MITIGATION OF MEMBRANE FOULING IN

MICROFILTRATION & ULTRAFILTRATION OF ACTIVATED

SLUDGE EFFLUENT FOR WATER REUSE

A thesis submitted in fulfilment of the requirement for the degree of

Doctor of Philosophy (PhD)

Sy Thuy Nguyen

Bachelor of Engineering (Chemical), RMIT University, Melbourne

School of Civil, Environmental & Chemical Engineering

Department of Chemical Engineering

RMIT University, Melbourne, Australia

December 2012

i
 DECLARATION

I hereby declare that:

 the work presented in this thesis is my own work except where due acknowledgement

has been made;

 the work has not been submitted previously, in whole or in part, to qualify for any other

academic award;

 the content of the thesis is the result of work which has been carried out since the

official commencement date of the approved research program

Signed by

Sy T. Nguyen

ii
 ACKNOWLEDGEMENTS

Firstly, I would like to thank Prof. Felicity Roddick, my senior supervisor, for giving me the
opportunity to do a PhD in water science and technology and her scientific input in every stage
of this work. I am also grateful to Dr John Harris and Dr Linhua Fan, my co-supervisors, for
their ponderable advice and positive criticisms.

The Australian Research Council (ARC) and RMIT University are acknowledged for providing
the funding and research facilities to make this project and thesis possible.

I also wish to thank the staff of the School of Civil, Environmental and Chemical Engineering,
the Department of Applied Physics and the Department of Applied Chemistry, RMIT
University, for the valuable academic, administrative and technical assistance.

Lastly, my parents are thanked for the encouragement and support I most needed during the
time this research was carried out.

Sy Nguyen
March 2012

iii
 LIST OF PUBLICATIONS

Refereed journals
S. T. Nguyen, F. A. Roddick and J.L. Harris (2010), Membrane foulants and fouling
mechanisms in microfiltration and ultrafiltration of an activated sludge effluent, Water Science
& Technology, 62(9), 1975-1983.

S. T. Nguyen, F. A. Roddick (2010), Effects of ozonation and biological activated carbon

filtration on membrane fouling in ultrafiltration of an activated sludge effluent, Journal of
Membrane Science, 363(1-2), 271-277.

S. T. Nguyen and F. A. Roddick (2011), Chemical cleaning of ultrafiltration membrane fouled

by an activated sludge effluent, Desalination & Water Treatment, 34, 94-99.

S. T. Nguyen, F. A. Roddick (2013), Pre-treatments for removing colour from secondary

effluent: Effectiveness and influence on membrane fouling in subsequent microfiltration,
Separation & Purification Technology, 103, 313-320.

Refereed conference proceedings

Membrane foulants and fouling mechanisms in microfiltration and ultrafiltration of an activated
sludge effluent, in Proceedings & Abstracts of the 7th IWA World Congress on Water
Reclamation and Reuse, 20-25 September 2009, Brisbane, Australia.

S. T. Nguyen, F. A. Roddick and J.L. Harris (2010), Multi-cycle test for evaluating membrane
fouling propensity in ultrafiltration of a secondary effluent, in Proceedings & Abstracts of the
5th IWA Young Water Professionals Conference, 5-7 July 2010, Sydney, Australia.

S. T. Nguyen and F. A. Roddick, Chemical cleaning of ultrafiltration membrane fouled by an

activated sludge effluent, in Proceedings of the 6th Conference of the Aseanian Membrane
Society, 22-26 November 2010, Sydney, Australia.

iv
 TABLE OF CONTENTS

SUMMARY ................................................................................................................................................................ 1

CHAPTER 1 . INTRODUCTION ............................................................................................................................ 4

1.1. BACKGROUND OF THE PROJECT ......................................................................................................................... 4

1.2. RATIONALE ....................................................................................................................................................... 4
1.3. RESEARCH OBJECTIVES ..................................................................................................................................... 5

CHAPTER 2 . LITERATURE REVIEW ................................................................................................................ 8

2.1. OVERVIEW OF CONVENTIONAL MUNICIPAL WASTEWATER TREATMENT PROCESSES .......................................... 8

2.1.1. Preliminary treatment .............................................................................................................................. 8
2.1.2. Primary treatment .................................................................................................................................... 8
2.1.3. Secondary treatment ................................................................................................................................. 8
2.1.4. Tertiary treatment .................................................................................................................................... 9
2.2. MEMBRANE FILTRATION IN WATER AND WASTEWATER TREATMENT ................................................................ 9
2.2.1. Membrane classification ........................................................................................................................ 10
2.2.2. Membrane materials and structures ....................................................................................................... 11
2.2.3. Membrane configurations ...................................................................................................................... 12
2.2.4. Operation modes of membrane processes .............................................................................................. 14
2.2.5. Feedwater for membrane processes for water reuse .............................................................................. 14
2.3. MEMBRANE FOULING AND FACTORS AFFECTING FOULING .............................................................................. 16
2.3.1. Effect of feedwater content ..................................................................................................................... 17
2.3.2. Effect of solution chemistry .................................................................................................................... 21
2.3.3. Effect of membrane characteristics ........................................................................................................ 22
2.3.4. Effect of membrane configuration .......................................................................................................... 23
2.3.5. Effect of operating conditions ................................................................................................................ 24
2.4. RESISTANCES IN MEMBRANE FILTRATION AND FOULING MODELS ................................................................... 24
2.4.1. Fouling models for dead-end filtration .................................................................................................. 25
2.4.2. Fouling models for cross-flow filtration ................................................................................................. 28
2.4.3. Concentration polarisation .................................................................................................................... 29
2.4.4. Resistance-in-series model ..................................................................................................................... 30
2.5. FOULING MITIGATION ..................................................................................................................................... 31
2.5.1. Pre-treatment of the feedwater ............................................................................................................... 31
2.5.2. Selection of membrane types and surface modification ......................................................................... 40
2.5.3. Optimising module configuration and process conditions ..................................................................... 40
2.5.4. Membrane cleaning ................................................................................................................................ 41
2.6. EVALUATING THE FOULING POTENTIAL OF FEEDWATER .................................................................................. 46
2.6.1. Modified fouling index (MFI) ................................................................................................................. 46
2.6.2. Unified membrane fouling index (UMFI) ............................................................................................... 47

v
 2.7. CHARACTERISATION OF FILTRATION MEMBRANES .......................................................................................... 48
2.7.1. Contact angle measurement ................................................................................................................... 49
2.7.2. Streaming potential ................................................................................................................................ 49
2.7.3. Atomic force microscopy (AFM) ............................................................................................................ 49
2.8. TECHNIQUES FOR FOULING CHARACTERISATION ............................................................................................. 50
2.8.1. Characterisation of organic matter ........................................................................................................ 50
2.8.2. Characterisation of foulant materials on membranes ............................................................................ 51

CHAPTER 3 . MATERIALS AND METHODS ................................................................................................... 53

3.1. WASTEWATER SOURCES.................................................................................................................................. 53

3.1.1. Raw and ozonated effluent and BAC filtrate .......................................................................................... 53
3.1.2. Coagulated samples ............................................................................................................................... 56
3.1.3. Samples pre-treated with an anion exchange resin (AER) ..................................................................... 56
3.1.4. Samples pre-treated with a powdered activated carbon (PAC) ............................................................. 57
3.2. ANALYTICAL METHODS FOR WASTEWATER SAMPLES ..................................................................................... 58
3.2.1. General characteristics .......................................................................................................................... 58
3.2.2. Fractionation of EfOM ........................................................................................................................... 60
3.2.3. Sample AMWD ....................................................................................................................................... 62
3.3. MEMBRANE ANALYSIS .................................................................................................................................... 64
3.3.1. ATR-FTIR spectroscopy ......................................................................................................................... 64
3.3.2. SEM/EDX spectroscopy ......................................................................................................................... 64
3.3.3. Contact angle measurement ................................................................................................................... 64
3.4. DEAD-END FILTRATION EXPERIMENTS ............................................................................................................ 65
3.4.1. The membranes used in dead-end filtration ........................................................................................... 65
3.4.2. Dead-end filtration system ..................................................................................................................... 65
3.4.3. Operating conditions and general procedure......................................................................................... 66
3.4.4. Single-cycle filtration ............................................................................................................................. 67
3.4.5. Multi-cycle filtration............................................................................................................................... 69
3.5. CHEMICAL CLEANING OF THE FOULED POLYMERIC MEMBRANES .................................................................... 69
3.5.1. Membrane fouling and cleaning procedure ........................................................................................... 69
3.5.2. Determination of the cleaning efficiency ................................................................................................ 70
3.5.3. Membrane analysis................................................................................................................................. 71
3.6. CROSS-FLOW FILTRATION EXPERIMENTS......................................................................................................... 72
3.6.1. The cross-flow filtration system and ceramic membranes...................................................................... 72
3.6.2. Operating conditions for cross-flow filtration tests................................................................................ 73
3.6.3. Determination of the membrane foulants and the effect of feed pre-treatments on the permeate flux in
cross-flow filtration .......................................................................................................................................... 74
3.6.4. Effect of rapid membrane back-flushing on the flux performance ......................................................... 74
3.7. REPRODUCIBILITY OF FILTRATION EXPERIMENTS ............................................................................................ 75
3.8. ANALYSIS OF FLUX DATA FOR THE IDENTIFICATION OF FOULING MECHANISMS .............................................. 76

vi
CHAPTER 4 . DEAD-END FILTRATION OF THE RAW ACTIVATED SLUDGE EFFLUENT ................ 78

4.1. EFFECT OF SUSPENDED SOLIDS (SS) ON FLUX DECLINE IN MF AND UF ........................................................... 79

4.2. IDENTIFICATION OF THE EFOM COMPONENTS RESPONSIBLE FOR FOULING ..................................................... 81
4.2.1. AMWDs of dissolved EfOM in the feed and the permeates .................................................................... 81
4.2.2. Foulant identification by ATR-FTIR spectroscopy ................................................................................. 83
4.2.3. Fouling potentials of individual EfOM fractions.................................................................................... 86
4.3. DOM REJECTION AND COLOUR REMOVAL BY THE MEMBRANES...................................................................... 89
4.4. FLUX DECLINE AND FOULING MECHANISMS IN MF AND UF OF THE RAW EFFLUENT ....................................... 90
4.4.1. Flux decline and fouling mechanisms in MF.......................................................................................... 90
4.4.2. Flux decline and fouling mechanisms in UF .......................................................................................... 92
CHAPTER 4 SUMMARY ........................................................................................................................................... 93

CHAPTER 5 . EFFECT OF COAGULATION ON MEMBRANE FOULING IN DEAD-END FILTRATION

................................................................................................................................................................................... 95

5.1. EFFECT OF COAGULATION ON MEMBRANE FOULING IN SINGLE-CYCLE MF AND UF ........................................ 96

5.1.1. Single-cycle MF...................................................................................................................................... 96
5.1.2. Single-cycle UF .................................................................................................................................... 103
5.1.3. Multi-cycle filtration............................................................................................................................. 106
5.2. REMOVAL OF DOC AND TRUE COLOUR BY COAGULATION-MF/UF HYBRID ................................................. 108
CHAPTER 5 SUMMARY ......................................................................................................................................... 109

CHAPTER 6 . EFFECT OF OZONATION & BIOLOGICAL ACTIVATED CARBON FILTRATION ON

MEMBRANE FOULING IN DEAD-END FILTRATION ................................................................................ 110

6.1. EFFECT OF OZONATION AND BAC FILTRATION ON CHARACTERISTICS OF THE RAW EFFLUENT ..................... 111
6.1.1. Effect on general characteristics .......................................................................................................... 111
6.1.2. Effect on EfOM fractions ...................................................................................................................... 112
6.1.3. Effect on AMWD of dissolved EfOM .................................................................................................... 114
6.2. EFFECT OF OZONATION AND BAC FILTRATION ON MF FLUX ........................................................................ 118
6.3. EFFECT OF OZONATION AND BAC FILTRATION ON UF FLUX ......................................................................... 121
6.3.1. UF with the HFM-180 membrane ........................................................................................................ 121
6.3.2. UF with the HFK-141 membrane ......................................................................................................... 122
6.4. EFFECT OF OZONATION AND BAC FILTRATION ON HYDRAULICALLY IRREVERSIBLE FOULING ...................... 123
6.4.1. Results from single-cycle filtration tests ............................................................................................... 123
6.4.2. Results from multi-cycle filtration tests ................................................................................................ 124
6.5. POTENTIAL EFFECT OF OZONATION AND BAC FILTRATION ON BIOFOULING PROPENSITY OF DOWNSTREAM
MEMBRANES ........................................................................................................................................................ 127

CHAPTER 6 SUMMARY ......................................................................................................................................... 128

CHAPTER 7 . EFFECT OF ADSORPTION PRE-TREATMENTS WITH A POWDERED ACTIVATED

CARBON AND AN ANION EXCHANGE RESIN ON MEMBRANE FOULING ......................................... 130

7.1. EFFECT OF ADSORPTION PRE-TREATMENTS ON RAW EFFLUENT CHARACTERISTICS ....................................... 131

vii
 7.2. EFFECT OF ADSORPTION PRE-TREATMENTS ON AMWD OF THE RAW EFFLUENT ........................................... 131
7.3. EFFECT OF ADSORPTION PRE-TREATMENTS ON MEMBRANE FOULING ............................................................ 134
7.3.1. Effect on total fouling ........................................................................................................................... 134
7.3.2. Effect on hydraulically irreversible fouling.......................................................................................... 139
7.4. PRACTICABILITY CONSIDERATIONS OF THE ADSORPTION PRE-TREATMENTS IN COMPARISON WITH OZONATION
............................................................................................................................................................................ 140
CHAPTER 7 SUMMARY ......................................................................................................................................... 142

CHAPTER 8 . CHEMICAL CLEANING OF FOULED UF POLYMERIC MEMBRANES ........................ 144

8.1. REMOVAL OF FOULANTS BY BACKWASHING ................................................................................................. 144

8.2. CLEANING EFFICIENCIES OF DIFFERENT CHEMICALS ..................................................................................... 144
8.3. MEMBRANE CLEANLINESS SHOWN BY ATR-FTIR SPECTROSCOPY AND EDX SPECTROSCOPY ...................... 150
CHAPTER 8 SUMMARY ......................................................................................................................................... 152

CHAPTER 9 . MEMBRANE FOULING IN CROSS-FLOW FILTRATION ................................................. 154

9.1. FLUX DECLINE AND MEMBRANE FOULANTS IN CROSS-FLOW MF AND UF OF THE RAW EFFLUENT ................ 155
9.1.1. Effect of SS on flux decline ................................................................................................................... 155
9.1.2. Membrane foulants in cross-flow MF and UF ..................................................................................... 157
9.1.3. DOC and colour rejection by the ceramic membranes ........................................................................ 158
9.2. EFFECT OF RAPID BACK-FLUSHING ON FLUX PERFORMANCE IN CROSS-FLOW FILTRATION ............................ 158
9.3. EFFECT OF FEED PRE-TREATMENTS WITH OZONE AND THE AER ON PERMEATE FLUX ................................... 162
9.4. EFFECT OF FEED COAGULATION WITH ALUM ON FLUX PERFORMANCE IN CROSS-FLOW MF .......................... 169
CHAPTER 9 SUMMARY ......................................................................................................................................... 171

CHAPTER 10 . CONCLUSIONS & RECOMMENDATIONS ......................................................................... 173

10.1. COMPONENTS IN THE EFFLUENT CONTRIBUTING TO FLUX DECLINE AND FOULING MECHANISMS ................ 173
10.2. EFFECT OF COAGULATION ON MEMBRANE FOULING .................................................................................... 174
10.3. EFFECT OF OZONATION AND BIOLOGICAL ACTIVATED CARBON (BAC) FILTRATION ON MEMBRANE FOULING
............................................................................................................................................................................ 175
10.4. EFFECTS OF ADSORPTION PRE-TREATMENTS ON MEMBRANE FOULING ........................................................ 175
10.5. EFFECT OF RAPID MEMBRANE BACK-FLUSHING ON FLUX PERFORMANCE IN CROSS-FLOW FILTRATION ....... 176
10.6. CHEMICAL CLEANING OF THE FOULED MEMBRANES ................................................................................... 176
10.7. IMPLICATIONS OF THESE FINDINGS .............................................................................................................. 176
10.8. RECOMMENDATIONS FOR FUTURE WORK .................................................................................................... 177

REFERENCES ...................................................................................................................................................... 179

APPENDICES ........................................................................................................................................................ 207

APPENDIX A. CONTACT ANGLES OF THE FLAT-SHEET POLYMERIC MEMBRANES ........................................................ 208

APPENDIX B. EXAMPLE OF DATA PROCESSING FOR A FILTRATION EXPERIMENT........................................................ 209
APPENDIX C. SAMPLE CALCULATION OF THE CLEANING EFFICIENCY OF THE MEMBRANE CLEANERS ......................... 213

viii
APPENDIX D. EFFECT OF OZONATION AND BAC FILTRATION ON THE EFFLUENT CHARACTERISTICS AND THE UMFI FOR
TOTAL FOULING FOR THE HFM-180 MEMBRANE .................................................................................................... 217

APPENDIX E. EFFECT OF OZONATION AND BAC FILTRATION ON THE DOC CONCENTRATIONS OF THE EFOM
FRACTIONS AS DETERMINED BY RESIN FRACTIONATION............................................................................................ 220

APPENDIX F. LC-OCD CHROMATOGRAMS OF THE SAMPLES LISTED IN TABLE 6.3 BUT NOT INCLUDED IN FIGURE 6.2
............................................................................................................................................................................ 221
APPENDIX G. PARTICLE SIZE DISTRIBUTIONS AND ZETA POTENTIALS OF THE RAW EFFLUENT AND THE OZONATED
EFFLUENT ............................................................................................................................................................. 222

APPENDIX H. LC-OCD CHROMATOGRAMS OF THE OZONATED EFFLUENT AND ITS MF PERMEATE WITH THE XUNP-003
MEMBRANE (SAMPLES LISTED IN TABLE 7.2 BUT NOT INCLUDED IN FIGURE 7.3) ..................................................... 224
APPENDIX I. CHARACTERISTICS OF THE SAMPLES USED FOR THE UF TESTS SHOWN IN FIGURE 7.4 ........................... 225
APPENDIX J. EFFECT OF CYCLIC CHEMICAL CLEANING WITH SDS AND TERG-A-ZYME ON THE PERMEATE FLUX OF THE
HFM-116 MEMBRANE........................................................................................................................................... 226
APPENDIX K. SEM IMAGES OF THE VIRGIN AND CHEMICALLY CLEANED HFM-116 MEMBRANE SAMPLES ................. 227
APPENDIX L. CHARACTERISTICS OF THE SAMPLES USED IN CROSS-FLOW FILTRATION TESTS ..................................... 228

ix
 LIST OF FIGURES

Figure 2.1. Schematic presentations of (a) dead-end mode and (b) cross-flow mode ................ 14
Figure 2.2. Schematic presentation of different membrane fouling models ............................... 27
Figure 3.1. The wastewater treatment process at the WWTP ..................................................... 53
Figure 3.2. EfOM fractionation procedure (three-fraction protocol) .......................................... 60
Figure 3.3. EfOM fractionation procedure (four-fraction protocol) ........................................... 62
Figure 3.4. Filtration experimental set-up for the flat-sheet membranes .................................... 66
Figure 3.5. A schematic diagram of the cross-flow filtration XLAB5 pilot unit ........................ 72
Figure 3.6. Reproducibility of (a) MF and (b) UF flux data for dead-end filtration ................... 75
Figure 3.7. Reproducibility of flux data for cross-flow filtration ............................................... 76
Figure 4.1. Effect of SS on (a) MF flux and (b) UF flux of the raw AS effluent........................ 79
Figure 4.2. AMWDs of the EfOM in the raw effluent, the permeates and the backwash water 82
Figure 4.3. ATR-FTIR spectra of the virgin and fouled membranes .......................................... 85
Figure 4.4. Flux data for MF and UF of different fractions in the raw AS effluent.................... 86
Figure 4.5. LC-OCD chromatograms of the three EfOM fractions of the raw AS effluent........ 88
Figure 4.6. Plots of t/V versus V for MF of the raw AS effluent ................................................. 90
Figure 4.7. Filtration law analysis for the MF membranes ......................................................... 91
Figure 4.8. Plots of t/V versus V for UF of the raw AS effluent ................................................ 93
Figure 5.1. Plots of (a) t/V versus V and (b) corresponding J/J0 versus v for MF of the raw and
coagulated samples with the VVLP membrane .......................................................................... 98
Figure 5.2. Plots of (a) t/V versus V and (b) corresponding J/J0 versus v for MF of the raw and
coagulated samples with the XUNP-003 membrane ................................................................... 99
Figure 5.3. AMWDs of the raw and coagulated effluent .......................................................... 100
Figure 5.4. AMWDs of the feed and the permeates from MF of the raw effluent with the VVLP
and XUNP-003 membranes....................................................................................................... 102
Figure 5.5. Removal of biopolymers and HS in the coagulated effluent by the VVLP and
XUNP-003 membranes ............................................................................................................. 103
Figure 5.6. Plots of (a) t/V versus V and (b) corresponding J0 /J versus v for UF of the raw and
coagulated samples with HFM-180 membrane ......................................................................... 104
Figure 5.7. Plots of (a) t/V versus V and (b) corresponding J0 /J versus v for UF of the raw and
coagulated samples with the HFK-141 membrane.................................................................... 105
Figure 5.8. Flux data for multi-cycle MF (top) and calculation of UMFI for hydraulically
irreversible fouling (bottom) for the XUNP-003 membrane..................................................... 107

x
Figure 5.9. Flux data for multi-cycle UF with the raw and the coagulated effluent for the HFM-
180 membrane ........................................................................................................................... 108
Figure 6.1. EfOM fractions in raw and ozonated AS effluents and in BAC filtrate ................. 114
Figure 6.2. Effects of ozonation, BAC filtration, and UF on AMWD of the raw AS effluent . 116
Figure 6.3. Effect of ozonation and BAC filtration on flux of the VVLP membrane ............... 119
Figure 6.4. Effect of ozonation and BAC filtration on flux of the XUNP-003 membrane ....... 119
Figure 6.5. Particle size distributions of the raw and ozonated AS effluent ............................. 120
Figure 6.6. Flux data for UF of the raw and ozonated effluent and the BAC filtrate with the
HFM-180 membrane ................................................................................................................. 121
Figure 6.7. Flux data for UF of the raw and ozonated effluent and the BAC filtrate with the
HFK-141 membrane .................................................................................................................. 123
Figure 6.8. Effect of ozonation on permeate flux of the HFM-180 membrane in multi-cycle
filtration ..................................................................................................................................... 126
Figure 6.9. Effect of ozonation on permeate flux of the HFK-141 membrane in multi-cycle
filtration ..................................................................................................................................... 127
Figure 7.1. AMWDs of the raw, ozonated, AER-treated, and PAC-treated effluents .............. 132
Figure 7.2. Effect of ozonation and adsorption pre-treatments on permeate flux of the XUNP-
003 membrane ........................................................................................................................... 135
Figure 7.3. Effect of adsorption pre-treatments on the rejection of EfOM components by the
XUNP-003 membrane ............................................................................................................... 137
Figure 7.4. Flux data for UF of the raw, ozonated, AER-treated, and PAC-treated effluent
samples with the HFM-180 membrane ..................................................................................... 138
Figure 7.5. Flux data for UF of the raw, ozonated, AER-treated, and PAC-treated effluent
samples with the HFK-141 membrane ...................................................................................... 138
Figure 7.6. Flux data for multi-cycle MF of the raw, PAC-treated and AER-treated effluents
with the XUNP-003 membrane ................................................................................................. 140
Figure 8.1. Cleaning efficiencies (ERW) of different cleaners on the fouled HFM-116 membrane
................................................................................................................................................... 146
Figure 8.2. Changes in WFR of NaOCl, SDS, and Terg-a-zyme over three UF cycles with the
HFM-116 membrane ................................................................................................................. 149
Figure 8.3. Effect of cyclic cleaning with Terg-a-zyme on the permeate flux in UF of the raw
effluent with the HFM-116 membrane ...................................................................................... 150
Figure 8.4. ATR-FTIR spectra of the virgin, fouled, and cleaned HFM-116 membranes ........ 151
Figure 8.5. EDX spectra of the virgin, fouled, and backwashed HFM-116 membranes .......... 152

xi
Figure 9.1. Flux data for (a) MF (TMP = 20 kPa) and (b) UF (TMP = 30 kPa) of the raw AS
effluent with and without SS at cross-flow velocity 1.8 m s-1 .................................................. 156
Figure 9.2. AMWDs of the raw AS effluent and the permeates in cross-flow filtration .......... 157
Figure 9.3. Effect of rapid back-flushing and increasing cross-flow velocity on flux performance
in MF of the raw AS effluent .................................................................................................... 160
Figure 9.4. Effect of rapid back-flushing and increasing cross-flow velocity on flux performance
in UF of the raw AS effluent ..................................................................................................... 161
Figure 9.5. Effect of pre-treatments for colour removal with ozone and the AER on permeate
flux in cross-flow MF ................................................................................................................ 164
Figure 9.6. Effect of pre-treatments for colour removal with ozone and the AER on permeate
flux in cross-flow UF ................................................................................................................ 164
Figure 9.7. Effect of (a) ozonation and (b) adsorption pre-treatment with the AER on the
rejection of proteins and polysaccharides of the ceramic MF membrane ................................. 166
Figure 9.8. Effect of (a) ozonation and (b) adsorption pre-treatment with the AER on the
rejection of different EfOM components of the ceramic UF membrane .................................. 167
Figure 9.9. Effect of ozonation on the permeate flux in cross-flow MF with rapid back-flushing
................................................................................................................................................... 168
Figure 9.10. Effect of ozonation on the permeate flux in cross-flow UF with rapid back-
flushing ...................................................................................................................................... 169
Figure 9.11. Effect of coagulation with alum on permeate flux in cross-flow MF without
membrane back-flushing ........................................................................................................... 170
Figure 9.12. Effect of coagulation with alum on permeate flux in cross-flow MF with
membrane back-flushing ........................................................................................................... 171

xii
 LIST OF TABLES

Table 2.1 Some characteristics of different membrane types...................................................... 10

Table 3.1. Characteristics of the raw AS effluent used in this study .......................................... 54
Table 3.2. Characteristics of the AER used in this study ............................................................ 56
Table 3.3. Characteristics of the PAC used in the study ............................................................. 57
Table 3.4. Characteristics of the polymeric membranes used in this study ................................ 65
Table 3.5. The membrane cleaners investigated ......................................................................... 70
Table 3.6. Characteristics of the ceramic membranes used in cross-flow filtration experiments 72
Table 4.1. Hydraulic resistances for MF and UF of the raw effluent with and without SS ........ 80
Table 4.2. DOM rejection and colour removal by the polymeric membranes ............................ 89
Table 5.1. Characteristics of the raw and the coagulated effluent samples used in single-cycle
dead-end filtration tests ............................................................................................................... 96
Table 5.2. Effect of coagulation on the hydraulic resistances from total fouling and irreversible
fouling in MF of the effluent, and calculated pure water flux recovery by backwashing......... 101
Table 5.3. Values of UMFI for total fouling and resistances from total fouling and hydraulically
irreversible fouling for UF of the raw and coagulated effluent ................................................. 106
Table 5.4. Effect of coagulation on the UMFI for hydraulically irreversible fouling ............... 108
Table 6.1. Effects of ozonation and BAC filtration on the characteristics of the raw effluent . 111
Table 6.2. Effect of ozonation and BAC filtration on the characteristics of the VHA and SHA
fractions in the raw effluent....................................................................................................... 113
Table 6.3. Effects of ozonation, BAC filtration, MF with the VVLP membrane and UF with the
HFM-180 membrane on the concentrations of EfOM components (in mg C L-1) as determined
by LC-OCD ............................................................................................................................... 117
Table 6.4. Effect of ozonation and BAC filtration on resistances from total fouling and
hydraulically irreversible fouling (for the flux data shown in Figures 6.3, 6.4, 6.6, and 6.7) .. 124
Table 6.5. Effect of ozonation on UMFI for hydraulically irreversible fouling ....................... 125
Table 7.1. Effect of ozonation and adsorption pre-treatment with the AER and the PAC on the
characteristics of the raw effluent ............................................................................................. 131
Table 7.2. Effect of ozonation and the adsorption pre-treatments with and without MF on the
concentrations of dissolved EfOM components (in mg C L-1) as determined by LC-OCD ..... 133
Table 7.3. Effect of adsorption pre-treatments and ozonation on resistance by total fouling and
resistance by hydraulically irreversible fouling ........................................................................ 139

xiii
Table 8.1. Characteristics of the raw effluent sample used in the membrane cleaning
experiments ............................................................................................................................... 145
Table 8.2. Effect of chemical cleaning with NaOCl, SDS, and Terg-a-zyme on the rejection of
DOC and UV-absorbance of the HFM-116 membrane............................................................. 150
Table 9.1. Characteristics of the raw effluent samples used in cross-flow MF and UF tests ... 155
Table 9.2. Average flux and accumulated specific permeate volume at the end of 7-hour MF and
UF tests with the raw AS effluent at various operating conditions ........................................... 162
Table 9.3. Characteristics of the raw, ozonated, and AER-treated effluent samples used in cross-
flow MF and UF tests ................................................................................................................ 163
Table 9.4. Effect of ozonation and pre-treatment with the AER on the rejection of biopolymers
and HS (measured in mg C L-1) by the MF and UF ceramic membranes ................................. 165

xiv
 NOMENCLATURE

ACH: Poly(aluminium) chlorohydrate HS: Humic substances

AER: Anion exchange resin LC-OCD: Liquid chromatography with UV detection
AFM: Atomic force microscopy and organic carbon detection
AMWD: Apparent molecular weight distribution MBR: Membrane bioreactor
AOC: assimilable organic carbon MF: Microfiltration
AS: Activated sludge MFI: Modified fouling index
ATR-FTIR: Attenuated total reflection-Fourier MW: Molecular weight
transform infrared spectroscopy MWCO: Molecular weight cut-off
BAC: Biological activated carbon NEU: Hydrophilic neutral fraction
BOD: Biological oxygen demand NF: Nanofiltration
BV: bed volume NOM: Natural organic matter
CDOC: Chromatographic dissolved organic carbon OCD: Organic carbon detection/detector
CHA: Hydrophilic charged fraction PAC: Powdered activated carbon
COD: Chemical oxygen demand PACl: Poly(aluminium) chloride
-1
Da: Dalton (= g mol ) PAN: Polyacrylonitrile
DAX-8: Acrylic ester resin PVDF: Polyvinylidene fluoride
DOC: Dissolved organic carbon PS: polysulfone
DOM: Dissolved organic matter PES: polyethersulfone
DBPs: Disinfection by-products (DBPs) PEEK: Polyetheretherketone
EDTA: Ethylenediaminetetraacetic acid PTFE: Polytetrafluoroethylene
EDX: Energy dispersive X-ray spectroscopy PVP: Polyvinyl pyrrolidone
EEM: Excitation-emission matrix RO: Reverse osmosis
EfOM: Effluent organic matter SDS: Sodium dodecyl sulphate
EPS: Extracellular polymeric substances SEC: Size exclusion chromatography
HFM-116: polyvinylidene fluoride membrane with a SEM: Scanning electron microscopy
molecular weight cut-off of 100 kDa (supplied by SHA: Slightly hydrophobic acid fraction (=
Koch Membrane Systems) Transphilic fraction)
HFM-180: polyvinylidene fluoride membrane with a SOC: Synthetic organic compounds
molecular weight cut-off of 100 kDa (Koch SMP: Soluble microbial products
Membrane Systems) SS: Suspended solids, in this thesis SS is defined as
HFK-141: polyethersulfone membrane with a the particulate matter retained by a 0.5 μm glass-
molecular weight cut-off of 30 kDa (Koch fibre filter
Membrane Systems) SUVA: specific UV absorbance [L m-1mg-1]
HOC: Hydrophobic organic carbon TFC: Thin-film composite
HPI: Hydrophilic fraction (= hydrophilic charged TMP: Transmembrane pressure [Pa]
fraction + hydrophilic neutral fraction) TOC: Total organic carbon
HPO: Hydrophobic acid fraction (= very TPI: Transphilic fraction
hydrophobic acid fraction) TSS: Total suspended solids

xv
UF: Ultrafiltration J BW : Pure water flux of the fouled and then
UMFI: Unified membrane fouling index
backwashed membrane [L m-2 h-1]
UV: Ultraviolet
J c : Pure water flux of the membrane after chemical
UVA254: UV absorbance at the wavelength 254 nm
[cm-1] cleaning [L m-2 h-1]

UVD: UV detection/detector J  : Critical flux in cross-flow filtration models [m

VCF: Volume concentration factor s-1]
VHA: Very hydrophobic acid fraction j c : Critical flux for the complete blocking model in
VVLP: Flat-sheet polyvinylidene fluoride membrane
cross-flow filtration [m s-1]
with pore size 0.1 μm (supplied by Millipore)
ji : Critical flux for the intermediate blocking model
XUNP-003: Hollow-fibre polyvinylidene fluoride
membrane with pore size 0.1 μm (supplied by Pall in cross-flow filtration [m s-1]

Corporation) j s : Critical flux for the cake filtration model in

XAD-4: Polyaromatic resin cross-flow filtration [m s-1]
IRA-958: Strong anion exchange resin
K s : Hermia’s parameter for standard blocking
model filtration in cross-flow filtration
Symbols
k : Constant in dead-end filtration models
A : Effective membrane surface area [m2]
k CB : Constant for the complete blocking models in
A0 : Reference membrane area for the calculation of
dead-end filtration [s-1]
the MFI, equal to dynamic viscosity at 20°C, equal
k IB : Constant for the intermediate blocking model
to 13.8  10-4 m2
in dead-end filtration [s-1]
C b : Particle/solute concentration in the bulk feed
k SB : Constant for the standard blocking (or pore
C f : Concentration of the membrane foulant in the
adsorption) model in dead-end filtration (s-1)
feed
k CF : Constant of the cake filtration model in dead-
E RW : Cleaning efficiency of the membrane cleaner
end filtration [s-1]
[%]
mc : Mass of the cake
fn( J , n) : Function relating the permeate flux at
n : Constant in dead-end and cross-flow filtration
time t and at time t = 0 in cross-flow filtration for
models, n = 0, 1, 1.5, and 2 correspond to the cake
the different models
filtration, intermediate pore blocking, standard pore
G : Constant in the cake filtration model for cross- blocking, and complete pore blocking models,
flow filtration [s m-2] respectively
-2 -1
J : Permeate flux [L m h ] P : Transmembrane pressure [Pa]
-2
J 0 : Pure water flux of the virgin membrane [L m P0 : Reference transmembrane pressure for the
-1
h ] calculation of the modified fouling index (MFI),
J  : Permeate flux at the end of the filtration equal to 2.0 bar
-2 -1
experiment [L m h ] Ra : Hydraulic resistance exerted by adsorptive
fouling [m-1]

xvi
Rc : Hydraulic resistance exerted by the cake layer t : Filtration time [s]
[m-1] V : permeate volume [m3]

R̂c : Specific resistance of the cake formed on the V : Permeate volumetric flow rate at time t [m3 s-1]

membrane surface V0 : Permeate volumetric flow rate at time t = 0 [m3

Rcp : Hydraulic resistance exerted by concentration s-1]
v : Accumulated specific permeate volume, i.e., total
polarisation [m-1]
permeate volume per unit membrane area [L m-2]
Rif : Hydraulic resistance exerted by hydraulically
WFR : Pure water flux recovery by chemical
irreversible fouling [m-1]
cleaning [%]
Rm : Hydraulic resistance exerted by the virgin
membrane, Greek letters
R p : Hydraulic resistance exerted by pore blocking  0 : Membrane surface porosity at time t =0
[m-1]  : Dynamic viscosity [Pa s]
Rres : Hydraulic resistance exerted by residual  0 : Reference dynamic viscosity for the calculation
-1
fouling after chemical cleaning [m ] of the MFI, equal to dynamic viscosity at 20°C [Pa s]
Rrf : Hydraulic resistance exerted by hydraulically  : Differential osmotic pressure [Pa]
reversible fouling [m-1]  : Blocked area per unit volume of filtrate [m-1]

Rtf : Hydraulic resistance exerted by total fouling

[m-1]

xvii
 SUMMARY

Membrane fouling has been a major obstacle in microfiltration (MF) and ultrafiltration (UF)
of secondary effluent. Feed pre-treatment is a common practice for fouling mitigation.
However, the diversity in the characteristics of effluent organic matter (EfOM) and the
complexity of the fouling phenomenon make it difficult to predict the effectiveness of
different pre-treatments. Therefore, this research sought to obtain a more detailed
understanding of the effects of various feed pre-treatment methods on membrane fouling in
MF and UF of secondary effluent, using an activated sludge (AS) effluent in Victoria
(Australia) as an example. This effluent had a brownish yellow colour, which may limit
customers’ willingness to purchase the product water for reuse. The pre-treatments
investigated included coagulation (without removal of the coagulated matter prior to MF or
UF), ozonation, ozonation followed by biological activated carbon (BAC) filtration, and
adsorption with an anion exchange resin (AER) and a powdered activated carbon (PAC).

The research was divided into two parts. The first part dealt with dead-end filtration using
polyvinylidene fluoride (PVDF) and polyethersulfone (PES) membranes. Two PVDF MF (0.1
μm) membranes, one of which was less hydrophobic than the other, and two hydrophobic
PVDF UF membranes of different molecular weight cut-offs (MWCO) (100 kDa and 50 kDa)
were used. A relatively hydrophilic PES UF membrane (MWCO = 30 kDa) was also
investigated. The second part was dedicated to cross-flow filtration with ceramic membranes.
Single-channel tubular membranes made of -alumina (0.1 μm) and zirconia (20 nm) were
used.

The fouling behaviour of the raw (i.e., non-pre-treated) effluent was studied first. Analysis of
the feed and the MF and UF permeates with liquid chromatography with organic carbon
detection (LC-OCD) revealed that biopolymers, which contained proteins and
polysaccharides (molecular weight (MW) >>20,000 Da), and humic substances (HS,
MW=1,000–20,000 Da) were the EfOM components responsible for fouling of all the
membranes investigated. The contribution of biopolymers to fouling was much greater than
that of HS, as indicated by the higher rejection of the former by the membranes. HS caused
hydraulically irreversible fouling, whereas biopolymers contributed to both hydraulically
reversible and irreversible fouling. The roles of biopolymers and HS in causing fouling were

1
further confirmed by the results obtained from the study on the effects of the feed pre-
treatments on the membrane fluxes and membrane cleaning experiments.

Flux data analysis showed that for dead-end MF, fouling by the raw effluent was governed by
pore adsorption and pore blocking during the first 20 minutes of filtration and then followed
the cake filtration model. For the PVDF UF membranes, fouling switched from pore
adsorption and pore blocking to cake filtration after the first 5 minutes of filtration. Fouling of
the PES UF membrane could be described by the cake filtration model over the course of
filtration.

The effect of coagulation on membrane fouling in dead-end filtration with the polymeric
membranes was then investigated. Coagulation with poly(aluminium) chlorohydrate (ACH)
and alum at 2.5 mg Al3+ L-1 significantly improved the permeate flux and reduced
hydraulically irreversible fouling. This was attributed to the coagulants removing the colloidal
organic matter, including some biopolymers and HS responsible for fouling, from the raw
effluent. Unlike MF fouling by the raw effluent, MF fouling by the coagulated effluent
followed the cake filtration mechanism from the beginning of filtration. The MF flux
improvement was greater for alum than ACH, whereas the two coagulants had a similar flux
enhancing effect for UF. Flux improvement by coagulation was also observed in cross-flow
MF tests with the alumina membrane.

Coagulation pre-treatment did not reduce the true colour of the effluent to the desired level
(20–25 Pt-Co units) unless very high coagulant doses (approximately 40 mg Al3+ L-1) were
used. Therefore, ozonation and adsorption pre-treatment with AERs and PAC were
considered. It was found that the colour removal target could be achieved by
(i) Ozonation of the raw effluent at 10 mg O3 L-1 for 10 minutes,
(ii) Pre-adsorption of the raw effluent with an AER at 10 mL L-1 with a 20-minute
contact time (1 bed volume of the resin treated 1500 bed volumes of the raw
effluent), or
(iii) Pre-adsorption with a PAC (Acticarb PS1300) at 150 mg L-1 for 30 minutes.

Dead-end filtration tests with the raw and ozonated effluent (after ozone had decomposed)
showed that ozonation led to significant flux improvement, which was attributed to the
breakdown of some biopolymers and HS to compounds with lower MW and hydrophobicity.

2
However, as this pre-treatment generated some low MW compounds (MW<500 Da), it may
accelerate membrane biofouling. Consequently, BAC filtration of the ozonated effluent was
considered. This additional pre-treatment led to further flux improvement in MF and in UF
with the 100 kDa membrane, but did not affect the flux of the tighter UF (30 kDa) membrane.
The hydraulically irreversible fouling was reduced after ozonation whereas BAC filtration did
not affect this type of fouling. BAC filtration showed potential to reduce biofouling since the
filter removed the majority of the low MW compounds generated during ozonation.

Pre-treatments with the PAC and the AER neither reduced nor worsened membrane fouling in
dead-end filtration. This was due to the fact that although the AER and the PAC removed
more than 50% of HS from the raw effluent, they did not remove biopolymers – the largest
contributors to membrane fouling. Similarly, cross-flow filtration tests showed that while
ozonation improved, pre-adsorption with the AER did not affect the flux performance of the
ceramic membranes.

For cross-flow MF and UF, rapid back-flushing the ceramic membranes at 600 kPa for 1
second every 12 minutes resulted in significant flux improvement. The application of rapid
membrane back-flushing in combination with coagulation or ozonation pre-treatment showed
the ability to maintain high permeate fluxes in cross-flow filtration of secondary effluent.

Chemical cleaning of the 50 kDa PVDF and 30 kDa PES membranes fouled by the raw
effluent was conducted. The most effective cleaners were sodium dodecyl sulphate, Terg-a-
zyme, and NaOCl. The results verified that hydrophobic attraction was the major force
keeping the hydraulically irreversible foulants and the membranes together and that HS
played a role in causing internal adsorptive fouling.

This work showed that only the pre-treatments which led to the removal (coagulation) or
breakdown in the structures (ozonation) of biopolymers reduced membrane fouling.
Therefore, pre-treatment for fouling reduction in MF and UF of secondary effluent should
target the biopolymers (i.e., proteins and polysaccharides). Furthermore, less hydrophobic MF
membranes would likely to benefit more from feed pre-treatments for fouling reduction than
more hydrophobic counterparts, as hydraulically irreversible fouling of the formers is likely to
be less severe and more easily mitigated.

3
 CHAPTER 1. INTRODUCTION

1.1. Background of the project

Australia has experienced long-term and severe droughts over the last decade and is expected
to face these events more frequently in the future (Australian Government - Bureau of
Meteorology 2008). As a result, water authorities are seeking alternative water sources to
meet the increasing demand due to increasing population. It is recognised that reusing
municipal wastewater will significantly increase Australian water availability (Victorian
Government - Department of Sustainability and Environment 2004). The Australian Federal
and State Governments therefore have initiated many water reuse and seawater desalination
projects. One of these was the upgrade of the treatment processes at a municipal wastewater
treatment plant (WWTP) in Victoria. This WWTP treats sewage via 4 stages: preliminary,
primary, and secondary treatments followed by disinfection. A small amount (< 5%) of the
disinfected secondary effluent is used as recycled water, mainly for irrigation and cleaning,
and the rest is discharged to the ocean. The purpose of this upgrade was to enable the WWTP
to produce “Class A” recycled water, which is suitable for use in new housing estates,
agriculture, and industry. It would also help to reduce flow to the ocean outfall and enable the
WWTP to meet increasingly more stringent regulatory requirements on the quality of
discharged water.

Microfiltration (MF) and ultrafiltration (UF) of the secondary effluent to remove suspended
solids (SS) and pathogens were considered as tertiary treatment options for the recycled
water. A major concern with this effluent was its brownish yellow colour, which may limit
customer willingness to purchase and reuse the effluent. Therefore, the WWTP also aimed to
reduce the true colour (i.e., colour after filtration through a 0.45 μm membrane) of the AS
effluent (which can range over 65–120 Pt-Co units) by approximately 75–80%, so that the
final effluent would have a true colour of 15–25 Pt-Co units, which appears almost colourless.

1.2. Rationale
Membrane processes for the treatment of water and wastewater have been widely accepted
due to their high quality product water and affordability. However, for this technology to be
used efficiently, membrane fouling needs to be minimised. Fouling of filtration membranes

4
results in a reduction in water throughput, necessitating process downtime, membrane
cleaning, and more frequent membrane replacement.

The fouling phenomenon is complex and is influenced by several factors, such as feedwater
and membrane characteristics and operating conditions (Mulder 1996). Feed pre-treatment is
commonly practised to reduce membrane fouling. However, the complexity of the fouling
phenomenon makes it difficult to predict the effectiveness of pre-treatment methods on
fouling reduction. For example, both beneficial and adverse effects of coagulation on
membrane fouling have been reported (Adham et al. 2006). Furthermore, whereas secondary
effluents at some WWTPs have a brownish yellow colour (most likely due to the high
concentration of humic substances), information regarding the effectiveness of different pre-
treatment methods for colour removal on fouling reduction in MF and UF of municipal
secondary effluent is scarce.

To date the vast majority of the membranes used in water and wastewater treatment are made
of polymeric materials. However, it has been predicted that ceramic membranes will become
competitive alternatives in the next decade (Pearce 2011). Previous studies on MF and UF of
water and municipal wastewater using ceramic membranes have focussed on the permeate
flux and quality. The effects of feed pre-treatments on membrane fouling have received little
attention. With the growing use of ceramic membranes, it is useful to have more detailed
understanding of their fouling behaviour in these applications.

1.3. Research objectives

The purpose of this study was to identify the components responsible for membrane fouling
and the fouling mechanisms in MF and UF of the brownish yellow secondary effluent
collected from the WWTP mentioned in Section 1.1. The effectiveness of different pre-
treatments on fouling reduction was also investigated. Dead-end filtration with polymeric
membranes and cross-flow filtration with ceramic membranes was examined. This study
focussed on the use of PVDF, PES, alumina, and zirconia membranes, which are the most
widely used membranes in water and wastewater treatment. The pre-treatments under
investigation included:

(i) Coagulation
(ii) Ozonation
(iii) Ozonation followed by biological activated carbon (BAC) filtration

5
 (iv) Adsorption with a powdered activated carbon (PAC), and
(v) Adsorption with an anion exchange resin (AER).

Coagulation was investigated as it is the most widely used pre-treatment for MF and UF of
water and wastewater. The other pre-treatments were considered for their ability to remove
colour-imparting compounds from the secondary effluent and their impact on the filtration
process.

The research was divided into 2 parts. The first part dealt with dead-end MF and UF with the
polymeric membranes. The research questions to be addressed included:
(a) What were the components in the effluent responsible for flux reduction during MF
and UF of the secondary effluent? Of these, which components were responsible for
hydraulically reversible and irreversible fouling?
(b) What were the predominant fouling mechanisms in dead-end MF and UF of the
effluent, as suggested by the flux data and the identity of the foulants?
(c) Which of the pre-treatment methods listed above reduced membrane fouling and what
were the reasons for the fouling reduction? Which methods could reduce the true
colour of the raw effluent to 20–25 Pt-Co units while improving the MF and UF flux?
(d) What were the most effective cleaning agents for the fouled membranes?

The second part was dedicated to cross-flow MF and UF with the alumina and zirconia
membranes with the following research questions:
(a) Were the membrane foulants in cross-flow filtration with the ceramic membranes
different or similar to those in dead-end filtration with the polymeric membranes?
(b) Would coagulation, ozonation and adsorption pre-treatment with the AER improve the
flux performance in cross-flow MF and UF?
(c) Rapid membrane back-flushing with the back-flush device being a piston driven by air
pressure has been used to improve the flux performance in cross-flow MF of water
and wastewater from industrial sources (e.g., yeast suspensions and electroplating
wastewater) with tubular ceramic membranes. Would this technique also be useful in
cross-flow MF and UF of the secondary effluent?

It is expected that the findings will contribute to the knowledge pool in the area of membrane
fouling mechanisms and fouling mitigation. They will help to explain why one feed pre-

6
treatment method is better than others in terms of fouling reduction and hence will assist
WWTPs in the selection of the pre-treatment methods suitable for their purposes.

7
 CHAPTER 2. LITERATURE REVIEW

2.1. Overview of conventional municipal wastewater treatment processes

The treatment of municipal wastewater is typically accomplished via 4 stages: preliminary,
primary, secondary, and tertiary.

2.1.1. Preliminary treatment

The purpose of this treatment is to remove large solids and grit from the wastewater to
prevent damage to the rest of the unit operations. This is usually accomplished by screening.
Coarse screens with openings of 6 mm or larger are used to remove large solids and rags,
whereas fine screens may be used to remove smaller particles (Rowe & Abdel-Magid 1995;
Vesilind 2003; Hammer & Hammer 2008).

2.1.2. Primary treatment

Primary treatment aims to reduce the SS level of the wastewater and to remove floating
materials such as fat, oil, and grease. It is usually done using settling tanks (or clarifiers),
wherein suspended particles heavier than water settle by gravity under quiescent conditions
and floating materials are removed by skimming. The settling of particulates can be enhanced
by pre-aeration or chemical coagulation (Vesilind 2003).

2.1.3. Secondary treatment

This involves a biological treatment process in which micro-organisms consume dissolved

organic matter (DOM) in wastewater to grow and reproduce, hence reducing the biological
oxygen demand (BOD) of the wastewater (Vesilind 2003). Secondary treatment can be done
using fixed film, suspended, or lagoon systems.

In fixed film systems, microbes grow on fixed objects (such as rock, sand, or plastic) and are
exposed repeatedly to the wastewater for adsorption of organic material and to the atmosphere
for oxygen (Vesilind 2003). In suspended systems (e.g. activated sludge), microbes are
suspended in wastewater by agitating the wastewater for several hours after which the
biomass is settled out as sludge in a secondary clarifier. Lagoons are shallow basins which

8
hold the wastewater for several months to allow natural degradation of sewage (Vesilind
2003).

2.1.4. Tertiary treatment

Secondary effluent may contain contaminants (e.g., nitrogen and phosphorus species,
disinfection by-products, colour, pathogens, metals and salts) at levels higher than regulatory
requirements and therefore require further treatment before being discharged to local waters
or recycled. These treatment processes can be physical (e.g., membrane filtration and granular
media filtration), biological (e.g., oxidation ponds), or chemical (e.g., precipitation of
phosphorus, iron and manganese) (Rowe & Abdel-Magid 1995; Vesilind 2003).

2.2. Membrane filtration in water and wastewater treatment

Nowadays, a variety of membrane processes (MF, UF, nanofiltration (NF), and reverse
osmosis (RO)) have been used to treat water and wastewater at many facilities worldwide.
Examples of municipal wastewater treatment plants using membrane filtration to treat
secondary effluent include West Basin Water Recycling Plant in Los Angeles, California
(design flow = 45420 m3/day), Chandler Wastewater Reclamation Plant in Chandler, Arizona,
Scottsdale Water Campus in Scottsdale, Arizona (design flow = 45420 m3/day), the NEWater
plants in Singapore, and the three advanced wastewater treatment plants at Bundamba,
Luggage Point, and Gibson Island of the Western Corridor recycled water project in
Queensland, Australia. In these treatment plants, MF or UF is used as a pre-treatment for NF
or RO.

There has been growing interest in the integration of membrane processes into existing water
and wastewater treatment facilities for various reasons, such as to meet more stringent
regulatory requirements for reuse and discharge and/or increase plant capacity (Water
Environment Federation 2006). The advantages of membrane filtration over conventional
filtration using granular media include superior quality of product water, small size, low
energy consumption, easier maintenance and capacity extension, and the capability of
handling wide fluctuations in feedwater quality (Clever et al. 2000).

MF and UF membranes have also been used widely in membrane bioreactors (MBRs). In
MBRs, the activated sludge process and membrane filtration are used in combination to

9
achieve biological treatment and solids/liquid separation (Le-Clech et al. 2006). As the focus
of this study is on membrane fouling in MF and UF of secondary effluent, membrane fouling
in MBRs will be briefly discussed in this literature review.

2.2.1. Membrane classification

Membranes are usually classified according to their pore size, the molecular weight (MW) of
the rejected material, the applied pressure, and their water affinity. The pore size, MW cut-off
(MWCO), and operating pressure of the 4 different types of membranes are given in Table
2.1.

MF membranes are capable of removing SS, some protozoan cysts (e.g., the pathogens
Giardia and Cryptosporidium), and bacteria (3–6 log removal). UF can remove some
colloidal materials, dissolved macromolecules, and some viruses. NF and RO membranes are
capable of removing dissolved solids (including ions); the separation capability of NF
membranes lies between that of UF and RO membranes (Stephenson et al. 2000). MF and UF
are conducted at much lower transmembrane pressure (TMP) than NF and RO and therefore
are often called low pressure membrane filtration.

Table 2.1 Some characteristics of different membrane types

(Water Environment Federation 2006)
MF UF NF RO
MWCO (Da) > 100,000 > 2,000–100,000 300–1000 100–200
Pore size (μm) 0.1–10 0.01–0.1 0.001–0.01 < 0.001
Operating pressure (psi) 1 – 30 3 – 80 70–220 800–1200
Operating pressure (kPa) (7–207) (20–550) (480–1500) (5500–8250)

Based on separation mechanism, filtration membranes can be categorised as either dense (NF
and RO membranes) or microporous (MF and UF membranes) (Mulder 1996). The separation
mechanism of MF and UF membranes is size exclusion (or sieving), i.e., components larger
than the pore size are prevented from permeating through the membrane. For NF and RO
membranes, besides size exclusion, separation also relies on the large differences in the
solubility and diffusion rates of water and solutes. The much slower diffusion rate of solutes
than water across the membranes allows the production of a solute-free permeate stream

10
(Bellona et al. 2004; Water Environment Federation 2006). Based on water affinity,
membranes can be divided into 2 groups, namely hydrophobic membranes and hydrophilic
membranes (Mulder 1996).

2.2.2. Membrane materials and structures

Membranes can have symmetric or asymmetric structures. Symmetric microporous

membranes are cast from one material and their pore size can be either uniform (isotropic) or
varied (anisotropic). For anisotropic membranes, the surface having the smaller pore size is
used as the filtering surface. Integral asymmetric membranes (or “skinned” membranes)
consist of a very thin (<1 µm) layer (skin) and a thicker (up to 100 µm) porous layer that adds
support and is capable of high water flux. A more recent development is thin-film composite
(TFC) membranes, made by bonding a thin layer of cellulose acetate, polyamide or other
acetate-containing compounds to a thicker porous substrate (Water Environment Federation
2006).

Filtration membranes available on the market are produced from a diverse range of materials,
which can be organic or inorganic. The most widely used organic materials include
polyvinylidene fluoride (PVDF), polysulfone (PS), polyethersulfone (PES), polyacrylonitrile
(PAN), polypropylene, cellulose acetate, polyamide, and polytetrafluoroethylene (PTFE). PS
and PES have been used extensively as the base material for the majority of commercial
membranes used in water treatment plants in the United States (Adham et al. 2006). Each
membrane material has its advantages and disadvantages in terms of costs, thermal stability,
chemical resistance, and biodegradability (Stephenson et al. 2000; Water Environment
Federation 2006).

Polymeric MF membranes can be produced from PTFE, PVDF, polypropylene, polyethylene,

cellulose esters, polycarbonate, PS, PES, polyimide, poly (ether imide), aliphatic polyamide,
and polyetheretherketone (PEEK). The first 4 polymers are hydrophobic whilst the remainder
are hydrophilic (Mulder 1996). Different from MF membranes, which can be symmetric or
asymmetric, UF, NF and RO membranes generally have an asymmetric structure with a much
denser top layer and consequently require a higher applied pressure to overcome their higher
membrane hydraulic resistance. UF membranes are normally made from PS, PES, sulfonated
PS, PVDF, PAN (and related block-copolymers), cellulosics (e.g., cellulose acetate), aliphatic

11
polyamide, poly (ether imide), and PEEK. Typical materials used in the production of RO
membranes include cellulose esters (most commonly cellulose diacetate and cellulose
triacetate), aromatic polyamide, and polybenzimidazoles (Mulder 1996).

Ceramic membranes, the majority of which are made from aluminium oxide (alumina, Al2O3),
titanium oxide (titania, TiO2), and zirconium oxide (zirconia, ZrO2), are also available.
Ceramic membranes generally have better thermal, chemical, and mechanical resistances, and
longer service life than their polymeric counterparts (up to 15 years compared to 3-5 years)
(Owen et al. 1995; Heijman & Bakker 2007). However, they require careful handling as the
material is brittle. Unlike the polymeric membranes, the high thermal and chemical
resistances of ceramic membranes make it possible to clean the fouled membranes at high
temperatures and with high concentrations of cleaning agents, therefore complete flux
recovery can be achieved with simple chemical cleaning (Bottino et al. 2001) with a lower
risk of membrane degradation. The high chemical resistance also allows ceramic membranes
to treat ozonated water without being damaged by high residual ozone concentrations
(Schlichter et al. 2004). Another advantage of ceramic membranes compared with polymeric
membranes is their higher operating fluxes (Heijman & Bakker 2007). Furthermore, with
ceramic membranes it is possible to incorporate oxidation processes, such as UV irradiation,
with filtration to remove contaminants in water and wastewater. In these applications,
titanium oxide acts as a photocatalyst which initiates the oxidation of contaminants (Djafer et
al. 2010). Ceramic membranes have been used widely in the food, beverage, and dairy
industries and in the treatment of industrial wastewater (Daufin et al. 2001; Finley 2005).
However, because they are more expensive than polymeric membranes, their use in water and
municipal wastewater treatment is still rather limited. Interestingly, they have been used
widely in Japan for the production of drinking water. Recently, decreasing cost has led to a
growing interest in the use of ceramic membranes in the water industry (Ciora & Liu 2003;
Heijman & Bakker 2007; Mueller & Witte 2008; Pearce 2011).

2.2.3. Membrane configurations

Membrane geometry (or configuration) and the way in which the individual membranes are
housed to produce modules are very important factors affecting the overall process
performance. Common membrane configurations in wastewater treatment include plate-and-

12
frame, spiral-wound cylinder, hollow-fibre, tubular, and rotating flat-plate (Stephenson et al.
2000).

The plate-and-frame configuration consists of a series of flat membrane sheets and support
plates. Many membrane sheets are welded to a support plate on both sides, forming a
membrane cartridge. These cartridges are often housed in a container (cassette) in which they
are slid into grooves for support. The cartridges are typically spaced 10 mm apart. Feedwater
is filtered as it flows between and parallel to the cartridges (Water Environment Federation
2006).

The spiral-wound configuration consists of several flat membrane sheets, each of which is
sealed on three sides and the open side is glued to a perforated plastic collection tube. A
flexible permeate spacer is placed between two successive membrane sheets. Flow through
the spiral-wound configuration is outside-in (Stephenson et al. 2000).

In the hollow-fibre configuration, millions of fibres are looped into a U-shape and the ends
are potted into a special epoxy resin which serves as a tube sheet (Vesilind 2003). The
preferred direction of water flow in hollow-fibre systems is “outside in”, i.e., the feedwater
enters the fibres from the fibre external wall and the permeate flows inside the fibre lumen
and is collected at the fibre end. This configuration can be either immersed (as in membrane
bioreactors) or contained in pressure vessels, which isolate the feedwater and the permeate.
With its high packing densities and low pressure requirements, which tend to result in low
life-cycle cost, this configuration has the potential to become the most common membrane
configuration for large municipal wastewater treatment facilities (Water Environment
Federation 2006).

In the tubular configuration, the membranes are cast on a support tube which is contained in a
pressure vessel. Membranes of this configuration are usually ceramic run in cross-flow mode
where the feedwater flows axially within the tube, while the permeate flows radially through
the porous structure of the tube wall. The permeate is collected outside the tube while the
concentrate continues to flow through it (Water Environment Federation 2006). Dead-end
filtration using monolith type tubular membranes has also been practised (Heijman & Bakker
2007; Mueller & Witte 2008).

13
2.2.4. Operation modes of membrane processes

Membrane filters may be operated in either dead-end or cross-flow mode, as shown in Figure
2.1 (Mulder 1996). In the “dead-end” mode, the pressure driving the liquid flow is applied
perpendicular the membrane surface.
In cross-flow filtration, the feedwater is forced to flow at fairly high velocity in the direction
tangential to the membrane surface. A thin cake layer may form due to the deposition of the
solutes in the feedwater on the membrane. High liquid velocity is used to wash away some of
the rejected materials from the cake layer, preventing it from building up and therefore
reducing fouling (Mulder 1996).

Filtration can be operated in either constant-pressure mode or constant-flux mode (Howell et

al. 2002). Constant-flux operation requires changing the applied pressure to maintain a
constant permeate flux.

Feedwater

Retentate
Feedwater

Membrane Membrane

Permeate Permeate

(a) (b)

Figure 2.1. Schematic presentations of (a) dead-end mode and (b) cross-flow mode
(Mulder 1996)

2.2.5. Feedwater for membrane processes for water reuse

In WWTPs, the feedwater for membrane processes is usually secondary effluent. The organic
content of this effluent is commonly known as effluent organic matter (EfOM), which
contains various complex and heterogeneous compounds. EfOM can be classified into 3
different groups according to their origins (Drewes & Fox 1999):
(i) Refractory natural organic matter (NOM) originating from drinking water,

14
 (ii) Synthetic organic compounds (SOC) generated during domestic use and
disinfection by-products (DBPs) produced during water and wastewater treatment,
and
(iii) Soluble microbial products (SMP) derived from the biological processes of
wastewater treatment. Compounds in SMP include polysaccharides, proteins,
aminosugars, nucleic acids, extracellular enzymes, antibiotics, steroids, and
structural components of cells (Barker & Stuckey 1999).

To reduce membrane fouling the secondary effluent is usually pre-treated to remove the
foulants. The most effective pre-treatment method varies with the characteristics of the
secondary effluent.

Different from MF and UF of secondary effluent, in which the feedwater contains EfOM and
a relatively small amount of SS (normally less than 25 mg L-1), the feed to the filtration
process in MBRs is a mixed liquor containing (Jiang et al. 2003; Le-Clech et al. 2006; Meng
et al. 2009):
(i) The activated sludge (mixed liquor suspended solids, MLSS) with high
concentrations of SS (3000–31000 mg L-1 (Brindle & Stephenson 1996)),
(ii) Bound extracellular polymeric substances (bound EPS, consists of
polysaccharides, proteins, nucleic acids, lipids and humic acids which adhere to
the surface of microbial cells in the sludge), and
(iii) Colloidal and soluble organic matter, including soluble EPS (i.e., SMP, originating
from substrate metabolism of micro-organisms).

15
2.3. Membrane fouling and factors affecting fouling
Membrane fouling can be defined as the build-up of contaminants on the membrane which
causes an increase in the TMP requirement (in constant-flux filtration) or a reduction in the
permeate flux through the membrane (in constant-pressure operation). It can take place at the
surface (macro-fouling) or inside the pore (pore fouling or micro-fouling).

There are many ways in which a membrane is fouled, including (Barger & Carnahan 1991;
Caothien et al. 2003):
 Deposition of particulates and suspended solids on the membrane (particulate fouling).
This type of fouling can be mitigated by membrane backwashing and/or air scrubbing;
 Precipitation of DOM on the membrane surface or in membrane pores (organic
fouling);
 Accumulation of biological growth in the system and/or its attachment to the
membrane (biofouling). Extracellular polymeric substances (EPS) produced by the
attached micro-organisms can form a viscous slimy gel, which retards the permeate
flow;
 Physical or chemical reactions between some components of the feedwater and the
membrane surface.

The major cause of fouling of MF and UF membranes is organic fouling, resulting from the
accumulation of DOM in the feedwater in the membrane pores or on the membrane surface.
Biofouling may occur in long-term operation. Membrane fouling in MBRs is more
complicated than that in MF and UF of secondary effluent, due to the presence of activated
sludge and bound EPS in the mixed liquor (Le-Clech et al. 2006).

Fouling is normally classified as reversible and irreversible fouling. Foulants in hydraulically

reversible fouling generally exist as a cake layer on the membrane surface and can be
removed by hydraulic cleaning such as backwashing the membrane with the permeate.
Irreversible fouling is probably caused by the adsorption or pore plugging of solutes in and
within the membrane pores (Zularisam et al. 2006). Chemically irreversible fouling is the
proportion of fouling that cannot be removed by chemical cleaning of the membrane.

16
The nature and extent of fouling is affected by interactions of many factors, including the
characteristics of the feedwater and the membranes, membrane configuration, and operating
conditions.

2.3.1. Effect of feedwater content

Dissolved organic matter, inorganic materials (such as silica, alumino-silicates, iron, and
aluminium), SS, and micro-organisms (bacteria and fungi) are potential foulants in membrane
filtration of surface water and wastewater.

The fouling potential of DOM is governed by solution chemistry, membrane characteristics,

and operational conditions. Feedwaters with higher DOM concentrations do not always cause
more severe fouling, indicating that some organic fractions have a higher fouling potential
than others (Water Environment Federation 2006; Her et al. 2008). However, there have been
disagreements about which component is mainly responsible for membrane fouling.

EfOM can be separated into 3 fractions: hydrophobic, transphilic, and hydrophilic fractions.
Humic substances (HS), the major component of NOM, are predominantly hydrophobic and
generally divided into 3 fractions: humic acid, fulvic acid, and humin. Humic acid and fulvic
acid are anionic polyelectrolytes with negatively charged carboxyl (COOH), methoxyl
carbonyl (C=O), and phenolic hydroxyl and alcoholic hydroxyl (OH) functional groups.
Humic acid is not soluble in water at pH below 2 but becomes soluble at higher pH. Fulvic
acid is water soluble whereas humin is water insoluble at all pH values (MacCarthy & Suffet
1989). Humic acid contains fewer carboxylic and hydroxyl functional groups and is more
hydrophobic than fulvic acid. The MW of aquatic humic acid is between 2000 Da and 5000
Da whereas MW of fulvic acid is generally less than 2000 Da (Sutzkover-Gutman et al.
2010). Humic acid is therefore considered to be colloidal. There are more colour centres on
humic acid molecules than on fulvic acid molecules (Vik & Eikebrokk 1989).

Many researchers investigated the fouling potentials of the fractions extracted from bulk
NOM and EfOM solutions (using resin fractionation) and found that the hydrophilic (non-
humic) components, which contained polysaccharide-like and protein-like compounds, were
the major foulants in membrane filtration (Jarusutthirak et al. 2002; Lee et al. 2004)
Jarusutthirak and Amy (2006) suggested that these foulants originated from SMP generated

17
during the biological processes of wastewater treatment. Fan et al. (2001) rated the fouling
potential of NOM fractions on hydrophobic PVDF MF membranes as hydrophilic neutral >
hydrophobic acids > transphilic acids > hydrophilic charged. These findings were supported
by more recent studies which analysed the feeds and the permeates in dead-end UF of EfOM
with size exclusion chromatography with organic carbon detection and found that the
components retained by the membranes were protein-like and polysaccharide-like compounds
(Laabs et al. 2006; Haberkamp et al. 2007; Henderson et al. 2011).

Nevertheless, other researchers found that the hydrophobic fraction of DOM (predominantly
HS) was the most important component contributing to fouling (Schäfer 1999; Shon et al.
2006). The high aromaticity, adsorptive capacity, hydrophobicity, and high molecular weight
of humic acid were believed to be responsible for its higher tendency to deposit on membrane
surface (Zularisam et al. 2006). Aoustin and co-workers (2001) studied the fouling of UF
membranes and found that the humic acid fraction was responsible for irreversible pore
adsorption and plugging, whereas fulvic acid and the hydrophilic NOM caused a smaller and
mostly reversible flux decline. Combe et al. (1999) and Yuan and Zydney (1999; 2000) also
indicated that humic acid was an important foulant in MF and UF. Recently, Kim and
Dempsey (2010) found that organic acids (which probably contained humic acid, as indicated
by their high UVA254), fouled MF and UF membranes more severely than any other
components (including colloidal organic matter and hydrophilic bases/neutrals) in a municipal
wastewater. The contradictory findings mentioned above could be attributed to differences in
water sources and membrane characteristics. They also illustrate the need for identifying the
foulants for a particular wastewater and membrane type.

Braghetta et al. (1998), on the other hand, suggested that interactions between hydrophobic
and hydrophilic components might have significant influence on the foulant behaviour of the
whole NOM solution. Therefore, fractionation of NOM into hydrophobic and hydrophilic
fractions might not be a meaningful tool for investigation of foulant materials. This was
supported by the study conducted by Jermann et al. (2007), which revealed that alginate (a
polysaccharide) and humic acid acted synergistically, causing more severe hydraulically
irreversible fouling on a relatively hydrophobic PES 100 kDa UF membrane than did the
individual components. The flux decline caused by the polysaccharide was much more severe
than by the humic acid. However, whereas most of the fouling by alginate was reversible (i.e.,
could be removed by backwashing the fouled membrane), a significant proportion of fouling

18
by humic acid was irreversible. The more severe irreversible fouling by humic acid than
alginate was attributed to adsorption of humic acid on the membrane. The fact that most
alginate retained on the membrane could be removed by backwashing, however, appears to
contradict the result obtained by Kimura et al. (2004), who found that the polysaccharide-like
compounds were responsible for irreversible fouling in UF of Chitose River water (Japan)
with a 750 kDa PS membrane. Fan et al. (2001) also reported that the hydrophilic neutral
fraction, which contained polysaccharide-like and protein-like compounds, caused more
severe irreversible fouling than the hydrophobic acid (i.e., humic acid) fraction in MF of
fractionated surface waters.

The presence of inorganic particles in the feedwater was found to affect the fouling
behaviours of organic substances. Inorganic particles such as clay minerals competed with
organic components to adsorb onto the membrane surface or in the pores. On the other hand,
the high surface area of clay minerals might facilitate the adsorption of DOM on the inorganic
layer. This phenomenon could have great effect on fouling characteristics. It could result in
enhanced deposition of inorganic particles on the membrane and a reduction in the sorption of
DOM onto the membrane, which increases membrane permeability (Zularisam et al. 2006).
This hypothesis is supported by the studies conducted by Mallevialle et al. (1989) and Kaiya
et al. (1996). In their study, Mallevialle and co-workers noticed a fouling layer consisting of
an organic matrix that served as cement for inorganic constituents. Kaiya et al. (1996)
analysed the fouling layer formed on a PP MF hollow-fibre membrane during the filtration of
Lake Kasumigaura (Japan) water and found that organic matter and manganese were the
predominant foulants.

The combined effect of organic and inorganic matter was also observed by many other
researchers. For example, Schäfer et al. (2000) found that the precipitation of calcium-organic
complex was the primary reason for MF membrane fouling caused by NOM extracted from a
surface water. Mo and Huang (2003) investigated the foulant characteristics in a coagulation-
MF combination process used for the treatment of a polluted river water and realised that
fouling on the inner surface was mainly due to biofouling, whereas fouling on the exterior
surface could be attributed to a combined effect of micro-organisms, organic (mainly low
MW molecules), and inorganic matter (mainly calcium). It is generally understood that the
presence of Ca2+ at sufficient concentrations in HS solutions induced more rapid fouling as
the fouling layer formed was dense and had low water permeability (Al-Amoudi et al. 2007).

19
Researchers attributed this rapid fouling to Ca2+ binding the negatively charged HS molecules
to form aggregates and/or bridging these molecules with the negatively charged membrane
surface (Hong & Elimelech 1997; Cho et al. 2000; Ahn et al. 2008).

For MBRs, as the feed to the filtration process is the mixed liquor, membrane fouling is
affected by the characteristics of the sludge (e.g., MLSS concentration and particle size,
viscosity, and temperature) as well as the characteristics (concentration and composition) of
bound EPS and soluble EPS (Meng et al. 2009). Contradictory findings have been reported on
the effect of MLSS concentration on membrane fouling in MBRs. Many studies found that
membrane fouling increased as the MLSS concentration was increased (Yamamoto et al.
1989; Cicek et al. 1999; Chang & Kim 2005; Yigit et al. 2008). However, others reported the
opposite trend (Defrance & Jaffrin 1999; Brookes et al. 2006) or insignificant impact of
MLSS concentration on membrane fouling (Hong et al. 2002; Le-Clech et al. 2003; Lesjean
et al. 2005).

Bound EPS has been recognised as the major components keeping activated sludge flocs in a
three-dimensional matrix. Some researchers found that bound EPS made a significant
contribution to membrane fouling in MBRs (Cho et al. 2005; Ahmed et al. 2007). However,
others showed that these components had no effect on membrane fouling (Rosenberger &
Kraume 2003). Although different findings on the importance of bound EPS in fouling of
MBRs have been reported, Meng et al. (2009) suggested that fouling mitigation in MBRs
should focus on controlling these components (apart from SMP, as discussed in the next
paragraph), since the concentration of bound EPS is closely related to many characteristics of
activated sludge which have significant impacts on MBR fouling (e.g., sludge volume index,
viscosity, hydrophobicity, surface charge and flocculation ability).

Several MBR studies have shown that SMP, particularly polysaccharide-like and protein-like
compounds, played a much more important role than the MLSS and bound EPS in causing
membrane fouling (Fawehinmi et al. 2004; Evenblij et al. 2005; Lesjean et al. 2005;
Rosenberger et al. 2005; Ji & Zhou 2006; Rosenberger et al. 2006; Iritani et al. 2007). Some
studies have also demonstrated that the polysaccharide-like components in SMP made a larger
contribution to fouling than protein-like compounds (Rosenberger et al. 2006; Yigit et al.
2008). The role of HS in fouling in MBRs has been less well studied, as researchers have
tended to focus on EPS (Le-Clech et al. 2006). Lyko et al. (2007) indicated that HS

20
contributed to membrane fouling by forming complexes with metal cations which deposited
on the membrane surface. Hong et al. (2012) analysed the foulants in MBRs using
fluorescence spectroscopy and liquid chromatography with organic carbon detection (LC-
OCD) and found that the relative proportions of biopolymers and HS varied with the
hydraulic retention time. Operating MBRs at a longer hydraulic retention time (4 hours)
increased the accumulation of biopolymers in the foulant layer, whereas at shorter hydraulic
retention times (1–2 hours), HS was found to be the dominant components of the foulant
layer.

It is noteworthy to mention that despite significant differences in the characteristics of surface

water, secondary effluent, and MBR mixed liquor, the majority of recent studies on
membrane fouling have shown that polysaccharide-like and protein-like compounds are the
major organic foulants in MF and UF of these feedwaters.

So far the vast majority of the investigations into the identity of the foulants have been
conducted in dead-end filtration mode using polymeric membranes. Intuitively, it can be
inferred that the components of the DOM responsible for fouling in dead-end filtration would
also be the membrane foulants in cross-flow filtration with ceramic membranes. Previous
studies, however, have not attempted to validate this inference, particularly with municipal
secondary effluents. Information regarding the effectiveness of feed pre-treatments on fouling
reduction in cross-flow MF and UF of municipal wastewater with ceramic membranes is also
scarce.

2.3.2. Effect of solution chemistry

Solution chemistry (pH, ionic strength, concentration of multivalent cations, and water
hardness) was found to contribute significantly to the nature and extent of fouling. It greatly
affected the structure and hydraulic resistance of the foulant layer via controlling the charge
and configuration of organic macromolecules (Seidel & Elimelech 2002). It was observed that
HS tended to aggregate more at low pH and high multivalent cation (particularly Ca2+)
concentrations. At neutral pH and low ionic strength, the molecules tended to stretch to more
linear forms (Ghosh & Schnitzer 1980).

21
Several studies revealed that membrane fouling by NOM was intensified in solutions of low
pH, high ionic strength and high divalent cation concentration. This phenomenon could be
explained as due to changes in intra- and inter-molecular electrostatic interactions between
organic molecules, particularly those containing negatively charged functional groups such as
carboxylic, phenolic and carbonyl. It is believed that increasing ionic concentration shielded
the charges on solute molecules and facilitated their coiling and aggregation, which resulted
in the accumulation of these molecules on the membrane surface (Zularisam et al. 2006).

Based on the UV absorbance at 254 nm (UVA254) of surface waters at different pH values,

Maartens et al. (1999) suggested that low pH favoured the precipitation of the humic acid and
fulvic acid fractions of NOM since in such conditions the carboxyl (COOH) groups of these
fractions were protonated, forming large and less soluble complexes.

Interactions among feedwater components, pH, and solution chemistry were also observed by
Adham et al. (2006). These researchers carried out experiments on 5 different river waters
and found that no single water quality parameter, such as pH, alkalinity, hardness, total
organic carbon (TOC) concentration, UVA254, specific UV absorbance (SUVA), or silica
concentration could be correlated alone to the extent of fouling. They also noticed that the
most important foulants were larger than 100 kDa and some were smaller than 10 kDa. There
was little or no foulant material between 10 and 100 kDa. Analysing the foulants at Waikato
Water Treatment Plant (New Zealand), the researchers concluded that silica at high
concentrations in raw water could combine with aluminium from coagulants to form kaolinite
(Al2Si2O5(OH)4), a major inorganic foulant (Adham et al. 2006).

2.3.3. Effect of membrane characteristics

Membrane characteristics such as MWCO, pore size, surface charge and

hydrophobicity/hydrophilicity are important factors affecting the rejection, types of retained
solutes, rate of flux decline, and fouling mechanisms.

Electrostatic repulsion and hydrophobic attraction are the two major forces controlling the
interaction between NOM/EfOM and membranes in MF and UF. Electrostatic repulsion
between DOM and the membrane surface is the result of the likeness in the surface charge of
these two constituents. In the pH range of typical natural water and wastewater (pH 5–8),

22
most membranes and DOM components carry net negative surface charges, and therefore
DOM and membranes tend to electrostatically repel each other (Liu et al. 2001). For this
reason, Cho and co-workers (2000) suggested that membranes with negatively charged
surfaces should be selected to reduce fouling caused by DOM.

On the other hand, hydrophobic attraction is the attraction between molecules with similar
chemical structures as a result of van der Waals force between these molecules. As a general
rule, organic compounds with less than one charged functional group per every twelve carbon
atoms have a great tendency to adhere to hydrophobic surfaces (such as hydrophobic
membranes), since the energy of hydrophobic attraction is greater than the electrostatic
repulsion energy (Liu et al. 2001). HS in water and wastewater is hydrophobic and therefore
tends to have strong adhesion to hydrophobic membranes (Zularisam et al. 2006). In general,
hydrophobic attraction (and therefore fouling propensity) tends to increase with increasing
MW and mass to charge ratio of DOM components (Liu et al. 2001). Similarly, previous
studies found that in MF and UF of the same DOM source, hydrophobic (e.g., unmodified
PVDF, PS, and PES) membranes had a higher fouling tendency than hydrophilic (e.g.,
cellulose-based and hydrophilised PVDF and PES) membranes due to their higher DOM
adsorption (Mulder 1996; Nakatsuka et al. 1996; Fan et al. 2001; Gray et al. 2008).
Hydrophobic attraction between DOM and membranes has also been found to be greater at
high ionic strength and high divalent cation concentrations. It is hypothesised that high ionic
strength of water/wastewater causes the compression of the “double electric layer” of
colloidal organic matter, which in turn reduces its repulsion from negatively charged
membranes. Divalent cations (e.g., Ca2+) can act as a “salt bridge” between negatively
charged DOM and negatively charged membranes by charge neutralisation and therefore
increase the tendency of membrane fouling (Liu et al. 2001).

The adsorption of DOM on the membrane surface can also alter the hydrophilicity and surface
charge of membranes. The surface of hydrophilic membranes became more negatively
charged as a result of rapid adsorption of HS (Childress & Elimelech 1996).

2.3.4. Effect of membrane configuration

Membrane configuration is a critical factor affecting the performance of the filtration process.
Configurations that are easy to clean and create high turbulence in the water flow are better in

23
combating fouling. High turbulence promotes mass transfer on the feed side and therefore
reduces the accumulation of matter near the membrane surface.

Other selection criteria for a good membrane configuration include (Stephenson et al. 2000):
o High membrane area to module bulk volume ratio,
o Easy to modularise,
o Low cost per unit membrane area, and
o Low energy consumption per unit volume of product water.

2.3.5. Effect of operating conditions

Operation in “dead-end” mode tends to result in rapid fouling due to the fast build-up of a
cake layer of retained materials. In cross-flow mode, the build-up of the cake layer is slowed
by the high-velocity liquid flow.

Operating conditions such as applied pressure, cross-flow velocity, and backwash frequency
also greatly affect the rate and extent of fouling (Cheryan 1986; Crozes et al. 1997;
Sheikholeslami 1999). The starting flux in MF and UF also affects the rate of flux decline. It
has been found that operating at higher fluxes resulted in more rapid fouling (Field et al.
1995).

2.4. Resistances in membrane filtration and fouling models

The total hydraulic resistance in membrane filtration includes resistances exerted by the
membrane, pore blocking, pore adsorption, the cake layer, and by concentration polarisation
(Van den Berg & Smolders 1990). The membrane hydraulic resistance is a membrane
constant and is independent of the feed composition and applied pressure. Pore blocking may
occur in porous membranes, when solutes of the same size as the membrane pore block the
pore entrance. Adsorption of solute molecules on the membrane surface or within membrane
pores, if it takes place, also contributes to the total resistance (Mulder 1996). Cake formation
occurs when the feed contains particles larger than the membrane pores. Pore blocking, pore
adsorption, and cake formation are considered to be the 3 mechanisms of membrane fouling
caused by DOM, whereas concentration polarisation, although causing flux decline, is not
considered as a fouling mechanism. This will be discussed further in Section 2.4.3.

24
2.4.1. Fouling models for dead-end filtration

Membrane fouling during constant-pressure unstirred dead-end MF and UF is thought to

proceed through successive or simultaneous steps as follows (Bowen et al. 1995):
(i) Particles reaching the membrane block the smallest pores
(ii) Particles cover the inner surfaces of bigger pores
(iii) Some newly arrived particles attach to the particles that are already on the
membrane whereas others directly obstruct some of the membrane pores
(iv) Cake begins to form.

The general empirical equation for these steps is (Bowen et al. 1995):
n
d 2t  dt 
2
 k  (2.1)
dV  dV 
where
V is the total permeate volume collected at time t (m3),
k and n are constants depending on the mechanism involved.
The following models represent the integrated form of this equation for different values of n .

2.4.1.1. Complete blocking filtration model ( k  k CB , n = 2)

This model is based on the assumption that particles reaching the membrane seal the pores but
do not accumulate on each other. This situation often occurs when the particle size is of the
same order as the pore size.
The linear form of the complete pore blocking is
V  V0  k CBV (2.2)
where
V is the permeate volumetric flow rate at time t (m3 s-1),
V0 is the permeate volumetric flow rate at time t = 0, which can be estimated by filtration
with pure (DOC-free) water, and
k CB is the constant for the complete blocking models (s-1)

25
2.4.1.2. Intermediate blocking model (long-term adsorption) ( k  k IB V0 , n = 1)

This model assumes that new particles can attach to particles that have already blocked some
pores or they can directly block some of the pores. It is described by the following equation:
1 1 k IB
  t (2.3)
V V0 V0

where k IB is the constant for the intermediate blocking model (s-1)

2.4.1.3. Standard blocking model (direct adsorption) ( k  2k SB V0 ) , n = 3/2)

This model is also called the “pore adsorption” or “pore constriction” model. It is derived for
the case in which the pore size is reduced by particles depositing on the internal pore wall.
This situation is likely to occur when the particle size is much smaller than the membrane
pore size.
The linear form of standard pore blocking model is
t k SB 1
 t (2.4)
V V0 V0

where k SB is the constant for the standard blocking model (s-1).

2.4.1.4. Cake filtration model ( k  k CF 2V0 , n = 0)

2

This model accounts for the case in which particles are large and cannot enter most pores or
all membrane area is already blocked. Particles are retained by sieving and form a cake on the
membrane surface.
The linear form of cake filtration model is
t k 1
 CF2 V  (2.5)
V 4V0  V0

where k CF is the constant of the cake filtration model (s-1)

A schematic presentation of the above fouling models is shown in Figure 2.2.

26
 Particles in
feed solution

Membrane
pore wall

(a) Complete blocking (b) Intermediate blocking

(c) Standard blocking (d) Cake filtration

Figure 2.2. Schematic presentation of different membrane fouling models

(Bowen et al. 1995)

Bowen et al. (1995) hypothesised that where the pore size was uniform and larger than the
molecules to be deposited, membrane fouling would start with a standard blocking step
followed by complete blocking, intermediate blocking, and cake filtration. If the molecules
were much larger than the pores, fouling would begin with complete blocking followed by
intermediate blocking and then cake filtration.

Some researchers found that pore blocking, pore adsorption, and cake formation were
responsible for fouling of MF membranes. For UF membranes, flux decline and enhanced
rejection were attributed to internal pore adsorption of organic components in the feedwater
(Zularisam et al. 2006). It was suggested that the initial rapid flux decline in MF and UF of
biologically treated sewage effluent could be attributed to pore blocking and adsorption of

27
EfOM onto the membrane matrix, whereas the subsequent slow flux decline was caused by
cake formation on the membrane surface (Lim & Bai 2003; Kim et al. 2005).

2.4.2. Fouling models for cross-flow filtration

The fouling models developed originally for dead-end filtration were modified to apply to
cross-flow filtration (Field et al. 1995). This was done by adding terms representing the
removal rate of particles from the membrane surface due to the cross-flow. The development
of these models also led to the concept of critical flux, which is an important parameter in
membrane filtration practice. Critical flux is the flux above which fouling starts to occur. It is
reached where there is a balance between the convective force exerted on particles in the feed
towards the membrane and particle back-transport. In cross-flow MF and UF, back-transport
of particles is facilitated by the cross-flow (Gésan-Guiziou et al. 2002; Bacchin et al. 2006).
The critical flux concept is not applicable to dead-end filtration due to the absence of particle
back-transport in this filtration mode. The general form of the equations for the four fouling
mechanisms in cross-flow filtration is (Field et al. 1995; Field & Wu 2011):


dJ
dt
 
 k J  J  J 2 n (2.6)

where J is volumetric flux at time t and J  is the critical flux (m s-1).

The equations relating the permeate flux, filtration time, and critical flux for the individual
models are given below:

2.4.2.1. Complete pore blocking model (n = 2)

The equation describing the relationship between the flux and the filtration time for this
fouling model is
  J 0 t 
J   J 0  j c  exp   j c (2.7)
 0 
where
J 0 is volumetric flux at time t = 0 (m s-1),

j c is the critical flux for the complete blocking law (m s-1),

 is the blocked area per unit volume of filtrate (m-1),

 0 is the membrane surface porosity at time t = 0
If J is lower than j c , no flux decline is expected.

28
2.4.2.2. Intermediate pore blocking model (n = 1)
For the intermediate blocking filtration model, the relationship between the flux and the
filtration time is given by

1  J 0  ji J 
t  ln   (2.8)
ji  J 0 J  ji 

where ji is the critical flux for the intermediate blocking law (m s-1)

2.4.2.3. Standard pore blocking model (n = 1.5)

This type of fouling is not mediated by back-diffusion of the membrane surface and the
relationship between t and V is the same as in the case of dead-end filtration (equation 2.4).
In terms of flux, this relationship can be written as
1 1  K 
  s A t (2.9)
J J0  2 
where
K s is the Hermia’s parameter for standard blocking model and
A is the membrane surface area (m2)

2.4.2.4. Cake filtration model (n = 0)

1   J J 0  js   1 1 
Gt  ln    j s    (2.10)
j s2   J 0 J  js   J J 0 
where
G is a constant (s m-2) and
j s is the critical flux for the cake filtration law (m s-1)

2.4.3. Concentration polarisation

During the operation of membrane filtration, there is always a convective flow of the solutes
in the feedwater towards the membrane surface and a diffusive backflow of these components
to the bulk feed solution. Before a steady state is established, the forward flow is greater than
the backflow, which results in the accumulation of the solutes at the membrane-solution
interface. Consequently, the concentration of the solutes at the membrane surface is higher
than that in the bulk solution. This phenomenon is known as concentration polarisation
(Matthiasson & Sivik 1980).

29
Concentration polarisation can result in a lower permeate flux and lower or higher solute
retention. It causes an increase in osmotic pressure near the membrane surface, which reduces
the effective TMP and consequently leads to a decrease in the permeate flux. This effect of
concentration polarisation is most significant for salts and small organic molecules because
these components exert a large osmotic pressure. The reduction in permeate flux caused by
concentration polarisation is therefore most significant for NF and RO membranes.

The contribution of concentration polarisation to flux decline in low-pressure membrane

filtration of NOM has been found to be relatively low, as MF and UF membranes are not
effective in removing salts and small organic molecules. Yuan and Zydney (2000)
investigated membrane fouling in MF and UF of Aldrich and Suwannee River humic acids
and found that concentration polarisation contributed to less than 15% of the total flux decline
for the 100 kDa membrane and less than 10% for the 300 kDa and 0.16 μm membranes.
Concentration polarisation is a reversible phenomenon as its effect on flux decline can be
eliminated by reducing the TMP (and hence reducing flux) (Bacchin et al. 2006).

2.4.4. Resistance-in-series model

The resistance-in-series model is the result of a phenomenological approach to fouling. It was

based on the fact that flux decline is contributed by different factors, including pore
adsorption, pore blocking, cake formation, and concentration polarisation.

According to this model, the membrane resistance and the resistances by fouling and by
concentration polarisation act in series to retard the permeate flow. Consequently, the
permeate flux can be written as (Barger & Carnahan 1991; Mulder 1996):
P  
J (2.11)
 Rm  Ra  R p  Rc  Rcp 
where
 is the osmotic pressure difference across the membrane,
Rm , Ra , R p , Rc , and Rcp are the resistances exerted by the membrane, adsorption of foulants

on membrane, pore blocking by foulants, cake layer formed on the membrane surface, and
concentration polarisation, respectively. The membrane resistance Rm can be obtained from
experiments on pure water feed.

30
Because in MF and UF of NOM and EfOM the rejected solutes are high MW compounds and
their concentrations are low, concentration polarisation and osmotic pressure difference can
usually be neglected. If the osmotic pressure difference across membrane  is negligible,
equation 2.11 is reduced to
P
J (2.12)
 Rm  Ra  R p  Rc  Rcp 

Many other versions of the resistance-in-series model have been used. For example, one
model divides the total resistance causing flux decline ( Rtf ) into reversible resistance ( Rrf ,

which includes reversible fouling and concentration polarisation effect) and irreversible
resistance ( Rif , which accounts for irreversible fouling). The permeate flux is then expressed

as (Aimar & Howell 1989):

P P
J  (2.13)
 Rm  Rtf   Rm  Rrf  Rif 
where Rtf is the resistance by total fouling

2.5. Fouling mitigation

Several approaches have been utilised to control membrane fouling. These include:
 Pre-treatment of the feedwater,
 Modification of membrane properties,
 Optimisation of module configuration and operating conditions,
 Periodic membrane cleaning,
 Evaluation of system performance using pilot plant, and
 Use of predictive models

2.5.1. Pre-treatment of the feedwater

Secondary effluent is usually pre-treated to reduce membrane fouling and/or improve the
permeate quality before being introduced to MF or UF processes. Appropriate pre-treatment
methods are selected based on the components to be removed and the extent of their removal.
Common pre-treatment methods for MF and UF include (Water Environment Federation
2006; Huang et al. 2009):
 Coagulation (usually with lime, alum, or ferric salts) and flocculation

31
  Adsorption (most commonly used material is powdered activated carbon)
 Pre-oxidation (using ozone)
 Pre-filtration (using large pore size membranes, granular media, filter cloth, etc.)
In practice, these methods can be applied in combination, for example, coagulation followed
by pre-filtration, coagulation followed by adsorption with powdered activated carbon (PAC)
(Abdessemed et al. 2000; Guo et al. 2004), and ozonation followed by coagulation (Lehman
& Liu 2009).

2.5.1.1. Coagulation and flocculation

For coagulation to take place the repulsive forces between like-charged particles must be
overcome, i.e., the colloidal dispersion must be destabilised. In wastewater treatment, this is
usually carried out by the addition of coagulants such as alum (aluminium sulphate), ferric
chloride, or appropriate cationic polyelectrolytes. In flocculation, particles formed as a result
of coagulation are further contacted to form larger aggregates called flocs. Polyelectrolytes
that accelerate the rate of floc formation and make flocs mechanically more stable can be used
as flocculant aids (Kolarik & Booker 1995).

It is thought that the removal of NOM by coagulation occurs via two major mechanisms
(Gregor et al. 1997; Wang et al. 2010):
(i) ‘Charge neutralisation’ of negatively charged NOM by cationic metal hydrolysis
species which leads to the formation insoluble charge-neutral products,
(ii) ‘Sweep coagulation’ in which NOM is removed by entrapment or adsorption on
metal hydroxide precipitate (e.g., Al(OH)3 and Fe(OH)3).
Charge neutralisation operates at quite low metal concentrations (typically a few µM for near
neutral pH conditions) where the formation of metal hydroxide precipitate is minimal. For
‘sweep coagulation’ to take place, the coagulant dose must be sufficiently high to cause rapid
precipitation of insoluble metal hydroxide.

Feed coagulation is a common pre-treatment method to reduce fouling of MF and UF

membranes. It is also used to enhance the removal of cysts, bacteria, viruses, algae, turbidity,
silt, colour, DOM, and inorganic compounds from the feedwater. However, both beneficial
and adverse effects of coagulation on membrane flux performance have been reported,
reflecting the complexity of the relationships between water contents, coagulants (type and

32
dose), and membranes. Therefore, coagulant type and dose must be carefully selected to avoid
acceleration of membrane fouling and maximise the usefulness of coagulation (Adham et al.
2006). Howe and Clark (2006) found that coagulation exacerbated fouling in MF of a surface
water and attributed this effect to “under-dosing” of the coagulant, which led to the formation
of fine flocs having sizes close to the pore sizes of MF membranes.

Previous studies on MF and UF of secondary effluent showed that coagulation led to

significant improvement in the permeate flux (Kim et al. 2005). Haberkamp et al. (2007)
attributed the improved flux to the removal of the high MW components (i.e., biopolymers,
which contained proteins and polysaccharides, and HS) of the EfOM.

Pre-hydrolysed inorganic coagulants such as poly(aluminium) chloride (PACl) and

poly(aluminium) chlorohydrate (ACH) have been used in a number of drinking water
treatment plants in place of alum due to their lower alkalinity consumption and lower sludge
production. The use of PACl and ACH was found to reduce breakage of coagulated flocs and
reduce aluminium concentration in the treated water (Gebbie 2001). Practical experience in
drinking water treatment plants indicated that ACH woked effectively over a broader pH
range (5.0 – 8.0) and produced less sludge compared with alum (5.0 – 6.5), the most widely
used coagulant (Jiang & Graham 1996; Gebbie 2001). These characteristics make ACH an
attractive alternative for alum in coagulation of secondary effluent, particularly those with pH
greater than 7.0 (such as the AS effluent being investigated in this study, pH 7.2 –7.8).

There are two process configurations of coagulation pre-treatment for MF/UF: coagulation
followed by sedimentation and in-line coagulation (Huang et al. 2009a). In the first
configuration, the coagulated materials are allowed to settle in sedimentation basins prior to
the membrane units. In-line coagulation usually uses lower coagulant doses and no removal of
the coagulated matter is done before MF or UF (Choi & Dempsey 2004; Zheng et al. 2012).
In-line coagulation is usually the preferred configuration as it reduces the process size and
therefore improves process economics (Judd & Hillis 2001; Sharp & Escobar 2006).

Coagulation can be accomplished with or without prior pH adjustment of the feed water. pH
adjustment is normally practised when optimal removal of DOM and contaminants (e.g.,
phosphorus species) is required (Water Environment Federation 2006).

33
2.5.1.2. Ozonation
Ozone is a strong oxidant that preferentially oxidises electron-rich compounds containing C–
C double bonds and aromatic alcohols (Von Gunten 2003). Ozonation has been used widely
in drinking water production to remove colour, tastes and odours. However, the application of
this process in sewage treatment is quite limited (Oneby et al. 2010). It was found that
ozonation reduced chemical oxygen demand (COD), BOD, and colour, and increased the
dissolved oxygen concentration in municipal secondary effluent without causing any harmful
effects on toxicity and mutagenicity (Paraskeva et al. 1998).

Lee et al. (2005) found that ozonation reduced membrane fouling in MF of a municipal
secondary effluent by decomposing high MW organic materials to compounds of lower MW.
Ozonation at high concentrations (~5.0 mg O3/mg SS, 15-minute contact time) also resulted in
the breakdown of large SS particles and converted the particle size distribution (PSD) of the
SS to a smaller and narrower range (Zhu et al. 2008). The SS with narrower PSD was thought
to create a more porous gel layer which improved the flux performance. At lower doses (~
0.55 mg O3/mg SS, 2-minute contact time), ozone was found to have a coagulation-aid effect
as it enhanced particle destabilisation and consequently improved particle aggregation by
coagulation. Reduced membrane fouling could also be due to the change in the SS zeta
potential, which reduced the affinity between the SS and membrane surface (Zhu et al. 2008).
On the other hand, Zhu et al. (2009) found that ozonation pre-treatment worsened membrane
fouling in dead-end MF of a secondary effluent with a ceramic membrane (pore size = 0.1
μm). They attributed this phenomenon to the increase in the concentration of large organic
molecules after ozonation and suggested this increase may have resulted from the lysis of the
bacteria in the effluent.

As ozonation kills micro-organisms, it has the potential of reducing biological fouling of the
downstream membrane systems. However, the treatment may lead to the release of
biodegradable substances from broken bacterial cells and the breakdown of high MW DOM
in water to low MW compounds (e.g., organic acids, alcohols, aldehydes, ketones). These
events may facilitate bacterial regrowth (Van der Kooij et al. 1989; Miettinen et al. 1998) and
therefore result in accelerated membrane biofouling. This is likely to happen when ozonation
and residual ozone removal are performed continuously (Voutchkov 2008). This problem can
be addressed by introducing a biological activated carbon (BAC) filtration process between
the ozonation process and the membrane system.

34
In BAC filtration, the biodegradable materials in the ozonated water are consumed by the
microbial flora attached to activated carbon (Volk et al. 1993; Lambert & Graham 1995;
Swietlik et al. 2004). The operating life of the activated carbon in BAC filtration can be as
long as 10 years, resulting in a low need for carbon regeneration (Wricke et al. 1996).
Ozonation followed by BAC filtration has rarely been used in sewage treatment and therefore
little research on their effects on membrane fouling has been done. The combined ozonation-
BAC filtration process was trialled at the WWTP from which the samples for this study were
collected.

2.5.1.3. Adsorption with ion exchange resins

Ion exchange resins are cross-linked polymers with fixed functional groups. They can be
cationic or anionic ion exchangers. Anion exchange resins (AERs) have been used extensively
in water and wastewater treatment (Zagorodni 2007). Typical AERs are made up of a polymer
matrix (most commonly polyacrylic or polystyrene cross-linked with divinylbenzene) with

N  CH 3 3 ) and exchangeable

 
fixed cationic functional groups (such as  NH 3 , NH 2 , or

anionic groups (e.g., chloride or hydroxide ions) (Zagorodni 2007). During contact with water
or wastewater, the DOM of which is predominantly in the anionic form, the negatively
charged DOM molecules displace the anionic groups to attach to the cationic functional group
of the AER.

AERs have been found very effective in the removal of DOM and colour in water. Many
AERs have the advantage of easy regeneration. Those with higher water contents have been
found to be more efficient in DOC removal due to the presence of more active sites. Strong
base resins tend to be more effective for DOM removal than weak base resins (Bolto et al.
2002).

Water pH may greatly influence the effectiveness of the ion exchange process. Croué et al.
(1999) suggested that at high pH, ion exchange was the dominant mechanism for the removal
of DOM, whereas at neutral pH, physical adsorption also played a role. Physical adsorption is
believed to be responsible for the removal of neutral organic compounds (Bolto et al. 2002;
Cornelissen et al. 2008). The presence of anions and cations in water has been found to affect
the ion exchange process. It has been reported that the presence of sulphate and nitrate in
water reduced NOM removal by AERs, as these anions competed with NOM for active sites

35
on the resins (Boyer & Singer 2006). Bicarbonate and chloride may increase the NOM
removal, as a result of dehydration of hydrophobic acids - a phenomenon referred to as the
‘salting out’ effect (Croué et al. 1999). The presence of iron in the water to be treated may
lead to the formation of an iron dioxide coating on the resins and therefore reduce their ion
exchange capacity (Hongve et al. 1999).

Regarding the DOM components removed by AERs, researchers have found that most AERs
used for water treatment are effective for the removal of aromatic hydrophobic DOM
components (humic and fulvic acids) with low to intermediate MWs (500–4000 Da) (Allpike
et al. 2005; Boyer & Singer 2005; Zhang et al. 2006; Humbert et al. 2007). AERs are also
capable of removing transphilic and hydrophilic NOM components (Croué et al. 1999).
However, they are not effective for the removal of large molecules such as polysaccharides
and proteins) (Croué et al. 1999; Humbert et al. 2007; Cornelissen et al. 2008).

Cornelissen et al. (2008) studied the removal of NOM in surface water by nine AERs and
found that the resins preferentially removed HS, hydrolysates of HS (building blocks) and low
MW neutral compounds with removal ranging from 1% to 60%. The resins tested showed
high removals of low MW neutral compounds (approximately 50%), which was attributed to
physical adsorption rather than ion exchange. The researchers also observed an increase in the
concentration of biopolymers (i.e., polysaccharides and proteins) after the ion exchange
treatment and attributed this increase to the washing out of polymeric compounds stemming
from the manufacturing process of the AERs. The results indicate that some AERs may need
appropriate treatment before being used for the removal of DOM.

Magnetic ion exchange (MIEX) has been used widely in the treatment of surface water for
the removal of NOM. Tests on surface water showed that MIEX was most effective in the
removal of NOM with MWs between 500 and 1500 Da (Humbert et al. 2005). Fearing et al.
(2004) reported that fulvic acids, due to their lower MW (800 Da), were removed more easily
by MIEX than humic acids (MW = 1060 Da). Similarly, Zhang et al. (2006) found that
MIEX removed the majority of low MW (500–1000 Da) components of EfOM collected
from a municipal WWTP.

36
Although pre-treatment with AERs can remove a large proportion of DOM from water and
wastewater, previous studies showed that this does not always result in fouling reduction in
MF and UF. For example, fouling reduction by adsorption pre-treatment with MIEX was
observed by Zhang et al. (2006) and Kim and Dempsey (2010), while no effect on fouling
was observed by Humbert et al. (2007). It should be noted that in their studies, Zhang et al.
and Kim and Dempsey conducted filtration tests with secondary effluent, whereas Humbert et
al. worked on surface water.

Zhang et al. (2006) attributed fouling reduction by pre-treatment with MIEX to the removal
of EfOM from the secondary effluent. However, the identity of the EfOM components
responsible for fouling was not provided. Kim and Dempsey (2010) demonstrated that
organic acids (which, according to their fractionation protocol, contained humic acids and
fulvic acids) were the major membrane foulants in MF and UF of secondary effluent.
Consequently, pre-treatment with ion exchange resins effective in the removal of organic
acids, such as MIEX, IRA-958, and diethylaminoethyl cellulose, resulted in significant
fouling reduction. On the other hand, Humbert et al. (2007) explained the ineffectiveness of
pre-treatments with MIEX and IRA-958 for fouling reduction as due to the poor removal of
NOM components with MWs between 20 kDa and 80 kDa. The inconsistencies among the
results reported in the literature suggest that further studies need to be carried out to obtain
better understanding of the identity of the membrane foulants and the effect of pre-treatment
with ion exchange resins on fouling.

2.5.1.4. Adsorption with activated carbon and other adsorbents

Powdered activated carbon (PAC) is used widely for the removal of taste, odour, DOM, and
synthetic organic chemicals. Many water treatment plants use PAC as a pre-treatment for the
feedwater to MF and UF systems. These plants utilise the membranes to separate the spent
PAC, micro-organisms, and other particulate matter, from the finished water (Adham et al.
2006).

Previous studies have shown that adsorption of DOM on PAC is greatly influenced by the
surface chemistry (particularly the surface charge) and the pore size distribution of the
activated carbon (Summers & Roberts 1988a; Kilduff et al. 1996; Newcombe & Drikas 1997;
Bjelopavlic et al. 1998). In general, the DOM components removed by microporous PAC

37
(pore size < 2 nm) are those of low MWs (< 600 Da), whereas mesoporous PAC (pore size =
2–50 nm) are capable of adsorbing DOM components with a range of MWs (from 200 Da to
larger than 1000 Da) (Newcombe et al. 2008). It has also been reported that low MW
components of NOM are adsorbed on PAC to a greater extent than larger compounds due to
the size exclusion effect (Newcombe et al. 2008).

Many commercial PAC products have been found very effective in removing aromatic
compounds, including humic and fulvic acids (Adham et al. 1991; Kim et al. 2007; Kim et al.
2008a; Song et al. 2009), and micropollutants, such as pharmaceuticals and cyanotoxins
(Campos et al. 1998; Newcombe et al. 2008; Saravia & Frimmel 2008; Campinas & Rosa
2010). Regarding to studies which used PAC adsorption as a pre-treatment to MF/UF, Lin et
al. (1999) reported that the PAC they used (C-5260, Sigma, mesoporous, average pore size =
25 nm, surface area = 730 m2 g-1) was effective for the removal of NOM components between
300 Da and 17,000 Da, but was not able to adsorb components outside this range. On the
other hand, Haberkamp et al. (2007) found that other mesoporous PAC products (Norit SA
Super, Norit AZO, and Donau Carbon Carbopal MB 4) were capable of adsorbing EfOM
components of a wide range of MWs, including biopolymers (MW >> 20,000 Da), HS (MW
= 1,000–20,000 Da), building blocks (MW = 300–500 Da) and low MW neutral compounds
(MW < 350 Da).

Although PAC can be highly effective in removing DOM, using PAC as a pre-treatment for
membrane systems does not guarantee a reduction in membrane fouling. For example, Carroll
et al. (2000) observed that while PAC adsorption was able to remove nearly 70% of the DOM
from Moorabool River water (Victoria, Australia), the treatment had little impact on the
permeate flux in dead-end MF. However, Lin and co-workers (2001) and Mozia et al. (2006)
found that PAC adsorption intensified fouling of UF membranes by humic acid. Lin et al.
(2001) attributed this effect to (i) the higher fouling potential of the HS remaining after PAC
adsorption than that of the untreated humic acid solution and (ii) the increased hydrodynamic
resistance caused by the dense organics-PAC mixture on the membrane surface. Their second
reason for the aggravated fouling was supported by Zhang et al. (2003), who also found that
PAC particles contributed to fouling in UF of a surface water, and Dialynas and
Diamadopoulos (2008), who used a municipal wastewater as the UF feed. Saravia et al.
(2006) suggested that to avoid membrane pore plugging by PAC particles, PAC with narrow

38
particle size distributions and particle sizes at least 100 times larger than the membrane pore
size should be used.

Improvement in MF and UF flux performance resulting from pre-treatment with PAC was
observed with surface water by Konieczny and Klomfas (2002) and with municipal
wastewater by Haberkamp et al. (2007) and Lee et al. (2007). Haberkamp et al. (2007)
attributed the increased flux performance to the removal of biopolymers (which contained
proteins and polysaccharides) by the PAC (SA Super, Norit). Lee et al. (2007), however, did
not attempt to elucidate the mechanisms of the fouling reduction observed.

The differences in these findings may be due to differences in the PAC used (e.g., surface
charge, pore size, particle size, and dosage) and differences in characteristics of the feeds and
the membranes. They also illustrate the need to investigate the effect of this pre-treatment on
membrane fouling for the AS effluent under consideration.

Pre-treating surface waters with heated iron oxide particles or heated aluminium oxide
particles was found to reduce fouling of UF membranes (Chang et al. 1998; Zhang et al.
2003; Cai et al. 2008) by surface waters. It has been proposed that heated aluminium oxide
particles can be pre-deposited on the membrane surface to form a fouling-resistant layer (Kim
et al. 2008a) or combined with pre-coagulation to simultaneously reduce fouling and remove
ortho-phosphorus for MF and UF in wastewater reclamation (Kim et al. 2008b). Koh et al.
(2006) showed that treating natural lake water with polysulfone colloids at 5–100 mg L-1
resulted in significant fouling reduction in UF and attributed this effect to the selective
adsorption of the organic foulants (MW = 20–200 kDa) on the colloids. It should be noted
that these novel adsorbents have only been tested with surface waters in laboratories or pilot-
scale plants.

2.5.1.5. Pre-filtration
Tsujimoto et al. (1998) observed a reduction in irreversible fouling in UF of a surface water
after pre-filtration the feed with granular activated carbon filters. However, the mechanism for
the fouling reduction was unclear. Zheng et al. (2010) found that pre-filtration of a secondary
effluent with a slow sand filter resulted in biodegradation of some biopolymers responsible
for fouling of the subsequent PES UF membrane.

39
The use of hydrophobic low pressure membranes with coarse pore matrices to remove
potential foulants has been proposed but rarely implemented since cost-effective measures
need to be developed to restore their removal capacity. Other materials, e.g., filter cloth,
string-wound polyester, and stainless steel, have also been used as the media for pre-filters
(Huang et al. 2009a).

2.5.2. Selection of membrane types and surface modification

Fouling can be reduced by using membranes with narrow pore size distributions or using
hydrophilic rather than hydrophobic membranes. This is particularly useful if proteins are the
foulant because proteins tend to adsorb more strongly on hydrophobic surfaces (Wilf & Alt
2000). If the feed contains charged colloids, using negatively charged membranes can help to
reduce fouling because most colloids are negatively charged (Mulder 1996).

Modification of the membrane surface to change its properties has been practised to reduce
fouling. This can be done by various means such as low temperature plasma activation, UV
irradiation, and redox radical grafting. UV irradiation can be carried out alone or with
hydrophilicity-increasing agents such as dextrans (Nyström & Järvinen 1991) or N-vinyl-2-
pyrrolidinone (Zhang et al. 2009). Redox radical grafting has been reported to increase
hydrophilicity and reduce surface roughness, which in turn reduce adsorption of organic
matter (Sheikholeslami 1999). Co-polymerisation of the base polymers with hydrophilic
materials such as poly(ethylene glycol), poly(ethyleneimine), and ultrahigh MW poly(styrene-
alt-maleic anhydride) has been reported to improve the hydrophilicity and reduce protein
adsorption on the membranes (Ma et al. 2007; Zhu et al. 2007). Surface hydrophilisation of
PVDF membranes has also been done by dispersing titanium dioxide nanoparticles into
PVDF solutions (Oh et al. 2010) and surface coating with poly(vinyl alcohol) (Du et al.
2009).

2.5.3. Optimising module configuration and process conditions

Proper selection of module configuration and operating conditions can be very effective in
reducing fouling. Concentration polarisation can be reduced by manipulating the flux and the
mass transfer coefficient. The mass transfer coefficient can be increased by increasing the
cross-flow velocity or changing the module configuration (decreasing module length and/or
increasing the hydraulic diameter) (Mulder 1996). Turbulence promoters and pulsating flow

40
can be used to break the boundary layer, hence improving the mass transfer coefficient
(Belfort et al. 1994; Mulder 1996).

In cross-flow filtration, operating at high cross-flow velocity (which consumes more energy)
and low flux increases the mass transfer coefficient, hence reducing concentration polarisation
and membrane fouling (Belfort et al. 1994; Mulder 1996). Fouling can be reduced to zero or
negligible if filtration processes are operated below the critical fluxes (sub-critical flux
operation) (Field et al. 1995; Bacchin et al. 2006). Compared with above-critical flux
operation, this strategy requires a much larger membrane surface area for the same water
productivity and hence higher capital investment. However, sub-critical flux operation has
some advantages, including lower energy consumption and less usage of chemicals for
membrane cleaning. Currently, in practice, to increase the water productivity, membrane
filtration processes are usually operated above the critical flux and thus allow fouling. Fouling
is then eliminated by frequent back-flush or back-shock, or by removing the force holding the
cake to the membrane surface (Bacchin et al. 2006). Howell et al. (2002) noticed that using
regular intermittent flux resulted in reduced long-term fouling.

2.5.4. Membrane cleaning

Membrane cleaning is needed to restore the permeate flux loss due to fouling. It can be
accomplished by hydraulic, mechanical, or chemical means. Cleaning methods used are
dependent on the foulant characteristics, membrane material, and membrane configuration.

2.5.4.1. Hydraulic and mechanical cleaning

Common hydraulic cleaning methods include backwashing and alternate pressurising and
depressurising. The term “backwashing” is normally applied to dead-end filtration. During
backwashing, permeate is flushed from the permeate side to the feed side to remove the
fouling layer on the membrane surface. Air scouring may also be applied prior to
backwashing to loosen the deposit layer (Water Environment Federation 2006). Adham et al.
(2005) surveyed a number of drinking water treatment plants in the United Kingdom (UK)
and found that backwash frequency varied from 5 to 96 per day and could last from 10
seconds to 10 minutes. In MF and UF of secondary effluents, backwashing is normally carried
out every 15–30 minutes (Leslie et al. 1999; Decarolis et al. 2001; Laabs et al. 2006; Water
Environment Federation 2006).

41
In cross-flow filtration, a similar technique called “back-flushing” has been used to reduce
fouling (Redkar & Davis 1995). Back-flushing is normally applied for 5–10 seconds every 3–
15 minutes. When back-flushing is applied for a very short duration (typically less than 1
second) and at a much higher frequency (once every few seconds), it is called “back-pulsing”
(Redkar & Davis 1995; Sondhi et al. 2000; Gabrus & Szaniawska 2009). Back-flushing and
back-pulsing are particularly useful in cross-flow filtration with tubular ceramic membranes
since the membranes are able to withstand much higher pressures than polymeric membranes
(Sondhi & Bhave 2001).

So far the vast majority of the studies involving back-flushing and back-pulsing in cross-flow
filtration has focussed on water and wastewater from industrial sources. Matsumoto et al.
(1988) reported significant flux improvement with back-flushing with the permeate (for 5
seconds every 3 minutes) in MF of yeast suspensions. Nipkow et al. (1989) observed a 42%
flux improvement with back-flushing in MF of Clostridium thermosulfurogenes in maltose
medium. Back-pulsing improved the flux performance in cross-flow MF and UF of yeast
suspensions (Redkar et al. 1996), bacterial lysate (Parnham & Davis 1996), oil emulsions
(Srijaroonrat et al. 1999), bovine serum albumin solutions (Kuberkar & Davis 2001), and
synthetic electroplating wastewater (Sondhi et al. 2000). However, information regarding the
effectiveness of back-flushing in reducing fouling in cross-flow MF and UF of surface water
and municipal wastewater remains scant. From an economic point of view, back-pulsing is
not beneficial for these applications due to the high permeate loss and energy demand.

Mechanical cleaning is normally done with oversized sponge balls and can only be used for
tubular membrane configurations in the absence of turbulence promoters or other internal
obstructions (Belfort et al. 1994; Mulder 1996).

42
2.5.4.2. Chemical cleaning
Chemical cleaning is used to remove contaminants which are not removed by hydraulic and
mechanical cleaning methods. Chemically enhanced backwash (CEB) is used to stabilise or
maintain the membrane permeability and therefore is often referred to as “maintenance”
cleaning. In CEB the cleaning solution is added to the backwash water and cleaning is
normally carried out at relatively low cleaner concentrations over a short time, and at ambient
temperature. CEB is automatically scheduled daily to weekly. On the other hand, cleaning-in-
place (CIP) is needed to recover the flux after long term operation. CIP is usually carried out
with higher cleaner concentrations, longer soaking time, and possibly at elevated
temperatures. It is normally applied when the membrane permeability falls below an
acceptable value (Porcelli & Judd 2010). According to Adham et al. (2005), the frequency of
CIP in the UK drinking water treatment plants varied from 0.2 to 50 per year with a median of
4 per year.

The most important consideration in the selection of cleaning agents is the compatibility of
the cleaners and the membranes to avoid membrane damage. The performance of membrane
cleaners needs to be evaluated based not only on their ability to restore the membrane flux,
but also on the ability to maintain the product water quality. This is because although some
chemicals can restore membrane flux, they may significantly reduce the rejection of viruses,
bacteria, and some chemical compounds of concern (Al-Amoudi & Lovitt 2007). The
mechanisms of foulant removal by different chemicals are complex and it has been realised
that the chemistries of foulants, membrane cleaners, and membrane materials may not be
reliable factors for the estimation of cleaning efficiency (Porcelli & Judd 2010).

Cleaning efficiency is affected by many factors, including cleaner concentration, temperature,

cleaning time, and hydrodynamic conditions (Liu et al. 2001). Concentration of cleaning
agents can affect both the equilibrium and rate of reactions. It needs to be sufficient to
maintain a reasonable reaction rate as well as to overcome the mass transfer barrier imposed
by the fouling layer (Liu et al. 2001).

According to Liu et al. (2001), temperature can affect the effectiveness of membrane cleaning
by (a) changing the equilibrium of chemical reactions, (b) changing the reaction kinetics, and
(c) changing the solubility of the foulants and/or reaction products during cleaning. In general,
higher cleaning temperature improves the cleaning efficiency. Increasing temperature

43
increases diffusive mass transfer, foulant solubility, and reactivity of the functional groups of
the foulants. High temperature also promotes swelling and mechanical destabilisation of
organic foulants or biofilm layers (Liu et al. 2001).

Dynamic cleaning in which cleaning solutions are circulated through the system can be more
effective than static cleaning such as soaking because circulation promotes contact between
cleaning agents and foulants and reduces the distance of diffusion by bringing fresh chemicals
close to the foulants (Liu et al. 2001).

In general, acids and chelating agents (e.g., ethylenediamine tetraacetic acid (EDTA) and
citric acid) are effective for cleaning membranes fouled by inorganic scale. Some organic
materials such as proteins and polysaccharides may be hydrolysed by inorganic acids (Liu et
al. 2001). The “salt bridge” effect of divalent cations on HS can cause a denser and more
adhesive fouling layer. Chelating agents and surfactants can form complexes with divalent
cations and therefore improve the cleaning efficiency on membranes fouled by organic
foulants (Al-Amoudi & Lovitt 2007). EDTA has been found very effective for cleaning
membranes fouled by humic acid in the presence of Ca2+ (Hong & Elimelech 1997) while
citric acid is effective for restoring flux of membranes fouled by iron oxides because it not
only dissolves the oxides but also forms complexes with iron (Liu et al. 2001). Due to their
low costs, hydrochloric acid (HCl) and sulphuric acid (H2SO4) are widely used to clean
mineral scales on fouled membranes (Porcelli & Judd 2010).

Bases, oxidants, and surfactants are used to clean membranes fouled by DOM. Alkalis clean
membranes via 2 mechanisms: hydrolysis of proteins and solubilisation of HS (Liu et al.
2001). The hydrolysis of proteins can cause disruption in their tertiary structures and they can
therefore be converted to peptides which are soluble in the bulk cleaning solutions. Alkaline
conditions also increase the negative charge and solubility of HS; this increases the
electrostatic repulsion between HS and the membranes and facilitates the removal of HS from
the membranes (Liu et al. 2001). At high pH, DOM molecules tend to have a stretched, linear
configuration due to the repulsion between negatively charged functional groups (Stumm &
Morgan 1996; Hong & Elimelech 1997). This molecular configuration creates a looser fouling
layer that allows easier access for chemicals to penetrate the inner portion of fouling layer.
The increase in mass transfer leads to improvement in cleaning efficiency. The most
commonly used alkali is sodium hydroxide (NaOH).

44
The oxidation of organic compounds in water and wastewater by oxidising agents generates
oxygen-containing functional groups (e.g., ketone, aldehyde, and carboxylic acid). The
presence of these functional groups generally increases the hydrophilicity of their parent
compounds and therefore reduces the adhesion of fouling materials to the membranes. The
disinfection effect of oxidants has also been found to reduce membrane biofouling. Sodium
hypochlorite (NaOCl) is the most commonly used membrane cleaning oxidant. Hydrogen
peroxide (H2O2) is used when the membranes (e.g., those made of polypropylene) are not
chlorine tolerant or when there are concerns about health and environmental impacts of the
chlorinated organic matter generated. The chlorine in NaOCl may cause swelling of
membranes, which increases the mass transfer of cleaning agents to the membrane surface
(Zeman & Zydney 1996). Cleaning with oxidising agents may result in degradation of
membranes. Chlorination of the C-S bonds in PES membranes was observed at chlorine
concentration above 150 mg L-1 at low to neutral pH levels. As a result, the membrane
mechanical strength was reduced significantly (Arkhangelski et al. 2007). Prolonged
exposure to NaOCl was found to cause ageing of a PVDF membrane, leading to reductions in
the membrane hydrophobicity and hydraulic resistance and changes in the pore size and pore
structures (Puspitasari et al. 2010).

Surfactants have both hydrophilic and hydrophobic structures. They can form micelles with
polysaccharides, proteins, and HS in water and help to separate these materials from the
membranes. Some surfactants interfere with hydrophobic interactions between bacteria and
membranes or disrupt functions of bacterial cell walls and therefore reduce fouling due to
biofilm formation (Paul & Jeffrey 1984; Ridgway et al. 1985; Rosenberg & Doyle 1990).

Enzymes have been used to clean polypropylene membranes in some water treatment plants
in the United States. However, their use in drinking water treatment is constrained by the high
cost and potable water regulation (Porcelli & Judd 2010).

Mixtures of membrane cleaning agents and sequential cleaning with different chemicals have
also been used (Porcelli & Judd 2010). Cleaning models developed using statistical methods
have been used to optimise membrane cleaning efficiency.

45
2.6. Evaluating the fouling potential of feedwater
The silt density index (SDI) and the modified fouling index (MFI) have been used widely for
evaluating the RO membrane fouling potential of a feedwater. As pointed out by Schippers
and Verdouw (1980), SDI is not proportional to particle concentration and therefore cannot be
used to predict the rate of flux decline due to particulate fouling. The MFI and the unified
membrane fouling index (UMFI) can be used to evaluate the MF and UF fouling potential of
surface water and wastewater, provided that fouling is governed by the cake filtration
mechanism.

2.6.1. Modified fouling index (MFI)

The MFI was developed based on the cake filtration equation for constant-pressure dead-end
filtration (Schippers & Verdoux 1980). It is expressed as:

t Rm C b Rˆ c
  V (2.14)
V AP 2 A 2 P
where
t is the filtration time,
V is the permeate volume,
P is the TMP,
 is the dynamic viscosity of the water being tested,
A is the membrane surface area,
Rm is the flow resistance exerted by the membrane,

R̂c is the specific resistance of the cake ( Rˆ c  ARc mc where Rc and mc are the resistance

and the mass of the cake, respectively), and

C b is the particle concentration in the feed.

The slope of the straight line in equation 2.14 is defined as the MFI:
Rˆ c Cb
MFI  (2.15)
2 A 2 P

The MFI is normalised to standard reference values P0 = 2.0 bar,  0 = dynamic viscosity at

20°C, and A0 = 13.8  10-4 m2. It is determined using 0.45 μm MF membranes in constant-

pressure dead-end filtration mode.

46
Feedwater having a higher value of MFI will have a higher fouling potential. It has been
proven that MFI is proportional to particle concentration (Schippers & Verdouw 1980). The
major drawback of MFI lies in the fact that it is based solely on cake filtration and does not
include the contribution of other factors to fouling. Also, the MFI method does not take into
account the dependence of the cake resistance on the pressure (Mulder 1996). Furthermore,
the MFI does not take into account fouling by particles smaller than 0.45 µm (Schippers et al.
1981). To overcome this drawback, a new fouling index, called MFI-UF, was introduced. The
MFI-UF is measured with a 13 kDa PAN membrane at 2.0 bar (Boerlage et al. 2002). More
recently, Khirani et al. (2006a) proposed MFI-NF, which involved the use of hydrophilic PES
NF membranes (500–1500 Da).

2.6.2. Unified membrane fouling index (UMFI)

The concept of UMFI was developed based on the resistance-in-series model and Darcy’s
law, and is written as (Huang et al. 2008; Lozier et al. 2008):

J0  Rˆ c C f  V   Rˆ C 
 1     1   c f v (2.16)
J  Rm  A   Rm 
   
where
J 0 is permeate flux at time t = 0. For constant-pressure filtration, it can be determined by
performing filtration with ultra-pure water (such as Milli-Q water) at the test pressure until
quasi-steady state flux is reached,
J is the water flux through the membrane for the water being tested,
C f is the foulant concentration in the feed, and

v is the accumulated permeate volume per unit membrane area

Equation 2.16 indicates that a plot of the reciprocal of the normalised flux ( J 0 / J ) versus v is
a straight line with the y-intercept being 1. The slope of this line is defined as the UMFI:

R̂c C f
UMFI  (2.17)
Rm
Compared with the MFI, the UMFI has the advantage of being applicable to both constant-
pressure and constant-flux filtration. Furthermore, it has been found better than the MFI in
predicting fouling in pilot-scale filtration using bench-scale data (Lozier et al. 2008).
However, like MFI, UMFI is also established based on the assumption that fouling is

47
governed by cake formation. To extend its usability, Huang et al. (2009b) suggested that the
index could be calculated by using only the first and the last data points of filtration tests
therefore disregarding the fouling mechanisms involved.

The UMFI defined in equation 2.17 is used to evaluate the total fouling potential of a
feedwater and is also called UMFI for total fouling. It is applied to a single long period of
filtration or the first filtration cycle in multi-cycle filtration (interspersed by backwashing)
experiments. In the latter case, it is named “UMFIi”.

According to Lozier et al. (2008), there are other types of UMFI, including the UMFI for
hydraulically irreversible fouling and the UMFI for chemical cleaning. The UMFI for
hydraulically irreversible fouling is calculated by plotting the values of J 0 / J at the

beginning of each filtration cycle against accumulated specific permeate volume ( v ) and
taking the slope of the straight line. In another method of calculation, only values of J 0 / J

for the first and the final filtration cycles are plotted against v . This latter method is intended
to minimise the effect of high variability between cycles in a run (Huang et al. 2009b). The
UMFI for chemical cleaning describes the chemically irreversible fouling for bench-scale
end-of-run backwash operational mode. It is calculated by plotting J 0 / J at the beginning of

the fouling filtration cycle and J 0 / J at the start of the filtration following chemical cleaning

against v (Lozier et al. 2008).

2.7. Characterisation of filtration membranes

Surface properties, structure, and morphology of membranes have been examined by contact
angle and streaming potential measurements, scanning electron microscopy (SEM) with
energy dispersive X-ray (EDX) spectroscopy, tunnelling electron microscopy, atomic force
microscopy (AFM), X-ray photoelectron spectroscopy (XPS), electron spin resonance, and
Raman spectroscopy. Among these techniques, contact angle and streaming potential
measurements, SEM, and AFM are the most widely used (AWWA Membrane Technology
Research Committee 2005; Wyart et al. 2008).

48
2.7.1. Contact angle measurement

Contact angle has been used widely to characterise the surface energy or hydrophobicity of a
membrane surface. It can be determined using techniques such as sessile drop, Wilhelmy
plate, captive bubble, and solvent uptake (Washburn analysis) with manual goniometers or
computerized dynamic contact angle measuring systems with digital image analysis. An ideal
hydrophilic surface has a contact angle of 0° (Clark 1998). In membrane research, it is
generally accepted that membranes with contact angles > 50° are hydrophobic membranes
(Lozier et al. 2008).

2.7.2. Streaming potential

Measuring membrane surface charge and pore charge can give insights to fouling mechanisms
because the charge determines the electrostatic interaction between the membrane and the
foulant. Streaming potential has been used to estimate the zeta potential on both clean and
fouled membrane surfaces (Zhu & Nyström 1998).

The zeta potential of membrane surfaces and membrane pores can be estimated using a
streaming potential analyser and the Helmholtz-Smoluchowski equation, which gives the
relationship between the measurable streaming potential and the zeta potential (Elimelech et
al. 1994; Clark 1998; Soffer et al. 2002). Most membranes used in the water industry are
negatively charged (AWWA Membrane Technology Research Committee 2005).

2.7.3. Atomic force microscopy (AFM)

AFM utilises the mechanical interactions between the membrane surface and a probe scanned
parallel to the mean plane of the surface to obtain 3-D topographic images. It allows
characterisation of surface roughness, pore size, and porosity in clean and fouled membranes
(Clark 1998; Chennamsetty et al. 2006; Wyart et al. 2008).

49
2.8. Techniques for fouling characterisation

2.8.1. Characterisation of organic matter

In membrane fouling studies, DOM in the feed and the permeate has been characterised using
different techniques. The analyses commonly used to investigate fouling of MF and UF
membranes by surface water and municipal wastewater are listed below:

2.8.1.1. Dissolved organic carbon concentration and specific UV absorbance

Organic matter can be characterised using simple techniques such as dissolved organic carbon
(DOC) concentration and UV absorbance at 254 nm. The ratio of this absorbance value
(UVA254) to the DOC gives the specific UV absorbance (SUVA), which is an indication of
the relative amount of humic and non-humic fractions in the water (Weishaar et al. 2003). The
percentage of DOC removal can be used as a measure of the effectiveness of the membrane in
rejecting organic matter (Roddick et al. 2007). Nevertheless, Howe and Clark (2006) realised
that the amount of DOC in the feedwater was not always a reliable indicator of its fouling
potential because lower DOC concentration did not always correspond to less severe
membrane fouling. This finding suggested that some specific fractions of DOC were
responsible for fouling and the proportion of such fractions in the DOC varied from one water
source to another.

Rapid fractionation using XAD-4 and DAX-8 resins, fluorescence excitation-emission matrix
(EEM) spectroscopy, size exclusion chromatography (SEC) with UV and DOC detection,
attenuated total reflection-Fourier transform infrared (ATR-FTIR) spectroscopy, and
biodegradable organic carbon (BDOC) determination can be used to obtain further
information about size, structure, and functionality of dissolved organic matter in the
feedwater and in the permeate.

2.8.1.2. Rapid fractionation

Rapid fractionation of organic constituents of the feed using XAD-4 and DAX-8 resins allows
the separation of these components into the hydrophobic (HPO, containing mainly HS),
transphilic (TPI), and hydrophilic (HPI, containing polysaccharides) fractions. The
hydrophilic fraction can be further separated into the charged (CHA) and neutral (NEU)

50
fractions (Chow et al. 2004). Rapid fractionation facilitates the study of the fouling potential
of organic fractions of different hydrophobicity (Roddick et al. 2007).

2.8.1.3. Fluorescence spectroscopy

Fluorescence excitation-emission matrices (EEMs) give information regarding fluorescent
DOM in water and wastewater. They allow differentiation of fluorescent humic acid-like and
fulvic acid-like materials, proteinaceous materials, and SMP in the feed, permeate, and
backwashed effluent. Fluorescence spectroscopy has been proven useful for the identification
of the membrane foulants which were removed by backwashing and cleaning with NaOH
(Henderson et al. 2011). Peldszus et al. (2011) applied principal component analysis to the
fluorescence EEMs of a biofiltered natural water and found that the concentration of the
protein-like substances was highly correlated with the hydraulically irreversible fouling of the
UF membrane treating this water.

2.8.1.4. High performance size exclusion chromatography

High performance size exclusion chromatography (HPSEC) with DOC and UV detection
permits the determination of the apparent molecular weight distribution (AMWD) of the non-
UV-absorbing and UV-absorbing molecules in water. It also gives information about the
concentration of different molecular weight ranges and their chemical nature (Her et al. 2002;
Allpike et al. 2005; Roddick et al. 2007).

2.8.1.5. Biodegradable dissolved organic carbon

Biodegradable dissolved organic carbon (BDOC) gives a measure of the biodegradability of
the dissolved organic components in the feedwater and is an indication of the feedwater’s
biofilm formation and biofouling potential. It is determined by incubating the water sample
with biologically active sand for 5–7 days under aerobic conditions and measuring DOC
concentration daily. The highest DOC reduction observed during this time is taken as the
BDOC concentration.

2.8.2. Characterisation of foulant materials on membranes

2.8.2.1. Attenuated total reflection-Fourier transform infrared (ATR-FTIR) spectroscopy

ATR-FTIR spectroscopy is an in situ technique that can be used to examine chemical
structures of fouling layers. It is based on the infrared vibrational “absorption” spectra of

51
materials. Different peaks in the ATR-FTIR spectra correspond to compounds of different
functionalities (Oldani & Schock 1989; Shon et al. 2004).

2.8.2.2. Scanning electron microscopy (SEM) and energy dispersive X-ray (EDX)
spectroscopy
High resolution images of clean and fouled membrane surfaces and cross sections can be
obtained from SEM and electron probe microanalysis. The resolution of the images can be as
high as 7Å, which is achieved with low-voltage field emission. SEM can also give
information on the elemental content of the sample if used in combination with EDX
spectroscopy (Clark 1998). Some researchers have used SEM/EDX to analyse foulants on the
membrane surface (Al-Halbouni et al. 2008). The major disadvantage of SEM is the need for
drying and coating of the samples because these processes are likely to introduce artefacts.
This problem is resolved by environmental scanning electron microscopy (ESEM), which is
conducted under low vacuum conditions (Roddick et al. 2007).

52
 CHAPTER 3. MATERIALS AND METHODS

3.1. Wastewater sources

3.1.1. Raw and ozonated effluent and BAC filtrate
The WWTP from which the secondary effluent was collected utilised a conventional
wastewater treatment process, which is described in Figure 3.1.

Wastewater
influent Grit to landfill Primary treatment Secondary treatment

Screening and Effluent

Screening Grit sedimentation to Activated
Clarifiers holding
removal remove solids sludge basin

Preliminary treatment
Sludge recycling Micro-
Sludge screens
Sludge

Disinfection
Sludge Sludge Sludge
drying digestion thickening
Outfall
pumping
Biosolids station
Biogas
(used as fuel for Recycled
pumping station and water
power station)
Discharge
to ocean

Figure 3.1. The wastewater treatment process at the WWTP

In preliminary treatment, large objects such as cotton buds, rags and other rubbish in the
sewage influent were removed by screening with fine screens. The wastewater was then
aerated to remove finer particles (e.g., grit and sand). It was then transferred to the primary
clarifiers, where heavy objects settled and formed “sludge” at the bottom of the tanks and
lighter materials (e.g., oil and fat) floated to the top of the tanks. The sludge and the floating
debris were pumped to the digesters, where they were broken down by bacteria. The
wastewater from the primary clarifiers was passed through the secondary treatment process,

53
which was an activated sludge (AS) system, for the breakdown of organic material by micro-
organisms. This wastewater, after being pumped to secondary clarifiers, was passed through
3-mm micro-screens and then disinfected by chlorination. A small proportion (< 5%) of the
chlorinated effluent was used as recycled water and the rest was discharged to the ocean.
Because the secondary treatment at the WWTP was an AS system, in this thesis the secondary
effluent being studied will also be called “AS effluent”.

The characteristics of the raw AS effluent are shown in Table 3.1 (data obtained from 36
samples collected from June 2008 to September 2011).

Table 3.1. Characteristics of the raw AS effluent used in this study

Table 3.2
Parameter Value Median Mean Standard deviation
pH 7.2 – 7.9 7.6 7.6 0.2
Total DOC (mg L-1) 10.5 – 15.9 13.8 13.5 1.4
UVA254 (cm-1) 0.29 – 0.44 0.39 0.37 0.04
SUVA (L m-1mg-1) 2.1 – 3.8 2.6 2.8 0.44
True colour (mg Pt-Co L-1
65 – 133* 103 101 15
or Pt-Co unit)
Turbidity (NTU) 1.9 – 6.6 4.0 4.0 1.2
TSS (mg L-1) 2.1 – 10.9 4.2 4.5 1.9
Conductivity (μS cm-1) 900 – 1060 980 983 49
HPO (% of total DOC) 41 – 49
TPI (% of total DOC) 21 – 27
HPI (% of total DOC) 28 – 32
* Values of 80–110 Pt-Co units were frequently observed

At a small trial plant within the WWTP, the raw AS effluent was ozonated at a dose of 10 mg
O3 L-1 for 10 minutes and the ozone transfer efficiency was consistently greater than 96%.
The ozone concentrations in the treatment process were measured (by the WWTP staff) using:
(i) a process water flowmeter, (ii) a process gas (ozone/oxygen) flowmeter and an ozone
concentration analyser for the process gas, (iii) a contactor off-gas flowmeter and an off-gas
ozone analyser, and (iv) an ozone analyser for the concentration of residual dissolved ozone in
the process water (usually zero). These measurements were used to perform a mass balance
for ozone around the ozonation process and determine the ozone transfer efficiency (which
was consistently greater than 96%).

54
The ozonated effluent was then filtered through a BAC column. The depth of the BAC bed
(particle size = 8 × 16 mesh or 2.36 × 1.0 mm) was 2 m and the column was operated in
downflow mode at 12 m h-1 and an empty bed contact time (EBCT) of 10 minutes. The
growth of the microbial community in the BAC column occurred naturally, since ozonation of
the raw effluent did not completely inactivate the biomass (the raw effluent contained residual
biomass from the upstream activated sludge plant). The first BAC filtrate sample used in this
study was collected when the BAC column had been in operation for approximately 3
months. As observed by the WWTP staff, the adsorption capacity of the activated carbon was
exhausted by the time this research was conducted. Therefore, the removal of dissolved EfOM
by the BAC filter observed in this study was essentially due to biodegradation.

Samples of raw and ozonated effluent and BAC filtrate were collected from the trial plant and
transported to RMIT University, where they were stored at 4C when not in use. Analyses and
filtration experiments were done within 4 days of sample collection. Since ozonation of water
and wastewater is known to cause breakdown of DOM structures and increase the
biodegradability of water samples (Takahashi et al. 1995; Camel & Bermond 1998; Von
Gunten 2003), the experiments with the ozonated effluent and the BAC filtrate were
conducted soon after sample collection to minimise effects of biodegradation on the
characteristics and fouling potentials of these samples. Monitoring the flux data of the raw
effluent on day 1 and day 4 showed that the effect of sample storage on the flux data was
minimal.

55
3.1.2. Coagulated samples
Coagulated samples were prepared in the laboratory at RMIT University. Coagulation was
performed with the raw effluent in a jar test apparatus (PB-700, Phipps & Bird). The
coagulants were aluminium sulphate (alum, supplied by BDH Chemicals) and
(poly)aluminium chlorohydrate (ACH, supplied by Omega Chemicals as MEGAPAC 23).
The tested coagulant doses were 1.0 mg Al3+ L-1 and 2.5 mg Al3+ L-1. The coagulant doses
were selected based on the recommendation by the WWTP staff and common practice in
municipal wastewater treatment, which tends to use in-line coagulation with low coagulant
doses (normally less than 5 mg Al3+ L-1) (Decarolis et al. 2001; Qin et al. 2004; Zheng et al.
2012). Note that in this study, “in-line coagulation” refers to coagulation without subsequent
removal of the coagulated matter by settling (Choi & Dempsey 2004; Wang et al. 2008).
After the addition of the coagulants (without pH adjustment), the samples were mixed at 160
rpm (velocity gradient (G) = 180 s-1) for 2 minutes and then at 30 rpm (G = 24 s-1) for 20
minutes. In coagulation practice, the rapid mixing time may vary from 30 seconds to 2
minutes (with G = 100–400 s-1) whereas the slow mixing time may range from 90 seconds to
20 minutes (G = 25–90 s-1) (Choi & Dempsey 2004; Howe et al. 2006; Wang et al. 2008).
The mixing conditions selected for this study were based on those used by Choi and Dempsey
(2004) and Wang et al. (2008). The coagulated samples were then subjected to MF or UF
without prior removal of the coagulated matter to simulate in-line coagulation (Choi &
Dempsey 2004).

3.1.3. Samples pre-treated with an anion exchange resin (AER)

Attempts were made to reduce the colour of the raw AS effluent and it was found that apart
from ozonation, colour reductions could be achieved by treating the raw effluent with a
commercial strong base AER or a PAC. The characteristics of the AER (given by the
manufacturer) are presented in Table 3.2.

Table 3.3. Characteristics of the AER used in this study

Functional group Quaternary ammonium
Matrix Polyacrylic
Mean pore size 45 nm
Charge capacity 0.52 meq mL-1
Water content 50–60%

56
Initial jar tests were conducted with AER doses of 5 and 10 mL L-1 to identify the resin
dosage which gave a colour reduction approximately 75–80%. The AER volume was
determined using a 10 mL measuring cylinder, in which the resin was allowed to settle for 1
hour. A plastic pipette was used to adjust the resin volume in the cylinder to the required
doses. After adding the resin to the AS effluent, the jar contents were stirred at 100 rpm (G =
100 s-1) for 20 minutes. The resin was then allowed to settle for 10 minutes and the
supernatant was retained for testing.

As in practice the AER is normally used to treat approximately 1500 bed volumes (BVs) of
water before it is regenerated, the resin was reused without regeneration until the total volume
of the AS effluent treated reached 1500 BVs. The combined treated AS effluent at each
particular dose could be considered as being equivalent to that produced from plant operation.
The colour removals at resin doses of 5 mL L-1 and 10 mL L-1 were 65% and 80%,
respectively. The AS effluent sample treated with 10 mL AER L-1 was used in the MF and UF
tests.

3.1.4. Samples pre-treated with a powdered activated carbon (PAC)

Acticarb PS1300 (supplied by Activated Carbon Technologies) was selected for its high
adsorption activity and high colour removal (as per manufacturer’s information). The pore
size distribution of the PAC was determined from nitrogen gas adsorption data at 77 K with a
Micromeritics Accelerated Surface Area and Porosimetry (ASAP) 2000 unit. Prior to the
measurement, the sample was degassed at 250°C overnight under vacuum. The total surface
area of the PAC was calculated using the Brunauer-Emmett-Teller (BET) method. Volumes
of micropores and mesopores were obtained from Barrett-Joyner-Halenda (BJH) pore area
and pore volume analysis. The characteristics of Acticarb PS1300 are shown in Table 3.3.

Table 3.4. Characteristics of the PAC used in the study

Origin Coal
Iodine number (mg g-1) > 1200
Average particle size (μm) 20–30
Micropore (< 2 nm) volume (cm3 g-1) 0.15
3 -1
Mesopore (2–50 nm) volume (cm g ) 0.52
Surface area (m2 g-1) 1429

57
PAC adsorption of HS, the major colour-causing compounds in waters and municipal
wastewaters, may take several days to reach equilibrium (Summers & Roberts 1988b).
However, such a long contact time is not available in the wastewater treatment plant.
Therefore, in this study, the contact time for PAC adsorption was limited to 30 minutes. The
adsorption was carried out in a jar tester with a stirrer speed of 100 rpm (G = 100 s-1).
Preliminary tests showed that for a contact time of 30 minutes, a PAC dose of 150 mg L-1 was
needed to achieve 80% colour removal.

3.2. Analytical methods for wastewater samples

3.2.1. General characteristics
Except for the MF and UF permeates, the dissolved organic carbon (DOC) concentration, UV
absorbance and true colour were obtained from samples which were vacuum-filtered through
0.45 μm cellulose acetate membranes (C045A047A, Advantec MFS) using a procedure
similar to that described by Karanfil et al. (2003). According to Advantec MFS (2001), this
type of membrane contains little aqueous extractables and has low protein binding capacity.
The vacuum flask and the filter holder were cleaned thoroughly with Milli-Q water prior to
filtration with the samples. To minimise leaching of organic carbon and UV254-absorbing
components from the filters, the membranes were rinsed with 600 mL of Milli-Q water before
use and filtrates were collected after an initial portion (approximately 10 mL) was wasted.

3.2.1.1. pH, conductivity, turbidity

The pH, conductivity, turbidity were measured using a Mettler Toledo pH meter, a Hach
Sension5 conductivity meter, and a Hach 2100P Turbidimeter, respectively.

The pH meter was calibrated with standard solutions of pH 4.0, 7.0, and 10.0. Calibration of
the conductivity meter was carried out with 500, 1413 and 2760 μS cm-1 potassium chloride
(KCl) standard solutions. Formazin standards (20, 100, and 800 NTU) were used to calibrate
the turbidimeter.

3.2.1.2. UV absorbance, true colour, and residual ozone concentration

UV absorbance was measured with a Unicam UV/Vis spectrophotometer with Milli-Q water
as the reference. A Hach spectrophotometer (model DR/4000) was used to determine the
sample colour in Pt-Co units according to the Standard Method 2120 C (American Public
Health Association 2005) and residual ozone concentration in the ozonated effluent samples

58
using the Indigo method (Hach Company 2003). In this method (Method 8311), the indigo
reagent reacted immediately and quantitatively with ozone. The blue colour of indigo was
bleached in proportion with the concentration of ozone in the sample. No residual ozone was
detected in the ozonated effluent and hence did not affect the membrane flux performance.

3.2.1.3. DOC concentration

DOC concentration was measured with a Sievers 820 TOC analyser. The samples, except for
the MF and UF permeates obtained from filtration tests, were filtered through 0.45 μm
cellulose acetate membranes prior to this analysis. The instrument used 6.0 M phosphoric acid
(H3PO4) to acidified the sample to pH 2 prior to analysis and aqueous ammonium persulphate
((NH4)2S2O8) as the oxidiser. It operated on the ultraviolet light-persulphate oxidation method
to determine the total carbon and total inorganic carbon in the sample. The total organic
carbon was calculated by subtracting the latter from the former. Internal calibration was done
by the manufacturer. External calibration for total organic carbon and inorganic carbon was
performed using potassium hydrogen phthalate (1, 5, 10, and 25 mg C L-1) and sodium
carbonate (1, 5, and 25 mg C L-1), respectively. The accuracy of the measurement was quoted
as  3% of the reading from the instrument.

3.2.1.4. Total suspended solid (TSS) concentration

TSS concentration was determined according to the Standard Method 2540 D (American
Public Health Association 2005) using 0.5 μm glass fibre filter (GC50, Sterlitech Corporation,
Washington, USA). For the raw and ozonated effluent, 1 L of the sample was used to
determine the TSS concentration. The measurement was repeated using 1.5 L of the samples
for those with TSS concentrations lower than 2.5 mg L-1 (to follow the Standard Method 2540
D). For the BAC filtrate, 5 L of the sample was used and therefore the amount of TSS
retained by the 0.5 μm glass-fibre filter was less than 2.5 mg.

3.2.1.5. Particle size distributions and zeta potentials

The size distributions and the zeta potentials of the particles in the raw and ozonated effluent
samples were measured with a Mastersizer 2000 and a Zetasizer 2000 (Malvern Instruments,
UK), respectively. Mastersizer 2000 uses low angle and wide angle light scattering as well as
back scattering to determine the particle properties. The particle size distribution was
calculated by the Malvern software, using the intensity of the scattered light obtained from

59
detectors in the optical unit. The calculation was based on the Mie theory, assuming that
100% of the volume was within the measurable range. The volume median diameter, defined
as the value below the 50th percentile of the particle size, was used to quantify the results.
Three replicates of particle size measurement and five replicates of zeta potential
measurement were performed for each sample and the average results were reported.

3.2.2. Fractionation of EfOM

Rapid fractionation following the procedure described by Lee et al. (2004) (Figure 3.2) was
used to separate the EfOM in the samples into 3 fractions: HPO, TPI, and HPI. The fractions
were recovered and used in MF and UF experiments to determine their relative fouling
potentials. Fractionation was carried out using glass columns containing 25 mL of the resins.
The sample flow rate passing through the resins was 3 BVs hr-1.

Whole
EfOM,
pH 2
0.1 N NaOH

DAX-8

HPO
TPI + HPI
0.1 N NaOH

XAD-4

TPI

HPI

Figure 3.2. EfOM fractionation procedure (three-fraction protocol)

Samples of the raw AS effluent, ozonated effluent, and BAC filtrate were filtered through
0.45 μm cellulose acetate membranes, adjusted to pH 2 with 4.0 N HCl, and then fed onto
columns of Superlite DAX-8 resin, which retained the HPO fraction. This fraction contained

60
mostly very hydrophobic acids (humic and fulvic acids) and some hydrophobic neutrals.
EfOM not adsorbed by DAX-8 resin was fed onto Amberlite XAD-4 resin, which adsorbed
the TPI fraction. This fraction was made up of weakly hydrophobic organic matter. EfOM not
retained by XAD-4 resin was HPI EfOM and consisted of non-humic EfOM such as proteins,
amino acids, and carbohydrates as well as organic colloids (Carroll et al. 2000; Shon et al.
2006; Lozier et al. 2008).

To recover the HPO and TPI fractions of the raw AS effluent to use in MF and UF
experiments, these fractions were desorbed from the resins using 200 mL of 0.1 N NaOH
followed by Milli-Q water so that the total volumes of HPO and TPI fractions collected were
equal to the volumes of the feeds. The recoveries of the HPO and TPI fractions from the
resins ranged from 90% to 97%, thus the concentrations of the collected HPO and TPI
fractions were representative of the HPO and TPI fractions in the non-fractionated effluent.
Before being used in the MF and UF tests, the three EfOM fractions obtained from the raw
AS effluent were diluted with Milli-Q water to a DOC concentration of 1.3 mg L-1, and
adjusted to pH 7.6 and a conductivity of 1060 μS cm-1 (same as for the non-fractionated
sample) using 4.0 N HCl and 4.0 M NaCl solution. The DOC concentration of 1.3 mg L-1 was
selected to ensure that the conductivities of the fractions could be adjusted to 1060 μS cm-1.

It should be pointed out that the resin fractionation method used in this study (as well as in
many previous studies (Nilson & DiGiano 1996; Jarusutthirak et al. 2002; Gray et al. 2004;
Shon et al. 2006)) is not able to achieve 100% recovery of the fractions. The small proportion
of the dissolved EfOM not recovered from the resins may have considerable contribution to
membrane fouling. Furthermore, the pH adjustments may have caused some irreversible
changes in the molecular structures of the EfOM. Therefore, the investigation of the fouling
potentials of the three EfOM fractions was intended to serve as a complementary method for
foulant identification. The identity of the EfOM components responsible for fouling was
further verified using the results obtained from other analyses, including AMWDs of the feeds
and the MF and UF permeates, ATR-FTIR spectra of the EfOM retained on the fouled
membranes (Chapter 4), feed pre-treatment studies (Chapters 5–7), and membrane cleaning
study (Chapter 8).

The four-fraction fractionation protocol (Figure 3.3) described by Chow et al. (2004) was
used to investigate the effect of ozonation and BAC filtration on EfOM properties. In this

61
fractionation scheme, the very hydrophobic acid (VHA) fraction and slightly hydrophobic
acid (SHA) fraction were equivalent to the HPO and TPI fractions, respectively (Carroll et al.
2000; Chow et al. 2004). The HPI fractions in the raw and ozonated effluent and the BAC
filtrate were further separated into hydrophilic charged material (CHA, adsorbed on
Amberlite IRA-958 ion exchange resin) and hydrophilic neutrals (NEU, not adsorbed on IRA-
958 resin).

Whole EfOM, pH 2

0.1 N NaOH

DAX-8

VHA
SHA + CHA + NEU

0.1 N NaOH

XAD-4

SHA
CHA + NEU

IRA-958

NEU
Figure 3.3. EfOM fractionation procedure (four-fraction protocol)

3.2.3. Sample AMWD

Sample AMWD was determined using liquid chromatography with organic carbon detection
(LC-OCD). The instrument was an LC-OCD system (Model 8, DOC-Labor) equipped with an
organic carbon detector (OCD) and a UV detector (UVD). The system design was based on
the Gräntzel thin-film UV reactor (Huber et al. 2011). In this system, organic carbon in the
sample was oxidised to carbon dioxide by UV irradiation at 185 nm with a persulphate

62
solution. The released carbon dioxide was quantified using a non-dispersive infrared detector
(Ultramat 6E, supplied by Siemens, Munich, Germany).

Potassium hydrogen phthalate was used to calibrate the DOC detector. The LC column was
Toyopearl TSK HW-50S (length: 250 mm, internal diameter: 20 mm) filled with 30 μm
hydroxylated methacrylic resin. The mobile phase was a phosphate buffer (2.5 g L-1 KH2PO4
and 1.5 g L-1 Na2HPO4.2 H2O, pH 6.4), running at an isocratic flow rate of 1.1 mL min-1.
Calibration of the OCD and UVD were done using potassium hydrogen phthalate. Suwannee
River Humic Acids Standard II and Suwannee River Fulvic Acids Standard II from the
International Humic Substances Society (IHSS) were used as reference standards for MW of
HS. The sample injection volume was 1000 μL. Samples were filtered (0.45 μm cellulose
acetate) and diluted with Milli-Q water to DOC concentrations below 5 mg L-1 prior to
analysis. The software program ChromCalc (DOC-Labor, Karlsruhe) was used for data
acquisition and processing. ChromCalc divided the chromatograms into 5 different areas
corresponding to biopolymers (which contained polysaccharides and proteins, MW >> 20,000
Da), humic substances (HS, MW = 1,000–20,000 Da), building blocks (breakdown product of
HS, MW = 300–500 Da), low MW organic acids and low MW HS (MW < 350 Da), and low
MW neutrals (mono-oligosaccharides, alcohols, aldehydes and ketones, MW < 350 Da). LC-
OCD analysis gives information regarding the organic materials which are eluted from the LC
column. The DOC that contributes to the chromatograms is called “chromatographic DOC
(CDOC)”. For most water and wastewater samples, a small proportion of the total DOC is
retained on the resin in the LC column as a result of hydrophobic interaction between the
DOC and the resin (Huber et al. 2011). This fraction of the total DOC is called “hydrophobic
organic carbon (HOC)” and usually not considered in studies which utilise LC-OCD analysis
to investigate membrane fouling, as it is not useful for the interpretation of the flux data
(Laabs et al. 2006; Haberkamp et al. 2007; Lozier et al. 2008; Zheng et al. 2010; Henderson
et al. 2011). Therefore, the HOC fractions of the samples analysed with LC-OCD were not
included in this thesis. For these samples, the HOC fractions accounted for 3–10% of the total
DOC (i.e., the sum of CDOC and HOC).

The analysis was carried out at the University of New South Wales, Sydney, Australia. The
effects of coagulation, ozonation, BAC filtration and adsorption pre-treatments with the AER
and the PAC on the AMWD of the EfOM in the raw effluent were tested on two batches of
samples collected on different dates and similar results for each pre-treatment were obtained.

63
According to Huber et al. (2011), LC-OCD analysis has been proven highly sensitive and
reproducible.

3.3. Membrane analysis

3.3.1. ATR-FTIR spectroscopy
A Spectrum 100 FT-IR spectrometer (PerkinElmer) was used to analyse the foulant layer on
the membranes. This spectrometer was equipped with a PerkinElmer® Universal ATR
Accessory with a diamond ATR crystal and utilised the Atmospheric Vapour Compensation
(AVC) technology to eliminate interferences from water vapour and carbon dioxide in the
atmosphere. FTIR spectra of the clean and fouled membranes were collected over the wave
number range of 650–4000 cm-1 using the ATR method. The resolution was adjusted to 4.0
cm-1. The virgin membranes had Milli-Q water passed through them for 45 minutes to remove
preservatives. The fouled membranes were obtained after filtration by the virgin membranes
with the 0.5 μm pre-filtered effluent samples to avoid possible interferences of SS on the
FTIR spectra. Prior to ATR-FTIR analysis, the membranes were dried overnight in a
desiccator at room temperature. Three different areas on each membrane sample were
scanned. The 3 spectra collected for each membrane sample were essentially superimposed.

3.3.2. SEM/EDX spectroscopy

SEM images and energy dispersive X-ray (EDX) spectra of dried virgin and fouled HFM-116
membrane samples were obtained with a microscope (Quanta 200, FEI) equipped with an
EDAX’s Si(Li) X-ray detector. The analysis was carried out at an accelerating voltage of 15
kV and a magnification of 5000 (for SEM images) and 200 (for EDX spectra). The SEM
images and EDX spectra (the latter gave information on the elements present on the
membranes) were collected for 3 different areas on the membrane samples and no differences
between these areas were noticed.

3.3.3. Contact angle measurement

The contact angles of the virgin flat sheet membranes were measured with an OCA20 contact
angle meter (DataPhysics Instruments GmbH) using the sessile drop method with Milli-Q
water. The water droplet volume was 10 μL and the contact angles (computed using the
Young-Laplace equation) were taken within 5 seconds. Five measurements were carried out
for each membrane and the average values were reported (details are included in Appendix
A). Although contact angle measurements are method- and operator-dependent, it allows

64
comparison of the degree of hydrophobicity of different membranes on a relative basis,
provided that the protocol used is well-defined and reproducible (Lozier et al. 2008).

3.4. Dead-end filtration experiments

3.4.1. The membranes used in dead-end filtration
The membranes used in dead-end filtration tests are listed in Table 3.4. The primary focus of
this study was PVDF membranes, as they are the most widely used membranes in MF and UF
of water and wastewater. Flat-sheet and hollow-fibre MF membranes were tested to determine
if membrane configuration had any significant effect on fouling, as a previous study on
surface water found that flat-sheet membranes fouled more rapidly than hollow-fibre
counterparts (Howe et al. 2007). A relatively hydrophilic UF membrane made of proprietary
PES was also tested.

Table 3.5. Characteristics of the polymeric membranes used in this study

Nominal
Pure water flux
pore Contact
Code Supplier Material Configuration (L m-2 h-1 bar-1)
size or angle ()
at 22°C
MWCO
VVLP Millipore PVDF 0.1 μm Flat-sheet 2000  150 622
XUNP-003 Pall PVDF 0.1 μm Hollow-fibre 2200  100 > 90
HFM-180 KMS PVDF 100 kDa Flat-sheet 1000  100 842
HFM-116 KMS PVDF 50 kDa Flat-sheet 150  12 682
HFK-141 KMS PES 30 kDa Flat-sheet 220  20 412

Supplier information, exact value not given

Koch Membrane Systems

The VVLP membrane was made of modified PVDF to increase the hydrophilicity (Millipore
2008). The HFM-180 and HFM-116 membranes were made of proprietary semi-permeable
PVDF. Each XUNP-003 module was made of 32 PVDF membrane fibres (inner diameter =
0.8 mm, outer diameter = 1.0 mm) potted in a PES shell, giving a total filtration area (outside-
in mode) of 0.0082 m2. According to the manufacturer, XUNP-003 is a hydrophobic
membrane with a contact angle greater than 90° (no specific value was provided).

3.4.2. Dead-end filtration system

Filtration experiments with the flat-sheet membranes were carried out using a stirred dead-end
filtration cell (Amicon 8050, effective membrane area 0.00134 m2) in the arrangement shown

65
Figure 3.4. Tests with the MF hollow-fibre (XUNP-003) membrane used the same
arrangement, except that the filtration cell was replaced with the hollow-fibre modules. The
applied TMP was regulated using compressed nitrogen gas. The permeate mass was recorded
every minute using a top-loading electronic balance (Explorer, Ohaus, accuracy ±0.01 g) and
logged by computer and was used to compute the permeate flux.

1. Nitrogen tank
2. Feed reservoir
3. Amicon stirred cell
4. Magnetic stirrer
5. Electronic balance
6. Computer

2 5 6
1

Figure 3.4. Filtration experimental set-up for the flat-sheet membranes

3.4.3. Operating conditions and general procedure

All filtration tests were carried out in the pressurised mode and at a water temperature of 221
ºC. The applied TMP was 70 kPa for MF and 110 kPa for UF, which was regulated using
compressed nitrogen gas. The conditions were selected based on those used by Adham et al.
(2006) and Howe et al. (2007) in bench-scale MF and UF of surface water. For the tests with
the flat-sheet membranes, a stirrer speed of 60 rpm was maintained to keep the cell contents
well mixed. The hollow-fibre membrane modules were run in dead-end outside-in mode.

A new flat-sheet membrane was used for each filtration run. Four hollow-fibre (XUNP-003)
modules were used. The fouled XUNP-003 membranes were regenerated by soaking in a
mixture of NaOH (0.025 N) and NaOCl (200 ppm available chlorine) at 35 ºC for 2 hours.
This chemical cleaning (recommended by the manufacturer) allowed full recovery of the
membrane pure water flux. The cleaned membranes were then used in further tests. During

66
the experimental program, each XUNP-003 module was reused 5–6 times and no
deterioration in the DOC rejection and colour rejection was observed.

Prior to filtration, each membrane was soaked in Milli-Q water for 2 hours and then had 1 L
of Milli-Q water passed through them to remove membrane preservatives (or cleaning agents
in the case of the XUNP-003 membrane). The pure water fluxes ( J 0 ) of the membranes were
obtained by taking the average of the Milli-Q water fluxes in the 8th, 9th, and 10th minutes,
when the membranes could be considered as in quasi-steady state (Fan 2002).

3.4.4. Single-cycle filtration

Single-cycle tests are a quick method to study membrane fouling by water and wastewater.
Filtration tests without intermittent backwashing were conducted with the raw and the pre-
treated effluent samples until the specific permeate volume ( v ) reached 200 L m-2 (MF) and
150 L m-2 (UF). The corresponding filtration time for the raw effluent with the VVLP
membrane ranged from 75 to 120 minutes, depending on the sample batch. These values for
the HFM-180, HFM-116, and HFK-141 were 60–90, 150–170, and 90–110 minutes,
respectively. The feeds and the permeates were collected for further analyses to identify the
membrane foulants (with LC-OCD) and evaluate the rejections of UVA254 and true colour.
The flux data obtained were used to assess the effect of the pre-treatments on the flux
performance. Three parameters were used for this purpose: the normalised flux, the resistance
by total fouling ( Rtf ) and the UMFI for total fouling (sample calculations for these parameters

are shown in Appendix B).

The normalised flux was defined as the ratio of the permeate flux during MF or UF with the
effluent samples ( J ) to the pure water flux of the membrane ( J 0 ). The resistance by total
fouling was calculated as
P
Rtf   Rm (3.1)
J 
where
J  is the permeate flux at the end of the filtration test,
P is the TMP
 is the viscosity of the wastewater (taken as 0.95810-3 Pa.s , i.e., same as the viscosity of
pure water at 22 °C), and

67
Rm is the resistance exerted by the virgin membrane,

P
Rm  (3.2)
J 0
Plots of reciprocal of the normalised flux ( J 0 / J ) versus specific permeate volume ( v ) were
constructed to obtain the UMFIs for total fouling, which are the slopes of the curves. The
UMFI for total fouling was used to quantitatively evaluate the effect of the different feed pre-
treatment methods on the flux performance. The lower the UMFI for total fouling, the better
the flux performance.

Then, to investigate the effect of different feed pre-treatments on hydraulically irreversible

fouling, the fouled membranes were backwashed with the permeates and their pure water
fluxes were measured. The backwash procedure for the flat-sheet membranes was as follows:
(i) Rinsing the membrane surface with Milli-Q water
(ii) Placing the membranes upside down in the filtration cell and filtering the permeate
at the operating pressure for 5 minutes
(iii) Returning the membranes to their processing orientation and measuring the pure
water flux of the backwashed membranes ( J BW ).
Backwashing of the fouled hollow-fibre membrane was done by passing the permeate in the
reverse direction (i.e., inside-out) at the operating pressure (70 kPa) for 5 minutes.

The resistances by irreversible fouling ( Rif ) were calculated using the resistance-in-series

model (Kimura et al. 2004; Al-Amoudi & Lovitt 2007) with the following equation:
P
Rif   Rm (3.3)
J BW

In a separate experiment, the HMF-116 and HFK-141 membranes fouled by the raw AS
effluent were backwashed with Milli-Q water at 110 kPa for 5 minutes. The backwash water
was then subjected to LC-OCD analysis (after being filtered through a 0.45 μm cellulose
acetate membrane) to identify the EfOM components responsible for hydraulically reversible
fouling.

68
3.4.5. Multi-cycle filtration
In real plant operation, MF and UF of secondary effluent are normally carried out in multi-
cycle filtration mode (15–30 minutes filtration followed by backwashing for approximately
10 seconds to a few minutes) (Leslie et al. 1999; Decarolis et al. 2001; Laabs et al. 2006;
Water Environment Federation 2006). Some researchers have suggested that flux data
obtained with virgin membranes may not be representative for every filtration cycle (Howe et
al. 2007). Therefore, in this study the effect of the pre-treatments on membrane fouling was
also investigated with the backwashed membranes. Six filtration cycles, each of which lasted
for 20 minutes and was followed by 1-minute backwashing with the permeate at the operating
pressure, were conducted.

3.5. Chemical cleaning of the fouled polymeric membranes

The aim of this part of the study was to identify the best cleaning agents for the polymeric
membranes fouled by the raw AS effluent and obtain more insights to the interactions
between the foulants, the membranes and the cleaning agents. The 50 kDa PVDF (HFM-116)
membrane was selected for its high retention of the foulants, which facilitated the tracking of
these compounds on the membrane during the fouling and cleaning processes using ATR-
FTIR spectroscopy and SEM/EDX spectroscopy. The three most effective cleaners for the
HFM-116 membrane were also tested on the PES 30 kDa (HFK-141) membrane.

3.5.1. Membrane fouling and cleaning procedure

The virgin membranes were fouled with the raw AS effluent at 110 kPa until the accumulated
specific permeate volume (v) reached 150 L m-2 (which took approximately 150 minutes).
The fouled membranes were then backwashed with the permeate at 110 kPa for 5 minutes and
their pure water flux was measured with Milli-Q water. The backwashed membranes were
then soaked (with gentle shaking on an orbital shaker at 100 rpm) in 75 mL of the cleaning
agent solutions at 24 °C for 45 minutes. The cleaning agents used for cleaning of the fouled
HFM-116 membrane are shown in Table 3.5. These included an acid (HCl), a base (NaOH), a
chelating agent (tetrasodium ethylenediaminetetraacetic acid (Na4EDTA)), two oxidising
agents (NaOCl and H2O2), a surfactant (sodium dodecyl sulphate or SDS), and a protease-
based cleaner (Terg-a-zyme, used widely in the fermentation industry). The concentrations
were selected based on common practice as well as the pH limit (pH 1.5–10.5) and chlorine

69
tolerance (200 ppm) of the HFM-116 membrane. Milli-Q water was used to dissolve/dilute
the cleaning agents to the target concentrations.

Three fouling and cleaning cycles were also conducted with the 3 most effective cleaners on
the HFM-116 membrane to determine if accumulation of residual fouling after chemical
cleaning occurred. These cleaning agents were also tested on the 30 kDa PES (HFK-141)
membrane (pH limit = 2.0–11.0).

Table 3.6. The membrane cleaners investigated

Chemical Concentration pH
Sodium hydroxide (NaOH) 0.3 mM 10.5
Hydrochloric acid (HCl) 10 mM 2.0
Na4EDTA 0.4 mM 10.5

Sodium hypochlorite (NaOCl) 200 ppm as chlorine 10.5*

Hydrogen peroxide (H2O2) 1.0 wt.% 4.5
H2O2/NaOH mixture pH 10.5 1.0 wt.% H2O2 10.5
Sodium dodecyl sulphate (SDS) 10 mM 6.8
Terg-a-zyme 1.0 wt.% 9.6
*
Adjusted with a small amount of HCl

3.5.2. Determination of the cleaning efficiency

Determination of the pure water flux of the chemically cleaned membranes was done after
rinsing them thoroughly with Milli-Q water. The pure water flux recovery ( WFR ) was
defined as the ratio of the pure water flux of the fouled membrane after hydraulic cleaning
( J BW ) or chemical cleaning ( J c ) to that of the virgin membrane ( J 0 ). The effectiveness of

chemical cleaning was evaluated using the cleaning efficiency, E RW (Al-Amoudi & Lovitt
2007), with
Rif  Rres
E RW   100 (3.4)
Rif

where Rres is residual resistance after chemical cleaning,

P
Rres   Rm (3.5)
J c
Sample calculations for WFR and E RW are included in Appendix C.

70
3.5.3. Membrane analysis
Samples of the virgin, fouled, backwashed, and chemically cleaned membranes (after being
dried overnight in a desiccator at room temperature) were examined with ATR-FTIR
spectroscopy and SEM/EDX spectroscopy to determine if the results from these analyses
could provide explanations for the differences in the cleaning efficiencies of the chemicals.

71
3.6. Cross-flow filtration experiments

3.6.1. The cross-flow filtration system and ceramic membranes

Cross-flow filtration tests were carried out with single-channel tubular ceramic membranes
(Table 3.6) in inside-out mode. The inner diameter and outer diameter of the ceramic tubes
were 7.0 mm, and 10.0 mm, respectively. The filtration surface area (based on inner diameter)
was 0.005 m2. The membranes were installed to be vertical in an XLAB5 pilot rig (Figure
3.5). The filtration rig and the membranes were supplied by Pall Corporation.

Table 3.7. Characteristics of the ceramic membranes used in cross-flow filtration experiments
Code Material Pore size Pure water flux* (L m-2 h-1 bar-1)
Ceramic MF -alumina 0.1 μm 1830  90
Ceramic UF Zirconia on alumina 20 nm 340  10
support
*
Measured at 22ºC

Relief valve
Retentate
valve
Permeate valve
P2 BF3
Membrane

Feed
Heat tank
exchanger

Cooling P1
water in
BF3: Back-flush device
Pump P1, P2: Pressure gauge

Figure 3.5. A schematic diagram of the cross-flow filtration XLAB5 pilot unit

The filtration system was equipped with a progressing cavity pump (PCM, France), a heat
exchanger (for maintaining a constant feed temperature), and a back-flush device (BF3). The

72
cross-flow velocity could be varied by adjusting the pump speed. The filtration pressure was
recorded as the average of pressures at the inlet (P1) and outlet (P2) of the membrane module.
The main component of the back-flush device was a cylindrical piston, which was driven by
air pressure. The permeate reservoir volume for back-flushing was 3 mL. The back-flush
frequency and duration were set by a timer. Back-flushing could be applied once every 12
minutes to once every 30 seconds and the back-flush duration could be varied from 0.5 to 12
seconds. When back-flushing was activated, the piston moved rapidly to push the permeate
back to the tube lumen.

3.6.2. Operating conditions for cross-flow filtration tests

All filtration tests were conducted at a water temperature of 221 ºC (which was maintained
by the heat exchanger). The majority of the tests were run at a cross-flow velocity of 1.8 m s-1.
The corresponding Reynolds number was 12000, which was in the turbulent flow region of
the regime at which most commercial membrane modules operate (1000 < Reynolds number
< 15000) (Belfort et al. 1994). Progressing cavity pumps may cause pulsations in the flow due
to internal leakages, which may affect the rate of membrane fouling. However, this problem
was not observed with the pump used in this study, as the TMP remained steady during the
filtration tests. This was probably due to the fact that the pump was new and operated at a
capacity much lower than its maximum capacity (approximately 20% and 30% of the
maximum pumping capacity at cross-flow velocities of 1.8 m s-1 and 2.7 m s-1, respectively).

To align with the current pilot plant practice, tests were conducted at a starting flux (i.e., pure
water flux) of 350 L m-2 h-1 (MF) and 100 L m-2 h-1 (UF). This corresponded to a TMP of 20
kPa for MF and 30 kPa for UF. For comparison, Mueller and Witte (2007) used a silicon
carbide membrane (0.5 μm) to run pilot MF tests with a highly turbid filter backwash water
(turbidity = 197 NTU) at a cross-flow velocity of 3.4 m s-1 and the permeate flux varied from
102 to 280 L m-2 h-1. They also carried out tests in dead-end mode with an alumina UF
membrane (pore size = 50 nm) at fluxes between 100 and 300 L m-2 h-1 and an alumina MF
membrane (0.2 μm) at fluxes of 100–400 L m-2 h-1. In this study, the starting fluxes of 350 L
m-2 h-1 (MF) and 100 L m-2 h-1 (UF) were selected to account for the smaller pore sizes of the
membranes, the much lower turbidity of the AS effluent (1.9–5.6 NTU) than the filter
backwash water, and the (expected) fouling reduction by the tangential flow.

73
Two MF and two UF membranes were used. The fouled membranes were regenerated by
soaking in NaOCl solutions (approximately 1000 ppm available chlorine) at 70C for 45
minutes. This cleaning procedure allowed full recovery of the membrane pure water flux. The
regenerated membranes were used in further experiments. Prior to the tests with the effluent
samples, Milli-Q water was passed through the membranes at 70 kPa for 30 minutes to
remove the membrane preservatives/cleaning agents and then at the operating pressure (20
kPa for MF and 30 kPa for UF) for 15 minutes to determine the pure water flux.

3.6.3. Determination of the membrane foulants and the effect of feed pre-treatments on the
permeate flux in cross-flow filtration
Cross-flow MF and UF tests without intermittent back-flushing were conducted with the raw,
coagulated, ozonated and AER-treated effluent samples to determine the effect of these pre-
treatments on the flux performance. The coagulant tested was alum at 2.5 mg L-1. The ozone
and AER dosages were 10 mg O3 L-1 with 10-minute contact time and 10 mL AER L-1 with
20-minute contact time (1 bed volume of the AER treated 1500 bed volumes of the raw
effluent), respectively. Each MF and UF run lasted for 2 hours and the corresponding VCF
was 1.3 for MF and 1.2 for UF. The feeds and the permeates were analysed to identify the
foulants (with LC-OCD) and evaluate the rejections of DOC, UVA254 and true colour by the
membranes.

3.6.4. Effect of rapid membrane back-flushing on the flux performance

The effectiveness of rapid back-flushing on fouling reduction was investigated with the raw,
coagulated, and ozonated effluent. Membrane backwashing (or back-flushing) results in
additional equipment and energy costs and a loss in permeate production (Bacchin et al.
2006). These costs should be minimised, particularly in MF and UF of secondary effluent,
where the end product has a relatively low economic value. Therefore, this study focussed on
the use of rapid back-flushing (duration = 1 second) in cross-flow filtration as an analogue of
backwashing in dead-end filtration (note that in dead-end filtration, backwashing normally
lasts from 10 seconds to several minutes). Back-flushing was applied to MF and UF at the
lowest frequency (12 min-1) and the lowest feed cross-flow velocity (1.8 m s-1) available to
the filtration system. Some tests (without back-flushing) were also carried out at a cross-flow
velocity of 2.7 m s-1 to compare the effect of increased cross-flow velocity and rapid back-
flushing on the flux performance.

74
3.7. Reproducibility of filtration experiments
The reproducibility of the flux curves was tested by performing duplicate experiments with
the raw AS effluent of the same batch (Figures 3.6 and 3.7). The nearly superimposed curves
indicate good reproducibility and any difference in the physical structure among the
membrane samples of the same type had negligible effect on the filtration result. For the
polymeric MF membranes, the deviations in the normalised flux (i.e., ratio of the pure water
flux of the virgin membrane to the flux of the fouled membrane) between the duplicate runs
were less than 5%. For the other membranes, these values were less than 3%.

1
Raw + VVLP (test 1)
(a)
Raw + VVLP (test 2)
0.8
Raw + XUNP-003 (test 1)
J/J0

Raw + XUNP-003 (test 2)

0.6

0.4

0.2

0
0 50
v (L m ) 100 150 200
-2

1
Raw + HFM-180 (test 1)
(b)
Raw + HFM-180 (test 2)
0.8 Raw + HFM-116 (test 1)
Raw + HFM-116 (test 2)
Raw + HFK-141 (test 1)
0.6 Raw + HFK-141 (test 2)
J/J0

0.4

0.2

0
0 50 -2
v (L m ) 100 150

Figure 3.6. Reproducibility of (a) MF and (b) UF flux data for dead-end filtration

75
 1
Raw + MF (test 1)
Raw + MF (test 2)
0.8
Raw + UF (test 1)
Raw + UF (test 2)
0.6
J/J0
0.4

0.2

0
0 50 -2 100 150 200
v (L m )

Figure 3.7. Reproducibility of flux data for cross-flow filtration

(Cross-flow velocity = 1.8 m s-1, TMP = 20 kPa for MF and 30 kPa for UF)

Tests on the effects of the pre-treatments on the permeate flux (and effluent characteristics)
were performed in duplicate using samples of the same batch and the results were confirmed
with a minimum of one batch of samples collected on another day. The daily fluctuations in
the characteristics of the raw effluent resulted in differences in the flux data for samples
collected on different days. However, for each of the feed pre-treatment methods, the same
effect on membrane fouling was observed for samples of different batches. On some
occasions, this was further confirmed by the results obtained from multi-cycle experiments
using the same samples.

3.8. Analysis of flux data for the identification of fouling mechanisms

Identification of the fouling mechanisms in dead-end MF and UF was done by fitting the flux
data to equations 2.2–2.5. Attempts were also made to identify the fouling mechanisms in
cross-flow filtration using the integral method for flux data analysis described by Field and
Wu (2011). This method is based on equation 2.6 with the assumption that fouling is
governed by complete pore blocking (n = 2) or pore adsorption (n = 1.5) in the initial period
of filtration and then by intermediate pore blocking (n = 1) or cake filtration (n = 0).
However, it was realised that the application of this method to the flux data obtained from the
cross-flow filtration system used in this study was not theoretically robust. The reason was
that the formation of a cake layer would eventually reduce the internal diameter of the
membrane tube and therefore increase the cross-flow velocity of the feed. As a result, the

76
accuracy of the calculation of the critical fluxes using the equations derived theoretically by
Field and Wu (2011) (equations 3.6 and 3.7) could not be assured. Consequently, the fouling
mechanisms in cross-flow filtration of the secondary effluent were not determined in this
study.

ln J 0 J  v
n = 1: = K n 1    K n 1 J n1 (3.6)
t t
1 1 v
n = 0:  = K n 0    K n 0 J n0 (3.7)
Jt J 0 t t
where
J n1 is the critical flux for the intermediate pore blocking model,

J n0 is the critical flux for the cake filtration model, and

K n is a constant.

77
 CHAPTER 4. DEAD-END FILTRATION OF THE RAW ACTIVATED SLUDGE
EFFLUENT

The overall objective of this study was to investigate the effects of different feed pre-
treatment methods on membrane fouling in MF and UF of the effluent. To understand why
one pre-treatment reduced while others worsened or did not affect fouling, knowledge of the
fouling behaviour of the raw (i.e., non-pre-treated) effluent was needed. Therefore, this part of
the study was dedicated to fouling in dead-end MF and UF of the raw AS effluent. The
components responsible for fouling and the predominant fouling mechanisms were identified.
Because the raw effluent contained SS (defined as the particulate matter retained by a 0.5 μm
glass-fibre filter in this study) and EfOM, it was necessary to determine the relative
contributions of these components to the flux decline. The rejections of DOC, UV-absorbing
and colour-bearing compounds of the membranes were also evaluated.

Dead-end filtration experiments were conducted using PVDF (VVLP, XUNP-003, HFM-180
and HFM-116) and PES (HFK-141) membranes and the raw AS effluent with and without SS.
The EfOM components responsible for fouling were identified by analysing the feed and the
permeates with LC-OCD. ATR-FTIR spectroscopy was used to identify the compounds
retained on the fouled membranes. The relative fouling potentials of the three fractions of the
EfOM (i.e., the HPO, TPI and HPI fractions) were also determined. These fractions were
obtained from the resin fractionation process described in Figure 3.2 (Chapter 3, section
3.2.2). The fractions were diluted with Milli-Q water to a DOC concentration of 1.3 mg L-1,
and adjusted to pH 7.6 and a conductivity of 1060 μS cm-1 (same as for the non-fractionated
EfOM) prior to the MF and UF tests.

The flux data obtained from MF and UF of the raw effluent samples were analysed using the
dead-end filtration models to determine the predominant fouling mechanisms in MF and UF.

78
4.1. Effect of suspended solids (SS) on flux decline in MF and UF
The flux data for the raw AS effluent before and after the removal of SS are shown in Figure
4.1.

1 Whole sample + VVLP

(a)
0.5 micron pre-filtered + VVLP
0.8 Whole sample + XUNP-003
0.5 micron pre-filtered + XUNP-003

0.6
J/J0

0.4

0.2

0
0 100 v (L m )
-2 200 300

1
Whole sample + HFM-180
(b)
0.5 micron pre-filtered + HFM-180
0.8 Whole sample + HFM-116
0.5 micron pre-filtered + HFM-116
Whole sample + HFK-141
0.6
0.5 micron pre-filtered + HFK-141
J/J0

0.4

0.2

0
-2
0 50 v (L m ) 100 150
Figure 4.1. Effect of SS on (a) MF flux and (b) UF flux of the raw AS effluent
(Raw effluent collected on 26 November 2009, DOC = 14.3 mg L-1,
turbidity = 3.8 NTU, UVA254 = 0.406 cm-1, true colour = 103 Pt-Co units)

The removal of SS from the raw AS effluent reduced the rate of flux decline on the MF
(VVLP and XUNP-003) and the loose UF (HFM-180) membranes, but had no effect on the
tighter UF membranes (HFM-116 and HFK-141). The results indicated that particles larger
than 0.5 μm made some contribution to the flux reduction on the MF (VVLP and XUNP-003)

79
and the loose UF (HFM-180, 100 kDa) membranes. This could be attributed to the high initial
permeability of these membranes (i.e., low membrane resistance, Table 4.1), which resulted in
the hydraulic resistance exerted by the cake layer formed by the SS on the membrane surfaces
making up a significant proportion of the total hydraulic resistance during MF and UF. On the
other hand, the high hydraulic resistance of the tight UF membranes made the resistance
exerted by the SS particles minor compared with the total resistance caused by these
membranes and the EfOM.

Table 4.1. Hydraulic resistances for MF and UF of the raw effluent with and without SS

Rm Rtf * Rif *
Case
( 10-10 m-1) ( 10-10 m-1) ( 10-10 m-1)
Whole sample + VVLP 18.1 471 11.3
0.5 μm pre-filtered sample + VVLP 17.2 66.7 10.5
Whole sample + XUNP-003 15.9 673 36.0
0.5 μm pre-filtered sample + XUNP-003 15.8 373 37.7
Whole sample + HFM-180 41.0 472 55.1
0.5 μm pre-filtered sample + HFM-180 40.9 359 51.4
Whole sample + HFM-116 242 702 183
0.5 μm pre-filtered sample + HFM-116 237 677 172
Whole sample + HFK-141 173 548 87.2
0.5 μm pre-filtered sample + HFK-141 169 515 91.9
* -2 -2 -2
Calculated for v = 300 L m (VVLP), 200 L m (XUNP-003) and 150 L m (the other membranes)

It was noticed that for the MF and the loose UF membranes, although the total fouling ( Rtf )

caused by the sample without SS was significantly lower than that caused by the whole raw
effluent sample, the extent of hydraulically irreversible fouling caused by these two samples
were fairly similar (Table 4.1). This indicated that most of the flux decline caused by the SS
was hydraulically reversible, since backwashing removed the majority of the SS deposited on
the membrane surfaces.

The removal of SS from the raw AS effluent resulted in a larger flux improvement in MF with
the flat sheet membrane (VVLP) than with the hollow-fibre membrane (XUNP-003),
suggesting that membrane configuration may have some influence on the results obtained
from bench-scale experiments, although similar trends could be expected. Similar results were
reported by Howe et al. (2007) for MF and UF of surface water. Other factors which may

80
contribute to this difference included differences in the degree of hydrophobicity (contact
angle of the VVLP membrane = 62º, whereas that of the XUNP-003 was greater than 90º) and
differences in the surface and pore morphology of the two membranes.

4.2. Identification of the EfOM components responsible for fouling

The EfOM components responsible for fouling were identified using 3 different techniques:
LC-OCD analysis, fouling potentials of the HPO, TPI and HPI fractions, and ATR-FTIR
spectroscopy.

4.2.1. AMWDs of dissolved EfOM in the feed and the permeates

The feed (after pre-filtration through a 0.45 μm cellulose acetate membrane) and the
permeates collected from the filtration tests with the 0.5 μm pre-filtered sample were analysed
with LC-OCD to determine the components retained on the MF and UF membranes. The
chromatograms obtained (Figure 4.2) showed that biopolymers (MW>>20,000 Da), which
included polysaccharides and proteins, and HS were the compounds most significantly
retained by the membranes and therefore potentially responsible for fouling. Similar results
were obtained in recent studies utilising LC-OCD for the identification of the membrane
foulants in UF of secondary effluent (Zheng et al. 2009; Henderson et al. 2011). It should be
noted that materials retained by membranes are not necessarily the membrane foulants. In this
thesis, the role of biopolymers in causing membrane fouling was further verified by the fact
that the feed pre-treatments which led to the removal of biopolymers (coagulation, Chapter 5)
or breakdown in the molecular structures of biopolymers (ozonation, Chapter 6) significantly
reduced membrane fouling. On the other hand, the contribution of HS to fouling was
supported by the fact that cleaning the fouled membranes (after backwashing) with sodium
dodecyl sulphate and Terg-a-zyme gave high pure water flux recoveries and the cleaning
solutions turned yellow (an indication of the presence of HS) after use (Chapter 8).

Polysaccharides, due to their neutral charges, were likely to foul the membranes mainly by
pore blocking and cake formation. Adsorption of some polysaccharides on the membranes
was also possible (Jermann et al. 2007). Proteins are amphoteric (possessing both negatively
and positively charged functional groups) and therefore can interact with the membranes
through dipole interactions or by forming a cake layer on the membrane surfaces (Lozier et al.

81
2008). Some proteins can be hydrophobic in nature due to the presence of hydrophobic amino
acids (e.g., proline) in their structures and can adsorb on the membranes through hydrophobic
attraction. This hypothesis is supported by the presence of a small proportion of biopolymers
in the HPO and TPI fractions extracted from the raw effluent (Section 4.2.3).

HS Building blocks
12

Low MW HS and low MW acids

OCD response (relative units)

10
Biopolymers
Raw
8 Raw + VVLP
Raw + XUNP-003
6 Raw + HFM-180

4 Low MW neutrals

0
20 40 60 80 100
Retention time (min)

HS Building blocks
12

(b) Low MW HS and low MW acids

OCD response (relative units)

10
Raw
8 Raw + HFM-116
Raw + HFK-141
HFM-116 backwash water
6 Biopolymers HFK-141 backwash water

4
Low MW neutrals
2

0
20 40 60 80 100
Retention time (min)

Figure 4.2. AMWDs of the EfOM in the raw effluent,

the permeates and the backwash water
(a) MF and loose UF membranes (feed: DOC = 14.2 mg L-1, UVA254 = 0.343 cm-1) and
(b) Tight UF membranes (feed: DOC = 13.2 mg L-1, UVA254 = 0.402 cm-1)
(The dotted lines define the areas under the peaks, the backwash water samples were
diluted with Milli-Q water to a volume same as the permeate volume)

82
Some HS (MW=1,000–20,000 Da) were also retained, most likely by hydrophobic
interaction with the membranes (Liu et al. 2001; Jarusutthirak et al. 2002). This was
supported by the higher HS retention of the XUNP-003 membrane (more hydrophobic)
than the VVLP membrane, and the lower HS retention of the HFK-141 membrane
(relatively hydrophilic and smaller nominal pore size) than the HFM-116 and HFM-180
membranes (hydrophobic and larger pore size). This trend in HS rejection was in agreement
with the trend in colour removal, i.e., the less hydrophobic membranes gave lower colour
removals (see Section 4.3). HS were found to be responsible for the colour of the raw AS
effluent as the true colour of the HPO fraction accounted for approximately 80% of the true
colour of the non-fractionated EfOM.

The role of HS in causing adsorptive fouling was supported by two other phenomena. First,
the colour rejection by the membranes was observed from the first minute of filtration.
Second, cleaning the fouled membranes with sodium dodecyl sulphate and Terg-a-zyme
gave high pure water flux recoveries and the cleaning solutions turned yellow after use
(Chapter 8).

The higher retention of biopolymers than HS (Figure 4.2) indicated that proteins and
polysaccharides played a more important role than HS in causing fouling. Pore adsorption
and pore blocking by these components were responsible for irreversible fouling, since
backwashing removed a proportion of the biopolymers and essentially no HS from the
fouled membrane (Figure 4.2(b), the backwash water chromatograms).

In summary, LC-OCD analysis of the feed and the permeates showed that biopolymers
(i.e., proteins and polysaccharides) and HS were the EfOM components responsible for
membrane fouling. HS contributed mainly to hydraulically irreversible fouling, whereas
biopolymers contributed to both reversible and irreversible fouling.

4.2.2. Foulant identification by ATR-FTIR spectroscopy

To further confirm the results obtained from LC-OCD analysis, the ATR-FTIR spectra of
the virgin membranes and the membranes fouled by the 0.5 μm pre-filtered effluent were
collected to identify the EfOM components retained on the membranes. As seen in Figures
4.3, the foulants on the membranes showed peaks around 1069 cm-1, indicating the

83
presence of C–O stretching of polysaccharides and/or polysaccharide-like materials (Cho et
al. 1998, Jarusutthirak et al. 2002). The peaks around 1720 cm-1 were indications of
carboxylic groups in HS. The peaks at 1540 cm-1 suggested the existence of N–H bonds and
C=N stretching of secondary amides in the foulant layer. The peaks around 1640 cm-1
corresponded to stretching vibration of C=O connected to primary amides of proteins. The
peaks near 2920 cm-1 were attributed to aliphatic C–H stretching of HS. The broad bands
around 3000-3650 cm-1 were due to O–H stretching of hydroxyl groups (in
polysaccharides) and N–H stretching of amides (Cho et al. 1998; Howe et al. 2002;
Jarusutthirak et al. 2002). Consequently, it was concluded that proteins, polysaccharides
and HS were the major membrane foulants in MF and UF of the raw effluent.

84
 0.3
1069 (a)
Virgin VVLP
0.25
Fouled VVLP
0.2 1640
Absorbance
0.15 1540

0.1 1720
3000-3650
2920
0.05

0
500 1000 1500 2000 2500 3000 3500 4000
-1
-0.05 Frequency (cm )

0.2 1060 (b)

Virgin HFM-180

0.15 Fouled HFM-180

1640
Absorbance

1540 3000-3650
0.1
2920
1720
0.05

0
500 1000 1500 2000 2500 3000 3500 4000
-1
-0.05 Frequency (cm )

0.65
(c) Virgin HFK-141
0.55 1540
Fouled HFK-141
1070
0.45
1640
Absorbance

0.35

0.25
1720
3000-3650
0.15
2920
0.05

-0.05 500 1000 1500 2000 2500 -1 3000 3500 4000

Frequency (cm )

Figure 4.3. ATR-FTIR spectra of the virgin and fouled membranes

(a) VVLP, (b) HFM-180, and (c) HFK-141

85
4.2.3. Fouling potentials of individual EfOM fractions

The EfOM fractions obtained from resin fractionation of the raw effluent (using the
procedure described in Figure 3.2, Section 3.2.2) were also used in MF and UF tests to gain
more insights into the characteristics of the membrane foulants. Prior to the filtration tests,
the HPO, TPI and HPI fractions of the raw effluent were diluted with Milli-Q water to a
DOC concentration of 1.3 mg L-1, and adjusted to pH 7.6 and a conductivity of 1060 μS
cm-1 (same as for the non-fractionated sample). The flux data obtained are shown in Figure
4.4.

1 1
VVLP HFM-180
0.9 HPO HPO
TPI TPI
0.8 0.8
HPI J/J 0 HPI
0.7
J/J0

0.6
0.6

0.5
0.4
0.4

0.3
0.2
0 150 300 -2 450 600 750
0 150
v (L m ) 300 450
-2
v (L m )

1 1 HFK-141
HPO
HFM-116 TPI
0.9 HPI 0.9
J/J0
J/J0

0.8 0.8

HPO
0.7 TPI
0.7
HPI
0.6
0.6
-2 0 50 100 -2 150 200 250
0 50 v (L m ) 100 150 v (L m )

Figure 4.4. Flux data for MF and UF of different fractions in the raw AS effluent
(DOC of each fraction = 1.3 mg L-1)

For this AS effluent, the HPI fraction had the highest fouling potential for all the
membranes investigated (Figure 4.4). A similar finding for natural surface waters was
reported by Carroll et al. (2000) and Lee et al. (2004). This finding, however, is different
from the result obtained by Shon et al. (2006) who found that the HPI fraction of an AS
effluent had the lowest UF fouling potential. Possible reasons for this difference include

86
differences in the properties of the EfOM of the two samples and the membranes (Shon and
co-workers used 17,500 Da sulfonated polysulfone membranes). The higher fouling
potential of the HPI fraction than the other two fractions could be attributed to the fact that
the majority of polysaccharides and proteins in the raw effluent were concentrated in this
fraction (Figure 4.5). The components with MW between 1,000 Da and 20,000 Da in the
HPI fraction are not likely to be HS due to their very low UV response (at 254 nm) and the
fact that the HPI fraction was colourless (true colour = 0). The presence of a small
proportion of biopolymers in the HPO and TPI fractions indicated that some proteins and/or
polysaccharides in this AS effluent were hydrophobic and therefore could contribute to
adsorptive fouling of the membranes.

The HPO fraction was the most hydrophobic fraction, followed by the TPI fraction, as
indicated by their UV absorbance (Figure 4.5 (a)). Adsorption on the hydrophobic
membranes (via hydrophobic attraction) was therefore expected to be strongest for the HPO
fraction. In addition, the fact that the HPO fraction contained some HS of higher MW than
the TPI fraction (as indicated by the lower retention time of the HS peak on the HPO curve
in Figure 4.5 (b)) may have contributed to the higher fouling potential of the HPO fraction,
since higher MW compounds are likely to narrow the membrane pores more rapidly.
Although adsorptive fouling by the HPI fraction was expected to be lower than that by the
other two fractions (as indicated by the lower UVA254 and therefore lower hydrophobicity
of the HPI fraction), the HPI fraction showed the highest fouling potential. This can be
explained by the fact that pore adsorption was dominant during the initial period of
filtration (as shown in Section 4.4); however, as filtration proceeded, the cake filtration
mechanism, in which polysaccharides and proteins formed a cake layer on the membrane
surface, became the governing fouling mechanism.

The fact each EfOM fraction contained compounds of various MWs indicated that the
fractions were not “pure” in terms of molecular sizes of the constituents. This is probably
one of the reasons for the different results reported in the literature regarding the relative
fouling potentials of the different EfOM (and NOM) fractions isolated by resin
fractionation.

87
 0.18 HS
(a)
0.16 Building blocks

0.14 Low MW HS and

UVD response (relative units)

HPO
low MW acids
0.12 TPI
HPI
0.1
26 28 30 32
0.08

0.06
Low MW neutrals
0.04
Biopolymers
0.02

0
20 40 60 80 100
Retention time (min)

1.2 Building blocks

(b) Low MW neutrals

HS
1
OCD response (relative units)

Low MW HS and
0.8
low MW acids
HPO
TPI
0.6
HPI

0.4 Biopolymers

0.2

0
20 40 60 80 100
Retention time (min)

Figure 4.5. LC-OCD chromatograms of the three EfOM fractions of the raw AS effluent
(a) UVD response and (b) OCD response (DOC of each fraction = 1.3 mg L-1)

88
4.3. DOM rejection and colour removal by the membranes
The DOM rejection and true colour removal by the membranes in MF and UF of the raw
AS effluent are given in Table 4.2. The results were collected from filtration tests with 5
samples collected from June 2008 to June 2010. The turbidity of the permeates was
consistently lower than 0.2 NTU, much lower than the requirement of 2.0 NTU for “Class
A” recycled water.

Table 4.2. DOM rejection and colour removal by the polymeric membranes
Membrane % DOC removal % UVA254 removal % colour removal
VVLP 4 – 9% 2 – 6% 3 – 8%
XUNP-003 8 – 16% 7 – 13% 15 – 22%
HFM-180 10 – 17% 15 – 21% 24 – 33%
HFM-116 15 – 18% 21 – 25% 35 – 43%
HFK-141 9 – 18% 8 – 12% 9 – 15%

The low DOC removals by the membranes are consistent with previous studies which used
membranes of comparable pore sizes to filter surface water (Adham et al. 2006; Lozier et
al. 2008; Byun et al. 2011) and municipal secondary effluent (Laabs et al. 2006;
Haberkamp et al. 2008; Nguyen et al. 2009; Zheng et al. 2010).

89
4.4. Flux decline and fouling mechanisms in MF and UF of the raw effluent

4.4.1. Flux decline and fouling mechanisms in MF

As shown in Figure 4.1, rapid fouling of the membranes occurred within the first few
minutes of filtration. Plots of t / V versus V were constructed to determine if fouling of the
MF membranes by the raw AS effluent could be described by the cake filtration model.

30000

25000

20000
t /V (s/L)

15000

10000

5000 Raw + VVLP

Raw + XUNP-003
0
0 0.05 0.1 0.15 0.2 0.25 0.3
V (L)

Figure 4.6. Plots of t/V versus V for MF of the raw AS effluent

(Data for XUNP-003 were normalised to a membrane area of 0.00134 m2)
(Raw effluent: DOC = 14.2 mg L-1, turbidity = 3.8 NTU,
UVA254 = 0.343 cm-1, colour = 90 Pt-Co units)

The non-linear sections of the curves in Figure 4.6 suggested that fouling in MF of the raw
effluent was not governed by the cake filtration mechanism in the first 20 minutes. Because
the effluent contained EfOM with various molecular sizes (from smaller than 350 Da to
much larger than 20,000 Da, as shown in the LC-OCD chromatogram in Figure 4.2) as well
as SS, it was highly likely that pore adsorption and pore blocking occurred simultaneously
in this initial period (Bowen et al. 1995; Duclos-Orsello et al. 2006). To identify the
dominant fouling mechanism in this period, the flux data were analysed using the equations
2.2–2.4 and the results are shown in Figure 4.7. The linearity of the plot of t V versus t
confirmed that pore adsorption played a significant role during the first 20 minutes of MF.

90
 25
Raw + VVLP
20 Raw + XUNP-003

V (mL/min) 15

10

5
(a)

0
0 20 40 60 80 100 120 140
V (mL)

0.16

0.14
(b)

0.12
t /V (min/mL)

0.1

0.08

0.06

0.04
Raw + VVLP
0.02
Raw + XUNP-003
0
0 5 10 15 20
t (min)

0.5
(c)
0.4
1 / V (min/mL)

0.3

0.2

0.1 Raw + VVLP

Raw + XUNP-003
0
0 5 10 15 20
t (min)

Figure 4.7. Filtration law analysis for the MF membranes

(a) Complete pore blocking, (b) standard pore blocking, and (c) intermediate pore blocking
(t : filtration time, V: permeate volume,V : permeate volume per minute)

91
The best description of the sequence for MF fouling by the raw AS effluent was
(i) Adsorption of particles smaller than the membrane pores (including
polysaccharides, proteins, and HS) on the internal pore walls (standard
blocking) and some contribution of intermediate pore blockage during the first
15–20 minutes, followed by
(ii) Formation of a cake layer from particles larger than the pores, which included
proteins, polysaccharides, and particles larger than 0.45 μm.
(iii) The cake layer formed then acted as a secondary membrane for filtration of the
rest of the sample

This implies that in an operation protocol similar to that used in MF plants, i.e.,
filtration for 15–30 minutes followed by backwashing, the governing fouling
mechanism in the first filtration cycles for the virgin (or chemically cleaned)
membranes would be the pore adsorption mechanism with some contribution of pore
blocking. In the subsequent cycles, fouling (of the backwashed membranes) would be
dominated by the cake filtration mechanism.

4.4.2. Flux decline and fouling mechanisms in UF

The UF flux data were also analysed using same approach used in MF. The plots of t / V
versus V for UF of the raw AS effluent are shown in Figure 4.8. The deviation from
linearity of the first few data points on the curves for the HFM-180 and HFM-116
membranes indicated fouling of these membranes was not controlled by the cake filtration
mechanism in the first 5 minutes of filtration. This can be explained as during this time
pore blocking and pore adsorption on these membranes were significant. The contribution
of pore adsorption to the fouling of these hydrophobic membranes was supported by their
relatively high HS rejection (Figure 4.2 (b)) and high colour removal (Table 4.2) observed
from the beginning of filtration.

The early dominance of the cake filtration mechanism in UF was attributed to the small
pore size of the UF membranes. Fouling of the hydrophilic HFK-141 membrane could be
described by the cake filtration mechanism from the beginning of filtration, probably
because HS adsorption on this membrane was much less significant than the deposition of
polysaccharides and proteins on the membrane surface. This was supported by the lower

92
HS retention (as shown in Figure 4.2), lower colour removal (Table 4.2), and the higher
pure water flux recovery by backwashing of the HFK-141 membrane (approximately 67%)
than the HFM-116 membrane (56%).

48000

40000

32000
t/V (s/L)

24000

16000
Raw + HFM-180
8000 Raw + HFM-116
Raw + HFK-141
0
0 0.05 0.1 V (L) 0.15 0.2 0.25 0.3

Figure 4.8. Plots of t/V versus V for UF of the raw AS effluent

Chapter 4 Summary
This part of the study investigated fouling of the PVDF and PES membranes in dead-end
MF and UF of the raw AS effluent. It was found that both SS and EfOM contributed to the
flux decline in filtration of the raw AS effluent with the MF and the 100 kDa UF
membranes. However, SS did not contribute to flux decline of the tighter UF membranes
(HFM-116 and HFK-141).

The results from LC-OCD analysis of the feed and the permeates showed that biopolymers,
which contained proteins and polysaccharides, and HS were the EfOM components
responsible for membrane fouling. This was supported by the ATR-FTIR spectra of the
organic matter retained on the fouled membranes. As the membranes investigated retained a
higher amount of biopolymers than HS, it was concluded that the contribution of
biopolymers to membrane fouling was greater than that of HS. Furthermore, filtration tests
with the EfOM fractions isolated from the raw effluent revealed that the HPI fraction
(which contained the majority of the biopolymers of the non-fractionated EfOM) caused the
most severe flux decline, followed by the HPO fraction and the TPI fraction. The role of

93
biopolymers in causing membrane fouling was also supported by the fact that the feed pre-
treatments which led to the removal of biopolymers (coagulation, Chapter 5) or breakdown
in the molecular structures of biopolymers (ozonation, Chapter 6) significantly reduced
membrane fouling.

Backwashing removed a proportion of the biopolymers and essentially no HS from the

fouled membranes. Therefore, it was concluded that polysaccharides and proteins
contributed to both reversible and irreversible fouling, whereas fouling caused by HS was
mainly irreversible. The role of HS in causing hydraulically irreversible fouling was further
confirmed (in Chapter 8) by the fact that cleaning the fouled membranes with sodium
dodecyl sulphate and Terg-a-zyme achieved high pure water flux recoveries and the
cleaning solutions turned yellow (an indication of the presence of HS) after use.

Results from flux data analysis using the dead-end filtration models indicated that MF
fouling by the raw AS effluent was controlled by the pore adsorption mechanism in the first
20 minutes of filtration and then followed the cake filtration model. Fouling of the
hydrophobic UF (HFM-180 and HFM-116) membranes was dominated by the pore
adsorption mechanism in the first 5 minutes of filtration, whereas fouling of the relatively
hydrophilic UF (HFK-141) membrane could be described by the cake filtration model
throughout the course of filtration.

94
 CHAPTER 5. EFFECT OF COAGULATION ON MEMBRANE FOULING IN
DEAD-END FILTRATION

As seen in Chapter 4, MF and UF of the raw effluent experienced rapid flux decline which
needed to be mitigated to improve the water productivity. The most widely used method for
fouling mitigation in MF and UF of secondary effluent is coagulation (Water Environment
Federation 2006). Therefore, in this chapter the effect of this pre-treatment on membrane
fouling in dead-end MF and UF of the raw AS effluent was investigated. Coagulation with
alum and ACH was tested. ACH was considered for several reasons. First, practical
experience in drinking water treatment plants indicates that ACH is effective in coagulation
over a broader pH range (5.0 – 8.0) compared with alum (5.0 – 6.5), the most widely used
coagulant. The pH of the raw AS effluent in this wastewater treatment plant varied from 7.2
to 7.8. Therefore, the use of ACH may result in better membrane flux and not necessitate
preliminary pH adjustment. Second, ACH does not contain sulphate ions, which reduces
the risk of odour and corrosion problems in the treatment system. These problems are more
likely to occur if alum is used, since an increase in sulphate concentration of the wastewater
will increase hydrogen sulphide production (Gebbie 2001). Third, ACH is known to
consume much less alkalinity than alum (Duan & Gregory 2003). Coagulation with ACH
will therefore result in a smaller reduction in pH of the AS effluent and may eliminate the
need for pH adjustment of the MF/UF permeates. Also, coagulation with poly(aluminium)
coagulants has been found to produce less sludge than coagulation with alum (Jiang &
Graham 1996; Gebbie 2001). ACH is a pre-hydrolysed inorganic coagulant used in some
drinking water treatment plants. However, this coagulant has rarely been used in
wastewater treatment and therefore few studies on its effectiveness on reduction of
membrane fouling by secondary effluents have been conducted.

Following common practice in municipal wastewater treatment, which tends to use in-line
coagulation with low coagulant doses (Decarolis et al. 2001; Qin et al. 2004), and the
recommendation by the operators at the wastewater treatment plant, alum and ACH were
tested at relatively low doses: 1.0 and 2.5 mg Al3+ L-1. The coagulated effluent samples
were immediately subjected to MF or UF without prior removal of the coagulated flocs to
simulate in-line coagulation. Filtration tests were conducted with the raw and the
coagulated samples using the hydrophobic PVDF (VVLP, XUNP-003, HFM-180 and
HFM-116) membranes and the PES (HFK-141) membrane. The flux data were collected

95
and analysed to evaluate the effect of coagulation on the flux performance. The EfOM of
the raw and the coagulated effluent were analysed with LC-OCD to determine the reasons
for the flux data observed.

In wastewater treatment plants, MF and UF are usually conducted in multi-cycle mode in

which the membrane is backwashed with the permeate 15–30 minutes (Leslie et al. 1999;
Decarolis et al. 2001; Laabs et al. 2006; Water Environment Federation 2006).
Furthermore, some researchers have suggested that flux data obtained with virgin
membranes may not be representative for every filtration cycle (Howe et al. 2007).
Therefore, the effect of coagulation on membrane fouling was also investigated with the
backwashed membranes. Six filtration cycles, each of which lasted for 20 minutes and was
followed by 1-minute backwashing with the permeate at the operating pressure, were
conducted.

5.1. Effect of coagulation on membrane fouling in single-cycle MF and UF

5.1.1. Single-cycle MF

Single-cycle MF and UF tests were conducted with the raw and coagulated effluent with
the characteristics shown in Table 5.1. Coagulation (performed in the jar tester using the
procedure described in section 3.1.2) with either alum or ACH at 2.5 mg Al3+ L-1 removed
approximately 10% of the DOC and 20% of the true colour from the raw effluent.

Table 5.1. Characteristics of the raw and the coagulated effluent samples used in single-
cycle dead-end filtration tests
(Sample collected on 19 February 2009)
Raw + 1.0 mg Raw + 2.5 mg Raw + 2.5 mg
Parameter Raw
Al3+ (ACH) L-1 Al3+ (ACH) L-1 Al3+ (alum) L-1
pH 7.8 7.8 7.6 7.2
DOC (mg L-1) 14.2 13.3 12.9 12.7
UVA254 (cm-1) 0.343 0.324 0.299 0.295
SUVA (L m-1 mg-1) 2.42 2.44 2.32 2.32
True colour (Pt-Co units) 90 79 72 70
Turbidity (NTU) 5.2 N/M* N/M* N/M*
TSS (mg L-1) 6.2 N/M* N/M* N/M*
* Not measured

96
The flux data for MF of the raw and coagulated effluent samples are shown Figures 5.1 and
5.2 for the VVLP and XUNP-003 membranes, respectively. To compare the fouling
mechanisms of the different samples, plots of t V versus V were compared. The linearity
of the curves for the coagulated samples indicated that unlike MF fouling by the raw
effluent, MF fouling by the coagulated samples was dominated by the cake filtration
mechanism from the beginning of filtration (Figures 5.1 (a) and 5.2 (a)). This can be
explained by the addition of the coagulants to the raw AS effluent destabilising the
colloidal particles (including some membrane foulants) and so facilitating the aggregation
of the colloids to form larger particles (or flocs) (Kolarik & Booker 1995). This reduced
pore blocking and pore adsorption and allowed early dominance of the cake filtration
mechanism. At the pH of this AS effluent (pH 7.8) and the low coagulant doses (mg
Al3+/mg DOC = 0.18/1.0 for 2.5 mg Al3+ L-1), coagulation was likely to proceed through a
combination of ‘sweep coagulation’ mechanism (the dominant mechanism at pH > 7) and
charge neutralisation mechanism, since there were insufficient polynuclear aluminium
species and amorphous aluminium hydroxide precipitate (Wang et al. 2008). In the ‘sweep
coagulation’ mechanism, EfOM was removed by entrapment and/or adsorption on the
aluminium hydroxide precipitate, which was a hydrolysis product of alum and ACH. In the
charge neutralisation mechanism, cationic aluminium species and anionic EfOM interact
electrostatically to form insoluble charge-neutral particles (Gregor et al. 1997).

97
 34000
Raw
30000 Raw + 1.0 mg Al (ACH)/L
26000 Raw + 2.5 mg Al (ACH)/L
Raw + 2.5 mg Al (alum)/L
t/V (s/L) 22000

18000 (a)

14000

10000

6000
VVLP
2000
0 0.1 V (L) 0.2 0.3

0.8
(b) Raw
Raw + 1.0 mg Al (ACH)/L
Raw + 2.5 mg Al (ACH)/L
0.6
Raw + 2.5 mg Al (alum)/L
J0 /J

0.4

VVLP
0.2

0
0 50
v (L m ) 100 150 200
-2

Figure 5.1. Plots of (a) t/V versus V and (b) corresponding J/J0 versus v for MF of the raw
and coagulated samples with the VVLP membrane

98
 4500
(a)
Raw
4000
Raw + 1.0 mg Al (ACH)/L
3500 Raw + 2.5 mg Al (ACH)/L
3000 Raw + 2.5 mg Al (alum)/L
t/V (s/L)
2500
XUNP-003
2000

1500

1000

500

0
0 0.45 0.9 V (L) 1.35 1.8

0.8
(b) Raw AS effluent
Raw + 1.0 mg Al (ACH)/L
Raw + 2.5 mg Al (ACH)/L
0.6
Raw + 2.5 mg Al (alum)/L
J0 /J

0.4

XUNP-003
0.2

0
0 50 -2 100 150 200
v (L m )

Figure 5.2. Plots of (a) t/V versus V and (b) corresponding J/J0 versus v for MF of the raw
and coagulated samples with the XUNP-003 membrane

To better understand the mechanisms of fouling reduction by coagulation, the EfOM of the
raw and the coagulated samples were analysed with LC-OCD. The results obtained (Figure
5.3) showed that coagulation removed some biopolymers and HS in the raw effluent, which
reduced the amount of foulants in contact with the membranes and therefore improved the
flux. As can be seen in Figures 5.1 and 5.2, at the same dose (2.5 mg Al3+ L-1), the flux
improvement was slightly higher for alum than ACH. Since the removal of biopolymers
and HS by the two coagulants were comparable (Figure 5.3), the higher flux improvement

99
by alum was probably due to the higher porosity of the cake layer formed on the membrane
surface by the alum-coagulated sample (Wang et al. 2008). Increasing the coagulant dose
from 1.0 mg Al3+L-1 to 2.5 mg Al3+L-1 resulted in better flux improvement (Figures 5.1 and
5.2), which was attributed to the increase in the amount of biopolymers and HS removed
by coagulation (Figure 5.3).

HS Building blocks
12
Low MW HS and low MW acids
10
OCD response (relative units)

Biopolymers
Raw
8
Raw + 1.0 mg Al (ACH)/L
Raw + 2.5 mg Al (ACH)/L
6
Raw + 2.5 mg Al (alum)/L

4
Low MW neutrals

0
20 30 40 50 60 70 80 90 100
Retention time (min)

Figure 5.3. AMWDs of the raw and coagulated effluent

(Raw: DOC = 14.2 mg L-1, UVA245 = 0.343 cm-1, true colour = 90 Pt-Co units)

To determine the effect of coagulation on hydraulically irreversible fouling, the fouled

membranes were backwashed with the permeate at 70 kPa for 5 minutes and their pure
water fluxes were measured. The calculated pure water flux recoveries (WFR) by
backwashing and the hydraulic resistances for the different cases are shown in Table 5.2.
The data showed that coagulation pre-treatment led to significant reductions in the
resistance from hydraulically irreversible fouling (Rif). This indicated a reduction in pore
adsorption and pore blocking of the membranes and could be attributed to the lower
concentrations of biopolymers and HS in the coagulated samples and the early formation of
the cake layer which reduced the contact between these compounds and the membrane
pores.

100
Table 5.2. Effect of coagulation on the hydraulic resistances from total fouling and
irreversible fouling in MF of the effluent, and calculated pure water flux recovery by
backwashing

Rtf * Rif *
Case WFR (%)
( 10-10 m-1) ( 10-10 m-1)
Raw + VVLP 402 11.6 62
3+ -1
Raw + 1.0 mg Al (ACH) L + VVLP 265 2.1 89
Raw + 2.5 mg Al3+(ACH) L-1 + VVLP 209 1.0 95
Raw + 2.5 mg Al3+(alum) L-1 + VVLP 159 1.0 95
Raw + XUNP-003 520 35.7 31
3+ -1
Raw + 1.0 mg Al (ACH) L + XUNP-003 285 25.5 40
3+ -1
Raw + 2.5 mg Al (ACH) L + XUNP-003 186 22.0 45
Raw + 2.5 mg Al3+(alum) L-1 + XUNP-003 134 21.8 45
* Calculated for v = 200 L m-2

As seen in Table 5.2, the XUNP-003 membrane experienced much more severe irreversible
fouling than the VVLP membrane. Backwashing restored the pure water flux of the
membrane fouled by the raw effluent to approximately 62% of the pure water flux of the
virgin VVLP membrane. The corresponding value for the XUNP-003 membrane was only
31%. Factors which potentially contributed to this difference included the differences in the
hydrophobicity, configuration, surface roughness and pore size distribution of the two
membranes. Previous studies on surface water found that hydrophobic membranes suffered
more severe irreversible fouling than hydrophilic counterparts (Fan 2002; Gray et al. 2008).
The XUNP-003 membrane was a hydrophobic membrane with a contact angle greater than
90º, whereas the VVLP membrane was significantly less hydrophobic (contact angle
approximately 62º). The greater irreversible fouling of the XUNP-003 membrane was also
consistent with its higher rejection of biopolymers and HS (Figure 5.4), UV absorbance and
true colour (Table 4.2, Chapter 4) than the VVLP membrane.

The fact that VVLP was a flat-sheet membrane whereas XUNP-003 was a hollow-fibre
membrane likely contributed significantly to the difference in the extent of hydraulically
irreversible fouling suffered by the two membranes, as membrane configuration affected
the hydrodynamics of the backwash and consequently affected backwashing efficiency.
During backwashing, the fouled PVDF flat-sheet was turned upside down in the filtration
cell and the permeate was forced downwards through the membrane. In this case, the

101
permeate flow was expected to be distributed evenly across the membrane surface.
Different from the flat-sheet membrane, the hollow-fibre module comprised 32 PVDF
fibres (inner diameter = 0.8 mm, outer diameter = 1.0 mm, length = 102 mm) potted in a
PES shell. Backwashing of the hollow-fibres involved forcing the permeate to flow axially
(in the inside-out direction) in 32 individual fibres of 102-mm length. Therefore, even flow
distribution of the permeate in the fibres may not have been achieved.

HS Building blocks
12

Low MW HS and low MW acids

OCD response (relative units)

10
Biopolymers
Raw
8
Raw + VVLP
Raw + XUNP-003
6

4 Low MW neutrals

0
20 30 40 50 60 70 80 90 100
Retention time (min)

Figure 5.4. AMWDs of the feed and the permeates from MF of the raw effluent with the
VVLP and XUNP-003 membranes

The reduction in the hydraulically irreversible fouling resulting from the coagulation pre-
treatment was greater for the VVLP membrane than the XUNP-003 membrane (Table 5.2).
For the VVLP membrane, coagulation at 2.5 mg Al3+ L-1 reduced the resistance from
irreversible fouling by approximately 90%, whereas this value for the more hydrophobic
membrane (XUNP-003) was only 36%. The large reduction in the irreversible fouling for
the VVLP membrane indicated that coagulation removed the majority of biopolymers and
HS responsible for pore adsorption and pore blocking of this membrane. However, a
significant proportion of the foulants responsible for irreversible fouling of the XUNP-003
membrane still remained in the coagulated effluent. This hypothesis was supported by the
fact that the XUNP-003 membrane retained more biopolymers and HS of the coagulated
EfOM than the VVLP membrane (Figure 5.5).

102
 HS Building blocks
12

Low MW HS and low MW acids

OCD response (relative units) 10

8
Raw + 2.5 mg Al (ACH)/L
Biopolymers Raw + 2.5 mg Al (ACH)/L + VVLP
6
Raw + 2.5 mg Al (ACH)/L + XUNP-003

Low MW neutrals
2

0
20 40 60 80 100
Retention time (min)

Figure 5.5. Removal of biopolymers and HS in the coagulated effluent by the VVLP and
XUNP-003 membranes

The fairly similar resistances by irreversible fouling ( Rif ) of the samples coagulated with

alum and ACH at 2.5 mg Al3+ L-1 indicated that the cake layers formed by the two samples
on the membrane surfaces were equally easy to remove by backwashing. This observation
supported the hypothesis that the difference in the MF flux of these samples was due to the
difference in the cake porosity. Similar to total fouling, the reductions in hydraulically
irreversible fouling were lower for the lower coagulant dose (Table 5.2) and can be
attributed to the lower removal of the membrane foulants (Figure 5.3).

5.1.2. Single-cycle UF

The flux data obtained from the single-cycle UF tests with the raw and the coagulated
samples are shown in Figures 5.6 and 5.7. To utilise the UMFI concept, plots of the
reciprocal of the normalised flux (J0 /J) versus the accumulated specific permeate volume
(v) were constructed and the slopes of the curves were taken as the UMFI for total fouling
(Huang et al. 2009b), which are shown in Table 5.3. (The UMFI for total fouling was not
computed for the MF flux data since the plot of J0 /J versus v for the raw effluent deviated
significantly from linearity (due to the dominance of the pore adsorption and pore blocking
mechanisms in the first 20 minutes))

103
 30000
Raw
26000 Raw + 1.0 mg Al (ACH)/L
Raw + 2.5 mg Al (ACH)/L
22000 Raw + 2.5 mg Al (alum)/L
t/V (s/L)
18000
HFM-180
14000

10000

6000 (a)

2000
0 0.1 V (L) 0.2 0.3

15
Raw
Raw + 1.0 mg Al (ACH)/L
Raw + 2.5 mg Al (ACH)/L
10 Raw + 2.5 mg Al (alum)/L
J 0/J

HFM-180

(b)

0
-2
0 50 v (L m ) 100 150

Figure 5.6. Plots of (a) t/V versus V and (b) corresponding J0 /J versus v for UF of the raw
and coagulated samples with HFM-180 membrane

104
 30000
Raw
Raw + 2.5 mg Al (ACH)/L
25000

t/V (s/L)
HFK-141
20000

15000
(a)
10000
0 0.05 V (L) 0.1 0.15 0.2

4
Raw
Raw + 2.5 mg Al (ACH)/L

3
J0 /J

HFK-141

(b)
1
-2
0 50 v (L m ) 100 150

Figure 5.7. Plots of (a) t/V versus V and (b) corresponding J0 /J versus v for UF of the raw
and coagulated samples with the HFK-141 membrane

Coagulation improved the permeate flux for UF but did not change the prevailing fouling
mechanism, i.e., fouling was still governed by cake filtration but with a lower UMFI for
total fouling (Figures 5.6 and 5.7 and Table 5.3). Different from MF, alum and ACH
performed similarly in terms of UF flux enhancement. This was probably due to the smaller
pore size of the UF membranes, which made the effect of the difference in cake layer
porosities on the membrane flux insignificant.

The flux improvement by coagulation indicated that if MF and UF are conducted at a

constant flux and rising TMP, as in pilot and full-scale plants, this pre-treatment would
reduce not only the starting operating TMP but also the TMP rise required to sustain the

105
same flux as fouling occurs. This would in turn lower the energy requirement for the
membrane filtration process.

Table 5.3. Values of UMFI for total fouling and resistances from total fouling and
hydraulically irreversible fouling for UF of the raw and coagulated effluent

UMFI Rtf * Rif *

(m2 L-1) ( 10-10 m-1) ( 10-10 m-1)

Raw + HFM-180 0.075 470 55.2
3+ -1
Raw + 1.0 mg Al (ACH) L + HFM-180 0.055 363 38.4
3+ -1
Raw + 2.5 mg Al (ACH) L + HFM-180 0.044 280 31.1
Raw + 2.5 mg Al3+(alum) L-1 + HFM-180 0.042 269 32.7
Raw + HFK-141 0.017 438 67.5
3+ -1
Raw + 2.5 mg Al (ACH) L + HFK-141 0.010 250 49.2
* Calculated for v = 150 L m-2

Similar to MF, coagulation also reduced the hydraulically irreversible fouling of the UF
membranes (Table 5.3). This indicated an opportunity for reducing the frequency of
membrane chemical cleaning and would help to extend the membrane lifetime.

5.1.3. Multi-cycle filtration

Multi-cycle filtration experiments with the raw and the ACH-coagulated effluent were
conducted to determine if the results obtained from single-cycle tests were applicable to a
filtration protocol analogous to that used in MF and UF plants. Six filtration cycles, each of
which lasted for 20 minutes and was followed by 1-minute backwashing with the permeate,
were carried out. The flux data from these tests are presented in Figures 5.8 and 5.9. As
seen in these figures, the highest irreversible fouling occurred in the first filtration cycle,
indicating that pore adsorption and pore blocking were most severe for the virgin
membranes. The increase in the hydraulically irreversible fouling with successive cycles (as
indicated by the increase in the value of J0 /J) resulted from the accumulation of residual
fouling which was not removed by backwashing.

Although the experimental arrangements had some limitations in simulating the membrane
modules used in practice, most notably in the backwashing procedure, the results obtained

106
indicated that fouling reduction by coagulation observed in single-cycle tests was also
achieved in multi-cycle runs. It is important to clarify this point because in practice
filtration with backwashed membranes occupies the majority of the filtration time. Virgin
membranes are only available at the time of purchase and chemical cleaning is done once
every day (chemically enhanced backwash) or every few months (cleaning-in-place).

14
XUNP-003
12

10

8
J 0/J

2 Raw AS effluent
Raw + 2.5 mg Al (ACH)/L
0
0 100 200 300 400 500 600 700
-2
v (L m )

6
XUNP-003 y = 0.0114x + 1
5
y = 0.0056x + 1
4
J 0/J

1 Raw AS effluent
Raw + 2.5 mg Al (ACH)/L
0
0 100 200 300 -2
400 500 600
v (L m )

Figure 5.8. Flux data for multi-cycle MF (top) and calculation of UMFI for hydraulically
irreversible fouling (bottom) for the XUNP-003 membrane
(UMFI for hydraulically irreversible fouling was calculated using the first data points of the
first cycle (v = 0, J0/J = 1) and the last cycle of filtration)
(Samples collected on 27 March 2009, Raw: DOC = 12.0 mg L-1, UVA254 = 0.400 cm-1,
true colour = 112 Pt-Co units, turbidity = 3.9 NTU)

107
 10 HFM-180

6
J0 /J
4

2
Raw AS effluent
Raw + 2.5 mg Al (ACH)/L
0
0 100 200 -2 300 400 500
v (L m )

Figure 5.9. Flux data for multi-cycle UF with the raw and the coagulated effluent for the
HFM-180 membrane
(Samples collected on 27 March 2009, Raw: DOC = 12.0 mg L-1, UVA254 = 0.400 cm-1,
true colour = 112 Pt-Co units, turbidity = 3.9 NTU)

To quantitatively evaluate the effect of coagulation on the hydraulically irreversible

fouling, the UMFI for hydraulically irreversible fouling was calculated using the two-point
method (Huang et al. 2009b) and the results are included in Table 5.4. Feed coagulation
with ACH at 2.5 mg Al3+L-1 reduced the UMFI for hydraulically irreversible fouling by
approximately 50% for the XUNP-003 membrane and 45% for the HFM-180 membrane.

Table 5.4. Effect of coagulation on the UMFI for hydraulically irreversible fouling
UMFI for hydraulically irreversible fouling (m2 L-1)
Membrane
Raw Raw + 2.5 mg Al3+ (ACH) L-1
XUNP-003 11  10-3 5.6  10-3
HFM-180 7.1  10-3 3.9  10-3

5.2. Removal of DOC and true colour by coagulation-MF/UF hybrid

Coagulation with alum or ACH at 2.5 mg Al3+ L-1 removed approximately 20% of the true
colour and 10% of the DOC from the raw AS effluent. Alum was slightly better than ACH
in removing colour-imparting compounds. The alum-treated effluent usually had a true

108
colour 1–3 Pt-Co units lower than the ACH-treated effluent. Compared with the permeates
from MF or UF treatment, the permeates from the coagulation-MF/UF processes typically
had a lower DOC concentration (0.2–1.2 mg DOC L-1) and lower colour (7–20 Pt-Co units
lower). However, they still had an unpleasant yellow colour (60–85 Pt-Co units for a raw
effluent of 90–110 Pt-Co units). Coagulation pre-treatment of the raw effluent at 2.5 mg
Al3+ L-1 therefore was not able to address the high colour issue. It was found that a very
high coagulant dose of 40 mg Al3+ L-1 was needed to achieve the 75–80% colour removal
target desired by the WWTP. The use of such a high coagulant dose is impractical due to
the low economic value of the product water and additional problems such as high
aluminium concentrations in the permeate and high sludge volume.

Although changes in the pH of the raw effluent by coagulation were small due to the low
coagulant doses, it was seen that the pH reduction by coagulation with alum was larger than
that by ACH. At a dose of 2.5 mg Al3+ L-1, the pH of the alum-coagulated sample was 7.2
whereas the pH of the ACH-coagulated sample was 7.6, indicating lower alkalinity
consumption by ACH.

Chapter 5 Summary
The effect of coagulation with ACH and alum on fouling mechanisms and flux performance
for MF and UF of the AS effluent was investigated. Coagulation was found to modify the
MF fouling mechanism to fit the cake filtration model from the beginning of filtration. It
did not change the governing fouling mechanism of UF, but lowered the UMFI for total
fouling. The flux improvement by coagulation was attributed to the ability of the coagulants
to remove some biopolymers and HS from the raw effluent. The MF flux improvement was
higher for alum than ACH, whereas the two coagulants had a similar flux enhancing effect
in UF. Coagulation also reduced irreversible fouling (i.e., pore adsorption and pore
blocking) of the membranes and this effect was greater for the VVLP membrane than the
more hydrophobic XUNP-003 membrane. Results from multi-cycle filtration tests showed
that coagulation reduced fouling of the virgin as well as the hydraulically backwashed
membranes. Coagulation pre-treatment, however, did not reduce the true colour of the MF
and UF permeates to the desired level (20–25 Pt-Co units) unless very high coagulant doses
(approximately 40 mg Al3+ L-1) were used.

109
 CHAPTER 6. EFFECT OF OZONATION & BIOLOGICAL ACTIVATED
CARBON FILTRATION ON MEMBRANE FOULING IN DEAD-END
FILTRATION

Although colour is not a criterion for “Class A” recycled water in Victoria, Australia,
reducing the colour intensity of the treated AS effluent will improve customers’ willingness
to purchase and reuse the product water. As mentioned in Chapters 4 and 5, neither MF and
UF nor coagulation-MF/UF was able to reduce the colour of the raw AS effluent to the
desired level (i.e., approximately 20–25 Pt-Co units). Ozonation was therefore trialled at
the WWTP for this target. Preliminary tests showed that an ozone dose of 10 mg L-1 with a
10-minute contact time was sufficient to achieve the colour removal target. As ozone
oxidised a proportion of the dissolved EfOM in the raw AS effluent to low MW compounds
(<500 Da) which may be a source of assimilable organic carbon (AOC), a BAC filter was
installed downstream of the ozonation process to remove the AOC as well as suspended
solids in the ozonated effluent. The WWTP also considered the use of a low-pressure
membrane filtration process to remove the particulate matter and pathogens from the BAC
filtrate. Therefore, this part of the study investigated the effect of ozonation (at a dose of 10
mg O3 L-1 with a 10-minute contact time) and ozonation followed by BAC filtration on the
characteristics and MF and UF filterability of the raw effluent.

110
6.1. Effect of ozonation and BAC filtration on characteristics of the raw effluent

6.1.1. Effect on general characteristics

The effects of ozonation and BAC filtration on the DOC concentration, UVA254, true colour
and turbidity were investigated on 6 different sample batches and similar trends were
observed. Table 6.1 shows the characteristics of the raw and ozonated effluent and the BAC
filtrate for the samples collected on 10 August 2009. The characteristics of the other sample
batches used in the experiments in this chapter are included in Appendix D.

Table 6.1. Effects of ozonation and BAC filtration on the characteristics of the raw effluent
(Samples collected on 10 August 2009)
Ozonated
Parameter Raw AS effluent BAC filtrate
AS effluent
pH 7.6 7.5 6.8
DOC (mg L-1) 14.0 15.4 13.7
UVA254 (cm-1) 0.379 0.247 0.210
SUVA (L m-1 mg-1) 2.71 1.60 1.53
True colour (Pt-Co units) 109 28 25
Turbidity (NTU) 3.8 3.2 0.79
TSS (mg L-1) 3.7 3.6 0.2

Ozonation at a dose of 10 mg O3 L-1 for 10 minutes resulted in 75–80% reduction in the

true colour of the raw AS effluent, which greatly improved its visual appearance (Table
6.1). The colour removal was attributed to the oxidation of colour-imparting compounds
(i.e., compounds with conjugated C–C double bonds or aromatic rings, particularly HS) in
the raw AS effluent to form colourless materials, as shown by a 34–47% reduction in the
UVA254 of the raw AS effluent. The lower SUVA of the ozonated effluent indicated a
significant reduction in the hydrophobicity of the dissolved EfOM after ozonation.
Ozonation increased the DOC level in the raw AS effluent by 0.5–1.7 mg C L-1. This was
attributed to DOC released from the breakdown of colloidal organic matter to lower MW
compounds (Wang & Pai 2001; Lehman & Liu 2009) and/or breakdown of micro-
organisms (e.g., bacteria, algae, protozoa). BAC filtration removed a proportion of the
dissolved EfOM in the ozonated effluent yielding a filtrate with DOC level slightly lower
than the DOC level of the raw effluent. Since the adsorption capacity of the activated
carbon was exhausted before this study was conducted (see section 3.1.1), the removal of

111
DOC by the BAC filter was due to biodegradation of the EfOM by the micro-organisms
growing on the filter. The small decrease in the true colour (1–3 Pt-Co units) of the
ozonated effluent after BAC filtration indicated that a small proportion of HS in the
ozonated effluent was consumed by the microbes on the BAC. This was further confirmed
in section 6.1.3. As identified by the WWTP staff, the reduction in pH after BAC filtration
(0.3–0.7 pH unit) was due to nitrification taking place in the BAC column. The BAC filter
also retained the particulate matter producing a filtrate with low turbidity and TSS. Overall,
the combination of ozonation and BAC filtration was very effective for removing true
colour and turbidity from the raw AS effluent.

6.1.2. Effect on EfOM fractions

Previous research has investigated the effect of ozonation on the characteristics of different
NOM fractions. In these studies, ozone was applied to the fractions isolated from the NOM
(Marhaba et al. 2000; Swietlik & Sikorska 2004) and EfOM (Zhu et al. 2010). However,
the effect of ozonation (and ozonation followed by BAC filtration) of the bulk EfOM on the
relative concentrations of the fractions has not been studied. It is therefore interesting to
determine this effect for the AS effluent under consideration. The result of this analysis
provided an explanation for the effect of ozonation on the MF and UF flux, which is
discussed in Section 6.2. Furthermore, it can be regarded as a contribution to the general
knowledge of wastewater treatment.

As seen in Figure 6.1 (the results obtained from another sample set, which showed similar
trends, are included in Appendix E), ozonation resulted in a reduction in the VHA fraction
and increases in the other EfOM fractions. The majority of the additional DOC resulting
from ozonation was SHA and CHA in nature. Some of these additions could be products of
the oxidation of VHA (mainly humic and fulvic) material. A proportion of the VHA
fraction in the raw AS effluent was oxidised to less hydrophobic compounds, which led to a
reduction in the concentration of the VHA fraction and an increase in the SHA fraction
after ozonation.

Although resin fractionation procedures involved large changes in the pH of the samples
which may affect the structures and so properties of NOM and EfOM, the fractions
extracted from the resins (after neutralisation) have been used in several studies to obtain
insights to the effects of different water treatment processes on NOM and EfOM (Marhaba

112
et al. 2000; Wong et al. 2002; Gray et al. 2004; Swietlik et al. 2004; Shon et al. 2006; Zhu
et al. 2010).

In this study, the VHA and SHA fractions of the raw and ozonated effluent were recovered
from the resins by extraction with 0.1N NaOH and then adjusted to the original pH of the
samples (pH 7.6) with 4.0 N HCl (as described in section 3.2.2) to investigate the effect of
ozonation on their hydrophobicity. The recovery for the VHA and SHA fractions was 92–
96%. The DOC concentration, UVA254 and SUVA of the extracted fractions are shown in
Table 6.2. For comparison purposes, it was assumed that the changes (if there were any) in
the UV absorbance of these fractions caused by the changes in the pH during the
fractionation and extraction process were similar for the raw and the ozonated effluent. The
reductions in the SUVA of the VHA and SHA fractions after ozonation indicated that the
VHA and SHA fractions of the ozonated effluent were significantly less hydrophobic than
those of the raw effluent. The reductions in the SUVA of these fractions were also
consistent with the colour removal by ozonation since VHA and SHA accounted for
approximately 90% of the true colour of the raw AS effluent.

Table 6.2. Effect of ozonation and BAC filtration on the characteristics of the VHA and
SHA fractions in the raw effluent
(Samples collected on 10 August 2009)
Fraction VHA SHA
True True
SUVA SUVA
DOC UVA254 colour DOC UVA254 colour
(L m-1 (L m-1
(mg L-1) (cm-1) (Pt-Co (mg L-1) (cm-1) (Pt-Co
mg-1 ) mg-1 )
units) units)
Raw
6.4 0.229 3.57 82 2.7 0.065 2.38 12
effluent
Ozonated
5.0 0.145 2.9 N/M* 3.9 0.059 1.52 N/M*
effluent
*
Not measured

It is known that ozone decomposes humic acid to give low MW substances, such as
carboxylic acids (formic, acetic, oxalic, and glyoxylic acids, etc.) and aldehydes, and so
increases the biodegradability of the treated water (Takahashi et al. 1995; Camel &
Bermond 1998; Von Gunten 2003). In the ozonated effluent, the low MW acids may have
been isolated in the CHA fraction due to their charge. The increase in the NEU fraction was

113
attributed to the breakdown of colloidal organic matter and VHA and SHA fractions. The
soluble microbial products released from broken microbial cells may also contribute to this
increase.

18
VHA 15.4
16 SHA 14.0
CHA 13.7
14 NEU
Total
DOC (mg L )
-1

12

10

8 6.9

6 5.4
4.7
4.1 4.2
3.6
4 2.9
2.2 2.0 2.3 2.6 2.2
2

0
Raw effluent Ozonated effluent BAC filtrate

Figure 6.1. EfOM fractions in raw and ozonated AS effluents and in BAC filtrate
(Samples collected on 10 August 2009)

The removal of DOC from the ozonated effluent by the BAC filter was attributed to its
consumption by the microbial flora on the activated carbon. Fractionation showed
reductions in the VHA and CHA components of the ozonated effluent after BAC filtration.
Since the concentration of the SHA and the NEU fractions remained unchanged after BAC
filtration, it was deduced that the consumption by micro-organisms and/or adsorption on
BAC of these compounds was probably counterbalanced by SHA and NEU materials
released by micro-organisms.

6.1.3. Effect on AMWD of dissolved EfOM

LC-OCD analysis allowed better understanding of the effects of ozonation and BAC
filtration on the raw AS effluent. Figure 6.2 (a) shows the chromatograms obtained with
UV detection (at 254 nm). The reductions in the UV absorbance of biopolymers, HS, and
building blocks after ozonation indicated the breakdown in the molecular structures of these

114
compounds. HS showed the largest reduction in the UV absorbance, which can be
explained by the large number of C–C double bonds and electron-rich aromatic rings in
these molecules, which makes them very reactive to ozone (Von Gunten 2003). The
reduction in the UV absorbance of biopolymers was small and likely to originate from the
breakdown in the structures of some proteins, particularly those containing aromatic amino
acids such as tryptophan and tyrosine (Cataldo 2003). BAC filtration of the ozonated
effluent led to small reductions in the UV absorbance of the biopolymers and HS in the
ozonated effluent, which can be attributed to adsorption and/or biodegradation taking place
on the BAC filter.

The chromatograms obtained with organic carbon detection (OCD) are shown in Figure 6.2
(b). The DOC levels of different compound groups were calculated by integrating the areas
under the peaks using DOC-Labor ChromCalc software (Huber et al. 2011) and the results
are shown in Table 6.3. The chromatograms of the samples listed in Table 6.3 but not
shown in Figure 6.2 are included in Appendix F.

115
 2 HS Building blocks
(a)
Low MW HS
UVD response (relative units) 1.5 0.03

Raw
0.02

0.01
Ozonated

1 0
BAC filtrate
26 28 30 32

0.5 Low MW neutrals

Biopolymers

0
20 40 60 80 100
Retention time (min)

14 HS Building blocks
(b)
12 Low MW acids and low MW HS
OCD response (relative units)

10
Raw
8 Ozonated
Biopolymers BAC filtrate
Raw + HFM-180
6 Ozonated + HFM-180

4 Low MW neutrals

0
20 40 60 80 100
Retention time (min)

Figure 6.2. Effects of ozonation, BAC filtration, and UF on AMWD of the raw AS effluent
(a) UV response and (b) DOC response
(The dotted lines define the areas under the peaks)
(Samples collected on 10 August 2009)

116
Table 6.3. Effects of ozonation, BAC filtration, MF with the VVLP membrane and UF
with the HFM-180 membrane on the concentrations of EfOM components (in mg C L-1) as
determined by LC-OCD
Building Low MW Low MW
Sample Biopolymersa HSb
blocksc acidsd neutralse
Raw 2.4 6.7 2.1 0.0 2.6
Ozonated 2.5 5.8 3.4 0.1 2.9
BAC filtrate 2.3 5.5 2.9 0.0 2.3
Raw + VVLP 1.9 6.4 2.0 0.0 2.5
Ozonated + VVLP 2.2 5.6 3.2 0.1 3.0
Raw + HFM-180 1.6 5.5 2.0 0.0 2.5
Ozonated + HFM-180 2.1 5.4 3.3 0.1 2.9
BAC filtrate + HFM-180 2.0 5.3 2.8 0.0 2.2
a
Proteins, protein-like material, polysaccharides, polysaccharide-like material, amino sugars, MW >> 20,000
Da
b
MW = 1,000–20,000 Da
c
Breakdown products of HS, MW = 300–500 Da
d
Monoprotic organic acids, MW < 350 Da
e
Include mono-oligosaccharides, alcohols, aldehydes, ketones, MW < 350 Da

Integrating the data in Figure 6.2 and Table 6.3, ozonation resulted in
(i) a small increase in the concentration of biopolymers, which probably came from
the breakdown of colloidal organic matter and/or release of intracellular matter
from micro-organisms,
(ii) partial breakdown of biopolymers, as can be seen in Figure 6.2 (a),
(iii) a small reduction in the amount of HS due to the breakdown of HS to lower
MW compounds, and
(iv) an increase in biodegradable materials, i.e., building blocks, low MW acids and
low MW neutrals, which were breakdown products of HS and possibly
ozonolysis of micro-organisms.

During BAC filtration, some of the biopolymers, HS, building blocks, low MW acids, and
low MW neutrals in the ozonated AS effluent were consumed by the micro-organisms on
the BAC. Although HS are generally regarded as non-biodegradable, the breakdown of C–
C double bonds by ozonation may have improved their biodegradability (Buls 2000),
making them amenable to consumption by the microbes on the BAC.

117
6.2. Effect of ozonation and BAC filtration on MF flux

Figures 6.3 and 6.4 show that ozonation improved the flux through the MF membranes. As
fouling of the MF membranes in the initial period of filtration was governed by the pore
adsorption mechanism (Chapter 4), the higher permeate flux of the ozonated effluent than
the raw AS effluent from the beginning of filtration indicated a reduction in the adsorption
of the foulants on the membranes. This was supported by the fact that the membranes
retained a lower amount of biopolymers and HS in the ozonated effluent than in the raw
effluent (Table 6.3, results for the XUNP-003 membrane with another batch of samples is
presented in section 7.2, Chapter 7). The lower adsorption capacity of the ozonated
biopolymers and HS on the membranes can be attributed to the reduction in their
hydrophobicity (as shown by the reductions in their UV responses in Figure 6.2 (a) and the
conversion of some VHA material to SHA (section 6.1.2)), which reduced hydrophobic
interactions with the membranes. It was also possible that ozonation caused changes in the
secondary and tertiary structures of some proteins (Cataldo 2003), allowing them to pass
through the membranes thus reducing the rate of pore blocking and cake formation. Ozone
may also react with -D-glycosidic linkages on polysaccharides and depolymerise the
molecules (Wang et al. 1999). The fact that the concentration of biopolymers did not
decrease after ozonation indicated that the breakdown products of these compounds were
larger than 20 kDa and therefore remained in the biopolymers peak on the LC-OCD
chromatogram of the ozonated effluent.

118
 1
VVLP Raw
Ozonated
BAC filtrate
0.8 0.5 micron pre-filtered Raw
0.5 micron pre-filtered Ozonated
0.5 micron pre-filtered BAC filtrate
0.6
J/J0

0.4

0.2

0
0 50 100 150 -2 200 250 300 350
v (L m )
Figure 6.3. Effect of ozonation and BAC filtration on flux of the VVLP membrane
(Samples collected on 10 August 2009)

1 XUNP-003 Raw
Ozonated
BAC filtrate
0.8 0.5 micron pre-filtered Raw
0.5 micron pre-filtered Ozonated
0.5 micron pre-filtered BAC filtrate
0.6
J/J0

0.4

0.2

0
0 50 -2
v (L m ) 100 150 200

Figure 6.4. Effect of ozonation and BAC filtration on flux of the XUNP-003 membrane
(Samples collected on 10 August 2009)

BAC filtration of the ozonated effluent further reduced the rate of fouling. Since the
ozonated AS effluent and the BAC filtrate had fairly similar fouling potentials after the
removal of particles larger than 0.5 µm (Figures 6.3 and 6.4, open symbols), the reduction

119
in fouling by BAC filtration was due mainly to the decrease in the TSS level (Table 6.1).
This was consistent with the small reductions in the amounts of membrane foulants after
BAC filtration (Table 6.3).

Previous studies show that ozonation could change the size distribution (Zhu et al. 2008)
and surface charge (or zeta potential) (Chandrakanth & Amy 1996; Zhu et al. 2008) of
organic particles in water and wastewater. These changes may affect the cake permeability
in membrane filtration and therefore affect the flux (Fu & Dempsey 1998). Figure 6.5
shows the particle size distributions of the raw and the ozonated effluent samples (results of
the three replicates are included in Appendix G). Ozonation resulted in the breakdown of
particles larger than 300 µm in the raw AS effluent. However, the flux data for the samples
before and after the removal of particles larger than 0.5 µm (Figures 6.3 and 6.4) indicated
that for this AS effluent this change had little influence on the membrane flux compared
with the breakdown of dissolved EfOM. The zeta potentials of the raw effluent and the
ozonated effluent were -13.21.4 mV and -14.31.5 mV, respectively, indicating that
ozonation did not cause significant changes in the electric charges of the colloidal particles
in the raw effluent.

8
Raw
7
Ozonated
6
Volume %

0
1 10 100 1000 10000
Particle size (micron)
Figure 6.5. Particle size distributions of the raw and ozonated AS effluent
(Samples collected on 10 August 2009)

120
Ozonation of water/wastewater is known to produce compounds containing carboxylic and
phenolic groups which may have a higher tendency to form aggregates with divalent
cations (e.g., Ca2+ and Mg2+) than the parent compounds and therefore may worsen
membrane fouling (Zhu et al. 2010). However, although ozonation of this AS effluent
produced some low MW organic acids (Table 6.3), the results from the MF and UF tests
showed that any such negative effect on the membrane flux was either absent or
compensated by the breakdown in the structures of the membrane foulants (i.e.,
polysaccharides, proteins and HS).

6.3. Effect of ozonation and BAC filtration on UF flux

6.3.1. UF with the HFM-180 membrane

Similar to MF, ozonation and BAC filtration also reduced fouling on the HFM-180
membrane. The UMFIs for total fouling of the HFM-180 membrane were computed for the
raw AS effluent, ozonated AS effluent, and BAC filtrate using forced linear regression (y-
intercept was set at 1.0 to follow equation 2.16 and facilitate flux comparison). The results
were the slopes of the fitted lines (Figure 6.6) in m2 L-1.

15
Raw
Ozonated
BAC filtrate
0.5 micron pre-filtered Raw
0.5 micron pre-filtered Ozonated
10 0.5 micron pre-filtered BAC filtrate
J0 /J

2
Raw: y = 0.082x + 1, R = 0.987
2
Ozonated: y = 0.056x + 1, R = 0.994
2
BAC filtrate: y = 0.040x + 1, R = 0.993
0
-2
0 50 v (L m ) 100 150

Figure 6.6. Flux data for UF of the raw and ozonated effluent and the BAC filtrate
with the HFM-180 membrane
(Samples collected on 10 August 2009)

121
Ozonation reduced the UMFI of the raw AS effluent by approximately 30%, which can be
explained by the breakdown of biopolymers and HS to lower MW compounds (Figure 6.2
and Table 6.3). The lower amounts of biopolymers and HS retained by the membrane
indicate that the fouling by these compounds was reduced after ozonation.

BAC filtration resulted in a 30% reduction in the UMFI of the ozonated AS effluent.
Similar to MF, the flux improvement by BAC filtration was attributed to the removal of
particles larger than 0.5 µm by the filter.

For the samples presented in Table 6.1 and Figure 6.6, the overall reduction in the fouling
rate by the ozonation/BAC filtration pre-treatment was approximately 50% as measured by
the UMFI. The flux improvement implies that if UF is conducted at a constant flux and
rising TMP, as in pilot and full-scale plants, the combined pre-treatment would
significantly reduce the increasing rate of TMP and therefore lower the energy requirement
for the membrane filtration process.

For the 6 sample batches collected from March 2009 to August 2009, ozonation reduced
the UMFI for total fouling of the HFM-180 membrane by the raw AS effluent by 22–40%.
BAC filtration further reduced the UMFI for the ozonated effluent by 17–29%. Overall, the
combination of ozonation and BAC filtration reduced the UMFI for total fouling by the raw
AS effluent by 35–50%. The characteristics and UMFI for total fouling for these samples
are included in Appendix D.

6.3.2. UF with the HFK-141 membrane

Ozonation also reduced fouling of the tighter UF (HFK-141, 30 kDa) membrane (Figure
6.7). However, different from the loose UF (HFM-180) membrane, BAC filtration did not
further improve the permeate flux of the HFK-141 membrane. This was consistent with the
fact that the removal of particles larger than 0.5 µm did not have any significant effect on
the flux decline for this membrane (Chapter 4, Section 4.1).

122
 4 2
Raw: y = 0.017x + 1, R = 0.989
2
Ozonated: y = 0.012x +1, R = 0.998
2
BAC filtrate: y = 0.011x + 1, R = 0.997
3
J0 /J

2
Raw
Ozonated
BAC filtrate
1
-2
0 50 v (L m ) 100 150

Figure 6.7. Flux data for UF of the raw and ozonated effluent and
the BAC filtrate with the HFK-141 membrane
(Samples collected on 16 June 2009,
their characteristics are shown in Appendix D)

6.4. Effect of ozonation and BAC filtration on hydraulically irreversible fouling

6.4.1. Results from single-cycle filtration tests

The hydraulic resistances exerted by total fouling ( Rtf ) and hydraulically irreversible

fouling ( Rif ) were calculated for the flux data in Figures 6.3, 6.4, 6.6 and 6.7, and the

results are shown in Table 6.4.

Ozonation led to significant reductions in the hydraulically irreversible fouling of the

membranes. This was attributed to the smaller amount of biopolymers and HS retained on
the membranes (as suggested by the data in Table 6.1 for the VVLP and HFM-180
membranes). Because the amounts of biopolymers and HS removed by the BAC filter were
small, this treatment did not further reduce the proportion of internal fouling. Similar to
coagulation pre-treatment (Chapter 5), the reduction in hydraulically irreversible fouling by
ozonation was greater for the less hydrophobic MF membrane (VVLP, 75% reduction in
Rif) than the more hydrophobic counterpart (XUNP-003, 35% reduction in Rif). The results
indicate that mitigation of hydraulically irreversible fouling is more difficult to achieve
with more hydrophobic membranes.

123
Table 6.4. Effect of ozonation and BAC filtration on resistances from total fouling and
hydraulically irreversible fouling (for the flux data shown in Figures 6.3, 6.4, 6.6, and 6.7)
Rm Rtf Rif
Case
( 10-10 m-1) -10
( 10 m ) -1
( 10-10 m-1)
Raw + VVLP 18.2 471 11.2
Ozonated + VVLP 17.8 247 2.80
BAC filtrate + VVLP 17.5 94.0 3.11
Raw + XUNP-003 16.9 515 37.6
Ozonated + XUNP-003 17.0 248 24.7
BAC filtrate + XUNP-003 17.3 228 24.2
Raw + HFM-180 38.0 486 52.4
Ozonated + HFM-180 41.0 369 34.9
BAC filtrate + HFM-180 40.5 289 31.8
Raw + HFK-141 174 607 67.5
Ozonated + HFK-141 173 468 43.2
BAC filtrate + HFK-141 170 443 45.2

6.4.2. Results from multi-cycle filtration tests

To investigate the effect of ozonation on membrane flux in a filtration protocol similar to

that used in pilot and full-scale plants, multi-cycle filtration tests (20-min filtration
followed by 1-min backwashing) were conducted with the raw and ozonated effluent using
the XUNP-003, HFM-180 and HFK-141 membranes. Data from these tests (Figures 6.8
and 6.9) show that fouling reduction by ozonation was also achieved in multi-cycle
filtration (Data for the XUNP-003 membrane are shown in Figure 7.2, Chapter 7, where
ozonation and adsorption with the AER and the PAC are compared).

The UMFI for hydraulically irreversible fouling was calculated using the two-point method
(Figure 6.8 (b)) and the results (Table 6.5) show a reduction in this value after ozonation.
Ozonation therefore improved water productivity and reduced the accumulation rate of
residual fouling that cannot be removed by backwashing. The reduction in the hydraulically
irreversible fouling achieved with the ozonation/BAC filtration pre-treatment indicated an
opportunity for reducing the frequency of membrane chemical cleaning and this would help
to extend the membrane lifetime.

124
Table 6.5. Effect of ozonation on UMFI for hydraulically irreversible fouling
UMFI for hydraulically irreversible fouling (m2 L-1)
Membrane
Raw Ozonated
XUNP-003 11  10-3 6.4  10-3
HFM-180 7.6  10-3 4.4  10-3
HFK-141 4.1  10-3 2.0  10-3

125
 12
(a)
10

6
J0 /J

2 Raw
Ozonated
0
0 50 100 150 200 250 300 350 400 450
-2
v (L m )

3.5
(b) y = 0.0076x + 1
3

2.5
J0 /J

2 y = 0.0044x + 1

1.5
Raw
Ozonated
1
0 100 -2 200 300 400
v (L m )

Figure 6.8. Effect of ozonation on permeate flux of the

HFM-180 membrane in multi-cycle filtration
(a) Specific permeate volume versus the reciprocal of normalised flux and
(b) Calculation of the UMFI for hydraulically irreversible fouling using the first data points
of the first cycle (v = 0, J0 /J = 1) and the last cycle of filtration
(Samples collected on 24 April 2009, their characteristics are shown in Appendix D)

126
 2.8

2.6

2.4

2.2

2
J0 /J

1.8

1.6

1.4
Raw
1.2
Ozonated
1
0 50 100 150 -2
200 250 300 350
v (L m )

Figure 6.9. Effect of ozonation on permeate flux of the

HFK-141 membrane in multi-cycle filtration
(Samples collected on 24 April 2009, their characteristics are shown in Appendix D)

6.5. Potential effect of ozonation and BAC filtration on biofouling propensity of

downstream membranes
Ozonation of the raw AS effluent increased the concentrations of low MW organics
(building blocks, low MW acids, and low MW neutrals) (Table 6.3). If MF or UF is used to
directly treat the ozonated effluent, this could facilitate bacterial growth in the membrane
unit and accelerate membrane biofouling, particularly when complete disinfection by
ozonation is not achieved (Miettinen et al. 1998). Furthermore, the membranes retained a
lower amount of biopolymers in the ozonated effluent than in the raw effluent and were not
able to retain the low MW compounds (Table 6.3). Thus, ozonation caused increases in the
concentrations of potentially biodegradable organic materials in the permeate, which would
increase the risk of biofouling of downstream RO membranes if RO is used to further treat
this permeate. The addition of the BAC filter downstream of the ozonation process
potentially reduced the risk of biofouling since the filter removed the majority of the low
MW organic compounds generated during ozonation (Table 6.3). However, the
concentrations of biopolymers and building blocks in the permeate from the BAC filtrate
were still higher than those in the permeate from the raw AS effluent (Table 6.3). Overall,

127
ozonation of the raw AS effluent improved the MF and UF flux but increased the risk of
biofouling of these membranes and further membranes downstream. BAC filtration of the
ozonated effluent reduced but did not completely eliminate this risk.

Chapter 6 Summary

The effect of ozonation and ozonation followed by BAC filtration on the characteristics and
MF and UF filterability of the raw AS effluent was investigated. Ozonation at a dose of 10
mg O3 L-1 led to a 75–80% reduction in the true colour of the raw AS effluent, which was
attributed to the oxidation of HS, the main chromophores. Rapid fractionation showed that
after ozonation some of the VHA material in the raw effluent was converted to SHA, CHA,
and possibly NEU compounds. Some additional NEU material in the ozonated AS effluent
probably originated from the breakdown of high MW colloidal organic matter and micro-
organisms. BAC filtration removed some of the VHA and CHA fractions from the ozonated
effluent, and lowered the turbidity of the effluent to less than 1.0 NTU.

LC-OCD analysis showed that ozonation caused the oxidation of a small proportion of HS
in the raw effluent to building blocks, low MW acids, and low MW neutrals. Some of these
oxidation products and a small amount of the oxidised HS were then utilised by the micro-
organisms on the filter media during BAC filtration.

Similar to the raw AS effluent, biopolymers and HS were the major membrane foulants in
MF and UF of the ozonated effluent and the BAC filtrate. Ozonation improved the MF and
UF permeate flux by breaking down a proportion of the membrane foulants to compounds
with lower MW and hydrophobicity. BAC filtration of the ozonated effluent removed the
majority of the particles larger than 0.5 μm and therefore led to further flux improvement in
MF and in UF with the 100 kDa membrane (HFM-180). However, BAC filtration did not
further improve the flux through the tighter UF membrane (HFK-141, 30 kDa). The UMFI
for total fouling was used to quantitatively evaluate the effectiveness of ozonation and
BMF on UF flux. The hydraulically irreversible fouling of the membranes was reduced
after ozonation while subsequent BAC filtration did not affect this type of fouling. The
reduction in hydraulically irreversible fouling by ozonation was significantly greater for the
less hydrophobic MF (VVLP) membrane than the more hydrophobic counterpart (XUNP-
003).

128
As ozonation of the raw AS effluent generated low MW organic components (i.e., building
blocks, low MW acids and low MW neutrals), this pre-treatment may accelerate biofouling
of the MF and UF membranes and further downstream membranes. The addition of the
BAC filter downstream of the ozonation process showed potential to reduce the risk of
biofouling since the filter removed a significant proportion of the low MW organic
compounds generated during ozonation.

129
 CHAPTER 7. EFFECT OF ADSORPTION PRE-TREATMENTS WITH A
POWDERED ACTIVATED CARBON AND AN ANION EXCHANGE RESIN ON
MEMBRANE FOULING

As mentioned in Chapter 2, the yellow colour of the raw AS effluent is an ongoing issue
which potentially limits customers’ willingness to purchase the treated effluent for reuse
purposes. The fact that the colour-bearing compounds in the raw effluent were isolated in
the VHA and SHA fractions (Chapter 6), which is known to contain mainly HS (Croué
1999), indicated that adsorption pre-treatment with appropriate AERs or PAC may help to
address this problem. PAC adsorption has long been used for the removal of HS (Summers
& Roberts 1988b). HS molecules are aromatic compounds carrying negative charges at
neutral pH and were therefore expected to undergo anion exchange with AERs (Bolto et al.
2002).

A commercial AER and a PAC product capable of achieving the colour removal target (75–
80%) were found. They were selected for their ready availability in Victoria (Australia) and
the fact that they have been used for water and wastewater treatment for many years. The
effect of adsorption pre-treatment with the PAC (Acticarb PS1300, Activated Carbon
Technologies) and the AER (at the doses which gave 75–80% colour reduction) on
membrane fouling was investigated and the results are presented in this chapter. The
hydrophobic PVDF MF (XUNP-003, 0.1 μm) and UF (HFM-180, 100 kDa) membranes
and the slightly hydrophilic PES UF (HFK-141, 30 kDa) membrane were used in this study.
A direct comparison between ozonation and these adsorption pre-treatments is provided
which may be useful for other WWTPs which are facing similar colour issues.

130
7.1. Effect of adsorption pre-treatments on raw effluent characteristics
As seen in Table 7.1, the 80% colour removal target was achieved by treating the raw AS
effluent with ozone at 10 mg L-1, the AER at 10 mL L-1 or the PAC at 150 mg L-1.
However, different from ozonation, which caused a small increase in the DOC
concentration of the effluent, adsorption with the PAC and the AER removed
approximately 57% and 37%, respectively, of the DOC from the raw AS effluent,
respectively. Also, the reduction in the UVA254 after ozonation (42%) was significantly
lower than after adsorption with the PAC (74%) and the AER (63%).

Table 7.1. Effect of ozonation and adsorption pre-treatment with the AER and the PAC on
the characteristics of the raw effluent
(Samples collected on 08 July 2011)
Raw + 10 mL Raw + 150 mg Ozonated (Raw
Parameter Raw
AER L-1 PAC L-1 + 10 mg O3 L-1)
pH 7.8 7.7 7.8 7.7
DOC (mg L-1) 15.1 9.5 6.5 15.6
UVA254 (cm-1) 0.389 0.144 0.10 0.227
SUVA (m-1mg-1 L) 2.58 1.52 1.54 1.46
True colour (Pt-Co units) 100 22 21 22
Turbidity (NTU) 1.9 N/M* N/M* 1.8
*
Not measured

7.2. Effect of adsorption pre-treatments on AMWD of the raw effluent

The LC-OCD chromatograms of the raw and pre-treated AS effluent samples (Figure 7.1)
provided insights into the effects of the pre-treatments on the dissolved EfOM. The DOC
concentrations of the different organic components, calculated by integrating the areas
under the peaks, are shown in Table 7.2.

131
 2.5 HS Raw
(a) Ozonated
Building AER-treated
2 0.06
blocks
UVD response (relative units) 0.04
PAC-treated

0.02

1.5 Low MW acids and low MW HS

0
26 28 30 32

Biopolymers
0.5 Low MW neutrals

0
20 30 40 50 60 70 80 90 100
Retention time (min)

15 HS Building blocks
(b)
Low MW acids and low MW HS
OCD response (relative units)

10
Raw
Ozonated
Biopolymers
AER-treated
PAC-treated
5

Low MW neutrals

0
20 30 40 50 60 70 80 90 100
Retention time (min)

Figure 7.1. AMWDs of the raw, ozonated, AER-treated, and PAC-treated effluents
(a) UVD response, (b) OCD response
(The dotted lines define the areas under the peaks)
(Samples collected on 08 July 2011)

132
Table 7.2. Effect of ozonation and the adsorption pre-treatments with and without MF on
the concentrations of dissolved EfOM components (in mg C L-1) as determined by LC-
OCD
Building Low MW Low MW
Sample Biopolymers HS
blocks acids neutrals
Raw 1.3 8.0 2.4 0.0 2.2
Ozonated 1.4 7.6 2.7 0.2 2.4
AER-treated 1.3 2.5 2.4 0.0 2.2
PAC-treated 1.3 3.6 0.7 0.0 0.8
Raw + XUNP-003 0.2 7.0 2.2 0.0 2.2
Ozonated + XUNP-003 0.7 7.2 2.7 0.2 2.3
AER-treated + XUNP-003 0.3 2.0 2.3 0.0 2.2
PAC-treated + XUNP-003 0.2 2.8 0.7 0.0 0.8

The data in Figure 7.1 and Table 7.2 showed that the EfOM components adsorbed on the
PAC included HS, building blocks, and low MW neutrals. The forces keeping the EfOM
molecules attached to the PAC may include (Newcombe et al. 2008) (i) electrostatic
attraction between PAC surface and EfOM, (ii) hydrophobic attraction between graphitic
carbon surface and hydrophobic portions of EfOM molecules, (iii) hydrogen bonding
between functional groups on EfOM and those on the carbon surface, and (iv) -
interactions between the carbon surface and the aromatic EfOM. The simultaneous
operation of these mechanisms, and more importantly, the presence of both mesopores and
micropores could be the reasons for the PAC removing not only building blocks and low
MW neutrals but also HS. Mesopores (2–50 nm) probably provided adsorption sites for the
HS molecules (aquatic fulvic acids generally have an average diameter of 2 nm whereas
humic acids are larger molecules (Sutzkover-Gutman et al. 2010)). This was supported by
the fact that at the same dosage and contact time, another PAC product (provided by Tiwest
Sales Australia) having a similar surface area (1427 m2 g-1), a lower mesopore volume
(0.33 cm3 g-1) and a higher micropore volume (0.25 cm3 g-1) than the PAC used in this
study (Acticarb PS1300, surface area = 1429 m2 g-1, mesopore volume = 0.52 cm3 g-1,
micropore volume = 0.15 cm3 g-1) gave only 48% colour removal. Building blocks and low
MW neutrals were likely to adsorb in both mesopores and micropores (< 2 nm).
Approximately 75% of the building blocks and 65% of low MW neutral compounds in the
raw AS effluent were removed by the PAC treatment.

133
The AER adsorbed a larger proportion of HS than the PAC. However, it was not as
effective as the PAC in removing building blocks and low MW neutrals (Figure 7.1 and
Table 7.2). The higher selectivity of the AER than the PAC for HS was attributed to the
electrostatic attraction between the resin and the negatively charged HS molecules. During
the adsorption test, it was noticed that the colour removal of the AER reduced with the
passage of increasing volume of the treated sample. The colour of the effluent after the first
100 bed volumes was 8 Pt-Co units whereas after treatment of 1500 bed volumes, the
colour of the combined treated effluent was 22 Pt-Co units due to increasing saturation of
the adsorption sites with increasing volume.

7.3. Effect of adsorption pre-treatments on membrane fouling

7.3.1. Effect on total fouling

Whole samples and samples pre-filtered through 1.0 μm glass-fibre filters (GF/B,
Whatman) were used in MF and UF experiments to determine if the PAC particles and
residual AER particles influenced the membrane flux.

The flux data for the XUNP-003 membrane are shown in Figure 7.2. MF of the whole
samples and the 1.0 μm pre-filtered samples gave similar results (Figure 7.2 (b)), indicating
that the PAC particles and residual AER particles had little effect on the flux. This was
further confirmed by MF of a suspension containing 150 mg PAC per litre of Milli-Q
water, which showed no fouling by the PAC particles. Therefore, it can be concluded that
the changes in the dissolved EfOM contents resulting from the adsorption pre-treatments
did not affect fouling of the membrane.

134
 1
Raw
(a) Ozonated
0.8 AER-treated
PAC-treated

0.6
J/J0

0.4

0.2

0
0 50 100 150 200
-2
v (L m )

1
(b) 1.0 micron pre-filtered Raw
1.0 micron pre-filtered Ozonated
0.8
1.0 micron pre-filtered AER-treated
1.0 micron pre-filtered PAC-treated
0.6
J/J0

0.4

0.2

0
0 50 -2 100 150 200
v (L m )
Figure 7.2. Effect of ozonation and adsorption pre-treatments
on permeate flux of the XUNP-003 membrane
(a) MF of whole samples and (b) MF of pre-filtered samples
(Samples collected on 08 July 2011)

As seen in Figure 7.2, ozonation significantly improved the membrane flux, whereas pre-
treatments with the PAC and the AER neither reduced nor worsened membrane fouling.
The flux improvement by ozonation was attributed to the breakdown of proteins,
polysaccharides and HS to less hydrophobic and lower MW compounds (as discussed in
Chapter 6). The ineffectiveness of the adsorption pre-treatments in fouling mitigation can

135
be explained by the PAC and AER not removing biopolymers and the HS which
contributed to membrane fouling from the raw AS effluent. The reasons for the inability of
the PAC to remove biopolymers may include: (i) the preferential adsorption of the HS and
the low MW compounds on the PAC led to the saturation of the PAC with these molecules
and therefore no adsorption sites were available for the biopolymers and/or (ii) the larger
size of the biopolymers molecules than the pores of the PAC. The ineffectiveness of the
AER for removing biopolymers was very likely due to their lack of negatively charged
functional groups at the sample pH (Humbert et al. 2007; Huber et al. 2011) as well as the
larger size of the biopolymers than the AER pore size (45 nm). Different from this study, a
previous study found that some PAC products (used at 50–1000 mg L-1) removed both
biopolymers and HS from a secondary effluent (Haberkamp et al. 2007).

The biopolymers in the raw, AER-treated, and PAC-treated effluents had similar fouling
potentials, as shown by the similar rejections by the membrane (Figure 7.3 and Table 7.2).
The membrane retained a lower amount of HS in the AER-treated (0.5 mg C L-1) and PAC-
treated (0.8 mg C L-1) effluents than in the raw effluent (1.0 mg C L-1) (Table 7.2),
indicating that the HS in the pre-treated effluents probably had a lower fouling potential
than the HS in the raw effluent. However, this did not result in any improvement in the MF
flux.

136
 15 HS Building blocks

OCD response (relative units) Low MW acids and low MW HS

Raw
10
Raw + XUNP-003
AER-treated
Biopolymers AER-treated + XUNP-003
PAC-treated
PAC-treated + XUNP-003
5

Low MW neutrals

0
20 30 40 50 60 70 80 90 100
Retention time (min)

Figure 7.3. Effect of adsorption pre-treatments on the rejection

of EfOM components by the XUNP-003 membrane
(Samples collected on 08 July 2011)

On the other hand, although ozonation led to a small increase in the biopolymers
concentration (most likely due to the breakdown of microbial cells in the raw effluent), the
oxidation of biopolymers and HS led to significant fouling reduction. As seen in Table 7.2,
after ozonation the rejection of these compounds by the MF membrane reduced
significantly. (For clarity of Figure 7.3, the LC-OCD chromatogram of the permeate from
MF of the ozonated effluent, i.e., the sample “Ozonated + XUNP-003” in Table 7.2, is
included in Appendix H). Since the removal of a large proportion of the HS (70% by the
AER and 55% by the PAC) did not reduce membrane fouling, it can be concluded that
although biopolymers comprised a small proportion of the total DOC in the AS effluent,
they made a much larger contribution to fouling than did HS. Therefore, pre-treatments for
fouling reduction should target the removal of polysaccharides and proteins rather than HS.

Experiments with the HFM-180 and HFK-141 UF membranes gave similar flux results, i.e.,
the adsorption pre-treatments did not affect the permeate flux (Figures 7.4 and 7.5).

137
 0.6
Raw
Ozonated
0.5
AER-treated
PAC-treated
0.4
J/J0

0.3

0.2

0.1

0
0 50 -2 100 150
v (L m )

Figure 7.4. Flux data for UF of the raw, ozonated, AER-treated,

and PAC-treated effluent samples with the HFM-180 membrane
(Samples collected on 01 April 2011, their characteristics are shown in Appendix I)

Raw
Ozonated
0.8 AER-treated
PAC-treated
J/J0

0.6

0.4

0.2
0 50 -2 100 150
v (L m )

Figure 7.5. Flux data for UF of the raw, ozonated, AER-treated, and PAC-treated effluent
samples with the HFK-141 membrane
(Samples collected on 08 July 2011)

138
7.3.2. Effect on hydraulically irreversible fouling

The resistances by total fouling ( Rtf ) and hydraulically irreversible fouling ( Rif ) were

calculated for the samples in Figures 7.2, 7.3 and 7.5 and the results are presented in Table
7.3. As expected, the adsorption pre-treatments did not affect the irreversible fouling of the
membranes.

Table 7.3. Effect of adsorption pre-treatments and ozonation on resistance by total fouling
and resistance by hydraulically irreversible fouling
Case Rtf * ( 10-10 m-1) Rif * ( 10-10 m-1)
Raw + XUNP-003 516 36.0
AER-treated + XUNP-003 510 36.9
PAC-treated + XUNP-003 543 37.4
Ozonated + XUNP-003 342 27.1
Raw + HFM-180 664 54.5
PAC-treated + HFM-180 641 52.8
AER-treated + HFM-180 646 55.1
Ozonated + HFM-180 534 35.9
Raw + HFK-141 576 89.8
AER-treated + HFK-141 558 83.5
PAC-treated + HFK-141 562 86.9
Ozonated + HFK-141 391 66.9
*
Calculated for v = 200 L m-2 (MF) and v = 150 L m-2 (UF)

The flux data obtained from multi-cycle MF of the raw, PAC-treated, and AER-treated
samples with the XUNP-003 membrane are shown in Figure 7.6. The results show that the
pre-treatments did not affect the permeate flux of the virgin membrane or the hydraulically
backwashed membrane.

139
 14

12

10

8
J0 /J

Raw
2 Ozonated
AER-treated
PAC-treated
0
0 50 100 150 200 250 300 350 400 450 500 550
-2
v (L m )

Figure 7.6. Flux data for multi-cycle MF of the raw, PAC-treated and AER-treated
effluents with the XUNP-003 membrane
(Samples collected on 08 July 2011)

In terms of biofouling, since PAC adsorption removed a significant proportion of building

blocks and low MW neutrals from the raw AS effluent (Table 7.2), this pre-treatment may
reduce membrane biofouling. The pre-treatment with the AER probably would not affect
membrane biofouling propensity as it was ineffective in removing these low MW
compounds.

7.4. Practicability considerations of the adsorption pre-treatments in comparison with

ozonation
Of the three pre-treatments investigated for colour removal, ozonation is the only method
that produces no waste stream. Although this method involves the use of the toxic gas
ozone, it is a safe and practical method with the current technologies. The adsorption
methods, while utilising non-toxic materials, have the disadvantage of producing waste
streams, which need further treatments before disposal.

140
For the AER, the spent resin needs to be regenerated, normally with brine. This process
produces a waste stream with very high salinity, DOC concentration, and true colour (Bolto
et al. 2002). NF or RO can be used to separate the brine from the EfOM and the brine can
be reused for resin regeneration. The NF or RO retentate, which would contain the
concentrated EfOM and probably elevated concentrations of toxic and/or bioaccumulative
organic substances (such as pharmaceuticals and personal care products), would need
further treatment before it can be disposed of to the environment. These further treatments
add to the technical complexity, and therefore the cost may be greater.

For PAC adsorption, the PAC particles in the treated effluent may be allowed to settle in
the adsorption tank prior to MF/UF. It is also possible to run MF or UF of the pre-treated
effluent without prior removal of the PAC, most likely in the submerged membrane
configuration (Khirani et al. 2006b). The spent PAC will need to be collected for reuse,
regeneration or disposal. However, because its colour removal capacity will decrease after
successive cycles, fresh PAC must be added to maintain the colour removal target. In
addition, PAC can be a health and safety concern (although problems in handling PAC can
be avoided by using the granular version of the same activated carbon). It should be pointed
out that the PAC dose required for the colour removal target in this study was very high
compared with typical dosages for the removal of micropollutants from water (less than 30
mg PAC L-1 (Newcombe et al. 2008)).

The main advantage of the adsorption pre-treatments was their high DOC removals and
therefore production MF and UF permeates with low DOC concentrations. Prior to being
discharged for reuse, the permeates would be disinfected, normally by chlorination, and a
low EfOM concentration would likely result in low concentrations of disinfection by-
products (DBPs) in the product water (Reckhow et al. 1990; Singer 1999; Liang & Singer
2003). Ozonation pre-treatment may have both beneficial and adverse effects on the
formation of DBPs. Ozonation has been found to reduce the total organic halide formation
potential during the chlorination of municipal wastewater (Wert et al. 2007). However, this
pre-treatment may lead to the formation of bromate (a suspected carcinogen), N-
nitrosodimethylamine (NDMA, a carcinogenic, mutagenic and clastogenic chemical)
(Hollender et al. 2009), and other potentially toxic oxidation by-products (Wert et al.
2007). Despite of this complication, the advantage (if there is any) of the adsorption pre-
treatments over the ozonation pre-treatment in terms of reducing DBPs formation is of

141
minor significance in the current context, as the DBPs formation during chlorination of
municipal wastewater has been found very low and the parameter is currently not a
requirement for “Class A” recycled water, which is only used non-potable purposes (EPA
Victoria 2002).

AERs have been used widely in surface water treatment for the removal of NOM
(Ødegaard et al. 1999). The installation cost for adsorption pre-treatment with AERs is
relatively low and operation is simple (Hongve et al. 1999). Similarly, PAC adsorption
systems are relatively inexpensive to build. However, prolonged use of PAC at high doses
can be very uneconomical due to the need for replacing the spent PAC (Newcombe et al.
2008). The capital investment for ozonation systems is probably higher than that for the
adsorption processes (Oneby et al. 2010). However, the elimination of the need for
replacing the spent PAC or treating the AER concentrate makes it a more attractive and
probably a less expensive pre-treatment option. Furthermore, ozonation also provides the
disinfection effect which cannot be achieved with the adsorption pre-treatments.

Overall, for the AS effluent investigated, ozonation is the best pre-treatment for colour
removal, as it keeps the whole treatment process simple while reducing membrane fouling.

Chapter 7 Summary
The effect of adsorption pre-treatments for colour removal on membrane fouling in MF and
UF of the AS effluent was investigated in comparison with ozonation. It was found that like
ozonation (10 mg O3 L-1, 10-minute contact time), a colour removal of 75–80% was also
achieved with adsorption pre-treatment with the AER (10 mL L-1, 20-minute contact time,
1 BV of the AER treated 1500 BVs of the raw effluent) or the PAC (Acticarb PS1300, 150
mg L-1, 30-minute contact time). Of these, ozonation was the only method which reduced
membrane fouling. The reduction in fouling resulted from the breakdown in the molecular
structures of the membrane foulants (i.e., biopolymers and HS). Pre-treatments with the
PAC and the AER neither reduced nor increased fouling, as the components of the EfOM
removed by the PAC and the AER did not make significant contributions to membrane
fouling. As ozonation generated some building blocks, low MW acids and low MW
neutrals, it may increase the risk of membrane biofouling. As mentioned in Chapter 6, BAC
filtration of the ozonated effluent showed potential to alleviate this problem. Adsorption

142
pre-treatment with the PAC may reduce biofouling as the PAC removed a significant
proportion of the building blocks and low MW neutrals from the raw effluent. Advantages
and disadvantages of the three pre-treatments in terms of practicability were also discussed.

143
 CHAPTER 8. CHEMICAL CLEANING OF FOULED UF POLYMERIC
MEMBRANES

Although feed pre-treatments can reduce fouling and periodic membrane backwashing can
significantly recover the flux by displacing some foulants, the accumulation of
hydraulically irreversible foulants over long-term runs requires chemical cleaning of
membranes to maintain their flux performance. This chapter investigates the ability of
different cleaning agents to restore the performance of the 50 kDa PVDF membrane (HFM-
116) fouled by the raw AS effluent. Analyses of the fouled, backwashed, and chemically
cleaned membranes with FTIR spectroscopy and SEM/EDX spectroscopy provided insights
to the differences in the cleaning efficiency of the chemicals. The HFM-116 membrane was
selected for this study for its high retention of biopolymers and HS, which facilitated the
tracking of these compounds on the membrane during the fouling and cleaning processes
using FTIR spectroscopy and SEM/EDX spectroscopy. The membrane cleaners
investigated included an acid (HCl), a base (NaOH), a chelating agent (Na4EDTA), two
oxidising agents (NaOCl and H2O2), a surfactant (sodium dodecyl sulphate), and a
protease-based cleaner (Terg-a-zyme). The three most effective cleaners for the HFM-116
membrane were also tested on the 30 kDa PES membrane (HFK-141).

8.1. Removal of foulants by backwashing

As shown in Chapter 4, backwashing removed some proteins and polysaccharides retained
by the fouled membranes, which could be attributed to these high MW compounds being
larger than the membrane pores and forming a cake layer on the membrane surface.
However, backwashing was not effective in displacing HS from the membranes, suggesting
that HS adhered to the membrane surface and/or pore walls through strong hydrophobic
attraction.

8.2. Cleaning efficiencies of different chemicals

The membrane fouling and cleaning procedures are described in Chapter 3 (section 3.6).
Briefly, filtration of the raw AS effluent (collected on 19 February 2010, its characteristics
are shown in Table 8.1) was conducted at 110 kPa using samples of the virgin HMF-116
membrane. The filtration was stopped when the accumulated specific permeate volume (v)
reached 150 L m-2 (which took approximately 150 minutes). The fouled membranes were

144
backwashed with the permeate at 110 kPa for 5 minutes and their pure water flux was
measured with Milli-Q water. The backwashed membranes were then soaked in solutions
of the cleaning agents (with gentle shaking in an orbital shaker at 100 rpm) at 24 °C for 45
minutes. The pure water flux of the chemically cleaned membrane was determined after
rinsing them thoroughly with Milli-Q water.

Table 8.1. Characteristics of the raw effluent sample used in the membrane cleaning
experiments
Parameter Value
pH 7.8
DOC (mg L-1) 13.2
UVA254 (cm-1) 0.402
SUVA (m-1mg-1 L) 3.04
True colour (Pt-Co units) 113
Turbidity (NTU) 3.5

The cleaning efficiencies ( E RW ) of the chemicals for the fouled HFM-116 membrane are

shown in Figure 8.1 (sample calculations of E RW are given in Appendix C). The E RW of
NaOH was low, indicating that its hydrolytic effect (on polysaccharides and proteins) and
solubilisation effect (on HS) (Liu et al. 2001) was not sufficient for effective foulant
removal. HCl gave an E RW of 9% for the HFM-116 membrane, which was attributed to the
mild oxidation of some proteins and polysaccharides (Liu et al. 2001).

145
 100 93
87 88

80
67

60
E RW (%)

40
28 25
23
20 9

)
2)
)

8)
.5)

6)
.5)

.5
.5

4.5

6.
10

9.
H
10

10

10
,p

H
H
H

pH

,p

,p
,p
M

,p
,p
,p

M
m

.%

.%
e,

O2
M

m
rin
0

wt

wt
m

H2
(1

0
lo

.0

1.0
.4
.3

(1
l

ch

t. %
HC

(1
(0
(0

e(
m

SD
O2
H

TA

ym
pp
O

.0
H2
ED
Na

-z
(1
00

-a
4

(2

H
Na

rg
aO
Cl

Te
/N
O
Na

O2
H2

Figure 8.1. Cleaning efficiencies (ERW) of different cleaners

on the fouled HFM-116 membrane

The higher cleaning efficiency of the chelating agent Na4EDTA than NaOH (both at pH
10.5) indicated that some of the HS remaining on the membrane after backwashing was
probably in the form of humic-calcium complexes and Na4EDTA recovered the membrane
flux by simultaneously removing calcium from these complexes (through ligand exchange)
(Hong & Elimelech 1997) and dissolving the foulants.

NaOCl and H2O2 are both oxidising agents which are believed to remove membrane
foulants by oxidation reactions (Liu et al. 2001; Laabs et al. 2006; Porcelli & Judd 2010).
Oxidation breaks down functional groups of organic foulants to carboxyl, ketonic, and
aldehyde groups, making the foulants more susceptible to hydrolysis at high pH. As a result
of these reactions, the foulants become water soluble and detach from the membrane. In
this study H2O2 gave a much lower E RW than NaOCl, although its concentration (10,000
ppm or 294 mM) was much higher than that of NaOCl (200 ppm chlorine or 5.6 mM).
Since the E RW of the H2O2/NaOH mixture (pH 10.5) was higher than the E RW of H2O2 and
NaOH (Figure 8.1), the low efficacy of H2O2 may be attributed to the acidic nature of this
chemical. Chlorine may also have caused swelling of membranes, which increased the mass

146
transfer of the cleaning agent to the membrane surface (Zeman & Zydney 1996), resulting
in a higher E RW for NaOCl. Membrane cleaning with NaOCl and H2O2 was also

conducted at 40ºC and the E RW of NaOCl and H2O2 at this temperature was 98% and 77%,
respectively. Thus the higher temperature improved cleaning efficiency and NaOCl was a
better cleaning agent overall. On the other hand, Arnal et al. (2009) found that NaOCl
performed better than H2O2 at 25ºC, but the reverse happened at 40ºC when cleaning a
polysulfone UF membrane fouled by a surface water. It should be noted that although the
oxidative power of H2O2 is typically greater than that of OCl¯, exceptions have been
observed, depending on the characteristics of the solution and the compounds to be
oxidised (Crittenden et al. 2006).

SDS and Terg-a-zyme were the two most effective cleaning agents. The high E RW of SDS
confirmed that hydrophobic attraction was the major force keeping the hydraulically
irreversible foulants and the membrane together. It is generally understood that during
membrane cleaning with SDS that the hydrophobic tails of the surfactant molecules adsorb
to the foulant molecules and their hydrophilic heads are orientated towards the aqueous
phase (Gecol 2006). This reduces the hydrophobicity of the foulants and they are
solubilised into the aqueous phase. At SDS concentrations higher than the critical micelle
concentration (8.36 mM in deionised water at 20°C (Li & Elimelech 2004)), as in this
study, micelles also form in the cleaning solution. These micelles diffuse into the fouling
layer, dissociate and adsorb as monomers on the foulant molecules and enhance foulant
solubilisation (Li & Elimelech 2004). The fact that the SDS solution and Terg-a-zyme
solution turned yellow during the membrane cleaning experiment confirmed that HS played
a role in causing hydraulically irreversible fouling.

Terg-a-zyme is a blend of the anionic surfactant sodium dodecyl benzene sulphonate

(SDBS, 10–20 wt.%) and a protease enzyme (Bacillus licheniformis Subtilisin Carlsberg)
(Alconox 2006, 2011). At the Terg-a-zyme concentration used in this study (1.0 wt.%), the
SDBS concentration was between 2.9 and 5.7 mM, much higher than the critical micelle
concentration of SDBS (approximately 0.5 mM in deionised water at 24 °C (Weiss et al.
2007)). Therefore, the mechanism of foulant removal by SDBS was expected to be similar
to that of SDS (as described above). The higher E RW of Terg-a-zyme than SDS may be due
to the higher pH (which facilitated the dissolution of the foulants (Liu et al. 2001)) and/or

147
the surfactant and the enzyme acting together to solubilise the foulants. The enzyme may
have broken down some of the protein foulants, making them soluble in the cleaning
solution and/or more readily solubilised by the surfactant.

NaOCl, SDS, and Terg-a-zyme also showed high cleaning efficiencies for the fouled HFK-
141 membrane. The values of E RW of these cleaners were 96%, 103%, and 108%,
respectively. The increases in the pure water flux of the membrane after cleaning with SDS
and Terg-a-zyme can be attributed to these chemicals clearing the pores of materials that
were left from the membrane manufacturing process and were not removed by cleaning
with Milli-Q water (Nyström & Zhu 1997). The material was probably polyvinyl
pyrrolidone (PVP), which has been used widely in the initial casting of PES membranes as
a pore forming agent and a hydrophilising agent (Arkhangelski et al. 2007). It is very likely
that PVP was used in manufacturing the HFK-141 membrane, as the membrane was more
hydrophilic (contact angle = 412º (Table 3.4)) than non-modified PES (contact angle  60º
(Cheryan 1986)). It was also possible that the adsorption of the surfactants made the
membrane become more hydrophilic and therefore had a slightly higher pure water flux
compared with the virgin membrane (Nyström & Zhu 1997). These hypotheses were
supported by the fact that cleaning the virgin HFK-141 membrane with SDS and Terg-a-
zyme increased the pure water flux of the membrane by approximately 5%. The possibility
of SDS or Terg-a-zyme causing damage of the membrane was low, since the pHs of these
cleaners (pH 6.8 for SDS and pH 9.6 for Terg-a-zyme) were within the pH limit of the
membrane (pH 2.0–11.0), and virgin membranes were used in the tests. Furthermore, the
concentrations of the cleaners investigated (10 mM SDS and 1.0 wt.% Terg-a-zyme) were
those commonly used in practice (Alconox 2006; Al-Amoudi & Lovitt 2007).

From economic and practical viewpoints, NaOCl is superior to SDS and Terg-a-zyme since
it is less expensive and does not create foams like the other two. However, the use of SDS
or Terg-a-zyme may reduce the risk of membrane degradation caused by NaOCl in long-
term use.

Three fouling and cleaning cycles were also conducted on the HFM-116 membrane with
the three most effective cleaners and accumulation of residual fouling was observed (Figure
8.2).

148
 Cycle 1
Cycle 2
100 Cycle 3

80
WFR (%)
60

40

20

0
NaOCl SDS Terg-a-zyme

Figure 8.2. Changes in WFR of NaOCl, SDS, and Terg-a-zyme over three UF cycles with
the HFM-116 membrane

Although this did not appear to affect the permeate flux in UF of the AS effluent in the first
three filtration cycles (Figure 8.3, data for SDS and Terg-a-zyme are included in Appendix
J), it can be inferred that such a reduction in the permeate flux would occur in long-term
operation and therefore cleaning-in-place (CIP) with longer soaking time and/or higher
cleaning agent concentrations would be needed. Elevated temperature (within
manufacturer’s specifications) may also be used to increase the cleaning efficiency and
reduce the cleaning time. The rejections of DOC and UV-absorbance (UVA254) of the
virgin membranes and the chemically cleaned membranes were comparable (Table 8.2),
indicating that high flux recoveries were not associated with degradation of the membranes
over the 3 chemical cleaning cycles.

149
 180
Virgin membrane
160
After first chemical cleaning
140
After second chemical cleaning
120 After third chemical cleaning
J (L m h )
-1

100
-2

80

60

40

20

0
-2
0 50 v (L m ) 100 150

Figure 8.3. Effect of cyclic cleaning with Terg-a-zyme on the permeate flux in UF of the
raw effluent with the HFM-116 membrane

Table 8.2. Effect of chemical cleaning with NaOCl, SDS, and Terg-a-zyme on the rejection
of DOC and UV-absorbance of the HFM-116 membrane
(Feed: DOC = 13.2 mg L-1 and UVA254 = 0.402 cm-1)
Permeate Membrane after Membrane after
from UF Membrane after first
Virgin membrane second chemical third chemical
with chemical cleaning
cleaning cleaning
DOC UVA254 DOC UVA254 DOC UVA254 DOC UVA254
(mg L-1) (cm-1) (mg L-1) (cm-1) (mg L-1) (cm-1) (mg L-1) (cm-1)
NaOCl 10.7 0.317 10.9 0.322 10.8 0.320 10.7 0.319

SDS 10.7 0.314 10.5 0.311 10.5 0.312 10.5 0.310

Terg-a-
10.8 0.321 10.9 0.325 10.8 0.324 10.9 0.327
zyme

8.3. Membrane cleanliness shown by ATR-FTIR spectroscopy and EDX spectroscopy

Backwashing removed most of the foulants on the surface of the HFM-116 membrane,
leaving only traces of polysaccharides and proteins (Figure 8.4, peaks at 1640 cm-1 and
3000–3650 cm-1). Although the cleaning efficiencies of the chemicals investigated covered
a wide range (from 9% for HCl to more than 85% for NaOCl, SDS and Terg-a-zyme), the
FTIR spectra of the chemically cleaned membranes were very similar to the spectrum of the
virgin membrane. The results showed that FTIR spectroscopy was a useful tool for the

150
identification of organic foulants on the membrane surface but could not be used to
differentiate between the virgin and the cleaned membranes.

0.35
Virgin
0.3 Fouled
1069 Backwashed
0.25 Cleaned with NaOH
1540
Absorbance

0.2

0.15 1640
3000-3600
0.1
1720
2920
0.05

0
500 1000 1500 2000 2500 3000 3500 4000
-1
-0.05 Frequency (cm )

Figure 8.4. ATR-FTIR spectra of the virgin, fouled, and cleaned HFM-116 membranes

SEM images of the virgin, fouled and cleaned membranes were collected. The surface of
the fouled membrane was covered with the foulant layer, which made the pores invisible.
After backwashing, the pores on the membrane surface became visible, together with some
remaining trace foulants. The SEM images of the chemically cleaned membranes were very
similar to that of the virgin membrane (see Appendix K). SEM therefore did not allow
differentiation of the membranes treated with the different cleaning agents.

The EDX spectrum of the virgin membrane (Figure 8.5) indicated the presence of carbon,
oxygen, and fluorine. The presence of the foulants on the membrane resulted in an increase
in the oxygen peak, a large reduction in the fluorine peak, and the appearance of the
nitrogen, phosphorus, and sulphur peaks. These changes were attributed to the presence of
polysaccharides, proteins and HS in the fouling layer. A trace amount of calcium was also
detected, indicating that calcium may have played a role in causing membrane fouling.
Calcium ion has been found to contribute to membrane fouling in UF by promoting the
aggregation of organic molecules in solution via the formation of calcium bridges between
the carboxylic groups of HS (Ahn et al. 2008). This has been discussed in section 8.2.

151
 7500 F Virgin
Fouled
Backwashed
6000 C

1000
Counts
4500
P S
Ca
500
O
3000
0
1 1.5 2 2.5 3 3.5 4 4.5

1500
N

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
keV

Figure 8.5. EDX spectra of the virgin, fouled, and backwashed HFM-116 membranes

The EDX spectrum of the backwashed membrane had a higher oxygen peak compared with
the virgin membrane (Figure 8.5) due to the presence of organic foulants remaining after
backwashing. The EDX spectra of the chemically cleaned membranes were very similar to
the spectrum of the virgin membrane and therefore the technique was not able to
differentiate the degree of cleanliness of these membranes.

Chapter 8 Summary
Backwashing and chemical cleaning of the PVDF (50 kDa) and PES (30 kDa) membranes
used for UF of AS effluent were investigated to gain insights to the interactions between
the foulants and the membranes. Hydraulic backwashing removed some of the
polysaccharides and proteins retained on the membranes, but was not effective for
removing HS. The alkaline cleaners were much more efficient than the acidic cleaners for
removing the hydraulically irreversible foulants. The most effective cleaning agents were
NaOCl, SDS, and Terg-a-zyme, with cleaning efficiencies greater than 85%. The high
cleaning efficiencies of the SDS (a surfactant) and Terg-a-zyme (a blend of an anionic
surfactant and a protease) indicated that hydrophobic attraction was the major force keeping
the hydraulically irreversible foulants and the membranes together. The contribution of HS
to hydraulically irreversible fouling was verified by the fact that these solutions turned

152
yellow during the cleaning experiments. Accumulation of residual fouling after chemical
cleaning was observed. Membrane cleanliness was assessed using a combination of flux
measurement (cleaning efficiency) and surface analyses (ATR-FTIR spectroscopy,
SEM/EDX), the former being a more reliable indicator.

153
 CHAPTER 9. MEMBRANE FOULING IN CROSS-FLOW FILTRATION

Previous chapters investigated fouling mechanisms and the effect of different feed pre-
treatments on membrane fouling by the AS effluent using PVDF and PES membranes in
dead-end mode. This aligned with current practice which favoured the use of polymeric
membranes and the dead-end filtration mode. Although membranes made from ceramic
have been used widely in the food, beverage, and dairy industries and in the treatment of
industrial wastewater, the high membrane costs have limited their use in water and
municipal wastewater treatment. Recently, decreasing membrane cost and improvements in
module design have made them promising alternatives for polymeric membranes. As a
result, research into fouling of ceramic membranes by surface water has increased in the
past 5 years. However, the number of studies on the fouling of ceramic membranes in MF
and UF of secondary effluent is still limited, particularly in cross-flow filtration mode.

In this chapter the fouling of single-channel tubular ceramic membranes in cross-flow MF

and UF of the raw AS effluent was investigated. The MF membrane was made of -
alumina and its pore size was 0.1 μm. The UF membrane had a pore size of 20 nm and
comprised a zirconia layer coated on an alumina support. The membrane foulants were
determined. The effects of rapid back-flushing of the membrane and feed pre-treatments
with ozone, the AER, and alum on the permeate flux were also studied. To align with
current pilot plant practice, the vast majority of the MF and UF tests were conducted at a
cross-flow velocity of 1.8 m s-1 and at a starting flux of 350 L m-2 h-1 (MF, TMP = 20 kPa)
and 100 L m-2 h-1 (UF, TMP = 30 kPa). The corresponding Reynolds number was 12000,
within the regime (1000–15000) in which most commercial membrane modules operate
(Belfort et al. 1994).

154
The characteristics of the raw AS effluent used in cross-flow MF and UF tests are shown in
Table 9.1. To avoid the influence of sample aging on the results, different sets of samples
were used throughout the study. The characteristics of the samples collected on particular
dates are included in Appendix L.

Table 9.1. Characteristics of the raw effluent samples used in cross-flow MF and UF tests
Parameter Value
pH 7.3 – 7.8
DOC (mg L-1) 12.1 – 15.1
UVA254 (cm-1) 0.37 – 0.40
SUVA (L m-1mg-1) 2.6 – 3.2
True colour (Pt-Co units) 81 – 108
Turbidity (NTU) 1.9 – 3.5

9.1. Flux decline and membrane foulants in cross-flow MF and UF of the raw effluent

9.1.1. Effect of SS on flux decline

MF and UF of the raw AS effluent with and without pre-filtration through 0.5 µm glass-
fibre filters were conducted to determine the effect of SS on the flux decline. The results
(Figure 9.1) showed that different from dead-end MF and UF with the HFM-180 membrane
(Chapter 4), the permeate flux in cross-flow MF and UF was not affected by the SS. This
can be attributed to the cross-flow preventing the deposition of SS on the membrane
surface, and therefore dissolved EfOM was solely responsible for the flux decline.

155
 350
(a) Raw
300 0.5 micron pre-filtered Raw

250
J (L m h )
-1

200
-2

150

100

50

0
0 20 40 60 80 100
t (min)

100
(b) Raw
0.5 micron pre-filtered Raw
80
J (L m h )
-1
-2

60

40

20
0 20 40 60 80 100
t (min)

Figure 9.1. Flux data for (a) MF (TMP = 20 kPa) and (b) UF (TMP = 30 kPa)
of the raw AS effluent with and without SS at cross-flow velocity 1.8 m s-1
(Sample collected on 01 April 2011, DOC = 14.3 mg L-1, UVA254 = 0.371 cm-1,
true colour = 110 Pt-Co units, turbidity = 6.6 NTU)

156
9.1.2. Membrane foulants in cross-flow MF and UF

To identify the EfOM components responsible for fouling of the membranes, the feed and
the permeates collected after 1 hour of MF and UF of the raw AS effluent were subjected to
LC-OCD analysis. The filtration conditions were: cross-flow velocity = 1.8 m s-1, TMP =
20 kPa (MF) and 30 kPa (UF), and volume concentration factor (VCF) = 1.3 (MF) and 1.2
(UF).

15 HS Building blocks

Low MW acids and low MW HS

OCD response (relative units)

10
Raw
Biopolymers Raw + MF
Raw + UF

Low MW neutrals

0
20 30 40 50 60 70 80 90 100
Retention time (min)

Figure 9.2. AMWDs of the raw AS effluent and the permeates in cross-flow filtration
(Raw effluent was collected on 08 July 2011, DOC = 15.1 mg L-1,
UVA254 = 0.389 cm-1, true colour = 100 Pt-Co units, turbidity = 1.9 NTU)

The chromatograms (Figure 9.2) showed that similar to dead-end MF and UF, biopolymers
(which contained proteins and polysaccharides) and HS were the compounds retained most
significantly on the membranes and were therefore the potential foulants. The role of
biopolymers in causing fouling of the ceramic membranes was further confirmed by the
fact that the pre-treatment which did not affect the biopolymers in the raw effluent (i.e. pre-
treatment with the AER) did not affect the permeate flux, whereas those leading to the
removal of biopolymers (coagulation) or breakdown in the molecular structures of these
components (ozonation) reduced membrane fouling (sections 9.3 and 9.4). Similar results

157
regarding the identity of the foulants were obtained by Mueller and co-workers (Mueller et
al. 2008; 2009) for UF of surface water using a titania membrane (pore size = 50 nm). In
their study, cleaning-in-place (CIP) of the fouled membrane was performed with NaOCl
and the CIP waste was examined with LC-OCD and fluorescence spectroscopy which
showed the dominance of biopolymers and HS in the CIP waste.

Biopolymers, with their large molecular sizes, were likely to foul the membranes mainly by
pore blocking and cake formation. Some proteins can be hydrophobic in nature (e.g., those
with high proline content) and therefore may adsorb on the membrane pore walls. The
retention of HS by the membranes was attributed to their hydrophobic interactions with the
membranes (Liu et al. 2001; Jarusutthirak et al. 2002). This was supported by the small
molecular sizes of HS (1,000–20,000 Da) and the fact that colour removal by the
membranes was observed from the beginning of filtration. Although the contact angles of
the membranes used in this study were not measured, the literature suggests that alumina
and zirconia membrane can be hydrophobic (contact angle >50°). For example, Redón et al.
(2005) and Koonaphapdeelert & Li (2007) reported contact angles of 55–80° for alumina
membranes. A contact angle of 75° for a zirconia membrane was reported by Dobrak et al.
(2010).

9.1.3. DOC and colour rejection by the ceramic membranes

The rejection by the MF membrane was 6–10% for DOC (approximately 0.8–1.5 mg C L-1)
and 8–10% (about 8 Pt-Co units) for true colour, which were comparable with the
rejections of the polymeric membranes of the same pore size (VVLP and XUNP-003
membranes). These values for the ceramic UF membrane were slightly higher: 12–15% for
DOC (1.6–2.4 mg C L-1) and 12–16% for colour (about 12 Pt-Co units). The turbidity of
the MF and UF permeates was between 0.1 and 0.2 NTU, much lower than the requirement
of 2.0 NTU for “Class A” recycled water.

9.2. Effect of rapid back-flushing on flux performance in cross-flow filtration

In the previous sections of this chapter fouling of the ceramic membranes without
intermittent removal of the cake formed on the membrane surface was studied. In practice,
to improve the flux performance, the cake layer is normally removed by backwashing the
membrane at a predetermined time interval or when the TMP exceeds a preset value (the

158
latter is applied to constant-flux operation). In typical membrane filtration plants, the
backwash duration may last from 10 seconds to 10 minutes (Adham et al. 2005). The
backwash pressure is limited by the physical properties of the membrane. This value can be
much higher for ceramic than polymeric membranes due to the greater strength of the
former. In cross-flow filtration with tubular ceramic membranes, a technique called “rapid
back-flushing” has been used to improve flux performance in MF of water and wastewater
from industrial sources (e.g., electroplating wastewater) (Redkar & Davis 1995; Sondhi et
al. 2000). In this method, the permeate is pushed back to the feed side at high pressure for a
very short duration (typically less than 1 second) (Sondhi et al. 2000). However, its
application to secondary effluent has been limited, probably due to the limited use of
ceramic membranes for the treatment of municipal wastewater. In this section, the
effectiveness of rapid back-flushing on fouling reduction in cross-flow MF and UF with the
ceramic membrane was investigated. Back-flushing was accomplished by a piston which
rapidly pushed the permeate back to the feed side when activated. The back-flush pressure,
frequency, and duration were 600 kPa, 12 min-1, and 1 second, respectively.

The flux data for MF of the raw AS effluent at a cross-flow velocity of 1.8 m s-1 and a TMP
of 20 kPa with and without membrane back-flushing are shown in Figure 9.3. It can be seen
that back-flushing the membrane with the permeate every 12 minutes resulted in significant
flux improvement and this effect was sustained during 7 hours of operation. The average
permeate flux at the end of the test with back-flushing (84 L m-2 h-1) was more than twice
that obtained without back-flushing (38 L m-2 h-1) (Table 9.2). Furthermore, the
accumulated specific permeate volume increased by nearly 90% with the application of
rapid back-flush. Comparable improvements in the end flux and accumulated specific
permeate volume were also achieved in UF (Figure 9.4 and Table 9.2). At the end of the 7-
hour UF tests, the flux without back-flushing was 33 L m-2 h-1, whereas the average flux
with back-flushing was approximately 60 L m-2 h-1. The accumulated specific permeate
volume was also increased by 80% with the use of rapid back-flushing.

159
 250
Raw, 1.8 m/s, without back-flush

200 Raw, 1.8 m/s, with back-flush

Raw, 2.7 m/s, without back-flush
J (L m h )

150
-1
-2

100

50

0
0 60 120 180 240 300 360 420
t (min)

Figure 9.3. Effect of rapid back-flushing and increasing cross-flow velocity on flux
performance in MF of the raw AS effluent
(TMP = 20 kPa, back-flush pressure, frequency and duration were 600 kPa, 12 min-1 and 1
second, respectively, sample collected on 23 May 2011)
(The lowest data points on the flux curve for the case “with back-flush”
corresponded to the permeate loss to back-flushing)

160
 100

80
J (L m h )

60
-1
-2

40

Raw, 1.8 m/s, without back-flush

20
Raw, 1.8 m/s, with back-flush
Raw, 2.7 m/s, without back-flush
0
0 60 120 180 240 300 360 420
t (min)

Figure 9.4. Effect of rapid back-flushing and increasing cross-flow velocity on flux
performance in UF of the raw AS effluent
(TMP = 30 kPa, back-flush pressure, frequency and duration were
600 kPa,12 min-1, and 1 second, respectively
(Sample collected on 15 April 2011)

Examination of the flux data in Figures 9.3 and 9.4 revealed that back-flushing was least
effective in restoring the permeate flux in the first 3–4 cycles (i.e., the first 36–48 minutes).
This could be explained by the limited ability of back-flushing to remove the foulants
adsorbed on the membrane pore walls or blocking the membrane pores. As filtration
proceeded, the sites on the membranes available for pore adsorption and pore blocking
became saturated and cake formation became the dominant mechanism causing the flux
decline. It was from this point that back-flushing was most effective in dislodging the cake
layer from the membrane surface.

In cross-flow filtration, flux improvement can also be achieved by operating at higher feed
cross-flow velocities. The flux data obtained at a cross-flow velocity of 2.7 m s-1 and the
same TMP (20 kPa for MF and 30 kPa for UF) showed that the flux improvement by back-
flushing was slightly higher than that achieved by increasing the cross-flow velocity by
50% (Figures 9.3 and 9.4 and Table 9.2).

161
Table 9.2. Average flux and accumulated specific permeate volume at the end of 7-hour
MF and UF tests with the raw AS effluent at various operating conditions

Average flux at the end of Specific permeate volume

Case
7- hour run (v) after 7- hour run

MF, 1.8 m s-1, 20 kPa,

38 L m-2 h-1 370
without back-flush

MF, 1.8 m s-1, 20 kPa,

84 L m-2 h-1 698
with back-flush

MF, 2.7 m s-1, 20 kPa,

75 L m-2 h-1 657
without back-flush

UF, 1.8 m s-1, 30 kPa,

32 L m-2 h-1 287
without back-flush

UF, 1.8 m s-1, 30 kPa,

60 L m-2 h-1 522
with back-flush

UF, 2.7 m s-1, 30 kPa,

57 L m-2 h-1 456
without back-flush

For the filtration conditions investigated, rapid back-flushing was probably more energy-
efficient than increasing cross-flow velocity. The energy requirements of the back-flush
device and the back-flush timer were small and the device only worked intermittently,
whereas the feed pump consumed a 3.37 (= 1.53) times higher amount of energy to increase
the feed cross-flow velocity by 50%. However, it should be pointed out that as rapid back-
flushing has not been practised widely in the filtration of water and municipal wastewater,
it may be difficult to scale up the back-flush device, which was basically a piston driven by
air pressure, for use in full-scale filtration plants.

9.3. Effect of feed pre-treatments with ozone and the AER on permeate flux
As mentioned in Chapter 6, ozonation was the preferred pre-treatment for its ability to
remove colour and improve the flux performance in dead-end MF and UF with the
polymeric membranes. Pre-adsorption with the AER was able to reduce the true colour of
the effluent to 20–25 Pt-Co units at a practical dose (10 mL L-1), whereas a high (and
probably uneconomical) PAC dose (150 mg L-1) was needed to achieve this colour removal
target. In this section, the effect of ozonation and adsorption pre-treatment with the AER on
the flux performance of the ceramic membranes in cross-flow filtration mode was

162
investigated. Relatively short (2-hour) MF and UF tests with the raw, ozonated, and AER-
treated effluents were conducted at a cross-flow velocity of 1.8 m s-1 and a TMP of 20 kPa
(MF) and 30 kPa (UF) without the application of back-flushing. The ozone dose was 10 mg
L-1 with a contact time of 10 minutes. The AER dose was 10 mL resin L-1 with a contact
time of 20 minutes and stirring at 100 rpm. One bed volume (BV) of the AER was used to
treat 1500 BVs of the raw AS effluent. The characteristics of the samples used are shown in
Table 9.3.

Table 9.3. Characteristics of the raw, ozonated, and AER-treated effluent samples used in
cross-flow MF and UF tests
(Samples collected on 08 July 2011)
Parameter Raw Ozonated AER-treated
pH 7.8 7.7 7.7
DOC (mg L-1) 15.1 15.6 9.5
UVA254 (cm-1) 0.389 0.227 0.144
SUVA (m-1mg-1 L) 2.58 1.46 1.52
True colour (Pt-Co units) 100 22 22
Turbidity (NTU) 1.9 1.8 Not measured

The flux data obtained (Figures 9.5 and 9.6) showed that similar to dead-end MF and UF
with the polymeric membranes (Chapter 7), ozonation of the raw AS effluent significantly
improved the flux performance of the membranes, whereas pre-treatment with the AER
neither reduced nor worsened membrane fouling.

163
 360
Raw
300 Ozonated
AER-treated
J (L m h ) 240
-1
-2

180

120

60

0
0 20 40 60 80 100 120
t (min)

Figure 9.5. Effect of pre-treatments for colour removal with

ozone and the AER on permeate flux in cross-flow MF
(Samples collected on 08 July 2011)

120
Raw
100 Ozonated
AER-treated
80
J (L m h )
-1
-2

60

40

20

0
0 20 40 60 80 100 120
t (min)

Figure 9.6. Effect of pre-treatments for colour removal with ozone

and the AER on permeate flux in cross-flow UF
(Samples collected on 08 July 2011)

The permeates from the raw, ozonated and AER-treated effluents collected after 1 hour of
filtration were subjected to LC-OCD analysis. The results (Figures 9.7 and 9.8 and Table

164
9.4) showed that more biopolymers and HS in the ozonated effluent were able to permeate
through the membranes than in the raw and the AER-treated effluents. Therefore, similar to
the case of the polymeric membranes (Chapters 6 and 7), the flux improvement by
ozonation in cross-flow filtration with ceramic membranes can be attributed the breakdown
of polysaccharides, proteins, and HS to less hydrophobic and lower MW compounds. The
AER removed nearly 70% of HS and essentially no biopolymers from the raw effluent. The
UF membrane retained a lower amount of HS in the AER-treated effluent than in the raw
effluent. However, this reduction in HS rejection was not adequate for flux improvement,
indicating that polysaccharides and proteins played a much more important role in causing
flux decline.

Table 9.4. Effect of ozonation and pre-treatment with the AER on the rejection of
biopolymers and HS (measured in mg C L-1) by the MF and UF ceramic membranes
Biopolymers
Sample Biopolymers HS HS rejection
rejection
Raw 1.3 8.0 - -
Raw + MF 0.4 7.5 0.9 0.5
Raw + UF 0.2 7.0 1.1 1.0
Ozonated 1.4 7.6 - -
Ozonated + MF 0.8 7.3 0.6 0.3
Ozonated + UF 0.6 7.0 0.8 0.6
AER-treated 1.3 2.5 - -
AER-treated + MF 0.4 2.1 0.9 0.4
AER-treated + UF 0.2 1.9 1.1 0.6

165
 15 HS Building blocks
(a)

OCD response (relative units)

Low MW acids and low MW HS

10
Raw
Raw + MF
Biopolymers
Ozonated
Ozonated + MF
5

Low MW neutrals

0
20 30 40 50 60 70 80 90 100
Retention time (min)

15
HS
(b) Building blocks
OCD response (relative units)

Low MW acids and low MW HS

10

Raw
Biopolymers Raw + MF
AER-treated
AER-treated + MF
5

Low MW neutrals

0
20 30 40 50 60 70 80 90 100
Retention time (min)

Figure 9.7. Effect of (a) ozonation and (b) adsorption pre-treatment with the AER on the
rejection of proteins and polysaccharides of the ceramic MF membrane
(Cross-flow velocity = 1.8 m s-1, TMP = 20 kPa, without back-flushing, VCF  1.3)
(Samples collected on 08 July 2011)

166
 15 HS Building blocks
(a)

OCD response (relative units)

Low MW acids and low MW HS

10
Raw
Raw + UF
Biopolymers Ozonated
Ozonated + UF

Low MW neutrals

0
20 30 40 50 60 70 80 90 100
Retention time (min)

15 HS Building blocks
(b)

Low MW acids and low MW HS

OCD response (relative units)

10
Raw
Raw + UF
Biopolymers
AER-treated
AER-treated + UF
5

Low MW neutrals

0
20 30 40 50 60 70 80 90 100
Retention time (min)

Figure 9.8. Effect of (a) ozonation and (b) adsorption pre-treatment with the AER
on the rejection of different EfOM components of the ceramic UF membrane
(Cross-flow velocity = 1.8 m s-1, TMP = 30 kPa, without back-flushing, VCF  1.2)
(Samples collected on 08 July 2011)

167
The effectiveness of ozonation in fouling reduction was also tested in the back-flushing
mode. The results (Figures 9.9 and 9.10) showed that flux improvement by ozonation was
also achieved in MF and UF operated in this mode.

250

Raw

Ozonated
200
J (L m h )
-1
-2

150

100

50
0 60 120 180 240 300 360 420
t (min)

Figure 9.9. Effect of ozonation on the permeate flux

in cross-flow MF with rapid back-flushing
(Cross-flow velocity = 1.8 m s-1, TMP = 20 kPa, back-flush pressure = 600 kPa,
back-flush frequency = 12 min-1, back-flush duration = 1 second)
(Samples collected on 09 September 2011, their characteristics are shown in Appendix L)

168
 100

Raw
90
Ozonated

80
J (L m h )
-1
-2

70

60

50

40
0 60 120 180 240 300 360 420
t (min)

Figure 9.10. Effect of ozonation on the permeate flux

in cross-flow UF with rapid back-flushing
(Cross-flow velocity = 1.8 m s-1, TMP = 30 kPa, back-flush pressure = 600 kPa,
back-flush frequency = 12 min-1, back-flush duration = 1 second)
(Samples collected on 30 September 2011, their characteristics are shown in Appendix L)

9.4. Effect of feed coagulation with alum on flux performance in cross-flow MF

Ozonation (with or without subsequent BAC filtration) was the preferred pre-treatment for
MF and UF for its ability to decolourise the raw effluent while improving the permeate
flux. The product water would be suitable for non-potable reuse purposes (EPA Victoria
2003). Another method for colour removal is to treat the raw AS effluent with RO
membranes. This method is energy-intensive. However, the final water product would have
superior quality, which may meet the requirements for potable reuse purposes (Water
Environment Federation 2006). If this option is selected, the raw AS effluent would need to
be pre-treated to reduce fouling of the RO membrane. Currently, the most widely used
processes for pre-treatment of RO feeds are coagulation-MF/UF followed by RO (Qin et al.
2004; Water Environment Federation 2006). Coagulation is intended to reduce fouling of
the MF or UF membranes. For secondary effluents, the most popular process configuration
is in-line coagulation (i.e., no allowance for the settling of coagulated matter) with low
coagulant doses. In this section, the effect of coagulation with alum at a dose of 2.5 mg Al3+

169
L-1 (without prior removal of the coagulated matter) on the permeate flux in MF (at a cross-
flow velocity of 1.8 m s-1 and TMP of 20 kPa) was investigated. Short (2-hour) tests were
conducted with the raw and the coagulated effluents without the application of rapid back-
flushing. Longer (7-hour) tests with rapid back-flushing every 12 minutes were also carried
out in an attempt to obtain flux profiles similar to those of pilot plants. The results (Figures
9.11 and 9.12) indicated that coagulation was an effective pre-treatment for fouling
reduction in cross-flow MF with and without the application of rapid back-flushing. Similar
to dead-end MF with the PVDF membrane (Chapter 5), the flux improvement by
coagulation can be attributed to the removal of a proportion of biopolymers and HS from
the raw AS effluent.

360
Raw
300
Coagulated

240
J (L m h )
-1
-2

180

120

60

0
0 20 40 60 80 100 120
t (min)

Figure 9.11. Effect of coagulation with alum on permeate flux in

cross-flow MF without membrane back-flushing
(Cross-flow velocity = 1.8 m s-1, TMP = 20 kPa)
(Sample collected on 30 September 2011)

170
 250

Raw

Coagulated
J (L m h ) 200
-1
-2

150

100

50
0 60 120 180 240 300 360 420
t (min)

Figure 9.12. Effect of coagulation with alum on permeate flux

in cross-flow MF with membrane back-flushing
(Cross-flow velocity = 1.8 m s-1, TMP = 20 kPa, back-flush pressure = 600 kPa,
back-flush frequency = 12 min-1, back-flush duration = 1 second))
(Samples collected on 23 May 2011)

Chapter 9 Summary
This chapter was dedicated to fouling and fouling mitigation in cross-flow MF and UF of
the AS effluent using tubular (single-channel) ceramic membranes. The MF membrane was
made of -alumina and its pore size was 0.1 μm. The UF membrane had a pore size of 20
nm and was made of zirconia coated on an alumina support. The majority of the tests were
carried out at a cross-flow velocity of 1.8 m s-1 and TMP of 20 kPa (MF) and 30 kPa (UF).

Tests with the raw AS effluent with and without pre-filtration through 0.5 μm glass-fibre
filters showed that different from dead-end MF and UF with the 100 kDa (HFM-180)
membrane, particles larger than 0.5 μm made no contribution to the flux decline in cross-
flow MF or UF. This was attributed to the cross-flow preventing the deposition of these
particles on the membrane surface. The membrane foulants were identified as biopolymers
(i.e., polysaccharides and proteins) and HS.

171
The effect of rapid membrane back-flushing, with the back-flush device being a piston
driven by air pressure, on the flux performance was studied. It was found that back-flushing
the membranes at 600 kPa for 1 second every 12 minutes resulted in significant flux
improvement, which was attributed to the removal of the foulant layer on the membrane
surface. However, it may be difficult to scale-up the back-flush system, as it has rarely been
used in the treatment of surface water and municipal wastewater.

Of the two feed pre-treatments for colour removal, ozonation (10 mg O3 L-1, contact time
10 minutes) reduced membrane fouling, whereas adsorption with the AER (10 mL resin
L-1, 20-minute contact time, 1 BV of the AER treated 1500 BVs of raw effluent) neither
increased nor reduced the permeate flux. Similar to dead-end MF and UF with the
polymeric membranes (Chapters 6 and 7), the flux improvement by ozonation was
attributed to the breakdown of some biopolymers (i.e., proteins and polysaccharides) and
HS to lower MW and less hydrophobic compounds. The ineffectiveness of the AER in
fouling reduction can be explained by its inability to remove polysaccharides and proteins
from the raw AS effluent.

Feed coagulation with alum at 2.5 mg Al3+ L-1 (without prior removal of the coagulated
matter) significantly improved the flux performance of the MF membrane. This indicated
an opportunity to utilise a coagulation/MF hybrid scheme with rapid back-flushing as a pre-
treatment for RO for the production of high-quality recycled water which may be suitable
for potable reuse purposes.

172
 CHAPTER 10. CONCLUSIONS & RECOMMENDATIONS

Fouling of polymeric and ceramic membranes by raw and pre-treated activated sludge (AS)
effluent was studied to obtain a more detailed understanding of the fouling mechanisms and
the effects of different feed pre-treatments on membrane fouling in microfiltration (MF)
and ultrafiltration (UF) of secondary effluent. The pre-treatments investigated included
coagulation without removal of the coagulated matter prior to MF or UF, ozonation, and
adsorption with an anion exchange resin (AER) or a powdered activated carbon (PAC). The
dosages of ozone, AER and PAC were selected to reduce the true colour of the raw effluent
by 75-80% (i.e., from 65–133 Pt-Co units to 15–25 Pt-Co units).

The research was divided into two parts. The first part dealt with dead-end filtration using
polyvinylidene fluoride (PVDF) and polyethersulfone (PES) membranes – the most widely
used membranes in surface water and municipal wastewater treatment. Two PVDF MF (0.1
μm) membranes, one was less hydrophobic (contact angle = 62º) than the other (contact
angle > 90º), and two hydrophobic PVDF UF membranes with different molecular weight
cut-offs (MWCO) (50 kDa and 100 kDa) were used. A relatively hydrophilic PES UF
membrane (MWCO = 30 kDa) was also investigated. The second part studied fouling and
fouling mitigation in cross-flow filtration with ceramic membranes. Single-channel tubular
membranes made of -alumina (pore size = 0.1 μm) and zirconia (pore size = 20 nm) were
used. The following conclusions were drawn from the experimental data:

10.1. Components in the effluent contributing to flux decline and fouling mechanisms
Both effluent organic matter (EfOM, materials smaller than 0.45 μm) and suspended solids
(SS, particles larger than 0.5 μm) contributed to the flux decline in dead-end filtration of the
effluent with the MF and the loose UF (100 kDa) membranes. However, SS did not
contribute to the flux reduction on the tighter UF membranes (PVDF 50 kDa and PES 30
kDa) or in cross-flow filtration. Most of the flux decline by the SS was hydraulically
reversible.

For both dead-end filtration with the polymeric membranes and cross-flow filtration with
the ceramic membranes, biopolymers, which contained proteins and polysaccharides
(molecular weight (MW)>>20,000 Da), and humic substances (HS, MW=1,000–20,000

173
Da) were the EfOM components responsible for fouling. The contribution of biopolymers
to membrane fouling was much greater than that of HS, as indicated by the higher rejection
of the former by the membranes. This was verified by the fact that the pre-treatments which
removed more than 50% of the HS from the raw effluent but did not affect the biopolymers
(i.e., pre-adsorption with the AER or the PAC) did not improve the membrane flux. Fouling
by HS was mainly hydraulically irreversible, whereas biopolymers contributed to both
hydraulically reversible and irreversible fouling. The majority of the biopolymers in the raw
EfOM were isolated in the hydrophilic fraction, which meant that this fraction had the
highest membrane fouling potential compared with the hydrophobic and transphilic
fractions.

Flux data analysis showed that for dead-end MF, fouling of the membranes by the raw
effluent was governed by pore adsorption and pore blocking during the first 20 minutes of
filtration and then followed the cake filtration model. For the hydrophobic PVDF UF (100
kDa and 50 kDa) membranes, fouling switched from pore adsorption and pore blocking to
cake filtration after the first 5 minutes of filtration. Fouling of the relatively hydrophilic UF
(PES 30 kDa) membrane could be described by the cake filtration model throughout the
course of filtration.

10.2. Effect of coagulation on membrane fouling

Coagulation with ACH and alum at 2.5 mg Al3+ L-1 significantly improved the permeate
flux and reduced hydraulically irreversible fouling in dead-end MF and UF. The MF flux
improvement was higher for alum than ACH, whereas the two coagulants had a similar flux
enhancing effect in UF. Lowering the ACH dose to 1.0 mg Al3+ L-1 reduced the extent of
fouling reduction by coagulation. Unlike MF fouling by the raw effluent, MF fouling by the
coagulated effluent could be described by the cake filtration mechanism from the beginning
of filtration. This was attributed to the coagulants destabilising the colloidal organic matter,
including some biopolymers and HS responsible for membrane fouling in the raw effluent,
and so causing the aggregation of these materials to form particles larger than the
membrane pores. The reduction in the concentrations of the foulants and the early
formation of the cake layer on the membrane surface enabled the improved flux and
reduced hydraulically irreversible fouling. The reduction in irreversible fouling was much
greater for the less hydrophobic MF membrane than the more hydrophobic counterpart.
Coagulation reduced the rate of flux decline, but did not change the governing fouling

174
mechanisms in UF. This pre-treatment also resulted in significant flux improvement in
cross-flow MF with the alumina membrane. However, a very high coagulant dose
(approximately 40 mg Al3+ L-1) was needed to reduce the true colour of the effluent to the
desired level (20–25 Pt-Co units).

10.3. Effect of ozonation and biological activated carbon (BAC) filtration on

membrane fouling
Ozonation of the raw effluent at 10 mg O3 L-1 for 10 minutes led to significant flux
improvement for both dead-end and cross-flow filtration, which was attributed to the
breakdown of a proportion of the biopolymers and HS to compounds with lower MW and
hydrophobicity. BAC filtration of the ozonated effluent removed the majority of the
particles larger than 0.5 μm and therefore led to further flux improvement in dead-end
filtration with the MF and the loose UF (PVDF 100 kDa) membranes. However, this
additional pre-treatment did not affect the flux of the tighter UF (PES 30 kDa) membrane.
The hydraulically irreversible fouling of the membranes was reduced after ozonation,
whereas BAC filtration did not affect this type of fouling. Similar to coagulation, the
reduction in hydraulically irreversible fouling by ozonation was significantly greater for the
less hydrophobic MF membrane than the more hydrophobic membrane. The unified
membrane fouling index (UMFI) was found a useful tool for quantitative evaluation of the
effectiveness of different feed pre-treatments on fouling reduction.

As ozonation of the raw effluent generated some low MW (<500 Da) compounds (which
were potentially biodegradable), this pre-treatment may accelerate membrane biofouling.
The BAC filter showed potential to reduce biofouling since the filter removed the majority
of the low MW compounds generated during ozonation.

10.4. Effects of adsorption pre-treatments on membrane fouling

Pre-treatments with the PAC and the AER to remove 75–80% of the true colour of the raw
effluent neither reduced, nor worsened membrane fouling in dead-end filtration. This was
attributed to the fact that although the AER and the PAC removed more than 50% of the
HS from the raw effluent, they did not remove biopolymers – the largest contributors to
membrane fouling. However, PAC adsorption may reduce membrane biofouling as it
removed a significant proportion of the low MW (<500 Da) compounds from the raw
effluent. Similarly, cross-flow MF and UF tests showed that adsorption pre-treatment with
the AER did not affect the permeate flux.

175
10.5. Effect of rapid membrane back-flushing on flux performance in cross-flow
filtration
Back-flushing the ceramic membranes at 600 kPa for 1 second every 12 minutes resulted in
significant flux improvement, which was attributed to the removal of the foulant layer on
the membrane surface. The opportunity to combine coagulation or ozonation pre-treatment
with rapid membrane back-flushing to maintain a high permeate flux in cross-flow filtration
with ceramic membranes was demonstrated.

10.6. Chemical cleaning of the fouled membranes

Chemical cleaning of the 50 kDa PVDF and 30 kDa PES membranes fouled by the raw
effluent was conducted. The most effective cleaning agents were NaOCl, sodium dodecyl
sulphate (SDS), and Terg-a-zyme. The high cleaning efficiencies of the cleaners containing
anionic surfactants (SDS and Terg-a-zyme) verified that hydrophobic attraction was the
major force keeping the hydraulically irreversible foulants and the membranes together.
The fact that these solutions turned yellow during the membrane cleaning experiment
confirmed that HS played a role in causing internal adsorptive fouling.

10.7. Implications of these findings

Of the pre-treatments investigated, only those which led to the removal (coagulation) or
breakdown (ozonation) of biopolymers reduced membrane fouling. Pre-treatments which
do not affect the biopolymers in the secondary effluent are unlikely to improve the flux
performance of the membranes. The results obtained in this study therefore imply that in
MF and UF of secondary effluent (whether with polymeric membranes in dead-end mode
or ceramic membranes in cross-flow mode), pre-treatment for fouling reduction should
target the removal or the breakdown of the molecular structures of biopolymers (i.e.,
proteins and polysaccharides). For low colour secondary effluent, in-line coagulation would
be the most appropriate pre-treatment for its simplicity and relatively low investment and
operating costs. For secondary effluent with high colour, ozonation would be the pre-
treatment of choice for its decolourising effect and potential fouling reduction. However,
the addition of BAC filters downstream of the ozonation process may be needed to alleviate
membrane biofouling. The BAC filters would retain the majority of the suspended solids
from the ozonated effluent; therefore further flux improvement can be achieved if the
downstream membranes are MF or loose UF membranes operated in dead-end mode. BAC

176
filtration is unlikely to have any significant impact on the permeate flux of tight UF
(MWCO  50 kDa) membranes and membranes operated in cross-flow mode.

The high effectiveness of SDS and Terg-a-zyme for removing hydraulically irreversible
foulants indicated that these chemicals are promising alternative cleaning agents,
particularly for membranes with low chlorine tolerance (e.g., those made from
polypropylene).

10.8. Recommendations for future work

As shown in this study, biopolymers and a small proportion of HS in the secondary effluent
were responsible for membrane fouling. Further analysis of these compounds is needed to
reveal their detailed chemical compositions (e.g., types and molecular structures of
polysaccharides and proteins) and their chemical reactivity. This knowledge would help to
tailor pre-treatment for fouling reduction more effectively.

This research showed that the pre-treatments which removed more than 50% of HS but did
not affect biopolymers did not reduce fouling in MF and UF of secondary effluent.
However, the abundance of HS in secondary effluent and the high reactivity of HS towards
common pre-treatment methods (including coagulation, oxidation, and adsorption with
AERs and PAC) make it difficult to target only biopolymers to improve the efficiency of
the pre-treatments. If novel pre-treatments with high selectivity for biopolymers can be
found, the effectiveness of pre-treatments for fouling reduction would increase
significantly.

According to the present study, less hydrophobic MF membranes are likely to benefit more
from feed pre-treatments than more hydrophobic counterparts, as the hydraulically
irreversible fouling of the formers tends to be less severe and more easily mitigated.
However, most of the membranes used for MF of secondary effluent are made from
hydrophobic materials (e.g., PVDF, PES, and polypropylene). Researchers have attempted
to reduce the hydrophobicity of these membranes by different methods, such as surface
coating with hydrophilic polymers (Du et al. 2009) and blending or grafting the base
materials with hydrophilic copolymers and/or nanoparticles (Zhao et al. 2007; Li et al.
2009). The resulting membranes showed higher fouling resistance compared with the non-
modified membranes (in most studies, fouling tests were done with bovine serum albumin

177
solutions). However, the use of these membranes has been limited at lab-scale level.
Therefore, it is suggested that future research should focus on testing the practicability of
these membranes in pilot-scale and full-scale MF and UF plants. It is envisaged that the use
of less hydrophobic membranes would help to achieve the greatest reduction in irreversible
fouling which can be obtained from feed pre-treatments such as coagulation or ozonation.

This study investigated the fouling of tubular ceramic membranes under particular MF and
UF operating conditions which were selected based on current practice (as well as the
capacity of the equipment). It is recommended that tests conducted under a wider range of
operating conditions, including cross-flow velocity, TMP, and back-flush pressure,
frequency and duration, are carried out in future studies to determine the optimal operating
conditions for cross-flow filtration of secondary effluent. The advantage of the rapid back-
flushing method used in this study over conventional back-flushing systems is the
elimination of the need for a back-flush pump. The rapid action of the back-flush device
also increases the time available for filtration. However, the use of this device, which was
essentially a piston driven by air pressure, in full-scale filtration plants has not been
reported. The practicability in scaling up this device would be an interesting subject for
further investigation.

To date the use of ceramic membranes for MF and UF of municipal wastewater still
remains at lab-scale and pilot-scale level. Therefore, it is not clear if these membranes can
be cost competitive alternatives for the polymeric counterparts. As the cost of ceramic
membranes has decreased considerably in the last few years (Pearce 2011), it is perhaps the
right time to re-evaluate the possibility of using these membranes for this application.

178
 REFERENCES

Abdessemed, D, Nezzal, G & Aim, RB 2000, 'Coagulation-adsorption-ultrafiltration for

wastewater treatment reuse', Desalination, vol. 131, no. 1-3, pp. 307-14.

Adham, S, Chiu, K-P, Gramith, K & Oppenheimer, J 2005, Development of a

Microfiltration and Ultrafiltration Knowledge Base, AWWA Research Foundation,
Denver, Colorado.

Adham, S, Chiu, K-P, Lehman, G, Howe, K, Marwah, A, Mysore, C, Clouet, J, Do-Quang,

Z & Cagnard, O 2006, Optimization of Membrane Treatment for Direct and Clarified
Water Filtration, AWWA Research Foundation.

Adham, SS, Snoeyink, VL, Clark, MM & Bersillon, J-L 1991, 'Predicting and verifying
organics removal by PAC in an ultrafiltration system,' Journal of American Water Works
Association, vol. 83, pp. 81-91.

Advantec MFS Inc. 2001, Membrane filters, Advantec MFS Inc., viewed 20 June 2008,
<http://www.advantecmfs.com/catalog/filt/membrane.pdf>.

Ahmed, Z, Cho, J, Lim, B-R, Song, K-G & Ahn, K-H 2007, 'Effects of sludge retention
time on membrane fouling and microbial community structure in a membrane bioreactor',
Journal of Membrane Science, vol. 287, no. 2, pp. 211-8.

Ahn, W-Y, Kalinichev, AG & Clark, MM 2008, 'Effects of background cations on the
fouling of polyethersulfone membranes by natural organic matter: Experimental and
molecular modelling study', Journal of Membrane Science, vol. 309, pp. 128-40.

Aimar, P & Howell, JA 1989, 'Effects of concentration boundary layer development on the
flux limitations in ultrafiltration', Chemical Engineering Research and Design, vol. 67, pp.
255-61.

Al-Amoudi, A & Lovitt, RW 2007, 'Fouling strategies and the cleaning system of the NF
membranes and factors affecting cleaning efficiency', Journal of Membrane Science, vol.
303, no. 1-2, pp. 4-28.

179
Al-Halbouni, D, Traber, J, Lyko, S, Wintgens, T, Melin, T, Tacke, D, Janot, A, Dott, W &
Hollender, J 2008, 'Correlation of EPS content in activated sludge at different sludge
retention times with membrane fouling phenomena', Water Research, vol. 42, no. 6-7, pp.
1457-88.

Alconox 2006, Fax Document #0413: Terg-a-zyme Technical Bulletin, Alconox, viewed 20
September 2010,
<http://www.alconox.com/Resources/StandardDocuments/TB/techbull_tergazyme.pdf>.

Alconox 2011, Material Safety Data Sheet – Terg-a-zyme, Alconox, viewed 15 January
2012,
<http://www.alconox.com/Resources/StandardDocuments/MSDS/msds_tergazyme_english
_ghs.pdf>.

Allpike, BP, Heitz, A, Joll, CA, Kagi, RI, Abbt-Braun, G, Frimmel, FH, Brinkmann, T,
Her, N & Gary, A 2005, 'Size exclusion chromatography to characterize DOC removal in
drinking water treatment', Environmental Science & Technology, vol. 39, no. 7, pp. 2334-
42.

American Public Health Association 2005, Standard Methods for the Examination of Water
& Wastewater, 21st edn, APHA, Washington.

Aoustin, E, Schäfer, AI, Fane, AG & Waite, TD 2001, 'Ultrafiltration of natural organic
matter', Separation and Purification Technology, vol. 22-23, pp. 63-78.

Arkhangelski, E, Kuzmenko, D & Gitis, V 2007, 'Impact of chemical cleaning on properties

and functioning of polyethersulfone membranes', Journal of Membrane Science, vol. 305,
no. 1-2, pp. 176-84.

Arnal, JM, Garcia-Fayos, B, Sancho, M & Verdu, G 2009, 'Ultrafiltration membrane

cleaning with different chemical solutions after treating surface water', Desalination and
Water Treatment, vol. 7, pp. 198-205.

Australian Government - Bureau of Meteorology 2008, An assessment of the impact of

climate change on the nature and frequency of exceptional climatic events, Commonwealth
of Australia, Canberra.

180
AWWA Membrane Technology Research Committee 2005, 'Committee Report: Recent
advances and research needs in membrane fouling', Journal of American Water Works
Association, vol. 97, no. 8, pp. 79-89.

Bacchin, P, Aimar, P & Field, RW 2006, 'Critical and sustainable fluxes: Theory,
experiments and applications', Journal of Membrane Science, vol. 281, no. 1-2, pp. 42-69.

Barger, M & Carnahan, RP 1991, 'Fouling prediction in reverse osmosis processes', paper
presented to The Twelfth International Symposium on Desalination and Water Re-Use,
Malta, April 15 -18.

Barker, DJ & Stuckey, DC 1999, 'A review of soluble microbial products (SMP) in
wastewater treatment systems', Water Research, vol. 33, no. 14, pp. 3062-82.

Belfort, G, Davis, RH & Zydney, AL 1994, 'The behavior of suspensions and

macromolecular solutions in crossflow microfiltration', Journal of Membrane Science, vol.
96, pp. 1-58.

Bellona, C, Drewes, JE, Xua, P & Amy, G 2004, 'Factors affecting the rejection of organic
solutes during NF/RO treatment - a literature review', Water Research, vol. 38, no. 12, pp.
2795–809.

Bjelopavlic, M, Newcombe, G & Hayes, R 1998, 'Adsorption of NOM onto activated

carbon: Effect of surface charge, ionic strength and pore volume distribution', Journal of
Colloid and Interface Science, vol. 210, pp. 271-80.

Boerlage, SFE, Kennedy, MD, Dickson, MR, El-Hodali, DEY & Schippers, JC 2002, 'The
modified fouling index using ultrafiltration membranes (MFI-UF): characterisation,
filtration mechanisms and proposed reference membrane', Journal of Membrane Science,
vol. 197, no. 1-2, pp. 1-21.

Bolto, B, Dixon, D, Eldridge, R, King, S & Linge, K 2002, 'Removal of natural organic
matter by ion exchange', Water Research, vol. 36, no. 20, pp. 5057-65.

Bottino, A, Capannelli, C, Borghi, AD, Colombino, M & Conio, O 2001, 'Water treatment
for drinking purpose: ceramic microfiltration application', Desalination, vol. 141, no. 1-3,
pp. 75-9.

181
Bowen, WR, Calvo, JI & Hernández, A 1995, 'Steps of membrane blocking in flux decline
during protein microfiltration', Journal of Membrane Science, vol. 101, no. 1-2, pp. 153-65.

Boyer, TH & Singer, PC 2005, 'Bench-scale testing of a magnetic ion exchange resin for
removal of disinfection by-product precursors', Water Research, vol. 39, no. 7, pp. 1265-76.

Boyer, TH & Singer, PC 2006, 'A pilot-scale evaluation of magnetic ion exchange
treatment for removal of natural organic material and inorganic anions', Water Research,
vol. 40, no. 15, pp. 2865-76.

Braghetta, A, DiGiano, FA & Ball, WP 1998, 'NOM accumulation at NF membrane

surface: Impact of chemistry and shear', Journal of Environmental Engineering, vol. 124,
no. 11, pp. 1087-98.

Brindle, K & Stephenson, T 1996, 'The application of membrane biological reactors for the
treatment of wastewaters', Biotechnology and Bioengineering, vol. 49, pp. 601-10.

Brookes, A, Jefferson, B, Guglielmi, G & Judd, SJ 2006, 'Sustainable flux fouling in a

membrane bioreactor: impact of flux and MLSS', Separation Science & Technology, vol.
41, pp. 1279-91.

Buls, J 2000, 'Making humic substances more biodegradable', Masters thesis, Montana
State University.

Byun, S, Taurozzi, JS, Alpatova, AL, Wang, F & Tarabara, VV 2011, 'Performance of
polymeric membranes treating ozonated surface water: Effect of ozone dosage', Separation
and Purification Technology, vol. 81, no. 3, pp. 270-8.

Cai, Z, Kim, J & Benjamin, MM 2008, 'NOM removal by adsorption and membrane
filtration using heated aluminum oxide particles', Environmental Science & Technology,
vol. 42, no. 2, pp. 619-23.

Camel, V & Bermond, A 1998, 'The use of ozone and associated oxidation processes in
drinking water treatment', Water Research, vol. 32, no. 11, pp. 3208-22.

Campinas, M & Rosa, MJ 2010, 'Removal of microcystins by PAC/UF', Separation and

Purification Technology, vol. 71, pp. 114-20.

182
Campos, C, Marinas, BJ, Snoeyink, VL, Baudin, I & Laîné, J-M 1998, 'Adsorption of trace
organic compounds in CRISTAL® processes', Desalination, vol. 117, pp. 265-71.

Caothien, S, Liu, C & O’Connell, P 2003, Reducing fouling and the cost of membrane
systems, Pall Corporation, viewed 29 April 2008,
<http://www.pall.com/pdf/wat_pdf_HEFMPaper6_30_03.pdf>.

Carroll, T, King, S, Gray, SR, Bolto, BA & Booker, NA 2000, 'The fouling of
microfiltration membranes by NOM after coagulation treatment', Water Research, vol. 34,
no. 11, pp. 2861-8.

Cataldo, F 2003, 'On the action of ozone on proteins', Polymer Degradation and Stability,
vol. 82, no. 1, pp. 105-14.

Chandrakanth, M & Amy, G 1996, 'Effects of ozone on the colloidal stability and
aggregation of particles coated with natural organic matter', Environmental Science &
Technology, vol. 30, no. 2, pp. 431-43.

Chang, I-S & Kim, S-N 2005, 'Wastewater treatment using membrane filtration - effect of
biosolids concentration on cake resistance', Process Biochemistry, vol. 40, pp. 1307-14.

Chang, YJ, Choo, KH, Benjamin, MM & Reiber, S 1998, 'Combined adsorption-UF
process increases TOC removal', Journal American Water Works Association, vol. 90, no.
5, pp. 90-102.

Chennamsetty, R, Escobar, I & Xu, X 2006, 'Characterization of commercial water

treatment membranes modified via ion beam irradiation', Desalination, vol. 188, no. 1-3,
pp. 203-12.

Cheryan, M 1986, Ultrafiltration Handbook, 1st edn, Technomic Publishing, Lancaster,

Pennsylvania.

Childress, AE & Elimelech, M 1996, 'Effect of solution chemistry on the surface charge of
polymeric reverse osmosis and nanofiltration membranes', Journal of Membrane Science,
vol. 119, no. 2, pp. 253-68.

183
Cho, J, Amy, G, Pellegrino, J & Yoon, Y 1998, 'Characterization of clean and natural
organic matter (NOM) fouled NF and UF membranes, and foulants characterization',
Desalination, vol. 118, no. 1-3, pp. 101-8.

Cho, J, Amy, G & Pellegrino, J 2000, 'Membrane filtration of natural organic matter:
factors and mechanisms affecting rejection and flux decline with charged ultrafiltration
(UF) membrane', Journal of Membrane Science, vol. 164, no. 1-2, pp. 89-110.

Cho, J, Song, KG, Yun, H, Ahn, KH, Kim, JY & Chung, TH 2005, 'Quantitative analysis of
biological effect on membrane fouling in submerged membrane bioreactor', Water Science
& Technology, vol. 51, no. 6-7, pp. 9-18.

Choi, KY & Dempsey, BA 2004, 'In-line coagulation with low-pressure membrane

filtration', Water Research, vol. 38, no. 19, pp. 4271-81.

Chow, C, Fabris, R & Drikas, M 2004, 'A rapid organic characterisation tool to optimise
water treatment processes', Journal of Water Supply: Research & Technology, vol. 53, no.
2, pp. 85-92.

Cicek, N, Franco, JP, Suida, MT, Urbain, V & Manem, J 1999, 'Characterization and
comparison of a membrane bioreactor and a conventional activatedsludge system in the
treatment of wastewater containing high-molecular weight compounds', Water Environment
Research, vol. 71, pp. 64-70.

Ciora, RJ & Liu, PKT 2003, 'Ceramic membranes for environmental related applications',
Fluid/Particle Separation Journal, vol. 15, no. 1, pp. 51-60.

Clark, MM 1998, 'Committe report: Membrane processes', Journal of American Water

Works Association, vol. 90, no. 6, pp. 91-105.

Clever, M, Jordt, F, Knauf, R, Räbiger, N, Rüdebusch, M & Hilker-Scheibel, R 2000,

'Process water production from river water by ultrafiltration and reverse osmosis',
Desalination, vol. 131, no. 1, pp. 325-36.

Combe, C, Molis, E, Lucas, P, Riley, R & Clark, M 1999, 'The effect of CA membrane
properties on adsorptive fouling by humic acid', Journal of Membrane Science, vol. 154,
no. 1, pp. 73-87.

184
Cornelissen, ER, Moreau, N, Siegers, WG, Abrahamse, AJ, Rietveld, LC, Grefte, A,
Dignum, M, Amy, G & Wessels, LP 2008, 'Selection of anionic exchange resins for
removal of natural organic matter (NOM) fractions', Water Research, vol. 42, pp. 413-23.

Crittenden, JC, Trussell, RR, Hand, DW, Howe, KJ & Tchobanoglous, G (eds) 2006, Water
Treatment Principles and Design, John Wiley and Sons, USA.

Croué, JP 1999, 'Isolation, fractionation, characterization and reactive properties of natural

organic matter', paper presented to The AWWA 18th Federal Convention, Adelaide,
Australia.

Croué, JP, Violleau, D, Bodaire, C & Legube, B 1999, 'Removal of hydrophobic and
hydrophilic constituents by anion exchange resin', Water Science & Technology, vol. 40,
no. 9, pp. 207-14.

Crozes, GF, Jacangelo, JG, Anselme, C & Laîné, JM 1997, 'Impact of ultrafiltration
operating conditions on membrane irreversible fouling', Journal of Membrane Science, vol.
124, no. 1, pp. 63-76.

Daufin, G, Escudier, J-P, Carrère, H, Bérot, S, Fillaudeau, L & Decloux, M 2001, 'Recent
and emerging applications of membrane processes in the food and dairy industry', Trans
IChemE, vol. 79, no. 2, pp. 89-102.

Decarolis, J, Hong, S & Taylor, J 2001, 'Fouling behavior of a pilot scale inside-out hollow
fiber UF membrane during dead-end filtration of tertiary wastewater', Journal of Membrane
Science, vol. 191, no. 1-2, pp. 165-78.

Defrance, L & Jaffrin, MY 1999, 'Reversibility of fouling formed in activated sludge

filtration', Journal of Membrane Science, vol. 157, no. 1-3, pp. 73-84.

Dialynas, E & Diamadopoulos, E 2008, 'Integration of immersed membrane ultrafiltration

with coagulation and activated carbon adsorption for advanced treatment of municipal
wastewater', Desalination, vol. 230, no. 1-3, pp. 113-27.

Djafer, L, Ayral, A & Ouagued, A 2010, 'Robust synthesis and performance of a titania-
based ultrafiltration membrane with photocatalytic properties', Separation and Purification
Technology, vol. 75, no. 2, pp. 198-203.

185
Dobrak, A, Verrecht, B, Van den Dungen, H, Buekenhoudt, A, Vankelecom, IFJ & Van der
Bruggen, B 2010, 'Solvent flux behavior and rejection characteristics of hydrophilic and
hydrophobic mesoporous and microporous TiO2 and ZrO2 membranes', Journal of
Membrane Science, vol. 346, no. 1-2, pp. 344-52.

Drewes, JE & Fox, P 1999, 'Fate of natural organic matter (NOM) during groundwater
recharge using reclaimed water', Water Science and Technology vol. 40, no. 9, pp. 241-8.

Du, JR, Peldszus, S, Huck, PM & Feng, X 2009, 'Modification of poly(vinylidene fluoride)
ultrafiltration membranes with poly(vinyl alcohol) for fouling control in drinking water
treatment', Water Research, vol. 43, no. 18, pp. 4559-68.

Duan, J & Gregory, J 2003, 'Coagulation by hydrolysing metal salts. ' Advances in Colloid
and Interface Science, vol. 101-102, pp. 475-502.

Duclos-Orsello, C, Li, W & Ho, C-C 2006, 'A three mechanism model to describe fouling
of microfiltration membranes', Journal of Membrane Science, vol. 280, no. 1-2, pp. 856-66.

Elimelech, M, Chen, WH & Waypa, JJ 1994, 'Measuring the zeta (electrokinetic) potential
of reverse osmosis membranes by a streaming potential analyzer', Desalination, vol. 95, pp.
269-86.

EPA Victoria 2002, Guideline for Environmental Management – Disinfection of Treated

Wastewater, EPA Victoria, viewed 15 February 2012,
<http://epanote2.epa.vic.gov.au/EPA/Publications.nsf/PubDocsLU/730?OpenDocument>.

EPA Victoria 2003, Guidelines for Environmental Management – Use of Reclaimed Water,
EPA Victoria, viewed 11 March 2008,
<http://epanote2.epa.vic.gov.au/EPA/Publications.nsf/d85500a0d7f5f07b4a2565d1002268f
3/64c2a15969d75e184a2569a00025de63/$FILE/464.2.pdf>.

Evenblij, H, Geilvoet, S, van der Graaf, J & van der Roest, HF 2005, 'Filtration
characterisation for assessing MBR performance: three cases compared', Desalination, vol.
178, pp. 115-24.

186
Fan, L, Harris, JL, Roddick, FA & Booker, NA 2001, 'Influence of the characteristics of
natural organic matter on the fouling of microfiltration membranes', Water Research, vol.
35, no. 18, pp. 4455-63.

Fan, L 2002, 'Interactions of Natural Organic Matter and Microfiltration Membranes in

Potable Water Treatment', PhD thesis, RMIT University.

Fawehinmi, F, Lens, P, Stephenson, T, Rogalla, F & Jefferson, B 2004, 'The influence of

operating conditions on EPS, SMP and bio-fouling in anaerobic MBR', paper presented to
Water Environment-Membrane Technology Conference, Seoul, Korea.

Field, RW, Wu, D, Howell, JA & Gupta, BB 1995, 'Critical flux concept for microfiltration
fouling', Journal of Membrane Science, vol. 100, no. 1-2, pp. 259-72.

Field, RW & Wu, JJ 2011, 'Modelling of permeability loss in membrane filtration: Re-
examination of fundamental fouling equations and their link to critical flux', Desalination,
vol. 283, pp. 68-74.

Finley, J 2005, 'Ceramic membranes: a robust filtration alternative', Filtration &

Separation, vol. 42, no. 9, pp. 34-7.

Fu, LF & Dempsey, BA 1998, 'Modelling the effect of particle size and charge on the
structure of the filter cake in ultrafiltration', Journal of Membrane Science, vol. 149, no. 2,
pp. 221-40.

Gabrus, E & Szaniawska, D 2009, 'Application of backflushing for fouling reduction during
microfiltration of yeast suspensions', Desalination, vol. 240, no. 1-3, pp. 46-53.

Gebbie, P 2001, 'Using polyaluminium coagulants in water treatment', paper presented to

The 64th Annual Water Industry Engineers and Operators’ Conference, Bendigo, Victoria,
Australia, September 5-6.

Gecol, H 2006, 'Chapter 2: The basic theory', in R Farn (ed.), Chemistry and Technology of
Surfactants, Blackwell Publishing, USA, pp. 24-45.

Gésan-Guiziou, G, Wakeman, RJ & Daufin, G 2002, 'Stability of latex crossflow filtration:

cake properties and critical conditions of deposition', Chemical Engineering Journal, vol.
85, pp. 27-34.

187
Ghosh, K & Schnitzer, M 1980, 'Macromolecular structures of humic substances', Soil
Science, vol. 129, no. 5, pp. 266-76.

Gray, SR, Ritchie, CB & Bolto, BA 2004, 'Effect of fractionated NOM on low-pressure
membrane flux declines', Water Supply, vol. 4, no. 4, pp. 189-96.

Gray, SR, Ritchie, CB, Tran, T, Bolto, BA, Greenwood, P, Busetti, F & Allpike, B 2008,
'Effect of membrane character and solution chemistry on microfiltration performance',
Water Research, vol. 42, no. 3, pp. 743-53.

Gregor, JE, Nokes, CJ & Fenton, E 1997, 'Optimising natural organic matter removal from
low turbidity waters by controlled pH adjustment of aluminium coagulation', Water
Research, vol. 31, no. 12, pp. 2949-58.

Guo, WS, Vigneswaran, S, Ngo, HH & Chapman, H 2004, 'Experimental investigation of

adsorption-flocculation-microfiltration hybrid system in wastewater reuse', Journal of
Membrane Science, vol. 242, no. 1-2, pp. 27-35.

Haberkamp, J, Ruhl, AS, Ernst, M & Jekel, M 2007, 'Impact of coagulation and adsorption
on DOC fractions of secondary effluent and resulting fouling behaviour in ultrafiltration',
Water Research, vol. 41, no. 17, pp. 3794-802.

Haberkamp, J, Ernst, M, Böckelmann, U, Szewzyk, U & Jekel, M 2008, 'Complexity of

ultrafiltration membrane fouling caused by macromolecular dissolved organic compounds
in secondary effluents', Water Research, vol. 42, no. 12, pp. 3153-61.

Hach Company 2003, DR/4000 spectrophotometer procedures manual, Hach Company.

Hammer, MJ & Hammer, MJJr 2008, Water and Wastewater Technology, 6th edn, Pearson
Education Inc., New Jersey.

Heijman, SGJ & Bakker, S 2007, Ceramic microfiltration as the first treatment step in
surface water treatment (report to the European Commission), TECHNEAU, Europe.

Henderson, RK, Subhi, N, Antony, A, Khan, SJ, Murphy, KR, Leslie, GL, Chen, V, Stuetz,
RM & Le-Clech, P 2011, 'Evaluation of effluent organic matter fouling in ultrafiltration
treatment using advanced organic characterisation techniques', Journal of Membrane
Science, vol. 382, no. 1-2, pp. 20-59.

188
Her, N, Amy, G, Foss, D & Cho, J 2002, 'Variations of molecular weight estimation by HP-
size exclusion chromatography with UVA versus online DOC detection', Environmental
Science & Technology, vol. 36, no. 15, pp. 3393-9.

Her, N, Amy, G, Chung, J, Yoon, J & Yoon, Y 2008, 'Characterizing dissolved organic
matter and evaluating associated nanofiltration membrane fouling', Chemosphere, vol. 70,
no. 3, pp. 495-502.

Hollender, J, Zimmermann, SG, Koepke, S, Krauss, M, McArdell, CS, Ort, C, Singer, H,

Von Gunten, U & Siegrist, H 2009, 'Elimination of organic micropollutants in a municipal
wastewater treatment plant upgraded with a full-scale post-ozonation followed by sand
filtration', Environmental Science & Technology, vol. 43, no. 20, pp. 7862-9.

Hong, S & Elimelech, M 1997, 'Chemical and physical aspects of natural organic matter
(NOM) fouling of nanofiltration membranes', Journal of Membrane Science, vol. 132, no.
2, pp. 159-81.

Hong, S, Aryal, R, Vigneswaran, S, Johir, MAH & Kandasamy, J 2012, 'Influence of

hydraulic retention time on the nature of foulant organics in a high rate membrane
bioreactor', Desalination, vol. 287, pp. 116-22.

Hong, SP, Bae, TH, Tak, TM, Hong, S & Randall, A 2002, 'Fouling control in activated
sludge submerged hollow fiber membrane bioreactors', Desalination, vol. 143, no. 1-3, pp.
219-28.

Hongve, D, Baann, J, Becher, G & Beckmann, OA 1999, 'Experiences from operation and
regeneration of an anionic exchanger for NOM removal', Water Science & Technology, vol.
40, no. 9, pp. 215-21.

Howe, KJ, Ishida, KP & Clark, MM 2002, 'Use of ATR/FTIR spectrometry to study fouling
of microfiltration membranes by natural waters', Desalination, vol. 147, no. 1-3, pp. 251-5.

Howe, KJ & Clark, M 2006, 'Effect of coagulation pretreatment on membrane filtration

performance', Journal of American Water Works Association, vol. 98, no. 4, pp. 133-46.

189
Howe, KJ, Marwah, A, Chiu, K-P & Adham, SS 2006, 'Effect of coagulation on the size of
MF and UF membrane foulants', Environmental Science & Technology, vol. 40, no. 24, pp.
7908-13.

Howe, KJ, Marwah, A, Chiu, K-P & Adham, SS 2007, 'Effect of membrane configuration
on bench-scale MF and UF fouling experiments', Water Research, vol. 41, no. 17, pp.
3842-9.

Howell, JA, Arnot, TC, Chua, HC, Godino, P, Hatziantoniou, D & Metsämuuronen, S
2002, 'Controlled flux behaviour of membrane processes', Macromolecular Symposia, vol.
188, no. 1, pp. 23-35.

Huang, H, Young, T & Jacangelo, J 2008, 'Unified membrane fouling index for low
pressure membrane filtration of natural waters: Principles and methodology',
Environmental Science & Technology, vol. 42, no. 3, pp. 714-20.

Huang, H, Schwab, K & Jacangelo, J 2009a, 'Pretreatment for low pressure membranes in
water treatment: a review', Environmental Science & Technology, vol. 43, no. 9, pp. 3011-
9.

Huang, H, Young, T & Jacangelo, J 2009b, 'Novel approach for the analysis of bench-scale,
low pressure membrane fouling in water treatment', Journal of Membrane Science, vol.
334, no. 1-2, pp. 1-8.

Huber, SA, Balz, A, Abert, M & Pronk, W 2011, 'Characterisation of aquatic humic and
non-humic matter with size-exclusion chromatography – organic carbon detection – organic
nitrogen detection (LC-OCD-OND)', Water Research, vol. 45, no. 2, pp. 879-85.

Humbert, H, Gallard, H, Jacquemet, V & Croué, J-P 2007, 'Combination of coagulation and
ion exchange for the reduction of UF fouling properties of a high DOC content surface
water', Water Research, vol. 41, no. 17, pp. 3803-11.

Iritani, E, Katagiri, N, Sengoku, T, Yoo, KM, Kawasaki, K & Matsuda, A 2007, 'Flux
decline behaviors in dead-end microfiltration of activated sludge and its supernatant',
Journal of Membrane Science, vol. 300, pp. 36-44.

190
Jarusutthirak, C, Amy, G & Croué, J-P 2002, 'Fouling characteristics of wastewater effluent
organic matter (EfOM) isolates on NF and UF membranes', Desalination, vol. 145, no. 1-3,
pp. 247-55.

Jarusutthirak, C & Amy, G 2006, 'Role of soluble microbial products (SMP) in membrane
fouling and flux decline', Environmental Science & Technology, vol. 40, no. 3, pp. 969-74.

Jermann, D, Pronk, W, Meylan, S & Boller, M 2007, ' Interplay of different NOM fouling
mechanisms during ultrafiltration for drinking water production', Water Research, vol. 41,
no. 8, pp. 1713-22.

Ji, L & Zhou, J 2006, ' Influence of aeration on microbial polymers and membrane fouling
in submerged membrane bioreactors', Journal of Membrane Science, vol. 276, no. 1-2, pp.
168-77.

Jiang, JQ & Graham, NJD 1996, 'Enhanced coagulation using Al/Fe(III) coagulants: effect
of coagulant chemistry on the removal of colour-causing NOM', Environmental
Technology, vol. 17, no. 9, pp. 937-50.

Jiang, T, Kennedy, MD, van der Meer, WGJ, Vanrolleghem, PA & Schippers, JC 2003,
'The role of blocking and cake filtration in MIBR fouling', Desalination, vol. 157, no. 1-3,
pp. 335-43.

Judd, SJ & Hillis, P 2001, 'Optimisation of combined coagulation and microfiltration for
water treatment ', Water Research, vol. 35, no. 12, pp. 2895-904.

Kaiya, Y, Itoh, Y, Fujita, K & S. Takizawa 1996, 'Study on fouling materials in the
membrane treatment process for potable water.' Desalination, vol. 106, no. 1-3, pp. 71-7.

Karanfil, T, Erdogan, I & Schlautman, MA 2003, 'Selecting filter membranes for measuring
DOC and UV254', Journal of American Water Works Association, vol. 95, no. 3, pp. 86-100.

Khirani, S, Aim, RB & Manero, M-H 2006a, 'Improving the measurement of the Modified
Fouling Index using nanofiltration membranes (NF–MFI)', Desalination, vol. 191, no. 1-3,
pp. 1-7.

191
Khirani, S, Smith, PJ, Manéro, M-H, Aim, RB & Vigneswaran, S 2006b, 'Effect of periodic
backwash in the submerged membrane adsorption hybrid system (SMAHS) for wastewater
treatment', Desalination, vol. 191, no. 1-3, pp. 27-34.

Kilduff, JE, Karanfil, T, Chin, YP & Weber W. Jr 1996, 'Adsorption of natural organic
polyelectrolytes by activated carbon: a size exclusion study', Environmental Science &
Technology, vol. 30, no. 4, pp. 1336-46.

Kim, H-S, Takizawa, S & Ohgaki, S 2007, 'Application of microfiltration systems coupled
with powdered activated carbon to river water treatment', Desalination, vol. 202, pp. 271-7.

Kim, HC & Dempsey, BA 2010, 'Removal of organic acids from EfOM using anion
exchange resins and consequent reduction of fouling in UF and MF', Journal of Membrane
Science, vol. 364, no. 1-2, pp. 325-30.

Kim, J, Cai, Z & Benjamin, MM 2008a, 'Effects of adsorbents on membrane fouling by

natural organic matter', Journal of Membrane Science, vol. 310, no. 1-2, pp. 356-64.

Kim, J, Deng, Q & Benjamin, MM 2008b, 'Simultaneous removal of phosphorus and

foulants in a hybrid coagulation/membrane filtration system', Water Research, vol. 42, no.
8-9, pp. 2017-24.

Kim, S-H, Moon, S-Y, Yoon, C-H, Yim, S-K & Cho, J-W 2005, 'Role of coagulation in
membrane filtration of wastewater for reuse', Desalination, vol. 173, no. 3, pp. 301-7.

Koh, LC, Ahn, WY & Clark, MM 2006, 'Selective adsorption of natural organic foulants by
polysulfone colloids: Effect on ultrafiltration fouling', Journal of Membrane Science, vol.
281, no. 1-2, pp. 472-9.

Kolarik, LO & Booker, N 1995, 'Fundamentals of Coagulation-Flocculation Processes in

Water Treatment', in LO Kolarik & AJ Priestley (eds), Modern Techniques in Water and
Wastewater Treatment, 1st edn, CSIRO Publishing, Melbourne.

Konieczny, K & Klomfas, G 2002, 'Using activated carbon to improve natural water
treatment by porous membranes', Desalination, vol. 147, no. 1-3, pp. 109-16.

Koonaphapdeelert, S & Li, K 2007, 'Preparation and characterization of hydrophobic

ceramic hollow fibre membrane', Journal of Membrane Science, vol. 291, no. 1-2, pp. 70-6.

192
Kuberkar, VT & Davis, RH 2001, 'Microfiltration of protein-cell mixtures with
crossflushing or backflushing', Journal of Membrane Science, vol. 183, no. 1-2, pp. 1-14.

Laabs, C, Amy, G & Jekel, M 2006, 'Understanding the size and character of fouling-
causing substances from effluent organic matter (EfOM) in low-pressure filtration',
Environmental Science & Technology, vol. 40, no. 14, pp. 4495-9.

Lambert, SD & Graham, NJD 1995, 'Removal of non-specific dissolved organic matter
from upland potable water supplies - II. Ozonation and adsorption', Water Research, vol.
29, no. 10, pp. 2427-33.

Le-Clech, P, Jefferson, B & Judd, SJ 2003, 'Impact of aeration, solids concentration and
membrane characteristics on the hydraulic performance of a membrane bioreactor', Journal
of Membrane Science, vol. 218, no. 1-2, pp. 117-29.

Le-Clech, P, Chen, V & Fane, AG 2006, 'Review: Fouling in membrane bioreactors used in
wastewater treatment', Journal of Membrane Science, vol. 284, no. 1-2, pp. 17-53.

Lee, CW, Bae, SD, Han, SW & Kang, LS 2007, 'Application of ultrafiltration hybrid
membrane processes for reuse of secondary effluent', Desalination, vol. 202, no. 1-3, pp.
239-46.

Lee, N, Amy, G, Croue, J-P & Buisson, H 2004, 'Identification and understanding of
fouling in low-pressure membrane (MF/UF) filtration by natural organic matter (NOM)',
Water Research, vol. 38, no. 20, pp. 4511-23.

Lee, S, Lee, K, Wan, WM & Choi, Y 2005, 'Comparison of membrane permeability and a
fouling mechanism by pre-ozonation followed by membrane filtration and residual ozone in
membrane cells', Desalination, vol. 178, no. 1-3, pp. 287-94.

Lehman, SG & Liu, L 2009, 'Application of ceramic membranes with pre-ozonation for
treatment of secondary wastewater effluent', Water Research, vol. 43, no. 2, pp. 2020-8.

Lesjean, B, Rosenberger, S, Laabs, C, Jekel, M, Gnirss, R & Amy, G 2005, 'Correlation

between membrane fouling and soluble/colloidal organic substances in membrane
bioreactors for municipal wastewater treatment', Water Science & Technology, vol. 51, pp.
1-8.

193
Leslie, GL, Mills, WR, Dawes, TM, Kennedy, JC, McIntyre, DF & Anderson, BP 1999,
'Meeting the demand for potable water in Orange County in the 21st century: The role of
membrane processes', paper presented to American Water Works Association Membrane
Technology Conference, Long Beach, California, February 28-March 3.

Li, J-H, Xu, Y-Y, Zhu, L-P, Wang, J-H & Du, C-H 2009, 'Fabrication and characterization
of a novel TiO2 nanoparticle self-assembly membrane with improved fouling resistance',
Journal of Membrane Science, vol. 326, no. 2, pp. 659-66.

Li, Q & Elimelech, M 2004, 'Organic fouling and chemical cleaning of nanofiltration
membranes: measurements and mechanisms', Environmental Science & Technology, vol.
38, no. 17, pp. 4683-93.

Liang, L & Singer, PC 2003, 'Factors influencing the formation and relative distribution of
haloacetic acids and trihalomethanes in drinking water', Environmental Science &
Technology, vol. 37, no. 13, pp. 2920-8.

Lim, AL & Bai, R 2003, 'Membrane fouling and cleaning in microfiltration of activated
sludge wastewater', Journal of Membrane Science, vol. 216, no. 1-2, pp. 279-90.

Lin, C-F, Huang, Y-J & Hao, OJ 1999, 'Ultrafiltration processes for removing humic
substances: effect of molecular weight fractions and PAC treatment', Water Research, vol.
33, pp. 1252-64.

Lin, C-F, Liu, S-H & Hao, OJ 2001, 'Effects of functional groups of humic substances on
UF performance', Water Research, vol. 35, no. 10, pp. 2395-402.

Liu, C, Caothien, S, Hayes, J, Caothuy, T, Otoyo, T & Ogawa, T 2001, 'Membrane

chemical cleaning: from art to science', paper presented to The AWWA Water Quality
Technology Conference, San Antonio, Texas, USA, March 4-7.

Lozier, J, Cappucci, L, Amy, G, Lee, N, Jacangelo, J, Huang, H, Young, T, Mysore, C,

Emeraux, C, Clouet, J, Croue, J-P & Heijmann, B 2008, Natural Organic Matter Fouling of
Low-Pressure Membrane Systems, AWWA Research Foundation.

Lyko, S, Al-Halbouni, D, Wintgens, T, Janot, A, Hollender, J, Dott, W & Melin, T 2007,

'Polymeric compounds in activated sludge supernatant – characterisation and retention

194
mechanisms at a full-scale municipal membrane bioreactor', Water Research, vol. 41, no.
17, pp. 3894-902.

Ma, X, Su, Y, Sun, Q, Wang, Y & Jiang, Z 2007, 'Preparation of protein-adsorption-

resistant polyethersulfone ultrafiltration membranes through surface segregation of
amphiphilic comb copolymer.' Journal of Membrane Science, vol. 292, no. 1-2, pp. 116-24.

Maartens, A, Swart, P & Jacobs, EP 1999, 'Feed-water pretreatment: methods to reduce

membrane fouling by natural organic matter', Journal of Membrane Science, vol. 163, no.
1, pp. 51-62.

MacCarthy, P & Suffet, IH 1989, 'Aquatic Humic Substances and Their Influence on the
Fate and Treatment of Pollutants', in IH Suffet & P MacCarthy (eds), Aquatic Humic
Substances: Influence on Fate and Treatment of Pollutants, American Chemical Society,
Washington, pp. xvii-xxx.

Mallevialle, J, Anselme, C & Marsigny, O 1989, 'Chapter 41: Effects of Humic Substances
on Membrane Processes', in IH Suffet & P MacCarthy (eds), Aquatic Humic Substances:
Influence on Fate and Treatment of Pollutants, 1st edn, American Chemical Society,
Washington, pp. 749-67.

Marhaba, TF, Doanh, V & Lippincott, RL 2000, 'Changes in NOM fraction through
treatment: a comparison of ozonation and chlorination', Ozone Science & Engineering, vol.
22, no. 3, pp. 249-66.

Matsumoto, K, Katsuyama, S & Ohya, H 1988, 'Crossflow filtration of yeast by

microporous ceramic membrane with backwashing', Journal of Fermentation Technology,
vol. 66, pp. 199-205.

Matthiasson, E & Sivik, B 1980, 'Concentration polarization and fouling', Desalination,

vol. 35, pp. 59-103.

Meng, F, Chae, S-R, Drews, A, Kraume, M, Shin, H-S & Yang, F 2009, 'Review - Recent
advances in membrane bioreactors (MBRs): Membrane fouling and membrane material',
Water Research, vol. 43, no. 6, pp. 1489-512.

195
Miettinen, IT, Vartiainen, T, Nissinen, T, Tuhkanen, T & Martikainen, PJ 1998, 'Microbial
growth in drinking waters treated with ozone, ozone/hydrogen peroxide or chlorine',
Ozone: Science & Engineering, vol. 20, no. 4, pp. 303-15.

Millipore 2008, Durapore Membrane Filter, Millipore, viewed 20 July 2008,

<http://www.millipore.com/catalogue/item/VVLP04700>.

Mo, L & Huang, X 2003, 'Fouling characteristics and cleaning strategies in a coagulation-
microfiltration combination process for water purification', Desalination, vol. 159, no. 1,
pp. 1-9.

Mozia, S, Tomaszewska, M & Morawski, AW 2006, 'Application of an ozonation -

adsorption - ultrafiltration system for surface water treatment', Desalination, vol. 190, no.
1-3, pp. 308-14.

Mueller, U & Witte, M 2007, Ceramic membranes – Case related protocol for optimal
operational conditions to treat filter backwash water (report to the European Commission),
TECHNEAU.

Mueller, U, Schaefer, R, Witte, M, Boeckle, K, Boesl, M, Eberhagen, I & Hambsch, B

2008, Interim report: Removal of particulate matter by ceramic membranes during surface
water treatment (report to the European Commission), TECHNEAU.

Mueller, U & Witte, M 2008, Ceramic membrane applications for filter backwash water
treatment (report to the European Commission), TECHNEAU, Germany.

Mueller, U, Schaefer, R, Zawadsky, C, Boeckle, K, Boesl, M, Eberhagen, I & Hambsch, B

2009, Removal of phages and nanoparticles by ceramic membranes, TECHNEAU.

Mulder, M 1996, Basic Principles of Membrane Technology, 2nd edn, Kluwer Academic
Publishers, Dordrecht, The Netherlands.

Nakatsuka, S, Nakate, L & Miyano, T 1996, 'Drinking water treatment by using

ultrafiltration hollow fibre membranes', Desalination, vol. 106, pp. 55-61.

Newcombe, G & Drikas, M 1997, 'Adsorption of NOM onto activated carbon: electrostatic
and nonelectrostatic effects', Carbon, vol. 35, no. 9, pp. 1239-50.

196
Newcombe, G, Hepplewhite, C, Bjelopavlic, M & David, C 2008, Research Report 16 -
Optimisation of Adsorption Processes, Water Quality Research Australia, Adelaide,
Australia.

Nguyen, T, Fan, L, Roddick, FA & Harris, JL 2009, 'A comparative study of microfiltration
and ultrafiltration of activated sludge-lagoon effluent', Desalination, vol. 236, no. 1-3, pp.
208-15.

Nilson, J & DiGiano, F 1996, 'Influence of NOM composition on nanofiltration', Journal of

American Water Works Association, vol. 88, no. 5, pp. 53-66.

Nipkow, A, Zeikus, JG & Gerhardt, P 1989, 'Microfiltration cell-recycle pilot system for
continuous thermoanaerobic production of exo-β-amylase', Biotechnology and
Bioengineering, vol. 34, no. 8, pp. 1075-84.

Nyström, M & Järvinen, P 1991, 'Modification of polysulfone ultrafiltration membranes

with UV irradiation and hydrophilicity increasing agents', Journal of Membrane Science,
vol. 60, no. 1-3, pp. 275-96.

Nyström, M & Zhu, H 1997, 'Characterisation of cleaning results using combined flux and
streaming potential methods', Journal of Membrane Science, vol. 131, no. 1-2, pp. 195-205.

Ødegaard, H, Eikebrokk, B & Storhaug, R 1999, 'Processes for the removal of humic
substances from water - an overview based on Norwegian experiences', Water Science &
Technology, vol. 40, no. 9, pp. 37-46.

Oh, SJ, Kim, N & Lee, YT 2010, 'Preparation and characterization of PVDF/TiO2 organic-
inorganic composite membranes for fouling resistance improvement', Journal of Membrane
Science, vol. 345, no. 1-2, pp. 13-20.

Oldani, M & Schock, G 1989, 'Characterization of ultrafiltration membranes by infrared

spectroscopy, ESCA, and contact angle measurements', Journal of Membrane Science, vol.
43, pp. 243-58.

Oneby, MA, Bromley, CO, Borchardt, JH & Harrison, DS 2010, 'Ozone Treatment of
Secondary Effluent at U.S. Municipal Wastewater Treatment Plants', Ozone: Science &
Engineering, vol. 32, no. 1, pp. 43-55.

197
Owen, G, Bandi, M, Howell, JA & Churchouse, SJ 1995, 'Economic assessment of
membrane processes for water and waste water treatment', Journal of Membrane Science,
vol. 102, no. 1-2, pp. 77-91.

Paraskeva, P, Lambert, SD & Graham, NJD 1998, 'Influence of ozonation conditions on the
treatability of secondary effluents', Ozone: Science & Engineering, vol. 20, no. 2, pp. 133-
50.

Parnham, CS & Davis, RH 1996, 'Protein recovery from bacterial cell debris using
crossflow microfiltration with backpulsing', Journal of Membrane Science, vol. 118, no. 1-
2, pp. 259-68.

Paul, JH & Jeffrey, WH 1984, 'The effect of surfactants on the attachment of estuarine and
marine bacteria on surface', Canadian Journal of Microbiology, vol. 31, pp. 224-8.

Pearce, GK 2011, Ceramic membranes for water treatment: Technical status and economic
updates, O2 Environmental Inc., viewed 15 October 2011, <http://www.o2env.com/wp-
content/uploads/2011/02/Ceramic-Membranes-for-Water-Treatment.pdf>.

Peldszus, S, Hallé, C, Peiris, RH, Hamouda, M, Jin, X, Legge, RL, Budman, H, Moresoli,
C & Huck, PM 2011, 'Reversible and irreversible low-pressure membrane foulants in
drinking water treatment: Identification by principal component analysis of fluorescence
EEM and mitigation by biofiltration pre-treatment', Water Research, vol. 45, no. 16, pp.
5161-70.

Porcelli, N & Judd, S 2010, 'Chemical cleaning of potable water membranes: A review',
Separation and Purification Technology, vol. 71, no. 2, pp. 137-43.

Puspitasari, V, Granville, A, Le-Clech, P & Chen, V 2010, 'Cleaning and ageing effect of
sodium hypochlorite on polyvinylidene fluoride (PVDF) membrane', Separation and
Purification Technology, vol. 72, no. 3, pp. 301-8.

Qin, J-J, Oo, MH, Lee, H & Kolkman, R 2004, 'Dead-end ultrafiltration for pretreatment of
RO in reclamation of municipal wastewater effluent', Journal of Membrane Science, vol.
243, no. 1-2, pp. 107-13.

198
Reckhow, DA, Singer, PC & Malcolm, RL 1990, 'Chlorination of humic materials:
byproduct formation and chemical interpretations', Environmental Science & Technology,
vol. 24, no. 11, pp. 1655-64.

Redkar, S & Davis, RH 1995, 'Cross-flow microfiltration with high-frequency reverse

filtration', AIChE Journal, vol. 41, no. 3, pp. 501-8.

Redkar, S, Kuberkar, VT & Davis, RH 1996, 'Modeling of concentration polarization and

depolarization with high-frequency backpulsing', Journal of Membrane Science, vol. 121,
no. 1-2, pp. 229-42.

Redón, R, Vázquez-Olmos, A, Mata-Zamora, ME, Ordóñez-Medrano, A, Rivera-Torres, F

& Saniger, JM 2005, 'Contact angle studies on anodic porous alumina', Journal of Colloid
and Interface Science, vol. 287, no. 2, pp. 664-70.

Ridgway, HF, Rigby, MG & Argo, DG 1985, 'Bacterial adhesion and fouling of reverse
osmosis membranes', Journal of American Water Works Association, vol. 77, pp. 97-106.

Roddick, FA, Nguyen, T, Fan, L & Harris, JL 2007, 'Gaining an understanding of the
fouling of microfiltration and ultrafiltration membranes', paper presented to Victoria
Branch AWA Regional Conference, Traralgon, November 2007.

Rosenberg, M & Doyle, RJ 1990, 'Microbial Cell Hydrophobicity: History, Measurement,

and Significance', in RJ Doyle & M Rosenberg (eds), Microbial Cell Surface
Hydrophobicity, ASM, Washington D.C., pp. 1-38.

Rosenberger, S & Kraume, M 2003, 'Filterability of activated sludge in membrane

bioreactors', Desalination, vol. 151, no. 2, pp. 195-200.

Rosenberger, S, Evenblij, H, te Poele, S, Wintgens, T & Laabs, C 2005, 'The importance of

liquid phase analyses to understand fouling in membrane assisted activated sludge
processes – six case studies of different European research groups', Journal of Membrane
Science, vol. 263, no. 1-2, pp. 113-26.

Rosenberger, S, Laabs, C, Lesjean, B, Gnirss, R, Amy, G, Jekel, M & Schrotter, JC 2006,

'Impact of colloidal and soluble organic material on membrane performance in membrane
bioreactors for municipal wastewater treatment', Water Research, vol. 40, no. 4, pp. 710-20.

199
Rowe, D & Abdel-Magid, IM 1995, Handbook of Wastewater Reclamation and Reuse, 1st
edn, CRC Press, New York.

Saravia, F, Naab, P & Frimmel, FH 2006, 'Influence of particle size and particle size
distribution on membrane - adsorption hybrid systems', Desalination, vol. 200, pp. 446-8.

Saravia, F & Frimmel, FH 2008, 'Role of NOM in the performance of adsorption -

membrane hybrid systems applied for the removal of pharmaceuticals', Desalination, vol.
224, pp. 168-71.

Schäfer, AI 1999, 'Natural Organics Removal Using Membranes', PhD thesis, The
University of New South Wales.

Schäfer, AI, Schwicker, U, Fischer, MM, Fane, AG & Waite, TD 2000, 'Microfiltration of
colloids and natural organic matter', Journal of Membrane Science, vol. 171, no. 2, pp. 151-
72.

Schippers, JC & Verdouw, J 1980, 'The modified fouling index, a method of determining
the fouling characteristics of water', Desalination, vol. 32, pp. 137-48.

Schippers, JC, Hanemaayer, JH, Smolders, CA & Kostense, A 1981, 'Predicting flux
decline of reverse osmosis membranes', Desalination, vol. 38, pp. 339-48.

Schlichter, B, Mavrov, V & Chmiel, H 2004, 'Study of a hybrid process combining

ozonation and microfiltration/ultrafiltration for drinking water production from surface
water', Desalination, vol. 168, no. 1-3, pp. 307-17.

Seidel, A & Elimelech, M 2002, 'Coupling between chemical and physical interactions in
natural organic matter (NOM) fouling of nanofiltration membranes: implications for
fouling control', Journal of Membrane Science, vol. 203, pp. 245-55.

Sharp, MM & Escobar, IC 2006, 'Effects of dynamic or secondary-layer coagulation on

ultrafiltration', Desalination, vol. 188, no. 1-3, pp. 239-49.

Sheikholeslami, R 1999, 'Fouling mitigation in membrane processes', Desalination, vol.

123, no. 1, pp. 45-53.

200
Shon, HK, Vigneswaran, S, Kim, IS, Cho, J & Ngo, HH 2004, 'Effect of pretreatment on
the fouling of membranes: application in biologically treated sewage effluent', Journal of
Membrane Science, vol. 234, no. 1, pp. 111-20.

Shon, HK, Vigneswaran, S, Kim, IS, Cho, J & Ngo, HH 2006, 'Fouling of ultrafiltration
membrane by effluent organic matter: A detailed characterization using different organic
fractions in wastewater', Journal of Membrane Science, vol. 278, pp. 232-8.

Singer, PC 1999, 'Humic substances as precursors for potentially harmful disinfection by-
products', Water Science & Technology, vol. 40, no. 9, pp. 25-30.

Soffer, Y, Gilron, J & Adin, A 2002, 'Streaming potential and SEM-EDX study of UF
membranes fouled by colloidal iron', Desalination, vol. 146, pp. 115-21.

Sondhi, R, Lin, YS, Zhu, W & Alvarez, F 2000, 'Cross-flow filtration of synthetic
electroplating wastewater by ceramic membranes using high frequency backpulsing',
Environmental Technology, vol. 21, no. 6, pp. 699-712.

Sondhi, R & Bhave, R 2001, 'Role of backpulsing in fouling minimization in crossflow

filtration with ceramic membranes', Journal of Membrane Science, vol. 186, no. 1-2, pp.
41-52.

Song, K-Y, Park, P-K, Kim, J-H, Lee, C-H & Lee, S 2009, 'Coupling effect of 17[beta]-
estradiol and natural organic matter on the performance of a PAC adsorption/membrane
filtration hybrid system', Desalination, vol. 273, no. 1-3, pp. 392-9.

Srijaroonrat, P, Julien, E & Aurelle, Y 1999, 'Unstable secondary oil/water emulsion

treatment using ultrafiltration: fouling control by backflushing', Journal of Membrane
Science, vol. 159, no. 1-2, pp. 11-20.

Stephenson, T, Judd, S, Jefferson, B & Brindle, K 2000, Membrane Bioreactors for

Wastewater Treatment, IWA Publishing, London, UK.

Stumm, W & Morgan, JJ 1996, Aquatic Chemistry: Chemical Equilibria and Rates in
Natural Waters, 3rd edn, Wiley-Interscience Publication, New York.

201
Summers, RS & Roberts, PV 1988a, 'Activated carbon adsorption of humic substances - II.
Size exclusion and electrostatic effects', Journal of Colloid and Interface Science, vol. 122,
no. 2, pp. 382-97.

Summers, RS & Roberts, PV 1988b, 'Activated carbon adsorption of humic substances - I.

Heterodisperse mixtures and desorption', Journal of Colloid and Interface Science, vol.
122, no. 2, pp. 367-81.

Sutzkover-Gutman, I, Hasson, D & Semiat, R 2010, 'Humic substances fouling in

ultrafiltration process', Desalination, vol. 261, no. 1-3, pp. 218-31.

Swietlik, J, Dabrowska, A, Raczyk-Stanisławiak, U & Nawrocki, J 2004, 'Reactivity of

natural organic matter fractions with chlorine dioxide and ozone', Water Research, vol. 38,
no. 3, pp. 547-58.

Swietlik, J & Sikorska, E 2004, 'Application of fluorescence spectroscopy in the studies of

natural organic matter fractions reactivity with chlorine dioxide and ozone', Water
Research, vol. 38, no. 17, pp. 3791-9.

Takahashi, N, Nakai, T, Satoh, Y & Katoh, Y 1995, 'Ozonolysis of humic acid and its
effect on decoloration and biodegradability', Ozone: Science & Engineering, vol. 17, no. 5,
pp. 511-25.

Tsujimoto, W, Kimura, H, Izu, T & Irie, T 1998, 'Membrane filtration and pre-treatment by
GAC', Desalination, vol. 119, no. 1-3, pp. 323-6.

Van den Berg, GB & Smolders, CA 1990, 'Flux decline in ultrafiltration processes',
Desalination, vol. 77, pp. 101-33.

Van der Kooij, D, Hijnen, WAM & Kruithof, JC 1989, 'The effects of ozonation, biological
filtration and distribution on the concentration of easily assimilable organic carbon (AOC)
in drinking water', Ozone Science & Engineering, vol. 11, no. 3, pp. 297-311.

Vesilind, AP (ed.) 2003, Wastewater Treatment Plant Design, 1st edn, Water Environment
Federation, USA.

Victorian Government - Department of Sustainability and Environment 2004, Victorian

Government White Paper: Our Water Our Future - Securing Our Water Future Together-

202
Chapter 5: Smarter Water Use in Our Cities and Towns, Victorian Government, viewed 29
April 2008,
<http://www.dse.vic.gov.au/CA256F310024B628/0/E0CB3954FFE9BA17CA256FFE0007
E17A/$File/Chptr5.pdf>.

Vik, EA & Eikebrokk, B 1989, 'Coagulation process for removal of humic substances from
drinking water', in IH Suffet & P MacCarthy (eds), Aquatic Humic Substances: Influence
on Fate and Treatment of Pollutants, American Chemical Society, Washington, pp. 385-
408.

Volk, C, Renner, P, Roche, H, Paillard, H & J.C.Joret 1993, 'Effects of ozone on the
production of biodegradable dissolved organic carbon (BDOC) during water treatment',
Ozone: Science & Engineering, vol. 15, pp. 389-404.

Von Gunten, U 2003, 'Review - Ozonation of drinking water: Part I. Oxidation kinetics and
product formation', Water Research, vol. 37, no. 7, pp. 1443-67.

Voutchkov, N 2008, Pretreatment Technologies for Membrane Seawater Desalination, 1st

edn, Australia Water Association, Australia.

Wang, G-S & Pai, S-Y 2001, 'Ozonation of dissolved organic matter in biologically treated
wastewater effluents', Ozone: Science & Engineering, vol. 23, no. 5, pp. 351-8.

Wang, J, Guan, J, Santiwong, SR & Waite, DT 2008, 'Characterization of floc size and
structure under different monomer and polymer coagulants on microfiltration membrane
fouling', Journal of Membrane Science, vol. 321, no. 1-2, pp. 132-8.

Wang, J, Guan, J, Santiwong, SR & Waite, TD 2010, 'Effect of aggregate characteristics

under different coagulation mechanisms on microfiltration membrane fouling',
Desalination, vol. 258, no. 1-3, pp. 19-27.

Water Environment Federation 2006, Membrane Systems for Wastewater Treatment, 1st
edn, McGraw-Hill, USA.

Weishaar, JL, Aiken, GR, Bergamaschi, BA, Fram, MS, Fujii, R & Mopper, K 2003,
'Evaluation of specific ultraviolet absorbance as an indicator of the chemical composition

203
and reactivity of dissolved organic carbon', Environmental Science & Technology, vol. 37,
no. 20, pp. 4702-8.

Weiss, E, Groenen-Serrano, K & Savall, A 2007, 'Electrochemical mineralization of

sodium dodecylbenzenesulfonate at boron doped diamond anodes', Journal of Applied
Electrochemistry, vol. 37, no. 11, pp. 1337-44.

Wert, EC, Rosario-Ortiz, FL, Drury, DD & Snyder, SA 2007, 'Formation of oxidation
byproducts from ozonation of wastewater', Water Research, vol. 41, no. 7, pp. 1481-90.

Wilf, M & Alt, S 2000, 'Application of low fouling RO membrane elements for reclamation
of municipal wastewater', Desalination, vol. 132, no. 1-3, pp. 11-9.

Wong, S, Hana, JV, King, S, Carroll, TJ, Eldridge, RJ, Dixon, DR, Bolto, BA, Hesse, S,
Abbt-Braun, G & Frimmel, FH 2002, 'Fractionation of natural organic matter in drinking
water and characterization by 13C cross-polarization magic-angle spinning NMR
spectroscopy and size exclusion chromatography', Environmental Science & Technology,
vol. 36, no. 16, pp. 3497-503.

Wricke, B, Petzoldt, H, Heiser, H & Bornmann, K 1996, 'NOM-removal by biofiltration

after ozonation - results of a pilot plant test', in N Graham & R Collins (eds), Advances in
Slow Sand and Alternative Biological Filtration, John Wiley & Sons, England, pp. 51-60.

Wyart, Y, Georges, G, Deumie, C, Amra, C & Moulin, P 2008, 'Membrane characterization

by microscopic methods: Multiscale structure', Journal of Membrane Science, vol. 315, pp.
82-92.

Yamamoto, K, Hissa, M, Mahmood, T & Matsuo, T 1989, 'Direct solid liquid separation
using hollow fibre membrane in an activated sludge aeration tank', Water Science &
Technology, vol. 21, pp. 43-54.

Yigit, NO, Harman, I, Civelekoglu, G, Koseoglu, H, Cicek, N & Kitis, M 2008, 'Membrane
fouling in a pilot-scale submerged membrane bioreactor operated under various conditions',
Desalination, vol. 231, no. 1-3, pp. 124-32.

Yuan, W & Zydney, AL 1999, 'Humic acid fouling during microfiltration', Journal of
Membrane Science, vol. 157, no. 1, pp. 1-12.

204
Yuan, W & Zydney, AL 2000, 'Humic acid fouling in ultrafiltration', Environmental
Science & Technology, vol. 34, no. 23, pp. 5043-50.

Zagorodni, AA 2007, Ion Exchange Materials: Properties and Applications, 1st edn,
Elsevier, Oxford.

Zeman, LJ & Zydney, AL 1996, Microfiltration and Ultrafiltration: Principles and

Applications, Marcel Dekker, New York.

Zhang, M, Li, C, Benjamin, MM & Chang, Y 2003, 'Fouling and natural organic matter
removal in adsorbent/membrane systems for drinking water treatment', Environmental
Science & Technology, vol. 37, no. 8, pp. 1663-9.

Zhang, M, Nguyen, QT & Ping, Z 2009, 'Hydrophilic modification of poly (vinylidene

fluoride) microporous membrane', Journal of Membrane Science, vol. 327, no. 1-2, pp. 78-
86.

Zhang, R, Vigneswaran, S, Ngo, HH & Nguyen, H 2006, 'Magnetic ion exchange

(MIEX®) resin as a pre-treatment to a submerged membrane system in the treatment of
biologically treated wastewater', Desalination, vol. 192, no. 1-3, pp. 296-302.

Zhao, Y-H, Zhu, B-K, Kong, L & Xu, Y-Y 2007, 'Improving hydrophilicity and protein
resistance of ooly(vinylidene fluoride) membranes by blending with amphiphilic
hyperbranched-star polymer', Langmuir, vol. 23, no. 10, pp. 5779-86.

Zheng, X, Ernst, M & Jekel, M 2009, 'Identification and quantification of major organic
foulants in treated domestic wastewater affecting filterability in dead-end ultrafiltration',
Water Research, vol. 43, no. 1, pp. 238-44.

Zheng, X, Ernst, M & Jekel, M 2010, 'Pilot-scale investigation on the removal of organic
foulants in secondary effluent by slow sand filtration prior to ultrafiltration', Water
Research, vol. 44, no. 10, pp. 3203-13.

Zheng, X, Plume, S, Ernst, M, Croué, J-P & Jekel, M 2012, 'In-line coagulation prior to UF
of treated domestic wastewater–foulants removal, fouling control and phosphorus removal',
Journal of Membrane Science, vol. 403-404, pp. 129-39.

205
Zhu, H & Nyström, M 1998, 'Cleaning results characterized by flux, streaming potential
and FTIR measurements', Colloids and Surfaces A: Physicochemical and Engineering
Aspects, vol. 138, pp. 309-21.

Zhu, H, Wen, X, Noguchi, M & Gan, Y 2009, 'Membrane fouling in the reclamation of
secondary effluent with an ozone-membrane hybrid system', Separation Science &
Technology, vol. 44, no. 1, pp. 121-30.

Zhu, H, Wen, X & Huang, X 2010, 'Membrane organic fouling and the effect of pre-
ozonation in microfiltration of secondary effluent organic matter', Journal of Membrane
Science, vol. 352, no. 1-2, pp. 213-21.

Zhu, HT, Wen, XH & Huang, X 2008, 'Pre-ozonation for dead-end microfiltration of the
secondary effluent: suspended particles and membrane fouling', Desalination, vol. 231, no.
1-3, pp. 166-74.

Zhu, L-P, Zhang, X-X, Xu, L, Du, C-H, Zhu, B-K & Xu, Y-Y 2007, 'Improved protein-
adsorption resistance of polyethersulfone membranes via surface segregation of ultrahigh
molecular weight poly(styrene-alt-maleic anhydride)', Colloids and Surfaces B:
Biointerfaces, vol. 57, no. 2, pp. 189-97.

Zularisam, A, Ismail, A & Salim, R 2006, 'Behaviours of natural organic matter in

membrane filtration for surface water treatment - a review', Desalination, vol. 194, no. 1-3,
pp. 211-31.

206
APPENDICES

207
 Appendix A. Contact angles of the flat-sheet polymeric
membranes

The contact angles of the virgin flat sheet membranes in this study were measured with an
OCA20 contact angle meter (DataPhysics Instruments GmbH) using the sessile drop
method with a Milli-Q water droplet volume of 10 μL. The contact angles (computed using
the Young-Laplace equation) were taken within 5 seconds. Five measurements were carried
out for each membrane. The results are shown in Table A1.

Table A1. Contact angle measurements for the flat-sheet polymeric membranes

Contact angle (º) VVLP HFM-180 HFM-116 HFK-141

Replicate 1 59.9 82.6 66.8 40.7
Replicate 2 62.4 83.7 70.1 37.3
Replicate 3 61.5 86.0 66.4 41.0
Replicate 4 64.0 84.6 69.7 43.1
Replicate 5 63.6 80.9 66.9 41.7
Average 62  2 84  2 68  2 41  2

Some images taken from the contact angle meter are shown in Figure A1.

Virgin VVLP, 62.4º Virgin HFM-180, 84.6º

Virgin HFM-116, 69.7º Virgin HFK-141, 41.0º

Figure A1. Images of water droplets on the virgin membrane surfaces taken by the OCA20
contact angle meter

208
 Appendix B. Example of data processing for a filtration
experiment

Date of experiment: 19 February 2010

Membrane type: HFM-180 (PVDF, MWCO = 100 kDa), filtration area = 13.4 cm2
Sample: Raw effluent
Operating conditions: transmembrane pressure = 110 kPa, temperature = 22 ºC

1. Filtration of Milli-Q water using a virgin membrane – Determination of pure water flux
of the virgin membrane J0

Table B1. Flux data for the determination of the virgin membrane’s pure water flux
Time, t (min) Permeate weight (g) Flux, (g/13.4 cm2/min)
0 0 -
1 23.04 23.04
2 46.05 23.01
3 68.98 22.93
4 91.84 22.86
5 114.81 22.97
6 137.73 22.92
7 160.5 22.77
8 183.38 22.88
9 206.23 22.85
10 229.1 22.87

J0 = Average of the last 3 flux data

= 22.87 (g/13.4 cm2/min)
= 1023.88 L m-2 h-1 (taking the density of the permeate as 1000 g L-1)
= 2.84  10-4 m3 m-2 s-1
Hydraulic resistance of the virgin membrane (Rm):
P 110,000( Pa)
Rm = = = 4.04  1011 (m-1)
J 0 0.000958( Pa.s)  2.84  10 (m m s )
4 3  2 1

209
2. Filtration of the raw AS effluent – Plot of J/J0 versus v curve and calculation of the
resistance by total fouling and the UMFI for total fouling

Table B2. Flux data from the filtration test with the raw AS effluent and the HFM-180
membrane
Accumulated
Time, t Permeate Flux, J Normalised flux,
specific permeate
(min) weight (g) -2
(L m-2 h-1) J/J0
volume, v (L m )
0 0 0 1023.88 (= J0) 1
1 9.43 7.04 422.24 0.412
2 16.56 12.36 319.25 0.312
3 22.68 16.93 274.03 0.268
4 28.10 20.97 242.69 0.237
5 33.05 24.66 221.64 0.216
… … … … …
88 200.67 149.75 57.76 0.056
89 201.97 150.72 58.21 0.057

v = permeate volume/membrane area = 201.97 cm3/13.4 cm2 = 150.72 (L m-2) at the end of
the run (89th minute)
Permeate flux at the end of the filtration run ( J  ):
J  = 58.21 L m-2 h-1 = 1.62  10-5 m3 m-2 s-1
Resistance by total fouling ( Rtf ):

P 110,000( Pa)
Rtf =  Rm =  4.04  1011
J  5 3  2 1
0.000958( Pa.s )  1.62  10 (m m s )

= 6.70  1012 (m-1)

The UMFI for total fouling = slope of the curve J 0 J versus v (Figure B1)
= 0.114 (m2 L-1)

210
 20
y = 0.1141x + 1
2
R = 0.9866
15

J0 /J
10

Raw + HFM-180

0
-2
0 50 v (L m ) 100 150

Figure B1. Calculation of the UMFI for total fouling

3. Determination of the pure water flux recovery (WFR) and the resistance by hydraulically
irreversible fouling (Rif)
The fouled membrane was backwashed with the permeate at 110 kPa for 5 minutes. The
pure water flux of the backwashed membrane ( J BW ) was then measured with Milli-Q
water. The data are shown in Table B3.

Table B3. Flux data for the determination of the backwashed membrane’s pure water flux
Time, t (min) Permeate weight (g) Flux, (g/13.4 cm2/min)
0 0 -
1 9.63 9.63
2 19.24 9.61
3 28.88 9.64
4 38.49 9.61
5 48.23 9.75
6 57.88 9.65
7 67.50 9.62
8 77.00 9.5
9 86.55 9.55
10 96.12 9.57

211
J BW = Average of the last 3 flux data
= 9.54 (g/13.4 cm2/min)
= 427.16 L m-2 h-1
= 1.19  10-4 m3 m-2 s-1

Pure water flux recovery by backwashing ( WFR ):

J BW 427.16
WFR    0.42
J0 1023.88

Hydraulic resistance by irreversible fouling ( Rif ):

P 110,000( Pa)
Rif =  Rm =  4.04  1011 = 5.64  1011 (m-1)
J BW 0.000958( Pa.s )  1.19  10  4 (m 3 m  2 s 1 )

212
 Appendix C. Sample calculation of the cleaning efficiency of
the membrane cleaners

Date of experiment: 09 July 2010

Membrane type: HFM-116 (PVDF, MWCO = 50 kDa), filtration area = 13.4 cm2
Sample: Raw effluent
Operating conditions: transmembrane pressure = 110 kPa, temperature = 22 ºC

1. Filtration of Milli-Q water using a virgin membrane – Determination of pure water flux
of the virgin membrane J0

Table C1. Flux data for the determination of the virgin membrane’s pure water flux
Time, t (min) Permeate weight (g) Flux, (g/13.4 cm2/min)
0 0 -
1 3.81 3.81
2 7.65 3.84
3 11.47 3.82
4 15.28 3.81
5 19.09 3.81
6 22.89 3.8
7 26.74 3.85
8 30.53 3.79
9 34.33 3.8
10 38.13 3.8

J0 = Average of the last 3 flux data

= 3.80 (g/13.4 cm2/min)
= 170.00 L m-2 h-1 (taking the density of the permeate as 1000 g L-1)
= 4.72  10-5 m3 m-2 s-1
Hydraulic resistance of the virgin membrane (Rm):
P 110,000( Pa)
Rm = = = 2.43  1012 (m-1)
J 0 0.000958( Pa.s)  4.72  10 (m m s )
5 3  2 1

213
2. Filtration of the raw AS effluent – Plot of J/J0 versus v curve and calculation of the
resistance by total fouling

Table C2. Flux data from the filtration test with the raw AS effluent and the HFM-116
membrane
Accumulated
Time, t Permeate Flux, J Normalised flux,
specific permeate
(min) weight (g) -2
(L m-2 h-1) J/ J0
volume, v (L m )
0 0 0 170.00 (= J0) 1
1 2.92 2.18 130.75 0.769
2 5.6 4.18 120.00 0.706
3 8.17 6.10 115.07 0.677
4 10.63 7.93 110.15 0.648
5 13.01 9.71 106.57 0.627
6 15.28 11.40 101.64 0.598
… … … …
149 194.43 145.10 42.09 0.248
150 195.36 145.79 41.64 0.245

Permeate flux at the end of the filtration run ( J  ):

J  = 41.64 L m-2 h-1 = 1.16  10-5 m3 m-2 s-1
Resistance by total fouling ( Rtf ):

P 110,000( Pa)
Rtf =  Rm =  4.72  1012
J  5 3  2 1
0.000958( Pa.s )  1.16  10 (m m s )

= 9.93 1012 (m-1)

3. Determination of the pure water flux recovery (WFR) and the resistance by hydraulically
irreversible fouling (Rif)
The fouled membrane was backwashed with the permeate at 110 kPa for 5 minutes. The
pure water flux of the backwashed membrane ( J BW ) was then measured with Milli-Q
water. The data are shown in Table C3.

214
Table C3. Flux data for the determination of the backwashed membrane’s pure water flux
Time, t (min) Permeate weight (g) Flux (L m-2 h-1)
0 0 -
1 1.97 88.21
2 3.98 90.00
3 5.97 89.10
4 7.99 90.45
5 10 90.00
6 12.04 91.34
7 14.06 90.45
8 16.09 90.90
9 18.13 91.34
10 20.17 91.34

J BW = Average of the last 3 flux data

= 91.2 L m-2 h-1
= 2.53  10-5 m3 m-2 s-1

Pure water flux recovery by backwashing ( WFR ):

J BW 91.2
WFR    0.54
J0 170

Hydraulic resistance by irreversible fouling ( Rif ):

P 110,000( Pa)
Rif =  Rm =  2.43  1012
J BW 4 3  2 1
0.000958( Pa.s )  2.53  10 (m m s )

Rif = 2.10  1012 (m-1)

4. Determination of the resistance by residual fouling (Rres) after chemical cleaning and the
cleaning efficiency (ERW)
The backwashed membrane was soaked in 75 mL of 1.0 wt.% Terg-a-zyme solution (with
gentle shaking in an orbital shaker at 100 rpm) for 45 minutes. The chemically cleaned
membrane was then rinsed with Milli-Q water and its pure water was measured with Milli-
Q water. The data are shown in Table C4.

215
Table C4. Flux data for the determination of the chemically cleaned membrane’s pure water
flux
Time, t (min) Permeate weight (g) Flux (L m-2 h-1)
0 0 -
1 3.53 158.06
2 7.07 158.51
3 10.62 158.96
4 14.18 159.40
5 17.74 159.40
6 21.31 159.85
7 24.9 160.75
8 28.48 160.30
9 32.05 159.85
10 35.65 161.19

Pure water flux of the chemically cleaned membrane ( J c ):

J c = Average of the last 3 flux data

= 160.4 L m-2 h-1
= 4.45  10-5 m3 m-2 s-1

Pure water flux recovery by chemical cleaning:

J c 160.4
WFR    0.94
J0 170
Resistance by residual fouling after chemical cleaning:
P 110000( Pa)
Rres   Rm   2.43  1012
J c 5 3  2 1
0.000958( Pa.s )  4.45  10 (m m s )

Rres  1.45  1011 (m-1)

Cleaning efficiency ( E RW ) of the 1.0 wt.% Terg-a-zyme solution:

Rif  Rres 2.10  1012  1.45  1011

E RW   100  = 93 %
Rif 2.10  1012

216
 Appendix D. Effect of ozonation and BAC filtration on the
effluent characteristics and the UMFI for total fouling for the
HFM-180 membrane

Table D1. Characteristics of the samples collected on 03 November 2008

Ozonated BAC
Parameter Raw effluent
effluent filtrate
pH 7.5 7.6 7.1
DOC (mg L-1) 13.4 15.3 11.9
UVA254 (cm-1) 0.327 0.186 0.150
SUVA (L m-1 mg-1) 2.44 1.22 1.26
True colour (Pt-Co units) 89 28 25
Turbidity (NTU) 5.6 5.2 0.9

Table D2. Characteristics of the samples collected on 07 November 2008

Ozonated BAC
Parameter Raw effluent
effluent filtrate
pH 7.6 7.7 7.3
DOC (mg L-1) 14.8 16.5 13.2
UVA254 (cm-1) 0.400 0.260 0.234
SUVA (L m-1 mg-1) 2.70 1.58 1.77
True colour (Pt-Co units) 118 27 24
Turbidity (NTU) 2.0 2.2 0.7

Table D3. Characteristics of the samples collected on 08 December 2008

Ozonated BAC
Parameter Raw effluent
effluent filtrate
pH 7.5 7.8 7.2
DOC (mg L-1) 13.8 14.2 11.1
UVA254 (cm-1) 0.353 0.199 0.172
SUVA (L m-1 mg-1) 2.56 1.40 1.55
True colour (Pt-Co units) 90 19 16
Turbidity (NTU) 3.8 3.3 0.8

217
Table D4. Characteristics and UMFI for total fouling for the samples collected on 13 March
2009
Ozonated BAC
Parameter Raw effluent
effluent filtrate
pH 7.8 7.6 7.1
DOC (mg L-1) 12.6 13.8 12.1
UVA254 (cm-1) 0.400 0.247 0.226
SUVA (L m-1 mg-1) 3.17 1.79 1.87
True colour (Pt-Co units) 123 27 25
Turbidity (NTU) 4.0 4.0 0.8
UMFI (m2 L-1) ( HFM-180 membrane) 0.147 0.112 0.092

Table D5. Characteristics and UMFI for total fouling for the samples collected on 27 March
2009
Ozonated BAC
Parameter Raw effluent
effluent filtrate
pH 7.8 7.7 7.3
DOC (mg L-1) 12.0 13.3 11.8
UVA254 (cm-1) 0.400 0.243 0.223
SUVA (L m-1 mg-1) 3.33 1.83 1.89
True colour (Pt-Co units) 112 26 25
Turbidity (NTU) 3.6 3.5 0.8
UMFI (m2 L-1) (HFM-180 membrane) 0.075 0.048 0.040

Table D6. Characteristics and UMFI for total fouling for the samples collected on 30 March
2009
Ozonated
Parameter Raw effluent BAC filtrate
effluent
pH 7.6 7.5 7.2
DOC (mg L-1) 14.2 15.5 14.0
UVA254 (cm-1) 0.343 0.206 0.190
SUVA (L m-1 mg-1) 2.42 1.32 1.39
True colour (Pt-Co units) 90 19 18
Turbidity (NTU) 5.5 5.3 0.8
UMFI (m2 L-1) (HFM-180 membrane) 0.157 0.116 0.087

218
Table D7. Characteristics and UMFI for total fouling for the samples collected on 24 April
2009
Ozonated BAC
Parameter Raw effluent
effluent filtrate
pH 7.6 7.6 7.2
DOC (mg L-1) 11.3 12.9 10.9
UVA254 (cm-1) 0.375 0.240 0.210
SUVA (L m-1 mg-1) 3.32 1.86 1.93
True colour (Pt-Co units) 118 28 26
Turbidity (NTU) 4.2 4.0 0.8
UMFI (m2 L-1) ( HFM-180 membrane) 0.074 0.056 0.042

Table D8. Characteristics and UMFI for total fouling for the samples collected on 16 June
2009
Ozonated BAC
Parameter Raw effluent
effluent filtrate
pH 7.2 7.3 6.5
DOC (mg L-1) 13.8 14.6 13.3
UVA254 (cm-1) 0.353 0.210 0.192
SUVA (L m-1 mg-1) 2.56 1.44 1.44
True colour (Pt-Co units) 90 19 16
Turbidity (NTU) 3.8 3.5 0.6
UMFI (m2 L-1) (HFM-180 membrane) 0.068 0.041 0.034

219
Appendix E. Effect of ozonation and BAC filtration on the DOC
concentrations of the EfOM fractions as determined by resin
fractionation

18
VHA
15.5
16 SHA
CHA 14.2 14.0
14
NEU
12 Total
DOC (mg L )
-1

10

8 7.2
5.8
6 5.1 4.7
4.6
4 2.9 2.7 2.4 2.4
1.9 2.2 1.8
2

0
Raw effluent Ozonated effluent BAC filtrate

Figure E1. Effect of ozonation and BAC filtration on the DOC concentrations
of the EfOM fractions of the raw effluent
(Samples collected on 30 March 2009)

220
Appendix F. LC-OCD chromatograms of the samples listed in
Table 6.3 but not included in Figure 6.2

HS Building blocks
14

12 Low MW HS and low MW acids

OCD response (relative units)

10
Raw
8 Raw + VVLP
Biopolymers Ozonated
6 Ozonated + VVLP

4 Low MW neutrals

0
20 30 40 50 60 70 80 90 100
Retention time (min)
Figure F1. AMWDs of the raw and ozonated effluent and the
corresponding permeates obtained from MF with the VVLP membrane

14 HS
Building blocks

12
Low MW acids and low MW HS
OCD response (relative units)

10

8 BAC filtrate
Biopolymers
BAC filtrate + HFM-180
6

4 Low MW neutrals

0
20 40 60 80 100
Retention time (min)

Figure F2. AMWDs of the BAC filtrate and the permeate obtained
from UF with the HFM-180 membrane

221
 Appendix G. Particle size distributions and zeta potentials of the
raw effluent and the ozonated effluent

7
Raw, replicate 1
6
Raw, replicate 2
5 Raw, replicate 3
Volume %

0
1 10 100 1000 10000
Particle size (micron)

Figure G1. Particle size distributions of the raw effluent collected on 10 August 2009

8
Ozonated, replicate 1
7
Ozonated, replicate 2
6 Ozonated, replicate 3
5
Volume %

0
1 10 100 1000 10000
Particle size (micron)

Figure G2. Particle size distributions of the ozonated effluent collected on 10 August 2009

222
Table G1. Effect of ozonation on the zeta potentials of the raw effluent
Zeta potential (mV) Raw effluent Ozonated effluent
Replicate 1 -13.4  1.2 -14.7  1.7
Replicate 2 -12.7  1.0 -14.5  1.3
Replicate 3 -13.4  2.1 -14.0  0.7
Replicate 4 -13.1  1.4 -14.1  2.1
Replicate 5 -13.3 1.5 -14.0  1.9
Average -13.2  1.4 -14.3  1.5

223
Appendix H. LC-OCD chromatograms of the ozonated effluent
and its MF permeate with the XUNP-003 membrane
(Samples listed in Table 7.2 but not included in Figure 7.3)

15 HS Building blocks
OCD response (relative units)

Low MW acids and low MW

10
Ozonated
Biopolymers Ozonated + XUNP-003

5
Low MW neutrals

0
20 30 40 50 60 70 80 90 100
Retention time (min)

Figure H1. LC-OCD chromatograms of the ozonated effluent and the corresponding
permeate from MF with the XUNP-003 membrane

224
 Appendix I. Characteristics of the samples used for the UF tests
shown in Figure 7.4

Table I1. Effect of ozonation and adsorption pre-treatment with the AER and the PAC on
the characteristics of the raw effluent samples (samples collected on 01 April 2011)
Raw + 10 mL Raw + 150 Ozonated (Raw +
Parameter Raw -1 -1
AER L mg PAC L 10 mg O3 L-1)
pH 7.3 7.5 7.4 7.4
DOC (mg L-1) 14.3 8.8 6.3 15.0
UVA254 (cm-1) 0.371 0.128 0.106 0.221
SUVA (m-1mg-1 L) 2.59 1.45 1.68 1.47
True colour (Pt-Co units) 110 24 26 24
Turbidity (NTU) 6.6 N/M* N/M* N/M*
*
Not measured

225
 Appendix J. Effect of cyclic chemical cleaning with SDS and
Terg-a-zyme on the permeate flux of the HFM-116 membrane

180
Virgin membrane
160 After first chemical cleaning
After second chemical cleaning
140
After third chemical cleaning
120
J (L m h )
-1

100
-2

80

60

40

20

0
0 50 -2 100 150
v (L m )

Figure J1. Effect of chemical cleaning with SDS on the permeate flux in UF of the raw
effluent with the HFM-116 membrane over three cleaning cycles

180
Virgin membrane
160 After first chemical cleaning
After second chemical cleaning
140
J (L m h )
-1

After third chemical cleaning

-2

120

100

80

60

40

20
-2
0 50 v (L m ) 100 150

Figure J2. Effect of chemical cleaning with Terg-a-zyme on the permeate flux in UF of the
raw effluent with the HFM-116 membrane over three cleaning cycles

226
Appendix K. SEM images of the virgin and chemically cleaned
HFM-116 membrane samples

(b)
(a)

(c) (d)

Figure K1. SEM images of the HFM-116 membrane

(a) Virgin, (b) Fouled by the raw effluent,
(c) After backwashing with the permeate, and (d) After cleaning with NaOH

227
 Appendix L. Characteristics of the samples used in cross-flow
filtration tests

Table L1. Characteristics of the raw effluent samples used in cross-flow filtration tests in
Chapter 9
DOC UVA254 True colour Turbidity
Sample collection date
(mg L-1) (cm-1) (Pt-Co unit) (NTU)
15 April 2011 14.8 0.388 81 2.6
23 May 2011 12.1 0.388 108 3.2
08 July 2011 15.1 0.389 100 1.9
09 September 2011 12.2 0.371 90 3.5
30 September 2011 14.2 0.406 104 2.6

Table L2. Characteristics of the ozonated effluent samples used for Figures 9.9 and 9.10
Sample collection DOC UVA254 True colour Turbidity
-1 -1
date (mg L ) (cm ) (Pt-Co unit) (NTU)
09 September 2011
11.9 0.249 22 3.5
(Figure 9.9)
30 September 2011
15.0 0.258 24 2.4
(Figure 9.10)

228