study of track-etched submerged membrane bioreactor … · study of track-etched submerged membrane...
TRANSCRIPT
STUDY OF TRACK-ETCHED SUBMERGED
MEMBRANE BIOREACTOR FOR DOMESTIC
WASTEWATER TREATMENT
OOI WAI KEONG B.Eng (Hons) NUS
A THESIS SUBMITTED FOR THE DEGREE OF MASTER OF ENGINEERING
DIVISION OF ENVIRONMENTAL SCIENCE AND ENGINEERING
NATIONAL UNIVERSITY OF SINGAPORE 2008
i
ACKNOWLEDGEMENTS
I would like to express my sincere appreciations to my supervisor Prof. Ong Say
Leong for his patient guidance and sincere advice on work and life. Special thanks
also to all the laboratory officers, friends and my family; as well as anyone who
has helped me in one way or another during my study.
Ooi Wai Keong
ii
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ...................................................................................................................... i
TABLE OF CONTENTS.........................................................................................................................ii
SUMMARY ...............................................................................................................................................vi
NOMENCLATURE..............................................................................................................................viii
LIST OF FIGURES................................................................................................................................... x
LIST OF TABLES.................................................................................................................................. xii
1 Introduction...................................................................................................................................1
1.1 Worldwide Water Shortage ..........................................................................................1
1.2 Wastewater Reclamation ...............................................................................................2
1.3 Membrane Bioreactor Development.........................................................................3
1.4 Membrane Development to Reduce Fouling in Wastewater Treatment...5
1.5 Project Objectives..............................................................................................................6
1.6 Project Scope .......................................................................................................................6
1.7 Organization of Thesis.....................................................................................................8
2 iter L ature Review..................................................................................................................... 10
2.1 r T eatment of Wastewater........................................................................................... 10
2.1.1 Anaerobic Treatment............................................................................................... 11
.1.2 2 Aerobic Treatment.................................................................................................... 12
2.2 Membrane Technology................................................................................................. 13
2.2.1. Track‐etched Membranes..................................................................................... 15
2.3 Membrane Bioreactor (MBR).................................................................................... 18
2.4 Membrane Fouling and Factors Affecting Membrane Fouling ................... 21
2.5 e M mbrane Characteristics.......................................................................................... 24
2.5.1 Configuration and Material ................................................................................... 24
2.5.2 Hydrophobicity .......................................................................................................... 25
2.5.3 Porosity and Surface Roughness ........................................................................ 26
iii
2 Pore Size........................................................................................................................ 27
2.6 lu
.5.4
S dge Characteristics .................................................................................................. 29
2.6.1 MLSS Concentration................................................................................................. 29
2.6.2 Extracellular Polymeric Substances (EPS) ..................................................... 30
2.6.3 Floc Structure and Size ........................................................................................... 31
.6.4 2 Dissolved Organic Matter (DOM) ....................................................................... 32
2.7 p O erating Conditions.................................................................................................... 34
2.7.1 Aeration and Cross Flow Velocity (CFV) ......................................................... 34
.7.2 2 SRT and HRT................................................................................................................ 35
2.8 e N ed for Nitrogen Removal........................................................................................ 36
2.8.1 Nitrification.................................................................................................................. 37
2.8.2 Impact of DO Concentration on Nitrification ................................................ 38
2.8.3 Relationship between pH and Nitrification ................................................... 40
2.8.4 Relationship between SRT and nitrification.................................................. 41
3 Materials and Methods........................................................................................................... 42
3.1 Overview ............................................................................................................................ 42
3.2 Experimental Setup ....................................................................................................... 42
3.3 Membrane Modules ....................................................................................................... 44
3.4 Operating Conditions.................................................................................................... 46
3.5 Innoculation Process..................................................................................................... 50
3.6 a S mpling Methods.......................................................................................................... 52
3.6.1 Sample Collection and Storage ............................................................................ 52
3.6.2 EPS ................................................................................................................................... 53
3.6.3 Total & Volatile Suspended Solids (TSS/VSS)............................................... 55
3.6.4 Chemical Oxygen Demand (COD)....................................................................... 55
3.6.5 Total Organic Carbon (TOC) ................................................................................. 56
3.6.6 Total & Soluble Nitrogen (TN/SN)..................................................................... 56
3.6.7 Specific Oxygen Uptake Rate (SOUR)................................................................ 57
3.6.8 Transmembrane Pressure and Flux.................................................................. 58
iv
3.6.9 urbidity ....................................................................................................................... 58
T
3.6.10 Molecular Weight (MW) Distribution............................................................ 58
3.6.11 Atomic Force Microscopy (AFM) ..................................................................... 59
3.6.12 Scanning Electron Microscopy (SEM)............................................................ 59
3.6.13 Contact Angle............................................................................................................ 60
3.6.14 Particle Size Distribution .................................................................................... 61
3.6.15 Zeta Potential ........................................................................................................... 62
3.6.16 Surface Charge ......................................................................................................... 62
3.6.17 Relative Hydrophobicity...................................................................................... 63
3.6.18 Microscopic Analysis............................................................................................. 63
4 Results and Discussions ........................................................................................................ 65
4.1 Membrane Fouling Behavior ..................................................................................... 65
4.2 e R actor Performance .................................................................................................... 68
4.2.1 COD and TOC Removal Efficiency....................................................................... 68
4.2.2 Turbidity Removal Efficiency............................................................................... 74
4.2.3 TN Removal Efficiency ............................................................................................ 75
.2.4 4 Nitrification performance...................................................................................... 78
4.3 ff E ects of Membrane Characteristics on Fouling .............................................. 79
4.3.1 Effects of Membrane Surface Charge on Fouling......................................... 79
4.3.2 Effects of Membrane Hydrophobicity on Fouling ....................................... 84
4.3.3 Effects of Surface Roughness, Porosity & Pore Structure on Fouling.85
.3.4 4 Effects of Pore Size & Particle Size Distribution on Fouling ................... 88
4.4 ff E ects of Sludge Characteristics on Fouling....................................................... 91
4.4.1 Effects of MLSS on Membrane Fouling............................................................. 92
4.4.2 Effects of Sludge Surface Charge on Fouling ................................................. 94
4.4.3 Effects of Soluble and Bound EPS on Membrane Fouling ........................ 95
4.4.4 Effects of Sludge Hydrophobicity ....................................................................... 99
4.4.5 Fractionation of DOC in the Supernatant and Permeates........................ 99
5 Conclusion and recommendations ................................................................................ 105
v
5.1 Conclusion ...................................................................................................................... 105
5.2
Recommendations ...................................................................................................... 107
References ................................................................................................................................ 109 6
vi
SUMMARY
This study was done to determine the feasibility of using track-etched membranes
(TEMs) in submerged membrane bioreactors (MBR) for the treatment of domestic
wastewater. The results obtained in this study illustrate the influence of membrane
characteristics on membrane fouling as well as on the organic and nitrogen removal
performance of the submerged MBRs for treating domestic wastewater.
Membrane characteristics such as membrane pore size, membrane surface charge,
porosity and membrane hydrophobicity were investigated. It was determined that
membrane surface charge and porosity may play a more significant role in terms of
fouling impact.
Biomass characteristics were not seen to be a major factor in this study as experiments
done on the biomass showed that the biomass exhibited constant values over the
course of the study. Running the 2 MBRs at two different sludge retention times
(SRTs) of 15 and 30 days in MBR 1 and MBR 2, respectively have also yielded data
that suggests sludge age plays a part in the excretion of different ratios of proteins and
carbohydrates and that this indirectly might impact on MBR performance.
It was found that membrane properties might only affect initial membrane fouling;
whereas operational and biomass characteristics have more impact on membrane
fouling in the case of long term operations. Other factors such as fractions of EPS
present in the mixed liquor as well as particle size of solutes in the mixed liquor might
also play a role in the fouling of the membranes.
vii
Carbonaceous and nitrogen removal of the track-etched membranes were satisfactory
with an average carbonaceous removal of more than 85% and an average nitrogen
removal of 10% for MBR 1 and 15% for MBR 2. Nitrification performance was
excellent as was the turbidity of the membrane permeates.
Further investigation is suggested on chemical cleaning of fouled membrane as in this
study, permeate fluxes of membranes were observed to increase after chemical
cleaning which suggested that chemical cleaning might affect the integrity of the
membrane.
Keywords:
domestic wastewater, fouling, membrane bioreactor (MBR), membrane characteristic,
performance, Track-Etched Membrane (TEM)
viii
NOMENCLATURE
AFM atomic force microscopy
AOB ammonia-oxidizing bacteria
ASP activated sludge process
COD chemical oxygen demand
CFV cross flow velocity
MBR membrane bioreactor
DO dissolved oxygen
DOC dissolved organic carbon
DOM dissolved organic matter
EPS extracellular polymeric substances
FS flat sheet
HF hollow fibre
HRT hydraulic retention time
MCE mixed cellulose esters
MF microfiltration
MLSS mixed liquor suspended solids
MLVSS mixed liquor volatile suspended solids
MWCO molecular weight cut off
NOB nitrite-oxidizing bacteria
OUR oxygen uptake rate
PCTE polycarbonate track-etched
PETE polyethelyene track-etched
PES polyethersulfone
PTFE polytetrafluoroethylene
PVDF polyvinylidene fluoride
PVP Polyvinylpyrrolidone
PVSK polyvinyl sulfate
ix
PWP pure water permeability
RO reverse osmosis
SCOD soluble chemical oxygen demand
SEM scanning electron microscopy
SMP soluble microbial products
SN soluble nitrogen
SOUR sludge oxygen uptake rate
SRT solid retention time
STOC soluble total organic carbon
TDS Total Dissolved Solids
TEM track-etched membrane
TMP transmembrane pressure
TN total nitrogen
TOC total organic carbon
TSS total suspended solids
VSS volatile suspended solids
UF ultrafiltration
x
LIST OF FIGURES
Figure 1‐1: Proportion of Earth's Water .....................................................................................1
Figure 1‐2: Flow Chart of Project Scope......................................................................................8
Figure 2‐1: Comparison between a conventional treatment train and an MBR.....19
Figure 2‐2: Comparison of submerged and sidestream configuration....................... 20
Figure 2‐3: Fouling potential flowchart (Judd and Judd, 2006) .................................... 23
Figure 3‐1: MBR undergoing testing of aerators and level sensors............................. 44
Figure 3‐2: Plate‐and‐frame membrane module used in this study. ........................... 46
Figure 3‐3: Pumps and control box used in setup ............................................................... 48
Figure 3‐4: Schematic diagram of MBR 1 set‐up .................................................................. 49
Figure 3‐5: Schematics diagram of MBR 2 set‐up ................................................................ 50
Figure 3‐6: Image of analyer measuring contact angle on glass.................................... 60
Figure 3‐7: Illustration of particle size analysis of feed .................................................... 61
Figure 4‐1: Fouling behavior for membranes in MBR 1.................................................... 65
Figure 4‐2: Fouling behavior for membranes in MBR 2.................................................... 66
Figure 4‐3: TCOD Removal Efficiency of all membranes for MBR 1 ............................ 70
Figure 4‐4: TCOD Removal Efficiency of all membranes for MBR 2 ............................ 70
Figure 4‐5: SCOD Removal Efficiency of all membranes for MBR 2............................. 71
Figure 4‐6: STOC Removal Efficiency of all membranes for MBR 1 ............................. 71
Figure 4‐7: Total TOC Removal Efficiency of all membranes for MBR 2 ................... 72
Figure 4‐8: STOC Removal Efficiency of all membranes for MBR 2 ............................. 72
Figure 4‐9: Comparison of permeate and supernatant TOC ........................................... 73
Figure 4‐10: Comparison of supernatant and permeate turbidity............................... 74
Figure 4‐11: DO and SOUR in MBR 2 ......................................................................................... 76
Figure 4‐12: Gram staining done on day 50 for MBR 2 ..................................................... 77
Figure 4‐13: Neisser staining done on biomass from MBR 2 on day 50 .................... 78
Figure 4‐14: Zeta Potential of Membranes used in MBR 1 .............................................. 79
xi
Figure 4‐15: Zeta Potential of Membranes used in MBR 2 .............................................. 81
Figure 4‐16: Zeta potential of supernatants and permeates........................................... 82
Figure 4‐17: Comparison of membrane hydrophobicities............................................... 84
Figure 4‐18: Surface image of each membrane from AFM analysis............................. 88
Figure 4‐19: Influent particle size distribution..................................................................... 89
Figure 4‐20: Surface image of each membrane from SEM analysis ............................. 90
Figure 4‐21: MLSS and MLVSS concentration of MBR 1 ................................................... 92
Figure 4‐22: MLSS and MLVSS concentration of MBR 2 ................................................... 93
Figure 4‐23: Protein/carbohydrate ratio for MBR 1 .......................................................... 96
Figure 4‐24: Protein/carbohydrate ratio for MBR 2 .......................................................... 97
Figure 4‐25: Fractionation of DOC in supernatant and permeates of MBR 1 ....... 101
Figure 4‐26: Mass percentage of fractions desorbed from membrane surface... 102
xii
LIST OF TABLES
Table 1.1: Comparing membranes by manufacturing procedure and structure .... 5
Table 2.1: Submerged MBR Versus Side‐stream MBR (Stephenson, 2000)............21
Table 2.2: Configuration and materials used in MBRs (Judd and Judd, 2006).......25
Table 2.3: Summary of results on pore size ..........................................................................27
Table 3.1: Properties of the MF membranes used in MBR 1.........................................45
Table 3.2: Properties of the MF membranes used in MBR 2..........................................45
Table 3.3: Operating conditions for MBR 1 ...........................................................................47
Table 3.4: Operating conditions for MBR 2 ...........................................................................47
Table 3.5: Influent characteristic for MBR 1 .........................................................................51
Table 3.6: Influent characteristic for MBR 2 .........................................................................52
Table 4.1: Total Nitrogen Removal ...........................................................................................75
Table 4.2: Nitrification Performance........................................................................................79
Table 4.3: Mean roughness values of membranes..............................................................85
Table 4.4: Pure water permeability (PWP) of membranes ............................................87
Table 4.5: Concentrations of EPS in MBRs 1 and 2 ............................................................95
Table 4.6: Concentrations of EPS in permeate and supernatant..................................98
able 4.7: EPS in dissolved organic foulants desorbed from membranes............104 T
1
1
Introduction
1
.1 Worldwide Water Shortage
Water is a finite resource and yet is an essential life-sustaining element. Drinking
water fit for life available on planet earth make up 1.7% of the total water supply with
the remainder comprising seawater. Of this 1.7%, 30.1% is ground water and slightly
more than half of it is saline (USGS, 2007) . This indicates that less than 1.5% of the
total water supply is available as fresh water for a world population that is still
experiencing high growth.
Figure 11: Proportion of Earth's Water
It is stated that, “Access to water for life is a basic human need and a fundamental
human right. Yet in our increasingly prosperous world, more than 1 billion people are
denied the right to clean water and 2.6 billion people lack access to adequate
sanitation.” (UNESCO, 2006).
2
In this surge towards economic progress, minimal priority is placed on environmental
concerns, leading to massive pollution. This has compounded the problem in making
available drinking water supplies even scarcer. Hence, newer technologies are
urgently required in order to solve this deficiency. Additionally, since a large
proportion of these 3.6 billion people are mainly living in the 3rd world countries,
technologies that are low in terms of costs and technical knowledge would be
beneficial.
1.2 Wastewater Reclamation
There are currently two sources of water that can be commonly reclaimed with our
present level of technology: seawater and wastewater. Comparing both, desalination is
cost and energy intensive, and can be applied mainly in coastal areas. Hence, reusing
wastewater is a sustainable water management approach that would result in a stable
supply of water in the long term (Ghayeni et al., 1998).
In comparison to seawater desalination, the advantage of reclaiming wastewater is
that typically a large proportion of a cities’ public water supply would end up as
domestic wastewater. This provides a reliable supply of water with slight variations of
water available at lower costs throughout the year as compared to desalination
(Khawaji, 2008). It was reported that total life cycle cost of producing reverse
osmosis (RO) water from secondary effluent was about 1.2 times cheaper than
seawater desalination. This is due to the lower total dissolved solids (TDS) present in
the secondary effluent compared to seawater, thus lowering the osmotic pressure
required to produce an equal amount of fresh water (Cote et al., 2005).
3
1
.3 Membrane Bioreactor Development
Numerous physical, chemical and biological processes have been explored and
developed over the years to treat wastewater with varying types and extent of
contamination. In recent years, membranes have grown to be a viable option.
Membrane Bioreactors (MBRs) is a variation of biological treatment incorporating the
use of membrane (Henze, 2001). The first attempt coupling membrane technology
with biological reactors for wastewater treatment in North America is the membrane
Sewage Treatment (MST) system developed by Dorr-Oliver Inc. in the 1960s. The
system was marketed in Japan under license to Sanki Engineering with some success.
By late 1980s and early 1990s, other companies such as Thetford Systems in the USA
and Zenon Environmental, now one of the leading membrane companies, developed
an MBR system that eventually led to the introduction of the first ZenoGem® iMBR
process in the early 1990s. In the mid-1990s, commercialized Kubota immersed flat-
sheet MBR was made available in Japan. Thereafter, Zenon and Kubota continued to
introduce more new technology into the market, becoming the big players in this
industry.
According to a technical market research report entitled “Membrane Bioreactors in
the Changing World Water Market”, from US-based Business Communications Co
Inc (BCC), the current global market is valued at an estimated US$216.6 million in
2005 and expected to approach $363 million in 2010.
In another market research conducted by Frost & Sullivan, it is predicted that the US
and Canadian MBR markets sum up to US$32.2 million, and projected to reach
US$89 million in 2010. Hence, the potential growth of MBR market can be perceived
as optimistic. This level of optimism arises due to the trade-off between several
4
drivers of MBR technology, as well as the possible limitations that generally hinders
the growth of the market (Research, 2006).
The following section lists the main drivers and restraints that play a major role in
influencing the MBR market (Judd and Judd, 2006).
i) New and more stringent enforcement on effluent discharge due to higher
awareness of water-borne diseases;
ii) Global water scarcity;
iii) More state incentives to promote wastewater technology;
iv) Decrease in investment costs due to technological advancements; and
v) Increasing confidence in and greater acceptability of MBR technology.
However, one of the main issues of membrane technology is fouling, which is the
accumulation of substances on membrane surfaces and/or within the membrane pores
that result in the deterioration of membrane performance. This drives up either
operational or capital costs through more frequent cleaning of membranes and
replacement of membranes in irreversible fouling respectively.
Fouling reduction is thus the focus of many studies and research. Many studies have
been done to investigate the effects of operational characteristics like hydraulic
retention time (HRT) and biomass characteristics such as mixed liquor suspended
solids (MLSS) (Choi et al., 2006). However, there has been lacking information with
regards to the impact of membrane material on MBR fouling.
5
1.4 Membrane Development to Reduce Fouling in Wastewater Treatment
The principle property of a membrane is the separation of 2 components. In
wastewater treatment, the membrane rejects pollutants consisting of suspended or
dissolved particles and allows solids-free water to be produced.
Membrane materials are very diverse, various membranes with properties such as
high surface porosity, anisotrophic structure, narrow pore size distribution as well as
ability to resist some thermal and chemical attack have been found to be suitable for
use in MBRs to treat wastewater.
There are two different types of membranes used in MBRs - organic and inorganic
membranes. Organic membranes are made of polymers while inorganic membranes
can be made of ceramics. The following table summarizes some existing types of
membranes as well as their production methods (Stephenson, 2000).
Table 1.1: Comparing membranes by manufacturing procedure and structure
Membrane Type Manufacturing Procedure Structure
Ceramic Pressing, sintering of fine powders 0.1-10 µm pores
Etched Polymers (Track-Etch)
Radiation followed by acid etching 0.5-10 µm cylindrical pores
Symmetric Microporous
Phase inversion reaction 0.05-5 µm pores
Integral Asymmetric Microporous
Phase inversion reaction followed by evaporation 1-10 nm pores at membrane surface
Composite Asymmetric Microporous
Application of thin film to microporous membrane 1-5 nm pores at membrane surface
6
In the most commonly used phase inversion, a polymeric solution is cast to produce a
thin layer of material. Precipitating the polymer in water produces the porous side of
the membrane; the permselective side of the membrane is then produced by
evaporation of the polymer solvent to produce the side of lower permeability,
resulting in an anisotropic membrane structure.
In track-etched membranes, the membranes are made by passing the membranes
through radiation and acid etching process, which results in uniform cylindrical pores
as well.
1.5 Project Objectives
This study seeks to evaluate the feasibility of using track-etched membranes in MBRs
for the treatment of municipal wastewater and to address the deficiencies in fouling
studies by:
i) Evaluation of fouling potential of track-etched membranes in MBRs treating
municipal wastewater.
ii) Comparison of operational performances of track-etched membranes in MBRs
treating municipal wastewater.
1
.6 Project Scope
In order to achieve the above objectives, the study is divided into two parts.
i. In part one, track-etched membranes are studied to investigate how the
7
membrane, sludge and operating characteristic affect the fouling potential in
MBRs for long term operations.
ii. Evaluate the impact of membrane characteristics on permeate flux, permeate
quality and membrane fouling.
iii. Investigate under the same sludge and operating conditions, the effects of
various membranes properties on membrane fouling. Membrane properties
addressed include:
i. Surface roughness,
ii. Surface charge of membrane,
iii. Membrane hydrophobicity,
iv. Membrane pore size, and
v. Its relation to the fouling performance of the membranes
In part two, the study is done under the same sludge and operating conditions:
i. Compare filtration performance of track-etched membranes with the
performance of the Polytetrafluoroethylene (PTFE) membrane in terms of
water quality and removal of nitrogenous and carbonaceous compounds.
ii. Deduce and evaluate the suitability of track-etched membranes for
treatment of municipal wastewater.
8
A flowchart summarizing the aforementioned scope of work is provided in Figure 1-2.
Figure 12: Flow Chart of Project Scope
1
.7 Organization of Thesis
The rest of the thesis is divided into the following four chapters:
i. Chapter 2 – Literature Review
This chapter provides a more detailed explanation of what has been
summarized in chapter 1. It will also present reviews of published literatures,
covering topics relevant to this study. Strengths and shortcomings of certain
studies will also be made.
9
ii. Chapter 3 - Materials and Methods
This chapter describes the bench-scale submerged MBR setup and its
operating conditions used in the study. It also comprises of the detailed
sampling procedures and various analytical methods used in the study.
iii. Chapter 4 – Results and Discussion
This chapter presents the findings obtained from the experiment carried out as
described in Chapter 3. A detailed discussion pertaining to the results obtained
in this study and that from other studies will be provided. The chapter will be
divided into two major separate parts covering the two phases as described in
the objective.
iv. Chapter 5 – Conclusions and Recommendations
This chapter summarizes the major conclusions from this study and based on
the findings obtained, recommendations will be provided for future studies.
10
2
Literature Review
2
.1 Treatment of Wastewater
Numerous processes have been explored and developed over the years to treat
wastewater depending on the type and extent of contamination. These include
physical, chemical and biological treatment processes. The variety of different
pollutants would dictate as to the type of treatment process to be applied.
In a municipal wastewater plant, physical and chemical processes would be coupled
with biological processes for wastewater treatment that are rich in organics. In cases
where organics are low in concentration, processes such as ion exchange, coagulation,
flocculation, clarification, filtration and membrane treatment may be carried out as the
treatment method.
Biological treatment can be further divided into 2 separate processes; aerobic and
anaerobic processes. For many years, aerobic treatment has been the dominant
treatment method for organic wastewaters while anaerobic treatment has been
traditionally applied for sludge digestion (Haandel and Lettinga, 1994). However,
with the need for wastewater reclamation, newer technologies have been developed to
produce higher quality effluent that is difficult to achieve using conventional
processes. These new technologies include membrane technology in both aerobic and
anaerobic systems.
11
2.1.1 Anaerobic Treatment
Anaerobic treatment is the treatment of wastewater without the use of elemental
oxygen. It involves a series of conversion from higher molecular weight organic
compounds, resulting in CH4, CO2 and other gases as final products.
Proliferation of this process has been rapid as advances were made, in part due to the
1970s oil crisis when energy costs escalated. Developments such as anaerobic filters
and upflow anaerobic sludge blanket resulted in increased efficiency and utilization of
anaerobic processes (Haandel and Lettinga, 1994). However, some inherent
limitations such as the low yield and long doubling times of the microorganisms
themselves, especially the acetogens and methanogens still exist. Thus to achieve the
aim of high efficiency, high concentration of biomass is needed in the reactor
(McCarty, 2001).
While anaerobic processes are able to achieve relatively good removal efficiencies,
effluents from anaerobic reactors rarely meet the discharge standards required. This is
because while it is able to remove a substantial amount of carbon material, nitrogen
concentration is unaffected. This means that polishing has to be carried out in order to
meet the discharge standards. Another option is to couple a membrane with the
anaerobic system but this system is currently still under research as opposed to
aerobic MBRs which have already been adopted widely in many countries (Jeison and
Van Lier, 2008).
12
2.1.2 Aerobic Treatment
Aerobic treatment can be defined as the process by which microorganisms decompose
complex organic compounds in the presence of oxygen and use the liberated energy
for reproduction and growth. Such processes include extended aeration, trickling
filtration, and rotating biological contactors. The dominant aerobic process in use
today is the conventional activated sludge process (ASP).
This process is able to degrade organic matter efficiently and is versatile in treating
nutrients such as nitrogen or phosphorus. This can be easily done through the addition
of anaerobic and anoxic tanks to the process train.
In the ASP, the mixed liquor concentration inside the aeration basin is achieved by
controlling the recycle sludge rate and sludge wastage rate. In an ASP, the sludge
retention time (SRT) and the hydraulic retention time (HRT) determine the aeration
basin volume, the mixed liquor concentration as well as the treatment efficiency. In
such a situation, it is difficult to decouple the SRT and HRT from each other (Metcalf
& Eddy.).
The main issue that can cause the ASP to fail is poor settling of biomass in the
secondary clarifier, whereby the sludge settling is highly dependant upon the
operational conditions in the aeration tank. (Jenkins, 2004). In cases where the sludge
is not able to settle properly, biomass may be washed out in the effluent. This in turn
creates two problems. Firstly, the effluent is unable to meet discharge standards and
secondly, the loss of biomass from the aeration tank means reduced removal
efficiency for the wastewater plant (Tandoi et al., 2006).
13
Another issue is the larger amounts of sludge generated in comparison to that of
anaerobic systems. Hence larger sludge treatment facilities are required.
If the final effluent is to be used for wastewater reuse, the treated effluent from the
ASP would still be required to go through tertiary filtration by microfiltration (MF) or
ultrafiltration (UF). The filtrate would then pass through the reverse osmosis (RO) for
removal of total dissolved solids (TDS) and residual pathogens.
2
.2 Membrane Technology
There are many different types of membranes (MF, UF, Nanofiltration (NF) and RO).
The degree of selectivity dependant on membrane pore size (Bhattacharyya and
Butterfield, 2003).
Membranes used in municipal wastewater such as MF tend to be of the coarsest sized
membranes and are capable of rejecting particulate materials. They are also
predominantly pressure-driven and have common elements of a purified permeate
product and a concentrated retentate waste. The rejection of these contaminants
however places a constraint on the membrane surface due to membrane fouling.
Membrane fouling refers to the gradual accumulation of rejected contaminants on the
membrane surface, leading to a reduction in flux at a given transmembrane pressure
(TMP), or conversely, an increase in TMP for a given flux, reducing the permeability
of the membrane which is the ratio of flux to TMP (Hillis et al., 2000). This in turn
exists as a limitation to the use of membranes for wastewater treatment. Such a
phenomenon, which is pertinent to all membrane-related processes, leads to a
reduction in productivity.
14
Besides the membrane characteristics that cause fouling, the membrane module
configuration in a MBR may also play a part in fouling due to the way proteins or any
substances are binded to the membranes (Ghosh and Wong, 2006). This is because the
mounting and orientation of the membrane in relation to the flow of water is crucial in
determining the overall performance.
The membrane should be configured so as to possess:
i. High membrane area to bulk volume ratio,
ii. High degree of turbulence for mass transfer promotion on the feed side,
iii. Low energy consumption and cost per unit membrane area,
iv. Design that facilitates cleaning, and
v. Modular design.
Various membrane configurations include;
i. Flat sheet (FS),
ii. Hollow fibre (HF),
iii. Tubular; capillary tube,
iv. Pleated filter cartridge, and
v. Spiral-wound.
However, only the first three configurations are used in MBRs due to their ability to
promote turbulence and their ease of cleaning.
15
2.2.1. Track-etched Membranes
Track-etched membranes have precisely controlled cylindrical pores and narrow pore
size distribution. The precise pore size of the screen filters is achieved by exposing
polymeric film to charged particles in a nuclear reactor, which leaves tracks in the
film. The pore density is controlled by the duration of time that the polymeric film is
exposed to the charged particles. The tracks on the polymeric film are then dissolved
with a chemical solution to form cylindrical pores. Varying the temperature, strength
and exposure time of the etching solution controls pore sizes (Yamazaki et al., 1996).
Track-etched membranes exhibit a number of properties that give the material a
unique overall performance. Exploiting different combinations of these properties has
allowed the development of applications by which track-etched membranes offer
alternatives that other membranes may not provide. Although most polymeric
materials such as polyvinylidene (PVDF) (Zhao et al., 1991) can be made using the
track-etched method, commercially, polycarbonate (PC) and polyester (PE) have been
most commonly used (Sterlitech Corporation, 2008).
Polycarbonate membranes are by nature more hydrophobic and hence typically have a
wetting agent, polyvinylpyrrolidone (PVP) that permits it to be hydrophilic for
polycarbonate with a contact angle of approximately 70 to 90 degrees. Polyester
membranes are by nature hydrophilic and hence do not have PVP coatings on them
(Osmonics, 2000).
Track-etched membranes do not have an inherent charge. They do, however, act in the
same way as the dielectric film in a capacitor, on which static electricity will build up.
16
This static charge will cause the membrane to act as if it has a charge, but that charge
is not actually a property of the membrane; it is the result of the environment in which
the membrane is placed in. The implication is that it is possible to induce charges in
these membranes and hence apply it to filter certain materials or liquids which are
either opposite in charge to the membrane and vice versa.
Track-etched membranes are inherently smooth, a property that is preserved
throughout the membrane manufacturing process. This smooth, flat surface is ideally
suited for microscopic applications. This factor (surface roughness) can be especially
important where a defined focus must be maintained or topographical defects may
interfere with quantification (e.g., epifluorescence). Thickness variations in track-
etched membranes are kept within ±1 µm with roughness not exceeding 50 nm (peak
to valley) (Apel, 1995).
Typically, the pore size of track-etched membranes can vary greatly depending on the
manufacturing procedure. This wide range of pore sizes allows them to be used for a
multitude of applications (Chittrakarn, 2002). They also have very controlled pore
size distribution due to the manufacturing process, which means pore size is usually
between –20 and 0% of the stated pore size. The intralot coefficient of variance for
pore size is typically 2–3%. In addition to pore size, pore density (or porosity) can be
controlled. It typically ranges from 1 x 105 to 6 x 108 pores/cm2, and can vary within
certain limits in relation to pore size (Koul et al., 1991).
The internal shape of the pores can also be controlled to allow the formation of a truly
cylindrical pore structure. An even pore structure is essential for controlled and
consistent filtration properties in the membrane—this is particularly true for
nanoporous track-etched membranes (membranes with a pore size less than 100 nm)
17
(Ferain and Legras, 2003). Although most track-etched membranes have a structure in
which the pores are perpendicular to the membrane surface, it is possible to create
parallel pores with controlled angles to the surface. Track-etched membranes are
manufactured commercially with either PC or PE polymers, both of which are
unreactive to most biological materials. This very low biological activity makes them
well suited for In Vitro Diagnostic (IVD) applications, in which very low levels of
binding or interference are required. The low biological activity of these track-etched
membranes also allows cells to grow on the membrane surface. In addition, the
physical stability of the materials against heat and chemical attack allows them to be
included in a number of different device formats and processes (Rao et al., 2003).
Besides this, superior strength of up to 3000 psi can be attained with the appropriate
holder.
The thin nature means that holdup volume or depth effects are practically nonexistent,
especially when compared with cellulose or glass fiber materials that can result in
considerable sample retention. However, the very thin nature of the materials can
cause problems during handling and processing, hence limiting large-scale
applications.
Due to the low porosity of the membrane, the liquid flow rate through the membrane
is very low for small-pore-sized materials and can be controlled by the choice of pore
size and pore density.
As in the case of other screen-type membrane filters compared to cartridge filters,
particle capture is more efficient comparing the same nominal pore sizes, possibly
indicating a more accurate separation cut-off (Vignati et al., 2006). Traditionally,
track-etched membrane is applied in laboratory filtration because small particles could
18
be retained on the membrane surface and analyzed. Track-etched membrane is
limitedly used in the process filtration, such as purification of deionized water in
microelectronics and filtration of beverages, due to a higher dirt loading capacity and
throughput required (Apel, 2001). In contrast, other membranes such as
polytetrafluoroethylene (PTFE) membrane has a structure of polymer nodules
connected by thin fibers (interwoven sponge-like microstructure) and a wider pore
size distribution than track-etched membranes. PTFE membrane is a porous and thick
membrane, meaning that particles are retained throughout the tortuous paths within
the membrane matrix as well as on the surface of the membrane.
In summary, the track-etched membrane has several advantages as compared to
conventional membranes used in water treatment.
i. Flexibility in manufacturing to cater for different cut-offs of specific
elements
ii. Ability of manufacturing to have even sized pores to have more
accurate cut-off of colloids and particles
iii. Can be produced in with pore sizes in parallel to a specific angle to the
surface.
iv. Can be inert versus heat, chemical and biological attacks
2
.3 Membrane Bioreactor (MBR)
Membrane Bioreactors (MBRs) are an evolution from the traditional ASP. It is a
technology that combines the conventional ASP with MF or UF. MBR operations are
similar to that of an ASP but, without the need for secondary clarification and tertiary
19
steps like sand filters, which does the job of solid liquid separation in an ASP. This
also means that several tertiary treatments can be combined to deliver excellent
quality effluent (Gunder and Krauth, 1998). In terms of performance, MBR can retain
not only bacteria and colloidal material, but also high molecular weight components
either present in the wastewater or released through bacterial lyses. This causes high
viscosity of the reactor content. In addition, high MLSS can be retained in the reactor.
This allows for more effective conversion of organics in comparison to the
conventional ASP, which is capable of maintaining a MLSS concentration of ~3000-
3500 mg/l. One of the consequences is that a high aeration capacity is required to
counteract fouling of the membrane and to account for heavily decreased oxygen
transfer coefficients (Le-Clech et al., 2003). Differences in the configuration for an
ASP and a MBR system are illustrated in Figure 2-1.
Figure 21: Comparison between a conventional treatment train and an MBR
20
Advantages for adapting an MBR system are,
i) Reduced footprint,
ii) High quality effluent due to complete solid-liquid separation,
iii) Independent control of SRT and HRT,
iv) Can be operated at low F/M ratios and sludge production, and
v) Ease of retrofitting existing plants, especially in the use of submerged MBR
systems.
However, in spite of the many advantages of MBR, the major problem has always
been the costs of membranes and membrane fouling (Judd, 2008).
Currently, there are two competing configurations for MBR, the submerged MBR
system and the side-stream MBR system (Figure 2-2).
Figure 22: Comparison of submerged and sidestream configuration
Each has its own advantages and disadvantages as shown in Table 2.1:
21
Table 2.1: Submerged MBR Versus Sidestream MBR (Stephenson, 2000)
With higher demands for water quality, MBR systems are seen as an important
technology to be used in both water and wastewater treatment in the near future (Judd,
2008).
2
.4 Membrane Fouling and Factors Affecting Membrane Fouling
In order to make the submerged MBR more viable, membrane fouling, a crucial
determinating factor of MBR performance must be reduced to the minimum (Le-
Clech et al., 2006). Membrane fouling is due to a series of physicochemical and
biological mechanisms, which relates to increased deposition of solid material onto
the membrane surface (blinding) and within the membrane structure (pore plugging).
The resistance imparted by these layers would then depend on the thickness of the
layer, feedwater composition and the flux throughout the membrane (Chang et al.,
2002).
As membranes foul, the permeability decreases, which results in less water
throughput per unit volume and hence increased cost either through additional
cleaning of the membranes or in the case of irreversible fouling, replacement of the
entire membrane unit.
22
Recent studies have been conducted on other alternative modifications for more
effective and economical methods to control membrane fouling in the MBR (Yang et
al., 2006). This includes: (1) modification of membrane module and reactor design
such as optimization of the membrane packing density and the location of aerator,
(2) permeate flux control and adoption of intermittent suction mode to reduce cake
formation on membrane surface, and (3) removal of the foulant using back-washing
and chemical cleaning.
Membrane characteristics, such as membrane material, pore size and hydrophobicity
are also important factors influencing the rate and extent of membrane fouling.
Particularly, results of fouling experiments using polymeric and/or ceramic
membranes revealed that there was a close relationship between membrane material
and membrane fouling in the MBR (Judd, 2004); (Yamato et al., 2006). However,
there is still little information available with regards to the impact of membrane
material compared to other characteristics such as sludge and operating conditions on
MBR fouling, and thus, further in-depth investigation is required for better
understanding. Fouling can be related to 3 factors in MBRs as shown in Figure 2-3.
23
Figure 23: Fouling potential flowchart (Judd and Judd, 2006)
As shown in Figure 2-3, sludge characteristics and operating conditions are inter-
related as either one can affect the other. Section 2.5 to 2.7 will describe in detail the
difference and how each factor affects fouling.
24
2
.5 Membrane Characteristics
Key membrane design parameters are (1) configuration, that is the geometry and flow
direction, (2) surface characteristics (normally represented by the pore size and
material but also including properties such as the surface charge, hydrophobicity and
porosity, pore tortuosity and shape, and crystallinity), and (3) inter-membrane
separation. In submerged MBRs, membranes in the upper molecular weight cut of
(MWCO) range of UF and the smaller pore-sized MF offer sufficient rejection and
reasonable fouling control under the conditions applied. It must be designed with the
following in mind (Stephenson, 2000):
i. Mechanically & chemically robust to withstand the stresses imposed during
the filtration and cleaning cycles.
ii. Readily modified to give a hydrophilic surface, which then makes them more
resistant to fouling.
iii. Manufactured at low cost, especially for submerged MBR since they are
operated at low fluxes and hence have typically larger membrane areas.
2.5.1 Configuration and Material
As stated earlier, submerged MBRs have been favoured over the sidestream
configuration due to operational reasons, which relates mainly to the usage of aeration
as a means of suppressing fouling through shear generation. Membrane configuration
is also a determining factor to the effectiveness of aeration as a means of fouling
control. Flat-sheet (FS) and hollow-fibre (HF) membranes comparison was done in
25
one study (Gunder and Krauth, 1998), which demonstrated that FS membrane could
attain higher permeability than HF membrane. However, another study showed that
HF membrane fouled slightly less than FS membrane of the same pore size, and
permeability was not recovered following cleaning with water (Hai et al., 2005). In
terms of materials, choices are limited as there are currently limited polymeric
membranes suitable for MBR requirements. Ceramic and other membranes are still
not widely used due to the prohibitive costs incurred during the manufacturing
process (Stephenson, 2000). Some examples of suitable polymeric membrane
materials used in MBR are shown in Table 2.2.
Table 2.2: Configuration and materials used in MBRs (Judd and Judd, 2006)
2.5.2 Hydrophobicity
It has been established that hydrophobic interactions take place between membrane
materials and solutes, microbial cells and Extracellular Polymeric Substances (EPS).
Membrane fouling is expected to be more severe with hydrophobic than hydrophilic
membranes. A conflicting study was conducted by (Fang and Shi, 2005), which
suggested that membranes of greater hydrophilicity are more vulnerable to deposition
of foulants with hydrophilic nature. However, it must be noted that the most
hydrophilic membrane is the most porous and this is also a factor contributing to the
26
fouling phenomenon. Conclusions from a study using membranes of similar
characteristics by (Chang et al., 2001a) is that solute rejection was mainly via
adsorption onto or sieving by the cake deposited on the membrane and to a lesser
extent, direct adsorption onto the surfaces of membrane pores and surface.
However, one main drawback of the above studies is that the time durations during
which the MBR were operated were short and hence may not reflect the true fouling
rejection of the membrane under extended operations.
2.5.3 Porosity and Surface Roughness
A study concluded membrane roughness and porosity were possible causes of fouling
in a study (Fang and Shi, 2005). Four MF membranes of similar pore sizes in the
range of 0.2 to 0.22 µm exhibited more fouling as porosity increased, whereas least
fouling was observed on the track-etched polycarbonate membrane. Relative pore
resistance was also found to be significant in the track-etched PC, PVDF, mixed
cellulose esters (MCE) and polyethersulfone (PES) membranes with relative pore
resistance of ~0%, 2%, 11% and 86%, respectively. With regards to surface
roughness, track-etched membranes have been known to possess one of the lowest
surface roughness among membranes due to its manufacturing techniques (this is a
distinctive characteristic acquired during the manufacturing process) and hence would
be ideal to test if surface roughness is a factor for membrane fouling.
27
2.5.4 Pore Size
Impact of pore size on membrane fouling is postulated to be related to the feed
solution characteristics, (Lee et al., 2005) especially the particle size distribution.
However, conflicting results exist on the pore size and hydraulic performance as
summarized in Table 2.3. This could have been a consequence of the changing and
complex nature of the biological suspensions in MBR systems and the large pore size
distribution of the membranes used (Chang et al., 2002); (Le-Clech et al., 2003).
Table 2.3: Summary of results on pore size
Membranes tested (Optimum Pore size in Bold)
Test Duration
Comments Reference
0.1, 0.22, 0.45 µm 20 h - (Zhang et al., 2006)
70kDa (~0.45 µm), 0.3 µm 8 h - (Choi et al., 2005a)
30kDa (~0.015µm) 2 h CFV = 0.1 m/s
(Choi et al., 2005b)
0.3 µm 2 h CFV = 0.35 m/s (Choi et al., 2005b)
0.1, 0.2, 0.4, 0.8 µm - - (Lee et al., 2005)
200kDa (~0.8 µm), 0.1, 1.0 µm 3 h Flux-step test (Le Clech et al., 2003)
0.3, 1.5, 3.0, 5.0 µm 25 min - (Chang et al., 2001b)
0.3, 1.5, 3.0, 5.0 µm 45 d - (Chang et al., 2001b)
0.4, 5.0 µm 1 d,
<50 d
No effect for <50 d;
5.0 µm for 1 d
(Gander et al., 2000)
0.01, 0.2, 1 µm (No effect) Few h Flux-step test (Madaeni et al., 1999)
0.05, 0.4 µm - - (Chang et al., 1994)
28
Experiments comparing MF and UF membranes at a crossflow velocity (CFV) of 0.1
m/s demonstrated that MF membranes provide a hydraulic resistance two times that of
a UF membrane. However, there was no difference in the Dissolved Organic Carbon
(DOC) rejection of both membranes after two hours. This indicates the presence of a
dynamic membrane layer that provided the perm-selectivity rather than the membrane
substrate itself (Choi et al., 2005b).
It is conventional thinking that smaller pores gives better membrane performance by
rejecting a wider range of materials, with regards to their size, thus, increasing cake
(or fouling layer) resistance. For membranes with larger pore sizes, the layer is more
readily removed and less likely to leave residual pore plugging or surface adsorption.
It is the latter and related phenomena, which causes reversible and irreversible
fouling. Gander et al (2000) discovered a study that in isotrophic membranes without
surface hydrophilicisation, there was greater initial fouling for larger pore membranes
under constant flux and significant flux decline when smaller pore-size membranes
were used over long term operation under constant pressure.
Comparison carried out on 4 membranes (Lee et al., 2005) showed that the largest
pore size used (0.8 µm) had a slightly higher supernatant concentration containing
higher molecular weight molecules while at the same time experienced less fouling.
Le-Clech et al. (2003) also showed that sub-critical fouling resistance and fouling rate
increased linearly with membrane resistance, which in turn was inversely proportional
to decreasing pore size. The results suggested that a dynamic layer of greater overall
resistance was observed with more selective membranes operating under sub-critical
conditions. This supports the statement that membranes with larger pore sizes tend to
have decreased deposition onto the membrane surface compared to internal
adsorption.
29
Higher pore aspect ratio (pore surface length/pore surface width) was also discovered
to be a factor affecting fouling. Reduced fouling was observed with higher pore aspect
ratios i.e., more elliptical and less circular (Kim et al., 2004).
However, it can be seen from Table 2.3 that none of the studies were operated for
long durations and membranes used were of different characteristics (e.g. materials,
pore size distribution etc.). Hence the results yielded may not conclude that membrane
pore size induces higher tendency of membrane fouling.
2
.6 Sludge Characteristics
MBR membrane fouling can be affected by the interactions between the membrane
and the biological suspensions in the reactor. The operating conditions and the
influent characteristics in turn influence the microbial consortia within the reactor.
2.6.1 MLSS Concentration
Numerous studies have been carried out to determine the effect of MLSS
concentration on fouling and it is debatable as to whether this parameter by itself
would be a major contributor to membrane fouling.
High MLSS concentrations can accelerate membrane fouling via rapid deposition of
sludge particles on the membrane surface (Sato and Ishii, 1991). Other studies
contradicted this theory by showing that critical flux was inversely related to MLSS
concentration from 0 to 10 g/l (Madaeni et al., 1999). Studies by Le-Clech et al.,
(2003) however, showed that there was no significant increase in critical flux with
30
MLSS ranging from 4 to 8 g/l, and a slight reduction in fouling when MLSS
concentration reached 12 g/l. A study also concluded that at MLSS levels below 6 g/l,
there was reduced membrane fouling, insignificant change between 8 to 12 g/l and
significant effect at MLSS concentration more than 15 g/l (Rosenberger et al., 2005).
Given the lack of a clear correlation between MLSS concentration and any specific
foulant characteristics, MLSS concentration appears to be a poor indicator of biomass
fouling propensity.
2.6.2 Extracellular Polymeric Substances (EPS)
Extracellular Polymeric Substances (EPS) is used as a general term, that encompasses
all classes of autochthonous macromolecules such as carbohydrates, proteins, nucleic
acids, lipids and other polymeric compounds found at or outside the cell surface. They
are responsible for the structural and functional integrity of the bioflocs or biofilm,
which are also the key components that determine the physiochemical and biological
properties. In addition, it also enhances the survival and robustness of the bacteria
against convection flow, as well as penetration of antimicrobial agents (Flemming and
Wingender, 2001).
Research has shown that a high amount of EPS could have a negative impact on the
membrane performance in a MBR (Chang and Lee, 1998); (Nagaoka et al., 1998);
(Nagaoka et al., 2000); (Rosenberger and Kraume, 2002).
These researchers had demonstrated that EPS, accumulated in the
aeration basin and on the membrane is responsible for the increase in membrane
filtration resistance. Having a higher amount of EPS will result in a greater flux
31
reduction, regardless of the physiological state of the activated sludge and the
membrane material; and that EPS content in activated sludge can be a probable
index for membrane fouling in MBR. Membrane autopsy performed by Cho
and Fane (2002) revealed that significant amount of EPS was present in the fouled
membrane. They reported that membrane fouling profile can be characterized by
initial conditional fouling, followed by an intermediate extended period of slow TMP
rise and finally, by a sudden transition to a rapid TMP rise. EPS was responsible for
the initial stage of fouling by pore closure and surface fouling of the membrane (Cho
and Fane, 2002).
However, one possible source of biasness could arise due to the fact that numerous
EPS extraction methods exist. This could inevitably lead to inaccurate quantification
and hence comparison of extracted EPS. This was observed when different EPS
extraction methods were compared (heating, EDTA, cation exchange resin and
formaldehyde) under various conditions. It was reported that effectiveness in
extracting EPS from biomass is dependent on the extraction method. Overall, the
method using formaldehyde plus sodium hydroxide (NaOH) was found to be the most
effective (Ge et al., 2007).
Thus, this has to be taken into consideration when comparing results obtained from
different EPS extraction methods.
2.6.3 Floc Structure and Size
A study carried out to compare the aggregate size distribution in ASP and MBR
sludge (Cabassud et al., 2004) suggests a distinct difference in terms of mean particle
32
sizes. ASP particles had a mean diameter of 160 µm while that in an MBR had a
bimodal distribution of 5 to 20 µm and 240 µm. The high concentrations of small,
particulate materials in the MBR can be postulated to be a result of complete retention
of suspended solids by the membrane.
Other studies suggest a similar trend when partial characterization was carried out for
MBR flocs up to a size of 100 µm. Floc sizes ranging from 10 to 40 um with a mean
of 25µm had been reported (Bae and Tak, 2005).
Given the large physical flocs size, it is impossible that pore plugging by the flocs can
occur. They will be impeded by drag forces and shear-induced diffusion from
depositing on the membrane surface. However, it is possible that EPS produced by the
flocs may indirectly affect the clogging of the membrane channels.
2.6.4 Dissolved Organic Matter (DOM)
DOM consists mainly of C, N and P, which are also the main elements of bacteria
cells. Majority of these DOM are cellular components and soluble microbial products
(SMP) that are released into the bulk solution during cell lysis, cell synthesis, grazing,
and diffusion through the cell membrane or other unknown biological purposes.
The concept of SMP in dissolved matter is a relatively new development with
researches started only in the recent few years. A study by Ng et al., (2005) has
shown that greater fouling was due to SMP rather than from the biomass, which was
at MLSS concentration of 4 g/l. This implies that SMP characteristics have a
significant impact on membrane permeability whereby it absorbed onto the membrane
surface, blocking membrane pores and/ or forming a gel structure on the membrane
33
surface. This fouling layer provided a nutrient source for biofilm development and
resist hydraulic flow to the membrane (Ng et al., 2005).
Experiments comparing the permeate and the SMP seem to suggest that SMP
materials are retained at or near the membrane; (Brookes et al., 2003; Evenblij and
van der Graaf, 2004; Lesjean et al., 2005).
SMP comprises humic and fulvic acids, polysaccharides, proteins, nucleic acids,
organic acids, amino acids, antibiotics, steroids, exocellular enzymes, siderophores,
structural components of cells and products of energy metabolism (Parkin and
McCarty, 1981). Hence it is tough to isolate and identify specific components of SMP
and to attribute a particular component as the fouling factor. Therefore, a practical
method is to use TOC and COD as a measurement of SMP concentration (Barker and
Stuckey, 1999).
Another interesting point to note is that many studies have shown a direct relation
between the carbohydrate level in SMP fraction and membrane fouling in a MBR
(Lesjean et al., 2005; Rosenberger et al., 2005). The hydrophilic nature of
carbohydrate may explain the apparently higher initial fouling propensity with mainly
hydrophilic membranes in MBR. (Drews et al., 2005), however, found that fouling
propensity of carbohydrate portion of SMP (SMPC) changes during unsteady MBR
operation, indicating that fouling may vary in large-scale plants. There has also been a
lack of studies on the hydrophobic protein fraction of the SMP. It can be presumed
that such materials will also play a part in fouling since significant amounts of protein
are retained by the membrane. Also majority of the studies stated above used mainly
synthetic wastewater (Rosenberger et al., 2005) , in which the organics are more
easily biodegraded and hence, SMP values may be lower than those in real systems
34
(Shin and Kang, 2003). Hence SMP production may be a function of the chemical
nature of the feed (Shin and Kang, 2003).
2
.7 Operating Conditions
2.7.1 Aeration and Cross Flow Velocity (CFV)
Stirring and aeration processes are major operations needed for effective wastewater
treatment, especially in aerobic MBR operating systems.
In a submerged MBR, aeration provides a two-fold effect. In addition to providing
oxygen to biomass and keeping solids in suspension, aeration is also used to scour the
membrane surface to minimize membrane fouling. Coarse bubble diffuser is generally
used in submerged MBR. The rising bubbles will provide a turbulent crossflow
velocity (approximately 1 m/s) over the surface of the membrane. This helps to
maintain flux through the membrane by reducing the build up of material at the
membrane surface and thereby increasing the operational cycle of the system{Meng,
2008}.
There is a corresponding relationship between the aeration rate and the desposition of
the cake layer. However, it has been demonstrated that there was an optimum value
for the cake removal and the cake-removal efficiency of aeration did not increase
proportionally with the increase in the airflow rate (Han et al., 2005).
Experiments have proven both extremes, low and high aeration intensity had a
negative influence on membrane permeability (Meng et al., 2007). High aeration
intensity would result in severe breakup of sludge flocs and promote the release of
35
colloids and solutes from the microbial flocs to the bulk solution. Under aeration
intensity of 150 l/h, Brownian diffusion was the main back transport mechanism for
membrane foulants, which was ineffective in cake layer removal efficiency. The cake
resistance resulting from an aeration rate of 150 l/h was more than two times that of
400 l/h and 800 l/h. this indicates that aeration has high impacts on the removal of
cake layer (Meng et al., 2007).
2.7.2 SRT and HRT
The SRT and HRT are related to the sludge production in a MBR. SRT can produce
significant effects on biomass properties in an MBR. With perfect solid-liquid
separation by the membrane, MBR can maintain high MLSS concentration and long
SRT. A higher biomass concentration would give rise to higher treatment efficiency.
To date, there are several conflicting reports on the effect of SRT on membrane
fouling of submerged MBR. Some researchers have reported that the MLSS
concentration in a long SRT operation will increase the membrane filtration resistance
(Yamamoto et al., 1989); (Shimizu et al., 1996); (Lee et al., 2003), while others have
demonstrated a reduced fouling rate at a longer SRT (Fan et al., 2000); (Lee et al.,
2001).
As both EPS and SMP are products associated with microbial activity, their
compositions and concentrations are influenced by the operating SRT. (Huang et al.,
2000) reported that accumulated soluble organic substances or SMP in MBR could
have a negative influence on the membrane permeability. On the other hand, higher
EPS content associated with shorter SRT operation was found to increase membrane
fouling (Nagaoka et al., 2000). Both EPS and SMP are the biological products
36
associated with microbial activities and the characteristic of biomass is often
controlled by process operating parameters such as SRT, HRT and F/M ratio.
2
.8 Need for Nitrogen Removal
The need for removal of nitrogen in the wastewater is an important treatment step
necessary in most wastewater plants. Nitrogen is the most common element on earth
with nitrogen gas making up ~78% of the earth’s atmosphere.
Nitrogen exists in wastewater in the form of organic nitrogen, ammonium, nitrite and
nitrate. Nitrogen in the wastewater is initially in the form of urea and is hydrolyzed by
bacteria to ammonium. Nitrogen entering the treatment plants normally consist of
approximately 60% ammonium, 40% organic ammonia and negligible amounts of
nitrite and nitrate.
Since ammonium is a macro-nutrient for algae and plants growth, discharging it
discriminately can result in devastating consequences on the receiving body.
Ammonium through nitrification can also impose stress on the oxygen demands in the
receiving body since it takes about 4.57g O2 to oxidise per gram of NH4+-N
(Rittmann and McCarty, 2001).
In addition, ammonia is corrosive to a certain extend and will limit its water
reclamation. Health concerns are also important as infants who consume high nitrate
concentration can suffer from Methaemoglobinaemia or commonly known as “Blue
Baby Syndrome”. In order to fufill the objective of reclaiming high quality water from
wastewater, a nitrification-denitrification process may be required in the treatment
process of wastewater.
37
2.8.1 Nitrification
It is important to note that nitrification does not remove nitrogen beyond 5 to 10% of
ammonium by the bacteria as substrate; it only removes the oxygen demand imposed
by the ammonium present in the wastewater. The only way to release nitrogen is
through reducing the oxidized nitrogen to nitrogen gas through denitrification.
Although in certain special cases such as high MLSS and high substrate
concentrations, total nitrogen removal as high as 30% may be possible in the aerobic
portion of an anoxic-aerobic MBR (Wang et al., 2005).
Nitrification is widely known to be carried out by two distinct groups of bacteria,
namely ammonia-oxidizing bacteria (AOB) which oxidize ammonia to nitrite and
nitrite-oxidizing bacteria (NOB) which oxidize nitrite to nitrate. Nitrifiers are
chemolithoautotrophs which are obligatory aerobes (Rittmann and McCarty, 2001). It
uses inorganic NH4+-N as the source of electrons with oxygen as the terminal electron
acceptor and it utilizes inorganic carbon for cell synthesis.
The biogeneration of nitrate from NH4+-N is a two-step process:
2NH4+ + 3O2 2NO3
- + 4H+ + 2H2O Δε = + 84kcal/mol (AOB) (2.1)
2NO2- + O2 2NO3
- Δε = +17.8kcal/mol (NOB) (2.2)
Overall:
NH4+ + 2O2 NO3
- + 2H+ + H2O Δε = + 91.8kcal/mol (2.3)
Various parameters influence the nitrification process. These parameters include
dissolved oxygen (DO), temperature, pH, organic loading and SRT.
38
2.8.2 Impact of DO Concentration on Nitrification
Nitrifiers are obligate aerobes and oxygen is strictly required for its growth. Based on
Equation 2.3, the theoretical oxygen demand for complete nitrification is 4.57 g O2/g
NH4+-N. This oxygen consumption is substantial when compared to the
corresponding need for organic decomposition. Thus the overall oxygen supply to
any nitrification process should be increased by this additional amount on top of that
required for the oxidation of the organic compounds. In reality, the oxygen demand
for nitrification should be between 4.25 and 4.57 g O2/g NH4+-N as Equation 2.3 does
not consider the synthesis of biomass and utilization of CO2 by nitrifiers (Eckenfelder
et al., 1992). Assuming the empirical formula of nitrifier is C5H7NO2, the overall
reaction of both AOB and NOB together with the synthesis of nitrifier cells can be
expressed as:
NH4+ + 1.815 O2 + 0.1304 CO2 → 0.0261 C5H7NO2 + 1.973 H+
+ 0.973 NO3− + 0.921 H2O (2.4)
Although nitrifiers are strict aerobes, it can survive under a prolonged period under
lack of oxygen conditions. This explains why nitrification can still be achieved in a
sequencing batch reactor (SBR) and pre-denitrification process where nitrifiers
experience alternate aerobic and anoxic environments. However, it is still unknown
to many researchers how nitrifiers survive under prolong anaerobic conditions, as well
as the duration that nitrifiers can survive in anaerobic condition without affecting its
nitrification performance. An understanding in this would definitely help to optimize
the process design, be it an SBR or pre-denitrification process.
39
It has been reported that low DO concentration can inhibit the nitrification process
during the activated sludge process. Rittmann and McCarty (2001) also reported that
continued operation with a DO level below the KO2, where KO2 is the saturation
constant for DO, (KO2 of AOB = 0.5 mg /L; KO2 of NOB = 0.68 mg/L), will lead to
biomass washout and high NH4+ concentration in the effluent. It is observed that
under a low DO level environment, NH4+ is unable to be fully oxidized to NO3
- , but
will only be partially oxidized to NO2-. This will result in nitrite accumulation as high
as 60 mg/l at a HRT of two to about four days (Hanaki et al., 1992).
(Ruiz et al., 2006) found that nitrite accumulation took place at DO concentration of
between 0.7 and 1.5 mg/L and that ammonia oxidation was affected at a DO
concentration of 0.5 mg/L. Thus a minimum DO concentration of 2.0 mg/L is usually
recommended for complete nitrification performance to be achieved in a treatment
process (Tchobanoglous et al., 1991); Eckenfelder and Grau, 1992). In addition,
Grady and Lim (1980) reported that heterotrophic bacteria have a maximum growth
rate of five times and yields of two to three times that of autotrophic nitrifying
bacteria.
Contradictory to the above, (Park and Noguera, 2004) have reported complete
nitrification occurring under a low DO concentration environment. They
demonstrated that given sufficient time for acclimatization, complete nitrification
could be achieved in a chemostat reactor with a DO concentration of as low as 0.12
mg/L.
This could indicate that DO concentrations reported are not representative of what is
in the mixed liquor. In a submerged MBR, there is often a much higher level of DO
concentration due to the need for aeration rate than that which is present in an ASP
40
system. It has been shown that nitrification can take place in an ASP system given the
right pH and SRT. Thus in an MBR with much higher DO than that provided in an
ASP, nitrification should not be inhibited by DO levels.
Studies discussed above did not mention the aeration devices and the efficiencies of
the devices as well as the viscosity of sludge and concentration of MLSS, given the
mass transfer from the device to water, there might be differences in the actual
amount of oxygen present in the reactors.
2.8.3 Relationship between pH and Nitrification
Nitrifiers are extremely sensitive to pH and operate within a narrow optimal range.
Tchobanoglous and Burton (1991) reported that the optimal pH range lies between 7.5
and 8.6 while USEPA (1975) suggested that nitrification rate can be assumed to be
constant at pH between 7.2 and 8.0. A study was conducted using pure culture
isolated from activated sludge to examine nitrification inhibition assays. It was found
that the optimal pH for Nitrosomonas and Nitrobacter are 8.1 and 7.9, respectively
(Grunditz and Dalhammar, 2001). Another study conducted showed that nitrification
rate in a biofilm reactor declined rapidly when pH fell below 7.0 and ceased at pH
within the range of 6.5 to 6.7 (Boller et al., 1994).
The effect of pH on nitrification can be linked to the availability of alkalinity in the
mixed liquor. Based on Equation (2.1), H+ ions are produced during ammonia
oxidization. These H+ ions will react with the alkalinity (OH-, HCO3-, CO3
2-) available
in the wastewater. Theoretically, nitrification will consume 8.64 mg HCO3-/mg NH4
+-
N. However, if insufficient alkalinity is available in the wastewater, pH could reduce
41
to 6.2 and nitrification would stop in activated sludge (Painter, 1970). Thus, addition
of lime or soda ash may be required to maintain pH at an optimal level when
alkalinity is insufficient. In domestic wastewater where the pH is normally above
neutral, pH should not be an influencing factor of nitrification in MBR (Nagaoka and
Nemoto, 2005).
2.8.4 Relationship between SRT and nitrification
The activity of the nitrifying bacteria in activated sludge is highly dependant on
temperature and SRT. It was observed that in an aerated reactor, near complete
nitrification was accomplished at 20oC with SRT of 2.7 days (Randall et al., 1992).
When the temperature decreased to 10oC, even with SRT of 5 days, the nitrification
effectiveness was lower than 65%. The effect of temperature in the range of 10 – 20oC
was not observed when SRT was 15 days. The increase of temperature by 1oC would
result in about 10% reduction of SRT required for nitrification (Sinkjaer et al., 1994).
Generally, SRT above 20 days eliminate unfavorable influence of low temperatures,
and thus stabilizing the nitrification process.
42
3
Materials and Methods
3
.1 Overview
Two MBRs were set up for the purpose of this study.
In MBR 1, the effects of membrane characteristics such as surface roughness, surface
membrane charge, membrane hydrophobicity, together with sludge and operating
characteristics on fouling potential were investigated. At the same time, comparison
was done on the filtration performance based on water quality and removal of
nitrogenous and carbonaceous compounds by the track-etched membranes and PTFE
membrane with the same pore size.
In MBR 2, the effect of pore sizes in relation to the fouling performance was
evaluated with respect to the sludge and operation characteristic. Filtration
performance based on water quality, organics and nitrogenous compounds were also
carried out as a basis for comparison with results from MBR 1.
3
.2 Experimental Setup
In the case of MBR 1, an existing reactor was modified by compartmentalizing into 3
sections using 3 removable baffle walls. Each compartment has a capacity of 8.1 L
with a total working volume at 24.3 L and sufficient freeboard of 20% of the working
height.
MBR 2 was newly constructed with its setup similar to the first reactor but with four
removable baffle walls and a smaller working volume of 7.5 L per compartment,
giving a total volume of 30 L with similar freeboard of 20% working height. The
43
reactor dimensions are 263 mm x 250 mm x 800 mm (LxBxH) with a thickness of 10
mm.
Both MBRs were fabricated using acrylic plastic. In order to ensure homogeneity of
the mixed liquor in both reactors, the membranes were placed in the same reactor.
Each membrane was placed parallel to each other in an equally spaced compartment,
which was partitioned using baffle plates. Each plate was sized at 230 mm in length
and 693 mm in height, with a thickness of 5 mm. This fabrication simulated a
complete-mixed flow reactor setup. Four sampling ports were each aligned parallel to
the partitions, 20 mm above the base (three in the case of MBR 1). The fabricated
membranes were then mounted in between two baffles plates located above each air
diffuser.
Three pairs of steel rods were used as level sensors. The water level in the reactor was
monitored using two conductive level controllers (Omron 61F-GP-N2, Japan) housed
in the control box. One for high water level cut off and the other for the low water
level cut off. A timer is also housed in the control box to control the suction pumps so
as to allow for membrane relaxation.
The setup is shown in Figure 3-1.
44
Figure 31: MBR undergoing testing of aerators and level sensors
3
.3 Membrane Modules
Three new membranes PCTE, PETE and PTFE with the same 0.1 μm pore size and
four new PVDF membranes of pore sizes 0.03 μm, 0.09 μm, 0.2 μm and 0.35 μm
were mounted carefully onto both sides of a membrane module using Araldite epoxy
glue. In order to ensure water and air tightness, the epoxy glue was applied both under
the membrane and two additional layers applied on top of the membrane edges by
carefully brushing with a smooth brush. A plastic tube was attached to the top of the
each module from where the effluent was discharged. The properties of the membrane
45
modules are summarized in Tables 3.1 and 3.2. An example of the membrane module
used in this study is shown in Figure 3-2.
Table 3.1: Properties of the MF membranes used in MBR 1
Item PETEa PCTEa PTFEa
Table 3.2: Properties of the MF membranes used in MBR 2
Manufacturer
Pore structure
Pore size (µm)
Effective Surface Area (m2)
GE-Osmonics
Cylindrical
0.1
0.11
GE-Osmonics
Cylindrical
0.1
0.11
Sumitomo Electric
Non-cylindrical
0.1
0.11
a PETE = Polyester; PCTE = Polycarbonate; PTFE = Polytetrafluoroethylene.
Item PVDF (Polyvinylidene)
Manufacturer
Pore structure
Pore size (µm)
Effective Surface Area (m2)
-
Cylindrical
0.03, 0.09, 0.2, 0.35
0.11
46
Figure 32: Plateandframe membrane module used in this study.
3
.4 Operating Conditions
MBR 1 was operated at a HRT of 8 h and a SRT of 15 d, while in MBR 2, a HRT of 8
h and a SRT of 30 d. Due to the large quantities of feed water required, MBR 1 was
operated 1st and MBR 2 was started later.
47
The operating conditions are shown in Tables 3.3 and 3.4.
Table 3.3: Operating conditions for MBR 1
Parameter (Reactor 1) Values
Volume (l) 24.3
SRT (day) 15
HRT (hr) 8
Feed flow rate (ml/min) 50.6
Suction rate (ml/min) 63.3
Air flow rate (l/min) 4-5
Sludge wastage flow (l/day) 1.62
Suction-Idle Time (Min-Min) 8-2
Table 3.4: Operating conditions for MBR 2
Parameter (Reactor 2) Values
Volume (l) 30
SRT (day) 30
HRT (hr) 8
Feed flow rate (ml/min) 62.5
Suction rate (ml/min) 19.5
Air flow rate (l/min) 4-5
Sludge wastage flow (l/day) 1
Suction-Idle Time (Min-Min) 8-2
Municipal wastewater collected from the Ulu Pandan Water Reclamation Plant was
sieved using a 1.0 mm fine screen. The wastewater was then fed to the MBR from a
continuous well-stirred 400 L feed tank. The feed wastewater was then fed into the
MBR using peristaltic pumps (Masterflex® L/S ® Standard Drive, Cole-Parmer
48
Company). The MBRs were operated at ambient temperature (25 to 27oC) with the
pH in the aerobic zone controlled using a pH controller (HD PH-P1, Etatron DS,
Italy) at 7.0 ± 0.1 using 0.25 M Na2CO3 as required. Excess activated sludge was
discharged daily from the MBRs through a sampling port to maintain the desired
SRT.
Individual membrane suction pumps for each membrane ensured that there was a
constant membrane flux from each membrane. Each membrane module was attached
to a pressure gauge (ZSE50F-02-22L, Japan) by a tube before it was connected to a
suction pump. The pumps and the control box that were used in operations are
illustrated in Figure 3-3.
Figure 33: Pumps and control box used in setup
The effluent pumps were controlled by a timer (Omron H3GR, Japan), which was
calibrated for an 8 min on, 2 min off cycle to allow for periodic membrane relaxation.
49
In a cross-flow filtration system like a submerged MBR, shear force is a critical
requirement to control the cake layer formation and to attain a stable flux. Hence, one
diffuser was installed beneath each membrane in both MBRs. The diffusers were
aligned parallel to the width of each membrane module so that shear forces were
induced by the turbulence of air and liquid to control cake layer formation. Placement
below the membrane module facilitates the upward movement of air and promotes
liquid turbulence. Compressed air controlled by airflow meters (Aalborg Instruments
& Controls, Inc., NY, USA) supplied through the air diffuser which act to provide
both good mixing of the mixed liquor within the aerobic zone and at the same time a
effective scouring on the membrane surface. The schematics of MBR 1 and MBR 2
are shown in Figures 3-4 and 3-5, respectively.
Figure 34: Schematic diagram of MBR 1 setup
50
Figure 35: Schematics diagram of MBR 2 setup
When fouling has occurred, both physical and chemical cleaning were carried out.
Physical cleaning was carried out by wiping the membranes with a soft sponge
followed by rinsing with tap water. Then the membranes were cleaned chemically by
immersing in 1% (w/v) HCl for 0.5 h followed by 0.5% (w/v) sodium hypochlorite
(NaClO) for 2 h.
3
.5 Innoculation Process
Both MBRs were seeded with activated sludge (filtered with 1-mm sieve) from the
aeration basin of a local Water Reclamation Plant. Domestic wastewater from the
same plant was collected weekly as feed for this study. The wastewater was stored at
51
4oC until its usage was required. In order to minimize any temperature effects on the
influent wastewater on the performance of the MBRs, the domestic wastewater would
be left to warm to ambient temperature over a time period of approximately 7-8 h
prior to its transferring into the feed holding tank.
The feed holding tank was equipped with top mount stirrer to keep the influent
wastewater homogenous. The MBR was then continuously fed with wastewater from
this common feed holding tank.
As the quality of domestic wastewater was always changing, influent characteristic
into the reactors were changed and in turn, would affect the characteristics of the
sludge such as MLVSS concentration in the reactors.
The influent characteristics for MBR 1 and 2 are as shown in Tables 3.5 and 3.6.
Table 3.5: Influent characteristic for MBR 1
Influent (Reactor 1) Values
pH 6.4 ± 1.3
COD (mg/l) 378.6 ± 171.5
TN (mg/l) 52.7 ± 5.9
TSS (mg/l) 244.0 ± 99.9
VSS (mg/l) 175.4 ± 51.7
Number of samples taken = 48
52
Table 3.6: Influent characteristic for MBR 2
Influent (Reactor 2) Values
pH 6.6 ± 0.9
COD (mg/l) 513.2 ± 254.6
TN (mg/l) 55.8 ± 8.5
TSS (mg/l) 488. ± 271.2
VSS (mg/l) 423.3 ± 177.3
Number of samples taken = 48
3
.6 Sampling Methods
3.6.1 Sample Collection and Storage
Feed and effluent samples were collected from their respective inlet and outlet
tubings. Mixed liquor samples were collected from the sampling port located at mid
height of each zone. Samples were always collected starting from the effluent ports
before proceeding upstream to the collection of sludge and finally to the feed. This
was to ensure minimal disturbances to the upstream process during sample collection
at the downstream.
To obtain the supernatant from the mixed liquor and soluble portion of the feed and
membrane effluents, the collected samples were centrifuged (Kubota highspeed
refrigerated centrifuge 6500, Japan) at 4oC for 10 minutes at 9,000 rpm before
undergoing filtration through a 0.45µm filter (Membrane Filter: GN-6 grid 47mm,
53
0.45µm, Pall Corporation, NY, USA). Samples were stored immediately at 4oC in the
cold room and analyzed within a week.
3.6.2 EPS
3
.6.2.1 EPS Extraction
EPS was extracted using the thermal treatment method with slight modification of the
method used by Forster (Ge et al., 2007). This method was known to be the most
effective extraction method for activated sludge, since it releases a large amount of
extra-cellular polymers from the flocs and causes less cellular disruption than other
methods such as EDTA and sodium hydroxide treatment. However, the breakdown of
cells is possible during thermal treatment, which might cause cell contents to be
released into the EPS solution. For measuring EPS, 25 ml of activated sludge was
centrifuged at 5000 g for 5 min at 4oC (Kubota highspeed refrigerated centrifuge
6500, Japan) then followed by filtration to remove the bulk solution. After discarding
the supernatant, the remaining sludge was washed and re-suspended in milli-Q water.
The extracted solution was obtained by placing the re-suspended liquid in a water
bath at 80oC for 10 min. After heat treatment, samples were then centrifuged at
12,000 rpm for 10 min at 4oC and the supernatant was separated by filtration for
subsequent proteins and carbohydrates measurement.
54
3
.6.2.2 Total Carbohydrate
Total carbohydrate concentration was determined using the phenol/sulphuric acid
assay (Dubois, 1956) using glucose (Amresco, Anhydrous D-Glucose) as the
calibration standard. Briefly, 2 mL of sample was first mixed with 1 mL of Phenol
Reagent (5% phenol solution), followed by 5 mL of concentrated sulphuric acid. The
solutions were mixed immediately by vortexing and allowed to stand at room
temperature for 30 min. Absorbance was measured at 490 nm using a
spectrophotometer (HACH, DR/4000U, CO, USA). Standards were prepared by
diluting various volume of glucose stock solution (100 µg/mL) in distilled water.
They were then used to obtain a calibration curve to calculate the carbohydrate
concentration.
3
.6.2.3 Protein
Protein concentration was determined using the modified Lowry’s Method (Lowry et.
al, 1951) with borvine serum albumin (BSA, Sigma-Aldrich, B-4287) as the
calibration standard. Briefly, 1 mL of sample was mixed with 5 mL of Assay Mix
using a vortex mixer. Assay Mix was prepared with 50 mL alkaline reagent (0.1 M
NaOH, 2% Na2CO3, 0.5% Na Tartrate and 0.5% Na Dodecylsulfate) and 0.5 mL
copper reagent (1% CuSO4.5H2O). After 10 min of incubation at room temperature,
0.5 mL of diluted Folin-Ciocalteu reagent (1 N) was added into the solution and well
mixed immediately using a vortex mixer. The solutions were allowed to stand at
room temperature for another 30 min. Absorbance was taken at 660 nm using a
spectrophotometer (HACH, DR/4000U, CO, USA). Standards were prepared by
55
diluting various volume of BSA stock solution (1 µg/µL) in distilled water. They were
then used to obtain a calibration curve to calculate the protein concentration.
3.6.3 Total & Volatile Suspended Solids (TSS/VSS)
Glass micofibre filter (GF/F, Whatman, UK) were rinsed with distilled water and
baked at 550°C for 20 min and weighed prior to analysis. Samples were filtered using
the filter and a vacuum pump. Sample was dried in an oven (MEMMERT ULM 6,
Germany) at 103 – 105°C for at least 1 h and allowed to cool in a desiccator before
being weighed again to measure the TSS. They were then ignited in a furnace
(Thermolyne 48000, IA, USA) at 550°C for 20 min followed by cooling and weighing
as previously described to get the VSS value. Total suspended solids and volatile
suspended solids were measured in accordance to the Standard Methods (APHA,
2005).
3.6.4 Chemical Oxygen Demand (COD)
For soluble COD, the samples were filtered with a 0.45 µm membrane filter (Pall,
USA) and a vacuum pump. The supernatant was then collected for analysis.
COD was conducted using the closed reflux method (Block heater: HACH COD
Heater, Model 16500-10, CO, USA) in accordance to the Standard Methods (APHA,
2005).
56
3.6.5 Total Organic Carbon (TOC)
Total Organic Carbon Analyzer (Shimadzu TOC-VCSH, Japan) with ASJ-V (Auto
Sampler and Injector) was used to determine the organic carbon concentration of the
samples. All samples were diluted to less than 25 mg/L before analysis. The method
used was 680oC catalytically-aided combustion oxidation.
For soluble TOC, the samples were filtered with a 0.45 µm membrane filter (Pall,
USA) and a vacuum pump. The supernatant was then collected for analysis.
This is in accordance to the Standard Methods (APHA, 2005).
3.6.6 Total & Soluble Nitrogen (TN/SN)
TN and SN concentrations were measured using a Shimadzu TOC analyzer
(Shimadzu TOC-VCSH, Japan) coupled with total Nitrogen Measuring Unit (TNM-10,
Shimadzu, Japan) with ASJ-V (Auto Sampler and Injector). T-N in the feed
wastewater was also quantified using the TOC-V instrument by the combustion (710
ºC) method with catalytic oxidation.
3
.6.6.1 Ammonia Nitrogen (NH3-N)
NH3-N was measured by the HACH DR/4000 spectrophotometer coupled with
HACH Test ‘N’ Tube Vials Kit for high (0 to 50.0 mg N/L) and low range (0 to 2.5
mg N/L) of NH3-N. The testing procedure was as stated by the manufacturer in the
HACH Method 10031 and 10023 respectively.
57
− −
3
.6.6.2 Nitrite and Nitrate ( and ) 2NO 3NO
−2NO and concentration were measured after filtering through a 0.45-µm pore
size GN-6 grid 47-mm membrane filter (Gelman Science, USA). Samples were
analyzed using the ion chromatography method (Ion Chromatograph: Dionex DX-
500, CA, USA) in accordance to the Standard Methods (APHA, 2005).
−3NO
3.6.7 Specific Oxygen Uptake Rate (SOUR)
The in-situ SOUR of the aerobic zone mixed liquor was measured in accordance to
the Standard Methods (APHA, 2005). Briefly, mixed liquor collected from the
aerobic zone was placed in a 300 mL BOD bottle with a magnetic mixer to produce
an entirely homogenous sample. The dissolved oxygen (DO) concentration was
monitored with a DO meter (YSI, Model-58, Europe) and recorded by a plotter
(Linseis L6012B, N.J, USA) until it reached below 1 mg/L. Oxygen Uptake Rate
(OUR) was taken as the slope of the linear portion of the DO decline curve versus
time and the SOUR was then calculated by dividing the OUR over the MLVSS
concentration of the sample in mg/L.
58
3.6.8 Transmembrane Pressure and Flux
The transmembrane pressure in kPa was registered once daily from the pressure gauge
which was connected to the permeate pipe from the membrane modules. The
permeate flux was consistently monitored to maintain it at desired flux.
3.6.9 Turbidity
Turbidity for membrane effluent and influent was determined using a Hach
Turbidimeter Model 2100N (Hach, CO, USA).
3.6.10 Molecular Weight (MW) Distribution
MW distribution were determined using a 50-ml stirred ultrafiltration cell (amicon®
model 8050, Millipore Corporation, MA, USA) using 44.5 mm Millipore disc
ultrafiltration membranes. Three membranes with nominal molecular weight cut offs
(MWCOs) of 100,000 (100K), 10,000 (10K) and 1,000 (1K) daltons were used in
succession with the highest MW first and lowest MW last. Pure nitrogen was used to
pressurize the cell. The pressure in the ultrafilter was kept constant at 30 psi. Samples
taken after each of the filters were analysed to determine specific TOC.
To fractionate the DOC in the supernatant and permeate waters, Supelite DAX-8
(Supelco, USA) and Amberlite XAD-4 (Rohm and Haas, Germany) resins were used
together with the Amicon stirred cell (Model 8050, Millipore, MA, USA) and
ultrafiltration membranes (Ultracel series, Millipore, MA, USA) mentioned above.
The organic matter in the waters was fractionated into three groups; namely,
59
hydrophobic (DAX-8 adsorbable), transphilic (XAD-4 adsorbable), and hydrophilic
(neither DAX-8 nor XAD-4 adsorbable) fractions.
3.6.11 Atomic Force Microscopy (AFM)
Surface roughness of a target membrane and its surface image were observed using an
AFM instrument (Dimension 3100TM, Veeco Metrology Group, NY, USA). The
membrane samples dried at room temperature were fixed with a double sided tape and
analyzed in tapping mode and phase contact to observe the membrane morphology
and to measure the membrane surface roughness. Small pieces were cut from each
membrane, attached onto metal disks using a double-side tape and set to a magnetic
sample holder located on top of the scanner tube. All the AFM analyses were
undertaken at room temperature of 25 ± 1 °C.
3.6.12 Scanning Electron Microscopy (SEM)
Surfaces of all membranes were analyzed using a Scanning Electron Microscope
(SEM) (JEOL Model JSM-5600LV, Japan) with a resolution of 3.5 nm
(magnification: ×25 to ×300,000). All samples were dried prior to doing SEM so as to
eliminate the vaporization of moisture in the vacuum environment during SEM
scanning. This was done using a critical point dryer in the case of sludge samples and
an oven at 37oC for approximately 30 min to 1 h for virgin membranes.
60
Membrane samples were then placed and fixed using double-sided tape on a circular
stub. It was then placed in a sputter coater to coat the samples with a layer of platinum
prior to SEM viewing.
3.6.13 Contact Angle
The contact angle measurement for membrane hydrophobicity was carried out
according to the sessile drop technique by a goniometer (VCA Optima Systems, AST
Product Inc., USA). A drop (0.5–1 µL) of deionized water with a microsyringe was
placed on the membrane surface. The contact angle of a membrane was calculated
from the average of the contact angles at both left and right sides of the drop. Contact
angle cannot give an absolute value of the membrane hydrophilic/hydrophobicity but
allows a comparison between membranes of interest. An example of a measurement is
shown in Figure 3-6.
Figure 36: Image of analyer measuring contact angle on glass
61
3.6.14 Particle Size Distribution
The particle size distribution was analyzed by collecting the samples from feed after
centrifuging and filtration. Five different pore size cellulose nitrate membrane filters
(Whatman, UK) were used in the analysis, namely 5 μm, 1 μm, 0.65 μm, 0.45 μm and
0.1 μm. After centrifuging the feed samples at 9000 rpm for 10 min at 4oC, the sample
was first filtered through the 5 μm filter as shown in Figure 3-7. A portion of the
sample was collected while the remaining filtrate was filtered through a 1-μm filter.
The samples were filtered through filters in descending order of pore size and the
respective filtered samples were collected. The samples were then analyzed by
measuring TOC concentrations using the methods as stated previously in 3.6.5.
Figure 37: Illustration of particle size analysis of feed
62
3.6.15 Zeta Potential
Zeta potential of the membrane surface was determined using an electro kinetic
analyzer (EKA, Anton Paar, Austria) which adopted the streaming potential method
for surface charge analysis. The zeta potential measurement was carried out using a
solution of 10 mM NaCl at pH values ranging from 2 to 10 and at about 25°C.
Detailed protocol for zeta potential measurement is described elsewhere (Childress
and Elimelech, 1996).
The zeta potential of supernatant of the mixed liquor and membrane permeate was
quantified using a ZetaPals (Brookhaven Instruments, USA) instrument. This zeta
potential analyzer is designed for use with macromolecular solutions and suspensions
of particles with diameters varying from 0.005 µm to 30 µm.
3.6.16 Surface Charge
The surface charge of microbial floc was determined by the titration method (Morgan
et al., 1990) which used polybrene and polyvinyl sulfate (PVSK) as the cationic and
anionic standards, respectively. A mixed liquor sample (1 mL) was diluted to 100 mL
with Milli-Q water and mixed with an excess amount (5 mL) of 0.001 N polybrene
standard solution. Standard solution of 0.001 N PVSK was used to titrate against the
excess amount of polybrene to a colorimetric endpoint (blue to pink/purple) indicated
by a few drops of toluidine blue. An equal volume (1 mL) of polybrene diluted with
the same amount of Milli-Q water (100 mL) was used as a blank.
63
The surface charge was calculated according to equation (3-1):
Surface meqech (arg 000,1)()1 ××
×−=−
MVNBAMLVSSg (3‐1)
where, A is the amount of PVSK (mL) added to the sample, N the normality of PVSK,
B the amount of PVSK (mL) added to blank, V the amount of sample (mL) used, and
M is the concentration of MLVSS (g L-1).
3.6.17 Relative Hydrophobicity
Relative hydrophobicity of sludge was measured by mixing with n-hexane and
allowing phases to form (Rosenberg et al., 1980). The measurement was performed
according to the method modified by (Chang and Lee, 1998). Sludge relative
hydrophobicity is expressed as the ratio of MLSS concentration in the aqueous phase
after emulsification (MLSSa) to MLSS concentration in the aqueous phase prior to
emulsification (MLSSb).
lativeRe 100)1((%) ×−=b
a
MLSSMLSS
cityHydrophobi (3-2)
3.6.18 Microscopic Analysis
Due to some performance issues in MBR 2, microscopic analysis was carried out. A
light microscope (Olympus BX41, Japan) was used to analyze the microbial types in
the activated sludge. A 10x ocular was used, together with objectives that magnify the
64
image 10, 40 and 100x. Resulting magnification is hence 100x, 400x and 1000x times
respectively. A wet mount was prepared for observing the living activated sludge.
Gram staining was also performed in the analysis of biotic community. Gram staining
first colors the bacteria blue using carbol gentian violet, then cells were washed with
alcohol to differentiate between Gram-positive and Gram-negative bacteria. Safranin
was then used to counter stain the Gram-negative bacteria so that it can be visible
under microscopic analysis. Neisser staining requires a mixture of methylene blue and
crystal violet and followed by addition of Chrysoidin Y. After rinsing off with tap
water, the neisser positive cells can be identified with a microscope.
65
4 Results and Discussions
4
.1 Membrane Fouling Behavior
In this study, both MBRs were operated with a constant flux to maintain a HRT of 8 h
throughout the study. Membrane fouling is indicated by an increase in its suction
pressure (TMP). The membranes’ suction profiles in both reactors over a period of
120 days are shown in Figures 4-1 and 4-2. The membranes were operated for a
period of 3 cycles and 2 cycles in MBRs 1 and 2, respectively, after each physical and
chemical cleaning process. Operations of the MBR were generally smooth except
between day 40 to 60 for MBR 2 whereby the air compressor was not functioning
properly and hence resulted in intermittent air scouring and air supply.
0
20
40
60
80
100
0 20 40 60 80 100 120Time (day)
TMP
(kPa
)
PETE 0.1 µm PCTE 0.1 µmPTFE 0.1 µm
Figure 41: Fouling behavior for membranes in MBR 1
66
Noticeable fouling was observed to occur almost immediately upon start-up for the
PETE membrane for the first cycle in MBR 1. Thereafter the initial fouling following
physical and chemical cleaning process was not as significant as that during cycle 1.
Gradual increase of TMP was observed instead after the first and second cleaning. In
contrast, the PCTE and PTFE membranes exhibited a gradual increase in TMP
followed by a steeper increase after the first 10 d during cycle 1. Fouling for the
PCTE was also observed to be less severe compared to the PETE and PTFE
membranes.
0
20
40
60
80
100
0 20 40 60 80 100 120Time (day)
TMP
(kPa
)
PVDF 0.03 µm PVDF 0.09 µmPVDF 0.2 µm PVDF 0.35 µm
Figure 42: Fouling behavior for membranes in MBR 2
Conditional fouling was observed at beginning of the run for all membranes in
MBR 2 with the 0.35 µm PVDF membrane having a slightly larger increase in TMP,
67
following which, there was a lag of about 15 d before TMP increased gradually at an
increasing rate until day 40 when TMP increased drastically. The drastic increase in
TMP could not be directly attributed to a fault in the air compressor which occurred
between day 40 and 60 as a similar fouling trend was observed in the second run with
no equipment fault.
In all cases, fouling trend was similar, except that the 0.35 μm PVDF membrane had a
higher TMP at all times.
After each run, it was observed that all the membranes were coated with a thick layer
of biofilm and membrane surfaces were not observable by visual inspection.
After subjecting the membranes to chemical cleaning using 3% hypochlorite acid, it
was observed that in subsequent runs, the membranes had a lower conditional fouling,
indicating increased permeability. The starting TMP measured was also slightly lower
for subsequent runs in both MBR 1 and 2 as compared to run 1. This could be due to
the fact that membrane integrity had been affected by chemical cleaning, as had been
indicated in many studies, especially in the cases of track-etched membranes, due to
their manufacturing method. (Lawrence et al., 2006); (Al-Amoudi and Lovitt, 2007).
Another study also showed that chemical cleaning can increase the permeability of
new membranes (Weis et al., 2003). It is possible that chemical reaction between the
chemical agents and foulants occurred by changing the morphology of the foulant or
altering the surface chemistry of fouling layer to remove the foulants from the
membrane surface.
Since the membranes were placed in the same reactors with similar mixed liquor
effects and operational conditions, it is possible to postulate that differences in the
effects of fouling could be due to membrane material characteristics such as porosity,
68
membrane hydrophobicity and surface roughness in the case of MBR 1 and
membrane pore size based on the results obtained from MBR 2.
4
.2 Reactor Performance
General water quality parameters, such as TOC, COD, NH3-N and turbidity in the
feed wastewater, supernatant and permeates were measured throughout the entire
study. Each MF membrane was able to maintain at least 70% removal efficiency for
Total COD (TCOD) throughout the operation. Operations in MBR 2 were hampered
by a faulty air compressor between day 40 and 60 and removal performance of the
membranes was affected even though there was no direct correlation with the fouling
performance of the membranes.
4.2.1 COD and TOC Removal Efficiency
Organics removal was observed to be similar for both COD and TOC removal
efficiency as shown in Figures 4-3 to 4-8. TCOD removal efficiencies were generally
good for all membranes with an average of more than 85% removal efficiency.
Comparing soluble TOC (STOC) values from Figures 4-6 and 4-8, it is observed that
soluble TOC (STOC) removal efficiency was approximately between 60 to 70% for
each membrane, indicating that there was no significant difference in the
performance.
However, as noted in Figures 4-4, 4-5, 4-7 and 4-8, there were unstable performance
in MBR 2 due to operational problems during the period from day 40 to 60 and
69
removal performance subsequently recovered till about day 70. Due to the rainy
weather during certain periods especially between Day 30 to 70 of the operations for
MBR 1, there were large variations in the influent characteristics and this had an
effect on the performance of the reactors especially on MBR 1 operated during the
rainy season.
A general trend was observed for all the membranes. The removal efficiencies rise
towards the end of the cycle for the SCOD and STOC in both MBRs, indicating the
possible presence of a biofilm buildup or biocake which promoted the removal of the
soluble portions. After each cleaning cycle, it was noted that TCOD and SCOD
removal efficiencies dropped to about 80% and 50%, respectively, in both MBRs.
For TCOD results in MBR 1, no clear conclusion could be drawn as run 1 and run 3
of MBR 1 showed improving rejection of carboneous materials while run 2 showed a
reverse trend. This might suggest that the TCOD and TOC rejection in track-etched
membrane in MBRs were influenced by the wastewater influent characteristic. Due to
the inclement weather, particulate materials might be lesser at certain periods of time
and this might have contributed to the different treatment efficiencies over different
cycles as well.
Since MBR 2 was operated during the dryer periods of the year, removal efficiency
was not expected to vary much. However, it was observed that during day 40 to 60,
the removal performance was badly affected, with some membranes having good
removal performance while others having low removals. Removal performance
indicated by Figures 4-4 and 4-7 shows that for TCOD and TOC removal, there was a
decrease in the MBR performance as well as large fluctuations. A reasonable
assumption would be the intermittent aeration resulting in less carboneous treatment
70
at certain periods. In the case of the soluble removal, this was an indication that some
membranes could have thicker biofilm formed on the membrane surface, thus
achieving more removal than usual. The reverse was true for other membranes due to
the intermittent aeration which could result in certain membranes having different
cross-flow velocities and hence different membrane scouring effect.
70
75
80
85
90
95
100
0 20 40 60 80 100 120
Time (d)
Tot
al C
OD
Rem
oval
Eff
icie
ncy
(%)
PETE 0.1 µmPCTE 0.1 µmPTFE 0.1 µm
Run 1 Run 2 Run3
Rainy weather
Figure 43: TCOD Removal Efficiency of all membranes for MBR 1
70
75
80
85
90
95
100
0 20 40 60 80 100 120
Time (d)
Tot
al C
OD
Rem
oval
Eff
icie
ncy
(%)
PVDF 0.03 µmPVDF 0.09 µmPVDF 0.2 µmPVDF 0.35 µm
Run 1 Run 2
Figure 44: TCOD Removal Efficiency of all membranes for MBR 2
71
70
75
80
85
90
95
100
0 20 40 60 80 100 120
Time (d)
Solu
ble
CO
D R
emov
al E
ffic
ienc
y (%
)
PVDF 0.03 µmPVDF 0.09 µmPVDF 0.2 µmPVDF 0.35 µm
Run 1 Run 2
Figure 45: SCOD Removal Efficiency of all membranes for MBR 2
40
50
60
70
80
90
100
0 20 40 60 80 100 120
Time (d)
Solu
ble
TO
C R
emov
al E
ffici
ency
(%)
PETE 0.1 μmPCTE 0.1 μmPTFE 0.1 μm
Run 1 Run 2 Run3
Figure 46: STOC Removal Efficiency of all membranes for MBR 1
72
50
55
60
65
70
75
80
85
90
95
100
0 20 40 60 80 100 120
Time (d)
Tot
al T
OC
Rem
oval
Eff
icie
ncy
(%)
PVDF 0.03 µm
PVDF 0.09 µm
PVDF 0.2 µm
PVDF 0.35 µm
Run 1 Run 2
Figure 47: Total TOC Removal Efficiency of all membranes for MBR 2
20
30
40
50
60
70
80
90
100
0 20 40 60 80 100 120
Time (d)
Solu
ble
TO
C R
emov
al E
ffic
ienc
y (%
)
PVDF 0.03 µm
PVDF 0.09 µm
PVDF 0.2 µm
PVDF 0.35 µm
Run 1 Run 2
Figure 48: STOC Removal Efficiency of all membranes for MBR 2
73
3456789
1011
Waters
Conc
entr
atio
n (m
g.L
-1) Supernatant (MBR 1)
PETE 0.1 µm (MBR 1)PCTE 0.1 µm (MBR 1)PTFE 0.1 µm (MBR 1)Supernatant (MBR 2)PVDF 0.03 µm (MBR 2)PVDF 0.09 µm (MBR 2)PVDF 0.2 µm (MBR 2)PVDF 0.35 µm (MBR 2)
Figure 49: Comparison of permeate and supernatant TOC
A comparison of permeate and supernatant TOC was illustrated in Figure 4-9. For
MBR 1, it was observed that supernatant TOC was higher than permeate TOC.
However, in MBR 2, it was observed that supernatant TOC was lower than the
permeate TOC. The higher TOC in supernatant of MBR 1 compared to the
supernatant of MBR 2 could be attributed to the fact that MBR 2 was operated at a
longer SRT and had a higher MLSS concentration; hence resulting in better removal
of carbonaceous materials. Generally, it is observed based on the mean values that
performance in MBR 2 was better than MBR 1.
The permeate TOC in MBR 2 was observed to have a much larger variation as
compared to MBR 1. This indicated a possibility that the cake layer formation
changes substantially (indicating the effect of cake layer for TOC removal) for the
membranes in MBR 2 as compared to MBR 1 and this could either be an effect of the
different sludge ages of 15 d in MBR 1 and 30 d in MBR 2. Alternatively, such a
phenomenon could also be due to a difference in membrane materials, leading to
adhesion differences of materials onto the different membrane surfaces and hence the
difference in permeate TOC in both MBRs.
74
4.2.2 Turbidity Removal Efficiency
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Samples
NTU
Supernatant (MBR 1)PETE 0.1 µm (MBR 1)PCTE 0.1 µm (MBR 1)PTFE 0.1 µm (MBR 1)Supernatant (MBR 2)PVDF 0.03 µm (MBR 2)PVDF 0.09 µm (MBR 2)PVDF 0.2 µm (MBR 2)PVDF 0.35 µm (MBR 2)Tap Water
Figure 410: Comparison of supernatant and permeate turbidity
Turbidity values shown in Figure 4-10 indicate that permeate quality was better in
MBR 1 than in MBR 2. There was also a larger standard deviation in the values for
the MF membranes in MBR 2 with average values slightly lower than the supernatant
values. This further underscores the point that the pores in all the membranes in MBR
2 could have suffered from some clogging of small particulate matters. It is possible
that during the air compressor failure, there was little or no aeration, causing the
formation of biofilm. Thereafter, biofilm formation could have intensified and
thickened, causing the particulates to be sucked into the pores and subsequently into
the membrane internal structure, thus causing this different between the values from
MBR 1 and MBR 2.
Based on the experience in MBR 1, an excellent removal with quality comparable to
tap water is achievable with membranes of pore size of 0.1 µm. Water quality in terms
of turbidity should not pose a problem in future operations using track-etched
membranes.
75
4.2.3 TN Removal Efficiency
There was no anoxic tank in the MBR design in this study to facilitate denitrification
process. Hence, the only possible route for TN removal is through assimilation by
bacteria. Comparison was carried out between MBR 1 and MBR 2 which had a SRT
of 15 d and 30 d, respectively.
Table 4.1: Total Nitrogen Removal
Total Nitrogen Removal (%)
MBR 1 (SRT = 15d) 2 (SRT = 30d)
Membrane pore size (µm)
PETE 0.1
PCTE 0.1
PTFE 0.1
PVDF 0.03
PVDF 0.09
PVDF 0.2
PVDF 0.35
Mean n=30 10.1 ± 4.3
11.0 ± 5.2
10.6 ± 4.5
13.0 ± 12.2
18.0 ± 13.1
19.0 ± 11.1
17.9 ± 12.9
It was observed that MBR 1 achieved a total nitrogen removal of about 10%. This
could be attributed to bacteria assimilation in the reactor. However, in MBR 2, there
was a much higher overall total nitrogen removal which was as high as 30% between
days 40 to 70. Comparison with (Wang et al., 2005) showed that a total nitrogen
removal in an aerobic section of an MBR of up to 40% at SRT of 15 d may be
attainable under certain conditions such as carbon and nitrogen loading rates, high
mixed liquor concentration and existence of large flocs which allows for anoxic
conditions inside the floc conditions under high mixed liquor concentration.
The higher nitrogen removal observed in MBR 2 could be due to the breakdown in
the air compressor which resulted in dead zones due to inadequate mixing of the
mixed liquor and intermittent air supply over a period of about 20 d. This is confirmed
by DO concentration in MBR 2 between days 40-60 which dropped below 2 mg/l at
76
certain days. (Figure 4-11). Therefore, the existence of microflocs and hence anoxic
zones within the reactor could have led to a slight increase in denitrification in MBR 2
during this period. Statically, this could be observed by a larger standard deviation in
the values for MBR 2 as compared to MBR 1.
0
2
4
6
8
10
0 20 40 60 80 100 120 140
Time (D)
DO
And
SO
UR
DO (mg/L)
SOUR (MBR 2) (mg O2 /g VSS .hr)
Figure 411: DO and SOUR in MBR 2
Further evidence that the high total nitrogen removal in MBR 2 could be attributed to
the presence of dead zones shown by the presence of filamentous bacteria using gram
and neisser staining techniques (Figures 4-12 and 4-13) during Day 50 of the
operation. The presence of considerable amounts of filamentous microorganisms with
small flocs could be an indication of oxygen limitations within the flocs. This could
indicate anoxic conditions developed in the reactors during the periods of intermittent
and unstable aeration.
77
Figure 412: Gram staining done on day 50 for MBR 2
78
Figure 413: Neisser staining done on biomass from MBR 2 on day 50
4.2.4 Nitrification performance
Nitrification conversion to nitrate was nearly 100% for all membrane permeates as
shown in Table 4.2. The results in MBR 1 and 2 clearly demonstrated that operating
at SRTs of 15 d and 30 d respectively did not affect the nitrification performance.
Additionally, there was unlikely to be a significant effect of the biofilm on the
membrane with regards to nitrification as compared to the MLSS.
79
Table 4.2: Nitrification Performance
Ammonia-Nitrogen conversion to Nitrate (%)
Membrane pore size (µm)
PETE 0.1
PCTE 0.1
PTFE 0.1
PVDF 0.03
PVDF 0.09
PVDF 0.2
PVDF 0.35
Mean n=30 99.7 ± 0.1
99.5 ± 0.2
99.6 ± 0.5
99.8 ± 0.3
99.9 ± 0.3
99.9 ± 0.1
99.9 ± 0.1
4
.3 Effects of Membrane Characteristics on Fouling
4.3.1 Effects of Membrane Surface Charge on Fouling
Zeta potential measurement for each membrane was undertaken to evaluate surface
charge at different pH regions as shown in Figure 4-14.
-40
-30
-20
-10
0
10
0 2 4 6 8 10
pH
Zet
a po
tent
ial (
mV
)
PETE 0.1 μm PCTE 0.1 μmPTFE 0.1 μm
Figure 414: Zeta Potential of Membranes used in MBR 1
80
These results showed that the PETE membrane had an isoelectric point at near pH 3,
while the isoelectric points of the PCTE and PTFE membranes did not fall within pH
2–10. This indicates that the PETE membrane is more pH-dependent and negatively
charged at a neutral pH region compared to the PCTE and PTFE. A more negatively-
charged membrane can associate easily with positive divalent ions, such as calcium.
The divalent ions form bridges between the negatively-charged hydrophilic part of
humic acid and the membrane, fostering a faster build-up of cake/gel layer on the
membrane surface.
In the case of the PVDF membranes as shown in Figure 4-15, the isoelectric points
were almost similar at around pH 3 and hence differences in performance of the
rejection of particulate materials could only be due to the difference in the pore sizes.
Difference in pore sizes indirectly impacts membrane fouling. The streaming zeta
potential results are not in agreement with Kim et al. (1997), which showed that the
zeta potential of track-etched membranes with smaller pore sizes of less than 0.05 µm
became constant at about -6 mV at pH more than 5 while those of larger pore sizes
showed constant values at pH > 6. It was also shown that isoelectric points increased
with rising pore size (Kim et al., 1997). However, this was not the case in this study.
This could be a case of manufacturing differences since the amount of charged
functional groups at the openings and within pores are affected by the amount of
etching time in track-etched membranes. In this study, it was observed that both the
PVDF and PETE membranes had similarities in their iso-electric points and with
reference to Figures 4-1 and 4-2, both membrane types had a rapid initial fouling
stage in their operations, further enforcing the observation that certain membranes
might indeed foster a faster buildup of cake layer as compared to other membranes.
81
-40
-30
-20
-10
0
10
0 2 4 6 8 10
pH
Zet
a po
tent
ial (
mV
)
PVDF 0.03 µm PVDF 0.09 µmPVDF 0.2 µm PVDF 0. 35 µm
Figure 415: Zeta Potential of Membranes used in MBR 2
However, analysis of the fouled layer on different pore sized membranes by Kim et
al. (1997) showed that percentage decrease in the zeta potential is higher for the
membrane of the largest pore size of 0.1 µm.
While analysis of fouled membrane was not carried out, results from the analysis of
the zeta potential of permeates from the membranes can be used in lieu of the fouled
membranes results.
82
-16
-12
-8
-4
0Water sample
Zet
a po
tent
ial (
mV
)
Supernatant (MBR 1)
PETE 0.1 μm (MBR 1)
PCTE 0.1 μm (MBR 1)
PTFE 0.1 μm (MBR 1)
Supernatant (MBR 2)
PVDF 0.03 μm (MBR 2)
PVDF 0.09 μm (MBR 2)
PVDF 0.2 μm (MBR 2)
PVDF 0.35 μm (MBR 2)
Figure 416: Zeta potential of supernatants and permeates
The zeta potential analyzer used can measure the surface charge of colloids with
diameters of 0.005 µm to 30 µm and thus, the zeta potential was able to evaluate the
extent of colloidal particles rejected by the membrane and its foulants layer.
To evaluate and quantify particulate rejection by membrane filtration during the
operation, zeta potentials of the supernatant and permeates were measured: -11.0 ± 3.6
for supernatant in MBR 1, -1.8 ± 1.6 for PETE membrane permeate, -2.8 ± 1.2 for
PCTE membrane permeate, -2.1 ± 1.9 for PTFE membrane permeate, -10.5 ± 0.7 for
supernatant in MBR 2, -4.8 ± 1.1 for 0.03 μm PVDF membrane permeate, -6.93 ±
0.8 for 0.09 μm PVDF membrane permeate, -7.6 ± 0.8 for 0.2 μm PVDF membrane
permeate and -10.6 ± 1.6 for 0.35 μm PVDF membrane permeate (Figure 4-16).
Different zeta potentials of the membrane permeates were likely due to the membrane
surface charge aforementioned. From the comparison of supernatant values, zeta-
83
potential was not a function of its MLSS concentration. It is probably due to the
charge of the layer deposited on the membrane. As discussed earlier, zeta potential of
permeate from the membrane of larger pore size was shown to decrease more (from
~-10 mV to ~-2 mV). This implies that repulsion of particulates and charged solutes
would be weaker than other membranes. This would explain the positive relationship
between the pore size and the permeate zeta potential, suggesting that there is less
repulsion from the larger pore sized membranes and hence the cake layer on the larger
pore sized membrane was less negatively charged.
In addition, the faster build-up and compaction of cake/gel layer resulting from the
bridging effect was likely to lead to less passage of colloid components through
membrane as was seen in the differences in the zeta potential for different permeates
for pore sizes. Hence, membrane surface charge was a factor influencing initial MBR
fouling. However, for long term operations, the surface charge is altered by the
biofilm or the fouled compounds on the membrane surface and that would be the layer
influencing the fouling after the initial fouling. Secondly, membrane charges might
also impact long-term operations if the amount of initial fouling is substantial and
hence membranes have to be operated at a higher flux to achieve the same throughput.
This could affect the operational costs of the systems.
84
4.3.2 Effects of Membrane Hydrophobicity on Fouling
40
44
48
52
56
60
64
68
72
76
Membrane Type
Mem
bran
e H
ydro
phob
icity
(Deg
rees
)
PETE 0.1 μm (MBR 1)
PCTE 0.1 μm (MBR 1)
PTFE 0.1 μm (MBR 1)
PVDF 0.03 μm (MBR 2)
PVDF 0.09 μm (MBR 2)
PVDF 0.2 μm (MBR 2)
PVDF 0.35 μm (MBR 2)
Figure 417: Comparison of membrane hydrophobicities
From the results of membrane hydrophobicity and sludge hydrophobicity, it is
expected that the PVDF, PETE and PCTE, being more hydrophobic membranes, were
more favorable for hydrophobic interaction compared to the PTFE membrane (Figure
4-17). However, the PTFE showed a faster increase in filtration resistance than the
PCTE membrane. Chang and Lee (1998) have demonstrated that permeate fluxes of
more hydrophobic membranes declined more significantly during UF filtration of
activated sludge. In addition, more hydrophobic sludges led to more significant
increase in cake resistance due to the hydrophobic and waxy nature of the foaming
sludge surface. (Meng et al., 2006) showed that higher EPS concentration and relative
hydrophobicity of bulking sludge could boost membrane fouling during batch
filtration tests of MBR sludge. However, findings from these studies were obtained
from the result of batch-filtration tests and not long-term operation. (Jang et al., 2007)
also studied the effect of hydrophobicities of sludge and EPS on MBR fouling.
85
However, it did not clearly reveal the contribution of sludge hydrophobicity on
membrane fouling. Consequently, regardless of the importance of sludge
hydrophobicity and membrane hydrophobicity on the early stage of the activated
sludge filtration, the characteristics are likely to be a minor influencing factor in the
membrane fouling during a long-term operation since the biofilm layer would be the
main interaction with the sludge rather than between the membrane and sludge layer.
4.3.3 Effects of Surface Roughness, Porosity & Pore Structure on
Fouling
The smoother track-etched membranes, have pore openings with straight-through
(non-interconnected) microstructure, whereas the rough PTFE membrane has a highly
interconnected pore structure. Their roughness values measured using AFM is
tabulated in Table 4.3.
Table 4.3: Mean roughness values of membranes
Mean roughness values of membranes (nm)
Membrane pore size (µm)
PETE 0.1
PCTE 0.1
PTFE 0.1
PVDF 0.03
PVDF 0.09
PVDF 0.2
PVDF 0.35
Mean 8.3 ± 1.5
11.6 ± 0.1
131.5 ± 31.9
10.18 ± 0.31
32.85 ± 0.49
27.7 ± 0.57
74.95 ± 4.6
There is little information in relation to the effect of membrane roughness on MBR
fouling. Some studies using NF and reverse osmosis membranes have revealed that
rough membrane surfaces are subjected to fouling occurrence than smooth surface
(Elimelech et al., 1997); (Vrijenhoek et al., 2001). However, the results of filtration
resistance in this work were not in agreement with the previous researches, suggesting
86
that surface roughness did not play an important role in membrane fouling in MBR
under long term operation. A possible reason could be the difference in application of
the NF membranes as well as the membrane configurations which were used. In the
flat-sheet configuration, the membrane was subjected to strong and constant scouring
on the membrane surface. The shear forces introduced by the air bubbles could be
more dominant in minimizing fouling than the smooth surface of the track-etched
membrane in the reduction of fouling.
Pore microstructure might also have an impact on membrane fouling in the MBR
process. As described previously, the PETE and PCTE membranes, which were
manufactured by the tract-etched process, had a dense structure with uniform
cylindrical pores.
In contrast, the PTFE membrane, which was made by physical stretching of the
polymer film followed by annealing at elevated temperature, had a structure with
polymer nodules connected by thin fibers. (Zydney and Ho, 2003) found that for a MF
membrane with straight-through pores, pore blockage had led to a rapid permeate flux
decline during protein filtration since pore clogging reduces fluid flow through the
blocked pores.
Results from Ho and Zydneys' studies (1999, 2000) showed that membrane fouling
was an inverse function of porosity for the track-etched membrane with low porosity.
However, membrane fouling is not influenced by porosity for the high porosity
Anopore membrane, which has a non-deformable honeycomb pore structure with no
lateral crossovers between individual pores. For a membrane with a highly
interconnected pore structure, the rate of flux decline was much smaller than that for a
87
track-etched membrane with straight-through microstructure, even for a membrane
with similar porosity, resulting from fluid flow under and around the blockage.
Due to the lack of information on the porosity of the given membranes, pure water
permeability (PWP) test was conducted on the membranes instead. PETE membrane
was found to have a lower PWP value, indicating that the membrane had the lowest
porosity among all the membranes.
Table 4.4: Pure water permeability (PWP) of membranes
PWP values (L m-2 h-1 kPa-1)
Membrane pore size (µm)
PETE 0.1
PCTE 0.1
PTFE 0.1
PVDF 0.03
PVDF 0.09
PVDF 0.2
PVDF 0.35
Mean 20.5 ± 2.9
26.7 ± 3.0
29.5 ± 7.5
47.1 ± 2.6
43.2 ± 3.1
41.5 ± 2.8
23.6 ± 4.5
In this study, the fouling rate tendency of the PETE and PCTE membranes was in
good agreement with the results from Ho and Zydneys’ studies (1999, 2000) because
the membranes did not have a high porosity like the Anopore membranes.
In addition, the differences between the PCTE and PTFE membranes were likely due
to the difference of target materials (the target water used in the previous studies
contained a sole protein component, not a mixture with solutes and particles present in
domestic wastewater). The accmulation of foulants on the PTFE membrane surface
was likely to be much more than that of the PCTE surface because the crevices on the
PTFE surface was observed to be more and larger compared to the PCTE surface in
the 3-D images by an AFM instrument as shown in Figure 4-18.
88
Figure 418: Surface image of each membrane from AFM analysis
(a) PETE, (b) PCTE and (c) PTFE membranes. The figures show AFM 3D images of membranes over an area of 5 µm × 5 µm.
Data from the PVDF membranes suggests that the PVDF membrane of larger pore
size had smaller porosity and hence potentially foul faster. However, there were
insignificant differences in the fouling of the different pore sized membranes shown
in Figures 4-1 and 4-2, indicating that membrane surface roughness might not play a
significant role in fouling.
4.3.4 Effects of Pore Size & Particle Size Distribution on Fouling
Analysis was done on the particle size distribution of the feed solution and results
were illustrated in Figure 4-19.
89
0
5
10
15
20
25
30
5 1 0.65 0.45 0.1
Membrane Filter Pore Size(μm)
% o
f par
ticle
s filt
ered
and
ret
aine
d b
y ea
ch m
embr
ane
Figure 419: Influent particle size distribution
Approximately 50% of the particles were larger than 0.1 μm as illustrated in Figure 4-
19. Hence, fouling is expected to have a more significant effect on the membranes
with larger pore sizes, since for the membranes with smaller pore sizes, there is less
pore blocking.
From the SEM pictures shown in Figure 4-20, it was observed that the pores were not
as uniform as expected, with some pores having merged and creating larger pores due
to the manufacturing process. This could also be a reason as to why the TCOD
removal performances of the PVDF membranes in MBR 2 were rather similar as the
particle cut off.
90
Figure 420: Surface image of each membrane from SEM analysis
From Top Left: (a) PETE, (b) PCTE, (c) PTFE, (d) Fouled PETE, (e) Fouled PTFE,
(f) 0.03 µm PVDF, (g) 0.35 µm PVDF, (h) Fouled 0.03 µm PVDF, (i) Fouled 0.35
µm PVDF membranes
As illustrated from the SEM pictures for the 0.03 μm and 0.35 μm membranes in
Figures 4-18f and 4-18g, respectively, the pore aspect ratio (measured as the length
along the major axis over the length along the minor axis) was not seen to play a
significant role in membrane fouling. In comparison with the SEM picture of the
PTFE membrane (Figure 4-20c), it is seen that the pore aspect ratio for the PTFE was
much larger than that of the PCTE and PETE track-etched membranes due to its
difference in manufacturing and its inherent structure. However, there is no evidence
to suggest that pore aspect ratio played a significant role in fouling in this study as
fouling performance of PTFE and PCTE membranes in MBR 1 were quite similar.
91
A separate study conducted to compare UF and MF showed that in a similar
environment with a crossflow velocity of 0.1 m/s, the MF membrane produced a
hydraulic resistance twice of that for the UF membrane. However, DOC rejection of
both membranes used was similar after 2 h of operation, indicating the formation of a
dynamic cake layer on the MF membrane (Choi et al. 2006).
In this study, there were no significant effects of pore size on membrane fouling.
Permeability was observed to decrease faster for the 0.35 μm membrane while the rest
of the membranes experienced the same fouling trend. However, this was probably
due to the fact that the 0.35 μm membrane had a much high membrane resistance and
nearly half as low porosity as compared to the rest of the membranes. Based on the
pictures taken of the pore sizes, it is seen that the membrane of smaller pore size had a
much thicker biofilm layer deposition (Figure 4-18h) than the 0.35 µm PVDF
membrane (Figure 4-18i). There were few studies conducted with constant flux
operations, additionally most studies were only for short duration which showed that
larger membrane pore sizes tend to foul faster. However, results gathered here agree
with Gander et al. (2000) in which there were higher initial fouling for membranes
having large pores. However, it was not observed that significant flux decline
occurred when small pore membranes were used over an extended time as opposed to
the previously mentioned study.
4
.4 Effects of Sludge Characteristics on Fouling
To elucidate the fouling mechanisms in the MBR process, characteristics of mixed
liquor, such as sludge concentration and surface charge, were monitored throughout
the operation. The sludge volume index (SVI) values measured in MBR 1 were
92
between 34 and 83 mL g-1, suggesting a good settling sludge. Therefore, it can be
considered that MBR 1 was not affected by the fouling factor associated with
excessive filamentous bacteria.
4.4.1 Effects of MLSS on Membrane Fouling
MLSS concentration was monitored over the course of operation of the MBRs and is
shown in Figures 4-21 and 4-22. MLSS concentration in MBR 1 was fluctuating and
low due to the wet weather. In contrast, MBR 2 was operated during the dryer periods
of the year and hence MLSS concentration was higher.
Figure 421: MLSS and MLVSS concentration of MBR 1
93
4
6
8
10
12
14
16
0 20 40 60 80 100 120Time (d)
Con
cent
ratio
n (g
l -1
)MLSS MLVSS
Figure 422: MLSS and MLVSS concentration of MBR 2
Many researches had reported the effect of MLSS concentration on MBR fouling but
there were still controversial reports. Le-Clech et al. (2003) reported that no
significant impact of MLSS was observed for a shift from 4 to 8 g L-1, but more
fouling arose with MLSS increase to 12 g L-1. Rosenberger et al. (2005) also
described that the general trend of MLSS increase on membrane fouling in the MBR
process for municipal wastewater treatment seemed to result in less fouling below 6 g
MLSS L-1, no significant impact between 8 and 12 g MLSS L-1, and more fouling
above 15 g MLSS L-1.
As stated earlier, there exist many conflicting studies on the effects of MLSS on
membrane fouling. In this study, the MLSS concentration of MBRs 1 and 2 ranged
94
from 1.6 g L-1 to 8.0 g L-1 (average value: 4.7 g L-1) and 9.4 g L-1 to 13.0 g L-1
(average value: 10.6 g L-1), respectively. However, there was no similar relationship
with fouling as reported by Le-Clech et al (2003).
Even though MLSS concentration in MBR 1 showed a gradual rise in values,
membrane permeability and fouling were not significantly affected. In MBR 2, the
MLSS was kept at a relatively constant MLSS concentration with no adverse fouling
effects as well. Thus, membrane permeability was not likely to be significantly
affected by the sludge concentration throughout the operation.
4.4.2 Effects of Sludge Surface Charge on Fouling
Surface charge of mixed liquor of both MBRs were measured by the titration method,
varying between -0.7 and -0.3 meq g-1 MLVSS. (Liu and Fang, 2003) showed that the
surface charge of conventional activated sludge ranged from -0.6 to -0.2 meq g-1
MLVSS. In Lee et al.'s studies, (2003), MBR sludges at different SRTs (20 d, 40 d
and 60 d) were between -0.7 and -0.4 meq g-1 MLVSS. The flocs surface charges and
bound EPS were attributed to the negative charges from ionization of anionic
functional groups.
This study confirmed that sludge surface charge was generally within a narrow band
and hence would not have any significance on the impact of this study.
95
4.4.3 Effects of Soluble and Bound EPS on Membrane Fouling
EPS is considered to be an important foulant and is suggested as a probable index for
membrane fouling (Chang and Lee, 1998). EPS in activated sludge can be divided
into two categories: (1) bound EPS extracted from microbial floc, consisting of
insoluble materials (sheaths, capsular polymers, condensed gel, loosely bound
polymers, and attached organic material), and (2) soluble EPS including soluble
macromolecules, colloids and slimes, considered as soluble microbial products
(Laspidou and Rittmann, 2002); Jang et al., 2007). Protein and carbohydrate in the
EPS are considered as the dominant components. Protein generally has a hydrophobic
tendency, whereas carbohydrate is more hydrophilic (Liu and Fang, 2003).
Table 4.5: Concentrations of EPS in MBRs 1 and 2
Item TOC (mg/g MLVSS)
Carbohydrate (mg glucose/g MLVSS)
Protein (mg BSA/g MLVSS)
Protein/ carbohydrate
MBR 1 (SRT=15d) 36.2 ± 14.3 15.9 ± 2.0 31.4 ± 3.4 2.0 ± 0.3
MBR 2 (SRT=30d) - 7.7 ± 1.0 18.2 ± 2.9 2.33 ± 0.3
The concentration evolutions of bound EPS components are shown in Table 4.5. It
can be seen that, for an MBR operated at a constant SRT, the concentrations of bound
EPS components varied throughout the operation. In MBR 1, Carbohydrate tended to
decrease slightly with operation time, whereas TOC, protein and protein/carbohydrate
ratio appeared to increase gradually (Figure 4-23). The finding by Lee et al. (2003)
showed that the filtration resistance caused by microbial flocs was strongly positively
correlated to the protein/carbohydrate ratio of bound EPS. Therefore, the gradual
increment of protein/carbohydrate ratio in MBR 1 shown in Figure 4-23 was likely to
96
be a factor that caused the PCTE and PTFE membranes to increasingly rapidly foul
after membrane cleaning. This could be due to excess unutilized substrate being
converted into both intracellular storage granules and EPS, thus resulting in higher
EPS per unit of biomass. Another possibility could be the changing interactions
between the increasing hydrophobic sludge due to the increasing proteins
concentration in the EPS and the slightly more hydrophobic PCTE and PETE
membrane surfaces.
0
1
2
3
4
0 20 40 60 80 100 120
Time (d)
Prot
ein
/ car
bohy
drat
e
Figure 423: Protein/carbohydrate ratio for MBR 1
97
0
1
2
3
4
0 20 40 60 80 100 120
Time (d)
Prot
ein
/ car
bohy
drat
e
Figure 424: Protein/carbohydrate ratio for MBR 2
In MBR 2, however, protein/carbohydrate ratio appeared to be following a decreasing
trend (Figure 4-24). In the case of MBR 2, biomass at a longer SRT would have
underwent endogenous respiration and cell lysis. This would have led to the release of
EPS into the bulk solution and further hydrolyzed into SMP. These substrates could
then provide added sustenance to the biomass population for maintenance processes,
resulting in a decrease in both EPS and SMP content under long SRTs.
From earlier results, it is known that the membranes used in this study are mainly
hydrophobic with the PTFE being most hydrophilic. Therefore fouling is expected to
be significantly affected if there are more hydrophilic organic factions in the MBR.
This is observed from the protein/carbohydrate ratio where there was a decreasing
98
protein/carbohydrate ratio in MBR 2. This implies that fouling would be retarded,
which was shown by Figures 4-1 and 4-2, where the membranes in MBR 2 were
shown to foul slower than the membranes in MBR 1. This is another suggestion that
depending on the amount of hydrophilic or hydrophobic portions in the sludge EPS,
there would be different effects on both hydrophilic and hydrophobic membranes.
Table 4.6: Concentrations of EPS in permeate and supernatant
Item TOC (mg/L)
Carbohydrate (mg/L)
Protein (mg/L)
Protein/ carbohydrate
Supernatant 1 (MBR 1) 9.9 ± 1.1 10.4 ± 1.6 10.0 ± 1.7 0.98 ± 0.24
PETE 0.1 µm (MBR 1) 7.4 ± 0.5 6.8 ± 0.6 7.2 ± 1.3 1.06 ± 0.22
PCTE 0.1 µm (MBR 1) 7.8 ± 0.8 7.0 ± 0.8 7.5 ± 1.7 1.08 ± 0.25
PTFE 0.1 µm (MBR 1) 7.4 ± 0.5 6.8 ± 0.5 7.5 ± 1.4 1.11 ± 0.22
Supernatant 2 (MBR 2) - - - -
PVDF 0.03 µm (MBR 2) - 7.9 ± 1.8 11.1 ± 3.2 1.31 ± 0.34
PVDF 0.09 µm (MBR 2) - 7.7 ± 1.85 10.9 ± 2.8 1.35 ± 0.34
PVDF 0.2 µm (MBR 2) - 9.0 ± 2.75 9.2 ± 2.6 1.03 ± 0.34
PVDF 0.35 µm (MBR 2) - 8.9 ± 2.3 8.1 ± 2.8 0.90 ± 0.34
Soluble EPS concentrations in the supernatant and permeates are summarized in
Table 4.6. The level of protein/carbohydrate ratio of soluble EPS was found to be
much lesser than that of bound EPS. This was likely due to the fact that affinity
between protein and microbial flocs could be higher than that between carbohydrate
99
and flocs due to the hydrophobicities and surface charges of protein and flocs. In
addition, the PCTE membrane had lower rejection rates for protein and TOC of
soluble EPS compared to the PETE and PTFE membranes. This observation also
confirms less cake deposit (layer) associated with the PCTE membrane than the PETE
and PTFE membranes.
Similarly in MBR 2, an observation was made that cake deposit layer was more
prominent for membranes with smaller pore size (shown in Figure 4-18h) as it is
capable of a larger rejection rate for both of the membranes with pore sizes smaller
than 0.1 µm.
4.4.4 Effects of Sludge Hydrophobicity
Hydrophobic microbial flocs lead to better flocculation and generally minimal
interaction with hydrophilic membranes. The sludge relative hydrophobicity ranged
from 1.6% to 57% in this study, indicating slightly lesser hydrophobicity compared to
the normal activated sludge in the previous research (Chang and Lee, 1998). From
this result, it can be considered that the influence of sludge hydrophobicity on
membrane fouling is not likely to be significant in this study.
4.4.5 Fractionation of DOC in the Supernatant and Permeates
Fractionation of DOC was done on samples from MBR 1. DOC component in the
supernatant and permeates was fractionated using DAX-8 and XAD-4 resins,
followed by DOC measurement for each fractionated water sample. The isolation
100
procedure allowed determination of hydrophobic (HPO), transphilic (TPI) and
hydrophilic (HPI) fractions. The HPO fraction comprised of the least polar and
generally highest molecular weight (MW) humic materials of the three fractions. The
TPI fraction is of intermediate polarity and lower MW, and the HPI fraction consists
of the most polar and lowest MW organic matter (Quanrud et al., 2003). This analysis
aimed to examine the hydrophobic and hydrophilic compositions of DOC in the
supernatant and permeate, which could be affected by the interaction (adsorption)
between the organic matter and membrane hydrophobicity. Membrane hydrophobicity
can be considered an important player for membrane fouling in the MBR system.
Solutes, colloids and microbial cells interact preferentially with more hydrophobic
membranes, resulting in more severe membrane fouling. However, it can be seen
from Figure 4-25 that the composition of each fraction between the supernatant and
permeates was not changed significantly throughout the operation, implying that
membrane hydrophobicity might not be a dominant influencing factor on MBR
fouling in this study.
101
Figure 425: Fractionation of DOC in supernatant and permeates of MBR 1
After the termination of the MBR 1 operation, the organic foulants desorbed from the
fouled membranes were separated into the HPO, TPI and HPI fractions as well. The
irreversible foulants were extracted from the fouled membranes using 0.1 M NaOH
after acidic cleaning. The mass percentage in each fraction is illustrated in Figure 4-
26.
102
Figure 426: Mass percentage of fractions desorbed from membrane surface
The dissolved organic foulant desorbed from the PCTE membrane was primarily
hydrophobic fraction, whereas the PETE membrane had much less hydrophobic
fraction than expected regardless of the membrane with the similar contact angle. As a
result, the fractionation analyses exhibited no clear relationship between membrane
hydrophobicity and membrane fouling tendency. He et al., (2005) studied the effect of
membrane hydrophobicity in the MBR system during comparison of ultrafiltration
membranes with different hydrophobic characteristics. In the study, unexpectedly, a
less hydrophobic membrane showed a higher fouling rate than the more hydrophobic
membranes in the experiment. In the rejection experiment of bovine serum albumin,
in addition, the membrane hydrophobicity did not affect the adsorption properties
except for the adsorption interaction in the early stage of the filtration (Nakamura and
Matsumoto, 2006). Therefore, Le-Clech and his coworkers (2006) stressed that
103
membrane hydrophobicity is expected to play a minor role during extended filtration
periods regardless of its significance on the early stage of the fouling formation. Once
the membrane is initially fouled, the properties of the membrane would become
secondary to those of the sludge components covering the membrane surface. Thus,
the faster formation of cake/gel layer on the PETE membrane was likely to more
strongly influence on the amount of dissolved organic materials attached to the
membrane. Aforementioned earlier, a faster membrane fouling causes a large amount
of foulants to get attached to the membrane in a shorter time. This results in the quick
build-up and compaction of fouling layer on the membrane surface irrespective of
hydrophobicity of fouling materials. The amount of adsorbed foulants during the
operation is dependent upon not membrane hydrophobicity but permeability when the
initial permeate flux of a target membrane is determined at a higher permeate flux
than the average design flux.
The discrepancy of HPO and HPI fractions among the membranes could be explained
by the different amounts of biopolymers adsorbed to the membrane. The biopolymers
like EPS can fill the void spaces between the cell particles in the cake layer. The
biopolymers can be readily deposited onto the membrane surfaces by permeation
drag, and not readily detached by shear force due to its low back transport velocity.
Proteins of EPS are generally hydrophobic, while carbohydrate of EPS is more
hydrophilic (Liu and Fang, 2003). EPS analysis of dissolved organic foluants
desorbed from the target membranes was performed to investigate the cause of
different HPO and HPI fractions in the desorbed organic foulants (Table 4.7).
104
Table 4.7: EPS in dissolved organic foulants desorbed from membranes
Item DOC (mg/m2)
Carbohydrate (mg/ m2)
Protein (mg/ m2)
Protein/ carbohydrate
PETE 24.6 8.0 77.0 9.6
PCTE 67.3 13.0 192.9 14.8
PTFE 43.6 9.5 91.1 9.6
The desorbed DOC mass per unit area of the membrane surface was larger for the
PCTE and PTFE membranes than for the PETE membrane, which was not consistent
with the permeability result of each membrane. This was attributable to the fact that
the fractionation measurement of irreversible dissolved foulants was performed after
target membranes were thoroughly fouled. In addition, protein/carbohydrate ratio of
the PCTE membrane was much higher than those of the PETE and PTFE membranes,
suggesting that higher protein content contributed to higher HPO fraction in the
desorbed foulants.
105
5
Conclusion and recommendations
The feasibility study for using track-etched membranes for wastewater treatment was
carried out using 2 MBRs. Experimental analysis was done to compare the
performance between track-etched membranes and asymetric PTFE membrane.
Studies were also done to determine factors which influence membrane fouling. This
study evaluated the impact of membrane properties such as pore structure on
submerged MBR fouling over long-term operation treating actual municipal
wastewater.
5
.1 Conclusion
The following conclusions on membrane characteristics on fouling were drawn:
The PETE membrane with the lowest PWP value exhibited the most rapid increase of
the filtration resistance, whereas the PCTE membrane with the intermediate PWP
value had the lowest fouling rate of the membranes. The PWP values for the PVDF
membranes were twice as high as that for the PCTE and PETE membranes. This was
also verified by the lower TMP for the PVDF membranes compared to the PCTE and
PETE membranes. However, PVDF and PETE membranes showed a rapid rise in
TMP initially as compared to PCTE and PTFE membranes.
This implies that membrane porosity played an important role in the track-etched
membrane filtration.
Although the PTFE membrane with interwoven sponge-like microstructure had the
highest PWP value, it showed more rapid increase of the filtration resistance
compared to the PCTE. This was likely due to more filling of foulants in localities of
the rougher surface of the PTFE membrane.
106
Fouling on membrane pore is very likely to be dependent on the influent particle
sizes. Besides that, for short term operations, smaller pore-sized membranes may not
foul that quickly as compared to larger pore membranes, but the opposite might be
true for long term operations. The results in this study showed that larger pore size
membranes would be more susceptible to initial fouling due to the particles clogging
the pores.
Sludge characteristics, such as SVI, sludge concentration, surface charge and
hydrophobicity, exhibited normal condition in this study comparable to other studies,
suggesting that they were not likely to be a key role in membrane fouling.
Surface charge of the PETE membrane and its foulants layer contributed to a lower
zeta potential of the membrane permeate, indicating a better particulate rejection.
Membrane hydrophobicity was not likely to play a significant role in membrane
fouling during a long-term operation, instead it is likely to play an indirect role by
affecting the initial fouling stage in causing rapid adhesion of materials to the
membranes in certain cases and possibility reduced throughput operationally of the
membranes
Gradual increment of protein/carbohydrate ratio in bound EPS was likely to
contribute to the result that the PCTE and PTFE membranes increasingly rapidly
fouled after membrane cleaning. When operating at a higher SRT,
protein/carbohydrate ratio showed a decreasing trend, this also coincided with a
longer fouling cycle. This suggests that proteins are the key foulant in this study due
to the interaction between the hydrophobic protein fractions and the slightly more
hydrophobic membranes.
107
DOC fractionation in the supernatant and permeates showed that the composition of
each fraction in the waters investigated was not changed significantly throughout the
operation, supporting that membrane hydrophobicity might not be a dominant player
influencing on MBR fouling. The organic foulants desorbed from the PCTE
membrane dominantly contained hydrophobic fraction, which was likely due to the
EPS protein released from biomass attached to the membrane.
The following conclusions on membrane performance were drawn:
i. Membranes performance in terms of carbonaceous removal efficiency fluctuated
with respect to the influent characteristic. However, it was able to achieve at least
a minimum of 75% removal while maintaining an average of 85% removal
efficiency throughout the study.
ii. Nitrification performance was excellent as expected, given the long SRT for the
nitrifiers to grow. Turbidity removal was a little poorer in MBR 2. However
overall, the performance of all the track-etched membranes was acceptable.
5.2 Recommendations
Results of permeability exhibited that the permeate flux profiles of the PCTE and
PTFE membranes were almost similar, whereas the PETE membrane had the most
rapid flux decline rate. The faster permeability decline and lower permeate TOC of
the PETE could be explained by the fact that a higher permeate flux than an average
design flux had led to a fast fouling due to a larger amount of fouling materials to get
attached into and onto the membrane in a shorter time. This suggests that we should
108
not operate a membrane at a higher permeate flux than the average design flux in the
MBR process.
Further research should be carried out to investigate of the fractions of EPS that affect
the membrane fouling by analysing its components and its range in molecular weight
distribution to further understand the impact of membrane fouling using different pore
sized membranes.
Effect of chemical cleaning on track-etched membranes should be further investigated
as this study showed that membrane integrity might be altered after chemical cleaning,
which results in an increase in permeability.
109
6 References
UNESCO), U. N. E. S. A. C. O. (2006) The 2nd UN World Water Development M, W. W. A. (Ed.).
(Report: 'Water, a shared responsibility'. IN PROGRA USGS), United. States. Geological. Services. (2007) (http://ga.water.usgs.gov/edu/watercyclegwstorage.html as of 3rd July 2008. AL‐AMOUDI, A. & LOVITT, R. W. (2007) Fouling strategies and the cleaning ystem of NF membranes and factors affecting cleaning efficiency. Journal of
esMembran Science, 303, 6‐28. PEL, P. (2001) Track etching technique in membrane technology. Radiation
r tAMeasu emen s, 34, 559‐566. PEL, P. Y. (1995) Heavy‐particle tracks in polymers and polymeric track Amembranes. Radiation Measurements, 25, 667‐674. BAE, T. H. & TAK, T. M. (2005) Interpretation of fouling characteristics of ltrafiltration membranes during the filtration of membrane bioreactor mixed
uliquor. Journal of Membrane Science, 264, 151‐160. ARKER, D. J. & STUCKEY, D. C. (1999) A review of soluble microbial products
R , 3063‐3082. B(SMP) in wastewater treatment systems. Water esearch, 33 BHATTACHARYYA, D. & BUTTERFIELD, D. A. (2003) New insights into embrane science and technology : polymeric and biofunctional membranes, m
Amsterdam ; Boston, Elsevier. OLLER, M., GUJER, W. & TSCHUI, M. (1994) Parameters affecting nitrifying
c Bbiofilm rea tors. Water Science and Technology, 29, 1‐11. BROOKES, A., JEFFERSON, B., LE‐CLECH, P. & JUDD, S. (2003) Fouling of embrane bioreactors during treatment of produced water. International m
Membrane Science and Technology Conference (IMSTEC). Sydney, Australia. CABASSUD, C., MASSE, A., ESPINOSA‐BOUCHOT, M. & SPERANDIO, M. (2004) Submerged Membrane Bioreactors: Interactions between membane filtration nd biological activity. Proceedings of Water EnvironmentMembrane Technology aConference. Seoul, Korea. CHANG, I. S., BAG, S. O. & LEE, C. H. (2001a) Effects of membrane fouling on olute rejection during membrane filtration of activated sludge. Process sBiochemistry, 36, 855‐860. CHANG, I. S., CHOO, K. H., LEE, C. H., PEK, U. H., KOH, U. C., KIM, S. W. & KOH, J. H. (1994) Application of ceramic membrane as a pre‐treatment in anaerobic‐digestion of alcohol‐distillary wastes. Journal of Membrane Science, 90, 131‐139.
110
pP
CHANG, I. S., GANDER, M., JEFFERSON, B. & JUDD, S. J. (2001b) Low‐cost embranes for use in a submerged MBR. Process Safety and Environmental m
Protection, 79, 183‐188. CHANG, I. S., LE CLECH, P., JEFFERSON, B. & JUDD, S. (2002) Membrane fouling in embrane bioreactors for wastewater treatment. Journal of Environmental
i 1mEngineer ng Asce, 28, 1018‐1029. CHANG, I. S. & LEE, C. H. (1998) Membrane filtration characteristics in embrane‐coupled activated sludge system ‐ the effect of physiological states of m
activated sludge on membrane fouling. Desalination, 120, 221‐233. CHILDRESS, A. E. & ELIMELECH, M. (1996) Effect of solution chemistry on the urface charge of polymeric reverse osmosis and nanofiltration membranes. sJournal of Membrane Science, 119, 253‐268. CHITTRAKARN, T., BHONGSUWAN, T., WANICHAPICHART, P., NUANUIN, P., CHONGKUM, S., KHONDUANGKAEW, A., BORDEEPONG, S., (2002) Nuclear track‐etched pore membran production using neutrons from the Thai research reactor TRR‐1/M1. Songklanakarin Journal of Science and Technology. CHO, B. D. & FANE, A. G. (2002) Fouling transients in nominally sub‐critical flux operation of a membrane bioreactor. Journal of Membrane Science, 209, 391‐403. CHOI, H., ZHANG, K., DIONYSIOU, D. D., OERTHER, D. B. & SORIAL, G. A. (2005a) ffect of permeate flux and tangential flow on membrane fouling for wastewater Etreatment. Separation and Purification Technology, 45, 68‐78. CHOI, H., ZHANG, K., DIONYSIOU, D. D., OERTHER, D. B. & SORIAL, G. A. (2005b) nfluence of cross‐flow velocity on membrane performance during filtration of Ibiological suspension. Journal of Membrane Science, 248, 189‐199. CHOI, H., ZHANG, K., DIONYSIOU, D. D., OERTHER, D. B. & SORIAL, G. A. (2006) ffect of activated sludge properties and membrane operation conditions on
6Efouling characteristics in membrane bioreactors. Chemosphere, 3, 1699‐1708. OTE, P., SIVERNS, S. & MONTI, S. (2005) Comparison of membrane‐based Csolutions for water reclamaton and desalination. Desalination, 182, 251‐257. DREWS, A., VOCKS, M., IVERSEN, V., LESJEAN, B. & KRAUME, M. (2005) Influence of unsteady membrane bioreactor operation on eps formation and filtration esistance. International Congress on Membranes and Membrane Process (ICOM). rSeoul, Korea. ECKENFELDER, W. W., GRAU, P. & INTERNATIONAL ASSOCIATION ON WATER POLLUTION RESEARCH AND CONTROL . CONFERENCE (1992) Activated sludge rocess design and control : theory and practice, Lancaster, PA, U.S.A., : Technomic ub. Co.
111
ELIMELECH, M., ZHU, X. H., CHILDRESS, A. E. & HONG, S. K. (1997) Role of membrane surface morphology in colloidal fouling of cellulose acetate and omposite aromatic polyamide reverse osmosis membranes. Journal of cMembrane Science, 127, 101‐109. EVENBLIJ, H. & VAN DER GRAAF, J. (2004) Occurrence of EPS in activated sludge rom a membrane bioreactor treating municipal wastewater. Water Science and fTechnology, 50, 293‐300. FAN, X. J., URBAIN, V., QIAN, Y. & MANEM, J. (2000) Ultrafiltration of activated ludge with ceramic membranes in a cross‐flow membrane bioreactor process. sWater Science and Technology, 41, 243‐250. ANG, H. H. P. & SHI, X. L. (2005) Pore fouling of microfiltration membranes by Factivated. Journal of Membrane Science, 264, 161‐166. FERAIN, E. & LEGRAS, R. (2003) Track‐etch templates designed for micro‐ and anofabrication. Nuclear Instruments & Methods in Physics Research Section BnBeam Interactions with Materials and Atoms, 208, 115‐122. FLEMMING, H. C. & WINGENDER, J. (2001) Relevance of microbial extracellular olymeric substances (EPSs) ‐ Part I: Structural and ecological aspects. Water pScience and Technology, 43, 1‐8. GANDER, M. A., JEFFERSON, B. & JUDD, S. J. (2000) Membrane bioreactors for use n small wastewater treatment plants: membrane materials and effluent quality. iWater Science and Technology, 41, 205‐211. GE, L. Y., DENG, H. H., WANG, H. W., MA, L. M. & LIU, Y. (2007) Comparison of xtraction methods for quantifying extracellular polymers in activated sludges. eFresenius Environmental Bulletin, 16, 299‐303. GHAYENI, S. B. S., BEATSON, P. J., SCHNEIDER, R. P. & FANE, A. G. (1998) Water reclamation from municipal wastewater using combined microfiltration reverse smosis (ME‐RO): Preliminary performance data and microbiological aspects of osystem operation. Desalination, 116, 65‐80. GHOSH, R. & WONG, T. (2006) Effect of module design on the efficiency of embrane chromatographic separation processes. Journal of Membrane Science, m
281, 532‐540. GRUNDITZ, C. & DALHAMMAR, G. (2001) Development of nitrification inhibition
n assays usi g pure cultures of Nitrosomonas and Nitrobacter. Water Research, 35, 433‐440. GUNDER, B. & KRAUTH, K. (1998) Replacement of secondary clarification by embrane separation ‐ Results with plate and hollow fibre modules. Water cience and Technology, 38, 383‐393. mS
112
Jt
HAANDEL, A. C. V. & LETTINGA, G. (1994) Anaerobic sewage treatment : a ide apractical gu for regions with hot climate, Chichester ; New York, : J. Wiley.
HAN, S. S., BAE, T. H., JANG, G. G. & TAK, T. M. (2005) Influence of sludge etention time on membrane fouling and bioactivities in membrane bioreactor rsystem. Process Biochemistry, 40, 2393‐2400. HANAKI, K., HONG, Z. & MATSUO, T. (1992) Production of nitrous‐oxide gas uring denitrification of waste‐water. Water Science and Technology, 26, 1027‐d1036. HE, Y. L., XU, P., LI, C. J. & ZHANG, B. (2005) High‐concentration food wastewater reatment by an anaerobic membrane bioreactor. Water Research, 39, 4110‐t4118. ENZE, M. (2001) Wastewater treatment : biological and chemical processes, New H
York, Springer. HILLIS, P., ROYAL SOCIETY OF CHEMISTRY (GREAT BRITAIN), EUROPEAN DESALINATION SOCIETY. & SOCIETY OF CHEMICAL INDUSTRY (GREAT RITAIN) (2000) Membrane Technology in Water and Wastewater Treatment, BCambridge, UK, Royal Society of Chemistry. HO, C. C. & ZYDNEY, A. L. (1999) Theoretical analysis of the effect of membrane orphology on fouling during microfiltration. Separation Science and m
Technology, 34, 2461‐2483. HO, C. C. & ZYDNEY, A. L. (2000) A combined pore blockage and cake filtration odel for protein fouling during microfiltration. Journal of Colloid and Interface m
Science, 232, 389‐399. UANG, X., LIU, R. & QIAN, Y. (2000) Behaviour of soluble microbial products in a H
membrane bioreactor. Process Biochemistry, 36, 401‐406. JANG, N., REN, X., KIM, G., AHN, C., CHO, J. & KIM, I. S. (2007) Characteristics of oluble microbial products and extracellular polymeric substances in the
bsmembrane ioreactor for water reuse. Desalination, 202, 90‐98. JEISON, D. & VAN LIER, J. B. (2008) Anaerobic wastewater treatment and embrane filtration: a one night stand or a sustainable relationship? Water
527‐532. mScience and Technology, 57, JENKINS, D., RICHARD, M. G., DAIGGER, G. T., (2004) Manual on the causes and ontrol of activated sludge, bulking, foaming, and other solids separation problems,
a o d cBoca R ton, Fla., L n on : Lewis. UDD, S. (2004) A review of fouling of membrane bioreactors in sewage reatment. Water Science and Technology, 49, 229‐235.
113
JUDD, S. (2008) The status of membrane bioreactor technology. Trends in l y,Biotechno og 26, 109‐116.
JUDD, S. & JUDD, C. (2006) The MBR book : principles and applications of embrane bioreactors in water and wastewater treatment, Amsterdam ; Oxford, m
Elsevier. HAWAJI, A. D., KUTUBKHANAH, I. K. & WIE, J. M. (2008) Advances in seawater K
desalination technologies. Desalination, 221, 47‐69. KIM, J., JANG, M., CHIO, H. & KIM, S. (2004) Characteristics of membrane and odule affecting membrane fouling. Proceedings of Water Environment
mMembrane Technology Conference. Seoul, Korea. KIM, K. J., FANE, A. G., NYSTROM, M. & PIHLAJAMAKI, A. (1997) Chemical and electrical characterization of virgin and protein‐fouled polycarbonate track‐tched membranes by FTIR and streaming‐potential measurements. Journal of
eeMembran Science, 134, 199‐208. KOUL, S. L., CHADDERTON, L. T., FINK, D. W. & CRUZ, S. A. (1991) ON THE AILORING OF POLYMERS BY RADIATION. Nuclear Tracks and Radiation TMeasurements, 19, 825‐828. LASPIDOU, C. S. & RITTMANN, B. E. (2002) A unified theory for extracellular olymeric substances, soluble microbial products, and active and inert biomass.
36pWater Research, , 2711‐2720. LAWRENCE, N. D., PERERA, J. M., IYER, M., HICKEY, M. W. & STEVENS, G. W. (2006) The use of streaming potential measurements to study the fouling and leaning of ultrafiltration membranes. Separation and Purification Technology, 106‐112.
c48, LE CLECH, P., JEFFERSON, B., CHANG, I. S. & JUDD, S. J. (2003) Critical flux etermination by the flux‐step method in a submerged membrane bioreactor. dJournal of Membrane Science, 227, 81‐93. E‐CLECH, P., CHEN, V. & FANE, T. A. G. (2006) Fouling in membrane bioreactors
Lused in wastewater treatment. Journal of Membrane Science, 284, 17‐53. LE‐CLECH, P., JEFFERSON, B. & JUDD, S. J. (2003) Impact of aeration, solids oncentration and membrane characteristics on the hydraulic performance of a cmembrane bioreactor. Journal of Membrane Science, 218, 117‐129. LEE, J., AHN, W. Y. & LEE, C. H. (2001) Comparison of the filtration characteristics etween attached and suspended growth microorganisms in submerged embrane bioreactor. Water Research, 35, 2435‐2445.
bm
114
LEE, W., JEON, J.‐H., CHO, Y., CHUNG, K. Y. & MIN, B.‐R. (2005) Behavior of TMP ccording to membrane pore size. International Congress on Membranes and aMembrane Process (ICOM). Seoul, Korea. LEE, W., KANG, S. & SHIN, H. (2003) Sludge characteristics and their contribution o microfiltration in submerged membrane bioreactors. Journal of Membrane tScience, 216, 217‐227. LESJEAN, B., ROSENBERGER, S., LAABS, C., JEKEL, M., GNIRSS, R. & AMY, G. (2005) Correlation between membrane fouling and soluble/colloidal organic ubstances in membrane bioreactors for municipal wastewater treatment. Water sScience and Technology, 51, 1‐8. LIU, Y. & FANG, H. H. P. (2003) Influences of extracellular polymeric substances EPS) on flocculation, settling, and dewatering of activated sludge. Critical (Reviews in Environmental Science and Technology, 33, 237‐273. MADAENI, S. S., FANE, A. G. & WILEY, D. E. (1999) Factors influencing critical flux n membrane filtration of activated sludge. Journal of Chemical Technology and iBiotechnology, 74, 539‐543. CCARTY, P. L. (2001) The development of anaerobic treatment and its future.
149‐156. MWater Science and Technology, 44, MENG, F. G., SHI, B. Q., YANG, F. L. & ZHANG, H. M. (2007) Effect of hydraulic etention time on membrane fouling and biomass characteristics in submerged
and Enginermembrane bioreactors. Bioprocess Biosystems ering, 30, 359‐367. MENG, F. G., YANG, F. L., XIAO, J. N., ZHANG, H. M. & GONG, Z. (2006) A new nsight into membrane fouling mechanism during membrane filtration of bulking
uspension. Journal of Membrane Science, 285, 159‐165. iand normal sludge s METCALF & EDDY. Wastewater engineering: treatment and reuse (2002). MORGAN, J. W., FORSTER, C. F. & EVISON, L. (1990) A comparative‐study of the ature of biopolymers extracted from anaerobic and activated sludges. Water nResearch, 24, 743‐750. NAGAOKA, H., KONO, S., YAMANISHI, S. & MIYA, A. (2000) Influence of organic oading rate on membrane fouling in membrane separation activated sludge
S e , lprocess. Water ci nce and Technology, 41 355‐362. NAGAOKA, H. & NEMOTO, H. (2005) Influence of extracellular polymeric ubstances on nitrogen removal in an intermittently‐aerated membrane
W ,sbioreactor. ater Science and Technology, 51 151‐158. NAGAOKA, N., YAMANISHI, S. & MIYA, A. (1998) Modeling of biofouling by extracellular polymers in a membrane separation activated sludge system. Water Science and Technology, 38, 497‐504.
115
NAKAMURA, K. & MATSUMOTO, K. (2006) Properties of protein adsorption onto ore surface during microfiltration: Effects of solution environment and pmembrane hydrophobicity. Journal of Membrane Science, 280, 363‐374. NG, C. A., SUN, D., ZHANG, J., CHUA, H. C., BING, W. & FANE, A. (2005) Strategies to mprove the sustainable operation of membrane bioreactors. Proceedings of
oniInternational Desalinati Association Conference. Singapore. SMONICS, G. (2000) Do Poretics Track‐Etch Membranes have any type of
http://osmonics.custhelp.comOwetting agents on them? , as of 15th July 2008. PARK, H. D. & NOGUERA, D. R. (2004) Evaluating the effect of dissolved oxygen n ammonia‐oxidizing bacterial communities in activated sludge. Water oResearch, 38, 3275‐3286. PARKIN, G. F. & MCCARTY, P. L. (1981) A comparison of the characteristic of oluble organic nitrogen in untreated and activated‐sludge treated wastewaters. sWater Research, 15, 139‐149. QUANRUD, D. M., HAFER, J., KARPISCAK, M. M., ZHANG, H. M., LANSEY, K. E. & RNOLD, R. G. (2003) Fate of organics during soil‐aquifer treatment: A
sustainability of removals in the field. Water Research, 37, 3401‐3411. RANDALL, C. W., PATTARKINE, V. M. & MCCLINTOCK, S. A. (1992) Nitrification inetics in single‐sludge biological nutrient removal activated‐sludge systems. kWater Science and Technology, 25, 195‐214. RAO, V., AMAR, J. V., AVASTHI, D. K. & CHARYULU, R. N. (2003) Etched ion track olymer membranes for sustained drug delivery. Radiation Measurements, 36, p585‐589. ESEARCH, B. (2006) Membrane Bioreactors in the Changing World Water RMarket. ITTMANN, B. E. & MCCARTY, P. L. (2001) Environmental biotechnology : Rprinciples and applications, New York, McGraw‐Hill. ROSENBERG, M., GUTNICK, D. & ROSENBERG, E. (1980) Adherence of bacteria to ydrocarbons ‐ A simple method for measuring cell‐surface hydrophobicity. hFems Microbiology Letters, 9, 29‐33. ROSENBERGER, S., EVENBLIJ, H., POELE, S. T., WINTGENS, T. & LAABS, C. (2005) The importance of liquid phase analyses to understand fouling in membrane ssisted activated sludge processes ‐ six case studies of different European
2 aresearch groups. Journal of Membrane Science, 263, 113‐1 6. ROSENBERGER, S. & KRAUME, M. (2002) Filterability of activated sludge in membrane bioreactors. Desalination, 146, 373‐379.
116
RUIZ, G., JEISON, D., RUBILAR, O., CIUDAD, G. & CHAMY, R. (2006) Nitrification‐enitrification via nitrite accumulation for nitrogen removal from wastewaters. dBioresource Technology, 97, 330‐335. SATO, T. & ISHII, Y. (1991) Effects of activated‐sludge properties on water flux of ltrafiltration membrane used for human excrement treatment. Water Science uand Technology, 23, 1601‐1608. SHIMIZU, Y., OKUNO, Y., URYU, K., OHTSUBO, S. & WATANABE, A. (1996) Filtration characteristics of hollow fiber microfiltration membranes used in embrane bioreactor for domestic wastewater treatment. Water Research, 30, m
2385‐2392. SHIN, H. S. & KANG, S. T. (2003) Characteristics and fates of soluble microbial roducts in ceramic membrane bioreactor at various sludge retention times.
a , 37 pWater Rese rch , 121‐127. SINKJAER, O., YNDGAARD, L., HARREMOES, P. & HANSEN, J. L. (1994) haracterization of the nitrification process for design purposes. Water Science
, 30,Cand Technology 47‐56. TEPHENSON, T. (2000) Membrane bioreactors for wastewater treatment, SLondon, IWA. TERLITECH CORPORATION (2008) Commercial Track‐etched membranes.
.hShttp://www.2spi.com/catalog/spec_prep/filter3 tml. TANDOI, V., JENKINS, D., WANNER, J. R. & IWA SPECIALIST GROUP ON ACTIVATED SLUDGE POPULATION DYNAMICS. (2006) Activated sludge eparation problems : theory, control measures, practical experience, London ; sSeattle, IWA Publishing. CHOBANOGLOUS, G., BURTON, F. L. & METCALF & EDDY. (1991) Wastewater Tengineering : treatment, disposal, and reuse, New York, : McGraw‐Hill. VIGNATI, D. A. L., LOIZEAU, J. L., ROSSE, P. & DOMINIK, J. (2006) Comparative performance of membrane filters vs. high‐surface area filtration cartridges for he determination of element concentrations in freshwater systems. Water tResearch, 40, 917‐924. VRIJENHOEK, E. M., HONG, S. & ELIMELECH, M. (2001) Influence of membrane urface properties on initial rate of colloidal fouling of reverse osmosis and snanofiltration membranes. Journal of Membrane Science, 188, 115‐128. WANG, Y., HUANG, X. & YUAN, Q. P. (2005) Nitrogen and carbon removals from food processing wastewater by an anoxic/aerobic membrane bioreactor. Process Biochemistry, 40, 1733‐1739.
117
WEIS, A., BIRD, M. R. & NYSTROM, M. (2003) The chemical cleaning of polymeric F membranes fouled with spent sulphite liquor over multiple operational
lUcycles. Journa of Membrane Science, 216, 67‐79. YAMAMOTO, K., HIASA, M., MAHMOOD, T. & MATSUO, T. (1989) Direct solid‐iquid separation using hollow fibre membrane in an activated‐sludge aeration
ltank. Water Science and Technology, 21, 43‐54. YAMATO, N., KIMURA, K., MIYOSHI, T. & WATANABE, Y. (2006) Difference in embrane fouling in membrane bioreactors (MBRs) caused by membrane m
polymer materials. Journal of Membrane Science, 280, 911‐919. YAMAZAKI, I. M., PATERSON, R. & GERALDO, L. P. (1996) A new generation of rack etched membranes for microfiltration and ultrafiltration .1. Preparation
. tand characterisation Journal of Membrane Science, 118, 239‐245. YANG, Q. Y., CHEN, J. H. & ZHANG, F. (2006) Membrane fouling control in a ubmerged membrane bioreactor with porous, flexible suspended carriers. sDesalination, 189, 292‐302. ZHANG, J., CHUA, H. C., ZHOU, J. & FANE, A. G. (2006) Factors affecting the embrane performance in submerged membrane bioreactors. Journal of m
Membrane Science, 284, 54‐66. ZHAO, C. D., VATER, P. & BRANDT, R. (1991) Further‐studies on the production
t of PVDF nuclear rack microfilters. Nuclear Tracks and Radiation Measurements, 19, 829‐832. ZYDNEY, A. L. & HO, C. C. (2003) Effect of membrane morphology on system apacity during normal flow microfiltration. Biotechnology and Bioengineering, 3, 537‐543. c8