Denitrification in Membrane Bioreactors by

Anabela Duarte Fonseca

Thesis submitted to the Faculty of the Virginia Polytechnic Institute and State University in partial fulfillment of the requirements for the degree of Master of Science in Environmental Engineering

Dr. Clifford W. Randall, Chair Dr. John T. Novak Dr. Ann M. Stevens

September 14, 199 Blacksburg, Virginia Key words: denitrification, water treatment, membrane bioreactors, filtration, contactor, ion exchange

Copyright 1999, Anabela Duarte Fonseca

Denitrification in Membrane Bioreactors

Anabela Duarte Fonseca

(ABSTRACT) Three membrane bioreactors, a low flux filter (LFF), a diafilter (DF), and an ionexchange (IE) membrane bioreactor were used to treat water polluted with 50 ppm-N nitrate. The three systems were compared in terms of removal efficiency of nitrate, operational complexity, and overall quality of the treated water. In the low flux filter (LFF) membrane bioreactor an hemo-dialysis hollow fiber module was used and operated continuously for 29 days with a constant flux of permeate. The performance of the system was constant during the span of the experiment, which demonstrated that when the module was operated under constant low flux of permeate, the membrane filtration process was not affected by fouling. The removal rate of the LFF was 100% since the treated effluent did not contain nitrate or nitrite. The volumetric denitrification rate was 240 g-N day-1 m-3, which is within the range of denitrification rates obtained in tubular membrane modules. The treated effluent contained acetate, the carbon source of the biological process, and other inorganic nutrients, which showed that operating this ultrafiltration module at controlled flux did not improve the retention of these substances in the bioreactor. The same hemo-dialysis hollow fiber module employed in the LFF system was used in the diafilter (DF) membrane bioreactor. In the DF system, however, the membrane module was used as a contactor that separated the treated water and the bioreactor system, which allowed the transfer of solutes through the membrane porous structure and supported the growth of a biofilm on the membrane surface. The nitrate removal rate of the DF system increased from 76% to 91% during the 17 days assay. Unfortunately, this improvement could be attributed to microbial contamination of the water circuit because significant concentrations of the carbon source, acetate, nutrients, and nitrate were found in the treated effluent. The volumetric denitrification rate of the system was 200 g-N day1 m-3, and the surface denitrification rate was lower than values previously reported for contactor membrane bioreactors. The results hereby presented do not evidence any advantage of operating the Filtral 20 ® membrane module as a contactor instead of as a filter such as in the LFF system. On the other hand, the third system herein presented, the IE membrane bioreactor, demonstrated several advantages of a contactor configuration but with a non-porous ion exchange membrane module in place of the Filtral 20 ®. As in a contactor system, the anion membrane provided a surface for biofilm growth, facilitated the transport of nitrate, and prevented mixing of treated water and bioreactor medium. Compared to the two previous systems, the most remarkable result of the IE was the reduction of secondary pollution in the treated water. The concentrations of phosphate and ethanol were zero

and less than 1% of the concentration in the bioreactor, respectively. In addition, the IE system was less complex than the two other systems because the ion exchange membrane is non-porous. Therefore, unlike with porous contactors, it was not necessary to control the flux of treated water that could be lost through the bioreactor. The average surface denitrification rate of the IE system was 7.0 g-N day-1 m-2, which is higher than what had been reported for other contactor denitrification systems. However, because of the low surface to volume ratio of the membrane module that was used, the volumetric denitrification rate of the IE system was low, equivalent to 65 g-N day-1 m-3.

ii

Acknowledgments I thank my advisor, Dr. Clifford W. Randall, for his support, suggestions and guidance throughout these studies. I would also like to thank my committee members, Dr. Ann M. Stevens and Dr. John T. Novak for providing additional insights. My appreciation also is extended to many graduate students in the program for their help and advice. I would also like to thank Jodie Smiley, Marilyn Grender and Julie Petruska for their assistance with analytical methods and many other suggestions. Finally, I would like to thank my family for supporting my decision to join the ENE program at Virginia Tech and to Raja Mazumder for having first suggested it.

iii

Table of Contents ABSTRACT.................................................................................................................................................... I ACKNOWLEDGMENTS .................................................................................................................................. III TABLE OF CONTENTS ....................................................................................................................................IV LIST OF TABLES ............................................................................................................................................VI LIST OF FIGURES ......................................................................................................................................... VII CHAPTER 1- INTRODUCTION ......................................................................................................................... 1 PROBLEM STATEMENT ................................................................................................................................. 1 ALTERNATIVES ............................................................................................................................................ 1 APPROACH ................................................................................................................................................... 1 CHAPTER 2- LITERATURE REVIEW ............................................................................................................... 3 NITRATE POLLUTION ................................................................................................................................... 3 TREATMENT TECHNOLOGIES ....................................................................................................................... 3 ION EXCHANGE (IX) .................................................................................................................................. 3 BIOLOGICAL DENITRIFICATION .................................................................................................................. 4 COMBINED ION-EXCHANGE (IX) AND BIOLOGICAL DENITRIFICATION .......................................................... 5 REVERSE OSMOSIS AND ELECTRODIALYSIS ................................................................................................. 5 MEMBRANE BIOREACTORS .......................................................................................................................... 5 DENITRIFICATION IN FILTER MEMBRANE BIOREACTORS ............................................................................. 7 DENITRIFICATION IN CONTACTOR MEMBRANE BIOREACTORS ..................................................................... 8 CHAPTER 3 - MATERIALS AND METHODS................................................................................................... 10 POLLUTED GROUNDWATER ........................................................................................................................ 10 BIOREACTOR FEED ............................................................................................................................... ...... 10 LOW FLUX FILTER MEMBRANE BIOREACTOR (LLF) ................................................................................... 10 DIAFILTER MEMBRANE BIOREACTOR (DF)................................................................................................. 14 ION-EXCHANGE BIOREACTOR (IE) ............................................................................................................. 14 START-UP CULTURE .................................................................................................................................. 14 ANALYTICAL PROCEDURES ........................................................................................................................ 15 SAMPLE STORAGE ...................................................................................................................................... 16 CHAPTER 4 - RESULTS ................................................................................................................................. 17 LOW FLUX FILTER MEMBRANE BIOREACTOR (LFF) ................................................................................... 17 DIAFILTER MEMBRANE BIOREACTOR ......................................................................................................... 20 ION EXCHANGE MEMBRANE BIOREACTOR .................................................................................................. 24 CHAPTER 5 - DISCUSSION ............................................................................................................................ 28 LOW FLUX FILTER MBR ........................................................................................................................... 28 MEMBRANE FLUX .................................................................................................................................... 28 DENITRIFICATION RATE AND LOADING CAPACITY ...................................................................................... 28 SECONDARY POLLUTION .......................................................................................................................... 29 DIAFILTER-MBR ....................................................................................................................................... 29 DENITRIFICATION RATE AND SECONDARY POLLUTION ............................................................................... 29 POLLUTION WITH NITRITE ....................................................................................................................... 31 MICROBIAL CONTAMINATION OF THE TREATED WATER.............................................................................. 31 C/N REQUIREMENT ............................................................................................................................... .. 31 ION EXCHANGE-MBR ................................................................................................................................ 31

iv

DENITRIFICATION RATE ........................................................................................................................... 31 SECONDARY POLLUTION .......................................................................................................................... 32 WATER CORROSION ................................................................................................................................. 32 C/N REQUIREMENT ............................................................................................................................... .. 32 CHAPTER 6- CONCLUSIONS ......................................................................................................................... 33 CHAPTER 7- RECOMMENDATIONS FOR FUTURE RESEARCH...................................................................... 34 REFERENCES ................................................................................................................................................ 35 VITA .............................................................................................................................................................. 53

v

List of Tables Table 1- Head-space operational parameters for the analysis protocol of ethanol........................................ 15 Table 2- Operational settings of the HP GC for the analysis protocol of ethanol and carboxilic acids............................................................................................................................... 15 Table 3- Denitrification rate, removal and exclusion efficiency and in denitrification contactor membrane bioreactors..................................................................................................... 30 Table A 1- Technical characteristics of membrane module Filtral® 20 ....................................................... 38 Table A 2- Influent LFF-MBR...................................................................................................................... 39 Table A 3 - Biomedium LFF-MBR .............................................................................................................. 40 Table A 4- Treated water (effluent) of the LFF-MBR .................................................................................. 41 Table A 5 - Denitrification rate in the LFF-MBR......................................................................................... 42 Table A 6 - Influent (polluted water) of the DF-MBR.................................................................................. 43 Table A 7- Biofeed of the DF-MBR ............................................................................................................. 44 Table A 8 - Biomedium of the DF-MBR ...................................................................................................... 45 Table A 9- Effluent (treated water) of the DF-MBR .................................................................................... 46 Table A 10- Denitrification in the DF-MBR................................................................................................. 47 Table A 11- Influent (polluted water) of the EX-MBR ................................................................................ 48 Table A 12 - Biofeed of the EX-MBR .......................................................................................................... 49 Table A 13 - Biomedium of the EX-MBR.................................................................................................... 50 Table A 14 - Effluent (treated water) of the EX-MBR system ..................................................................... 51 Table A 15 - Denitrification in the EX-MBR ............................................................................................... 52

vi

List of Figures Figure 1- Schematic of a filter membrane bioreactor and detail of the filtration process through the membrane: the stripes in the membrane illustrate the porous structure; the white arrows through the membrane represent the bulk flux of permeate....................................................................................................................... ............... 6 Figure 2- Schematic of a contactor membrane bioreactor and detail solute (dark dots) transport through the membrane. ................................................................................................ 7 Figure 3- Schematic of the low flux filter membrane bioreactor. Legend: 1-membrane module; 2-bioreactor ; 3-permeate vessel; 4- nitrogen tank; P1-pump 1 (100 ml min-1); P2-pump 1 (100 ml min-1); P3-pump 1 (3 ml min-1); T- timer switch operated pump. .............................................................................................................. 11 Figure 4- Schematic of the diafiltration membrane bioreactor. Legend: 1-membrane module; 2-bioreactor ; 3-permeate vessel; 4- nitrogen tank; T-timer switch operated pump.................................................................................................................. ......... 12 Figure 5- Schematic of the ion exchange membrane bioreactor. Legend: 1-membrane module; 2-bioreactor ; 3- nitrogen tank; T-timer switch operated pump. ................................. 13 Figure 6- Concentration of nitrate and nitrite in the influent (polluted water) and in the effluent (treated water) of the LFF- MBR. Flow rate 3.3 ml min-1; hydraulic retention time (HRT) in bioreactor = 4 h .................................................................................. 18 Figure 7- Flow rate of influent (polluted water) and effluent (treated water) of the LFFMBR. ; hydraulic retention time (HRT) in bioreactor = 4 h ..................................................... 18 Figure 8- Concentration of acetate in the effluent (treated water) and in the biomedium of the LFF MBR. Flow rate 3.3 ml min-1; hydraulic retention time (HRT) in bioreactor = 4 h......................................................................................................................... 19 Figure 9 - Concentration of sulfate in the effluent (treated water) and in the biomedium of the LFF-MBR. Flow rate 3.3 ml min-1; hydraulic retention time (HRT in bioreactor = 4 h......................................................................................................................... 19 Figure 10- Denitrification rate per unit of reactor volume and filtration flux per unit of surface in the LFF-MBR. Flow rate 3.3 ml min-1; hydraulic retention time (HRT in bioreactor = 4 h........................................................................................................... 20 Figure 11- Concentration of nitrate and nitrite in the influent (polluted water) and effluent (treated water) in the DF-MBR. Hydraulic retention time (HRT) of bioreactor =1.62 day; HRT in the water circuit =1.7 h. ............................................................ 21 Figure 12- Concentration of acetate in the effluent (treated water) and in the biomedium of the DF- MBR. Hydraulic retention time (HRT) of bioreactor =1.62 day; HRT in the water circuit =1.7 h. ............................................................................................... 22 Figure 13- Concentration of sulfate in the influent (polluted water), the effluent (treated water) and biomedium of the DF- MBR. Hydraulic retention time (HRT) of bioreactor =1.62 day; HRT in the water circuit =1.7 h. ............................................................ 23 Figure 14- Surface and volumetric denitrification rate in DF-MBR. Hydraulic retention time (HRT) of bioreactor =1.62 day; HRT in the water circuit =1.7 h. .................................... 23 Figure 15-Concentration of nitrate and nitrite in the influent (polluted water) and in the effluent (treated water) of the IE-MBR. Hydraulic retention time (HRT) in the bioreactor =3.5 day; HRT in the water circuit =4.4 h. ........................................................ 25 Figure 16- Concentration of chloride, nitrate and bicarbonate in treated water effluent of the ion-exchange membrane bioreactor. Hydraulic retention time (HRT) in the bioreactor =3.5 day; HRT in the water circuit =4.4 h. ........................................................ 25 Figure 17-Concentration of phosphate in the effluent (treated water) and in the biomedium of the IE-MBR. Hydraulic retention time (HRT) in the bioreactor =3.5 day; HRT in the water circuit =4.4 h. ............................................................................... 26 Figure 18- Concentration of sulfate in the influent (untreated water),effluent (treated water) and biomedium of the ion-exchange MBR. Hydraulic retention time (HRT) in the bioreactor =3.5 day; HRT in the water circuit =4.4 h.......................................... 26

vii

Figure 19- Concentration of ethanol in the effluent (treated water) and in the biomedium of the IE-MBR Hydraulic retention time (HRT) in the bioreactor =3.5 day; HRT in the water circuit =4.4 h. ............................................................................................... 27 Figure 20- Surface denitrification rate in the IE-MBR. Hydraulic retention time (HRT) in the bioreactor =3.5 day; HRT in the water circuit =4.4 h. ........................................................ 27

viii

Chapter 1- Introduction Problem Statement The application of rich nitrogenous fertilizers to soils has led to contamination of many drinking water sources with nitrate. Because preventive policies and in-situ remediation of polluted watersheds have proven to be both difficult and expensive to implement [Clifford and Liu, 1993], pre-treatment of drinking water remains as the only effective alternative for this problem. Since nitrate is not removed during conventional drinking water treatment- i.e., coagulation, filtration and disinfection, other processes must be used in order to produce safe drinking water. Membrane bioreactors (MBRs) combine biological conversion and membrane selectivity and are among the most promising technologies to remove nitrate from drinking water. Membrane and energy costs, and the secondary pollution of the treated water with biological nutrients and products are very important issues in the development and optimization of denitrifying membrane bioreactors. Alternatives Denitrifying membrane bioreactors can be categorized as filter or contactor configurations. The main difference between the two configurations is the function of the membrane, which in the filter MBR is a filtration media and in the contactor MBR is a matrix through which the pollutant is transported from the water to the bioreactor, where is it made available for biological conversion. Both configurations are schematically represented in Figure 1 and Figure 2 in the Literature Review Section (pages 6 and 7). Biofilms tend to develop on the membrane surface of contactors and on that account some contactor MBRs are called membrane-biofilm systems. Approach The objectives of this research were to investigate the feasibility of using biological membrane systems for the removal of nitrate from water and to compare the performance of three types of membrane reactors. These studies were begun by investigating the application potential of a filtration module, Filtral 20®, that is commercially sold for hemo-dialysis. The membrane module is a hollow fiber unit with a very high ratio of surface to volume (6.8 m2/l), which is an advantage for both filtration and dialysis processes. Furthermore, the cost of each unit is substantially lower than that of other membrane modules because of the higher scale of production. However, the cost of a membrane unit is just a part of the total cost of membranes. The frequency of replacement of membranes because of their decreasing capacity due to fouling of the membrane is also a concern. Management of membrane fouling is particularly critical for the filtration capacity of the Filtral 20® because it was designed as a “single use” unit and the organic membranes in it are not resistant to chemical cleaning.

1

Filtral 20® was tested using two alternative configurations, i.e., as a filter MBR and as a contactor MBR. The low flux filter (LFF) membrane bioreactor was, as the name suggests, a filter membrane bioreactor that was operated with controlled low flux of permeate and, additionally, with recirculation of both permeate and biomedium. A schematic of the LFF system is shown in Figure 3, page 11 in the Materials and Methods section. The objective was to determine if maintaining a low and constant permeate flux could reduce the effect of fouling and hence allow for continuous use of the module. The diafilter (DF) membrane bioreactor had a contactor configuration analogous to the blood-dialysis equipment configuration for which Filtral 20® is commercially sold. A schematic of the DF system is shown in Figure 4, page 12 in the Materials and Methods section. In a contactor unit such as the Filtral 20® it is very important to have a homogeneous distribution of flow through the module because preferential paths can form. If this happens, then some of the available membrane surface will not be efficiently used. For this reason, the biomedium and the water were recirculated through the module. The objective was to determine if the transport of nitrate, which was the limiting factor in previously reported contactor MBRs, could be increased by using a module with a very high surface to volume ratio. Another MBR with a contactor configuration is hereby presented- the ion exchange (IE) MBR. The membrane module was a lab-scale unit with c.a. 0.0175 m2/l of surface to volume ratio. The IE-membrane bioreactor is based on the same principle as the “Combined ion-exchange/ biological denitrification” system [Van der Hoek et al, 1988]. The main difference being the use of an anion selective membrane instead of an ionexchange resin bed [Crespo and Reis, 1998]. A schematic of the system is shown in Figure 5, page 13 in the Materials and Methods section. One of the immediate advantages of the process is the possibility of operating the system continuously because it does not require regeneration off-line. A second advantage, also determined during this study, is that the membrane separated the treated water and the biomedium and therefore prevented microbial contamination, and reduced secondary pollution, of the treated water. The three systems, the LFF, the DF and, the IE membrane bioreactors, are discussed and compared in this report based on the removal efficiency of nitrate, the operational complexity, and the overall quality of the treated water.

2

Chapter 2- Literature Review Nitrate Pollution Nitrate pollution is an increasing problem worldwide that affects areas with intense agricultural activity and high population density. The intense application of rich nitrogenous fertilizers is the most frequent cause of nitrate groundwater pollution. Nevertheless, urban sewage effluents were found to be the source of pollution in 40% of surface waters polluted with nitrate [Wild, 1997]. Although the primary toxicity of nitrate is low, its presence in drinking water is a serious health hazard because at concentrations higher than 10 mg-N l-1 of nitrate, sufficient quantities of nitrite can be formed in the intestinal tract of infants and lead to acute asphyxiation, a syndrome known as methemoglobinemia. Other epidemiological studies have linked nitrate to congenital malformations and increased risk of cancer development [Bouchard et al, 1992]. Nitrate and nitrite are regulated by primary water standards. The U.S. EPA established maximum contaminant levels (MCLs) of 10 mg-N l-1 for nitrate and 1 mg-N l-1 for nitrite [EPA, 1999]. The European Union established an MCL of 50 mg-NO3-l-1 (11.3 mg-N l-1), a recommended level of 25 mg-N03- l-1 and an MCL for nitrite equal to 0.1 mg-N02 l-1 (0.03 mg-N l-1) [EC, 1980]. Treatment Technologies Nitrate is stable, very soluble in water, and not removed or transformed during conventional drinking water treatment processes, such as coagulation-flocculation, softening and filtration. More specific and sophisticated technologies, such as ion exchange (IX), biological denitrification, reverse osmosis, or electrodialysis must be used to remove nitrates from drinking water. Ion exchange and biological denitrification are the most frequently chosen technologies when nitrate removal is the only concern [Kapoor and Viraraghavan, 1997]. Combined reverse osmosis and electrodialysis are usually used to implement more complex treatment strategies, such as desalination, although there is an increased interest in electrodialysis using membranes that are nitrate selective [Indusekhar et al, 1991, Oldani et al, 1992]. Other more recent technologies include chemical, catalytic denitrification [Murphy, 1986], electrolytic denitrification and electrodeionization (EDI) [Salem et al, 1995] and are still in development. Ion exchange (IX) The ion exchange process involves flowing the polluted water through a strong base anion resin bed where nitrate is retained and replaced by another anion, i.e. a counter-ion. The counter-ion is usually chloride, and less frequently is bicarbonate. The resin bed is operated until saturation, after which it is regenerated off-line by converting it back to the counter-ion form. An IX process is usually operated with parallel resin beds that are sequentially saturated and regenerated. Nevertheless, IX can be operated continuously with loops [Kapoor and Viraraghavan, 1997] or with IX membrane modules [Korngold, 1991; Salem et al, 1995]. In either case, regeneration in IX results in production of a very

3

concentrated brine solution of nitrate and counter-ion. Regeneration and costs associated with the disposal of brines account for a substantial fraction of the operational costs. A 1.84 MGD IX plant in the United Kingdom spent more than 1,000 kg of salt per day and regeneration costs over a 20 plant-year life can be more than double of the initial equipment costs [Andrews and Harward, 1994]. Brines are usually either stored in landfills or discharged in the sea, which increases the risk of future watershed pollution and, hence, discourages the use of ion-exchange in many areas. In 1992, fifteen IX plants were operating in the U.S., where the process is preferred relative to other alternatives because of the simplicity of operation compared to biological denitrification [Clifford and Liu, 1993]. Furthermore, initial concerns regarding the release of synthetic substances from the resin material to the drinking water were dismissed after studies concluded that properly conditioned resins did not add organic compounds to the water. In fact, IX proved to have a positive impact upon the water quality by absorbing organic micro-pollutants such as aromatic compounds and herbicides [Dore et al, 1986]. Biological denitrification Biological denitrification is commonly used to remove nitrate from municipal and industrial waste waters both in Europe and the U.S. Nevertheless, the experience acquired in wastewater treatment denitrification was not immediately applied to drinking water because of several concerns, such as bacterial contamination, presence of residual organics and increased chlorine demand for the disinfection of the treated water [Bouwer and Crowe, 1988]. However, denitrification of drinking water is practiced in Europe, where pilot and full scale plants are currently in operation [Laîne, 1998, Liessens et al, 1993]. The selection of the process is mostly based on not adding chloride to the water and for producing only a relatively small amount of sludge as a waste product [Clifford and Liu, 1993]. Denitrification takes place under anoxic conditions and is a dissimilative biological process whereby nitrate serves as an electron acceptor, and is reduced to nitrogen gas, through a sequence of intermediates that include nitrite (NO2-), nitric oxide (NO), and nitrousoxide (N2O) [Benefield and Randall, 1982]. Denitrification can occur in both heterotrophic and autotrophic organisms [Zumft, 1997] but the heterotrophic process is usually preferred because it provides higher removal rates than the autotrophic process: respectively, 12-160 g-N m-3 hr-1 [Kapoor and Viraraghavan, 1997] and 5-62.5 g-N m-3 hr-1 [Gross and Treuter, 1986; Dries et al, 1988, Trouve and Chazal, 1999]. Unfortunately, heterotrophic denitrification has its own less positive aspects. Indeed, heterotrophic biological treatment requires an excess supply of exogenous nutrients, particularly organic carbon, which are seldom completely consumed. These nutrients remain in the treated water effluent, contributing to what is known as secondary pollution [Lemoine et al, 1991]. The removal of secondary pollutants requires extensive posttreatment which accrue substantially to the total cost of drinking water treatment and up to 5% of additional raw water wastage [Fuchs et al, 1997].

4

Significant research efforts have been devoted to engineering efficient biological systems for drinking water that overcome microbial contamination and secondary pollution of the treated water with organic residuals. Membrane bioreactors have been the object of some of these efforts and will be discussed in a later section of this review. Combined ion-exchange (IX) and biological denitrification Ion exchange and biological denitrification have been combined whereupon nitrate is removed by IX and the concentrated brine produced upon resin regeneration is denitrified in a bioreactor [Van der Hoek and Klapwijk, 1987, Clifford and Liu, 1993]. After denitrification, fresh chloride salt is added the brine, which is then recycled as the regenerating solution. The major advantage of this combined process is the reduction of the amount of brine produced, up to 95% [Van der Hoek et al, 1988, Clifford and Liu, 1992]. Furthermore, direct contact between water and microorganisms from the bioreactor can be avoided if the regenerating biomedium is filtered. Nevertheless, regular disinfection of the resin bed is required because the regenerated brine carries residual soluble organic substances that adsorb on the resin, where they support biofilm growth [Van der Hoek and Klapwijk, 1987]. Disinfection procedures must be carefully optimized since most disinfectant products will reduce the capacity and longevity of some resins [Van der Hoek et al, 1987 a,b ]. Reverse Osmosis and Electrodialysis Reverse osmosis (RO) of water is a pressure driven process whereby water molecules are forced through a membrane while ionic solutes are retained. RO can be used to remove nitrate from water. However, it is not specific for nitrate and hence reduces the total mineral content of water as well. Reverse osmosis is therefore more competitive when it is combined to remove excess hardness, sulfate or salt content. The most common problem associated with reverse osmosis is membrane fouling, which can be minimized by pre-treatment of the water. The operation costs for nitrate removal by reverse osmosis are about eight time more than for IX, which translates into U.S.A dollars (USD) as 12 per pound of nitrate-nitrogen removed [Kapoor and Viraraghavan, 1997]. In electrodialysis, ion transport is driven by the passage of a direct electric current through stacks of anion or cation specific membranes. Unlike for RO, currently, there are nitrate selective membranes that have increased the application potential of electrodialysis. Membrane fouling is a problem as much as it is in RO, and hence, the water must be pre-treated. A 1,000 m3day-1 treatment plant plant produces about 24 m3day-1 of a brine residue [Lutin et al, 1997]. The operational costs in 1995 were similar to those of RO, but have decreased significantly and were reported to be approximately USD 8.0 per pound of nitrate-nitrogen removed [Lutin et al, 1998]. Membrane bioreactors Membrane bioreactors (MBRs) combine biological or enzymatic conversion and membrane separation. Membrane bioreactors have a multitude of uses, and water treatment is one of them. In water treatment, membranes are specially sought for filtration, which enables the production of water that is free of microbes and particulate

5

material. Nevertheless, the application of membrane bioreactors for drinking and wastewater treatment for purposes other than filtration is increasing. Membrane bioreactors can be categorized as filter or contactor configurations. The main difference between the two configurations is the function of the membrane, which in the filter MBR is a filtration media and in the contactor MBR is a matrix through which individual substances are transported. Both configurations are schematically represented in Figure 1 and Figure 2. The filter configuration combines biological reaction followed by membrane filtration in order to produce a treated effluent that is free of particulate and colloidal material that was excluded by the pore size of the filtration membrane. On the other hand, a contactor configuration combines membrane absorption and transport followed by biological reaction. In most applications, the pollutant is removed from the water or from the air through the membrane structure and made available to microorganisms in the bioreactor. Examples of such systems are the membrane extractive bioreactor [Livingston, 1994] for solvent detoxification of waste-waters, the gas-liquid extraction membrane bioreactor [Beeton et al, 1991], and the membrane biofilm bioreactor [McCleaf and Schroeder, 1995].

Influen t ( co ntainin g po llutants )

R et ain ed su b sta nce N on re ta ine d su bst an ce

Po rou s m e m b ran e

M e m b ran e F ilte r

P er m ea te ( Effluen t)

f lo w o f p e rm ea te

B ior eac tor

Figure 1- Schematic of a filter membrane bioreactor and detail of the filtration process through the membrane: the stripes in the membrane illustrate the porous structure; the white arrows through the membrane represent the bulk flux of permeate

6

M e m bra n e

B ior eac tor feed

Po llu ta nt in wa te r

M e m b ran e C o n t act o r

B ior eac tor

E fflu ent

C on ve rte d p ollut an t in b io re act or m e diu m

Po llu ta nt in b io re act or m ed ium

Influen t ( with po llutant)

Figure 2- Schematic of a contactor membrane bioreactor and detail solute (dark dots) transport through the membrane.

Denitrification in Filter Membrane Bioreactors The filtration process in a filter membrane bioreactor can be classified either as dead-end, whereby the liquid is forced to flow against the membrane surface, or as cross-flow, whereby the liquid flows tangentially to the membrane surface. The cross-flow filter MBR is the type of system most frequently used for drinking water denitrification. In this type of system, an anoxic bioreactor and a cross flow filtration unit are coupled in parallel. The biomedium is pumped through the filtration module and flows tangentially to the membrane. The trans-membrane pressure (TMP) across the membrane drives the transport of fluid through the membrane, generating a permeate/filtrate. Biocrystal, a system marketed by Lyonnaise des Eaux, is one of the systems that is currently in use in Europe [Fuchs and Lebosse, 1998] Cross-flow filtration systems have high volumetric denitrification capacity, up to 72 g-N m-3 day-1 [Barreiros et al, 1998], are capable of producing a treated effluent that is free of bacteria [Chang et al, 1993], protozoa and their cysts, and remove up to 4 logs of virus counts [Jacangelo et al, 1996]. However, the treated effluent is the filtered biomedium, and therefore it contains incompletely degraded substrate or microbial soluble metabolites that were not rejected or retained by the membrane. Microfiltration membranes retain only particulate substances that are bigger than 0.02-10
m, which means that solutes that are not aggregated as colloids, or solids that do not exceed the cut off size of the pore, will not be retained. Ultrafiltration membranes will retain species that are bigger than 10-200 Å (Å = 10-10 m) which includes most macro-solutes, but not small molecules such as ions or the carbon source of the heterotrophic biological process. To conclude, filtration bioreactors are able to remove nitrate and produce a filtered effluent but cannot prevent secondary pollution of the treated water. Hence, one or more downstream treatment units are required to remove soluble organic carbon and eventually other soluble inorganic nutrients.

7

Submerged membrane processes are an alternative to cross flow filtration. In submerged membrane systems the membrane fibers are immersed directly in the bioreactor tank and a suction pump collects a permeate that filters through the membrane. The filtration is a dead-end process, whereupon a low flow vacuum pump pulls the biomedium through the membrane at a flux that is generally lower than in cross flow filtration, within 15-30 l m-2 h-1. The filtration flux performance is less than that of cross-flow filtration, but this is compensated by the advantage of reduced pumping energy requirements [Bouhabila et al, 1998]. The turbulence existing in the tank and the sweeping effect of aeration bubbles upon cake fouling removal are sufficient to maintain the flux since the trans-membrane pressure is much lower than in cross-flow filtration [Bouhabila et al, 1998. Zeeweed (Zenon Env., Canada) is a patented system that uses ultrafiltration membranes to remove turbidity and microorganisms or in place of a sedimentation basin to remove iron and manganese. Denitrification in Contactor Membrane Bioreactors A few contactor-MBRs were designed for drinking water denitrification. The main purpose of these systems was to segregate the drinking water and the microbial denitrifying culture, and therefore prevent microbial contamination and attempt to reduce secondary pollution of the treated water. As mentioned earlier, some authors named these systems as membrane-biofilms reactors because biofilms tend to form on the membrane surface. The “double flow immobilized-cell reactor” [Lemoine, 1991] and the “membraneimmobilized biofilm reactor” [McCleaf and Schroeder, 1995] are two primary contactor systems reported for denitrification of drinking water. The membranes used in these systems were porous membranes and not selected for any special selectivity towards any of the substances dissolved in the water or in the biomedium. This meant that these systems relied on dialysis (transport driven by the concentration gradient) as the main transport process through the porous membrane. In bench-scale assays, only one contactor membrane bioreactor, the doubled flow immobilized-cell reactor, claimed to consistently produce microorganism and acetate-free treated water [Lemoine et al, 1991]. For the remaining systems, bench scale studies revealed that the treated water contained incompletely degraded substrate and nutrients [Fuchs et al, 1997; Mansell and Schroeder, 1998]. Furthermore, although the membrane restricted microorganisms to the bioreactor, in the long run the system could not prevent growth of microorganisms in the treated water, where they were supported by the presence of organic and inorganic nutrients [Reising and Schroeder, 1996]. Secondary pollution in these systems could thus be explained based on the value of the diffusion coefficients of nitrate, other nutrients and the carbon source, particularly when the molecular size difference of nitrate and nutrients is similar, in which case biological pollutants may diffuse more quickly from the bioreactor to the water than nitrate can diffuse from the water to the bioreactor. Such could have been the cause for the presence of methanol treated water described by McCleaf and Mansell (1995) and by

8

Mansell and Schroeder, 1998. However, even when the diffusion coefficients of nitrate and the carbon source are very different, such as was proved by Lemoine in regard to acetate and nitrate [Lemoine et al, 1991], convection mixing through the porous membrane may explain the presence of dissolved organic substances in the water. If that is the case, then secondary pollution would be quite difficult to manage in porous contactors because convective transport is often an obligatory result of pumping solutions through a porous membrane. Overall, the only contactor using porous membranes that claimed to prevent secondary pollution was the double-flow immobilized MBR. This system used porous membranes that sandwiched an agar layer containing immobilized denitrifying microorganisms. The composite was non-porous overall because of the agar layer and hence, was resistant to convection driven mixing.

9

Chapter 3 - Materials and Methods Polluted groundwater Synthetically concocted groundwater polluted with nitrate was prepared by supplementing deionized water with 10 mg l-1 KCl, 5 mg l-1 K2SO4, 100 mg l-1 of KHCO3 and 363.6 mg l-1 KNO3. Bioreactor feed The bioreactor feed, biofeed, was prepared according to the nitrate load and the type of carbon source selected for the process: acetate or ethanol. The C/N ratio for acetate was selected to be 4 (molar ratio) and the C/N ratio for ethanol was selected to be 2.8 (molar ratio). The concentration of inorganic nutrients was defined by the amount of organic carbon. Per 0.52 gram of C (carbon) added: 0.11 g KH2PO4; 6.45 ml of salt solution (26 g l-1 NH4Cl, 2.5 g l-1 MgSO4.7H20 g l-1, 2.25 g l-1 Na2SO4 g l-1, 0.9 g l-1 CaCl2.2H2O); 2.42 ml of iron solution (0.77 gl-1 FeCl2.4H2O); 0.4 ml of micro-nutrient solution (0.1 g l-1 ZnSO4.7H2O, 0.03 g l-1 MnCl2.4H2O, 0.3 g l-1 H3BO3l, 0.2 g l-1 CoCl2.6H2O, 0.01 g l-1 CuCl2.2H2O, 0.02 g l-1 NiCl2.6H2O, 0.03 g l-1 NaMoO4.2H2O). Oxygen dissolved in the bioreactor feed was removed by heating, followed by purging with nitrogen gas. Low flux filter membrane bioreactor (LLF) The layout of the LFF is presented in Figure 3. The system combines a membrane cartridge, model Filtral® 20 (Hospal, France) and an anoxic biological reactor. Further membrane module specification are shown in Table A1. The biomedium of the biological reactor was recirculated through the inside of the fibers (lumen) of the dialyzer at a flow rate of 80 ml min-1. The total volume of biomedium was 700 ml. The bioreactor was fed continuously with 3 ml min-1 of polluted groundwater and 0.3 ml min-1 of biofeed. The solids retention time (SRT) in the system was 1.6 days. The temperature was maintained at 240C. Another pump, operated by a timer switch, removed biomedium every 4 hours. The carbon source in the biofeed was acetate The porous structure of the membrane compelled us to introduce a flow control system to prevent the recirculated water permeate from flowing back to the bioreactor. The flux of water through the dialyzer was controlled by setting three pumps in the recirculation water circuit, one before the membrane module and two after the membrane module. Pumps P1 and P2 (represented in Figure 3) recirculated the water through the shell side of the dialyzer at a flow rate of 100 ml min-1. Pump 3 was set to 3 ml min-1, which equaled the net flux permeate.

10

N2 P2

4

P3 B iofee d & Influen t W ater

E fflu ent W ater

1

3 2 T P1

B iow as te ( ce lls an d bio m ed ium )

Figure 3- Schematic of the low flux filter membrane bioreactor. Legend: 1-membrane module; 2-bioreactor ; 3-permeate vessel; 4- nitrogen tank; P1-pump 1 (100 ml min); P2-pump 1 (100 ml min-1); P3-pump 1 (3 ml min-1); T- timer switch operated pump.

1

11

N2 4 E fflu ent W ater T

1

3

B iofee d 2

T

B iow as te

Influen t w ater

Figure 4- Schematic of the diafiltration membrane bioreactor. Legend: 1-membrane module; 2-bioreactor ; 3-permeate vessel; 4- nitrogen tank; T-timer switch operated pump.

12

E fflu ent W ater

N2 3

T B iofee d 1 T

2

B iow as te ( ce lls an d bio m ed ium ) Influen t W ater

Figure 5- Schematic of the ion exchange membrane bioreactor. Legend: 1-membrane module; 2-bioreactor ; 3- nitrogen tank; T-timer switch operated pump.

13

Diafilter membrane bioreactor (DF) The layout of the DF is presented in Figure 4. The system combined a diafilter membrane cartridge, model Filtral® 20 (Hospal, France) and an anoxic biological reactor. Further membrane and module specifications are presented in Table A1. The biomedium of the biological reactor was recirculated through the inside of the fibers (lumen) of the dialyzer at a flow rate of 80 ml min-1. The total volume of biomedium was 700 ml. Biofeed was added to the bioreactor in fed-batch cycles every 4 hours. The flow rate of the biofeed was 432 ml day-1. The solids retention time (SRT) in the bioreactor system was 1.6 days. The temperature was maintained at 240C. The carbon source in the biofeed was acetate. A pump operated by a timer switch removed biomedium every 4 hours. The water circuit was continuously fed with polluted groundwater at a flow rate of 3 ml min-1. Analogous to the LFF, a flow of water in the membrane module had to be controlled by setting two pumps in the recirculation water circuit, one before the membrane module and one after the membrane module, both set at 100 ml min-1. Ion-exchange bioreactor (IE) The layout of the ion exchange bioreactor system is presented in Figure 5. The system combined composed of an ion exchange membrane module and an anoxic biological reactor. The membrane module enclosed a flat membrane Neosepta ACS (Tokuyama Corp., Japan). The membrane had an ion exchange capacity equal to 1.8-2.0 mili equivalents per gram of dry weight resing (meq g-1) and a thickness of 0.12-0.2 mm. The dialyzer had 80 ml of cell volume and 28.3 cm2 of membrane surface area. The biological reactor was coupled to one of the cells of the dialyzer. The total volume of the biomedium was 300 ml and it was continuously recycled between the bioreactor and the dialyzer at a flow rate of 150 ml min-1. The bioreactor was operated as a fed-batch, with feeding cycles every 4 hours and the SRT was 3.5 days. The carbon source in the biofeed was ethanol. The temperature was controlled at 240C and the pH and redox potential were measured on-line. A microbore peristaltic pump fed the synthetic polluted groundwater continuously at a flow rate of 0.3 ml min-1. Start-up Culture The bioreactors were inoculated with the effluent of a denitrifying continuously stirred reactor (CSTR) that had been inoculated with diluted activated sludge from the Blacksburg wastewater treatment plant (VA, U.S.A). The enrichment reactor had a SRT equal to 3.5 days and had been operating for a minimum of 15 days. The feed C/N ratio and nitrate load were the same as those of the membrane bioreactor for which the start-up culture was prepared. The start-up of the LFF and DF had a molar C:N equal to 4.0 and volumetric load equal to 330 g N-NO3 day-1 l-1. The start-up of the ion exchange MBR had a C:N equal to 2.8 (molar ratio) and a volumetric load equal to 100 g N-NO3 day-1 l-1.

14

Analytical procedures Potassium, ammonia and calcium were analyzed in a Dionex DX-120 system equipped with an ion supressor and column CS12 (10-30) (Dionex, U.S.A). The mobile phase was a solution of 20 mM of methane sulfonic acid at a flow rate of 1.0 ml min-1. Nitrate, nitrite, chloride, phosphate, and sulfate were determined in the same Dionex DX-120 system equipped with an Ion-Pac AS9-SC (Dionex,U.S.A). The anion analysis mobile phase (flow rate of 2.0 ml min-1) was a solution of 1.7 mM of NaHCO3 and 1.5 mM Na2CO3. A conductivity detector was used to detect and quantify the concentration of both ions and anions. Total organic carbon (TOC) and total inorganic carbon (TIC) were determined using a TC analyzer, Model 800 (Sievers, U.S.A). Ethanol was quantified by Head-Space GC. An Hewlet Packard Head Space autosampler 19395A (Hewlett Packard, U.S.A) was coupled to an Hewlett Packard gas-chromatographer Model 5880A equipped with a SP1200 column (Supelco, U.S.A). The protocol parameters of the HS-autosampler were as listed in Table 1 and the protocol for the operation of the GC is listed in Table 2. Table 1-Head-space operational parameters for the analysis protocol of ethanol. Oil bath temperature (o C) Injection loop temperature (o C) HS-gas carrier HS-gas carrier pressure (bar) Auxiliar gas pressure (bar) Servo air pressure (bar) Injection program start-end time (seconds)

70 74 N2 1.5