TOC removal becomes an easy surrogate for THMFP removal. Removal patterns of TOC do not, however, always follow the removal patterns of THMFP (Batchelor et al. 1987). Since the early work of Rook (Rook 1974, 1976, 1977) and Stevens (Stevens et al. 1976), aquatic humic substances (AHS) have maintained their position as the primary suspects for DBP precursors. The chemical basis for DBP formation from the reaction of chlorine and AHS is not well understood. AHS are agreed to be aromatic polymers with large numbers of functional groups. They are subdivided into two groups, humic acids and fulvic acids. Humic acids molecular weight (MW) generally ranges from 2000 to 10 000 Daltons but can be as large as 100 000 Daltons. Fulvic acid M W ranges from 500 to 2000. The relative contribution of humic acids and fulvic acids to the DBP formation appears to be water specific (Babcock and Singer 1979, Reckhow et al. 1990). Most TOC removal or THMFP control techniques do not remove organic material in both the low humic acid M W range and the fulvic acid M W range; usually one or the other is achieved. In fact, many treatments only remove organics with MWs greater than all AHS. Therefore, depending on the qualities of the water in question and the characteristics of the employed treatment process(es), it is possible to achieve a significant TOC removal, without removing DBP precursors (Reckhow et al. 1990, Amy et al. 1992). The present research evaluated the removal patterns of TOC and THMFP under different treatment conditions for Seymour Reservoir water in Vancouver, British Columbia. The TOC and THMFP removal patterns as measured for raw water, microfiltration (MF) 2 1.0 INTRODUCTION filtered water, coagulated and M F filtered water, ultrafiltration (UF) filtered water, and powdered activated carbon (PAC) treated and UF filtered water were evaluated with respect to their removal efficacy. In addition, the removal patterns were evaluated to determine i f DBP production was primarily humic acid- or fulvic acid-controlled, and to determine the usefulness of TOC as a surrogate parameter. Other surrogate parameters, UV254, specific U V A (SUVA) and differential U V were also evaluated for the same waters. 3 2.0 LITERATURE REVIEW 2.1 Chlorination The prime purpose of disinfecting public water supplies is to prevent the spread of waterborne diseases and to inactivate pathogenic organisms. The epidemiological relationship between water and disease had been suggested as early as 1854, when John Snow and John York demonstrated that the source of the epidemic of Asiatic cholera in London was the water at the Broad Street Pump. It was further shown that a nearby ruptured sewer was contaminating the well and that the sewer serviced the home of an infected person (Sawyer et al. 1994). The relationship between water and disease was further advanced with the establishment of the theory of spontaneous generation by Louis Pasteur (1854 to 1864), and the germ theory of disease by Robert Koch in 1876. Now, 150 years later, we know that all waters support biological communities. Because some microorganisms can be responsible for public health problems, control of the biological growth in the public water supply becomes one of the most important aspects of water treatment. Chlorine is by far the most commonly used disinfectant in the drinking water industry today. Of all surface water treatment plants in the United States, chlorine is used as a pre-disinfectant in more than 63% and as a post-disinfectant in more than 67% (EPA 1999). The first recorded uses of chlorination as a continuous process in water treatment were in a small town in Belgium (early 1900s) (EPA 1999) and in England (1904) (Sawyer et al. 1994). Since the introduction of chlorine as a continuous treatment method, waterborne diseases such as cholera and typhoid have been virtually eliminated. 4 2.0 LITERATURE REVIEW In 1974, researchers in the Netherlands and the United States demonstrated that THMs are formed as a result of chlorination (Rook 1974, Stevens et al. 1976, Rook 1977, EPA 1999). THMs are formed when chlorine reacts with N O M in the source water and the distribution system. Since 1974, many other classes of DBPs (resulting from the reaction of chlorine and NOM) have been identified. The issue with the formation of DBPs is one of public health. 2.2 Disinfection By-Products 2 . 2 . 1 Types of Disinfection By-Products and Their Health Effects While chlorination is effective in controlling many organisms, it reacts with N O M to form DBPs, which are halogen-substituted by-products. There are over 500 identified DBPs; of these, 23 are known or suspected to have an adverse health effect (Table 2.1). These 23 are divided into two major subdivisions: base extractables (BE) and acid extractables (AE). The BE division contains four groups: haloacetonitriles (HAN), trihalomethanes (THMs), chlorinated ketones (CK), and a miscellaneous group. The A E division contains two groups: haloacetic acids (HAAs), and chlorinated phenols (CP)(Blau et al. 1992, EPA 1999). Four disinfection residuals have also been identified, that are suspected of having an adverse health effect (Table 2.1). 5 2.0 LITERATURE REVIEW Table 2.1. List of Disinfection By-Products and Disinfection Residuals Suspected to Cause Adverse Health Effects Disinfectant Residuals Halogenated Organic By-Products Base or Acid Extractable Free Chlorine Trihalomethanes BE Hypochlorous acid Chloroform Hypochlorite ion Bromoform Dibromochloromethane Chloramines Bromodichloromethane Monochloramine Chlorine dioxide Haloacetic acids A E Monochloroacetic acid Dichloroacetic acid Trichloroacetic acid Monobromoacetic acid Dibromoacetic acid Haloacetonitriles BE Dichloroacteonitrile Bromochloroacetonitrile Dibromoacetonitrile Trichloroacetonitrile Haloketones B E 1,1 -Dichloropropanone 1,1,1 -Trichloropropanone Chlorophenols A E 2-Chlorophenol 2,4-Dihlorophenol 2,4,6-Trichlorophenol Chloropicrin B E Chloral hydrate Cyanogen chloride N-Organochlorides M X (3 -chloro-4-(dichloromethyl)-5 -hydroxy-2(5 H)-furanone) Adapted from (EPA 1999) 6 2.0 LITERATURE REVIEW Of the 6 chemical groups listed in Table 2.1, THMs are the most common and generally comprise 40% to 98% of the total DBP after chlorination of most surface waters (Amy et al. 1992, Blau et al. 1992). THMs were the first of the DBP discovered in 1974. In 1975, the USEPA presented the results of the National Organics Reconnaissance Survey (NORS), which showed that THMs could be found in almost every disinfected water at concentrations of up to 700 ug/L. Of the four THMs, chloroform is the most common, since the formation of the remaining three depend on bromide being present in the source water or distribution system. Chloroform (CHCI3) is a colorless, volatile liquid that is nonflammable. It is slightly soluble in water and is miscible with oils, ethanol, ether, and other organic solvents. Chloroform has a pleasant, non-irritating odor. It is unstable when exposed to air, light, and/or heat, which cause it to break down to phosgene, hydrochloric acid, and chlorine (Rayner-Canham et al. 1989). Chloroform is used primarily in the production of fluorocarbon-22. Fluorocarbon-22 is used in the production of fluoropolymers. Miscellaneous uses of chloroform (4%) include: as a solvent in the extraction and purification of some antibiotics, alkaloids, vitamins, and flavors; as a solvent for lacquers, floor polishes, artificial silk manufacture, resins, fats, greases, gums, waxes, adhesives, oils, and rubber; as an industrial solvent in photography and dry cleaning; as a heat transfer medium in fire extinguishers; and as an intermediate in the preparation of dyes. At least one grain fumigant mixture had contained chloroform with carbon disulfide. Chloroform, formulated with other ingredients, is used to control screwworm in animals. Its use as an anesthetic has been largely discontinued because it was linked (in the 1920s) to adverse effects such as liver failure, cardiac arrhythmia and ventricular fibrillation (NTP 2001 , Brereton 1998). 7 2.0 LITERATURE REVIEW The health effects of DBPs and disinfectants are generally evaluated with epidemiological studies and toxicology studies using laboratory animals. Results from toxicology studies have shown chloroform, bromoform, bromodichloromethane and - dichloroacteic acid to be carcinogenic in laboratory animals; and bromodichloromethane and the HAA5s cause reproductive and developmental effects in laboratory animals (NCI 1976, Jorgensen et al. 1985, EPA 1998). Epidemiological studies have also suggested the consumption of chlorinated surface water has a weak association to certain cancers (colon, bladder) or reproductive effects (spontaneous abortion or still births) or developmental effects (low birth weight)(EPA 1998, Dodds 1999). Table 2.2 indicates cancer risk classifications of both disinfectant residuals and DBPs as of January 1999. 8 2.0 LITERATURE REVIEW Table 2.2. Cancer Risk Classifications for Disinfectant Residuals and Disinfection By-Products Contaminant Cancer Classification* Hypochlorous acid Hypochlorite ion Monochloramine -Chlorine dioxide D Chloroform B2 Bromoform B2 Dibromochloromethane C Bromodichloromethane B2 Monochloroacetic acid -Dichloroacetic acid B2 Trichloroacetic acid C Dichloroacteonitrile C Bromchloroacetonitrile -Dibromoacetonitrile C Trichloroacetonitrile 1,1-Dichloropropanone 1,1,1 -Trichloropropanone -2-Chlorophenol D 2,4-Dihlorophenol D 2,4,6-Trichlorophenol B2 Chloropicrin -Chloral hydrate C Cyanogen chloride * GROUP A t Sufficient evidence in epidemiological studies to support casual association between exposure and cancer ' ' • GROUPB (.Limited evidence inepidemiologicalstadies (Bl) ancte GROUP C , Limited evidence from animal studies and inadequate or no data m humans > < GROUP D ' Inadequate or no human and animal evidence of carcinogenicity From (EPA 1999) 2.2.2 Factors Controlling T H M Formation Factors which control the rate of DBP formation include (Garcia-Villanova et al. 1997, Rebhun et al. 1997, Brereton 1998): • pH, 9 2.0 LITERATURE REVIEW • temperature, • chlorine dose, • chlorine residual, • reaction time, • bromide concentration, and • TOC quality and quantity. 2.2.2.1 pH Increased pH values produce increased T H M formation. T H M formation increases have been reported to range from 50% (Figure 2.1) to 100% (Garcia-Villanova et al. 1997) per unit of pH change. The lower the pH, the more of the non-ionized, more reactive HCIO form of hypochlorous acid is found, thus increasing the reaction rate with chlorine. Time (hours) Figure 2.1. Effect of pH on T H M Formation from 1 mg/L Humic Acid (Adapted from (Brereton 1998)) 10 2.0 LITERATURE REVIEW 2.2.2.2 Temperature Studies have shown an Arrhenius-type relationship between the T H M formation reaction rate and temperature (Rayner-Canham et al. 1989, Garcia-Villanova et al. 1997). Therefore, an increasing T H M formation rate would be expected to accompany increasing temperature. 2.2.2.3 Chlorine Dose and Chlorine Residual T H M formation is strongly dependent on chlorine concentration. Data show that T H M formation increases linearly with applied chlorine dose to a certain chlorine concentration (water specific), at which point T H M formation plateaus (Figure 2.2), and an increased applied chlorine dose does not show a significant T H M increase (Rebhun et al. 1997, Brereton 1998). Chlorine residual also affects T H M formation. Most researchers have found a linear relationship between chlorine consumption (applied chlorine - chlorine residual) and T H M formation with a reaction order greater than or equal to unity (Garcia-Villanova et al. 1997). 11 2.0 LITERATURE REVIEW Applied Chlorine Dose (mg/L) Figure 2.2. Effect of Chlorine Dose on T H M Formation (Adapted from (Brereton 1998)) 2.2.2.4 Reaction time The T H M reaction is controlled by slow reaction kinetics. In most waters, T H M production is rapid for the first 5 to 20 hours, after which time the production begins to level off. Whether production completely plateaus, or continues at a slower formation rate, is water dependent. (Rebhun et al. 1997) showed that an 18 mg/L dose incubated for 1 day produced 302 pg/L THMs, while an 18 mg/L dose incubated for 7 days produced 418 pg/L, a 116 pg/L or 38% increase in THMs. They also reported that the average T H M formation in the first hour after chlorine addition, for all applied chlorine doses, was 80% of the total (7 day) T H M formation. 12 2.0 LITERATURE REVIEW 2.2.2.5 Bromide concentration Bromide concentration affects both the rate of formation and T H M yield. During chlorination, bromide is oxidized to bromine which is more reactive than chlorine (Stevens et al. 1976, Greiner et al. 1992, Wells and Chadik 1998). Rebhun et al. (1997) reported that the THM/dissolved organic carbon (DOC) ratio for a bromide-free water ranged from 2 to 9.6 pg/mg, while the THM/DOC for a bromide containing water ranged from 19 to 45 pg/mg. (Wells and Chadik 1998) reported that the formation of brominated DBPs increases with increasing Br" concentration and contact time. 2.2.2.6 TOC Since THMs are a result of the reaction between N O M and chlorine, it follows that the quality and quantity of N O M in the source water would directly affect the reaction rate. The formation of T H M increases strongly with increasing amounts of N O M following a first order reaction (Garcia-Villanova et al. 1997). It has been shown that chlorine will react preferentially with AHS (see Section 2.3), which comprise 50% to 75% of the N O M in surface water. 2.3 Natural Organic Matter Many researchers have documented that N O M is the principal precursor of organic DBP formation ((EPA 1999). Of the total N O M , it is considered that AHS are the primary T H M precursors. AHS constitute 50% to 75% of N O M and are the largest fraction of natural organic matter in water (Figure 2.3). The concentration of humic substances in mountain 13 2.0 LITERATURE REVIEW streams, melt waters and large streams (ex: Seymour water shed) varies from 0.05 mg/L to 4 mg/L (Thurman 1985). AHS are non-volatile, coloured, polyelectric, organic acids and generally range in molecular weight from 500 to >10 000 Daltons (have been noted to weigh up to 100 000). Their elemental composition is approximately (Thurman 1985): • 50% carbon, • to 5% hydrogen, • 35 to 40% oxygen, • 1 to 2% nitrogen, • <1% sulfur and phosphorus. low-molecular weight acids 25% humic acids bases f 45% hydrophilic neutrals 15% hydrophobic neutrals 6% 5% Figure 2.3. Distribution of Surface Water TOC in Rivers of the United States (Malcolm 1990) 14 2.0 LITERATURE REVIEW AHS are polymeric, with the aromatic ring as the monomer (Amy et al. 1992). Functional groups include carboxylic acids, phenol hydroxyl, carbonyl, and hydroxyl (Thurman 1985). AHS are subdivided into two groups: humic acids and fulvic acids. Humic acid is the hydrophobic fraction that precipitates at pH 2.0 or less, and the hydrophilic fulvic acid always remains in solution. The humic acids tend to be larger in molecular weight (2000 to >10 000 Daltons) than the fulvic acids (500 to 2000 Daltons) and are often colloidal. Natural waters with a high AHS concentration have a high U V A , because the double bonds contained in the aromatic rings strongly absorb U V light at 254 nm. The S U V A , defined as the ratio of UV254 to TOC, is an indicator of the proportion of the humic substances in a water (Urfer et al. 1999). Water sources with a high amount of AHS typically have S U V A values >4 L/mgm (absorbance (1/m) over DOC (mg/L) greater than 4) (Vrijenhoek et al. 1998, Urfer et al. 1999). The difference in SUVA, before and after treatment or chlorination is a good measure of the amount of AHS removed during treatment or destroyed during chlorination. It is often used as a surrogate measurement for AHS concentration and/or THMFP (Najm et al. 1994). 2.4 Surrogate Parameters Control of DBPs is a major concern at any water treatment plant. There are several plant operational methods that can reduce T H M formation, such as the location of chlorine addition in the process and the use of pretreatment. Effective and cost efficient execution of these measures requires easy measurements of the DBP precursors or AHS in the water to be treated. The measurement of THMs requires the use of a gas chromatograph (GC), a trained 15 2.0 LITERATURE REVIEW technician to operate the GC and several days of incubation time. These factors make the T H M test expensive and time consuming. The direct measurement of AHS can also be quite complicated due to its heterogeneous nature. There are several available methods but they are slow and tedious or involve expensive equipment and a trained technician. A need for easy, inexpensive and reliable measurement has led to the use of surrogate parameters. Although surrogate parameters have limitations, they are used because they can be measured rapidly and inexpensively (EPA 1999). 2.4.1 TOC Knowing that AHS comprise 50 to 75% of N O M in surface waters, N O M is a viable surrogate parameter. N O M can be measured indirectly through TOC (Vickers et al. 1995). T H M formation models, based solely on TOC concentrations tend to have poor predictability (Brereton 1998). In addition, removal patterns of TOC do not follow the removal patterns of T H M (Batchelor et al. 1987). 2.4.2 UVA and SUVA The correlation between absorbance at UV254 and DBP formation has been well documented (Najm et al. 1994, Korshin et al. 1997, Vrijenhoek et al. 1998). The premise for the correlation is that, since the aromaticity of N O M is related to both U V absorbance and DBP formation, then absorbance and DBP formation should be related also ((Korshin et al. 1997)). As more aromatic molecules react with chlorine, fewer molecules are available in the sample to absorb U V (aromatic molecules absorb U V light preferentially at X,=254 nm). Therefore, UV254 and S U V A (UV254/TOC) become useful predictors for the DBP formation potential of a water (Edzwald et al. 1985, Korshin et al. 1997). Previous 16 2 . 0 L I T E R A T U R E REVIEW researchers have found linear relationships between T H M formation and UV254, with r 2 values equal to 0.95 (Korshin et al. 1997) and 0.93 (Najm et al. 1994). Korshin et al. (1997) showed that waters with higher SUVAs contained a high humic fraction in their N O M . 2.4.3 Differential UV As discussed above, the aromaticity of N O M is related to both U V absorbance and DBP formation. Aromatic molecules react with chlorine, and strongly absorb U V light. The production of DBP involves the destruction of aromatic molecules, leaving fewer aromatic molecules available in the sample to absorb U V . Korshin et al. (1997) and L i et al. (1998) suggested that differential U V (AUV), or the change in U V as a result of chlorination, may be a useful surrogate parameter for DBP formation potential because A U V represents the destruction of aromatic molecules to form DBP precursors. They have reported very strong correlations, r2=0.95 for AUV254 vs. chloroform formation (Korshin et al. 1997) and r2=0.99 for AUV272 vs. TOX formation (Korshin et al. 1997, L i et al. 1998). 2.5 NOM Removal 2.5.1 Membrane Processes Membrane technology for the treatment of water has become increasingly popular in recent years. Reasons for this advancement include: 1) the advent of more stringent regulations, 2) declining and deteriorating water resources, 3) the emphasis on water reuse, 4) land/space restrictions, and 5) advances in membrane research and technology (Pirbazari et al. 1992). 17 2.0 LITERATURE REVIEW Membrane filtration can be defined as a separation process, which (Ballew 1978) utilizes a device, or combination of devices, that can separate a mixture into two or more fractions, each of which contains different relative proportions of the components present in the original mixture. The membrane device is a discrete thin interface that moderates the permeation of chemical species in contact with it (Baker 2000). Membranes are characterized by their molecular weight cutoff (MWCO), a loosely defined term generally taken to mean the molecular weight of the globular protein molecule that is 90% retained or rejected by the membrane. One of the limitations of the M W C O classification system is that, outside of laboratory scale experiments, molecule shape is not controlled. Linear, water-soluble molecules will have a much lower rejection than a globular molecule of the same molecular weight (Figure 2.4). 18 2.0 LITERATURE REVIEW 1 Skin of UF Membrant Porous Substrate Globular Proteins Linear Polymer Pepsin Cytochrome C Polydextran Molecular Weight (1000 Daltons) 35 13 100 Rejection (%) 90 70 0 Figure 2.4. Shape of Molecule has an Effect on Molecule Rejection (Adapted from (Baker 2000)) Membranes are used to separate particles and molecules smaller than about 10 u.m. Microfilters retain particles and bacteria in the size range of 10 p.m to 0.02 pm. Ultrafiltration retains colloidal material and molecules with a molecular weight greater than 1000 Daltons. It can effectively remove material of size 0.02 ixm to 0.001 itm. Reverse osmosis filters are the most selective and can retain all suspended and most dissolved species. Only water and some small molecules (<200 Daltons) are able to pass through, since the reverse osmosis (RO) membrane can retain matter between 0.001 p:m and 0.0001 p.m (Ballew 1978). 19 2.0 LITERATURE REVIEW Nanofiltration (NF) operates between UF and RO. Figure 2.5 shows some common separation processes plotted against the size of the particle the process is capable of separating from the original solution and the property of the particle that allows removal by the process. M EM B R A N E C U T O F F C o l i o i d s O r g a n i c c o m p o u n d s A l g a e O r g a n i c m a c r o m o le c u le s D i s s o l v e d s a I t s B a c t e r i a P o l l e n s Y e a s t s H h H 1 10 0 p m 10 0.1 0.01 0.00 1 0.000 1 H a i r G i a r d i a C r y p t o s p o r i d i u m S m a He s t b a c t e r ia P o l i o v i r u s R e v e r s e o s m o s i s N a n o f i l t r a t i o n U I t r a f i l t r a t i o n M ic r o f i It r a t to n S a n d f Mt r a t i o n R e m o v e d by A q u a s o u r c e ultraf i l trat ion m e m b r a n e s Figure 2.5. M W C O of the Different Membrane Processes Compared to Common Particles (Courtesy of Aquasource/Denard) Membranes may be homogenous in composition and structure (isotropic) or may be chemically and physically heterogeneous (anisotropic). In addition, the membrane may be microporous (randomly distributed, interconnected pores) or dense (nonporous), uniform (one material) or composite (layers of different materials), or electrically charged (Baker 2000). 20 2.0 LITERATURE REVIEW Many studies (Taylor et al. 1987, Tan and Amy 1991, Blau et al. 1992, Jacangelo et al. 1995) have shown that effective N O M and DBP precursor reductions can be achieved by some of the membrane separation processes. While NF has shown much potential, UF and M F have shown only limited N O M and DBP removal without pretreatment. Siddiqui et al. (2000) studied NF, UF and MF units on the same waters. They reported TOC removals for the NF, UF and M F units as 84 to 98%, 25 to 32% and 18%, respectively; THMFP removals were reported to be 95 to 99%, 50% and virtually zero, respectively. Other researchers have reported approximate TOC and T H M removals for both M F and UF units to be 5 to 70% and 10%, respectively (Vickers et al. 1995). Low T H M removal can be attributed to the solute MWs and the membrane MWCOs. The M W of AHS ranges from 500 to 2000 Daltons for fulvic acids and 2000 to 10 000 Daltons for humic acids. The MWCOs of UF units are approximately 10 000 Daltons and greater; and the M W C O of MF units are in excess of 1 000 000 Daltons. Therefore, it is expected that humic and fulvic acids will not be retained by the membrane and there will not be a related THMFP reduction. 2.5.2 Coagulation Water utilities first began the practice of coagulation in efforts to improve particle removal. From this they discovered that coagulation improved post-filtration disinfection efficacy and, at a higher dosage or adjusted pH, coagulation was also effective for colour removal. The introduction of T H M control regulations in 1979, resulted in an increased interest in N O M removal, especially by coagulation (Jacangelo et al. 1995). Coagulation is effective for removing N O M and in fact, enhanced coagulation is the USEPA's recommended TOC removal mechanism (EPA 1998). 21 2.0 LITERATURE REVIEW As summarized by (Crozes et al. 1995, Jacangelo et al. 1995, Urfer et al. 1999), the major mechanisms by which N O M is removed by coagulation are: • colloid destabilization through charge neutralization, electrical double layer compression, enmeshment or bridging which allows particulate and colloidal removal; • precipitation as aluminum or iron humates or fulvates which allows removal of dissolved matter; and • coprecipitation by adsorption or occlusion on the metal hydroxide, which also allows for removal of dissolved matter. The degree of N O M removal depends on the type of coagulant employed, coagulant dose and pH. One coagulant may be significantly more effective than another for a given water. There have been several reports of iron salts outperforming alum (Crozes et al. 1995, Jacangelo et al. 1995). Polymers are not expected to perform as well as metal-salt coagulants as they can remove particulate and colloidal N O M (charge neutralization) but not dissolved N O M (have no sites for co-precipitation). The coagulant pH affects both the inorganic coagulating species and the N O M . The coagulating species are more positively charged and the N O M is more protonated at a lower pH (Crozes et al. 1995). Therefore, coagulation becomes more favourable and the coagulant demand decreases. Maximum N O M removal generally occurs at pH values between 5 and 6 (Sawyer et al. 1994, Jacangelo et al. 1995, Vickers et al. 1995). Coagulation tends to remove higher molecular weight molecules before the smaller ones. A water containing a high M W humic acid (5000 to 10 000 Daltons) would be a good candidate for chemical coagulation. Low M W fulvic acids are hydrophilic and not amenable to coagulation or adsorption (Amy et al. 1992, Vrijenhoek et al. 1998). Methods of TOC 22 2.0 LITERATURE REVIEW removal remove mainly particulate matter, colloidal matter and humic acids in the 5000 to 10 000 Dalton range. THMFP removal varies depending on the relative amounts of humic and fulvic acids. TOC removal patterns do not follow THMFP removal patterns, as shown in Tables 2.3 and 2.4. TOC removal will vary with the concentration of particulate and colloid matter present in the water, while the T H M removal will vary with the relative amount of humic acid with M W >5000 Daltons initially present. Table 2.3. Reduction of TOC and THMFP by Coagulation on Daytona Beach Aquifer Water Coagulant type Al Al A l Fe Fe Coagulant dose 10 20 30 „ 15 30 (mg/L) > ' TOC reduction (%) 25 29 •51 17 27 THM Reduction (%) •24 37 43 18 40 (Adapted from (Najm et al. 1991)) Table 2.4. Reduction of TOC and THMFP by Coagulation on I Coagulant type Al Al Fe Fe Fe Coagulant dose 1.25 2.5: ? ' 2.5 5 7.5 (mg/L) TOC reduction (%) 34 59 14 63 72 THM Reduction 47 66 ; 18 68 77 (%) waco Reservoir Water (Adapted from (Najm et al. 1991)) 23 2.0 LITERATURE REVIEW 2.5.3 PAC Adsorption on porous carbon was described as early as 1550 BC, in an ancient Egyptian papyrus and later by Hippocrates and Pliny the Elder, for therapeutic purposes. In the 18 th century, carbons made from blood, wood and animals were used for purifying liquids. Bone char was used for colour removal in the sugar industry in the 19 th century. During World War I (WWI), coconut char was used in American soldier's gas masks. After World War II (WWII), coal-based activated carbons were developed, as the ancestors of the carbons we use today (Calgon 2001, U.S. Army Corps of Engineers 2001). The adsorption process is based on the adsorbate molecule being concentrated on the surface of activated carbon. The adsorbate molecule is attracted to the activated carbon by Van der Waal's forces. Van der Waals' forces are additive, so the addition of each adsorbate molecule to the surface of the activated carbon molecule increases the attractive force. Typically 65% to 95% of commercially available PAC passes through a 325-mesh (44 pm) sieve; by contrast, the average particle size of granular activated carbon (GAC) ranges from 0.2 mm to 0.3 mm (Kassam et al. 1991, Najm et al. 1991). PAC is manufactured from many materials including wood, lignite and coal. The manufacturing process involves two phases. The first phase, carbonation, involves drying and heating to separate the raw product. The second phase, activation, involves exposing the carbon material to an activating agent such as steam. The duration of the activation phase determines the number and size of the pores formed on the carbon (U.S. Army Corps of Engineers 2001). The porosity and pore size will control the available surface area for adsorption. The greater the surface area, the greater the adsorption capacity. The apparent density of PAC ranges from 0.36 to 0.74 g/cm3 (U.S. Army Corps of Engineers 2001). 24 2.0 LITERATURE REVIEW The efficiency of PAC at removing the TOC present in most natural waters has not been reported to be very high (Najm et al. 1991), possibly because removal is limited by slow kinetics of adsorption and equilibrium. Larger M W organics have a slower rate of diffusion into the pores of the PAC particle and may be impeded by size exclusion phenomena, in which large molecules are unable to enter the small pores of the carbon. For example: a rhodamine B dye (MW=422) will take 5 hours to come to equilibrium with a PAC; a fulvic acid (MW=10 000) will take 17 hours to come to equilibrium with the same PAC; and a humic acid (MW=50 000) will take over 2 days (Najm et al. 1991). If sufficient contact time is not available, a larger PAC dose can be used to achieve a greater reduction, but there exists an upper boundary on the removal, dependent on the PAC and TOC (American Water Works Association 1990). Typical TOC adsorption capacties for surface water are 10 to 50 mg/g (American Water Works Association 1990); i f the PAC dose is 25 mg/L and the equilibrium capacity 50 mg/g, theoretically only 0.25 mg/L of TOC will be removed. Another factor which affects PAC performance is the quality of the mixing and contact between the PAC and water. The PAC must be added at a point and in a manner which ensure its contact with all the water. PAC particle size will also affect the TOC removal. (Najm et al. 1991) reported that, for a 15 minute contact time, a 500 pg/L solution of trichlorophenol was reduced to 25 pg/L by 14 pm carbon, but was only reduced to 275 pg/L by 100 pm carbon. Equilibrium was not achieved in the 15 minutes in either case. Although PAC is not always successful at reducing overall TOC concentrations, it has been shown to be effective for removing the smaller sized N O M (Farahbakhsh and Smith 2001). The removal of the smaller sized N O M can be effective for T H M removal i f the water contains a significant amount of fulvic acids. TOC removal patterns are not equivalent to THMFP removal patterns, as shown in Table 2.5. 25 2.0 LITERATURE REVIEW Table 2.5. Percent Reduction of TOC and THMFP by Adsorbtion on 50 mg/L of PAC Carbon A Carbon B TOC T H M TOC T H M Daytona Beach Aquifer Water 13 35 21 29 Ilwaco Reservoir Water 12 45 13 36 (Adapted from (Najm et al. 1991)) 2.6 Summary Chlorine is added to water as a disinfectant to protect human health. Halogenated DBPs are formed when N O M reacts with free chlorine or bromine in the presence of free chlorine. THMs are the most common and well studied of the over 500 identified DBPs. THMs make up over 50% of total DBPs in most water samples (Li et al. 1998). The organic compounds which act as DBP precursors are called AHS. AHS comprise two groups, humic acids and fulvic acids. Humic acids are larger (2000 to >10 000 Daltons), hydrophobic compounds which precipitate at a pH <2, while fulvic acids are smaller (500 to 2000 Daltons), hydrophilic and do not precipitate at any pH. Both humic acids and fulvic acids will contribute to the THMFP. A water will tend to contain primarily either humic acids or fulvic acids (Babcock and Singer 1979, Reckhow et al. 1990). For a humic acid-controlled water, treatment techniques which primarily remove molecules in the M W range of humic acids would be expected to remove the greatest THMFP. For a fulvic acid-controlled water, the treatment techniques which primarily remove molecules in the M W range of fulvic acids would be expected to remove the greatest THMFP. The M W C O for M F units are reported to be greater than > 1 000 000 Daltons, 26 2.0 LITERATURE REVIEW while the M W C O for UF units are reported to be in the range of 10 000 to 100 000. One would expect UF to make a minimal contribution to THMFP removal by removing any humic acids > 10 000 Daltons in weight, and one would expect MF to have a negligible effect on THMFP. Coagulation has been reported to remove organics in the M W range of humic acids, 5000 to 10 000. PAC has been reported to remove organics in the range of fulvic acids and light end humic acids, 1000 to 5000. Amy et al. (1992) showed, in a study of 8 waters whose average AHS MWs ranged from 600 to 15 000 Daltons, that alum removed a greater portion of the THMFP than PAC, in 7 out of 8 cases (see Table 2.6). Since TOC and T H M removal are so dependent on the quality of the water, it is very difficult to predict removals (Batchelor et al. 1987, Amy et al. 1992, Greiner et al. 1992, Brereton 1998). Removal efficiencies and the primary treatments used to achieve the removals need to be evaluated on a water to water basis. 27 2.0 LITERATURE REVIEW Table 2.6. Comparison of the Efficiency of Alum and PAC for THMFP and DOC Removal Average MW of Applied Treatment DOC Removal THMFP Removal Sample (Daltons) (%) (%) 8800 PAC 50 mg/L 13 2 Alum 5 mg/L 42 32 600 PAC 50 mg/L 32 12 Alum 5 mg/L 25 38 3600 PAC 50 mg/L 38 54 Alum 5 mg/L 39 35 9300 PAC 50 mg/L 29 38 Alum 5 mg/L 57 48 12 000 PAC 50 mg/L 45 11 Alum 5 mg/L 54 32 900 PAC 50 mg/L 39 20 Alum 5 mg/L 53 36 15 000 PAC 50 mg/L 28 14 Alum 5 mg/L 35 24 9400 PAC 50 mg/L 51 17 Alum 5 mg/L 57 19 (Adapted from (Amy et al. 1992)) 28 3.0 RESEARCH OBJECTIVES 3.1 General THMFP removal becomes increasingly important as the potential health effects of DBPs are realized, and as more stringent regulations are imposed in response to these realizations. It is currently an industry perception that TOC removal correlates more or less directly to THMFP removal and therefore TOC removal will sufficiently control THMFP. The present research evaluated the removal of TOC and THMFP for Seymour Reservoir, British Columbia water under the following treatment approaches: 1) raw water, no treatment, 2) MF-filtered water, 3) coagulated and MF-filtered water, 4) UF-filtered water and 5) PAC treated and MF-filtered water. The primary goal of the research was to show that the TOC removal and the THMFP removal do not follow a similar pattern and therefore could not be correlated. The specific objective was to determine i f T H M production was primarily humic acid- or fulvic acid-controlled. A TOC removal technique would be expected to remove organics in a specific M W range, which may not correspond to the M W range of the DBP precursors. As such, not all TOC treatment methods would be expected to produce a THMFP removal. Each of the evaluated treatment techniques was expected to remove a specific M W fraction of the total organics (humic acids or fulvic acids). The comparison of treatment techniques and THMFP allowed the determination of whether the DBP formation was controlled primarily by humic or fulvic acids and therefore, which treatment method would be effective at removing THMFP. 29 3.0 RESEARCH OBJECTIVES Additionally, the usefulness of TOC, UV254, S U V A and differential U V as surrogate parameters were also evaluated for the same waters. 3.2 Specific Tasks The major objectives of this study were to determine i f the removal of humic acids or fulvic acids would be more effective for DBP formation control and to evaluate several characteristic parameters of the effluent as THMFP surrogates. The specific tasks required to meet the objectives were identified as follows: • test the chloroform formation potential of bulk waters from different treatment processes under exposure to chlorine; • characterize the different bulk waters using an apparent molecular weight (AMW) UF characterization technique; • compare A M W characterizations with the applied treatment process to determine the M W fraction of organics, and thus THMFP, removed by the process; • compare TOC removals and THMFP removal for each of the bulk waters to determine the usefulness of TOC as a surrogate; • compare UV254 and THMFP potential for each of the bulk waters to determine the usefulness of UV254 as a surrogate; • compare S U V A and THMFP potential for each of the bulk waters to determine the usefulness of S U V A as a surrogate; 30 3.0 RESEARCH OBJECTIVES compare differential U V and THMFP potential for each of the bulk waters to determine the usefulness of differential U V as a surrogate; and confirm the effect of biofilm on chloroform formation as previously reported by Chan (2000). 31 4.0 GENERAL METHODS AND ANALYTICAL PROCEDURES This chapter describes the experimental methods, the sampling methods and experimental procedures applied in the study. For the most part, standard procedures were followed in accordance with Standard Methods (APHA et al. 1995). Any deviations from the Standard Methods are described. 4.1 Sample Source and Background Information 4.1.1 Source Water The Greater Vancouver Regional District supplies water to 2 million people in the city of Vancouver and surrounding cities. The water is supplied from three protected wilderness watersheds (Capilano, Coquitlam and Seymour) comprised of 586 square kilometers. The watersheds remain one of the last, highly-protected watershed areas in North America, meaning that the area is protected from the public and is not shared with any industry, agricultural or commercial enterprises. This eliminates pollution risks posed by things such as septic fields, fertilization, transportation corridors and logging. The water travels through 550 km of supply mains to 18 municipalities for distribution (GVRD 1995). Each reservoir is designed to supply up to 1.2 billion litres/day at peak demand. The water used in the present study was sourced from the Seymour reservoir. Seymour provides 40% of the GVRD's total water supply. The size of the watershed is 18000 hectares. The area of the main reservoir is 262 hectares with a depth of 19 metres, the reservoir volume, when full, is 30 billion litres. The watershed is located in a typical coastal 32 4.0 METHODS AND PROCEDURES mountain area, receiving rain and snow (GVRD 2000). Although there has been no disease outbreak associated with the Seymour water supply, the water continues to fail the Guidelines for Canadian Drinking Water Quality (Health and Welfare Canada 1993) for turbidity levels. Heavy rainfall in the watershed leads to erosion and landslides, which cause elevated turbidity in the reservoir basin. The Seymour source water exceeds the Canadian turbidity guideline of 5 N T U (Health and Welfare Canada 1993) an average of 18 days per year (GVRD 2000). The water is naturally soft and low in pH, alkalinity, colour, organic carbon and turbidity (Table 4.1). Table 4.1. Seymour Water Quality Parameters (Annual Averages 2000) Parameter (mg/L unless shown) Untreated Seymour Water Source Alkalinity as C a C 0 3 3.4 Dissolved Organic Carbon 1.6 Total Organic Carbon 1.7 True Colour (cu) 13 Conductivity (umhos/cm) 14 Hardness as CaC03 5.03 Iron 0.07 (dissolved) / 0.18 (total) Manganese <0.01 (dissolved) / 0.02 (total) Nitrogen - Nitrate as N 0.09 pH (pH units) 6.5 Total phosphorus <0.005 Residue 16 (dissolved) /18 (total) Sulphate 1.5 UV254 (abs/cm) 0.064 Total sodium 0.52 Temperature (°C) 3 to 17 Turbidity (NTU) 0.17 to 4.5 33 4.0 METHODS AND PROCEDURES 4.1.2 Pilot Scale Membrane Filtration and Pretreatment The Seymour water used in this study was collected from a pilot scale water treatment facility located at the base of the Seymour Falls Dam. The facility was first commissioned to investigate filtration options for the future Seymour filtration plant (Chan 2000). The study being conducted at the time of this research was titled, * Investigation of Ultrafiltration Performance on BC Drinking Water Quality. Membrane pilot units were provided by manufacturers and were operated under varying operating conditions and in conjunction with different pretreatment processes (Table 4.2). Table 4.2. Flux Optimization Results and Overall Water Quality Changes" Parameter Membrane A § Membrane Flux Optimization Flux (L/m2.hr) Pressure Increase (kPa/hr) Runtime Estimation (days) Maximum allowable TMP (kPa) Overall Water Oualitv Turbidity Removal (%) U V A Reduction (%) TOC Reduction (%) 110 145 185 0.028 0.101 0.300 245 60 18 205 205 205 95.6±1.9% 13.9%±3.9% 6.6%±1.4% 110 150 190' 0.048 0.051 0.084 54 45 26 97 97 97 94.1%±1.3% 25.7%±6.9% 16.3%±2.1% § US Filter/Memcor f DENARD/Aquasource JSource: (Farahbakhsh and Smith 2001) The research detailed in this paper evaluated the effluent from two membrane units. A summary of the two units is provided in Table 4.3. Process train schematics for both membranes are included in Figure 4.1. Each of the units was operated alone (without pretreatment) and in conjunction with a pretreatment (polyaluminum chloride or powdered activated carbon). Samples were collected from: 34 4.0 METHODS AND PROCEDURES the raw water, membrane filtered water (from Membrane A and Membrane B) with no pretreatment, water pretreated with polyaluminum chloride (2, 4, 8 mg/L dose) and subsequently filtered through Membrane A , and water pretreated with powdered activated carbon (5, 10, 15 , 20 mg/L dose) and subsequently filtered through Membrane B. 35 4.0 METHODS AND PROCEDURES a) Backwash Air Supply PreFilter Screens Raw Water To Waste Backwash Filter Tank Permeate b) Purge Outlet BackwQshwater Water Outlet Treated Water Overflow Chlorine Backwash Purnp Figure 4.1. Process flow diagrams for a) Membrane A: US Filter/Memcor, and b) Membrane B: DENARD/Aquasource 36 4.0 METHODS AND PROCEDURES Table 4.3. Membrane Manufacturer Details and Operating Conditions Parameter Membrane A § Membrane Membrane Process microfiltration ultrafiltration Molecular Weight Cutoff >1 000 000 10 000 to 100 000 (Daltons) Pore Size (pm) 0.2 0.01 Active Area (m2) 45 7.2 Membrane Configuration pressure type pressure type Hydrophobicity hydrophilic hydrophilic Operating Pressure (kPa) 103 to 240 47 to 150 Flow Direction outside in inside out Mode of Operation dead-end dead-end/crossflow Co-operated Pretreatment polyaluminum chloride powdered Activated Carbon 4.2 Experimental Design 4.2.1 Characterization of Samples Each water sample (raw and treated) was analyzed for the following experimental parameters (see Section 4.3 for methodology): • TOC, • U V absorbance (200 through 400 or 600 nm), and • Seven day chloroform formation potential. 37 4.0 METHODS AND PROCEDURES 4.2.2 Apparent Molecular Weight Distribution Selected samples were also characterized for apparent molecular weight distribution (AMW), determined by laboratory scale ultrafiltration. The technique described in Amy et al. (1987), Amy et al. (1992) and Laine et al. (1989), was used to process samples (already filtered through a 0.45 pm membrane filters) through a series of 200 mL stirred cells (Amicon magnetic stirring table Model MT2, Amicon stirred cell Model 202, Amicon stainless steel reservoir Model RS4), with UF membranes having molecular weight cutoffs (MWCOs) of 1000, 5000, 10 000 and 30 000 (Amicon Y M series Diaflo membranes). The technique involves a configuration of feed/influent (sample water), permeate/effluent (corresponding molecular weight fraction) and a filtrate/retentate. This parallel processing approach produces a series of corresponding A M W permeates. The permeates were analyzed for TOC, U V absorbance (200 to 400 nm) and seven day chloroform formation potential. 4.2.3 Simulated and Material Specific Simulated Distribution System Tests On two sampling dates, samples were recovered for simulated distribution (SDS) and material specific simulated distribution system (MS-SDS) tests. The SDS test (Method 5710C (APHA et al. 1995)) uses a lab scale method to estimate the T H M formation potential in a full-scale, distribution system. The SDS method attempts to mimic the conditions in the water distribution system, with the exception of exposure to pipe material and biofilm. The MS-SDS test, developed by Brereton (1998) is based on the same principal as the SDS test, but takes exposure to pipe material and biofilm into account. The samples in the MS-SDS test are incubated in PVC pipe containers which have an active biofilm inside, as opposed to 38 4.0 METHODS AND PROCEDURES the clean glass jars used for incubation in the SDS test. Brereton (1998) and Chan (2000) discuss the MS-SDS method and its implications in great detail. 4.3 Analytical Methods 4.3.1 Chlorine Residual Free chlorine residuals were measured by the DPD (N,N-diethyl-p-phenylenediamine) colourimetric method, using a field kit (Hach model CN-70). The instantaneous reaction between the DPD indicator (N,N-diethyl-p-phenylenediamine) and free chlorine, in the absence of iodide ion, produces a pink/red colour which can be matched to a calibrated coloured disk; (the greater the amount of free chlorine available for the reaction, the more intense the pink colour). Prior to the commencement of the work, the method was checked against a wet titration method (Method 4500B (APHA et al. 1995)). Field kit determinations were 78% to 85% of the wet titration determinations. Samples with free chlorine concentrations exceeding 3.5 mg/L were diluted with organic-free water, prior to determination. For free chlorine concentrations in the range 0.0 to 0.7 mg/L, the method detection limit and precision were equal to 0.02 mg/L. For free chlorine concentrations in the range 0.7 to 3.5 mg/L, the method detection limit and precision (repeated determinations of same sample) were equal to 0.1 mg/L (Brereton 1998). 4.3.2 Total Organic Carbon Total organic carbon (TOC) was measured using an automatic D O H R M A N Pheonix 8000 UV-Persulfate analyzer equipped with an autosampler. The analyzer employed the 39 4.0 METHODS AND PROCEDURES persulfate-ultraviolet oxidation method (Method 5310C (APHA et al. 1995)). The method can reportedly detect concentrations as low as 0.05 mg/L, although, for the concentration range 0.1 to 20 mg/L, the method detection limit was calculated to be 0.02 mg/L. The precision was calculated to be 87% of the standard deviation (0.006 mg/L). Each sample was analyzed three times. The calibration curve was constructed using dilutions of a standard solution (Tekmar D O H R M A N Certified Standard Aqueous Carbon Potassium Acid Phthalate 1000 ppm). A minimum of one sample of known concentration was analyzed for each run. The average recovery of the known sample over the term of the research was 109.43%. Blank samples were analyzed at the beginning (four blank samples, INITIAL) and at the end (one blank) of each run. The blank sample analyzed at the end of the run (FLNAL), was included to ensure that sample contamination from ambient carbon had not occurred. It was determined that if, tinfoil (maintained at 105°C for a minimum of one hour prior to use) was secured around the mouths of the sample vials, to protect the sample from contact with ambient air, the FINAL concentrations were only 0.003 mg/L higher, on average, than the INITIAL concentrations. 4.3.3 Ultraviolet Absorbance Ultraviolet (UV) absorbance was measured according to Method 5910B (APHA et al. 1995). Samples were analyzed on a Spectronic Unicam UV300 spectrophotometer, using a 1 cm cell length (quartz cuvette). Samples were not filtered prior to analysis. Absorption at each nanometer from 200 nm to 400 nm was measured on all samples and from 400 nm to 600 nm on selected samples. As this is a non-specific measurement, method detection limit can not be determined (APHA et al. 1995). The measured absorbances were used to calculate UV254 (absorbance at A,=254 nm), differential U V at 254 nm and 272 nm (absorbance at a 40 4.0 METHODS AND PROCEDURES specified wavelength before and after chlorination) and S U V A (absorbance at 254 nm over DOC concentrations). 4.3.4 Trihalomethane Formation Potential The trihalomethane formation potential, estimated using chloroform formation potential, is a means of comparing the potential trihalomethane concentration between samples by maintaining a controlled environment. The method does not consider any physical or chemical factors (pipe type, temperature, pipe turbulence, biological activity, etc.) acting on the sample, other than the chlorine addition. Trihalomethane formation potential was measured using Method 571 OB (APHA et al. 1995). Samples were placed in a 125 mL amber jar with buffer solution and chlorine dosing solution (5 to 11 CI2: TOC). The jars were sealed with PTFE lined septa and capped with open top screw caps. It should be noted that 125 mL bottles were used in place of the recommended 250 mL bottles. The temperature of the water at the time of collection ranged from approximately 3 °C to 8 °C, and the samples were stored at 4 °C until incubation (25 °C). The temperature increase from the storage temperature (4 °C) to the incubation temperature (25 °C) caused an increase in the water volume. In the 250 mL bottles, the volume increase was greater than the septa could accommodate. The expanding water broke the seal provided by the septa and leaks developed. Water and presumably gases were able to leak from the bottle and air was able to enter (bubbles formed in the sealed bottles). It was determined that the septa controlled the volume increase in the 125 mL bottles and therefore, this bottle size was used for the duration of the study. 41 4.0 METHODS AND PROCEDURES Duplicate samples (duplicating the entire procedure including sample collection) were typically analyzed for each sample. Analysis duplicates (duplicating only buffering, dosing and incubation) were typically analyzed for 1 in every ten samples. 4.3.5 Trihalomethanes Total trihalomethane concentrations were estimated using chloroform concentrations, which were determined by liquid-liquid extraction and gas chromatography (Method 6232B (APHA et al. 1995)). Pentane (Fischer Chemical, HPLC grade) was used as the extraction solvent (ratio 2 mL pentane: 10 mL sample). Background interferences, associated with the extraction solvent, were eliminated by passing the solvent through a column of basic alumina (Brockman Activity I, 60-325 mesh) which was heated to 230 °C for a minimum of 2 hours and then cooled prior to use. Chloroform concentrations were analyzed using a Hewlett Packard 6890 series gas chromatograph (GC) equipped with a Hewlett Packard 7673 autosampler. One uL volume aliquots were injected. Table 4.4 shows the GC operating conditions. Chloroform concentrations were determined by comparing the sample response peaks to standards response peaks. Three to five standards enveloping the expected sample concentrations were measured in each GC run and were used to construct a calibration curve. Solvent blanks were measured at the beginning and end of each run. Duplicate samples (duplicating the entire procedure including sample collection) were typically analyzed for each sample. Analysis duplicates (duplicating only chloroform extraction and analysis) were typically analyzed for 1 in every ten samples. The method detection limit was 0.23 pg/L. 42 4.0 METHODS AND PROCEDURES Table 4.4. Gas Chromatograph Operating Conditions for Chloroform Analysis Parameter Setting Capillary Column Type DB624 (J&W Scientific 125-1334) Length and diameter 28 m x 530 pm i.d. Film thickness 3.0 pm Carrier gas Helium @ 5.5 mL/min Make-up gas Nitrogen Combined flow rate (He + N 2 ) @ 60 mL/min Oven Temperature Initial temperature 30 °C, holding for 2 minutes Ramp 5 °C/min Final Temperature 90 °C Injector Type splitless Temperature 90 °C Detector Type Electron capture detector Temperature 260 °C 4.3.6 Glassware and Reagent Preparation Sample collection vials (40 mL, amber glass) and bottles (1.25 mL, 500 mL, 1 L and 2.3 L) were rinsed once with tap water, rinsed with ultra-pure water three times, dried at 350 °C for one hour and then rinsed with sample water three times, prior to sample collection. A l l volumetric and laboratory glassware was rinsed with ultra-pure water three times prior to use. Extraction and autosampler vials were rinsed with solvent (pentane) and fired at 350 °C for one hour. Plastic caps and septa were washed with detergent then rinsed with tap water and ultra-pure water and were dried at 105 °C. A Millipore Alpha-Q Ultra-pure water system was employed to provide organic-free water. The system uses the supplied municipal tap water passed through a purification pack, which includes activated carbon, ion exchange resin, and an organic scavenger. The system 43 4.0 METHODS AND PROCEDURES produces 18.2 megaohm-cm ultra-pure water (Type I water, ASTM) with a reported TOC concentration less than 10 ppb (Millipore Corporation 2000). A l l reagents, including dosing solution, quenching agents, calibration standards, travel blanks, reagent blanks, and dilutions were prepared with this water. In addition, this water was used in the preparation of all glassware. The chlorine dosing solution was prepared from ultra-pure water and stock hypochlorite solution (Javex Bleach, Colgate-Palmolive Inc.). The stock hypochlorite concentration was determined to be 5.38% on opening of the bottle on June 2000 and 5.20% on last use in March 2001. 4.4 Statistical Evaluation Two methods of statistical analysis were used to evaluate the obtained data. Relationship trends were evaluated by the calculation of correlation coefficients (r2). The difference between two means or a determination of whether or not two data sets were collected from different populations, was assessed using the Student's Mest. The statistical calculations and graphical illustrations were completed using Microsoft Excel 2000 (Microsoft Corp., Redmond, WA). 4.4.1 Correlation Coefficients Correlation coefficients are used to summarize a pair of variables. These are a measure of the strength of the linear association between the variables ((Gilbert 1987)). If the variables tend to increase together, the correlation is considered positive. If the variables tend to increase and decrease in opposition of each other, the correlation will be negative. As the 44 4.0 METHODS AND PROCEDURES correlation coefficient increases in magnitude, the points become more tightly concentrated about a straight line through the data, Figure 4.2. ' " C d r r - 0 2 0 C o r r - O O i 0 2 - 4 6 C a r r - 0 .iO 0 2 , 4 6-C a r r - 0 3 0 "jr*" _r-' O 2 4 ••€•,. C a r r - 0 i 0 0, 2 ,4 6 Cti'rr-'OiO)'' 4 6 0 . . . 2 4 , 6 . . ; C d " T OJBO. '.ft, 2 .4 6 . " C a r r - O . ' K .0 2 4 6. C a r r - 0,70 2 4 6 Figure 4.2. Examples of scatterplots showing the corresponding correlation coefficients ((Dallal 1999)) A correlation coefficient of 0.8 or greater was considered to be significant in this study. 4.4.2 Student's r-tests With large samples, confidence intervals for population means can be constructed by using only the sample mean, sample standard deviation, sample size, and the properties of the 45 4.0 METHODS AND PROCEDURES normal distribution. This is true regardless of the distribution of the individual observations. This is not true for small sample sizes (n<60). Even when small sample sets themselves follow a normal distribution exactly, the difference between the sample set and the population means tends to be greater than would be predicted by the normal distribution (Dallal 1999). William Sealey Gosset, pseudonym 'A Student of Statistics', discovered that when individual observations follow a normal distribution, confidence intervals for population means using a small sample set could be calculated like large sample sets. The difference being the multiplier used for the large sample set was replaced by a multiplier that grew larger as the sample size became smaller. Gosset also discovered that a similar method could be used to compare two population means provided the sample sets in both populations follow normal distributions and the population standard deviations were. The multiplier is replaced with a multiplier inversely dependent on sample size and the two sample standard deviations are combined (or pooled) to give a best estimate of the common population standard deviation. The multipliers in the case of small samples come from a distribution which Gosset named the t distribution. (Student's t distribution). The p values returned from the Student's Mest, indicate the probability that the two means being compared come from different populations (ie: the samples can be considered different). A/?=0.01, means a 99% probability that the two means came from different populations. Probabilities equal to and greater than 90% are considered significant. A l l Mests were performed with 95% confidence. 46 5.0 RESULTS AND DISCUSSION 5.1 Apparent Molecular Weight Characterization Apparent molecular weight characterization was attempted on several samples to aid in showing the approximate M W range of the organics removed by each treatment option and to determination which M W range organics were primarily responsible for the formation of THMs. Only four samples were characterized and processed according to the described protocol. The following section includes the reportable results and a discussion regarding the limitations of the procedure. 5.1.1 Factors Affecting M W C O UF membranes are usually anisotropic structures, which have a finely porous surface layer or skin, supported by a stronger, more open substrate. The finely porous layer (average pore diameter is 10 to 1000 A (Baker 2000) performs the separation of water and microsolutes from macromolecules and colloids; also the macroporous substrate provides physical strength. UF membranes are characterized by their molecular weight cutoff (MWCO), as discussed in Section 2.5.1, but the system as an analytical tool has limitations. Amy et al. (1992), who developed the method, state that the method does not take the rejection properties of the membranes into account. They discuss a study which concluded that the UF membranes underestimate the low-molecular weight, organic fractions. The pH of the feed solution is another factor that affects permeation through UF membranes, as the pH will affect the ionization of some molecules. Ionized molecules may 47 5.0 RESULTS A N D DISCUSSION tend to repel each other and the molecules may become extended and inflexible. Since non-ionized molecules may not be in a state of repulsion, they tend to be more flexible and will more readily pass through the membrane (Baker 2000). Membrane fouling is the effect of reversible and irreversible processes occurring at the membrane surface and in the membrane itself. The processes result in a reduction of permeate flux or a permeate of lesser quality (decrease in solute rejection). There are many causes of fouling including concentration polarization, precipitation of solute at the membrane surface (Figure 5.1), plugging of membrane pores, biological fouling and degradation of the membrane itself (American Water Works Association 1990). molecules in bulk us internal fouling Figure 5.1. Schematic Representing Membrane, Surface and Internal Fouling 48 5.0 RESULTS AND DISCUSSION 5.1.2 M W C O Determination To overcome the uncertainties associated with the M W C O classification of the flat sheet lab-scale UF membranes used in the classification procedure, an attempt was made to determine the initial effective M W C O of each membrane and how to assess the effective M W C O decreased with continued use. Polystyrene molecular weight standards (Polymer Standards, Polymer Laboratories) were employed to try to determine the effective M W C O of the membranes. Standards of known molecular weight (550, 1660, 4950 and 186 000 Daltons) and known concentration (0.75, 1.5 and 3 ppm) were measured for U V absorbance from 200 nm to 400 nm. The absorbance curves consistently peaked at X = 265 nm. The UV265 absorbance readings were used to calculate a standard curve for each molecular weight standard (Figure 5.2). A 3 ppm solution of each M W standard was filtered through each of the lab-scale UF membrane filters, to be used for the apparent molecular weight characterization. A sample of the filtrate was collected and measured for absorbance at A, = 265 nm. An estimate of the concentration in the filtrate was obtained using the calibration curves. From the concentration of the filtrate, an estimate of the percent retained by the UF membrane was calculated. The results are summarized in Table 5.1 and are shown in Figure 5.3. The results for the 10 000 M W C O filter are as expected, with a higher retention of solutes with MW>MWCO; however the percent retention was only 62%, while according to the manufacturer's M W C O , one might expect a percent retention greater then 90%. The results for the 100 000 M W C O filter showed low retention and the percent retention decreased for the solute with MW> M W C O . This shows that the M W C O classification is unreliable and one would never know the actual 49 5.0 RESULTS A N D DISCUSSION molecular weight of the molecules being retained and those passing the filter, without further characterization. 50 5.0 RESULTS AND DISCUSSION g 1.5 c re -P 1 o (fl < 0.5 M W 1 8 6 0 0 0 0.5 1.5 2 Concentration 2.5 3.5 S 1-5 c re •S 1 o (0 n < 0.5 M W 4 9 5 0 0.5 1.5 2 Concentration 2.5 3.5 Figure 5.2b. Molecular Weight Standards Calibration Curves (R =1.00 for both curves) 51 5.0 RESULTS AND DISCUSSION MEMBRANE MWC 10 000 Q UJ HI NOMINAL MWC , .SOLUTE, MW i f t B S S f i THAR WFiMBff AFH? WNQ - EXPECT *TLE 30% RETENTION g Q I 1 ITC M W «iMAI I PR T H A N M P M R R A N F MWC. - FXPPfIT A UOW RETENTION 0 20000 40000 60000 80000 100000 120000 140000 160000 160000 200000 SOLUTE MW MEMBRANE MWC 100 000 NOMINAL MWC f GREATER THAN 90% SOLUTE MW SMALLER THAN MEMBRANE MWC -^XPECT-AU3W-RETENT40N SOLUTE MW Figure 5.3 UF Filter M W C O Tests 52 5.0 RESULTS AND DISCUSSION A second test was completed to determine how the membrane M W C O changed with increased filtration. A 3 ppm, M W = 4950 solution was filtered through a 10 000 M W C O membrane under the following progressive conditions: • Filter new and clean, • 600 mL of a 3 ppm MW=4950 solution filtered through, • 2 L raw water filtered through, and • Filter had undergone a chemical cleaning intended to restore the filter to new and clean condition. Table 5.1. Percent Retained Results from Filtering Solution of Known Molecular Weight and Known Concentration Through the UF Membranes Filter YM1 YM3 YM10 YM100 0.45p M W C O (Daltons) 1000 3000 10 000 100 000 -150 000 Solute Molecular Weight % Retained M W = 580 14.0 17.3 5.4 3.4 2.6 M W = 1660 14.5 11.0 9.0 7.0 7.8 M W = 4950 NS NS 9.8 9.5 9.6 M W = 186 000 NS NS 61.7 3.6 6.3 NS - Not Sampled The results were obtained in the same manner as the effective M W C O test and are summarized in Table 5.2. As the results show, the percent retention doubled after 600 mL of solute and increased 3-fold, after filtration of one raw water sample. Chemical cleaning (Tergazyme© wash) did not restore the filter. To maintain consistency through out the project, a new filter would need to be used for each sample. This was beyond the budget of the project. 53 5.0 RESULTS AND DISCUSSION Table 5.2. Results of a 3 ppm MW=4950 Solution Progressively Filtered Through a 10 000 M W C O UF Membrane 3 ppm MW=4950 Solution Sample Conditions (progressive) Absorbance (UV265) % Retained by Filter Filter new and clean (YM10) 1.596 10 Collected after 600 mL of solution filtered 1.503 15 Collected after 2 L of raw water filtered 0.826 54 Collected after chemical cleaning 0.808 55 The results of the two tests reported above indicate that the UF membranes provided inconsistent results. In addition to this, approximately 2 to 24 hours were required to filter a sample through the membranes. The longer filtration times allow for ambient air contamination, as well as bacterial growth, both of which may alter the sample. Due to these factors, the A M W characterization was discontinued after only several samples had been processed. 5.1.3 A M W Characterization Results Only four samples were characterized and processed according to the described protocol (Section 4.2.2). A l l four of the reportable samples were recovered from the Membrane A (MF) unit. The samples were progressively filtered through UF membranes with the following MWCOs: • 100 000 Daltons, • 10 000 Daltons, and • 3000 Daltons; providing the following A M W categories: • > 100 000, 54 5.0 RESULTS AND DISCUSSION • 100 000 to 10 000, • 10 000 to 3000, and • <3000. The results are summarized in Table 5.3 and shown in Figure 5.4. As indicated in Figure 5.4, in the unpretreated sample, the concentration of TOC was found to be fairly evenly distributed. The chloroform formation potential increases with decreasing A M W , the largest potential being associated with the <3000 Dalton fraction. This indicates that the THMFP precursors are associated with the organics found in the light end of the humic acid range and the fulvic acid range. Ergo, complete removal of THMFP would require multiple treatment techniques. In the pretreated water samples, the greatest percent concentration of TOC is found in the two smaller fractions. This indicates that, coagulation removed organics >10 000 Daltons. The chloroform formation potential was found to increase with decreasing A M W and the greatest potential is in the smallest A M W fraction (with the exception of one sample for which the maximum potential was associated with very heavy organics in the 100 000-10 000 A M W range). Coagulation did not remove organics <10 000 Daltons; therefore, this fraction represents the un-removed DBP precursors. 55 5.0 RESULTS AND DISCUSSION Chloroform Formation Potential HO mg/L Coagulant • 5 mg/L B5mg /L B 5 mg/L > 100 000 100 000-10 000 10 000-3000 <3000 AMW Fractions TOC Concentration B 0 mg/L Coagulant • 5 mg/L B5mg/L • 5 mg/L > 100 000 100 000-10 000 10 000-3000 <3000 AMW Fractions Figure 5.4. A M W Distributions of TOC and Chloroform Formation Potential for Un-pretreated (0 mg/L PAC1) and Pretreated (5 mg/L PAC1) M F Water 56 5.0 RESULTS AND DISCUSSION Table 5.3. Measured concentrations of TOC and THMFP for A M W samples M W Fraction Parameter Sample (Applied Coagulant Dose) 0 mg/L 5 mg/L 5 mg/L 5mg/L Unfiltered TOC 1.44 0.51 0.52 0.42 THMFP 103.5 76.4 67.9 66.1 100 000- 10 000 TOC 1.29 0.46 0.48 0.39 THMFP 100.2 68.1 29.4 63.4 10 000-3000 TOC 0.72 0.42 0.40 0.30 THMFP 78.2 71.1 45.3 29.5 <3000 TOC 0.29 0.22 0.17 0.12 TMFP 41.9 45.8 27.3 15.3 TOC concentrations reported in mg/L THMFP reported in pg/L Replicate data were used to conduct a series of paired Mests to determine statistical differences in terms of confidence levels. The results are shown in Table 5.4. TOC concentrations in the different A M W fractions were shown to be statistically different at the 90% confidence level. The confidence levels associated with the THMFP differences ranged from 95% to 50%. 57 5.0 RESULTS AND DISCUSSION Table 5.4. Replicate Series and Statistical Analysis for Sample MJ29 § (5 mg/L coagulant and MF) Parameter Measurements AMW Fraction n TOC (mg/L) ± standard deviation THMFP (pg/L) ± standard deviation >100 000 Daltons 2 0.42 ±0.01 62.85 ± 0.63 <100 000 2 0.38 ±0.01 64.75 ±1.88 <10 000 2 0.29 ± 0.02 33.15 ± 5.17 Paired f-Test Significance Levels* Comparisons TOC THMFP >100 000 vs. <100 000 fXO.l p<0.5 <100 000 vs. <10 000 p<0.l p<0.5 >100 000 vs. <10 000 p<0.l p<0.05 § Membrane A: US Filter/fvlemcor, collected January 29, 2001 t A 90% confidence level corresponds to p=0.1 5.2 SDS and MS-SDS Tests 5.2.1 The MS-SDS Test and Theory The SDS test (Method 5710C (APHA et al. 1995)) is used to predict the effect of distribution system residence time on T H M development. The SDS (method described in Section 4.2.3) does not take into account any other physical parameters which may influence T H M formation. The MS-SDS test was developed by Brereton (1998) to account for the effect of the physical parameters: pipe material and in situ biofilm. The test is founded on the same principle as the SDS test but includes the internal pipe environment effect. The samples are incubated in material-specific pipe sections which have an active, in situ, biofilm lining. 58 5.0 RESULTS AND DISCUSSION 5.2.2 Results of the SDS and MS-SDS Tests Two sets of raw and membrane-filtered samples were collected and subjected to MS-SDS tests, to compare to the more conventional SDS results. The results indicated an increased chloroform yield in the MS-SDS results, compared to the SDS results, with the exception of one raw water sample. The average increase was the smallest for the raw water samples and largest for those filtered through Membrane B/UF unit. The results are summarized in Table 5.5. Table 5.5. SDS and MS-SDS Results Showing Increased Chloroform Production for the MS-SDS Tests S A M P L E Y I E L D (pg/L chloroform/mg/L Percent residual chlorine) increase (%) SDS TEST MS-SDS TEST F6 R A W 15.9 22.1 38 F6 R A W (duplicate) 10.6 19.4 83 F l l R A W 37.3 12.5 -67 average 18 F6 M E M B R A N E A (MF) 11.8 17.8 51 FI 1 M E M B R A N E A (MF) 42.5 50.2 18 FI 1 M E M B R A N E A (duplicate) 43.2 52.6 22 average 30 F6 M E M B R A N E B (UF) 10.2 19.3 89 FI 1 M E M B R A N E B (UF) 33.1 49.9 51 average 70 Chan (2000), also completed an MS-SDS study in 2000 on Seymour water. Her results showed a decreased T H M production in the PVC pipe samples (MS-SDS test) incubated for longer than 4 days, compared to the glass bottle incubated samples (SDS test). In comparision, the results reported here (incubated for 7 days) indicate an increased T H M 59 5.0 RESULTS AND DISCUSSION production for the MS-SDS test in 7 out of 8 samples. Chan also showed that for samples having undergone filtration treatment (sand/anthracite) and incubated longer than 4 days, the MS-SDS test resulted in lower T H M production than the SDS test. The results reported here indicate that, the greater the level of treatment (lowest to highest being raw, M F , and UF), the greater the increase in the T H M production of the MS-SDS test, compared to the SDS test. Chan attributed the decreases she measured to biological activity and bio-uptake in the pipe biofilm environment. It is possible that the results reported here do not reflect the results reported by Chan because the chlorine concentration, used for dosing the samples during the present tests, was great enough to kil l the existing biofilm. Therefore, there would be no active bio-uptake occurring at the pipe surface and there would be extra organic material (extracellular material) available as T H M precursor material. Chan suggested that, when the biofilm is experiencing "stress", humic substances previously adsorbed into the biofilm, are released and become available for T H M formation. More work needs to completed on the MS-SDS tests to confirm the results. Chan concluded that the pipe effect (pipe material and biofilm) negatively influenced THMFP (ie: the pipe effect reduces T H M concentrations). This study indicates that the pipe effect positively influences THMFP (pipe effect increases T H M concentrations) and that the degree of pretreatment also positively influences T H M production (the greater the level of treatment, the higher the T H M concentrations). 60 5.0 RESULTS AND DISCUSSION 5.3 TOC Removal 5.3.1 Stage 1 Disinfectants and Disinfectant By-Products Rule Amendments to the United States Safe Drinking Water Act (SDWA) in 1996 require the Environmental Protection Agency (USEPA) to develop rules to balance the risks between protecting the consumer from microbial contaminants and harming the consumer from potential health effects associated with DBPs. The Stage 1 Disinfectants and Disinfectant By-Products Rule (Stage 1 Rule) is one of the first rules under the amendment (EPA 1998). The rule establishes maximum disinfectant levels and specified removal percentages of organic materials (measured as TOC). The TOC removal is a matrix-type standard. The required removal ranges from 15% to 50%, depending on source water TOC concentration and source water alkalinity. The Stage 1 Rule mandates TOC removal because the organic materials (TOC) may react with disinfectants to form DBPs. Therefore, the ultimate goal in removing TOC is to reduce the DBP formation potential. The USEPA-recommended method of TOC removal is through enhanced coagulation using hydrolyzing metal salt coagulants, although alternate treatment techniques are acceptable. 5.3.2 Expected T O C Removal As mentioned above, the Stage 1 Rule mandates TOC removal because the natural organic matter (TOC) present in a water source may react with the applied disinfectant to form DBPs. Taylor et al. (1987) developed a model that predicted T H M formation as a function of TOC, pH, chlorine dose, temperature and reaction time. They demonstrated that 61 5.0 RESULTS AND DISCUSSION an increase in TOC would increase T H M production more than a similar change in any other variable. The quality and quantity of the TOC present in a water source is dependent on many factors specific to the water source (e.g. local vegetation, temperature). Many researchers have tried to summarize the distribution of TOC in waters (e.g. Figure 5.9) and it is generally agreed that humic substances comprise 50% to 75% of N O M in surface waters (Thurman 1985, American Water Works Association 1990, Chan 2000). Researchers estimate that the range of molecular weights of humic substances (humic acids plus fulvic acids) is from 500 Daltons to 100 000 Daltons with the median being in the range from 5000 to 10 000 Daltons (Thurman 1985). The efficiency of TOC removal will be a function of the M W distribution of the N O M and the applied removal process. N O M removal by membrane filtration, coagulation and PAC has been widely documented in the literature (Taylor et al. 1987, Tan and Amy 1991, Crozes et al. 1995, Vrijenhoek et al. 1998, Siddiqui et al. 2000, Farahbakhsh and Smith 2001), see Section 2.5 for further details. TOC removal efficiency by membrane filtration ranges from poor to very good. UF membranes have been shown to remove 5% to 57% of TOC (Taylor et al. 1987). On the other hand, M F membranes have shown a 10% (Vickers et al. 1995) to 18% removal (Siddiqui et al. 2000). The A M W tests (Section 5.1), and previous research, indicate that 10 to 50% TOC removal could be expected from coagulation. TOC removal by PAC has not been reported to be very high (Najm et al. 1991); previous research indicates that we can expect 10 to 20 % TOC removal. 62 5.0 RESULTS AND DISCUSSION 5.3.3 Results TOC removal results are shown in Figure 5.5 and summarized in Table 5.6. Since Membrane B's M W C O is 0.01 um compared to Membrane A ' s M W C O of 0.2 pm, a higher removal efficiency was expected. As indicated, Membrane B (UF) apparently resulted in higher average removal efficiencies (30.9%) than Membrane A (MF) (12.9%), but statistically, the differences were not significant. The TOC removal efficiencies from Membrane A effluent and Membrane B effluent were compared statistically, using a Mest. The difference between the two treatments was shown not to be significant (p=0.4). The average TOC concentrations from Membrane A effluent and Membrane B effluent were also compared statistically using a Mest, to average raw water TOC concentrations. Due to a large variance, the Mest indicated that the difference between the raw water and the Membrane B filtered effluent was not significant (/?=0.75). The difference between the raw water and the Membrane A filtered water was shown to be significant with a large confidence interval (p=0.1). T-tests were completed to compare each pretreatment dosage within each treatment (Tables 5.7 and 5.8). For Membrane A (MF) effluent, which was pretreated with coagulant (polyaluminum chloride), significant differences, p<0.00\, were shown between no pretreatment (dose=0 mg/L) and each PAC1 dose. This indicates that coagulant addition significantly reduces TOC concentrations. For Membrane B (UF), significant differences were shown between the filtered water (no pretreatment) and a PAC dose of 10 mg/L (p=0.05) and a dose of 20 mg/L (p=0.05). 63 5.0 R E S U L T S A N D D I S C U S S I O N TOC Removal • Membrane B (UF) & PAC • Membrane A(MF) & PACI -439--1004 80 ro > O E o X. o o membrane B 60 49 3 • • average removal 3 0 . 9 % membrane A average removal 12J9J%__ • 10 15 20 25 PAC or PACI dosage (mg/L) Figure 5.5. Percent TOC Removals for Each Treatment Option Table 5.6. Average TOC Concentrations for Each Treatment Option Raw Ultrafiltration Microfiltration Coagulant 0 (mg/L) 5 10 15 20 P A C 0 (mg/L) 2 4 8 T O C 1.66 1.60 1.04 1.09 1.47 0.96 1.39 0.31 0.50 0.29 n 24 9 6 1 2 3 17 1 3 2 64 5.0 RESULTS AND DISCUSSION Table 5.7. T-test Comparison Between all M F Treatments for TOC Removal M E M B R A N E B (microfiltration) Applied PAC1 Dosage (mg/L) 0 2 4 8 0 - pO.001 p<0.001 pO.001 2 p<0.001 - p=0.05 p=0.5 4 p<0.001 p=0.05 - p=0.10 8 pO.OOl p=0.5 £=0.10 -Table 5.8. T-test Comparison Between all UF Treatments for TOC Removal M E M B R A N E A (ultrafiltration) Applied P A C Dosage (mg/L) 0 5 10 15 20 0 - p=0.2 /?=0.05 p=0A p=0.025 5 p=0.2 p=0.6 p=0.15 ^=0.65 10 p=0.05 p=0.6 - p=0.05 p=0.2 15 p=0A p=0.15 p=0.05 - p=0.05 20 p=0.025 p=0.65 p=0.2 p=0.05 -The results indicate that the TOC removal efficiencies achieved by the UF membrane and the MF membrane were not significantly different when compared to each other. The average TOC concentrations of the finished water were shown not to be significantly different for UF treated water and the raw water. M F treated water and raw water were shown to be different. The addition of coagulant (at any of the tested doses) significantly reduced the TOC concentration in the water. The addition of PAC was shown to significantly reduce TOC only at the 10 mg/L and 20 mg/L dose. 65 5.0 RESULTS AND DISCUSSION 5.4 Chloroform Formation Potential Removal 5.4.1 Stage 1 Disinfectants and Disinfectant By-Products Rule With advances in the identification of chlorination by-products and their health effects, a greater emphasis on DBP control has been written into US drinking water regulations with rumours that Canada will soon follow. The USEPA's Stage 1 Disinfectants and Disinfectant By-Products Rule (Stage 1 Rule) reduces the allowable concentration for total THMs to 80 pg/L (EPA 1998). For many water supplies, treatment is required to achieve this low level of contamination. 5.4.2 Expected T H M F P Removal M F and UF can not achieve significant THMFP removals due to the M W of the T H M precursor material. Laine et al. (1989) demonstrated that the average A M W of most DBP precursors is too small for this material to be retained by a M F membrane, and UF retention would be minimal. The humic substances which serve as T H M precursors have MWs in the range of 500 Daltons to 100 000 Daltons, with the median being 5000 to 10 000 Daltons (Thurman 1985), while UF processes have a M W C O in the range of 100 000 Daltons and M F processes have a M W C O >1 000 000 Daltons. The A M W tests and previous research indicate that 10 to 50% THMFP removal could be expected from coagulation and a 30 to 45% THMFP removal by PAC. 66 5.0 RESULTS A N D DISCUSSION 5.4.3 Results: T H M F P Removal THMFP removal results (calculated as the difference in chloroform concentrations in the treated sample and chloroform concentrations in the raw water sample) are shown in Figure 5.6 and Table 5.9. As indicated, Membrane A (MF) appeared to produce a higher average removal (27.2%) than Membrane B (UF) (18.5%). Since the Membrane B M W C O is 0.01 pm compared to the Membrane A M W C O of 0.2 pm, the higher removal efficiency for Membrane A was not expected. Statistically, the difference is not significant. The THMFP removal for Membrane A and Membrane B were compared using a /-test. The difference between the two treatments was shown only to be significant at an 80% confidence interval (p=0.2). The average THMFP concentrations from Membrane A effluent and Membrane B water were compared, statistically, to average raw water THMFP concentrations, using a t-test. The Mest indicated that the difference between the raw water and the Membrane B filtered effluent was significant (£=0.05). The difference between the raw water and the Membrane A filtered water was shown to be highly significant (£=0.01). T-tests were completed to compare each pretreatment dosage within each treatment (Tables 5.10 and 5.11). For Membrane A (MF) effluent, which was pretreated with coagulant (polyaluminum chloride), significant differences, £<0.001, were shown between no pretreatment (dose=0 mg/L) and each PAC1 dose. A significant difference was also shown between the 4 mg/L and 8 mg/L PAC1 dose (p=0.05). This indicates that coagulant addition significantly reduced THMFP concentrations. For Membrane B (UF) effluent, significant differences were shown only between the filtered effluent (no pretreatment) and the highest PAC dose of 20 mg/L (£=0.1). 67 5.0 RESULTS AND DISCUSSION THMFP Removal • Membrane A (MF) & PACI • Membrane B (UF) & PAC PAC or PACI Dosage (mg/L) Figure 5.6. Percent THMFP Removals for Each Treatment Option Table 5.9. Average THMFP Concentrations for Each Treatment Option Raw Ultrafiltration Microfiltration Coagulant 0 (mg/L) 5 10 15 20 P A C 0 (mg/L) 2 4 8 THMFP 188.8 172.0 139.9 150.5 178.9 128.6 146.7 47.2 67.4 39.6 n 42 11 13 4 1 6 3 0 2 5 6 6 8 5.0 RESULTS AND DISCUSSION Table 5.10. 7-test Comparison Between all M F Treatments for THMFP Removal M E M B R A N E B (microfiltration) Applied PAC1 Dosage (mg/L) 0 2 4 8 0 - pO.OOl pO.001 p<0.001 2 p<0.001 - p=0.3 p=0.25 4 p<0.001 p=0.3 p=0.05 8 p<0.001 p=0.25 p=0.05 -Table 5.11. T-test Comparison Between all UF Treatments for THMFP Removal M E M B R A N E A (ultrafiltration) Applied P A C Dosage (mg/L) 0 5 10 15 20 0 - p=0.2 P=0.2 p=0.96 p=0.1 5 p=0.2 - p=0.65 p=0.55 p=0.2 10 p=0.2 p=0.65 - p=0.45 p=0.2 15 p=0.96 p=0.55 p=0A5 p=0.15 20 p=0.l p=0.2 p=0.2 p=0.15 The results indicate that although there was not a significant difference in THMFP removal between Membrane A and Membrane B, there was significant THMFP removal for UF and M F filtration systems. In addition, coagulant addition at any dose and PAC addition at 20 mg/L produced significant THMP removal. 5.4.4 Results: Chloroform Yield Reduction In addition to T H M reduction, the chloroform yield reduction was evaluated. The chloroform yield is defined as the chloroform production divided by the chlorine consumed 69 5.0 RESULTS AND DISCUSSION during the incubation period (pg/mg). The chloroform yield presents a normalized parameter by which to compare the samples. In using chlorination as a treatment technique, it is desirable to add an excess of chlorine (to provide a residual for distribution), while minimizing T H M formation. The yield parameter allows us to evaluate the effluents on this basis. A smaller yield is desirable because it indicates a smaller chloroform production over a larger chlorine consumption. Yield reduction results (calculated as the difference between chloroform yield in the treated sample and chloroform yield in the raw water sample) are shown in Figure 5.7 and Table 5.12. As indicated, Membrane A (MF), 27.5%, apparently showed a better average reduction than Membrane B (UF), 14.6%. Since Membrane B nominal M W C O was 0.01 pm compared to Membrane A M W C O of 0.2 pm, the higher reduction was not expected. Statistically the difference was not significant. The yield reductions from Membrane A and Membrane B were compared, using a Mest. The difference between the two treatments was shown not to be significant (p=0.65). The average yield for Membrane A and Membrane B were compared, statistically, using Mests to average raw water yield. The Mests indicated that the difference between the raw water and the Membrane B filtered effluent was significant (p=0.01). The difference between the raw water and the Membrane A filtered water was shown to be significant, at a large confidence interval (p=0.l). 7-tests were completed to compare each pretreatment dosage within each treatment (Tables 5.13 and 5.14). For Membrane A (MF) effluent, which was pretreated with coagulant (polyaluminum chloride), significant differences,p=0.05 top<0.00\, were shown between no pretreatment (dose^O mg/L) and each PACI dose. This indicates that coagulant addition significantly reduces the chloroform yield. For Membrane B (UF) effluent, significant 70 5.0 R E S U L T S A N D D I S C U S S I O N differences were shown only between the filtered effluent (no pretreatment) and a PAC dose of 5, 15 and 20 mg/L (p=0.l, 0.1, 0.05, respectively). Significant differences were also seen between the 5 mg/L and 10 mg/L PAC dose (p=0.l) and between the 10 mg/L and 20 mg/L PAC dose (p=0.1). 100 > o E cu OL 0) > -80 60 40 20 * 0 o -20 -40 -60 Chloroform Yield Removal M e m b r a n e A (MF) & PACI • M e m b r a n e B (UF) & PAC membrane A average^ removal 27.5% v membrane 5 R aver 10 ft 2rags_ removal 14.6% 20 • PAC or PACI Dosage (mg/L) 25 Figure 5.7. Percent Chloroform Yield Reductions for Each Treatment Option 71 5.0 RESULTS A N D DISCUSSION differences were shown only between the filtered effluent (no pretreatment) and a PAC dose of 5, 15 and 20 mg/L (p=0.l, 0.1, 0.05, respectively). Significant differences were also seen between the 5 mg/L and 10 mg/L PAC dose (p=0.1) and between the 10 mg/L and 20 mg/L PAC dose (p=0.1). Chloroform Yield Removal • Membrane A (MF) & PACI • Membrane B (UF) & PAC -60 1 PAC or PACI Dosage (mg/L) Figure 5.7. Percent Chloroform Yield Reductions for Each Treatment Option 71 5.0 R E S U L T S A N D D I S C U S S I O N Table 5.12. Average Yield for Each Treatment Option Raw Ultrafiltral tion Microfiltration Coagulant 0 (mg/L) 5 10 15 20 PAC 0 (mg/L) 2 4 8 Yield 1.59 12.5 12.1 10 15 11.1 11.2 4.5 6.4 3.3 n 42 11 13 1 4 6 30 2 5 6 Table 5.13. T-test Comparison Between all M F Treatments for Chloroform Yield Reduction M E M B R A N E B (microfiltration) Applied PACI 0 2 4 8 Dosage (mg/L) 0 - p<0.001 p<0.001 p<0.001 2 /K0.001 - p=0.3 p=0.25 4 pO.OOl p=0.3 p=0.05 8 pO.001 p=0.25 p=0.05 -Table 5.14. T-test Comparison Between all UF Treatments for Chloroform Yield Reductions M E M B R A N E A (ultrafiltr ation) Applied P A C Dosage (mg/L) 0 5 10 15 20 0 - p=0.2 p=0.2 p=0.96 p=0A 5 p=0.2 p=0.65 p=0.55 p=0.2 10 p=0.2 p=0.65 - p=0A5 p=0.2 15 p=0.96 p=0.55 p=0A5 p=0A5 20 p=0A p-0.2 p=0.2 p=0.15 -The results indicate that although there is not a significant difference in yield reduction between Membrane A and Membrane B, there is a significant yield reduction between UF/MF filtered water and raw water. In addition, coagulant addition at any dose and PAC addition at 5, 10 and 20 mg/L produced a significant yield reduction. 72 5.0 R E S U L T S A N D D I S C U S S I O N 5.5 Analysis of THMFP Surrogates 5.5.1 Analysis of Surrogates: TOC Correlation analysis was completed to determine the significance of the relationship between TOC concentration and THMFP. If the relationship was shown to be significant, TOC could be used as a surrogate parameter for THMFP. The relationship was shown not to be significant (r2=0.16). Correlation analysis was also completed for TOC concentration and chloroform yield. This relationship was also shown not to be significant (r =0.05). 5.5.2 Analysis of Surrogates: UV254 and S U V A There was an interest in the measurement of THMFP surrogate parameters. Statistical correlations between U V parameters and chloroform formation potential can be developed based on the collected data. If any of the correlations are shown to be significant, then that parameter could be considered as a suitable surrogate for THMFP. The correlations considered UV254 (ultraviolet absorbance measured at 254 nm), SUVA254 (UV absorbance at A=254 nm/TOC concentration) and differential U V at 254 nm and 272 nm (ultraviolet absorbance of the unchlorinated sample - U V A of the chlorinated sample at 254 or 272 nm) (Li etal. 1998) The results from this study indicated a weak correlation between UV254 and chloroform formation potential (Figure 5.8). Linear regression analysis showed the r2=0.57, which is less than the previously reported values ((Li et al. 1998)). 73 5.0 RESULTS AND DISCUSSION The S U V A results (Table 5.15) indicate that the raw water has a higher humic content (SUVA=8.25) than the UF-filtered water (SUVA=4.4) and the MF-filtered water (SUVA=5.5). Table 1.15 indicates that the UF-filtered water has a lower humic content than the MF-filtered water, which would be expected since the UF unit would be expected to remove more material in the humic molecular size range than would the M F unit. The SUVAs for Membrane A effluent and Membrane B effluent were compared, using a /-test. The difference between the two treatments was shown to be highly significant (p=0.025). The average S U V A results for each membrane effluent were compared to the raw water S U V A by means of a /-test. The differences between each effluent and raw water SUVAs were significant at a large confidence interval (p=0.1). The S U V A results show that the addition of coagulant does not appear to remove additional humic substances but, in fact, hindered its removal. However, other researchers have shown coagulation to be efficient in removing humic substances and S U V A (Vrijenhoek et al. 1998) and the THMFP removal results indicate that it is. The addition of PAC does not appear to alter the humic substance concentration. A linear regression analysis on the relationship between S U V A and chloroform formation, Figure 5.9, indicates that the relationship is weakly correlated (r =0.15). 74 5.0 RESULTS AND DISCUSSION Table 5.15. S U V A and Chloroform Formation Potential Results SOURCE OF PAC DOSE COAGULANT SUVA ± standard CHLOROFORM SAMPLE WATER (mg/L) DOSE (mg/L) deviation FORMATION POTENTIAL (ppb) ± standard deviation RAW 0 0 8.25 ±9 .4 179.9 ±46.6 MEMBRANE A 0 0 5.5 ±2.2 126.3 ±52.8 (MF) 0 2 8.1 ±0 .8 47.2 ± 0.8 0 4 6.4 ±3.5 59.8 ± 16.1 0 8 5.1 ±0 .3 41.5 ±2.2 MEMBRANE B 0 0 4.4 ± 1.4 173.3 ± 109.5 (UF) 5 0 5.7 ±0 .3 139.9 ±33.7 10 0 4.75 ± 1.4 92.9 ±50.2 15 0 5.2 ±0 .2 178.9 ±6.1 20 0 4.9 ± 1.5 128.6 ± 21.7 5.5.3 Analysis of Surrogates: Differential UV In this study, differential U V (ultraviolet absorbance of the unchlorinated sample -U V A of the chlorinated sample at 254 or 272 nm) was evaluated as a surrogate for THMFP at A=254 nm and A=272 nm. A wavelength 272 nm was selected to evaluate the quenched samples, because it is often used when there are interferences at 254 nm (e.g. sulfite which was used as the quenching agent) ((Li et al. 1998)). The results of linear regression analysis indicated that AUV254 (Figure 5.10, r2=0.49) is more highly correlated to chloroform formation potential than AUV272 (Figure 5.11, r =0.28) but that neither differential U V correlation described the relationship as well as UV254 (Figure 5.8, r2=0.57). 75 5.0 RESULTS AND DISCUSSION Figure 5.8. Chloroform Formation Potential vs. UV254 Chloro form Format ion Potent ia l Vs. SUVA Figure 5.9. Chloroform Formation Potential vs. SUVA 76 5.0 RESULTS AND DISCUSSION Figure 5.10. Chloroform Formation Potential vs. Differential U V at 254 Chlo ro fo rm Format ion Potent ial vs . Differential UV272 300 c £ o CL C o ra E 150 o u_ o o 100 50 j~i 2 r\ n o R -0.28 * • « • * 4 * * • * • • * 0 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 Differential Absorbance at 272 nm Figure 5.11. Chloroform Formation Potential vs. Differential U V at 272 77 6.0 Conclusions A primary objective of this research was to show that TOC and THMFP were not controlled by the same organic compounds, and, as such, their removals did not follow the same pattern. This means that one technique may be very effective at removing TOC but not THMFP in the same water or vice versa. This research showed that TOC removal did not equal THMFP removal. The A M W tests demonstrated that coagulation effectively removed organics above A M W of 10 000 Daltons but the greatest THMFP remained with organics of A M W <3000 Daltons. The pilot plant study also showed that TOC removal and THMFP removal were not equated. The microfiltration membrane removed an average of 12.9% of the raw water TOC and an average of 27.2% of the raw water THMFP, while the ultrafiltration membrane removed 30.9% of the raw water TOC but only 18.5% of the raw water THMFP. A secondary objective of this research was to show whether the raw water T H M production was humic acid- or fulvic acid-controlled. The T H M formation of the water appeared to be controlled by humic acids. The treatment process which removed organic material in the humic acid range (coagulation) showed the largest chloroform formation potential reductions. Additional conclusions are: 1) UV254 appeared to be the best THMFP surrogate parameter for the studied raw water. 78 6.0 CONCLUSIONS 2) M F and UF showed the same efficiency at TOC, THMFP and chloroform yield potential removal/reduction. Statistically, one was not more effective than the other at any removal. 3) Applied PACI as a pretreatment improved THMFP removal by up to 50% and chloroform yield reduction by up to 40%. 4) Applied PAC as a pretreatment did not improve THMFP removal or chloroform yield reduction. 5) M F showed a significant TOC removal compared to the raw water, UF did not. 6) Both M F and UF showed significant THMFP and chloroform yield removal/reduction. Further study in isolating the raw water humic and fulvic acids before and after treatment would aid in understanding the removal processes and in selected the best removal technology with respect to TOC and THMFP. 79 REFERENCES American Water Works Association (1990). Water Quality and Treatment, Mc-Graw Hil l , Inc. Amy, G.L., et al. (1992). "Molecular Size Distributions of Dissolved Organic Matter." Journal of the American Water Works Association 84(6): 67-75. A P H A , et al. (1995). Standard Methods for the Examination of Water and Wastewater. Published Washington, D.C. Babcock, D.B. and P.C. Singer (1979). "Chlorination and Coagulation of Humic and Fulvic Acids." Journal of the American Water Works Association 71(3): 149-152. Baker, R.W. (2000). Membrane Technology and Applications, McGraw-Hill. Ballew, H.W. (1978). Basics of Filtration and Separation, Nuclepore Corporation. Batchelor, B., et al. (1987). "Developing Haloform Formation Potential Tests." Journal of the American Water Works Association 79(1): 50-55. Blau, T.J., et al. (1992). "DBP Control by Nanofiltration: Cost and Performance." Journal of the American Water Works Association 84(12): 104-116. Brereton, J. (1998). Impacts of Tuberculated Iron and Surface Biofilm in Trihalomaethane Formation in Chlorinated Drinking Water. Department of Civil Engineering. Vancouver, B.C., University of British Columbia: 210. Calgon (2001). Ccc Products and Technology; Http :/AV ww. Calgoncarbon.Com/Industry/Producttech/Adsorption. Chan, K.C.W. (2000). The Effect of Biofilm on T H M Formation in Chlorinated Drinking Water of a PVC Pipe Distribution System. Civil Engineering. Vancouver, University of British Columbia. Crozes, G., et al. (1995). "Enhanced Coagulation: Its Effect on Nom Removal and Chemical Costs." Journal of the American Water Works Association 87:1:78: 78-89. Dallal, G.E. (1999). The Little Handbook of Statistical Practice. http://www.tufts.edu/%7Egdallal.htm. Dodds, L. (1999). "Trihalomethanes in Public Water Supplies and Adverse Birth Outcomes." Epidemiology 10(3): 233-237. 80 R E F E R E N C E S Edzwald, J.K., et al. (1985). "Surrogate Parameters for Monitoring Organic Matter and T H M Precursors." Journal of the American Water Works Association 77(4): 122-132. EPA, (1998). Stage 1 Disinfectants and Disinfectant by-Products Rule: Fact Sheet. EPA, (1999). Alternative Disinfectants and Oxidants Guidance Manual. EPA 815-R99-014. Farahbakhsh, K. and D. Smith (2001). Investigation of Ultrafiltration Performance on Be Drinking Water Quality, Seymour Pilot Plant Study, University of Alberta. Farahbakhsh, K. and D. Smith (2001). Membrane Pilot Studies - the Seymour Experience. Innovative Electrotechnologies in Water and Wastewater Treatment Seminar, Richmond, BC. Garcia-Villanova, R.J., et al. (1997). "Formation, Evolution and Modeling of Trihalomethanes in the Drinking Water of a Town: I. At the Municipal Treatment Utilities." Water Resources 31(6): 1299-1308. Gilbert, R.O. (1987). Statistical Methods for Environmental Pollution Monitoring. Van Nostrand Reinhold. Greiner, A.D. , et al. (1992). "Technical Note: Comparing Predicted and Observed Concentrations of DBPs." Journal of the American Water Works Association 84(11): 99-102. G V R D , (1995). Watershed Management. Protecting a Precious Resource. Burnaby, B.C. G V R D , (2000). Water - Frequently Asked Questions, www.gvrd.bc.ca. Health and Welfare Canada (1993). Guidelines for Canadian Drinking Water Quality. Jacangelo, J.G., et al. (1995). "Selected Processes for Removing Nom: An Overview." Journal of the American Water Works Association 87(1): 64-77. Jorgensen, T.A., et al. (1985). "Carcinogenicty of Chloroform in Drinking Water to Male Osborne-Mendel Rats and Female B6c3fl Mice." Fundamental and Applied Toxicology 5: 760-769. Kassam, K., et al. (1991). "Accumulation and Adsorption Capacity of PAC in a Slurry Recirculating Clarifier." Journal of the American Water Works Association 83(2): 69-78. Korshin, G.V., et al. (1997). "The Decrease of U V Absorbance as an Indicator of TOX Formation." Water Resources 31(4): 946-949. L i , C , et al. (1998). "Monitoring DBP Formation with Differential U V Spectroscopy." Journal of the American Water Works Association 90(8): 88-100. 81 REF EREN CES Malcolm, R.L. (1990). Factors to Be Considered in the Isolation and Characterization of Aquatic Humic Substances. Linkoping Proceedings; Humic Substances Conference. Millipore Corporation (2000). Lab Water Products Catalogue, www.millipore.com/labwater/products.nsf/docs/milliq. Najm, I.N., et al. (1994). "Evaluating Surrogates for Disinfection by-Products." Journal of the American Water Works Association 86(6): 98-106. Najm, I.N., et al. (1991). "Using Powdered Activated Carbon: A Critical Review." Journal of the American Water Works Association 83(1): 65-75. NCI, N.C.I. (1976). Report on Carcinogenesis Bioassay of Chloroform. Springfield, V A , National Cancer Institute. NTP, N.T.P. Eighth Report on Carcinogens. 2001. Pirbazari, M . , et al. (1992). "MF-PAC for Treating Water Contaminated with Natural And Synthetic Organics." Journal of the American Water Works Association 84(12): 95-103. Rayner-Canham, G., et al. (1989). Chemistry, Addison-Wesley Publishers Limited. Rebhun, M . , et al. (1997). "Formation of Disinfection by-Products During Chlorination of Secondary Effluent and Renovated Water." Water Environment Research 69(6): 1154-1162. Reckhow, D.A., et al. (1990). "Chlorination of Humic Materials: By-Product Formation and Chemical Interpretations." Environmental Science and Technology 24(11): 1655-1664. Rook, J.J. (1974). "Formation of Haloforms During Chlorination of Natural Waters." Water Treatment Examination 23: 234-343. Rook, J.J. (1976). "Haloforms in Drinking Water." Journal of the American Water Works Association 68(3): 168-172. Rook, J.J. (1977). "Chlorination Reactions of Fulvic Acids." Environmental Science and Technology 11: 478-482. Sawyer, C.N., et al. (1994). Chemistry for Environmental Engineering, McGraw-Hill, Inc. Siddiqui, M . , et al. (2000). "Membranes for the Control of Natural Organic Matter from Surface Waters." Water Resources 34(13): 3355-3370. 82 REFERENCES Stevens, A . A . , et al. (1976). "Measurement of Thm and Precursor Conentration Changes." Journal of the American Water Works Association 68(11): 546-554. Tan, L. and G. Amy (1991). "Comparing Ozonation and Membrane Separation for Colour Removal and Disinfection by-Product Control." Journal of the American Water Works Association 83(5): 74-79. Taylor, J.S., et al. (1987). "Applying Membrane Processes to Groundwater Sources for Trihalomethane Precursor Control." Journal of the American Water Works Association 79(8): 72-82. Thurman, E . M . (1985). Organic Geochemistry of Natural Waters. Martinus Nijhoff/Dr W. Junk Publishers. U.S. Army Corps of Engineers (2001). Adsorption Design Guide. Washington, DC, Department of the Army. Urfer, D., et al. (1999). "Modeling Enhanced Coagulation to Imrove Ozone Disinfection." Journal of the American Water Works Association 91(3): 59-73. Vickers, J.C., et al. (1995). "The Use of Membrane Filtration in Conjunction with Coagulation Processes for Improved Nom Removal." Desalination 102(1-3): 57-61. Vrijenhoek, E., et al. (1998). "Removing Particles and Thm Precursors by Enhanced Coagulation." Journal of the American Water Works Association 90(4): 139-149. Wells, W.W. and P.A. Chadik (1998). "Effect of Bromide Ion on Haloacetic Acid Formation During Chlorination of Biscayne Aquifer Water." Journal of Environmental Engineering 124(10): 932-938. Williams, D.T., et al. (1980). "Trihalomethane Levels in Candian Drinking Water." Environmental Science Resources 16(503). 83 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0063725/manifest

Comment

Related Items