USF Libraries
USF Digital Collections

Evaluation of the impact of membrane change at a membrane softening water treatment plant

MISSING IMAGE

Material Information

Title:
Evaluation of the impact of membrane change at a membrane softening water treatment plant
Physical Description:
Book
Language:
English
Creator:
Keen, Michael
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Reverse osmosis
Nanofiltration
Groundwater treatment
Blending
Process evaluation
Dissertations, Academic -- Environmental Engineering -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: At the water treatment plant in Dunedin, Florida, reverse osmosis membranes remove the hardness from groundwater sources. Reverse osmosis membranes remove salts, pathogens, and organics from the feed water but can create an aggressive permeate. The membranes strip most ions in the process and the resulting permeate, if not subjected to blending on post treatment, has a tendency to leach metals from lead and copper pipes in the distribution networks. To prevent such problems, the permeate needs to be blended with partially treated raw water or to be chemically treated to re-mineralize and add alkalinity back into the water. In the last decade nanofiltration treatment has gained an increasing foothold in the water treatment industry especially as a water softener.Although nanofiltration membranes also have a high removal rate for organics and pathogens, the separation process is more selective towards multivalent ions (e.g., Ca²⁺, and Mg²⁺) than monovalent (e.g., Na⁺) ions. Most membrane softening plants blend minimally treated raw water with the membrane permeate as a means to reduce the aggressiveness of the water. However, blending can cause issues with disinfection byproducts and pathogen re-introduction. With nanofiltration membranes, fewer mono-valent ions are rejected which creates a more stable permeate and can reduce the blended water ratio. Since it is unlikely that most plants that use membrane filtration for water softening will be able to stop blending entirely, any improvement or sustainability of water quality at a reduced blend ratio should be viewed favorably within the water treatment industry.The study evaluates three nanofiltration membranes: TFC-SR, NF-90, and ESNA1-LF in relation to the reverse osmosis TFC-S RO membrane currently in use at Dunedin. Water flux and salt rejection of the permeate water were compared using solutions of NaCl, MgSO₄ and CaCl₂. Since the Langelier Saturation Index (LSI) is one of the main tests of the blended finished water and is used to judge water quality prior to its release into the distribution system, this study created a 0%, 10%, 15%, 20%, 30%, and 100% blend ratio for each membrane to compare and contrast the change in the LSI. The TFC-SR membrane showed the most promise in lowering the blend ratio while improving the aggressiveness of the finished water by showing a lower rejection for divalent ions. The TFC-SR membrane also showed an improvement in the LSI relative to the other membranes over the total range of blend ratios.
Thesis:
Thesis (M.S.Env.E.)--University of South Florida, 2009.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Michael Keen.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 102 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002029663
oclc - 437033872
usfldc doi - E14-SFE0002961
usfldc handle - e14.2961
System ID:
SFS0027278:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Evaluation of the Impact of Membrane Change at a Membrane Softening Water Treatment Plant by Michael Keen A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Environmental Engineering Department of Civil and Environmental Engineering College of Engineering University of South Florida Date of Approval: April 10, 2009 Major Professor: Daniel Yeh, Ph.D. Jeffrey Cunningham, Ph.D. Vinay Gupta, Ph.D. Keywords: reverse osmosis, nanofiltration, groundwater treatment, blending, process evaluation Copyright 2009, Michael Keen

PAGE 2

Acknowledgements I would like to thank all of the employees at the D unedin Water Treatment Plant who were always willing to explain something or help out in any other way with a smile. I would particularly like to mention Paul Stanek and John Van Amburg. I would also like to show my appreciation to my lab mates for all of their help within and outside of the lab, Ana Lucia Prieto, Anh Tien Do, Caryssa Joustra, Dave Starman and Tim Ware. Special thanks to Russell Ferlita, Stev e Heppler, Michael Gerdjikian and George Sunderland for their exceptional help on this project. And for his support and help throughout this sometimes painful process, I would like to express my gratitude to Dr. Daniel Yeh. I would also like to thank my thesis committee members Dr. Jeffrey Cunningham and Dr. Vinay Gupta for their time and input. For their encouragement and support, I want to thank Olya Martysevic h and my parents Richard Keen and Cathy Tyler.

PAGE 3

i TABLE OF CONTENTS LIST OF TABLES..............................................................................................................v LIST OF FIGURES...........................................................................................................vi ABSTRACT.....................................................................................................................viii 1. INTRODUCTION..........................................................................................................1 1.1 Purpose.................................................................................................................... ..5 1.2 Research Objectives..................................................................................................5 2. PLANT OVERVIEW.....................................................................................................6 2.1 Pretreatment..............................................................................................................8 2.2 Post Treatment........................................................................................................11 2.3 Other Post Treatment..............................................................................................12 2.4 Concentrate Disposal..............................................................................................14 3. LITERATURE REVIEW.............................................................................................15 3.1 Membrane Filtration...............................................................................................15 3.2 High Pressure Filtration..........................................................................................15 3.3 Spiral Wound Membranes and Cross Flow Filtration............................................16

PAGE 4

ii 3.4 Comparison of Nanofiltration an d Reverse Osmosis Membranes..........................17 3.5 Nanofiltration Rejection Mechanisms....................................................................18 3.6 Nanofiltration Fouling............................................................................................22 3.6.1 Scaling..........................................................................................................23 3.6.2 Colloidal Matter...........................................................................................24 3.6.3 Organics.......................................................................................................26 3.6.4 Biofouling....................................................................................................26 3.6.5 Concentration Polarization...........................................................................27 3.6.6 Membrane Compaction................................................................................27 3.7 Blending..................................................................................................................28 3.7.1 Organics.......................................................................................................29 3.7.2 Disinfection Byproducts..............................................................................30 3.7.3 Chlorination.................................................................................................30 3.8. Scaling and Corrosion Prediction with LSI...........................................................31 3.8.1 Langelier Saturation Index (LSI).................................................................31 3.8.2 Lead and Copper Rule.................................................................................35 3.8.3 Stage 2 Disinfectants and Disinfection Byproducts Rule............................35 3.8.4 Long Term 2 Enhanced Surface Water Treatment Rule..............................36 4. METHODS AND MATERIALS..................................................................................37 4.1 Overview.................................................................................................................37 4.2 Membrane Materials...............................................................................................38

PAGE 5

iii 4.3 Flat Sheet Membrane System.................................................................................39 4.4 Phase I.................................................................................................................... .44 4.5 Phase II................................................................................................................... .46 4.6 Phase III..................................................................................................................47 5. RESULTS AND DISCUSSION...................................................................................49 5.1 Phase I.................................................................................................................... .49 5.2 Phase II................................................................................................................... .52 5.3 Phase III..................................................................................................................67 5.4 Implementation.......................................................................................................73 5.4.1 Membrane Properties...................................................................................73 5.4.2 Concentrate Disposal...................................................................................77 5.4.3 Operation and Maintenance.........................................................................77 5.4.4 Plant Reclassification...................................................................................79 6. CONCLUSION.............................................................................................................81 APPENDICES..................................................................................................................87 Appendix 1: Overview of the Dune din Well Water Collection System...........................88 Appendix 2: Diagram of the D unedin Water Treatment Plant.........................................89 Appendix 3: Overview of All the Meas urements and Locations at the DWTP................90

PAGE 6

iv Appendix 4: List of Each Wate r Quality Lab Test at DWTP...........................................92 Appendix 5: List of Drinking Wate r Monitoring Done at the DWTP..............................94 Appendix 6: Specification Sheet for KOCH TFC-S Membrane......................................95 Appendix 7: Specification Sheet for KOCH TFC-SR Membrane....................................96 Appendix 8: Specification Sheet for FILMTEC NF-90 Membrane.................................97 Appendix 9: Specification Sheet for HYDRANAUTICS ESNA1-LF Membrane...........98 Appendix 10: Historical Data of th e Dunedin Water Treatment Plant.............................99 Appendix 11: TDS vs. Conductivity Graphs..................................................................100

PAGE 7

v LIST OF TABLES Table 1: Water Hardness and TDS Categories.................................................................3 Table 2: Water Quality Data for 8/10/07 & 8/11/08.........................................................4 Table 3: Comparative Rejection Values.........................................................................21 Table 4: Potential Membrane Foul ing Sources and Control Strategies..........................23 Table 5: Operationa l Values at the DWTP.....................................................................37 Table 6: List of Membranes and Their Published Characteristics.................................39 Table 7: Membrane Permeability Coefficients and Resistance......................................56 Table 8: Percent Rejection Tables for TFC-S and TFC-SR...........................................58 Table 9: Percent Rejection Table for NF-90 and ESNA1-LF.........................................59 Table 10: Water Quality Data from LS I Versus Blend Ratio Experiments......................68 Table 11: Overview of the Measurements Performed at DWTP......................................90

PAGE 8

vi LIST OF FIGURES Figure 1: Pictures of RO Skid at the DWTP.....................................................................8 Figure 2: Pictures of Va rious Pretreatment Systems......................................................10 Figure 3: Water Flow Diagram.......................................................................................12 Figure 4: Pictures of Vari ous Post Treatment Systems..................................................13 Figure 5: Diagram of a Spiral Wound Membrane..........................................................18 Figure 6: Membrane Filtration Spectrum.......................................................................20 Figure 7: LSI Values vs. LSI Parameters........................................................................34 Figure 8: Separation System s Flow Cell Front and Back...............................................40 Figure 9: Overview Schema tic of Flat Sheet System.....................................................42 Figure 10: Overview of Flat Sheet Membrane System.....................................................43 Figure 11: DWTP Feed Pressure Over a Seven Year Period............................................49 Figure 12: TDS vs. Conductivity Composite Graph.........................................................51 Figure 13: TFC-S Intrinsic Water Flux Plot.....................................................................53 Figure 14: TFC-SR Intrinsic Water Flux Plot...................................................................53 Figure 15: NF-90 Intrinsic Water Flux Plot......................................................................54 Figure 16: ESNA1-LF Intrinsic Water Flux Plot..............................................................54 Figure 17: Percent Rejection vs. Transmem brane Pressure (TMP) (Per Membrane)......64 Figure 18: Percent rejection vs. Transm embrane Pressure (TMP) (Per Salt)...................65 Figure 19: Flux vs. Pressure Graphs (Per Salt-Membrane)..............................................66

PAGE 9

vii Figure 20: Composite Flux in Salt Solutions Per Membrane...........................................67 Figure 21: Percent Blend Ratio vs. LSI............................................................................69 Figure 22: Blend Ratio vs. LSI (@ 0.72)..........................................................................69 Figure 23: 1st and 2nd Stage Blend Ratios vs. LSI..........................................................72 Figure 24: Blend Ratio Relationship Overview................................................................76 Figure 25: Dunedin Well System Collection Map............................................................88 Figure 26: DWTP Plant Schematic...................................................................................89 Figure 27: Historical %Salt Removal Over Time.............................................................99 Figure 28: Historical Blend Flows....................................................................................99 Figure 29: TDS vs. Cond uctivity for 6/27/2008.............................................................100 Figure 30: TDS vs. Cond uctivity for 7/2/2008...............................................................100 Figure 31: TDS vs. Cond uctivity for 7/9/2008...............................................................101 Figure 32: TDS vs. Cond uctivity for 7/12/2008.............................................................101 Figure 33: TDS vs. Cond uctivity for 7/13/2008.............................................................102 Figure 34: TDS vs. Cond uctivity for 7/24/2008.............................................................102

PAGE 10

viii Evaluation of the Impact of Membrane Change at a Membrane Softening Water Treatment Plant Michael Keen ABSTRACT At the water treatment plant in Dunedi n, Florida, reverse osmosis membranes remove the hardness from groundwater sour ces. Reverse osmosis membranes remove salts, pathogens, and organics from the feed wa ter but can create an aggressive permeate. The membranes strip most ions in the process and the resulting permea te, if not subjected to blending on post treatment, has a tendency to leach metals from lead and copper pipes in the distribution networks. To prevent such problems, the permeate needs to be blended with partially treated raw water or to be chemically treated to re-mineralize and add alkalinity back into the water. In the la st decade nanofiltration treatment has gained an increasing foothold in the water treatment i ndustry especially as a water softener. Although nanofiltration membranes also have a high removal rate for organics and pathogens, the separation process is more selective towards multivalent ions (e.g., Ca2+, and Mg2+) than monovalent (e.g., Na+) ions. Most membrane softening plants blend minimally treated raw water with the membrane permeate as a means to reduce the aggressiveness of the water. However, blending can cause issues with disinfec tion byproducts and pathogen re-introduction. With nanofiltration membranes, fewer mono-va lent ions are rejected which creates a more stable permeate and can reduce the blende d water ratio. Since it is unlikely that

PAGE 11

ix most plants that use membrane filtration for water softening will be able to stop blending entirely, any improvement or sustainability of water quality at a reduced blend ratio should be viewed favorably within the wate r treatment industry. The study evaluates three nanofiltration membranes: TFC-SR, NF-90, and ESNA1-LF in relation to the reverse osmosis TFC-S RO membrane currently in use at Dunedin. Water flux and salt rejection of the permeate water were co mpared using solutions of NaCl, MgSO4 and CaCl2. Since the Langelier Satura tion Index (LSI) is one of th e main tests of the blended finished water and is used to judge water qual ity prior to its releas e into the distribution system, this study created a 0%, 10%, 15%, 20%, 30%, and 100% blend ratio for each membrane to compare and contrast the cha nge in the LSI. The TFC-SR membrane showed the most promise in lowering the blend ratio while impr oving the aggressiveness of the finished water by show ing a lower rejection for di valent ions. The TFC-SR membrane also showed an improvement in the LSI relative to the other membranes over the total range of blend ratios.

PAGE 12

1 1. INTRODUCTION The Dunedin Water Treatment Plant (DWTP), operated by the city of Dunedin, Florida, is a reverse osmosis (RO) water so ftening plant that curre ntly produces about 3.9 MGD (million gallons per day) of high quality water for the residents of the city. Groundwater taken from local wells which ta p into the Upper Floridan Aquifer (UFA) constitutes the source water. The raw water can be classified as hard to very hard with a hardness value of 160-190 mg/L CaCO3, and it has a low to moderate total dissolved solids (TDS) content of 580600 mg/L (Crittenden and M ontgomery Watson Harza, 2005). Hardness values are categorized in Tabl e 1. The hard water must be softened to reduce scaling throughout the distribution sy stem and in homes and offices. DWTP accomplishes this by removing the magnesium and calcium ions through RO membrane filtration. The raw water also has elevated levels of iron and manganese which needs to be removed via the greensand filter pretreatme nt process before the feed water reaches the RO membranes to prevent scaling since th ese ions can easily precipitate out of the water. Various water quality parameters ove r the past two years for the DWTP can be seen in Table 2. The DWTP currently has to blend its RO permeate in an 80/20 blend ratio with minimally treated raw water in order to minimize the aggressiveness of permeate when released into the distributi on system. Aggressive finished water can corrode the metal in the distribution system pipes, and the corro sion can cause serious health concerns if the metal is either lead or copper.

PAGE 13

2 However, the blending process has a poten tial to create prob lems of its own by increasing the likelihood of non-compliance with regards to disinfection byproducts (DBPs) such as haloacetic acids (HAAs) a nd trihalomethanes (THMs). The absence of RO membrane filtration on the bypass water means that some natural organic matter (NOM) in the raw water has the potential to serve as precursor to halogenated DBPs. DBP creation happens when NOM reacts with chlorine either in the chlorination prior to the raw water entry into the plant or after disi nfection in the post tr eatment. Studies have shown that about 25% of halogenated compounds formed are THMs and 18-20% HAAs (Reckhow and Singer, 1984; Fleischaker and Ramdtke, 1983). The Environmental Protection Agency (EPA) has placed limits on the amount of DBPs released to consumers and the environment. The limits were enacted because DBPs have been proven to be carcinogens linked to bladder and rectal cancers (Morris et al., 1992). Through the implementation of the Stage 1 Disinfection B yproducts Rule (Stage 1 DBP), the EPA has set maximum contaminant levels (MCLs) fo r total THMs and five HAAs. The agency set the MCLs at 80 parts per billion (ppb) for THMs lik e chloroform, bromoform, bromodichloromethane and dibromochlorom ethane. MCLs for monochloroacetic, dichloroacetic, trichloroacetic, monobromoace tic and dibromoacetic acids were set at 60 ppb (EPA, 1998). Another issue with blending minimally tr eated waters to meet finished water demands comes from the cost associated with chemically treating the blended water. Additional chlorine has to be added to pr operly disinfect the bl ended bypass water. Switching the DWTP from a RO to a nanof iltration (NF) membrane system has the

PAGE 14

3 potential to lessen blending needs by creati ng a less aggressive permeate while still meeting all the requirements under the EPAs Stage2 DBP rule. Membrane change can also have the added benefit of possibl y meeting the Long Term 2 Surface Water Treatment Rule (LT2) generally applied to pl ants treating surface water. Currently, the DWTP is classified as a groundwater trea tment plant. Under the current plant classification DWTP must test their groundwater wells for Escherichia coli which are quite costly. Being reclassified under th e LT2 would cut down on the required well monitoring under Florida Department of Environmental Protection (FDEP) rules governing the treatment of groundwater. If DW TP can get reclassified as a surface water treatment plant by the FDEP, it would only have to show the appropriate removal in the plants treatment train of certain pathogenic microorganisms like Cryptosporidium parvum to meet the rule requirements. This would save the DWTP the costs of monitoring the wells and let the plant maintain focus on the efficiency of the treatment processes. Table 1: Water Hardness and TDS Categories Hardness Range (mg/L of CaCO3) Soft 0 to <60 Moderately Hard 60 to <120 Hard 120 to <180 Very Hard >180 (Crittenden and Montgomery Watson Harza, 2005)

PAGE 15

4 Table 2: Water Quality Data for 8/10/07 & 8/11/08 Parameters Units Raw '07 Raw '08 Feed '07 Feed '08 Perm '07 Perm '08 Conc '07 Conc '08 Field Parameters: Specific Conductance umhos/cm 1005 1002 167 3840 Water Temp. C 25.1 25.1 25.3 25.4 pH 7.12 7.12 6.12 7.57 Inorganics Total Alkalinity as CaCO3 mg/L 160 190 120 180 10 23 470 930 Chloride mg/L 190 190 200 180 45 35 790 820 Fluoride mg/L 0.19 0.22 0.17 0.24 0.14 0.032 0.56 0.47 Nitrate (as N) mg/L 0.54 0.14 0.53 0.15 0.14 0.097 1.1 0.34 Sulfate mg/L 37 33 100 32 1.5 0.39 500 160 TDS mg/L 600 580 600 530 84 80 2600 2400 TOC mg/L 2 2.1 1.9 1.9 0.5 0.5 10 11 Total Phosphorus mg/L P 0.066 0.079 0.075 0.21 0.011 0.01 0.36 0.94 Turbidity NTU 0.85 0.95 0.1 0.05 0.15 0.05 0.1 0.1 Metals Barium mg/L 0.03 0.027 0.021 0.021 0.01 0.005 0.096 0.11 Calcium mg/L 90 96 100 93 6 4.6 480 460 Iron mg/L 0.61 0.43 0.02 0.02 0.02 0.02 0.02 0.02 Iron, Dissolved mg/L 0.1 0.41 0.02 0.02 0.02 0.02 0.02 0.02 Potassium mg/L 4.1 4.6 3.8 4.9 1.1 1.2 14 13 Magnesium mg/L 14 13 14 13 0.76 0.95 68 65 Manganese mg/L 0.02 0.018 0.01 0.019 0.01 0.01 0.041 0.095 Sodium mg/L 94 80 100 76 29 22 340 300 Dissolved Silica as SiO2 mg/L 26 27 25 27 6.3 6.1 110 115 Strontium mg/L 0.28 0.29 0.29 0.3 0.019 0.016 1.5 1.6 Data provided by Southern Analytical Laborator ies, Inc. 110 Bayview Blvd. Oldsmar, Fl 34677 Note: Raw = raw well water; Feed = water afte r pretreatment and an ti-scalant injection going to the RO membrane; Perm = memb rane permeate; and Conc = membrane concentrate.

PAGE 16

5 1.1 Purpose This project will look into the replacement of the current RO membranes with NF membranes at the Dunedin Water Treatment Pl ant in Dunedin, Florida. The study will focus on maintaining and improving water qua lity, creating non-aggressive finished water, and increasing savings in plant operations Using different blend ratios from three different NF permeates, this project hopes to show that effluent quality as defined by a corrosivity and scalability i ndex ( Langelier Saturation Index LSI) can be maintained or improved, and costs can be reduced. 1.2 Research Objectives The objectives of this research are: 1. To quantify the effectiveness of three NF membranes compared to the membrane currently being used in te rms of the cascadi ng impact on plant operations, blend ratio, finished water quality, and possible plant reclassification. 2. To analyze finished water quality as a function of the ble nd to permeate ratio using different NF membranes with respect to plant operations and the potential for plant reclassification.

PAGE 17

6 2. PLANT OVERVIEW The DWTP receives its raw water from a group of wells in Pinellas County. See Appendix 1 for an overview of the plants 21 wells and raw water collection system. The source water comes from Zone A of the Upper Floridan Aquifer (Carnahan et al., 1995). The shallowest and freshest permeable area, Z one A has an average depth of 180 ft with a range of 115 to 250 ft (Broska and Barnette, 1999). The plant is located at 1401 County Road 1, Dunedin, Florida, which is southwest of Tampa. The plan t has been operational since 1992. The finished water distribution system has over 7.5 miles of transmission piping, four 2-million gallon ground storage tanks, a nd approximately 138 miles of distribution piping (Dunedin, 1992). Curre ntly the DWTP can produce 9.5 MGD, but has been permitted by Southwest Florida Water Manage ment District (SWFWMD) for 6.6 MGD. From a peak demand of 4.7 MGD in 1998, the ye arly demand has steadily decreased over the years as stricter water conservation (due to increase in water rates) and a higher demand for reused water from the Dunedi n Wastewater Treatment Plant (DWWTP) began to affect the local water consump tion. The recent average daily demand falls below 3.9 MGD. During the daily operation of the DWTP, the pl ant operators take measurements of various water parameters by which they assess the quality of different streams and judge the efficiency of the various treatment trains. An overview of all the measurements taken at the various plant locations is shown in Appendix 3.

PAGE 18

7 The RO treatment train comprises four tw o-stage skids. The first stage includes twenty six pressure vessels and the second stage has thirteen. Each pressure vessel contains seven membrane elements. Each membrane element is a Koch TFC 9921-S polyamide spiral wound module (8 diameter 40 length). The 8 elements have been phased out of commercial production and replaced by the now common 8 diameter element. Any new 8 membrane modules will have to be specially made by the manufacturer. According to Rick Lesan, an R&D engineer with Koch Membrane Systems, brine seals can be placed on the smalle r 8 elements and made to fit in the 8 pressure vessels. During the two stage proce ss, about 75% of the f eed water is converted into permeate. The first stage recovers 50%, and another 50% of the first stage concentrate is recovered. Pictures of the RO skids and other plant components are shown in Figure 1 below. Currently, at the DWTP, the raw water is pre-treated by four processes before reaching the RO membranes. The processes are (in order): pre-chlorination, greensand filtration, cartridge filtration (5 micron cartridge filters), and anti-scalant injection. After cartridge filtration, some of the water bypasses the anti-scalant and RO processes so it can be blended with the RO membrane perm eate. Following the RO membrane process, the water is subjected to five post-treatm ent processes: blending, degasification for CO2 removal, fluoridation, pH adjustment, and disinfection for chlorine residual in distribution system. The RO concentrate is sent directly to th e Dunedin Wastewater Treatment Plant for disposal. An overvie w DWTP schematic can be found in Appendix 2.

PAGE 19

8 Four Skids at DWTP Blue Pressure Vessels on Skid 4 Spiral Wound Membrane Element End View of a Skid Figure 1: Pictures of RO Skid at the DWTP 2.1 Pretreatment Chlorine is added to the raw water to help complete the oxidation of hydrogen sulfide (H2S) which causes bad odors in water. Some treatment centers have established greensand filters as an effective method of sulfide removal (Boyle, 2005). By performing pre-chlorination on the raw groundwater, the DWTP improves the greensand filters by

PAGE 20

9 removing the initial oxidation demand with chlo rine instead of potassium permanganate which is added specifically for the greensa nd process (Boyle, 2005). After potassium permanganate is added to the water and has ti me to react with the greensand can oxidize, filter, and adsorb the contaminants (Boyle, 2005). Currently, two of the five greensand filters use a manufactured greensand called Greensand Plus TM. The other three use conventionally mined glauconite greensand. Both types of greensand have similar performance traits (Boyle, 2005) These traits include the oxidation of iron, manganese and sulfide, avoidance of THMs or HAAs production, minimization of turbidity and sulfide oxidizing bacteria, and reduction of a por tion of the color content of the raw water (Boyle, 2005). The next step is cartridge filtration (nominal 5 m) which removes particulates to protect the RO membranes downstream against imp action or deposition. The anti-scalant (polyacrylic acid Gene ral Electrics Betz Hypersperse MDC 700) injections reduce the scaling of the RO membranes caused by certain carbonate and sulfate compounds by allowing the foulants poten tial to exceed their solubility constant without precipitating out of solution. In Figure 2 be low, various sections of the pretreatment process are shown.

PAGE 21

10 Greensand Filters Cartridge Filters Feed Water Pumps Just Af ter Anti-scalant Injection Figure 2: Pictures of Vari ous Pretreatment Systems

PAGE 22

11 2.2 Post Treatment The first post-membrane treatment process is blending of RO permeate with water that bypasses the anti-sc alant injection and the RO membra nes. The blend is comprised of 80% permeate and 20% bypass water. Th e 80/20 blend ratio allows the bypass water to remineralize the permeate which has had most of the minerals and alkalinity removed during RO treatment. However, CO2 in the feed water is no t removed by RO due to its small size and neutral charge, and it has to be taken out at th e degasification post treatment stage (Schaefer, 2005). A diagram of the bypass water and permeate flows is shown in Figure 3. The bypass water splits o ff from the feed water after the cartridge filter and before the anti-scalant in jection. It combines with the 1st and 2nd stage permeate right after the RO process but before degasification. Blending stabilizes the aggressive water at a lower cost than injecting chemicals into the water. The blended water undergoes daily tests for corrosiveness and scalability as measured by the Langelier Saturation Index (LSI ). The LSI tool measures the potential of the water to form chemical scale and its abil ity to corrode the pipes in the distribution system. A positive LSI value means the wate r has the potential to form scale, and a negative value describes the corrosive nature of the water. The plant operators perform the test on the finished water in the storag e tanks, on the clearwell tanks, and on water collected from the farthest poi nt in the distributio n system. The types and quantity of the plant measurements, the location of the sa mples taken, and whether the water quality tests are performed in-house or by outside la boratories are shown in Appendices 3 and 4. Five water variables are needed to calculate the LSI. The variables are temperature,

PAGE 23

12 calcium hardness, total alkalinity, total disso lved solids, and pH. Both calcium hardness and total alkalinity are in terms of calcium carbonate. The DWTP currently aims for a slightly positive finished wa ter LSI. This positive numbe r means the water will more likely scale than corrode. Figure 3: Water Flow Diagram 2.3 Other Post Treatment The blended water goes through a series of post-treatment processes as seen in Appendix 2. The post-treatment train includes degasification, fluoridation, the injection of sodium hydroxide for pH adjustment, a nd chlorination. Some of the post treatment systems are shown in Figure 4 below. De gasification removes any residual hydrogen sulfide and CO2. Carbon dioxide is found in groundw ater and may also have been formed if the pH was lowered enough due to the injection of the anti-scalant prior to

PAGE 24

13 membrane treatment. If the pH was lowere d significantly it coul d cause the carbonate within the water to change to carbon dioxide. Fluoride in the form of hydrofluosilicic acid is injected into the water to help pr omote healthy teeth and reduce cavities. The adjustment of pH before the effluent reaches the distributi on system is to stabilize the water for public consumption. Chlorine inj ection disinfects any bi ological contaminants within the water. Pathogen s are usually removed through the membrane process but may be reintroduced when the permeate is blended with unfiltered water. Enough chlorine is added to create a residual disinfection throughout the distribution system. However, high doses of chlorine can also lead to the fo rmation of THMs or HAAs when it comes into contact with certain DBP precursors ( NOM) found in the bypass water. Degasification Towers Chemical Storage Tanks Figure 4: Pictures of Various Post Treatment Systems

PAGE 25

14 2.4 Concentrate Disposal In Dunedin, the concentrate of the DWTP flows by a direct pipeline to the Dunedin Wastewater Treatmen t Plant (DWWTP). Before th e concentrate leaves the DWTP, it undergoes a pH adjustment with so dium hydroxide to around 8.4. Because of the thorough pretreatment of the feed water in the greensand filters, we can assume that when concentrate gets to the DWWTP, it ha s low concentrations of iron (Fe), and manganese (Mn) as verified in Table 2.

PAGE 26

15 3. LITERATURE REVIEW 3.1 Membrane Filtration There are several types of membrane filtr ation currently used in municipal water treatment plants. Each type of membrane can be loosely defined by the types of material rejected (Crittenden and Montgomery Wa tson Harza, 2005). Although rejection mechanisms for the different types of memb ranes can be quite different, each of the membranes uses pressure to produce the permeate (Crittenden and Montgomery Watson Harza, 2005). Each membrane type uses the differences in permeability (of water constituents) as a separation mechanism (B aker, 2004). During the membrane process, water is pumped touching the surface of the membrane resulting in permeate and concentrate streams. The membrane material is designed to be highl y permeable to some components of the feed stream while being less permeable to others (Crittenden and Montgomery Watson Harza, 2005). During the filtration process, low permeability constituents of the solution stay on the feed side of the membrane while more permeable ones are passed through the membrane. The re sulting product stream is relatively free of impermeable constituents (Crittenden and Montgomery Watson Harza, 2005). 3.2 High Pressure Filtration The main difference between low and hi gh pressure filtration is the removal mechanisms. In low pressure filtration (microfiltration or ultrafiltration), the removal mechanism relies solely on particle size ex clusion (Crittenden and Montgomery Watson

PAGE 27

16 Harza, 2005). On the other hand, high pressure filtration (reverse osmosis or nanofiltration) relies mainly on diffusion and to some degree on size exclusion in the case of nanofiltration (Schaefer et al., 2005). In high pressure diffusion membranes, the water is separated from the solution by overcoming th e osmotic pressure within the solution. In these membranes, greater pressures are need ed to overcome the hi gher osmotic pressures in different solutions (Baker, 2004). In th e water treatment industry, reverse osmosis membranes are usually used to produce potab le water from saline or brackish waters (Crittenden and Montgomery Watson Harza, 2005). The nanofiltration membranes are generally used to soften hard water a nd freshen brackish water (Crittenden and Montgomery Watson Harza, 2005). Most of the high pressure membranes in drinking water treatment use a spiral wound design which enable cross-flow filtration (Schaefer et al., 2005). 3.3 Spiral Wound Membranes and Cross Flow Filtration In Figure 5, a detailed schematic of a spiral wound module (SWM) shows an internal view of the different layers that ma ke up the membrane element. Multiple leaves comprised of membrane sheets, feed channe l spacers and permeate collection material wrap around a central permeate tube. The me mbranes are glued on three sides with the fourth side providing the opening toward the f eed flow (Schwinge et al., 2004). Since the feed flows over the surface of the membrane, the filtration system is termed cross flow filtration (CFF). The feed cha nnel spacers act to separate the membrane leaves and cause interference to the feed flow, which helps the water to become turbulent and keeps fouling of the membrane surface down (Schwinge et al., 2004; Baker, 2004). As the

PAGE 28

17 water transfers across the membrane it then travels spirally around the permeate material and exits out of the porous permeate tube in the center (Schwinge et al., 2004). The SWM maximizes the active surf ace area of the membranes while reducing the size of the system which allow water treatment plants to keep their footprints small. 3.4 Comparison of Nanofiltration and Reverse Osmosis Membranes Reverse osmosis membranes accomplishes the separation of dissolved solutes from water without regard to valence char ge (Crittenden and Montgomery Watson Harza, 2005). RO can effectively remove most constitu ents from water, but is not selective in the removal (Schaefer et al., 2005). Unlike ty pical RO membranes, NF membranes have the ability to selectively reject certain elec trolytes and low molecular weight dissolved constituents (Bartels et al., 2008). Created during the 1960s, NF membranes are mainly used to soften water because they have the ab ility to selectively reje ct those ions like Ca2+ and Mg2+ that are the main causes of hardness (S chaefer et al., 2005). In Florida, there are many water treatment plants that use NF membranes to soften their groundwater, such as, Deerfield Beach, Hollywood, and Boca Raton. At these plants, the NF membranes have been packed in spiral wound modules (S WM). The NF membrane plants that treat hard water in Florida use arrays of SWM in pa rallel and in series to meet their permeate demands. Multiple membranes sit inside of pres sure vessels which connect to each other. Groups of pressure vessels connected in para llel are usually called stages. Stages can either be connected in parallel or in series depending on the permeate needs.

PAGE 29

18 Figure 5: Diagram of a Spiral Wound Membrane (Based on Koch Membrane Systems spiral wound membrane diagram) 3.5 Nanofiltration Rejection Mechanisms Nanofiltration (NF) uses pressure to separate the solutes from the solution. The effective pressure (Pe) is the difference in the change of operating pressure (P) and the change in the osmotic pressure ( ) from the concentrate to the permeate side of the membrane. Many phenomena can describe the transport of solutes across the membrane, but Pe is the driving force for water flux. Pe = ( P ) (1)

PAGE 30

19 NF membranes have been termed loose RO membranes or tight ultrafiltration (UF) membranes, but their solute removal m echanisms are uniquely different from either RO or UF (Sharma and Chellam, 2006; Sch aefer, 2005; Bartels et al., 2008). According to Schaefer et al. (2005), NF membranes have three unique prope rties that set them apart. These distinctive properties are a high rejection of negatively charged multivalent ions, varied rejections of sodium chloride, and a rejection of non-charged, dissolved materials and positively charged molecules based on size and shape. To accomplish this range of rejection, NF membranes apply both the sievi ng (steric hindrance) effect and the Donnan (electrostatic) effect (Wang et al., 2002; Schaep et al., 1999). See Figure 6 for the different membrane filtration spectrums and Table 3 for a list of comparative rejection values for different membrane types.

PAGE 31

20 Figure 6: Membrane Filtration Spectrum (Based on Osmonics Inc. spectrum chart)

PAGE 32

21 Table 3: Comparative Rejection Values Species RO Loose RO NF UF Sodium Chloride 99% 70-95% 0-70% 0% Sodium Sulfate 99% 80-95% 99% 0% Calcium Chloride 99% 80-95% 0-90% 0% Magnesium Sulfate >99% 95-98% >99% 0% Humic Acid >99% >99% >99% 30% Virus 99.99% 99.99% 99.99% 99% Bacteria 99.99% 99.99% 99.99% 99% (Based on a similar table (Schaefer, 2005), fr om Bjarne Nicolaisen of Osmonics, Inc.) The leading method used to describe the solute removal mechanism of NF membranes comes from the Donnan-steric partitioning pore mode l (DSPM) (Bowen et al., 1996; Schaep et al., 2001; Labbez et al., 2002; Labbez et al., 2003; Bandini and Vezzani, 2003). In the DSPM, the NF membra ne is considered a charged porous layer and takes into account three pa rameters: effective pore size, effective ratio of membrane thickness to porosity, and effective charge density (Peeters et al., 1998; Bandini and Vezzani, 2003; Mohammad and Takriff, 2003). However, the DSPM model has a problem in predicting the rejection of divalent ions (Vezzanni and Bandini, 2001; Schaep et al ., 2001). To help in the predictive model, dielectric exclusion (DE) portioning has been used to explain the high rejections encountered in such divalent ions as Mg2+ (Schaep et al., 2001; Bandini and Vezzani, 2003).

PAGE 33

22 The DE model is based on the difference betwee n the dielectric consta nt of the membrane and the bulk solution (Bandini and Vezzani, 2003). In DE, the separation mechanism does not ta ke into account the charge of the ion. The dielectric constant is the expression used to identify the degree that a material will concentrate electric flux (Bandini and Vezzani, 2003). Electric flux is the movement of charge through a material. The differing electr ostatic fields cause an interaction between the ions and the polymeric surface in which the dielectric constant of the aqueous solution is much higher than the surface. At the boundary between these two fields, the ions cause a charge of the same polarity as the reference ion thus repelling the charged ions independent of its sign. 3.6 Nanofiltration Fouling Outside of costs associated with the pressure required for membrane filtration one of the biggest problems encountered during me mbrane treatment comes from the constant fouling of the membranes. This causes d ecline of permeate flux and loss of product quality (Baker, 2004). Koros et al. (1996) defi ned fouling as the process resulting in loss of performance of a membrane due to deposition of suspended or dissolved substances on its external surfaces, at its pore openings, or within its pores. Schaefer et al. (2005) lists some of the causes and cont rol strategies of membrane fouling, as summarized in Table 4. Any type of fou ling on the membrane can lead to reduced recovery, higher operational costs, higher en ergy demand, increase of cleaning frequency, and a reduction in the useful life of the membrane element (Vrouwenvelder et al., 2003; Manttari et al., 1997; Bo nne et al., 2000).

PAGE 34

23 As seen in Table 4, most of the operational co ntrolling of fouling occurs before the water makes contact with the membrane. The preven tive treatment of the raw water is the key to limiting fouling of the membranes. Table 4: Potential Membrane Fouling Sources and Control Strategies (Schafer et al., 2005) Origins of Fouling Fouling Control Scaling: substances exceeding their solubility produ ct Operate below solubility limit, pretreatment: reduce pH to 4-6, low recovery, and anti-scalants. Pre-oxidation of metals. Deposition of colloidal matter or dispersed fines Pr etreatment using filtratio n, microfiltration (MF) or Ultrafiltration (UF) Organic fouling Pretreatment using filtration, MF, UF, ion exchange, ozone, enhanced coagulation or carbon adsorption Biofouling: colonization by bacteria Hydr odynamics, operation below critical flux, chemical cleaning, pretr eatment: disinfection or UF, MF, Hydrodynamics, operation below critical flux, chemical cleaning 3.6.1 Scaling Calcium carbonate, calcium sulfate, silica complexes, barium sulfate, strontium sulfate and calcium fluoride have been identifi ed as leading causes of scale formation on membranes (Baker, 2004). Scaling occurs when the concentration of one of these species exceeds the solubility c onstant and starts to precipitat e out of the solution onto the membrane. Scaling can be greatly affected by pH, temperature, fluid velocity, time and salt concentration in the concentrate (S chafer, 2005). Certain cations like Mg2+ and Ca2+ can increase the precipitation and colloidal formation of silica complexes

PAGE 35

24 (Sheikholeslami and Bright, 2002). It has also been noted in the same study that iron and manganese even at low concentrations can also increase the fou ling potential of silica compounds. However, the greensand filtration at the DWTP removes the majority of the iron and manganese during the pretreatment phase. Since groundwater results from the flow of surface water through different types of sediment it becomes naturally mineralized and can sometimes have significant levels of scale forming species. The DWTP ha s a high scaling potential because of the concentration of certain minerals in the water such as Mg2+ and Ca2+ which the plant was designed to remove. Nederlof et al. (2000) studied differe nt pretreatment methods for controlling membrane fouling and concluded th at scaling must be addressed with the addition of anti-scalants or pH adjustment. At one time in DWTP, anti-scalant was added and the pH of the feed water was adjusted but over time the plant operators have discontinued the process. W ithout pretreatment of the wa ter, plant operators could reduce the recovery of the membranes to cont rol scaling (Schafer et al., 2005). At a reduced recovery, the likelihood of a critical buil dup in scale forming species would be lowered because the concentrate would not be likely to reach a supe rsaturated state. However, many water treatment plants need to maintain a certain recovery to meet their areas water demand, and th erefore use chemical pretreatment methods. 3.6.2 Colloidal Matter Particles defined by their small size, stat e of hydration, and surface charge make up the foulant group known as fine colloid s (Viessmann et al., 2009). The negative surface charge of the suspended particulate keeps them from aggregating and falling out

PAGE 36

25 of the solution (Viesmann et al., 2009, Schafe r et al., 2005). Thes e charges affect the particulates through electros tatic double layer (EDL) interactions. The negatively charged particulates attract a covering of positively charged ions by an electrostatic attraction, and the stable layer of positive ions is surrounded by a movi ng diffuse zone of counterions. The attraction between ions is reduced in the diffusive zone the further away the ions roam from the stable laye r (Viessmann et al., 2009). The EDL of two similarly charged particles will repel each othe r, and with proximity the repelling force increases (Sawyer et al., 2003). Water with high ionic strength has the potential to compact the EDL thus increasing the ability of the particles to get together and at a certain point in the process the Van der W aals force can overcome the EDL repulsion and let the colloids form aggregates and settle on membrane surfaces (Sawyer et al., 2003). The Van der Waals force is the intermolecula r attractive force which all particles possess in varying strengths according to their composition and density. A strong enough crossflow velocity in the membrane treatment syst em can create turbulent flow and keep much colloidal matter from depositing on the membranes. In a membrane water treatment plant, the cross-flow velocity decreases as the water flows through th e pressure vessels. The reduction in cross-flow velocity closer toward the end membrane element means the water flow can become more laminar, which will increase the susceptibility to chemical or colloidal fouling (Gwon et al., 2003). The prevalent met hods of reducing any colloidal fouling at the DWTP are the application of cartridge filtration before the feed pumps and maintaining a strong cross flow velocity.

PAGE 37

26 3.6.3 Organics Studies have shown that humics, nonhumics, polysaccharides and proteins dominate organic membrane fouling (Violl eau et al., 2005). Orga nic material comes from human activities, natural organic matter (NOM), or compounds formed during disinfection processes. It may also be formed through the addition of compounds during the transmission or treatment of water (C rittenden and Montgomery Watson Harza, 2005). Since much of the water at the DWTP is treated minimally before it gets to the membranes and the source wells are well protected, the majority of the organic matter in the DWTP comes from NOM. According to Schafer et al. (2005), the NOM can form a gel on the surface of the membrane through ad sorption. NOM can also build up a cake layer through deposition by or ganic colloids or restrict the pores once the organic molecules have penetrated the membra ne. Total organic carbon (TOC) commonly measures the concentration of NOM in the wa ter. Like most groundwater, the raw water at DWTP has a low concentration of TO C as shown in Table 2. Although organic fouling can occur in municipal water treatmen t facilities, it is more common in other membrane applications such as industrial processes where RO membranes are used to treat a process stream (Baker, 2004). 3.6.4 Biofouling Biofouling (biological fouling) is the growth of biologi cal organisms on either the permeate or concentrate side surface of the membrane (Bak er, 2004). In water treatment facilities that use membranes, biofouling is hard to control becau se fouling can occur with only a few viable bacteria and can feed o ff of any organic materi al (usually a steady

PAGE 38

27 supply in the feed water) f ound in the water including dead bacteria (Flemming, 2002). At DWTP, studies by Carnahan et al. (1995) found that there was enough organic matter in the raw water to support Pseudomonas bacteria. Once attached to the surface of the membrane, biofilm is very hard to remove because the organisms excrete extra-cellular polymeric substances (EPS) that form a protective medium and adhesive for the microorganisms (Carnahan et al., 1995; Fl emming, 2002). At DWTP biofouling of the membranes tends to occur more heavily on the feed side of the membrane element because the majority of the NOM and bact erium are removed by the membrane (Sagiv and Semiat, 2005). DWTP uses cleaning protoc ols to remove the buildup of biofoulants. 3.6.5 Concentration Polarization Concentration polarization can significantly affect the operation of NF and RO membranes. Feed water at the DWTP has many constituents that make up the waters characteristics. Because these constituents permeate at differing rates, gradients of concentration can form on either the perm eate or concentrate side of the membrane surface in a process called concentration pol arization (Baker, 2004). In addition, the ions collecting at the boundary can change the osmotic pressure of the solution thus decreasing the water flux, but the placement of feed spacers and a significant cross flow velocity can mitigate the degree of concen tration polarization (S chafer et al., 2005). 3.6.6 Membrane Compaction Although membrane compaction does reduc e the water flux of a membrane, it should not be confused with fouling (Bert, 1969; Schafer et al., 2005). As pressure increases within a membrane filtration proce ss, water will travel through the membrane.

PAGE 39

28 According to Berts research, a newly created membrane lacks the ability to retain most of the water within its matrix as the water passes through and over time the increase in pressure used on a RO or NF membrane fo rces out the water reducing the membranes hydration. As water is forced out of a partic ular area in the membrane matrix, it affects the permeability of the membrane because wa ter flux is the movement of water through the membrane and a reduction in water content at any point leads to an overall reduction of permeability (Bert, 1969). To overcome th is issue, this study setup a pre-compaction routine to temper the membranes before any experiments by running water through a membrane at a high enough pressure and timescale (Schafer et al., 2005). 3.7 Blending As stated in Chapter 2 Plant Overview once the feed water passes through the membrane system, many water treatment plants (such as DWTP) will blend the water with minimally treated raw water known as bypass water. The blending ratios depend on the constituent characterization of the bypass a nd permeate waters. Other factors that could influence the blending ratio are wate r recovery needs, production costs, and regulatory constraints. Because a RO or tight NF filtration process can strip most everything out of the feed water leaving it ve ry aggressive, a good bl end will alleviate the amount of chemicals needed for treatment to correct for corrosivity or scaling if the pretreatment is sufficient. A proper bl end will reduce the costs associated with chemically treating the water by remineraliz ing it. Along with a slightly positive LSI, remineralization includes an increased bicarbon ate alkalinity and pH for the treated water (Withers, 2005). However, the blend has cer tain drawbacks such as introducing NOM

PAGE 40

29 back into the finished water which can lead to DBPs once chlorine is added for disinfection. DBP formation can cause problems when tryi ng to meet regulatory limits set by the EPA. 3.7.1 Organics The origin of NOM is complex and varied. NOM is derived from multiple sources in the natural environment including secretions from the metabolic activity of organisms (Crittenden and Montgomery Wats on Harza, 2005). NOM can also develop from the decay of organic matter or from excretions of life forms (Crittenden and Montgomery Watson Harza, 2005). Basically, NOM comprises four different types of organic matter: carbohydrates, lipids, amino acids or nucleic acids, and the products of abiotic and biotic reactions between other NOM or inorganic mol ecules (Crittenden and Montgomery Watson Harza, 2005). Humic s ubstances are a major component of NOM (50-80% of dissolved organic matter), a nd are known DBP precursors (Thurman, 1985; Chadik and Amy, 1983). Being very complex, NOM has usually been measured with a bulk indicator like total orga nic carbon (TOC) (Crittende n and Montgomery Watson Harza, 2005; Dalvi et al., 2000). The effectiven ess of NF in the removal of the type of NOM that acts as a DBP precursor has been documented (Smith et al., 2002; Chellam et al., 2000). Taylor et al. (1987) found that RO did not remove NOM precursors significantly more effectively than NF membra nes, but required greater pressure and had a reduced flux.

PAGE 41

30 3.7.2 Disinfection Byproducts Factors including TOC, bromide ion c oncentration, pH, temperature, ammonia concentration, and carbonate alkalinity aff ect the types and con centrations of DBPs (Garvey et al., 2003). Some of the most common disinfectants (chlorine, ozone, chloramines) used in drinking water crea te their own DBPs (Richardson, 1998). The EPA first regulated DBPs in 1979 with the THM rule and in 1998 it introduced the Stage 1 Disinfectants/Disinfection Byproducts ru le (Stage 1 D/DBP Rule) (U.S. EPA 1979, 1998). This rule created and adjusted maxi mum contaminant levels (MCLs) for certain known DBPs. In 2003, the EPA added the Stag e 2 Disinfectants/Disinfection Byproducts Rule (Stage 2 D/DBP Rule) which specifies that utilities will have to meet MCLs calculating a yearly average at the complian ce monitoring station instead of a yearly average over the whole network (U.S. EP A 2003, 1998). DBPs have been linked to certain cancers in animals and humans, and st udies suggest that th e exposure routes in humans can be through ingestion, inhalati on and dermal absorption (Lavoie, 2000; Aggazzotti et al., 1998; Xu et al., 2002). At the DWTP, chlorine is the only disinfection chemical added to the water after blending. 3.7.3 Chlorination DBP formation is also usually dependen t on chlorine dose rates and contact time (Dalvi et al., 2000). There are several reasons why chlori ne disinfection remains popular even though it can cause DBPs. The chlorina tion fact sheet affirms chlorination as a useful disinfection process and states that it is a well established technology. The fact sheet states that chlorination is presently more cost effective than other disinfectants in

PAGE 42

31 most cases, it can prolong protection throughout the distri bution system, and it offers flexible dosing control (U.S. EPA, 1999). Ho wever, the EPA also lists several drawbacks to chlorination like increased chloride conten t. In high chlorine demand systems higher chlorine concentrations are needed. Cryptosporidium parvum and Giardia lamblia have shown resistance to chlorine and long term chlorination e ffects on the environment are unknown. 3.8. Scaling and Corrosion Prediction with LSI 3.8.1 Langelier Saturation Index (LSI) The LSI measures a solutions ability to dissolve or deposit calcium carbonate and has been used in the water industry to predict waters tendency to ei ther corrode or scale (Gebbie, 2000). Both corrosion and scaling are factors that affect th e public health, and corrosion products that leach off of distribut ion pipes can shield microorganisms from disinfectants (Melidis et al., 2007). The speciation in water of the carbonate system is directly dependent on pH (Crittenden and Montgomery Watson Harza, 2005; Langelier, 1936). At the DWTP, the plant op erators try to maintain a sl ightly positive LSI number. A small amount of scale on the surface of the pipes can shield the pipe material from water thus giving it a certain amount of pr otection against corros ion. The reactions between calcium and carbonates are the prim ary focus of the LSI (Langelier, 1936; Withers, 2005). According to Langelier, the index is the difference between the pH of the solution and the pHs (pH of saturation). The pH of saturation is the equilibrium pH once all forms of alkalinity have been adjusted so that water is only saturated in calcium carbonate (Langelier, 1936; Withers, 2005). A negative number represents a corrosive

PAGE 43

32 nature. A positive number means that the wate r has the ability to scale in the form of calcium carbonate, while zero indicates that the water is balanced. The larger the positive or negative number the greater its ability to cr eate scale or corrode (Langelier, 1936). The following equations calculate the LSI. LSI = pH pHs (2) pHs = (9.3+A+B)-(C+D) (3) A = (log10(TDS)-1)/10 (4) B = -13.12*log10( C+273)+34.55 (5) C = log10(Ca2+ as CaCO3 mg/L)-0.4 (6) D = log10(Alkalinity as CaCO3 mg/L) (7) According to Equations 3-8, TDS, tota l alkalinity, calcium hardness, pH and temperature affect the outcome of the LSI valu es. The variables with in the LSI equations will have differing degrees of influence on the calculated outcomes. Below in Figure 7, a specific range of LSI values was compared to the individual variab les while keeping the other parameters constant. The LSI values in the figure were calcula ted with four of the five following constants: temperature at 25 C, pH at 7, TDS at 550 mg/L, total Alkalinity

PAGE 44

33 at 150 mg/L as CaCO3, and calcium hardness at 140 mg/L as CaCO3. The resulting graphs give an idea of how much influence the individual variables have in the outcome of the calculated LSI values. According to the graphs, the least influential parameter is TDS as it can fluctuate over many magnitude s of values while only minimally changing the LSI value. Unlike the other parameters, TDS has an inverse relationship with LSI in that at lower values the TDS will produce a mo re positive LSI. Temperature and pH each have linear relationships with th e LSI. As the values raise so does the LSI. However, pH has a greater influence since it can change th e LSI to a greater degree by only fluctuating within a small pH range. The temperature produces a measured change in LSI as it increases. Within normal operating temperature range of 20 C -25 C, the temperature will only minimally change the LSI. Finally, the calcium hardness and total alkalinity have a logarithmic relationshi p with LSI. Both variable s have a greater degree of influence at the lower concentrations, but th eir ability to significantly change the LSI lessens at higher concentrations.

PAGE 45

34 LSI vs. Termperature y = 0.0191x 1.0909 1.20 1.00 0.80 0.60 0.40 0.20 0.00 01020304050 Temperature (C)LSI LSI vs. TDS y = 0.0434Ln(x) 0.3332 0.65 0.64 0.63 0.62 0.61 0.60 0.59 200 400 600 800 1000 1200 1400 TDS (mg/L)LSI LSI vs. pH y = x 7.6073 1.20 1.00 0.80 0.60 0.40 0.20 0.00 0.20 0.40 0.60 0.80 68 pHLSI LSI vs.Total Alkalinity and Calcium Hardness as CaCO3y = 0.4343Ln(x) 2.7834 2.00 1.80 1.60 1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00 0100200300400500 Total Alkalinity and Calcium Hardness as CaCO3 (mg/L)LSI Total Alkalinity Calcium Hardness Figure 7: LSI Values vs. LSI Parameters

PAGE 46

35 3.8.2 Lead and Copper Rule In 1991, the U.S. EPA enacted the Lead and Copper Rule (LCR) for drinking water suppliers because of the adverse health effects of copper a nd lead corrosion (U.S. EPA, 1991). Neither metal is prevalent in drinking water, but copper pipes and lead solder can undergo an oxidation/reducti on reaction with water, dissolved oxygen, and other oxidants (Xiao et al., 2007). The reacti on can precipitate these metals in the water and will then come into contact with potable water consumers. To combat the aggressiveness of certain waters to leach the metals, the DWTP and other treatment centers use LSI or another such index to de termine the aggressiveness of the finished water. 3.8.3 Stage 2 Disinfectants and Di sinfection Byproducts Rule The Stage 2 Disinfectants and Disinfec tion Byproducts Rule (Stage 2 DBPR) enacted by the U.S. EPA in January 2006, focu ses on the reduction and elimination of DBPs in drinking water from both surface a nd groundwater sources. The main change from the Stage 1 Disinfectants and Disinf ection Byproducts Rule was the method of compliance in reporting contaminant concentr ations. Before the Stage 2 DBPR, many treatment plants averaged the DBP measuremen ts over their entire distribution system. This meant that many plants could actually exceed their maximum contaminant levels (MCLs) if the average repor ted value was lower (Richardson, 2003). According to the EPA, the Stage 2 DBPR covers the DBPs formed when water treatment plants use disinfectants like chlora mines and chlorine to reduce the pathogens in the finished water. The two most prevalent groups of DBPs are THMs and HAA5 which can form when chlorine interacts with NOM (Crittenden and Montgomer y Watson Harza, 2005). The

PAGE 47

36 DWTP uses chlorine as their disinfectant a nd therefore they have issues regarding DBP formation. According to the EPAs Stage 2 DBP Rule guidance manual, pH adjustment, filtration, NF and RO processes and chlorinati on adjustment are methods used to reduce DBP formation within water treatment plants. 3.8.4 Long Term 2 Enhanced Surface Water Treatment Rule In 2006 along with the Stage 2 DBP Ru le, the EPA created the Long Term 2 Enhanced Surface Water Treatment Rule (LT2 Rule) to minimize illnesses associated with certain pathogens in the drinking water that are resist ant to some disinfectants like chlorine. Among others, these pathogens include Giardia and Cryptosporidium Surface water and groundwater that is influenced by su rface water are subject to this rule. The rule classifies systems into one of four cat egories called bins (U.S EPA, 2006). The bin categories are determined by monitoring results for E. coli which is cheaper than monitoring tests for pathogens like Cryptosporidium The higher the bin the more removal the treatment plant must provide with the highest bins having to show a further (1.0 2.5 log) reduction in Cryptosporidium levels above the 3.0 log required by the LT2 for meeting turbidity requirements. A lthough the well water for the DWTP is not considered to be influenced by surface wa ter, the plant would like to explore the possibility of being placed under this rule as it might help in re ducing costs associated with well testing. Cu rrently, the plant mu st test wells for E. coli in their wells and so many tests can be cost prohibitive. Being cl assified under the LT2 Rule would allow the plant to show compliance by providing certain treatment processes approved by the EPA for reducing these pathogens.

PAGE 48

37 4. METHODS AND MATERIALS 4.1 Overview Historical data collected from DWTP wh ich consisted of reports from Southern Analytical Laboratories (SAL) and data gath ered by plant operators during the routine operation of the DWTP were used to chart th e historical changes in water quality and operational variables. This data can be f ound in graphical format in the Appendix 10. Data gathered from these sources and used for this thesis project include parameters such as cross-flow velocity, feed flow, and opera tional pressure ranges. Table 5 lists the parameters used in this study. Table 5: Operational Values at the DWTP Parameter Ranged ValueUnits Cross-Flow Velocity 13.7-12.8 m/min Feed Flow 1371-1280 GPM Feed Pressure 112-120 psi Feed pH 6.80 (+/0.1) The study comprised three phases. In pha se I, plant operational parameters and historical data were gathered. Samples for TDS analysis of the raw water, feed water, bypass water, permeate, and concentrate streams were also collected. In phase II, a flow

PAGE 49

38 cell system was designed and built to test four flat sheet membranes with distilled water and solutions of MgSO4, NaCl, and CaCl2. The test resulted in the identification of the flux in distilled water and the flux in thr ee different salt solutions using the four membranes. Flux is the flow of water through the membrane expressed as flow per area. In phase III, membrane performance was test ed using actual feed water from the DWTP. The permeate of the feed water from the four different membranes used in this study was blended at different volumetric proportions with the bypass water (post-cartridge filter). Total alkalinity, calcium hardness, pH, conduc tivity, and temperature were measured for each of the blended waters. Using this info rmation, calculations were made of the LSI for each blend. 4.2 Membrane Materials The project tested four membranes in cluding the KOCH membrane currently used at the DWTP; the project also tested three other membranes. The choice of membranes and a short list of their publis hed rejections and other specifi cations are listed in Table 6. These are based on the data sheets provide d by the manufacturers, which have been summarized in Appendices 6-9.

PAGE 50

39 Table 6: List of Membranes a nd Their Published Characteristics Manufacturer Filmtec Hydranautics Koch Koch Model Number NF90 ESNA1-LF TFC-S* TFC-SR2 Membrane Type Polyamide TFC Polyamide TFC Polyamide TFC Polyamide TFC Nominal Surface Area (m2) 37 37 38 35.8 NaCl Rejection % 85-95 ---MgSO4 Rejection % >97 -99.25 95 CaCl2 Rejection % -84-96 --Max Operating Press. (psi) 595 603 350 500 Typical Operating Press. (psi) --75-125 50-100 pH Range Continuous Operation 3-10 3-10 4-11 4-9 Free Chlorine Tolerance (ppm) <0.1 <0.1 <0.1 <0.1 Diameter (203 mm) 7.9 in 7.89 in 8 in 8 in *Current membrane used at the DWTP. 4.3 Flat Sheet Membrane System The flow cell system used for this st udy was designed and fabricated by Mr. Bob Riley of Separation Systems (San Diego, CA). It was constructed from 316 stainless steel and can sustain a pressure up to 800 psi. Stainless steel Swagelok fittings were used to connect the flow cell, meters, and valves The cell has two rubber O-ring seals as shown in Figure 8. The first seal surrounds the feed channel and the other wraps around the membrane area both of which help mainta in the integrity of the pressurized process during operation. Above the flow channel lies a sintered steel section slightly larger than the feed channel which allows the permeate to move outside of the flow cell. The sintered steel helps maintain the integrity of the membrane as the feed pressure is distributed evenly over the entire membrane. Inside the feed channel are an entry for

PAGE 51

40 feed flow and an exit for concentrate flow depending on how the flat sheet module is connected to the system. A flexible tube is attached to the permeate exit at the top of the flow cell to capture the permeate for collec tion and testing. The top and bottom portions of the flow cell are atta ched by six steel bolts. Permeate exit Flow ports inside feed channel Sintered Steel O-rings Figure 8: Separation Systems Flow Cell Front and Back Along with the fittings, all stainless steel tubing and the digital pressure transducer (S Model with digital readout ) came from Swagelok. The Swagelok pressure transducer and a Swagelok analog pressure gage was installed before the back pressure needle valve. The back pressure valve ma intains the pressure within the flat sheet module by reducing the aperture the water can flow through thus building up pressure. A schematic for the flat sheet system can be found in Figure 9. The system also uses Swagelok needle valves to control and adjust the water flow throughout the system. The system used a McMillan S-111 flow meter w ith a metering range between 0.5-5 L/min. For those sections of the system that di d not use Swagelok tubing, standard flexible

PAGE 52

41 tubing was used. A Hydra-Cell M-03 positive displacement pump with 3 gal/min flow capacity along with an Emerson 2-hp motor were used to pump the water through the system. The system uses a Polyscience P-se ries refrigerated recirculating chiller to control the temperature of the water duri ng system operation. The hp chiller can maintain refrigerated temperatures between -10 C to 40 C. The chiller used copper coils connected to the chiller reservoi r to transfer heat out of the system reservoir in a closed loop system. Figure 10 shows pictures of the system setup.

PAGE 53

42 Figure 9: Overview Schematic of Flat Sheet System

PAGE 54

43 Flat sheet module, Flow Meter,LabPro Datalogger Reservoir, copper coil and HydraCell pump Labtop computer with LoggerPro 3.1 Polyscience Chiller and tubing Figure 10: Overview of Flat Sheet Membrane System

PAGE 55

44 The system rerouted the water from the concentrate back into the reservoir to conserve ions. The data collection system wa s operated from a laptop and utilized Logger Pro 3.1 and LabPro software packages from Ve rnier. The data collection system gathered the feed/concentrate flow, reser voir temperature, pH of rese rvoir, and conductivity of the feed and permeates. Conductivity, temperature, and pH were collect ed using Verniers ph-bta pH probe, con-bta conductivity probe, and tmp-bta stainless steel temperature probe. The manual data collection consisted of permeate flow and pressure readings from the pressure transducer. A 10 mL gr aduated cylinder and a stop watch calculated the permeate flow by measur ing the time it took the perm eate to reach 3 mL. Two permeate flow measurements were made and then averaged. 4.4 Phase I Several goals were established for th is phase. The first goal consisted of collecting plant data from the historical databa ses used by the plant operators. From this data experimental parameters such as feed flow were calculated for the flat sheet membrane system. This allowed the flat shee t system to mimic as closely as possible the current operating pressures, feed flow, feed pH and cross-flow velocity as displayed in Table 5. To calculate the cros s-flow velocity of the flat sheet system the depth of the feed channel as well as the active area membra ne width had to be measured. A Cen-Tech digital caliper measured both parameters. Th e depth of the feed channel was taken from the bottom of the channel to the top of the feed channel rubber seal (O-ring). The O-ring had a high density and it was assumed to compress very little if any during the pressurization process. The active membrane area was measured from midpoint to

PAGE 56

45 midpoint of the feed channel o-ring. The f eed channel depth measured 2.72 mm and the active membrane width was 31.98 mm. Using the following equation, the feed flow (Qf) was calculated using the plants current cross flow velocity (VCF). The feed flow was calculated to be 1.2 L/min based on the averag ed cross flow velocity from DWTPs 2008 historical data. Both feed channel depth (FCD) and active membrane area width (Wcell) as stated previously were measured. cell CD f CFW F Q V (8) The second goal of Phase I was to quan tify the TOC from the different water flows as well as compare the measured conduc tivity with the TDS va lues to estimate a conversion factor. The samples taken from DWTP were stored in Boston Amber Round bottles from Fisher Scientific. Each water bottle was used multiple times, and each bottle always stored the same water source sample The sampling protocol consisted of running the water sample lines at the plant for five minutes. Then each bottle was rinsed with water from the sample port at least three times before the sample was taken. Once the samples made were transported to the Univer sity of South Florida campus, they were stored in a refrigerator until the proper experiments could be run. From the DWTP, samples of the raw water, feed water, post-cartridge filtrate (bypass), permeate, and concentrate were collected. These samples are plant-level samples and not taken from individual skids.

PAGE 57

46 TDS experiments were conducted based on the Standard Methods 2540C, and conductivity was measured us ing the Cole-Parmer conduc tivity probe model 1481-61. This is the same probe that plant operators us e at their on-site laboratory in the DWTP. 4.5 Phase II Phase II consisted of gathering flux data and rejection data us ing distilled water and three separate salt solutions. The four solutions were run through the flat sheet system. The three salt solutions of 500 mg/L were NaCl, MgSO4, and CaCl2. Each experiment maintained feed pH at 6.8 (+/0.3), feed flow at 1.2 L/min (+/0.04 L/min), and water temperature at 25 C (+/0.4 C). Before the membranes could be used each required a tempering preparation procedure. Each membrane used in the flat sheet module was soaked in distilled water for one hour prior to compaction. Once hydrated, the membranes were placed in the flat sh eet module. Membrane compaction entailed running distilled water through the membrane in a recirculation mode for a 24 hour period at 120 psi. A second flow cell was at tached to the first in series to double the quantity of membranes that could be compact ed at one time. After compaction each membrane was stored in a ZiplocTM bag with paper soaked in distilled water to keep it hydrated before and after each use. After compaction flux data were gathered using a solution of distilled water. After installa tion in the flow cell each membrane ran at pressures of 40, 60, 80, 100, 120 and 140 psi. The water flux (Jw) was calculated using the permeate flow (Qp) and active membrane area (Acell) in following equation: cell p WA Q J (9)

PAGE 58

47 Finally, the % rejection data were ga thered using three solutions of MgSO4, CaCl2, and NaCl. Each solution consisted of a 500 mg/L concentrati on of these salts. The solutions ran at the same pressures as the flux experiments (40, 60, 80, 100, 120, and 140 psi). Unlike the water flux experiment, conductivity probes were placed in the receptacles holding the feed and permeate solutions. The conductance of the solutions was the bulk parameter used in determining % Rejection from the following equation. 100 1 % f pC C R (10) Cp and Cf are the conductivity in S/cm. 4.6 Phase III The final phase of the pr oject entailed using feed a nd bypass water gathered from the DWTP to create different ratios of bl ended water. Feed water and bypass water samples were collected in 15 L buckets with the same wash and storage procedure used with the amber Boston rounds. Each bucket had a lid and was stored in a refrigerator in the USF lab. The feed water was introduced through each of the four membranes in the flat sheet module system at a constant pres sure of 120 psi. Once enough of the permeate was produced it was blended with the bypass wa ter at volumetric ratios of 0%, 10%, 15%, 20%, 30%, and 100%. The Vernier probes measured conductivity, pH, and temperature of the blended water. After th e blended water had been measured for these parameters, it was subjected to total alkali nity and calcium hardness tests using EPA approved Hach methods 8221 and 8222 respective ly. Using the tota l alkalinity, calcium

PAGE 59

48 hardness, temperature, pH, and conductivity m easurements of the bl ended water, the LSI was calculated for each blend ratio. Sin ce the TFC-SR membrane reported much different LSI values than any other membranes tested, an additional test blending test was performed. To simulate the tw o stage RO process, the membrane was used to treat 50% of the water feed water. 500 mL of th e permeate water from this portion of the experiment was stored in the refrigerator unti l needed. The 50% left in the container had been concentrated similar to the feed solution fed into DWTP s second stage RO process. Another 500 mL of the permeat e was collected from the concentrated solution. The first and second stage permeates were mixed at a 2/ 3 to 1/3 volumetric ra tio respectively. The amounts depended on what bypass blend was being created. Similar to the first blend experiment, the blend ratios we re 0%, 10%, 15%, 20%, 30%, and 100%.

PAGE 60

49 5. RESULTS AND DISCUSSION 5.1 Phase I The historical data show that, for the past several years, the feed pressure at the DWTP has steadily decreased (Figure 11). From 2001-2005 the feed pressure increased from about 100 psi to a maximum around 130 psi. However, the pressure fell and then leveled at around 120 psi where it has been for the last couple of years. The reason for the increase in pressure was probably due to fouling of the membranes. New membranes were installed during late 2001 to early 2002. It took a while for them to reach an optimum performance between DWTPs cl eaning protocols and daily fouling. Figure 11: DWTP Feed Pressure Over a Seven Year Period

PAGE 61

50 To calculate the LSI at the DWTP, the pl ant operators need to estimate the TDS. They do not measure TDS directly, but inst ead gather conductivity data from water samples using a conductivity meter at thei r onsite laboratory. Th e operators use the following equation to calculate the TDS value in mg/L based on their meter reading of conductivity. 61 0 cm S ty Conductivi L mg TDS (11) The value of 0.61 represents a standard that the DWTP has been using since the beginning of operations. This equation wa s provided by their local engineers and probably represents a standard based on liter ature review instead of the analytical relationship between TDS and act ual conductivity from the plan ts water. According to published resources, the slope of the TDS vs. conductivity plot can fall in the range of 0.5-0.9 (Crittenden and Montgomery Watson Ha rza, 2005). The TDS versus actual conductivity data from experiments run duri ng this project are placed in a composite graph seen in Figure 12. All conductivities were measured using the plants conductivity meter. For the daily graphs of TDS vs. conductivity see Appendix 11. With a composite slope of 0.71, the results showed that the current slope factor used by DWTP was underestimating the TDS. The calculated sl ope factor probably re presents the water better than the value currently in usage because it is determined from actual measurements.

PAGE 62

51 Error in the estimates of conductivity and TDS in Fi gure 12 could arise from improper calibration of the conductivity meter at the DWTP or improper drying of the sample during TDS measurements. However, the variability in well sources used to create the raw water and future chemical changes of the well water due to salt water intrusion or other reasons coul d have an effect on these results and periodic testing will have to be done to maintain accuracy. Both the higher slope (0.71) and the current slope value (0.61) were used in cal culating and reporting the results of the LSI in the next section. Figure 12: TDS vs. Conductivity Composite Graph

PAGE 63

52 5.2 Phase II In Phase II the object of the experiment was to determine the flux and % rejection of a 500 mg/L solution of NaCl, CaCl2 and MgSO4 for the various membranes. By determining the flux and rejections, the membranes bench mark performance was assessed. During the testing of the membrane s, experiments conducted at lower pressure values exhibited the most variability and error due to the constant fluctuation of pressure. The fluctuation was due to the amount of vibr ation in the system at those pressures. Between 40-60 psi, the system had a tendency to randomly increase or drop pressure and flow rate. Constant vigilance and adjustment s had to be maintained to ensure relative stability within the system. Both the back pressure and flow adjustment had to be constantly attuned using the appropriate needle valves. However, over time the ability to maintain pressure and stability was improved. To maintain the concentration of the salts over time, the permeate was recycled into the feed reservoir. Fi gures 13-16, show plots of flux versus transmembrane pressure for each membrane using distilled water. The slopes of each chart represents the permeate flux coefficient, which are listed in Table 7.

PAGE 64

53 Figure 13: TFC-S Intrinsic Water Flux Plot Figure 14: TFC-SR Intrinsic Water Flux Plot

PAGE 65

54 Figure 15: NF-90 Intrinsic Water Flux Plot Figure 16: ESNA1-LF Intrinsic Water Flux Plot

PAGE 66

55 The two KOCH membranes TFC-S and TFC-SR have similar permeability coefficients. The Film-Tec NF-90 had a slightly lower value while Dows ESNA1-LF had the lowest permeability coefficient. The permeability coefficient is important because it gives one of the first indications of membrane fouling. As the membranes become fouled the coefficient will fall. The value will also fall when the TDS increases in water since the water will have a higher osmotic pressure to overcome. Since the coefficient is dependent on pr essure, a higher coefficient re lates to more production of water per active membrane area. This means that membranes with lower coefficients will need higher pressures to achieve production rates similar to other membranes with a higher coefficient. In Table 7, membrane re sistance was also calculated. As expected, TFC-SR had the lowest resistance which indicat es that less pressure will have to be applied to raise the water recovery than any of the other membranes.

PAGE 67

56 Table 7: Membrane Permeability Coefficients and Resistance Membrane Water Permeability Coefficient (L/m2*hr*bar) Membrane Resistance (m-1) Permeability* Coefficient for a 500 mg/L NaCl Soln. (L/m2*hr*bar) Membrane Resistance (m-1) Permeability* Coefficient for a 500 mg/L MgSO4 Soln. (L/m2*hr*bar) Membrane Resistance (m-1) Permeability* Coefficient for a 500 mg/L CaCl2 Soln. (L/m2*hr*bar) Membrane Resistance (m-1) TFC S 6.95 1.62E-4 6.24 1.80E-4 6.21 1.81E-4 6.33 1.78E-4 TFC-SR 6.95 1.62E-4 9.98 1.13E-4 9.10 1.23E-4 8.94 1.26E-4 NF-90 6.82 1.65E-4 6.11 1.84E-4 6.27 1.79E-4 6.23 1.80E-4 ESNA1LF 5.56 2.02E-4 5.10 2.20E-4 5.06 2.22E-4 4.86 2.31E-4 *Data gathered from flux char ts in Figures 11-14 and 17.

PAGE 68

57 The % rejection for each membrane can be calculated using the following equation where is the density of water (g/cm3), A is the water permeability coefficient, B is the salt flux coefficient, and P and are the pressure and osmotic pressure across the membrane respectively (Baker, 2004). The equation is the result of the combination of the equations for water flux and the salt concentration of the permeate side of the membrane. % 100 * 1 P A B R (12) However, this study used a simplified ve rsion of Equation 12 w ith the use of the permeate (Cp) and feed (Cf) bulk parameter of conductance. The new equation follows below. % 100 1 F PC C R (13)

PAGE 69

58 Table 8: Percent Rejection Tables for TFC-S and TFC-SR Membrane Salts Operating Pressure (psi) %Rejection Membrane Salts Operating Pressure (psi) %Rejection TFC-S TFC-SR NaCl 40 85.7 NaCl 40 66.1 60 87.9 60 66.3 80 90.0 80 67.0 100 91.7 100 66.8 120 92.4 120 67.0 140 92.7 140 67.2 MgSO4 40 95.8 MgSO4 40 88.5 60 96.9 60 88.7 80 97.6 80 89.3 100 97.9 100 89.5 120 98.0 120 89.0 140 98.7 140 89.1 CaCl2 40 90.6 CaCl2 40 82.9 60 93.0 60 83.3 80 95.9 80 84.0 100 97.3 100 85.8 120 98.0 120 86.9 140 98.1 140 87.7

PAGE 70

59 Table 9: Percent Rejection Table for NF-90 and ESNA1-LF Membrane Salts Operating Pressure (psi) %Rejection Membrane Salts Operating Pressure (psi) %Rejection NF-90 ESNA1-LF NaCl 40 85.1 NaCl 40 83.5 60 86.8 60 89.1 80 89.9 80 91.4 100 91.9 100 92.6 120 92.4 120 92.8 140 93.0 140 93.1 MgSO4 40 95.5 MgSO4 40 94.6 60 95.6 60 96.7 80 96.4 80 97.1 100 96.8 100 97.4 120 97.1 120 97.7 140 97.3 140 97.5 CaCl2 40 35.7 CaCl2 40 95.1 60 56.7 60 96.4 80 73.6 80 97.4 100 81.0 100 98.5 120 85.4 120 98.9 140 88.5 140 98.9 In Tables 8 and 9 above the percent reje ction of the different salt solutions per membrane at different operating pressures are shown. See Figures 17 and 18 for graphical representations of the data in these tables. According to information in the tables and graphs, at the current plant operating pressure of 120 psi or 8.27 bar NF-90 and TFC-SR have lower rejections of magnesium and calci um ions. However NF-90 is closer to the other two membranes than to TFC-SR. TFCS and ESNA-LF have similar rejection of the same ions at the same operating pressure The rejection of monovalent sodium is

PAGE 71

60 much lower in TFC-SR membrane, but TF C-S, NF-90, and ESNA1-LF have similar rejections for the sodium ion. The data above suggest that both NF-90 and TFC-SR would give more mineralized water because of th eir lower rejections of divalent ions. At a lower operating pressure of 100 psi, the da ta suggests that NF90 and TFC-SR would continue producing lower rejection values th an either TFC-S or ESNA1-LF membranes. Overall rejection performa nce of each membrane can be seen in Figure 17. The data show us that both TFC-S and NF-90 reje ct the salts at about the same percentage over the last range of pressures. Lo oking at the data for NF-90, the CaCl2 numbers look different than would be expected based in comparison with the MgSO4 numbers for the same membrane. NF-90 rejection for calcium drops significantly in the lower pressure ranges and is probably due to the issues a ssociated with maintaining the proper water flow and pressure within the system. More tests will have to be conducted to see if the discrepancy is due to some error in the expe riment. The trends show that both NF-90 and TFC-S would perform at roughly similar rej ections even if operating pressures were lowered. Like the TFC-S, the NF-90 gives us slightly increasing reje ction over a range of pressures. Since rejection is dependent on c oncentration of solutes and not pressure, the aberration in the data probably has something to do with fluctuations in temperature or flow. The data may be adjusted by norma lizing it with the ap propriate variable. Singularly, the ESNA1-LF membrane increases its rejection of all the salts to one degree or another at increasing pressures but the re jections become more or less stable around 100 psi or 6.9 bar. The rejection for most of these salts follows the same increasing pattern as the rejection of calcium by NF-90 and the error probably follows a similar

PAGE 72

61 explanation. At higher pressure ranges, ESNA1 -LF gives slightly higher rejection values than the TFC-S and NF-90 membranes. Overall, the TFC-S, NF-90 and ESNA1-LF provided similar rejection of all salts between the 100-140 psi and 6.9-9.7 bar of operating pressure. This means that the only si gnificant change is s een through the use of TFC-SR. The performance of each membrane based on individual salt rejections is shown in Figure 18. For NaCl rejection, all me mbranes except for TFC-SR give similar rejections over the same operating pressure s. TFC-SR gives a significantly lower rejection at below 70% over the same range of pressures. This increased NaCl concentration in the finished water with the use of TFC-SR would not affect the hardness, but would increase the conductiv ity. However, of all the LSI parameters the TDS value has the least effect according to Figure 7. Increased levels of sodium may also have implications for finished water taste since no post treatment will remove excess salinity. For MgSO4, TFC-S, NF-90 and ESNA1-LF had comparable rejections. The NF90 and ESNA1-LF had slightly lower rejecti ons, but the differences are between 1-2%. Switching out the current membranes with either ESNA1-LF or NF-90 would not necessarily change the Mg2+ concentration and therefor e the total hardness to any significant degree. However, the usage of TFC-SR would increase the hardness due to Mg2+ because rejection would decrease by 7-9% across the relevant pressure ranges. At DWTP, according to Table 2 most of the har dness comes from calcium so the effect of less Mg2+ rejection might have a reduced amount of an impact than the rejection numbers would show. ESNA1-LF had th e highest rejection of CaCl2 than any other membrane so

PAGE 73

62 the hardness of the permeate would be even less than from TFC-S. Both the TFC-S and TFC-SR had similar rejections at lowe r pressures but between 100-140 psi and 6.9-9.7 bar, TFC-SR had around 4-6% lower rejections The differences in rejection by the TFCSR membrane would most likel y increase the total hardness of the permeate. Again the lower rejection values of CaCl2 for the NF-90 membrane are subject to scrutiny and further evaluations are needed to verify the anomalies. At the higher ranges, NF-90 falls between the performance of TFC-S and TFCSR. With the use of NF-90 and TFC-SR the rejection of CaCl2 is slightly lower than TFC-S at certain operating pressures. The flux versus pressure curves for e ach membrane per each salt in Figure 19 show that the highest flux for any salt soluti on is retrieved from the TFC-SR membrane. A composite graph in Figure 20 shows the fl ux versus pressure curves for all the membranes of Figure 19. The trend lines in th e graph have been approximated. Both the TFC-S and the NF-90 have similar fluxes across the different pressures, so no real benefit is seen in terms of permeate flux with a sw itch to NF-90. On the other hand, the ESNA1LF has the lowest flux across the same range which means that higher pressure would have to be used to achieve the same amount of flux as the TFC-S or NF-90 membranes. Trying to maintain the current level of permeate quality by using ESNA1-LF may have the effect of increasing costs for DWTP sin ce similar water recovery would mean higher feed pressures. Again TFC-SR shows th e most positive difference than any other membrane. The TFC-SR membrane gives the greatest amount of flux over the range of operating pressures.

PAGE 74

63 So at lower operating pressures, the TFCSR would provide great er flux but reduced percent rejection enabling it to mineralize th e water at a cheaper cost than any other alternative membrane.

PAGE 75

64 Figure 17: Percent Rejection vs. Transmem brane Pressure (TMP) (Per Membrane)

PAGE 76

65 Figure 18: Percent rejection vs. Transm embrane Pressure (TMP) (Per Salt)

PAGE 77

66 Figure 19: Flux vs. Pressure Gr aphs (Per Salt-Membrane)

PAGE 78

67 Figure 20: Composite Flux in Salt Solutions Per Membrane 5.3 Phase III The final phase of the project entailed computing the LSI values for various blend ratios for each membrane using DWTP fe ed water and bypass water. Analysis of the data would suggest which membrane would make the appropriate finished water with the smallest amount of blend. According to Figures 21 and 22, usi ng the corrected slope factor of 0.71 had negligible effect on the LS I numbers. In line with the performance of each membrane, the NF-90, TFC-S and ESNA1-LF had slightly different LSIs at the various blend ratios, while TFC-SR was the mo st divergent. Individual LSI parameters can be seen in Table 10.

PAGE 79

68 Table 10: Water Quality Data from LSI Versus Blend Ratio Experiments (*Calculated with adjusted slope factor) pH Temp ( C) Cond ( S/cm) TDS (mg/L) TDS* (mg/L) Total Alk (mg/L as CaCO3) Cal Hard (mg/L as CaCO3) LSI LSI* NF90 10% 6.92 25.1 127.6 77.8 90.6 28 32 -2.00 -2.01 15% 6.94 25.5 181.2 110.5 128.7 30 46 -1.80 -1.81 20% 6.99 25.4 221.3 135.0 157.1 48 56 -1.47 -1.48 30% 7.19 25.3 243.8 148.7 173.1 62 88 -0.97 -0.98 TFC-S 10% 6.70 24.7 192.4 117.4 136.6 28 32 -2.25 -2.25 15% 6.90 24.9 222.7 135.8 158.1 40 50 -1.70 -1.71 20% 7.07 25.1 235.4 143.6 167.1 50 64 -1.32 -1.33 30% 7.18 25.1 247.8 151.2 175.9 66 94 -0.93 -0.94 TFC-SR2 10% 7.32 25 261.1 159.3 185.4 90 112 -0.58 -0.59 15% 7.41 25.6 257.7 157.2 183.0 80 124 -0.49 -0.49 20% 7.41 25.2 260.4 158.8 184.9 88 132 -0.43 -0.43 30% 7.51 25.2 262.6 160.2 186.4 100 148 -0.22 -0.23 ESNA1LF 10% 6.67 25 170.4 103.9 121.0 36 36 -2.10 -2.11 15% 6.80 24.8 210.8 128.6 149.7 38 50 -1.82 -1.83 20% 6.92 24.8 230.7 140.7 163.8 46 74 -1.45 -1.46 30% 7.26 25.1 245.2 149.6 174.1 64 96 -0.85 -0.86

PAGE 80

69 Figure 21: Percent Blend Ratio vs. LSI Figure 22: Blend Ratio vs. LSI (@ 0.72)

PAGE 81

70 The LSI for TFC-SR is noticeably lower at all blends than any other membrane. Figures 21 and 22 follow the percent rejecti on values in that the most noticeable difference between the membranes was the TFC-SR element. Since the finished water at DWTP had been pH adjusted, degasified and chlorinated, the LSI values based on permeate and bypass water without benefit of any post treatment shown above will be more negative. After the additional post memb rane treatment changes of pH adjustment, CO2 removal, and chlorination the LSIs from the laboratory blends should increase in proportion to the finished water tested at DWTP if all chemical additions remain constant. The ability of the finished waters to become corrosive will remain or degrade for most of the membranes since their LSI valu es are not affected to the same degree as the TFC-SR. In all the blend ratios, the TFC-SR has th e strongest potential to reach the desired objective with the minimal effort. TFC-SR ra nges between -0.17 and -0.53 LSI. Of the three membranes TFC-S, NF-90 and ESNA1-LF, at the highest ratio of 30%, the NF-90 had the lowest LSI of -0.92 while ESNA1-LF was -0.81. However, these numbers indicate a higher blend ratio than is currently in use at DWTP. At blends of 20% to 15%, TFC-S gives the highest LSI valu es. Since this is the membrane currently in use, neither the NF-90 nor the ESNA1-LF would be a better alternative. However, they have higher LSI values at the 10% blend, but relative to the current TFC-S blending value at 20% their LSI both fall below 0.7-0.79 of the TFC-S LSI value. This means that the best candidate for membrane change at the DWTP is the TFC-SR.

PAGE 82

71 Not only does its LSI value increase over al l blend ratios, but it provides more mineralized finished water. The TFC-SR w ill be the main focus for the rest of the discussion due to the operational sim ilarities in the other three membranes.

PAGE 83

72 Figure 23: 1st and 2nd Stage Blend Ratios vs. LSI

PAGE 84

73 Figure 23 shows the 0% and 100% blend ra tio versus LSI for TFC-SR membrane. Also, the graph shows the combined first and second stage permeates blended at the same ratios. Since the objective of the DWTP is to produce a finished water with an LSI in the slightly positive range, the ble nds associated with the highe st LSI values should be the most likely to achieve these results w ith the minimal amount of post membrane treatment. All three experiments with si ngle stage permeate blends for the TFCSR maintained roughly the same trend. There was some fluctuation in LSI but much of the difference could be attributed to the fact that different feed waters were used in all three trials. Overall, the membrane showed that it can produce a more positive LSI even with different feed water characteris tics. Figure 23 shows that the 1st and 2nd stage blends tend to shift the trend in a more positive dir ection. However, the discrepancy between the two experiments could probably be attribut ed to the time it took to concentrate the feed water solution by 50%. During that time the feed and bypass water characteristics can change. Human error or equi pment issues cannot be ruled out. 5.4 Implementation 5.4.1 Membrane Properties The blend ratio impacts many different parts of the DWTP. Below in Figure 23, some of the more important relationships c oncerning the blend rati o are shown. Based on the experimental results, the most remarkable change in plant perfor mance would be with the TFC-SR membrane. The NF-90 membrane exhibited lower LSI values than the current DWTP membrane at blend ratios le ss than 20%. Even though NF-90 had a more positive LSI value than the current DWTP membrane at the 10% blend, the NF-90 LSI

PAGE 85

74 was almost 50% more negative than the TFC-S LSI value at the current 20% blend ratio. Although NF-90 has a small improvement to flux and slightly lower rejection of divalent ions at the higher operating pr essure ranges than TFC-S, switching to the NF-90 would not make any significant improvement to th e current setup. Like the NF-90, ESNA1-LF also has a more negative LSI than TFC-S at the 10% blend ratio, but the flux is lower and the MgSO4 rejection would not significantly ch ange. Again the ESNA1-LF or NF-90 membranes would have a relatively insignifi cant impact on rejecti on, recovery, or in minimizing the allowable blend ratio. Overall, the TFC-SR membrane would be the best candidate for change because the rejection of divalent ions would decrease along with NaCl. The change in rejection and recovery would affect the plants finished water by decreasing the aggressiveness of the water wh ile improving the rate of recovery at a lower pressure due to a higher flux in ionic solution. If the blend ratio is lowered or removed then the membranes will have to in crease recovery to maintain the same amount of finished water. With a potentially hi gher flux, the TFC-SR could probably accomplish it at a lower operating pressure. At higher recoveries, the potential for f ouling increases according to a review of the literature. If blending were removed fr om the treatment proce ss, the recovery would have to increase by 20%. This increase coul d have a negative effect on the system by requiring the plant operators to increase their membrane fouling treatments. Also, higher rates of recovery might carry over into a reduction in th e lifetime of the membrane because of increased usage.

PAGE 86

75 Testing will have to be carried out to dete rmine if increased fouling and a lowered life cycle of the membrane would happen if bl ending were reduced or eliminated.

PAGE 87

76 Figure 24: Blend Ratio Relationship Overview

PAGE 88

77 5.4.2 Concentrate Disposal The impact on concentrate disposal w ith the TFC-SR membrane would be a reduction in the TDS of the concentrate and th us a lowering of the ionic strength of the solution. This change would slightly lowe r hardness, alkalinity, and salinity as the concentrate would have reduced levels of Mg2+, Na, and Ca2+. Any scaling issues would be improved. The concentration of the ions would be reduced making it more difficult for the scale causing solutes to exceed their so lubility constant. Th is would also affect silica scaling as the lack of Mg2+ and Ca2+ would reduce the ability for silica to cause scaling. 5.4.3 Operation and Maintenance The most important aspect of any change in membrane or plant operation is the costs associated with it. This project did not look at actual costs in dolla rs because too many unknown factors would have made any estimat e similar to a guess. Case in point is the actual replacement costs for the membranes. There are two choices available. DWTP can either replace the membranes with 8 or 8 modules. Replacing the 8 membranes would mean finding a manufacturer with the capabilities but similar price structure to make it affordable. Since the TFC-SR is made by their current membrane manufacturer it would not be that difficu lt to repackage the replacements, but other membrane manufacturers may include consid erable cost increases. To replace the membranes with an 8 element would mean the use of brine seals or some other technology to make them fit in the current pres sure vessels. This fi x will have certain effects on plant processes that may include significant loss of pressure and possibly a reduction in flow through the pressure ve ssels. Maintenance due to fouling or

PAGE 89

78 mechanical issues might increase. Any co mparison between the two membrane sizes will have to include more research into manuf acturers and brine seal type technology than was in the scope of this paper. However, some broad estimates towards higher or lower costs can be made by looking at the probable effects a change in membrane might entail. According to the experimental results, TFC-SR could operate at lower pressures while maintaining an increase in the LSI values. Operating at lower pressures would pr obably translate into increased energy savings, yet some of the savings might be nega ted through increased product recovery if the blend ratios were lowered or removed altogether. The reduced bypass water would include cost savings in the reduction of chlorine disinfection as more if not all of the water woul d pass through the membranes at lower blend ratios. The blended water would have reduced amounts of pathogens and thus reduced need for disinfection, although residual di sinfection throughout the dist ribution system would have to be maintained. Since the LSI value for the TFC-SR treated water is more positive, reductions in pH control would probably come into effect. Other costs associated with the possible need to increase th e amount of anti-scalan t injected into the water might rise. Increased amount of feed water going into the system coupled w ith higher recoveries might increase the concentrations of the contaminants making it easier for them to overcome their solubility concentration.

PAGE 90

79 5.4.4 Plant Reclassification Any change to the membranes and blend ratio would also have an impact on the Stage 2 DBP Rule and the LT2 Rule. Changi ng to the TFC-SR membrane would affect the Stage 2 DBP Rule, since the membrane ha s the potential to decrease the amount of bypass water blend. A reduction in the amount of bypass water blended with permeate includes a reduction in NOM. Most NOM that reaches the RO/NF filters would be taken out at similar efficiencies due to the rem oval effects of these membranes. Less NOM means less DBP precursors making it easier for the plant to stay within the DBP MCLs. A change to TFC-SR woul d only have a positive effect on the LT2 Rule if the no bypass water was blended. The LT2 gives trea tment credits in log removal for various processes in the water treatment plant that affect the removal or inactivation of Cryptosporidium Any blend with minimally treat ed raw water negates the possible credit received by the membrane filtration process. According to the LT2 Rule, the DWTP would most likely be placed in bin one category which is the least restrictive in terms of showing log removal. Without mandated monthly tests for Cryptosporidium over a 24 month period, this assumption cannot be verified. However, the plants source water suggests that contamination by either Cryptosporidium or Giardia would be unlikely. If the DWTP was listed as a filtration treatment plant with the FDEP and categorized within bin one, then it would onl y have to provide the standard 3 log removal required for turbidity. The plant would most likely be considered a Filtered Treatment Center because filtration by Greensand and cartridge were used on both the feed and bypass streams.

PAGE 91

80 If the plant was able to be reclassifi ed and placed under the LT2 regulations Cryptosporidium testing would have to be implemented along with integrity tests on certain processes to verify the assessment of this paper.

PAGE 92

81 6. CONCLUSION Membranes ESNA1-LF, NF-90 and TFC-SR were compared to the existing membrane TFC-S as an evaluation of possi ble changes in the blend ratio and its subsequent effect on the DWTPs processes. In conclusion, the greate st change from any membrane replacement would be in the switch to the TFC-SR modules At all pressure ranges regardless of the solution TFC-SR maintained higher flux than any other membrane. At operating pressures of 100 psi and 120 psi the TFC-SR flux for the various salt solutions increas ed by approximately 45% above the current membrane. The TFC-SR showed lower rejection for the three salts tested than the TFC-S. Using TFC-SR at the 120 psi and the 100 psi operating pr essures the NaCl rejection fell around 27%, MgSO4 fell around 8.5%, and CaCl2 fell between 7.5% 5.2%. No other membrane tested had more positive change in membrane flux and rejection valu es than the TFC-SR. The decreased rejections values equate into a less aggressive permeate than is currently produced at the DWTP. The aggressiveness of the finished water was measured by the LSI. According to the experimental values, TFC-SR produced a more positive LSI over the 0% to 30% blend range in relation to TF C-S. At 0% blend, TFC-SR produced a slightly more positive and slightly more negative LSI than any other membrane at a 30% blend. The TFC-SR has the potential to e liminate blending at DWTP.

PAGE 93

82 LITERATURE CITED Aggazzotti, G., Fantuzzi, G., Righi, E., Predie ri, G. (1998). "Blood a nd breath analyses as biological indicators of expos ure to trihalomethanes in indoor swimming pools." The Science of the Total Environment 217: 155-163. Baker, R. W. (2004). Membrane Technology and Applications. Menlo Park, CA, Wiley. Bandini, S., Vezzani, D. (2003). "Nanofiltration modeli ng: the role of dielectric exclusion in membrane characterisation." Chemi cal Engineering Science 58(15): 3303-3326. Bartels, C., Wilf, M., Casey, W., Campbell, J. (2008). "New generation of low fouling Nanofiltration membranes." Desalination 221: 158-167. Bert, J. L. (1969). "Membrane compaction: a theoretical and experi mental explanation." Polymer Letters 7: 685-691. Bonne, P. A. C., Hofman, J.A.M.H., van der Hoek, J.P. (2000). "S caling control of RO membranes and direct treatment of surface water." Desalination 132: 109-119. Bowen, W. R., Mukhtar, H. (1995). "Characterisation a nd prediction of separation performance of Nanofiltration membranes." Bowen, W. R., Mukhtar, H. (1996). "Characterisation a nd prediction of separation performance of nanofiltration membranes." Journal of Membrane Science 112: 263-274. Bowen, W. R., Mohammad, A.W ., Hilal, N. (1997). "Characterisation of Nanofiltration membranes for predictive purposes -use of salts, uncharged solutes and atomic force microscopy." Journal of Membra ne Science 126(1): 91-105. Boyle, E. C. (2005). City of Dunedin Me mbrane Softening Plant WTP Alternative Pretreatment Evaluation. Broska, J. C., Barnette, H.L. (1999). Hydrogeology and Analysis of Aquifer Characteristics in West-Central Pinellas Count y, Florida. Tallahassee, FL, United States. Geological Survey: 99-185. Carnahan, R. P., Bolin, L., Suratt, W. ( 1995). "Biofouling of PVD-1 reverse osmosis elements in the water treatment plant of the City of Dunedin, Florid a." Desalination 102: 235-244.

PAGE 94

83 Chadik, P. A., Amy, G.L. (1983). "Removing trihalomethane precursors from various natural waters by metal coagul ants." JAWWA 75(10): 532-536. Chellam, S., Taylor, J. (2001). "Simplified analysis of contaminant rejection during ground and surface water nanofilt ration under the information collection rule." Water Research 35(10): 2460-2474. Crittenden, J. and Montgomery Watson Harza (F irm) (2005). Water treatment : principles and design. Hoboken, N.J., John Wiley. Dalvi, A. G. I., Al-Rasheed, R., Jave ed, M.A. (2000). "Hal oacetic acids (HAAs) formation in desalination processes fr om disinfectants." Desalination 129: 261-271. Dunedin, C. o. (1992). City of Dunedin Revers e Osmosis Water Treatment Facility. C. o. Dunedin. Dunedin, Fl. EPA, U. S. (1998). Stage 1 Disinfectants and Disinfection Byproducts Rule. U. S. E. P. Agency. EPA, U. S. (1999). Wastewater Technology Fact Sheet: Chlorine Disinfection. U. S. E. P. Agency. Washington, D.C. EPA, U. S. (2006). Long Term 2 Enhanced Surface Water Treatment Rule. U. S. E. P. Agency. EPA, U. S. (2006). Stage 2 Disinfectants and Disinfection Byproducts Rule. U. S. E. P. Agency. Fleischacker, S. J., Ramdtke, S.J. (1983). "F ormation of organic chlorine in public water supplies." JAWWA 75: 132. Flemming, H. (2002). "Biofouling in water systems -cases, causes and countermeasures." Applied Microbio logy and Biotechnology 59(6): 629-640. Gavrey, E. (2003). "Relationships between measures of NOM in Quabbin Watershed." JAWWA 95: 73-84. Gebbie, P. (2000). Water stability -what does it mean and how do you measure it? 63rd Annual Water Industry Engineers and Operators' Conference. Gwon, E., Yu, M., Oh, H., Ylee, Y. (2003). "Fouling characteristics of NF and RO operated for removal of dissolved matter from groundwater." Wate r Research 37: 29892997. Kuros, W. J., Ma, Y.H., Shimizu, T. (1996) "Terminology for membranes and membrane processes -IUPAC recommendations." J ournal of Membrane Science 120: 149-159.

PAGE 95

84 Labbez, C., Fievet, P., Szymczyk, A., Vi donne, A., Foissy, A., Pagetti, J. (2002). "Analysis of the salt retention of a titania membrane using the D SPM model: effect of pH, salt concentration and na ture Journal of Membra ne Science 208(1-2): 315-329. Labbez, C., Fievet, P., Szymczyk, A., Vi donne, A., Foissy, A., Pagetti, J. (2003). "Evaluation of the DSPM m odel on a titania membrane: m easurements of charged and uncharged solute retention, electrokinetic charge, pore size, and water permeability." Journal of Colloid and In terface Science 262(1): 200-211. Langelier, W. F. (1936). "The analytical control of anticorrosi on water treatment." JAWWA 28: 1500. Lvesque B., A. P., Tardif R., Charest-Ta rdif G., Dewailly ., Prud'Homme D., Gingras G., Allaire S. (2000). "Evaluation of the he alth risks associated with exposure to chloroform in indoor swimming pools." Journal of Toxicology and Environmental Health Part A 61: 225-243. Manttari, M., Jokinen, J.N., Nystrom, M. ( 1997). "Influence of filtr ation conditions on the performance of NF membranes in the filtra tion of paper mill total effluent." Journal of Membrane Science 137: 187-199. Melidis, P., Sanozidou, M., Mandusa, A., Ouz ounis, K. (2007). "Corrosion control by using indirect methods." Desalination 213: 152-158. Mohammad, A. W., Takriff, M.S. (2003) "Predicting flux and rejection of multicomponent salts mixture in Nanofiltra tion membranes." Desalination 157(1-3): 105111. Morris, R. D., Audet, A.M., Angelillo, I. F., Chalmers, T.C., Mosteller, F. (1992). "Chlorination, chlorination by-products and canc er: a meta analysis." American Journal of Public Health 82(7): 955-963. Nederlof, M. M., Kruithof, J.C., Taylor, J.S ., van der Kooij, D., Schippers, J.C. (2000). "Comparison of NF/RO membrane performa nce in integrated membrane system." Desalination 131: 257-269. Peeters, J. M. M., Boom, J.P., Mulder M.H.V., Strathmann, H. (1998). "Retention measurements of Nanofiltration membranes with electrolyte solutions." Journal of Membrane Science 145(2): 199-209. Plottu-Pecheux, A., Democrate, C., Houssa is, B., Gatel, D., Carvard, J. (2001). "Controlling the corrosiveness of bl ended waters." Desalination 139: 237-249. Reckhow, D. A., Singer, P.C. (1984). "The removal of organic halide precursors by preozonation and alum coagul ation." JAWWA 76(4): 151-157.

PAGE 96

85 Richardson, S. D. (1998). Dri nking water disinfection by-prod ucts. The Encyclopedia of Environmental Analysis and Remediation. R. A. Meyers. New York, Wiley. 3: 1398. Sagiv, A., Semiat, R. (2005). "Backwash of RO spiral wound membranes." Desalination 179: 1-9. Sawyer, C. N., McCarty, P.L., Parkin, G.F. (2003). Chemistry for environmental engineering and science. New Delh i, India, Tata McGraw Hill. Schaep, J., van der Bruggen, B., Vandecasteele, C., Wilms, D. (1998). "Influence of ion size and charge in Nanofiltration." Separa tion and Purification Technology 14: 155-162. Schaep, J. (1999). Nanofiltration for the re moval of ionic components from water. Heverlee, Belgium, Katholieke Universiteit. Ph.D. Schaep, J., Vandecasteele, C. Mohammad, A.W., Bowen, W.R. (2001). "Modelling the retention of ionic components for different Nanofiltration membranes." Separation and Purification Technology(22-23): 169-179. Schfer, A. I., A. G. Fane, et al. (2005). Nanofiltration: principles and applications. Oxford ; New York, Elsevier Advanced Technology. Schwinge, J., Neal, P.R., Wiley, D.E., Fletch er, D.F., Fane, A.G. (2004). "Spiral wound modules and spacers review and analysis." Journal of Membrane Science 242: 129-153. Sharma, R. R., Chellam, S. (2006). "Temperatu re and concentration effects on electrolyte transport across porous thin-film composite nanofiltration membranes: Pore transport mechanisms and energetics of permeation." Journal of Colloid and Interface Science 298(1): 327-340. Sheikholeslami, R., Bright, J. (2002). "Silica me tals removal by pretreatment to prevent fouling of reverse osmosis membranes." Desalination 143: 255-267. Smith, D., Falls, V., Levine, A., Macleod, B., Simpson, M. (2002). Nanofiltration to augment conventional treatment for removal of algal toxins, taste and ordor compounds, and natural organic matter. Water Qual ity Technology Conference. Seattle, WA. Taylor, J. S., Thompson, D.M., Carswell, J. K. (1987). "Applying membrane processes to groundwater sources for trihalomethane precursor control." JAWWA 79: 72-82. Thurman, E. M. (1985). Organic Geochemistry of Natural Wate rs, Martinus Nijhoff, Dr. W. Junk. Vezzani, D., Bandini, S. (2002). "Donnan e quilibrium and dielectric exclusion for characterization of Nanofiltration memb ranes." Desalination 149(1-3): 477-483.

PAGE 97

86 Viessman, W., Hammer, M.J., Perez, E.M ., Chadik, P.A. (2009). Water supply and pollution control. Upper Saddle River, NJ, Prentice Hall. Violleau, D., Essis-Tome, H., Habarou, H ., Croue, J.P., Pontie, M. (2003). "Fouling studies of a polyamide Nanofiltration membrane by selected natural organic matter." Desalination 173: 223-238. Vrouwenvelder, J. S., Kappelhof, J.W.N.M., Heijman, S.G.J., Schippers, J.C., van der Kooij, D. (2003). "Tools for fouling diagnosis of NF and RO membranes and assessment of the fouling potential of f eed water." Desalination 157: 361-365. Wang, X., Zhang, C., Ouyang, P. (2002). "The po ssibility of separating saccharides from a NaCl solution by using Nanofiltration in diafiltration mode." Journal of Membrane Science 204: 271-328. Withers, A. (2005). "Options for recarbona tion, remineralisation and disinfection for desalination plants." Desalination 179: 11-24. Xiao, W., Hong, S., Tang, Z., Seal, S., Tayl or, J.S. (2007). "Effects of blending on surface characteristics of copper corrosion products in drinking water distribution systems." Corrosion Science 49: 449-468. Xu, X., Mariano, T.M., Laskin, J.D., Weisel C.P. (2002). "Percutaneous Absorption of Trihalomethanes, Haloacetic Acids, a nd Haloketones." Toxicology and Applied Pharmacology 184: 19-26.

PAGE 98

87 APPENDICES

PAGE 99

88 Appendix 1: Overview of the Dune din Well Water Collection System Figure 25: Dunedin Well System Collection Map

PAGE 100

89 Appendix 2: Diagram of the Dunedin Water Treatment Plant Figure 26: DWTP Plant Schematic

PAGE 101

90 Appendix 3: Overview of All the Measu rements and Locations at the DWTP Table 11: Overview of the Measurements Performed at DWTP Measurement Plant Lab* Instrumentation** Wells Raw Effluent Clearwell Concentrate Feed 2 Cswy Blvd. Interstage Stage 1 Perm. Stage 2 Perm. Total Perm. Concentrate Green Sand Filters Flow x 1/m 1/m 1/m 1/m 1/m 6/d Pressure x 2/d, 1/m 1/m 1/m 1/m 1/m pH x x 1/6m 1/d 3/d 4/d 3/d 2/d, 1/m 1/d 1/m 1/m 1/m 1/m Temperature x 2/d Conductivity x 1/6m 1/d 3/d 4/d 1/d 2/d, 1/m 1d 1/d 1/d 1/d 1/d, 1/m Turbidity x 1/d 3/d 4/d 1/d 2/d Alkalinity x 1/6m 1/d 3/d 4/d 1/d Calcium Hardness x 1/d 3/d 4/ d 1/d 1/m 1/m 1/m 1/m 1/m Total Hardness x 1/d 3/d 4/d 1/d Free Cl x 6/d 12/d 1/d Total Cl x 12/d Chlorides x 1/m, 1/3m 3/d 4/d Flouride x 1/d 3/d 4/d Sulfates x 1/3 m Fe x 1/6m 1/d 3/d 4/d 6/d

PAGE 102

91 Appendix 3: (Continued) Table 11: (Continued) Measurement Plant Lab* Instrumentation** Wells Raw Effluent Clearwell Concentrate Feed 2 Cswy Blvd. Interstage Stage 1 Perm. Stage 2 Perm. Total Perm. Concentrate Green Sand Filters Mn x 1/6m 1/d 3/d 4/d 6/d Br x 1/6m NO3 x 1/6m Dis. Silica x 1/6m SO4 x 1/6m TDS x 1/3m, 1/6m As x 1/6m Ca x 1/6m Mg x 1/6m K x 1/6m Na x 1/6m HS x 1/6m TOC x 1/6m Pressure Differential x 1/d LSI x 3/d 4/d 1/d

PAGE 103

92 Appendix 4: List of Each Water Quality Lab Test at DWTP Site Time Location Type of Sample Testing Testing Location Wells: Monthly Chlorides, water levels Lab** Quarterly Sulfates, TDS, Chlorides Lab** Bi-Annual testing of production wells Conductivity (field), pH (field) Field Alkalinity: Total, Bicarb, Carb Lab** Br, Cl, NO3, Dis. Silica, SO4, TDS, AS, Ca, Fe, Mg, K, Na Lab** HS, TOC Lab** Skids 1-4 Daily Feed turbidity, pH, Conductivity, Temperature, Pressure In-house Interstage Conductivity In-house Stage1 Permeate Conductivity In-house Stage2 Permeate Conductivity In-house Total Permeate Conductivity In-house Concentrate Conductivity In-house Monthly Feed Pressure, Conductivity, Flow, Calcium Hardness, pH In-house 1st Stage Pressure, Flow, Calcium Hardness, pH In-house 2nd Stage Pressure, Flow, Calcium Hardness, pH In-house Permeate Pressure, Flow, Calcium Hardness, pH In-house Concentrate Pressure, Conductivity, Flow, Calcium Hardness, pH In-house Green Sand Filters 1-5 Daily Filters (1-5) Flow In-house Run Time In-house Pressure Differential In-house KMnO4 residual, levels In-house Daily Cartridge Filters 1-5 Fe, Mn In-house

PAGE 104

93 Appendix 4: (Continued) Misc. Raw Fe, Mn In-house Clearwell Fe, Mn In-house Effluent Fe, Mn In-house Raw Flouride In-house Clearwell Flouride In-house Effluent Flouride In-house Raw pH, Alkalinity, Calcium Hardness, Total Hardness, Cl, Turbidity, Conductivity In-house Clearwell pH, Alkalinity, Calcium Hardness, Total Hardness, Cl, Turbidity, Conductivity In-house Effluent pH, Alkalinity, Calcium Hardness, Total Hardness, Cl, Turbidity, Conductivity, LSI* In-house Concentrate pH (3x), Alkalinity, Calcium Hardness, Total Hardness, Cl, Turbidity, Conductivity, LSI* In-house 2 Causeway Blvd, (Farthest pt. in distr. Sys.) Cl, LSI*, pH Plant Rain levels In-house North Head Free & Total Cl South Head Free & Total Cl West Head*** N/A *LSI = Langlier Saturation Index (scale and corrosive test) **Testing done by Southern Analytical ***West headwork is not pre chlorinated

PAGE 105

94 Appendix 5: List of Drinking Water Monitoring Done at the DWTP Site Time Location Type of Sample Testing Distribution System 1/9 yrs N/A Asbestors Distr. Sys. 1/yr N/A Nitrate & Nitrite Distr. Sys. 1/yr N/A Inorganics Distr. Sys. 1/yr N/A THCs & HAA5 Distr. Sys. 1/3yrs N/A Secondary Contaminants Distr. Sys. 1/9yrs N/A Gro ss Alpha, Radium & Uranium Distr. Sys. 1/3yrs N/A Volatile Organinc Distr. Sys. 8/3yrs N/A Synthetic Organics Distr. Sys. # per mo./yr N/A Microbial Contaminants Distr. Sys. 1/3yrs N/A Beta particle & photon radioactivity **Testing done by Southern Analytical

PAGE 106

95 Appendix 6: Specification Sheet for KOCH TFC-S Membrane Name: KOCH TFC-S Type: Reverse Osmosis Product Specifications: Product Nominal Active Surface Area (m2) Product Water Flow Rate (m3/d) Stabilized Salt Rejection (%) TFC-S 38 MgSO4 35.2 99.25 Comments: 1000 mg/l MgSO4, 80 psi, 77F (25C) and 15% recovery. Dimensions Membrane Element Diameter Permeate Tube Diameter Membrane Element Length 8 (in) 1.5 (in) 40 (in) Operating Limits Membrane Type Polyamide TFC Max. Operating Temperature 45 C Max. Operating Pressure 350 psig Maximum Pressure Drop 10 psi pH Range, Continuous Operation 4-11 pH Allowoble Short Term Cleaning 2.5-11 Maximum Feed Flow 75 gpm Maximum Feed Silt Density Index SDI 5 Free Chlorine Tolerance <0.1 ppm (Based on the KOCH TFC-S specification sheet)

PAGE 107

96 Appendix 7: Specification Sheet for KOCH TFC-SR Membrane Name: KOCH TFC-SR Type: Nanofiltration Product Specifications: Product Nominal Active Surface Area (m2) Product Water Flow Rate (m3/d) Stabilized Salt Rejection (%) TFC-SR 35.8 MgSO4 58.7 95 NaCl 58.7 10-30 Comments: 5000 mg/l MgSO4, 96 psi 77F (25C) and 15% recovery. 2000 mg/l NaCl, 95 psi 77F (25C) and 15% recovery. Dimensions Membrane Element Diameter Permeate Tube Diameter Membrane Element Length 8 (in) 1.5 (in) 40 (in) Operating Limits Membrane Type Polyamide TFC Max. Operating Temperature 113 C Max. Operating Pressure 500 psig Maximum Pressure Drop 10/15 psi pH Range, Continuous Operation 4-9 pH Allowoble Short Term Cleaning 2-11 Maximum Feed Turbidity 1 NTU Maximum Feed Silt Density Index SDI 5 Free Chlorine Tolerance <0.1 ppm (Based on the KOCH TFC-S specification sheet)

PAGE 108

97 Appendix 8: Specification Sheet for FILMTEC NF-90 Membrane Name: FILMTEC NF-90 Type: Nanofiltration Product Specifications: Product Nominal Active Surface Area (m2) Product Water Flow Rate (m3/d) Stabilized Salt Rejection (%) NF90 37 NaCl 28.4 85-95 MgSO4 36 >97 Comments: 2,000 mg/l NaCl, 70 psi 77F (25C) and 15% recovery. 2,000 mg/l MgSO4, 70 psi, 77F (25C) and 15% recovery. Dimensions Membrane Element Diameter Permeate Tube Diameter Membrane Element Length 40 (in) 1.5 (in) 40 (in) Operating Limits Membrane Type Polyamide TFC Max. Operating Temperature 45 C Max. Operating Pressure 600 psig Maximum Pressure Drop 15 psig pH Range, Continuous Operation 3-10 pH Range, Short-Term Cleaning (30 min) 1-13 Maximum Feed Flow 70 gpm Maximum Feed Silt Density Index SDI 5 Free Chlorine Tolerance <0.1 ppm (Based on the FilmTec NF90 specification sheet)

PAGE 109

98 Appendix 9: Specification Sheet for HYDRANAUTICS ESNA1-LF Membrane Name: Hydranautics ESNA1-LF Type: Nanofiltration Product Specifications: Product Nominal Active Surface Area (m2) Product Water Flow Rate (m3/d) Stabilize d Salt Rejectio n (%) NF90 37 CaCl2 31 91 Comments: 500 mg/l CaCl2, 70 psi 77F (25C) and 15% recovery. Dimensions Membrane Element Diameter Permeate Tube Diameter Membra ne Element Length 7.99 (in) 1.125 (in) 36 (in) Operating Limits Membrane Type Polyamide TFC Max. Operating Temperature 45 C Max. Operating Pressure 600 psig Maximum Pressure Drop 10 psi pH Range, Continuous Operation 3-10 Minimum Ratio of Concentrate to Permeate Flow for any Element 5:1 Maximum Feed Flow 75 gpm Maximum Feed Silt Density Index SDI 5 Free Chlorine Tolerance <0.1 ppm (Based on the Hydranautics ESNA1-LF specification sheet)

PAGE 110

99 Appendix 10: Historical Data of the Dunedin Water Treatment Plant Figure 27: Historical %Salt Removal Over Time Figure 28: Historical Blend Flows

PAGE 111

100 Appendix 11: TDS vs. Conductivity Graphs Figure 29: TDS vs. Conductivity for 6/27/2008 Figure 30: TDS vs. Conductivity for 7/2/2008 /

PAGE 112

101 Appendix 11: (Continued) Figure 31: TDS vs. Conductivity for 7/9/2008 Figure 32: TDS vs. Conductivity for 7/12/2008

PAGE 113

102 Appendix 11: (Continued) Figure 33: TDS vs. Conductivity for 7/13/2008 Figure 34: TDS vs. Conductivity for 7/24/2008


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200385Ka 4500
controlfield tag 001 002029663
005 20101006175724.0
007 cr mnu|||uuuuu
008 090918s2009 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002961
035
(OCoLC)437033872
040
FHM
c FHM
049
FHMM
090
TA145 (Online)
1 100
Keen, Michael.
0 245
Evaluation of the impact of membrane change at a membrane softening water treatment plant
h [electronic resource] /
by Michael Keen.
260
[Tampa, Fla] :
b University of South Florida,
2009.
500
Title from PDF of title page.
Document formatted into pages; contains 102 pages.
502
Thesis (M.S.Env.E.)--University of South Florida, 2009.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
520
ABSTRACT: At the water treatment plant in Dunedin, Florida, reverse osmosis membranes remove the hardness from groundwater sources. Reverse osmosis membranes remove salts, pathogens, and organics from the feed water but can create an aggressive permeate. The membranes strip most ions in the process and the resulting permeate, if not subjected to blending on post treatment, has a tendency to leach metals from lead and copper pipes in the distribution networks. To prevent such problems, the permeate needs to be blended with partially treated raw water or to be chemically treated to re-mineralize and add alkalinity back into the water. In the last decade nanofiltration treatment has gained an increasing foothold in the water treatment industry especially as a water softener.Although nanofiltration membranes also have a high removal rate for organics and pathogens, the separation process is more selective towards multivalent ions (e.g., Ca, and Mg) than monovalent (e.g., Na) ions. Most membrane softening plants blend minimally treated raw water with the membrane permeate as a means to reduce the aggressiveness of the water. However, blending can cause issues with disinfection byproducts and pathogen re-introduction. With nanofiltration membranes, fewer mono-valent ions are rejected which creates a more stable permeate and can reduce the blended water ratio. Since it is unlikely that most plants that use membrane filtration for water softening will be able to stop blending entirely, any improvement or sustainability of water quality at a reduced blend ratio should be viewed favorably within the water treatment industry.The study evaluates three nanofiltration membranes: TFC-SR, NF-90, and ESNA1-LF in relation to the reverse osmosis TFC-S RO membrane currently in use at Dunedin. Water flux and salt rejection of the permeate water were compared using solutions of NaCl, MgSO and CaCl. Since the Langelier Saturation Index (LSI) is one of the main tests of the blended finished water and is used to judge water quality prior to its release into the distribution system, this study created a 0%, 10%, 15%, 20%, 30%, and 100% blend ratio for each membrane to compare and contrast the change in the LSI. The TFC-SR membrane showed the most promise in lowering the blend ratio while improving the aggressiveness of the finished water by showing a lower rejection for divalent ions. The TFC-SR membrane also showed an improvement in the LSI relative to the other membranes over the total range of blend ratios.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor: Daniel Yeh, Ph.D.
653
Reverse osmosis
Nanofiltration
Groundwater treatment
Blending
Process evaluation
690
Dissertations, Academic
z USF
x Environmental Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.2961