DOCTOR OF PHILOSOPHY

Applications of Membrane Aerated Biofilm Reactors for Wastewater Treatment

Murray, Simon

Award date:
2016

Awarding institution:
Queen's University Belfast

Link to publication

Terms of use
All those accessing thesis content in Queen’s University Belfast Research Portal are subject to the following terms and conditions of use

• Copyright is subject to the Copyright, Designs and Patent Act 1988, or as modified by any successor legislation
• Copyright and moral rights for thesis content are retained by the author and/or other copyright owners
• A copy of a thesis may be downloaded for personal non-commercial research/study without the need for permission or charge
• Distribution or reproduction of thesis content in any format is not permitted without the permission of the copyright holder
• When citing this work, full bibliographic details should be supplied, including the author, title, awarding institution and date of thesis

Take down policy

A thesis can be removed from the Research Portal if there has been a breach of copyright, or a similarly robust reason.
If you believe this document breaches copyright, or there is sufficient cause to take down, please contact us, citing details. Email:
openaccess@qub.ac.uk

Supplementary materials
Where possible, we endeavour to provide supplementary materials to theses. This may include video, audio and other types of files. We
endeavour to capture all content and upload as part of the Pure record for each thesis.
Note, it may not be possible in all instances to convert analogue formats to usable digital formats for some supplementary materials. We
exercise best efforts on our behalf and, in such instances, encourage the individual to consult the physical thesis for further information.

Download date: 12. Jan. 2020

Applications of Membrane Aerated Biofilm
Reactors for Wastewater Treatment

Simon Thomas Murray MEng

A thesis submitted for the Degree of Doctor of Philosophy

The School of Chemistry and Chemical Engineering

Queen’s University Belfast

2016
Acknowledgements

I would like to thank those who have been involved in the supervision of my project at
the various times during its duration: Dr Wendy McLoone (née McMinn), Professor
Stephen Allen, Professor Ronnie Magee, Professor David Rooney, Dr Elaine Groom,
and Mr Joel Ferguson. Their guidance, inspiration and most of all patience have been
instrumental in the completion of this project.

Without the funding provided through the QUESTOR research programme and the
guidance provided by the members of the QUESTOR Industrial Advisory Board this
project would not have been possible. In particular, I would like to thank Sam Irwin and
Karen McDowell of NI Water, John Toner of Williams Industrial Services and Paddy
McGuinness of Colloide Engineering who have supported the BioSettler project from
the beginning.

I am also indebted to the staff of the QUESTOR ATU for their assistance with various
aspects of this work; Dr Julie-Anne Hanna, Ciarán Prunty, David Parker, Patricia
McCrory, Kathyrn Rogers, Kerry Kelly and Dr Nick Johnston. A special mention must
go to the late Alex Marshall, whose unique brand of genius was essential to the
completion of this project.

I am also grateful to final year students who have assisted me in some of the
experimental work presented here – Ciarán Doyle, Barry McQuaid and Niall Moroney –
and to some of the technical staff of the School of Chemistry and Chemical Engineering
who have assisted in the construction and maintenance of the equipment used in this
work, especially Jackie O’Connor, Suzanne Evans, Martin Catney and Kirin Hill.

i
Summary
Despite being the subject of peer reviewed research since the mid-1980s, the
conservative nature of the wastewater treatment industry means that the commercial
application of membrane aerated biofilm reactors has not realized the potential that the
published research demonstrates.

The early research demonstrated the ability of membrane aerated biofilm reactors to
achieve good levels of pollutant removal from various types of wastewater, but also
exposed several weaknesses of the technology (i.e. cost of membranes, control of
biofilm thickness) which have prevented the concept of MABfRs being developed in
viable wastewater treatment technologies.

However, as membrane technology has developed, the cost of suitable membranes has
fallen, prompting the research community to revisit the concept. This later batch of
research has identified several niche applications where membrane supported biofilms
can be used for effective removal of pollutants from water.

Using the MABfR for the treatment of secondary effluent as a polishing step is another
niche application which has been identified and is examined in this work; leading to the
development of a patented treatment technology – the BioSettler.

ii
Table of Contents

Acknowledgements .............................................................................................................i

Summary ........................................................................................................................... ii

Table of Contents ............................................................................................................. iii

List of Figures .................................................................................................................... x

List of Tables...................................................................................................................xiv

Abbreviations and Symbols ............................................................................................xvi

List of Abbreviations...................................................................................................xvi

List of Symbols ......................................................................................................... xvii

1 Introduction ................................................................................................................ 1

1.1 General Introduction ............................................................................................ 1

1.2 Wastewater treatment .......................................................................................... 2

1.2.1 Wastewater treatment in Northern Ireland ................................................... 4

1.3 Drivers for upgrade of wastewater treatment works ........................................... 5

1.3.1 Legislative drivers ........................................................................................ 5

1.3.2 Other drivers for wastewater treatment upgrade .......................................... 8

1.4 Proposed technology – The BioSettlerTM ............................................................ 8

1.5 Aims and Objectives ......................................................................................... 11

1.6 Conclusions ....................................................................................................... 12

2 Review of wastewater sources, composition and treatment options ........................ 13

2.1 Wastewater and wastewater treatment .............................................................. 13

2.1.1 Sources of wastewater ................................................................................ 13

2.1.2 Wastewater characterisation....................................................................... 17

2.1.3 Current wastewater treatment technologies ............................................... 20

iii
 2.1.4 Biofilms ...................................................................................................... 24

2.2 Biology of wastewater treatment ....................................................................... 26

2.2.1 Aerobic heterotrophy ................................................................................. 27

2.2.2 Nitrification ................................................................................................ 28

2.2.3 Denitrification ............................................................................................ 30

2.2.4 Stoichiometry of wastewater biology......................................................... 31

2.3 Bubbleless Aeration .......................................................................................... 32

2.3.1 Historical Context ...................................................................................... 33

2.3.2 Mass transfer mechanism ........................................................................... 34

2.3.3 Modes of operation .................................................................................... 35

2.3.4 Evaluation of Oxygen Mass Transfer Coefficient ..................................... 36

2.3.5 Calculation of oxygen flux ......................................................................... 38

2.3.6 Bubble formation ....................................................................................... 38

2.3.7 Effect of attached biomass on oxygen transfer .......................................... 39

2.3.8 Design of hollow fibre membrane contactors ............................................ 40

2.4 Membrane Attached Biofilm Processes ............................................................ 41

2.4.1 Pollutant removal ....................................................................................... 42

2.4.2 Control of biofilm thickness....................................................................... 45

2.5 Biological azo dye wastewater treatment .......................................................... 45

2.6 Inclined Plate Settlers ........................................................................................ 48

2.6.1 Design Considerations ............................................................................... 49

2.6.2 Operational Issues ...................................................................................... 50

2.6.3 Modelling ................................................................................................... 50

2.7 Conclusion ......................................................................................................... 52

3 Materials and Methods ............................................................................................. 53

iv
 3.1 Mass Transfer Studies ....................................................................................... 53

3.1.1 Membrane Materials .................................................................................. 53

3.1.2 Membrane Modules ................................................................................... 54

3.1.3 Parameter measurement ............................................................................. 55

3.1.4 Nitrogen diffusion ...................................................................................... 56

3.1.5 Stirring equipment...................................................................................... 56

3.1.6 Pumping equipment ................................................................................... 56

3.1.7 Datalogging ................................................................................................ 57

3.1.8 Experimental Set-up ................................................................................... 57

3.1.9 Experimental procedure ............................................................................. 58

3.2 Membrane Aerated Biofilm Reactor Studies .................................................... 60

3.2.1 Reactor Set-ups .......................................................................................... 61

3.2.2 Synthetic media .......................................................................................... 63

3.2.3 Seeding of reactors ..................................................................................... 66

3.3 Effluent characterisation .................................................................................... 67

3.3.1 pH ............................................................................................................... 68

3.3.2 Ammoniacal Nitrogen ................................................................................ 68

3.3.3 Chemical Oxygen Demand ........................................................................ 68

3.3.4 Nitrate......................................................................................................... 69

3.3.5 Nitrite ......................................................................................................... 70

3.3.6 Turbidity..................................................................................................... 70

3.3.7 Colour......................................................................................................... 70

3.3.8 Biochemical Oxygen Demand (BOD5) ...................................................... 71

3.4 Effluent disposal ................................................................................................ 72

4 Mass Transfer Studies .............................................................................................. 73

v
 4.1 Saturation concentration .................................................................................... 73

4.1.1 Effect of membrane type ............................................................................ 74

4.1.2 Effect of inlet pressure ............................................................................... 76

4.2 Oxygen Transfer ................................................................................................ 77

4.2.1 General observations .................................................................................. 77

4.2.2 Calculation of average oxygen flux ........................................................... 79

4.2.3 Calculation of mass transfer coefficient..................................................... 79

4.3 Effect of air side flowrate .................................................................................. 80

4.4 Effect of inlet air pressure ................................................................................. 83

4.5 Effect of water side flowrate ............................................................................. 86

4.6 Conclusions ....................................................................................................... 94

5 Membrane Aerated Biofilm Reactor Studies – Part 1 .............................................. 95

5.1 Reactor start-up ................................................................................................. 95

5.2 General observations ......................................................................................... 96

5.2.1 pH ............................................................................................................... 96

5.2.2 Nitrite concentration................................................................................... 98

5.3 Stoichiometric model development ................................................................... 99

5.4 Pollutant removal ............................................................................................ 104

5.4.1 Removal Efficiency.................................................................................. 104

5.4.2 Removal rates ........................................................................................... 106

5.4.3 Total Nitrogen Removal ........................................................................... 107

5.5 Response to increased inlet pressure ............................................................... 110

5.6 Apparent oxygen flux ...................................................................................... 112

5.6.1 Assumptions ............................................................................................. 112

5.6.2 Reaction scheme....................................................................................... 114

vi
 5.6.3 Effect of inlet pressure ............................................................................. 115

5.6.4 Mass transfer coefficient .......................................................................... 118

5.7 Conclusions ..................................................................................................... 122

6 Membrane Aerated Biofilm Reactor Studies – Part 2 ............................................ 123

6.1 Pollutant removal ............................................................................................ 123

6.1.1 Chemical Oxygen Demand ...................................................................... 124

6.1.2 Ammoniacal nitrogen ............................................................................... 126

6.1.3 Total nitrogen ........................................................................................... 128

6.2 Removal rates .................................................................................................. 129

6.2.1 Oxygen consumption ............................................................................... 130

6.2.2 Relative oxygen consumption .................................................................. 132

6.2.3 Denitrification .......................................................................................... 136

6.3 Model development ......................................................................................... 139

6.3.1 Assumptions ............................................................................................. 140

6.3.2 Model inputs ............................................................................................ 142

6.4 Model validity ................................................................................................. 145

6.4.1 Normalised standard deviation method .................................................... 145

6.4.2 Sensitivity analysis ................................................................................... 149

6.5 Conclusions ..................................................................................................... 154

7 Design and operation of a pilot-scale BioSettlerTM................................................ 155

7.1 Pilot scale design ............................................................................................. 155

7.1.1 Justification of design .............................................................................. 157

7.1.2 Membranes ............................................................................................... 161

7.1.3 Auxillary equipment................................................................................. 164

7.2 Trial locations .................................................................................................. 165

vii
 7.2.1 Site 1 ........................................................................................................ 165

7.2.2 Site 2 ........................................................................................................ 165

7.3 Trial results ...................................................................................................... 166

7.3.1 Site 1 ........................................................................................................ 166

7.3.2 Trial 2 ....................................................................................................... 169

7.3.3 Solids Removal ........................................................................................ 170

7.3.4 Biochemical Oxygen Demand ................................................................. 171

7.3.5 Ammoniacal Nitrogen .............................................................................. 173

7.4 Potential performance of scaled up system ..................................................... 174

7.4.1 Pollutant removal ..................................................................................... 174

7.4.2 Energy consumption................................................................................. 176

7.4.3 Energy consumption of unit pollutant removal ........................................ 177

7.5 Conclusions ..................................................................................................... 178

8 Treatment of azo dye waste in the Membrane Aerated Biofilm Reactor – a

feasibility study .............................................................................................................. 179

8.1 Experimental conditions .................................................................................. 179

8.2 General observations ....................................................................................... 180

8.2.1 Biofilm establishment .............................................................................. 180

8.2.2 Method of colour removal........................................................................ 181

8.2.3 pH ............................................................................................................. 182

8.2.4 Ammoniacal nitrogen concentration ........................................................ 183

8.2.5 Nitrate concentration ................................................................................ 184

8.2.6 UV/vis spectrometry ................................................................................ 185

8.3 Colour removal ................................................................................................ 190

8.3.1 AO7 Removal rate.................................................................................... 191

viii
 8.4 COD removal .................................................................................................. 193

8.4.1 COD removal rate .................................................................................... 194

8.5 Worked example .............................................................................................. 196

8.6 Conclusions ..................................................................................................... 197

9 Conclusions and Further Work .............................................................................. 199

9.1 Summary of conclusions ................................................................................. 199

9.2 Recommendations of further work .................................................................. 200

References ...................................................................................................................... 202

ix
List of Figures

Figure 1-1: Simplified schematic of Activated Sludge Process (adapted from Gray,
2004) .................................................................................................................................. 3
Figure 1-2: Trickling filter at NIW Parkgate WwTW ....................................................... 3
Figure 1-3: Rotating Biological Contactor (WIS Ltd) ....................................................... 4
Figure 1-4: Aerial view of Belfast WwTW, highlighting the relative sizes of aerobic
treatment and settlement tanks (Adapted from GoogleMaps) ........................................... 8
Figure 1-5: Inclined plate showing projected settling area and footprint (adapted from
Metso Minerals, 2006) ....................................................................................................... 9
Figure 1-6: Structure of membrane aerated biofilms ....................................................... 10
Figure 1-7: Membrane location in the BioSettler ............................................................ 11
Figure 2-1: Diurnal variations in wastewater flowrate and strength for a typical WwTW
(Tchobanoglous & Burton, 1991b) .................................................................................. 15
Figure 2-2: Structure of Acid Orange 7 ........................................................................... 16
Figure 2-3: Schematic Inclined Plate Settler (Parkson, 2010) ......................................... 21
Figure 2-4: Inclined Plates ............................................................................................... 21
Figure 2-5: Molecular transport in dense membranes (left) by differences in permeation
and microporous membranes by molecular filtration ...................................................... 22
Figure 2-6: Membrane classification (Radcliff & Zarnadze, 2004) ................................. 23
Figure 2-7: General structure of Membrane Aerated Biofilms ........................................ 26
Figure 2-8: Steps in the reduction of nitrate (adapted from Madigan & Martinko, 2006)
.......................................................................................................................................... 30
Figure 2-9: Examples of microbial respiration and associated redox potentials (adapted
from Madigan & Martinko, 2006) ................................................................................... 31
Figure 2-10: Anaerobic and aerobic degradation pathways of azo dyes ......................... 46
Figure 3-1: Structure of silicone rubber (left) and polyethersulphone polymers ............. 53
Figure 3-2: Silicone rubber membrane module ............................................................... 54
Figure 3-3: Schematic diagram of experimental set-up ................................................... 57
Figure 3-4: Water reservoir tank, magnetic stirrer and Hach-Lange LDO Meter ........... 58
Figure 3-5: Schematic diagram of MABfR A Set-up ...................................................... 61

x
Figure 3-6: Schematic diagram of MABfR B set-up ....................................................... 62
Figure 3-7: Spent (left) and unused COD vials ................................................................ 69
Figure 3-8: Fresh (left) and used nitrate vials .................................................................. 69
Figure 3-9: Spectrophotometer cuvettes .......................................................................... 71
Figure 4-1: Temperature-saturation correlations (0.5 bar) ............................................... 75
Figure 4-2: Effect of inlet pressure on saturation oxygen concentration (20 °C) ............ 76
Figure 4-3: Change in dissolved oxygen over duration of mass transfer experiments
(Silicone Rubber, 0.5 bar inlet pressure, 2 lpm gas side flowrate, 550 ml min-1 water side
flowrate). .......................................................................................................................... 77
Figure 4-4: Bubble formation on membrane surface (Silicone Rubber, 0.5 bar inlet
pressure, 2 lpm gas side flowrate, 550 ml min-1 water side flowrate) ............................. 78
Figure 4-5: Plot of 𝒍𝒏𝑪 ∗ −𝑪𝒕𝑪 ∗ versus t (Silicone Rubber, 0.5 bar inlet pressure, 2 lpm
gas side flowrate, 550 ml min-1 water side flowrate) ....................................................... 80
Figure 4-6: Effect of air side flowrate on average oxygen flux (0.5 bar inlet pressure, 550
ml min-1 water flowrate – minimum and maximum value error bar) .............................. 81
Figure 4-7: Effect of air flowrate on overall mass transfer coefficient (0.5 bar inlet air
pressure, 550 ml min-1 water flowrate – minimum and maximum value error bars)....... 83
Figure 4-8: Effect of inlet pressure on average oxygen flux (2 lpm air flowrate, 550 ml
min-1 water flowrate, minimum and maximum value error bars) .................................... 84
Figure 4-9: Effect of inlet air pressure on mass transfer coefficient (2 lpm air flowrate,
550 ml/min water flowrate, minimum and maximum value error bars) .......................... 85
Figure 4-10: Effect of water flowrate on average oxygen flux (0.5 bar air pressure, 2 lpm
air flowrate, minimum and maximum value error bars) .................................................. 86
Figure 4-11: Effect of water flowrate on overall mass transfer coefficient (0.5 bar inlet
air pressure, 2 lpm air flowrate, minimum and maximum value error bars) ................... 87
Figure 4-12: ‘Wilson plot’, minimum and maximum value error bars ............................ 89
Figure 4-13: Sherwood number versus Reynolds number (0.5 bar, 2 lpm inlet flow,
minimum and maximum value error bars) ....................................................................... 91
Figure 5-1: Influent and effluent pH ................................................................................ 97
Figure 5-2: Bulk nitrite concentration (days 1-190) ........................................................ 98
Figure 5-3: fs/sludge age relationships ........................................................................... 101

xi
Figure 5-4: Pollutant removal efficiency in Runs 1-7 (standard error bars) .................. 104
Figure 5-5: Effect of inlet pressure on pollutant removal rates (standard error bars) .... 106
Figure 5-6: total nitrogen removal in Runs 1-7 (standard error bars) ............................ 108
Figure 5-7: Effect of inlet pressure on denitrification rate ............................................. 109
Figure 5-8: Variation in COD and Ammoniacal nitrogen concentrations in response to
changes in inlet air pressure. .......................................................................................... 111
Figure 5-9: Simplified reaction scheme ......................................................................... 114
Figure 5-10: Effect of inlet pressure on apparent oxygen flux (standard error bars) ..... 115
Figure 5-11: Effect of inlet pressure on overall oxygen mass transfer coefficient ........ 120
Figure 6-1: Influent and Effluent COD concentrations in Runs 8-12 ............................ 124
Figure 6-2: Average COD concentrations and percentage COD removal (standard error
bars) ................................................................................................................................ 125
Figure 6-3: Influent and effluent ammoniacal nitrogen concentrations during Runs 8-12
........................................................................................................................................ 126
Figure 6-4: Average ammoniacal nitrogen concentrations and percentage ammoniacal
nitrogen removal in Runs 8 - 12(standard error bars) .................................................... 127
Figure 6-5: Influent and effluent Total-N concentrations in Runs 8-12 ........................ 128
Figure 6-6: Apparent oxygen flux during Runs 8-12 ..................................................... 131
Figure 6-7: fraction of supplied oxygen utilized by aerobic heterotrophs at different
COD:Amm-N ratios ....................................................................................................... 134
Figure 6-8: Variation of denitrification rates with nitrate availability (standard error bars)
........................................................................................................................................ 138
Figure 6-9: Sensitivity analysis for inlet pressure .......................................................... 150
Figure 6-10: Sensitivity analysis for influent ammoniacal nitrogen concentration ....... 151
Figure 6-11: Sensitivity analysis for influent COD concentration................................. 152
Figure 6-12: Sensitivity analysis for nitrate nitrogen ..................................................... 153
Figure 7-1: CAD diagrams – cross sectional (left) and bird’s eye view ........................ 156
Figure 7-2: BioSettler prototype tank ............................................................................ 156
Figure 7-3: Inlet flow patterns........................................................................................ 158
Figure 7-4: Membrane arrangement in BioSettler ......................................................... 159
Figure 7-5: BioSettler plate spacing............................................................................... 159

xii
Figure 7-6: V-notch weirs in BioSettler tank ................................................................. 160
Figure 7-7: Cross flow operation ................................................................................... 162
Figure 7-8: Daicen Membrane Module .......................................................................... 162
Figure 7-9: Membrane module construction .................................................................. 163
Figure 7-10: Membrane sealing with epoxy resin.......................................................... 163
Figure 7-11: Membrane modules in BioSettler unit ...................................................... 164
Figure 7-12: Layout of Newtownbreda WwTW (Googlemaps). ................................... 166
Figure 7-13: Second influent intake point used at Site 1. .............................................. 167
Figure 7-14: Sludge buildup on membranes .................................................................. 168
Figure 7-15: Damage to membranes .............................................................................. 168
Figure 7-16: Sample location at Site 2 ........................................................................... 170
Figure 7-17: Solids concentrations during BioSettler trial 2 ......................................... 171
Figure 7-18: BOD concentrations during BioSettler trial 2 ........................................... 172
Figure 7-19: Ammoniacal nitrogen concentrations during BioSettler Trial 2 ............... 173
Figure 7-20: Zena membrane module ............................................................................ 175
Figure 8-1: Biofilm development after 12 days of operation ......................................... 180
Figure 8-2: Removed biomass initially (left) and during greatest colour removal ........ 182
Figure 8-3: Influent and effluent pH during Run 2 ........................................................ 182
Figure 8-4: Ammoniacal concentrations throughout dye investigation ......................... 184
Figure 8-5: Absorbance spectra for standard AO7 solutions ......................................... 186
Figure 8-6: Calibration curve for AO7 concentration .................................................... 187
Figure 8-7: Solutions of Acid Orange 7 (left), Fe (III) Chloride and Riboflavin (right) at
the concentration used in the influent media ................................................................. 188
Figure 8-8: Cumulative absorbance effects.................................................................... 189
Figure 8-9: Colour removal in Runs 1-3 ........................................................................ 190
Figure 8-10: Influent and effluent COD concentrations during Runs 1 - 3 ................... 194

xiii
List of Tables
Table 1-1: Level of Treatment mandated by the UWwTD (adapted from DEFRA, 2002)
............................................................................................................................................ 6
Table 2-1 : Typical composition of untreated municipal wastewater .............................. 14
Table 2-2: Typical analysis of Dye house wastewater (Wilkinson 2007) ....................... 16
Table 2-3: Typical distribution of Aerobic Heterotrophic Bacteria in Activated Sludge
(Bitton, 2005) ................................................................................................................... 28
Table 2-4: Comparison of pollutant removal rates achieved in selected MABfR studies
.......................................................................................................................................... 43
Table 2-5: Denitrification rates of various MABfR studies ............................................. 44
Table 3-1: Synthetic waste composition .......................................................................... 64
Table 3-2: Composition of stock solutions used with MABfR A .................................... 65
Table 3-3: Composition of synthetic dye waste used in MABfR B ................................. 66
Table 3-4: Wastewater analyses ....................................................................................... 67
Table 4-1: Comparison of saturation constants ................................................................ 74
Table 4-2: Saturation coefficients for 0.5 barg inlet pressure .......................................... 74
Table 4-3: Pressure flowrate pairs used in inlet air pressure investigation ...................... 84
Table 4-4: Experimental and calculated membrane mass transfer coefficients ............... 89
Table 4-5: Comparison of membrane mass transfer coefficients..................................... 90
Table 4-6: Obtained Sherwood number/ Reynolds number relationships ....................... 91
Table 4-7: Mass transfer correlations for hollow fibre modules ...................................... 92
Table 5-1: Conditions during start-up period ................................................................... 95
Table 5-2: Conditions throughout inlet pressure investigation ........................................ 96
Table 5-3: Average and standard deviation pH................................................................ 98
Table 5-4: Values of ae used in model development (McCarty, 1975) .......................... 100
Table 5-5: Cell decay rates (Manser et al., 2006) .......................................................... 100
Table 5-6: Mass ratios of substrates involved in microbial reactions in the MABfR .... 103
Table 5-7: Comparison of pollutant removal efficiencies.............................................. 105
Table 5-8: Comparison of pollutant removal rates ........................................................ 106
Table 5-9: Values used in calculation of gas side oxygen concentration ...................... 120
Table 5-10: Mass transfer coefficients at different experimental conditions ................. 121

xiv
Table 6-1: Pollutant loadings in Run 8-12 ..................................................................... 124
Table 6-2: COD and ammoniacal nitrogen removal rates in Runs 8-12 (standard errors)
........................................................................................................................................ 129
Table 6-3: Comparison of pollutant removal rates (standard errors where shown) ....... 130
Table 6-4: Obtained and predicted average oxygen fluxes ............................................ 131
Table 6-5: Average oxygen fluxes in runs 8-12 ............................................................. 132
Table 6-6: Rates of microbial processes and oxygen uptake (mean values, standard
errors) ............................................................................................................................. 133
Table 6-7: Oxygen usage in Runs 8-12 (mean values) .................................................. 134
Table 6-8: Denitrification rates in runs 8-12 (mean values, standard errors shown) ..... 136
Table 6-9: Comparison of denitrification rates .............................................................. 136
Table 6-10: Relationships used to calculate reaction rates in MABfR model ............... 144
Table 6-11: Experimental and calculated COD concentrations for Runs 1-12 .............. 146
Table 6-12: Experimental and calculated Amm-N concentrations for Runs 1-12 ......... 147
Table 6-13: Experimental and calculated Tot-N concentrations for Runs 1-12 ............ 148
Table 6-14: Adjusted NSD (%) ...................................................................................... 149
Table 7-1: Summary of prototype dimensions ............................................................... 157
Table 7-2: Pollutant removal obtained during Trial 1 (Standard errors shown) ............ 169
Table 7-3: Average pollutant removal rates ................................................................... 175
Table 7-4: Potential removal rates ................................................................................. 175
Table 8-1: Component concentrations varied during dye degradation studies .............. 180
Table 8-2: Average calculated AO7 concentrations in Runs 1-3 (standard errors shown)
........................................................................................................................................ 187
Table 8-3: AO7 removal rates in Runs 1-3 .................................................................... 191
Table 8-4: Comparison of AO7 removal rates in literature ........................................... 193
Table 8-5: COD removal rates in Runs 1 - 3 ................................................................. 194

xv
Abbreviations and Symbols

List of Abbreviations
abs absolute
Amm-N Ammoniacal Nitrogen
AnAOB Anaerobic Ammonia Oxidising Bacteria
AO7 Acid Orange 7
AOB Ammonia oxidising bacteria
AOD Ammoniacal Oxygen Demand
AU Absorbance units
BOD Biochemical Oxygen Demand
CFD Computational Fluid Dynamics
CSTR Continuous Stirred Tank Reactor
COD Chemical Oxygen Demand
DO Dissolved Oxygen
EMBR Extractive membrane biofilm reactor
FISH Fluorescent In-Situ Hybridization
HRT Hydraulic Retention Time
IPS Inclined Plate Settler
lpm litres per minute
MABfR Membrane Aerated Biofilm Reactor
Nit-N Nitrate nitrogen
NIW Northern Ireland Water
NOB Nitrite oxidising bacteria
NTU Nephelometric Turbidity Units
p.e. Population equivalent
PES Polyethersulphone
RBC Rotating Biological Contactor
rpm Revolutions per minute
SBR Sequential Batch Reactor
SR Silicone Rubber

xvi
Tot-N Total nitrogen
TOD Total Oxygen Demand
UWwTD Urban Wastewater Treatment Directive
VOC Volatile Organic Compound
WFD Water Framework Directive
WwT Wastewater Treatment
WwTW Wastewater Treatment Works

List of Symbols
a Specific membrane surface area
ae Cell yield coefficient
A Absorbance
Aλ Absorbance at wavelength λ
Am Membrane area
Am,req Required membrane area
Am,req,T Total required membrane area
b Bacterial cell decay rate
ΔC Concentration difference across membrane
CAO7 Acid Orange 7 concentration
CAO7,inf Influent Acid Orange 7 concentration
CAO7, eff Effluent Acid Orange 7 concentration
Ccon Consent concentration
Ceff Effluent concentration
Cj,exp,i Effluent concentration of component j determined experimentally for
experimental run i
Cj,calc,i Effluent concentration of component j calculated using the model for
experimental run i
Cinf Influent concentration

xvii
de external diameter of membrane fibre
di internal diameter of membrane fibre
DM diffusivity of oxygen in membrane material
Dw diffusivity of oxygen in water
ε membrane porosity
fAH Fraction of oxygen used for aerobic heterotrophy
fd Biodegradable fraction of microorganisms
fe fraction of electron donor used for energy
fnit Fraction of oxygen used for nitrification
fs fraction of electron donor used for cell formation
G Mass velocity
h hour
J Oxygen Flux
Ja Apparent oxygen flux
J Average oxygen flux
K Overall mass transfer coefficient
KL Liquid side mass transfer coefficient
KG Gas side mass transfer coefficient
KM Membrane mass transfer coefficient
L Loading
LAO7,req Required AO7 removal (loading basis)
Leff Effluent loading
LCOD COD loading
LAmm-N Ammoniacal nitrogen loading
LNit-N Nitrate nitrogen loading

m Mass flowrate
M Molar mass
MO2 Molar mass of oxygen
NO3--Navailable Total available nitrate nitrogen
OUR Oxygen uptake rate
OURAH Aerobic heterotrophs oxygen uptake rate

xviii
OURnit Nitrifiers oxygen uptake rate
Q flowrate
r Rate
rAO7 Acid Orange 7 removal rate
rAH Aerobic heterotrophy rate
ramm-N Ammoniacal nitrogen removal rate
rCOD COD removal rate
rden Denitrification rate
rnit Nitrification rate
rtot-N Total nitrogen removal rate
PR Pressure at reference conditions
SM Solubility of oxygen in polymer
T Temperature
TR Temperature at reference conditions
ts Sludge age
τ membrane tortuosity
V Volume
vc critical velocity
wk week
X Biomass density

xix
1 Introduction

This thesis is concerned with the use of membrane aerated biofilms as a means for
treatment of wastewater streams. In particular, it investigates the feasibility of using
such a method as a pre-treatment or polishing step in order to upgrade existing works –
helping meet the demands placed by changing populations and legislation.

This introduction will explain the need for upgrading existing works – justifying the
development of technologies such as that proposed by this project.

1.1 General Introduction

Wastewater is defined as any water which has been tainted by human activity. It is the
combination of liquid and water carried wastes generated by residences, commercial and
industrial buildings; along with stormwater and groundwater which enter the sewerage
system (Metcalf & Eddy, 2003).

Wastewater treatment was first introduced during the second half of the 19th Century, as
early hygienists made the link between outbreaks of cholera in Paris and London and the
open sewers that primarily collected waste. In response, city planners such as
Haussmann and Bazalgette incorporated subterranean sewers into the modernization
plans of Paris and London respectively, although their system served little purpose other
than to divert wastewater away from areas where drinking water was obtained.

Since these early attempts, introduced to protect the health of the rapidly growing urban
populations, the focus of wastewater treatment has changed: the primary objective of
wastewater treatment is now to minimize the effect of mankind’s use of water on the
environment, and to protect it for future generations.

1
1.2 Wastewater treatment
Wastewater treatment consists of a series of physical, chemical and biological
operations, by which solid contaminants are removed and dissolved contaminants are
converted by chemical and/or biological action to immiscible gaseous or solid phases
which are then easily removed.

The majority of dissolved contaminants are easily oxidised by the action of bacteria.
Larger works typically utilise the activated sludge process, whilst smaller facilities use
either trickling filters or rotating biological contactors (RBCs).

The activated sludge process is a suspended growth process, first introduced into the
U.K. at the beginning of the 20th Century. In the process, after primary treatments
(mainly solids removal), the wastewater is mixed with the return activated sludge to
form the mixed liquor – a suspension with typically 1500 – 3500 mg l-1 of biomass as
suspended solids.

After aeration, the mixed liquor flows into settlement tanks, where the velocity is slowed
to an extent where the solid particles drop to the bottom of the tank. The defining
characteristic of the activated sludge process is that a large portion of this biomass is
recycled back to the aeration tank (Figure 1-1); meaning that the mean cell residence
time (average time spent by biomass in the treatment system) is much greater than the
hydraulic retention time (average time spent by liquid in the treatment system) –
allowing the biological activity to be maintained at the level required to achieve full
treatment.

2
 Figure 1-1: Simplified schematic of Activated Sludge Process (adapted from Gray, 2004)

Trickling, or percolating, filters were first introduced in 1893 (Tchobanoglous & Burton,
1991a), and consisted of a bed of porous rock over which wastewater is percolated or
trickled, giving the name (Figure 1-2). An ecosystem develops on the surface of the
rock, consisting of a mixture of heterotrophic and autotrophic bacteria, fungi, algae and
larger organisms such as snails and worms. A modern trickling filter utilizes plastic
packing material in order to maximise the contact area and area available for biofilm
support. The system is passively aerated, with oxygen being transferred into the
wastewater through the voids in the packing material.

Figure 1-2: Trickling filter at NIW Parkgate WwTW

3
Rotating biological contactors (RBCs) were first developed in the 1920s and became
commercially available in 1965 (Gray, 2004). The basic design of an RBC consists of
discs of biofilm support such as PVC, polyethylene or a similar material, which are
mounted on a rotating horizontal shaft, positioned in such a way so as to be
approximately 40% submerged in the liquid. Passive aeration takes place when the
biofilm is above the level of the wastewater.

Figure 1-3: Rotating Biological Contactor (WIS Ltd)

A common feature of the activated sludge process, trickling filters and RBCs is a final
settler in which biomass is removed by the physical process of settling from the effluent
before discharge. Little or no biological treatment takes place in a standard settler, with
the biological treatment limited by the amount of oxygen which can be transferred
during the previous treatment steps.

1.2.1 Wastewater treatment in Northern Ireland

Northern Ireland Water (NIW), which operates as both a government owned company
and non-departmental public body, is responsible for provision of municipal wastewater
treatment. NIW treat approximately 120 million m3 of wastewater each year in 656
WwTW (NI Water, 2013). Due to the non-urban nature of Northern Ireland, the majority
of these works are small, serving a population equivalent (p.e) of less than 250, whilst
4
254 of these works are larger works, serving more than 250 p.e. (Fivelman, 2010).
Population equivalent is the theoretical amount of pollution produced by one person in a
day and is defined by British Water (2013) as 150 litres, containing 60 g of BOD and 8 g
of Ammoniacal Nitrogen.

The larger works mainly achieve treatment through the application of the activated
sludge process, whilst smaller works mainly use either trickling filters or rotating
biological contactors.

1.3 Drivers for upgrade of wastewater treatment works

Northern Ireland has suffered from a historical underinvestment in the wastewater
treatment infrastructure. As a result, at its formation in April 2007, NIW inherited a
treatment works infrastructure which was significantly underperforming in comparison
to those in other parts of the UK. In response to this, NIW launched a £290 million
spending plan on plant upgrade (Fivelman, 2010). There were several drivers for this
upgrade, as described in Sections 1.3.1 and 1.3.2.

1.3.1 Legislative drivers

In Northern Ireland, the discharge of wastewater effluent to any waterway or
underground strata is regulated in accordance with the Water (Northern Ireland) Order
1999 (HMG, 1999). Under these regulations, persons wishing to discharge water must
apply to the Northern Ireland Environment Agency (NIEA) for permission to do so.

A successful applicant will be issued a ‘Consent to Discharge’, which includes

conditions relating to the quality and quantity of effluent which can be discharged.
Similar systems operate in other parts of the U.K.

Three pieces of European legislation have had an effect on the values of these discharge
consents:

5
 a) The Urban Wastewater Treatment Directive
b) The Water Framework Directive
c) The Nitrates Directive

a) Urban Waste Water Treatment Directive

The Urban Waste Water Treatment Directive (UWwTD) (European Council, 1991a)
aimed to protect the environment from the adverse effects of wastewater discharges;
effectively by mandating the introduction of wastewater collection and setting a
minimum standard for the treatment of this wastewater.

The minimum standards set are a function of both the population equivalent (p.e.) of the
wastewater source and the status of the receiving waters, as summarised in Table 1-1.
The terms population equivalent and the various levels of wastewater treatment are
discussed more fully in Chapter 2.

Table 1-1: Level of Treatment mandated by the UWwTD (adapted from DEFRA, 2002)
Treatment Process Discharge Area
Preliminary Screening of large solids Fresh waters <2,000 p.e.
Grit removal Coastal waster <10,000 p.e.
Primary Settlement of Suspended Solids Coastal waters >10,000 p.e. in
less Sensitive Areas
Secondary Biological Treatment Fresh waters >2,000 p.e.
Coastal waters >10,000 p.e.
Tertiary Various Methods >10,000 p.e. to Sensitive areas

In Northern Ireland, the requirements of the UWwTD were transposed into The Urban
Waste Water Treatment Regulations (Northern Ireland) 1995 (HMG, 1995). These
regulations placed tighter numerical consents upon Northern Ireland Water (and their

6
predecessor DRD Water Service) and also provided a framework for measurement of
compliance (NIEA, 2014).

b) Water Framework Directive

The Water Framework Direction (WFD) (European Council, 2000), due to be fully
implemented by 2015, places the demand that all inland, coastal and maritime water
courses achieve ‘good’ ecological status. Good status is assessed using a combination of
biological, hydrological, physical parameters.

Transposed into The Water Environment (Water Framework Directive) Regulations

(Northern Ireland) (HMG, 2003), the WFD has led to a series of plans for river basin
management, taking a holistic view of all activities which affect the aquatic environment
(NIEA, 2009).

c) Nitrates Directive
Concern over the widespread eutrophication of water bodies throughout Europe lead to
the publication of the Nitrates Directive (European Council, 1991b). Eutrophication is
the over fertilisation of lakes and rivers, from sources such as farming, sewage and
industry, causing an accelerated growth of algae and other plants (DOE NI & DARD NI,
2004). This growth can form a barrier to oxygen transfer on the surface of waters, and
lead to detrimental effects on biodiversity.

Locally, implementation of the Nitrates Directive has focussed on agriculture (DARD &
DOE, 2010), but has also had an effect on wastewater, with the introduction of
numerical consents on the total nitrogen and total phosphorous concentrations of
wastewater effluents.

7
1.3.2 Other drivers for wastewater treatment upgrade
Wastewater treatment infrastructure is a significant consumer of electrical energy
through aeration and pumping equipment (Kadar & Siboni, 1998). As this leads to
sizeable operating costs and associated operational carbon emissions, wastewater
treatment companies are now pursuing more efficient water and wastewater treatment
technologies (Smyth et al., 2013).

1.4 Proposed technology – The BioSettlerTM

As previously discussed, pressures caused by the tightening of discharge consents due to
legislative changes and increased loading due to population growth mean that many
WwTW struggle to meet consents.

In a wastewater treatment works using the activated sludge process, the area used by
tanks providing biological treatment is relatively small to the area required by the tanks
providing final settlement. Figure 1-4 shows an aerial of Belfast WwTW; the main
treatment works for the city of Belfast designed to serve a population equivalent of
400,000. The footprint of the settlement tanks (red box) is approximately 3 times that of
the aeration lanes (green box). As previously discussed, settling tanks are also included
in WwTW operating using either trickling filters or RBCs.

Settlement

Treatment

Figure 1-4: Aerial view of Belfast WwTW, highlighting the relative sizes of aerobic treatment and
settlement tanks (Adapted from GoogleMaps)

The solids removal potential of final settlers can be boosted by the installation of an
array of inclined plates. These inclined plates reduce the vertical distance a solid particle

8
needs to fall to be removed from suspension and increases the available settling area,
boosting solids removal rates (Figure 1-5). The concept and history of inclined plate
settlers is discussed more fully in Chapter 2.

Figure 1-5: Inclined plate showing projected settling area and footprint (adapted from Metso
Minerals, 2006)

However, in order for the area occupied by the final settling tanks to be utilised for
biological treatment, oxygen must be introduced without bubbles as in the aeration
tanks, as the turbulence associated with bubbles interferes with the settlement process.

One possible way in which oxygen can be introduced without bubbles is through the use
of membranes. If a suitable tubular membrane is placed in a liquid stream and filled with
compressed air; oxygen will diffuse from the inside of the membrane (where it is in high
concentration), through the structure of the membrane material and into the liquid stream
(where it is in low concentration). If the air pressure is maintained below a critical point,
this process will occur without bubble formation.

If such a membrane is placed in a wastewater stream, a biofilm quickly forms upon the
membrane surface (Figure 1-6). A biofilm is simply a group of microorganism cells
grouped together on a surface like mould in a shower or plaque on teeth.

9
This biofilm consists of a mixture of heterotrophic bacteria, which utilise oxygen to
remove organic carbon compounds, and nitrifying bacteria, which utilise oxygen to
convert ammonia to nitrate. If the wastewater is of a significant strength, the biofilm will
consume all of the oxygen supplied to it and an anoxic layer will form in the area closest
to the bulk liquid. Denitrifying bacteria will occupy this area, meaning that such a
system can provide additional BOD, ammoniacal and total nitrogen removal.

Figure 1-6: Structure of membrane aerated biofilms

This thesis introduces the BioSettlerTM, a novel for wastewater treatment which has been
developed as described here, and has been patented (Groom et al., 2009). The BioSettler
combines the two existing technologies of membrane aerated biofilms and inclined plate
settlers, by incorporating membrane aeration (with associated biofilm) on the underside
of an inclined plate (Figure 1-7).

10
 Figure 1-7: Membrane location in the BioSettler

Although established in their own right, this project is believed to be unique in

combining the two systems into a single technology.

Employing such a system in wastewater treatments would yield two major advantages:
the technology would lead to increases in the potential of final settling tank to remove
solids and provide additional BOD, ammoniacal and total nitrogen removal.

Also, as the technology would be available as a retrofit package, it would be a

significantly less expensive option for upgrade of wastewater treatment (Rubino, 1996).
Additionally, this upgrade would be achieved without additional tanks being needed,
making the technology suitable for use in situations where footprint is limited.

1.5 Aims and Objectives

This work aims to:
 Investigate the various factors that affect the bubbleless transfer of oxygen to
water, specifically:
o Membrane type;
o Inlet air pressure;

11
 o Air flowrate;
o Water side turbulence.
 Develop an understanding of the treatment of municipal wastewater in the
MABfR, particularly of wastewater containing pollutant concentrations found in
non-consent meeting secondary effluent. This will involve:
o Investigation of the effect of inlet air pressure on pollutant removal;
o Investigation of the effect of variation in wastewater composition on
pollutant removal;
o Model development to allow prediction of MABfR performance.
 Explore the use of the membrane aerated biofilm for the treatment of industrial
wastewater, especially those originating from dye houses and containing azo
dyes.
 Demonstrate the BioSettler concept with a pilot scale plant at a municipal
WwTW.

1.6 Conclusions
Increasingly more demanding legislation governing the discharge of wastewater
effluent, coupled with pressures from increasing and transient populations mean that
upgrade of existing wastewater treatment infrastructure is required throughout the
developed world.

Additionally, increasing environmental awareness and energy costs necessitate the

development of wastewater treatment technologies which are more energy efficient that
the current state of the art they will replace whilst matching or even exceeding the
effectiveness of the treatment that they provide.

The use of Membrane Aerated Biofilms is one approach that is worthy of exploration in
an attempt to meet these goals. When utilised as part of the BioSettler technology,
membrane aerated biofilms can provide aerobic and anoxic conditions simultaneously
and in the same tank, allowing a variety of wastewater pollutants to be mineralised. This
thesis will explore the practicableness of this technology for wastewater treatment.

12
2 Review of wastewater sources, composition and treatment
options

2.1 Wastewater and wastewater treatment

Wastewater is defined as any water which has been tainted by human activity. It is the
combination of liquid and water carried wastes generated by residences, commercial and
industrial buildings; along with stormwater and groundwater which enter the sewerage
system (Tchobanoglous & Burton, 1991b).

If allowed to accumulate without treatment, several problems are caused. The organic
molecules contained in this tainted water (such as sugars, fats, proteins) will be acted
upon by microorganisms, consuming all available dissolved oxygen so that it is no
longer available for fish and other aquatic organisms and producing unpleasant odours
(e.g. hydrogen sulphide).

Additionally, wastewater may contain nutrients, which cause the excess growth of
aquatic plants; mutagenic and carcinogen compounds and pathogenic microorganisms
that originate in the digestion systems of humans and other domestic animals.

In order to protect human health and the environment from these threats, the discipline
of wastewater engineering has developed. The discipline, which involves chemistry,
biology, civil and chemical engineering, concerns itself with all aspects of the
wastewater infrastructure, from collection at domestic dwellings and industrial premises
where it is generated to its treatment and subsequent disposal or reuse.

2.1.1 Sources of wastewater

Almost every form of human activity generates wastewater. The composition and
flowrate will vary greatly over time and as a function of the activity that produces it.

13
This thesis focuses on a solution for the need for upgrade of current wastewater
treatment techniques and facilities, with an emphasis on two particular types of waste:
(i) Municipal wastewater;
(ii) Dye house wastewater.

2.1.1.1 Municipal wastewater

The term municipal wastewater refers to that wastewater which is produced from
domestic residences and commercial properties (such as restaurants, shops and offices).
Along with breakdown products of faecal matter and urine, municipal wastewater will
contain residues of food and other solid materials, laundry detergents and other cleaning
chemicals.

A typical analysis of untreated municipal wastewater is given in Table 2-1 (adapted from
Tchobanoglous & Burton, 1991b). The various parameters used for characterisation of
wastewater are discussed in more detail in Section 2.1.2.

Table 2-1 : Typical composition of untreated municipal wastewater

Concentration (mg l-1)
Contaminant Weak Medium Strong
Solids (total) 350 720 1200
BOD5 110 220 400
COD 250 500 1000
Nitrogen (Total as N) 20 40 85
Phosphorous (Total as P) 4 8 15
Grease 50 100 150

Municipal wastewater typically follows a diurnal pattern, with small volumes of

wastewater reaching the treatment works at night and large flows in the morning and
early evening (Healy & O'Flynn, 2011). Similar variations are seen in the daily variation
of wastewater strength, as illustrated in Figure 2-1. In times of heavy rainfall, especially
in areas where stormwater is also carried by the foul sewer, volume of wastewater

14
entering wastewater treatment units increases significantly, with an associated dilution
effect which reduces wastewater strength.

Figure 2-1: Diurnal variations in wastewater flowrate and strength for a typical WwTW
(Tchobanoglous & Burton, 1991b)

To aid with the design of wastewater treatment facilities, the strength and volumes of
wastewater produced by different activities are grouped into a notional unit called
population equivalent (p.e.). This corresponds to the average volume and strength of
wastewater produced per person at a typical domestic dwelling and currently is defined
at 150 litres of wastewater, containing 60 g of Biochemical Oxygen Demand and 8 g of
ammoniacal nitrogen as N (British Water, 2013).

2.1.1.2 Dye house wastewater

The process by which natural fibres are coloured is inherently inefficient, with 4 -12% of
dyes lost to wastewater during textile processing (Coughlin et al., 2002). In addition to
unused dye molecules, dye house effluent will also contain dissolved metallic species,
residues of adhesives chemicals used to adhere carpet to backing material and chemicals
protect which the product from attack by microorganisms and insects (Wilkinson 2007).

15
 Table 2-2: Typical analysis of Dye house wastewater (Wilkinson 2007)
Parameter Value
Chemical Oxygen Demand 1500 – 2000 mg l-1
Unused Dye 20 – 50 mg l-1
pH 3.5 – 7.0
Suspended Solids 5 - 20 mg l-1

Azo dyes; characterized by the presence of one or more azo bridges, nitrogen-nitrogen
double bonds (−𝑁 = 𝑁−) (Van der Zee et al. 2003a), are the class of dye used most
commonly industrially, accounting for approximately 70% of all dyestuffs used
(Coughlin et al., 2002). Although not widely used by commercial dye houses (Wilkinson
2007), the azo dye most commonly used in degradation studies in the literature is 1-
Phenylazo-2-naphthol-4’-sulfonic acid, commonly known as Acid Orange 7 (AO7) or
Orange II. The structure of AO7 is shown in Figure 2-2 (Fernandes et al., 2004) .

Figure 2-2: Structure of Acid Orange 7

In 2007, the world production of azo dyes was estimated at 500,000 tonnes (Pandey et
al., 2007). The production and dying processes are inherently inefficient, meaning at
least 4% of the produced dyes are wasted, and ends up in domestic and industrial
wastewater streams (Coughlin et al., 2002). Although this results in significantly less
volumes of wastewater than that produced by municipalities, the coloured nature gives
azo dye effluent a significant impact on the general public.

The molecular structure of azo dyes makes them resistant to fading by exposure to
sweat, soap, water, light and oxidizing agents (Davies et al., 2006). Whilst this makes

16
them ideal for use as a dye, it makes them resistant to aerobic degradation in the
activated sludge process – dye removal (if any) takes place via adsorption of azo dye
molecules onto the settled activated sludge (Coughlin et al., 2003).

2.1.2 Wastewater characterisation

Wastewater is characterised by the type and concentration of pollutants that it contains.
The pollutants of interest in this study are listed and described in Sections 2.1.2.1 -
2.1.2.4 below. The methods by which they are obtained are detailed in Chapter 3.

2.1.2.1 Biochemical oxygen demand

Wastewater typically contains a mixture of organic compounds which are oxidised by
the action of microorganisms. Identification and quantification of each individual
organic compound is a very onerous task, and is therefore of little use to the wastewater
industry.

Biochemical Oxygen Demand (BOD) is a parameter which, instead of quantifying the

concentration of individual pollutants present, measures the amount of oxygen which is
required by bacteria to oxidise them in a defined time period. A 5 day time period is
most commonly used; but longer time periods are sometimes employed for practical
reasons (e.g. laboratory schedules) or in wastewaters containing compounds which
undergo very slow hydrolysis processes (Tchobanoglous & Burton, 1991b).

2.1.2.2 Chemical Oxygen Demand

Chemical Oxygen Demand (COD) is another measure of organic matter in wastewater
that can be oxidised. Rather than a time consuming biological determination, the COD
test chemically oxidises the wastewater using a reagent such as an acidic dichromate
solution, as described by Equation 2-1:

17
 𝐶𝑛 𝐻𝑎 𝑂𝑏 𝑁𝑐 + 𝑑𝐶𝑟2 𝑂72− + (8𝑑 + 𝑐)𝐻 +
Equation 2-1
𝑎 + 8𝑑 − 3𝑐
→ 𝑛𝐶𝑂2 + 𝐻2 𝑂 + 𝑐𝑁𝐻4+ + 2𝑑𝐶𝑟 3+
2

The amount of COD present can then be related to the amount of chromate ion reduced
to Cr(III), as ascertained by spectroscopic measurement of colour change from orange to
green.

The value of COD is typically higher than that of BOD, as wastewaters commonly
contain organic substances which can be oxidised chemically but not biologically (e.g.
large molecules such as protein chains or lignin); inorganic substances are also oxidised
by chromate and certain organic substances can be toxic to the microorganisms used in
the BOD test.

Despite these drawbacks, the COD test takes considerably less time than the BOD test
(approximately 2 hours versus 5 days), and as such is used. Oftentimes, where the
composition of wastewater is relatively consistent, a steady ratio of BOD:COD can be
established, and COD used as an estimator for BOD.

2.1.2.3 Ammoniacal nitrogen

Ammoniacal nitrogen (a term that describes nitrogen in both the NH3 and NH4+ forms) is
a characteristic breakdown product of organic matter and urea. It is the most common
nitrogenous pollutant contained in wastewater, with an estimated 100,000 tonnes of
ammoniacal nitrogen being discharged to inland waters in the UK each year (DEFRA
2002).

The presence of ammoniacal nitrogen in wastewater presents a number of problems. At

high levels (~35 mg l-1 and greater) the distinctive strong unpleasant odour is detectable
by the human nose. The un-ionised form, NH3, can cause pH shifts and changes in the
solubility of toxic substances, adversely affected receiving waters. Ammoniacal nitrogen

18
will also undergo nitrification (biological conversion to nitrate), causing depletion of
dissolved oxygen in receiving water.

Exposures of 24 hours to ammoniacal nitrogen concentrations of as low as 0.2 mg l-1

have been reported as being toxic to fish (Woods 2003). As a result environmental
protection agencies such as the Northern Ireland Environmental Agency, who are
responsible the regulation of wastewater discharges, place stringent limits on those
discharges in areas which serve as habitat for fish (Stewart 2014).

2.1.2.4 Nitrate
Nitrate (NO3-) is the product of the biological oxidation of ammonia, is therefore
commonly found in wastewater. Although it is preferable to discharge nitrate nitrogen
rather than ammoniacal nitrogen to receiving waters as nitrate does not deplete the
dissolved oxygen, nitrate is still a pollutant (Grady et al. 1999).

Nitrate nitrogen can be converted by plants and algae into organic matter. When nitrate
is present in high concentrations, this can lead to the growth of ‘algal blooms’, which
form a blanket on the surface of the receiving water, which, in addition to appearing
unnatural and unsightly, can alter the temperature, light levels and oxygen availability,
leading to adverse effects on the aquatic ecosystem.

High nitrate nitrogen concentration in drinking water has been linked to ‘blue-baby
syndrome’, where nitrate interferes with the oxygen carrying capacity of haemoglobin in
the blood, potentially leading to death (US EPA, 2012).

2.1.2.5 Total Suspended solids

Wastewater contains a variety of solid materials ranging from large inorganic material
such as stones and rags to microscopic bacterial cells, which must all be removed from
the wastewater stream before discharge. Larger particles can cause blocking of channels

19
of water courses with associated flooding; whilst smaller particles can reduce the
sunlight availability for aquatic plants and affect the temperature of water courses.

Solid materials are removed from wastewater based on physical size and difference in
density. For example, wastewater is typically passed through a 6 mm screen on arrival at
a treatment works, whilst solid particle such as activated sludge flocs are removed by
providing quiescent conditions are sufficient residence time for them to settle to the
bottom of clarifier tanks.

2.1.3 Current wastewater treatment technologies

Wastewater treatment falls into three categories: physical, chemical and biological, with
the majority of wastewater treatment processes being a combination of the three
different categories used in sequence.

For example, a typical municipal wastewater treatment works may consist of the
following steps: screening (physical); grit removal (physical); coagulation and
flocculation (chemical and physical); primary settling (physical); activated sludge
(biological) and final setting (physical).

In addition to the biological processes explained in Sections 2.2.1 - 2.2.3, knowledge of

two physical unit operation technologies is essential for understanding of this thesis.
These technologies are (i) inclined plate settlers and (ii) membranes.

2.1.3.1 Inclined plate settlers

Analysis of the mechanisms by which settling occurs in both primary and secondary
settling tanks in wastewater treatment plants reveals that the flowrate of wastewater
through the settling tank that can be settled is proportional to its area and a critical
velocity, vc, as shown in Equation 2-2.
𝑄 = 𝐴𝑣𝑐 Equation 2-2

20
This equation implies that the greatest volume of wastewater could theoretically be
settled by a tank approaching infinite surface area. Although correct theoretically, this
approach is impractical, as the demand on plant footprint would be too great
(Tchobanoglous et al., 2004). A practical solution to increasing available settling surface
is to fit settling tanks with an array of overlapping inclined plates as shown in Figure 2-3
and Figure 2-4.

Figure 2-3: Schematic Inclined Plate Settler Figure 2-4: Inclined Plates
(Parkson, 2010)

These systems are called inclined plates settlers and can provide up to 10 m 2 for every
m2 of footprint (Parkson, 2010). Influent enters the plates through the side, with settled
solids sliding down the upper plate surface to be collected at the bottom, whilst clarified
effluent rises up the back of the inclined plate and is collected through an overflow weir
at the top in the same way as in a standard settling tank.

2.1.3.2 Membranes in wastewater treatment

The word membrane is used to describe a thin interface that acts as a selective barrier
which can be used to control the permeation rate of chemical species which come into
contact with it (Baker et al., 1991). Membranes affect separation through differences in

21
solubility and diffusivity or molecular size. Membranes occur in nature in biological
cells, and artificial membranes mimicking this phenomenon are commonly used for a
variety of purposes including controlled drug release, gas separations, electronic
applications and wastewater treatment.

Membranes are characterised by their composition and structure. The membranes of

interest in this thesis are categorised as being either nonporous dense membranes or
microporous membranes.

Dense membranes consist of dense, uniform, structures, through which permeants are
transported by diffusion under the driving force of a pressure, concentration or electrical
potential gradient. The separation of components by the membrane is related to their
transport rate in the membrane material, which is a function of their diffusivity and
solubility in the membrane material. Dense materials are commonly used in gas
separation applications and are formed from materials such as silicone rubber and
polytetrafluoroethylene.

Figure 2-5: Molecular transport in dense membranes (left) by differences in permeation and
microporous membranes by molecular filtration

22
Microporous membranes have structures which are similar to conventional filters,
having a rigid, highly voided structure with a random distribution of interconnected
pores (Baker et al., 1991). The pores are smaller than in conventional filters, ranging
from approximately 0.01 to 10 µm in diameter. Microporous membranes are further
differentiated from each other based on the size of particles or molecules they retain, as
illustrated in Figure 2-6.

Figure 2-6: Membrane classification (Radcliff & Zarnadze, 2004)

Molecules larger than the larger pores will be completely rejected by the membrane.
Molecules smaller than the largest pores, but larger than the smallest pores will be
partially rejected in relation to the pore size distribution. In general, only particles with
significant difference in molecular size can be separated using membranes.

Additionally, microporous membranes can be either hydrophobic or hydrophilic. In

hydrophilic membranes, the membrane pores fill with liquid whilst in hydrophobic
membranes; the pores remain gas-filled during operation. Hydrophilic membranes are
generally preferable for liquid-phase filtrations, whilst hydrophobic are more suited to

23
gas transfer operations due to the higher diffusivities of gases through other gases
compared to liquids.

2.1.4 Biofilms
The term biofilm is used to describe a colony of bacterial cells, held together in a matrix
of extra-cellular polymeric material, produced by the organisms themselves (Madigan &
Martinko, 2006). Biofilms were traditionally thought of as being associated with a
surface, but recently the term biofilm has been expanded to include granular sludge, with
the defining characteristic being the existence of substrate gradients (Morgenroth, 2008).

Biofilm is the most common form of bacterial life with mould grown in showers, plaque
on teeth being common examples (Madigan & Martinko, 2006). Four reasons have been
identified as to why biofilms form:
(a) Safety in numbers;
(b) Allows cells to remain in a favourable niche (e.g. close to a source of substrate);
(c) Allows bacterial cells to exist close together, facilitating intercellular
communication;
(d) It is the default way in which bacterial cells grow (Madigan & Martinko, 2006).

Significant volumes of research are concerned with the prevention of biofilm formation,
especially in medical applications, where biofilm formation can lead to the spread of
infection (Monroe, 2007), or in membrane separation processes, where biofilm
formation can lead to reduction in permeate flux (Dreszer et al., 2014).

2.1.4.1 Biofilms in wastewater

Due to their high biomass retention, biofilm systems are very suitable for bacterial
processes with involve slow growth rates (e.g. those for ammonia removal), and have
been in use since the early wastewater systems developed by Victorian engineers.

24
Nowadays, alongside the two well established wastewater treatment technologies
introduced in Chapter 1, there are two novel biofilm technologies which have yet to be
adopted by the wastewater treatment industry, but have been the subject of significant
published research, namely Extractive Membrane Biofilm Reactors (EMBRs) (e.g.
Livingston et al., 1998) and Membrane Aerated Biofilm Reactors (MABfRs) (e.g.
Stephenson et al., 2000).

Extractive membrane biofilm reactors are a relatively novel wastewater technology

which combines aspects of the trickling filter and separation membranes, and is used in
wastewater treatment as a replacement for settling (Stephenson et al., 2000). In an
EMBR, the biofilm acts as part of the separation membrane, adding biological treatment
as wastewater moves through it (Livingston et al., 1998).

This is particularly useful in the treatment of those wastewaters containing volatile

organic compounds (VOCs), as the biofilm separates the VOCs from the aerated
compartment of the bioreactor, preventing air stripping of VOCs to the atmosphere
(Pavasant et al., 1996).

2.1.4.2 Membrane attached biofilms

Membrane aerated biofilm reactors (MABfRs) are those systems in which dense or
microporous gas transfer membranes are used to transfer oxygen to bacteria present on
the membrane surface without the formation of bubbles (Stephenson et al., 2000). The
biofilm is attached to the membrane surface and wastewater is present on the outer
surface of the biofilm so that counter diffusion of oxygen and substrate occurs into the
biofilm.

25
 Figure 2-7: General structure of Membrane Aerated Biofilms

MABfRs can be operated with either pure oxygen or compressed air as the aeration gas,
with the economics of the system favouring compressed air due to the high cost of pure
oxygen production. Various different membrane types and arrangements are used as
discussed further in Section 2.4.

2.2 Biology of wastewater treatment

Aerobic (in the presence of elemental oxygen), anoxic (without elemental oxygen) and
anaerobic (in absence of oxygen) biological processes have historically been used for
wastewater treatment.

Due to their low energy usage and the possibility of generating biogas, anaerobic
processes are of increasing interest. This biogas can then undergo a combustion process
and be used to generate renewable heat and electricity.

26
However, the types of wastewater streams which are of sufficient strength to generate
significant volumes of biogas are limited – especially in Northern Ireland where sewers
typically carry both sewage and storm water. Additionally, there are associated problems
with odour nuisance caused by the co-production of hydrogen sulphide gas, with
associated opposition from those living in the areas surrounding treatment works.

In Scotland, for example, legislation has been introduced preventing the release of
odours from wastewater treatment works and other industrial sources. In response to
this, water companies had been forced to place covers on some wastewater treatment
units to prevent foul odours being released into the air.

Three biological processes are of interesting in this study

(i) Aerobic heterotrophy;
(ii) Nitrification;
(iii) Denitrification.

Aerobic heterotrophy and nitrification are aerobic processes, whilst the process of
denitrification takes place in anoxic conditions. These three processes are summarized in
Sections 2.2.1 - 2.2.3 below.

2.2.1 Aerobic heterotrophy

Heterotroph is the name given to those microorganisms who obtain their energy through
the oxidation of organic matter (Bitton, 2005). The process is described as aerobic when
the oxidation utilises elemental oxygen and can be described by the half reaction shown
as Equation 2-3, where organic matter is represented by {CH2O}, a theoretical molecule
of COD (McCarty, 1975).

{𝐶𝐻2 𝑂} + 𝑂2 → 𝐶𝑂2 + 𝐻2 𝑂 Equation 2-3

27
In addition to obtaining energy, heterotrophs use also organic carbon for cell synthesis,
and as organic matter is the most common dissolved pollutant in wastewater,
heterotrophs dominate wastewater treatment systems (Grady et al., 1999). Pseudomonas,
an extensively studied genus of bacteria due to their prevalence as an opportunistic
pathogen in humans, is the most commonly found in wastewater (Bitton, 2005). A
typical analysis of activated sludge is shown in Table 2-3.

Table 2-3: Typical distribution of Aerobic Heterotrophic Bacteria in Activated Sludge (Bitton, 2005)
Genus or Group % Total Isolates
Comamonas-Psuedomonas 50.0
Alcaligenes 5.8
Pseudomonas (fluorescent) 1.9
Paracoccus 11.5
Unidentified (gram-negative rods) 1.9
Aeromonas 1.9
Flavobacterium-Cytophaga 13.5
Bacillus 1.9
Micrococcus 1.9
Coryneform 5.8
Arthrobacter 1.9
Aureobacterium-Microbacterium 1.9

2.2.2 Nitrification
Nitrification is a two stage process by which ammoniacal nitrogen is converted to
nitrate-nitrogen by the action of autotrophic bacteria, and is the most common method
for the removal of ammoniacal nitrogen in traditional wastewater treatment.

It is a two stage process, with the two separate stages being carried out by different
groups of bacteria. Both stages are carried out by autotrophic bacteria, meaning than no
organic carbon is involved in the process; carbon dioxide is instead used as a carbon
source for cell synthesis.

28
First ammonia is converted to nitrite – a process carried out by bacteria known as
nitrosofyers or ammonia oxidizing bacteria (AOB). Nitrosomonas is the most common
AOB found in WwTW, but many other genera have been identified as being able to
carry out this stage including Nitrosococcus, Nitrosospira and Nitrosocystis.

The conversion proceeds via two steps as shown by the half equations below:

Equation 2-4
NH 3  O2  2e   NH 2 OH  H 2 O

NH 2 OH  H 2 O  1
O2  NO2  2H 2 O  H 
2
Equation 2-5

The second stage involves the conversion of the produced nitrite to nitrate. This step is
known as ‘true’ nitrification and the bacteria that carry it out are known as nitrifiers
(nitrate producers). Nitrobacter is the most common nitrifier found in WwTW, though
Nitrococcus, Nitrospira and Nitrocystis have also been isolated.

The conversion occurs in accordance with the half equation given in Equation 2-6:
NO2  1
2 O2  NO3 Equation 2-6

Under steady state conditions, nitrite is not accumulated. As such, it is an accepted

assumption to condition the action of AOB as the rate determining step (Gray, 2004).
The detection of significant amounts of nitrite is evidential to inhibition of the second
‘true’ nitrification step.

Nitrification is highly dependent on wastewater temperature, with a drop in temperature

from 20°C to 15°C resulting in a drop in nitrification effectiveness of 35%
(Komorowska-Kaufman et al., 2006).

29
2.2.3 Denitrification
Denitrification is the biological process by which microorganisms convert nitrite and
nitrate to elemental nitrogen. It is an anoxic process, meaning that it takes place in the
absence of elemental oxygen; and heterotrophic processes, requiring organic carbon for
cell synthesis and energy generation (Dincer & Kargi, 2000).

The process proceeds via the pathway shown in Figure 2-8 below, with both nitrate and
nitrite being acceptable starter species.

Figure 2-8: Steps in the reduction of nitrate (adapted from Madigan & Martinko, 2006)

Denitrification is carried out by a large range of microorganisms including Escherichia

coli and Pseudomonas sp. These microorganisms are facultative aerobes; they carry out
aerobic respiration when oxygen is present as an electron donor, and utilise other
chemical species (such as nitrate) when it is not. Use of oxygen as an electron donor is
preferable, as in this way the microorganisms gain most energy (Madigan & Martinko,
2006).

Reduction potential (redox potential) is a useful parameter for predicting which chemical
species are used as an electron donor in biological processes. The redox potentials most

30
suited to different biological processes are illustrated in Figure 2-9. Anoxic processes,
such as denitrification, take place at a redox potential of approximately +400 mV.
Aerobic processes take place at higher redox potentials, with anaerobic processes taking
place at negative redox potentials.

Figure 2-9: Examples of microbial respiration and associated redox potentials (adapted from
Madigan & Martinko, 2006)

2.2.4 Stoichiometry of wastewater biology

McCarty (1975) published a method for developing a stoichiometric equation for
reactions mediated by bacteria in wastewater. The method involves the combination of
three half reactions: one for the oxidation of the electron donor (Rd); one for the
reduction of the electron acceptor (Ra); one for the assimilation of new bacterial cells
(Rc). The overall R is obtained as described in Equation 2-7.

31
 R  Rd  f e Ra  f s Rc Equation 2-7

Where: fe = fraction of electron donor used for energy

fs = fraction of electron donor used for cell formation

In order to achieve a balanced equation:

fe  fs  1 Equation 2-8

fs and fe are functions of cell yield coefficient, cell decay rate, solids retention time
(sludge age) and the biodegradable fraction of microorganisms as described by the
relationship in Equation 5-3.
 f bt 
f s  ae 1  d s  Equation 2-9
 1  bt s 
Where: ae = cell yield coefficient
fd = biodegradable fraction of active microorganism
b = cell decay rate (day-1)
ts = solid retention time (days)

The method also presents the half equations for a variety of different nitrogen sources,
electron acceptors and electron donors which can be combined as required. Values for a e
for a selection of common electron donor/electron acceptor pairs are also reported.

The construction of such stoichiometric equations for the various microbial reactions
taking place in wastewater is extremely useful as it allows the establishment of a mass
balance of the various inputs to the system. As such, this approach has been employed in
many modelling studies, including those that investigated pollutant removal with
membrane attached biofilms (Ergas & Reuss, 2001, Shanahan & Semmens, 2004).

2.3 Bubbleless Aeration

Bubbleless or bubble-free aeration is achieved by placing a thin film of synthetic
membrane between the gas and liquid phases. Oxygen is transported through the
32
membrane material by virtue of a concentration gradient, and dissolves directly from the
surface of the membrane into the liquid phase (Côté et al., 1989).

2.3.1 Historical Context

One of the earliest examples of relevant published research is that conducted by Robb
(1968), in which the transfer of oxygen to water, through various polymer membranes,
was investigated. In a study of 27 polymeric materials, including polystyrene,
polyethylene, polyvinylchloride and polyamide-6, silicone rubber was found to perform
best. The study also suggests possible uses for a silicone rubber membrane based on the
permeability: as a membrane oxygen enricher, as an air regenerator for possible use in a
nuclear fallout situation and to extract oxygen from fresh water; but did not suggest any
uses of silicone membranes which have come into common use.

A later study by Yasuda & Lamaze (1972) found, that with hydrophobic porous
membranes, the rate of oxygen transfer into water was controlled by the liquid boundary
layer – the resistance of which can be greatly reduced by operation in the turbulent
regime.

Recent works have utilised a variety of materials for gas transfer into water-based
solutions/mixtures, for a variety of purposes. Most commonly, these materials have been
polyethylene (PE) (Brindle et al., 1998), polypropylene (PP) (Ahmed et al., 1996)
and polytetrafluoroethylene (PTFE) (Schneider et al., 1995), but less common materials
such as silicone rubber (Coté et al., 1989) and Gore-Tex® (Timberlake et al., 1988)
have also been utilised. In some studies, composite membranes have been used (e.g.
Ahmed et al., 2004) – combining gas transfer properties of one material with the
robustness of another – allowing higher pressures to be used and therefore larger gas
fluxes to be achieved.

With microporous membranes, i.e. those constructed from materials such as PP and PE,
the formation of bubbles restricts operation to low pressures. As transmembrane

33
concentration difference (a function of transmembrane pressure) is the driving force for
mass transfer through the membrane, the rate of mass transfer is therefore limited. Dense
membranes, constructed of materials such as silicone rubber, can be operated at higher
pressures without bubble formation occurring, leading to greater oxygen fluxes.

2.3.2 Mass transfer mechanism

Membrane materials are categorised into microporous and dense membranes based on
the mechanism of gas transfer. In dense membranes, oxygen is transported from the gas
side to the water side via a solution-diffusion mechanism; oxygen is absorbed into the
polymer on the gas and is transported by diffusion through the membrane wall to the
interface with the liquid. There it then dissolves into the liquid phase and is transported
away from the membrane surface via diffusion.

When calculating the total mass transfer resistance (1/K), the gas mass transfer
resistance can be considered to be negligible, meaning it can be expressed as the sum of
the liquid film resistance (1/KL) and the membrane resistance (1/KM):

1 1 1
  Equation 2-10
K KM KL

where: K = overall mass transfer coefficient (ms-1);

KM = membrane mass transfer coefficient (ms-1);

KL = liquid side mass transfer coefficient (ms-1).

The liquid film mass transfer resistance is a function of hydrodynamic variables and can
be estimated for a given design through empirical mass transfer correlations.

34
The estimation of the membrane resistance is different for dense and porous membranes.
For the case of a dense polymer membrane, when the gas solubility in the polymer can
be represented by a linear isotherm and the diffusion coefficient in the membrane is
constant, the mass transfer resistance through the membrane can be expressed as
Equation 2-11

1 l
 Equation 2-11
KM SMDM

where: l = membrane thickness (m);

SM = solubility coefficient of the gas in the polymer (m3m-2);

DM = diffusion coefficient of the gas in the polymer (m2s-1).

Much more complex mechanisms exist for the transfer of gas through a porous
membrane. If the total gas pressure is maintained below the bubble point of the porous
membrane, there should be essentially no total pressure difference across the membrane
and transport should take place via diffusion through membrane pores- especially true
for low solubility gases such as oxygen, nitrogen and carbon dioxide (Yang & Cussler,
1986). If the membrane is made from a hydrophobic material, the pores remain gas
filled, and the oxygen transfer occurs via a gas-gas diffusion or Knudsen flow
mechanism, depending upon parameters such as membrane morphology, the nature of
the gas mixture, and the total gas pressure. The mass transfer resistance of the
membrane (1/KM) in this situation is normally considered negligibly small compared to
the liquid film resistance (1/KL), as first suggested by Yasuda & Lamaze (1972).

2.3.3 Modes of operation

Literature contains details of two different modes of membrane operation – either in the
form of a flat sheet or hollow fibre tube. Tubular membranes have the advantage in the
amount of surface area they can provide for mass transfer, especially when combined

35
with others in the form of a fibre bundle. Additionally, hollow fibre membranes can be
operated either in dead end mode, where each individual fibre is sealed, or in flow
through mode (Fang et al., 2004).

Flow-through operation holds two major disadvantages: oxygen transfer efficiencies of

100% can never be achieved and Volatile Organic Compounds (VOCs) will be stripped
to the atmosphere. Dead-end operation can avoid these, but Côté et al. (1989)
recommended avoiding its use as it was compromised by water condensation on the gas
side of the membrane fibre. This was not found with flow-through mode, as any water
vapour entering the fibre was swept away instead of being allowed to accumulate.

More recent work by Fang et al. (2004), reported that condensation is unavoidable
regardless of whether flow-through or dead-end operation is utilised. Due to higher mass
transfer coefficients for water compared to gas, the gas stream will quickly become
saturated with water vapour and condensation will occur. Several solutions were
suggested to overcome this problem, including incorporation of sections of hydrophilic
microporous material at the effluent end of the fibres. This will allow condensate to
return to the external solution, provided internal gas pressure exceeds external water
pressure.

2.3.4 Evaluation of Oxygen Mass Transfer Coefficient

An accepted procedure determining the overall oxygen transfer coefficient is presented
by the American Society of Civil Engineers (ASCE, 1992). The method involves the
removal of dissolved oxygen (DO) from a known volume of water and reoxygenation to
a value near saturation. Measuring the DO concentrations at various times, t, during the
reaeration period allows the oxygen transfer coefficient to be found using Equation 2-12,
which can be linearised to find a value of KL at each determination point.

36
  C  Ct 
e  K L at   s  Equation 2-12
 C s  C0 
Where: KL = liquid side mass transfer coefficient (ms-1)
a = specific surface area (m2m-3)
Ct = oxygen concentration in bulk liquid at time t (mg l-1)
Cs = oxygen concentration in equilibrium with gas as given by Henry’s
Law
C0 = initial concentration (Ct at t = 0)

A modified form of this method is employed by Côté et al. (1989), in their study
involving the use of silicone rubber membranes for oxygenation of water. Due to the
different solubilities of oxygen and nitrogen in membrane materials, the value of the
Henry’s Law constant for this system cannot be easily determined. To avoid this
problem, the gas side oxygen concentration is used, allowing the overall mass transfer
coefficient to be obtained instead, as detailed in Equation 2-10

 C  Ct 
e Kat   g 
 C g  C0 
Equation 2-13

where: K= overall mass transfer coefficient, m/s

Cg = oxygen concentration in gas phase

Although use of this method eliminates the difficulties involved in calculation of

interfacial concentration required by the ASCE method, errors are introduced due to the
uncertainty of gas phase concentrations caused by pressure drop and changes in gas
composition along the length of the membrane as a result of diffusion and back diffusion
through the membrane material.

An alternative method, based purely on bulk liquid side concentrations, is presented by

Cussler (1997). The method requires knowledge of the saturation concentration at the
experimental conditions and obtains an average value of the overall mass transfer
coefficient according to Equation 2-14:

37
 Ct  C * (1  e Kat ) Equation 2-14

Where: C* = saturation oxygen concentration in bulk liquid at experimental

conditions

2.3.5 Calculation of oxygen flux

Average oxygen flux can be obtained by calculation from the obtained mass transfer
coefficient as introduced by Côté (1989) as shown in Equation 2-15.

J  KC Equation 2-15

Where: J = average oxygen flux

K = overall mass transfer coefficient
ΔC = concentration difference across membrane

The use of Equation 2-15 requires knowledge of the oxygen concentration on the air side
of the membrane, which can be difficult to obtain for the reasons outlined in Section
2.3.2. An alternative method is the calculation of oxygen flux directly from
experimental data using Equation 2-16:

(Ct  C0 )
J
a(tt  t0 ) Equation 2-16

2.3.6 Bubble formation

Equation 2-15 is valid only below a critical value of transmembrane pressure, which is a
function of the membrane geometries (Coté et al., 1989). Above this critical value,
several researchers have reported the formation of a layer of bubbles which act as an
additional mass transfer resistance and reduce oxygen flux (Casey et al., 1999, Coté et
al., 1989, Ahmed & Semmens, 1996). In order to avoid this during operation, any
increase in gas pressure should be accompanied by an increase in liquid side fluid
pressure.

38
Literature contains details of two possible causes of bubble formation. In investigations
where deoxygenation was achieved by the use of nitrogen gas, the water becomes
saturated with nitrogen, meaning any nitrogen that diffused through the membrane at the
beginning of the aeration period (where air was used on the gas side) would be unable to
enter the bulk liquid, leading to the formation of bubbles (Coté et al., 1989).

Alternatively, bubble formation has also been attributed to higher than saturation values
of gas concentration at the membrane/gas interface, with oxygen concentrations of up to
100 mg l-1 being reported (Casey et al., 1999).

However, in a MABfR, although bubbles were observed during startup with pressures
exceeded 0.5 bar, they were not seen once the biofilm had reached a thickness of
100µm, allowing higher pressures to be used (Casey et al., 1999).

2.3.7 Effect of attached biomass on oxygen transfer

The presence of a biofilm on the surface of an aeration membrane affects the mass
transfer through the membrane in number of different ways. Results from experimental
(e.g. dos Santos & Livingston, 1995) and modelling studies (Essila et al., 2000)
suggest that respiration of biomass near the membrane surface leads to an increase in the
rate of oxygen transfer through the membrane. This can attributed to the maintaining of
a near-maximum concentration difference across the membrane, giving higher fluxes as
suggested by Equation 2-15.

Additional studies by Casey et al. (2000a, 2000b) suggest that this ‘respiring effect’ is
applicable only to young, thin biofilms. In thicker biofilms, the accumulation of biomass
leads to an increase in the resistance of mass transfer, reducing the oxygen transfer rate.
This has consequences for pollutant removal as the biofilm mass transfer resistance
slows the diffusion of substrate to the oxygen rich areas of the biofilm and of oxygen to
substrate rich areas of the biofilm.

39
Côté et al. (1989), suggest that the presence of a biofilm on the membrane surface will
have a negative effect on the rate of oxygen transfer. The researchers did not carry out
investigations with active biomass, but postulate that adsorption into the membrane of
CO2 and other respiration products will decrease the oxygen diffusion coefficient in the
membrane material, slowing the oxygen transfer rate.

Shanahan & Semmens (2006) carried out one of the few investigations where the mass
transfer characteristics of the clean membrane in a MABfR were established prior to
establishment of a biofilm. Using a flat sheet membrane, local oxygen fluxes of the
clean flat sheet membrane was calculated from a correlation similar of the form in
Equation 2-15, developed in the work with a clean membrane, and compared to fluxes
calculated from conversion of ammoniacal nitrogen to nitrate. The researchers observed
reduction in fluxes in upstream sections of the membrane, where the presence of a
biofilm reduced turbulence, and increases in downstream areas where the biofilm
reduced the boundary layer. The implication of this work for tubular membranes, where
the boundary is known to be of less significance, is that oxygen transfer is reduced by
the presence of the biofilm.

2.3.8 Design of hollow fibre membrane contactors

When designing a hollow fibre membrane contactor for gas transfer, there are two
elements that must be considered in order to maximise mass transfer:

(i) Maintenance of as high as possible concentration difference

(ii) Obtaining high mass transfer coefficients through good design.

The factors for consideration in design of hollow fibre contactors are analogous to those
involved in the design of heat exchangers (Coulson et al., 1999b). Mass transfer
coefficients, like heat transfer coefficients, cannot be accurately calculated theoretically.

40
In the absence of theoretically derived design equations, empirical relations of the form
of Equation 2-17 are used. Ascertained through lab scale experiments, they can be used
for scale-up provided geometric similarity is maintained.

Sh  a Re b Sc c Equation 2-17
𝐾𝐿 𝑑
𝑆ℎ = Equation 2-18
𝐷
𝑑𝑣𝜌
𝑅𝑒 = Equation 2-19
𝜇
𝜇 Equation 2-20
𝑆𝑐 =
𝜌𝐷
Where: Sh = Sherwood number (dimensionless form of mass transfer coefficient)
Re = Reynolds number (dimensionless)
Sc = Schmidt number(dimensionless)
a,b,c = constants (dimensionless)
K = mass transfer coefficient (ms-1)
d = characteristic length (m)
D = diffusivity (m2s-1)
V = velocity (ms-1)
Ρ = density (kgm-3)
µ = viscosity (Pa.s)

The Schmidt number, found by divided the kinematic viscosity by the diffusivity of
oxygen in water, is constant for all water-oxygen systems. A value of 0.33 is widely
accepted for oxygen/water systems in the literature (e.g, Yang & Cussler, 1986, Coté et
al., 1989, Ahmed & Semmens, 1996, Vladisavljevic, 1999).

2.4 Membrane Attached Biofilm Processes

Many researchers have investigated the use of membrane aeration for wastewater
treatment. Timberlake et al. (1988) conducted one of the first such studies and achieved
significant nitrification of wastewater. Later studies by Brindle & Stephenson (1996)
and Yamagiwa et al. (1994), who used a “fibrous woven support” in proximity to the
membrane in order to increase the surface area available for biofilm attachment,
obtained excellent BOD removal, nitrification and denitrification. These studies have
41
displayed that these three key processes of wastewater treatment can successfully be
carried out simultaneously if correct process conditions can be maintained.

2.4.1 Pollutant removal

Membrane aerated biofilm reactors have been used to treat a variety of wastewaters
including domestic wastewater (e.g. Pankhania et al., 1999), artificial swine wastewater
(Terada et al., 2003), effluent containing acetonitrile (Li et al., 2008) and effluent
containing pharmaceuticals (Peng et al., 2015).

The studies most relevant to this work are those which investigated the use of MABfRs
in situations where aerobic heterotrophy and nitrification processes occurred
simultaneously, with or without denitrification also taking place. A selection of these
studies is discussed in Sections 2.4.1.1 and 2.4.1.2.

2.4.1.1 Aerobic processes

Timberlake et al. (1988) authored an early study on the use of MABfRs for wastewater
treatment. Using a Gore-Tex membrane and modest lumen pressures, up to 55% organic
carbon removal was achieved at rates up to 4.2 gTOC m-2day-1. Nitrification was also
obtained concurrently at rates up to 0.6 gN m-2day-1, with simultaneous denitrification
proceeding at the same rate.

Yamagiwa et al. (1994), using their “fibrous woven support” achieved simultaneous
organic carbon removal and nitrification from a wastewater with a composition similar
to that of secondary effluent (20 mg l-1, 4 mgN l-1 as ammonia). Conversation rates of
6.3 gTOC m-2day-1 and 2.2 gN m-2day-1 was obtained using lumen pressures between
19.6 and 29.4 kPa (gauge). The researchers also reported a limited effect of air pressure
on reaction rates.

42
Downing & Nerenberg (2008a) investigated the nitrification rate of a membrane aerated
biofilm in the presence and absence of BOD. The researchers obtained a nitrification
rate of 1.5 gN m-2day-1 in the absence of BOD in the bulk liquid. This decreased to 1.3
gN m-2day-1 in the presence of 1 gBOD m-3 in the bulk, and to 0.4 gN m-2day-1 when the
bulk BOD concentration was 10 g m-3. The observed decrease in nitrification rate was
attributed to increased competition for oxygen from heterotrophic bacteria. The
researchers also noted that nitrification in the MABfR was less inhibited by BOD than in
convention biofilms.

Satoh et al. (2004) carried out a microprobe study of nitrification in a MABfR. The
researchers obtained nitrification rates of approximately 0.5 gN m-2day-1, whilst
confirming through the use of the microprobes that the majority of nitrification took
place close the membrane surface. The location of nitrifiers on the membrane surface,
where oxygen concentrations are highest and BOD concentrations lowest, explains the
reduced inhibition observed by Downing & Nerenberg (2008a).

COD and ammoniacal nitrogen removal rates obtained in the most relevant MABfR
studies to this work are summarised in Table 2-4.

Table 2-4: Comparison of pollutant removal rates achieved in selected MABfR studies
Author rCOD ramm-N
(gCOD m-2day-1) (gN m-2day-1)
Timberlake (1988) 0.19 0.04
Pankhania et al. (1994, 1999) 15.08 n/a
Yamagiwa et al. (1994) 6.3 2.2
Semmens et al. (2003) 10.0 2.0

2.4.1.2 Denitrification
Denitrification has not been studied to the same extent as aerobic processes. It is
considered the more reliable of the two total nitrogen removal processes - consistently
high denitrification rates are achieved in wastewater treatment plants under high COD

43
loadings. Nitrification is the less reliable of the two processes, and therefore requires
more optimisation for effective nitrogen removal. As such it is the more studied process
(Yamagiwa & Ohkawa, 1994).

Several studies have reported denitrification rates in MABfRs where denitrification was
not the focus of the work. The achieved removal rates are summarised in Table 2-5.

Table 2-5: Denitrification rates of various MABfR studies

Author Denitrification rates
gN m-2day-1
Timberlake et al. (1988) 0.1 – 0.6
Semmens et al. 2.0
Satoh et al. (2004) 0.12 – 0.33
Downing & Nerenberg (2008a) 0.23 – 0.32

In a study using a membrane bioreactor (standard MBR - no aeration supplied by

membrane), He et al. (2009) investigated the effect of C:N ratio on simultaneous
nitrification and denitrification. The researchers found the highest rates of total nitrogen
removal occurred at the highest C:N ratios, which was attributed to the availability of
sufficient COD for complete denitrification to take place.

The most comprehensive study of total nitrogen removal using a MABfR was carried
out by Walter et al. (2005). Using a synthetic feed which did not contain nitrate
nitrogen, the researchers measured nitrogen removal rates at various C:N ratios.
Nitrogen removal rates of up to 2 kg m-3day-1 were achieved with the highest rates
obtained at the lowest C:N ratio – a result which is contrary to He et al. (2009).

Although not considered by the researchers, a likely explanation for this is that the
higher C:N ratios led to less oxygen being utilised for nitrification; effectively meaning
that the rate of nitrogen removal by denitrification was limited by nitrate availability.
Matsumoto et al. (2007) carried out a modelling study based on a plug flow MABfR and

44
found the minimum C:N ratio for complete denitrification to occur can be calculated
from stoichiometry to be 2.86.

In the previously mentioned microprobes study, Satoh et al. (2004) found denitrification
to occur just above the nitrification zone (i.e. further away from the membrane surface),
and obtained denitrification rates in the range 0.12 – 0.33 gN m-2day-1, when an organic
carbon loading rate of 1.0 gCOD m-2day-1 was used.

2.4.2 Control of biofilm thickness

Several authors have commented on the negative effect thick biofilms can have on the
rate of pollutant removal, as the excess biomass represents a substantial resistance to
diffusion of substrate into the biofilm.

Hwang et al. (2010) operated a two-stage membrane biofilm reactor, where the first
stage used pure oxygen as the lumen gas with the second stage using hydrogen; in a
similar way to the Rittmann group (e.g. Lee & Rittmann, 2002). The system utilised
periodic sparging of nitrogen gas, to maintain to steady biofilm thickness for optimal
nitrification and denitrification. Whilst limited effect was seen on the nitrification
performance, a 25% increase in denitrification rate was observed.

2.5 Biological azo dye wastewater treatment

A variety of treatment options, including adsorption (Walker et al., 2003); chemical
oxidation (Aleboyeh et al., 2009); enzymatic catalysis (Cristovao et al., 2009);
constructed wetlands (Davies et al., 2006) and various biological processes have been
considered for dye house wastewater treatment.

There is a general agreement in the literature that the cleavage of the azo bond takes
place under anaerobic or anoxic conditions, with the general reaction mechanism being
that which is described by Figure 2-10.

45
 Figure 2-10: Anaerobic and aerobic degradation pathways of azo dyes
(Van Der Zee & Villaverde, 2005)

As shown, azo dyes readily degrade in anaerobic conditions to form colourless aromatic
amines which are resistant to further anaerobic degradation (Ong et al., 2005)
and have been reported as being or potentially being mutagenic agents (Shaw et al.,
2002). These aromatic amines breakdown further under aerobic conditions (Van Der Zee
& Villaverde, 2005), meaning that azo dye waste requires a combination of aerobic and
anaerobic/anoxic conditions in order to achieve complete removal.

The decolourisation and degradation of azo dye waste has been extensively studied by
various researchers. Decolourisation, which involved the cleavage of the –N=N- azo
bond, has been achieved under anaerobic, anoxic and aerobic conditions by different
groups of bacteria (Pandey et al., 2007). To date, there are no published studies on the
use of membrane aerated biofilms for decolorisation of azo dye waste.

The first stage of the mechanism is the anaerobic reduction of the azo bond. Gingell &
Walker (1971) proposed a two-stage mechanism for this reduction as described by
Equation 2-21 and Equation 2-22.

46
 2𝑒 − + 2𝐻 + + (𝑅 − 𝑁 = 𝑁 − 𝑅 ′ ) → (𝑅 − 𝑁𝐻 − 𝑁𝐻 − 𝑅 ′ ) Equation 2-21

2𝑒 − + 2𝐻 + + (𝑅 − 𝑁𝐻 − 𝑁𝐻 − 𝑅 ′ ) → (𝑅 − 𝑁𝐻2 ) + (𝑅′ − 𝑁𝐻2 ) Equation 2-22

In the reduction, the azo compounds (the ‘R’ groups) are used as terminal electron
acceptors, forming the R groups into amine compounds. This action breaks the azo
bond, which is the dye’s chromophore, removing the colour of the wastewater (Sponza
& Isik, 2002).

Many of the amine compounds formed by this degradation are readily degraded under
aerobic conditions (Brown & Labouruer, 1983). Complete degradation of sulphonated
azo dyes may prove problematic as sulphonated aromatic amines are difficult to degrade
(Tan et al., 2000), but is possible in the presence of properly adapted consortium of
microorganisms (Thumheer et al., 1986).

The exact mechanism by which azo bond cleavage occurs is unknown. Due to their large
molecular weight and polar nature, azo dyes are unlikely to pass through the cell
membrane into bacteria cells (Levine, 1991). Pearce et al. (2006), amongst other
researchers, suggest it is due to the action of the enzyme azo reductase, which is
secreted by Shewanella sp. and other bacteria.

Alternatively, some authors suggest the azo bond reduction takes place outside of
bacterial cells through chemical reactions with substrates such as sulphide, which are
typically present in dye house effluent as it is a common additive to dye baths (Van Der
Zee et al., 2003a).

Additionally, chemical and biological reaction mechanisms have been shown to be

accelerated by the addition of redox mediating compounds, which donate electrons to
facilitate azo bond cleavage and are later regenerated. Van Der Zee et al. (2003a)
demonstrated decolourisation of AO7 was significantly accelerated in the presence of
the redox mediator riboflavin at a sub-stoichiometric level.

47
As previously stated, there are no published studies detailing the use of membrane
aerated biofilms for decolourisation and degradation of azo dyes, but studies of
biological processes utilising combined or sequential anoxic and aerobic conditions are
contained in the literature.

Coughlin et al. (2002) investigated the use of a laboratory scale rotating biological
contactor (RBC) and found the process capable of decolourising AO7 from a
concentration of approximately 55 mg l-1 to below detection limits. A synthetic
wastewater was used in which AO7 was the only possible source of COD. A
consistently low effluent COD was obtained, which the researchers attributed to the
complete mineralization of the azo dye.

To date, there has been only one example of azo dye decolourisation involving a
MABfR. Wang et al. (2012) also used Shewanella sp. and achieved removal of up to
98% of AO7 with influent concentrations of between 50 – 200 mg l-1. However, the
MABfR was operated in sequential batch mode, and the researchers reported the biofilm
attached to the membrane surface was “not competent for AO7 decolourisation”.
Additionally, the best removal efficiencies were achieved during a period when biomass
from an activated sludge plant was mixed with the influent, and it may be speculated
that adsorption of azo dye onto this biomass made a significant contribution to the dye
removal.

2.6 Inclined Plate Settlers

The development of lamella settlers can be traced to two seemingly unrelated origins.
Hazen (1904), in his seminal paper on wastewater settling, explained how settling rate
was dependent upon settling area rather than tank volume, and suggested settling tanks
therefore be split via a stack of horizontal plates in order to increase available settling
area. It would have been normal practice at this time to periodically remove solids with a
series of moving chains rather than self-cleaning with inclined tank floors. This is

48
acknowledged by Masschelein (1992) as the first description of the use of lamella plates
in solids removal processes.

Boycott (1920) observed that blood corpuscles settled faster in inclined tubes compared
to those that are vertical. This phenomena has become known as the Boycott Effect
(Acrivos & Herbolzheimer, 1979) and has led to the use of multiple inclined tubes or
plates for settling processes in wastewater treatment (Mace & Laks, 1978), the mining
industry (Cook & Childress, 1978), fertilizer production (Wenk, 1990) and in cell
separation processes (Janelt et al., 1997).

In addition to the greater settlement efficiency, lamella settlers are also very low in
energy costs, command low capital costs, offer large savings in space occupied and can
settle fine suspensions at a high rate (Saleh & Hamoda, 1999, Humpal & Chiesa, 1990).
A disadvantage of lamella settlers is that, given certain process conditions, the flow
channels can become blocked; this can be overcome by fine screening of the influent
upstream of the settler (Grady et al., 1999).

Historically, there has been very little interest in the research of inclined plate or lamella
settlers, reflected in paucity of published research on the topic.

2.6.1 Design Considerations

Despite being in use since the 1960s, there are no definitive equations governing the
design of a lamella or inclined plate settler (IPS). Designs are based on experience and
established industry procedures, and quite often compromises have to be made in order
to produce an economically viable and practicable solution.

Instead, the design process involves following best practice established through existing
installations, and it is recommended that bench and/or pilot scale testing is carried out
prior to commissioning of a full scale unit (Humpal & Chiesa, 1990).

49
2.6.2 Operational Issues
The inclined plates are typically spaced 2 inches apart and inclined at angles between
45° and 60° to the horizontal. With angles below 45°, operation can be compromised by
plugging of flow channels by settled material, whilst at angles greater than 60°, long
flow channels must be employed to achieve desired settlement.

When determining the optimum angle and selecting a plate material, the wastewater
characteristics such as flow rate, influent suspended solids, type of solids and whether or
not the waste stream is corrosive, are of importance – the density, size distribution,
abrasiveness and ‘stickiness’ (whether they will adhere to the plates) of the solids must
also be considered (Saleh & Hamoda, 1999). Materials such as polyethylene,
polyvinylchloride, cement and wood are commonly used (Probstein & Hicks, 1978).

More recently, de Hoxar (2000) introduced a settler in which moving plates, with a
spiral arrangement, are used. These spiral settlers offer improved solid removal
efficiency and produce thicker sludge than traditional lamella settlers, and additionally
can remove particles that are lighter than water, as well as those that are heavier.

2.6.3 Modelling
The hydrodynamic flow in lamella settlers is very complex (Demir, 1995). Early
attempts to characterise the flow (e.g. (Ponder, 1925)) based on kinematic arguments,
were unsuccessful; with predicted settling rates consistently higher than experimentally
measured values.

Probstein et al. (1978) introduced a model based on the existence of three stratified
layers: clarified layer, feed suspension layer and sludge layer. The researchers also
identified the existence of two modes of operation (‘subcritical’ and ‘supercritical’
modes) depending on the ratio of the thickness of the clarified layer to channel height.
The supercritical mode was shown to be more stable, and lead to the design of a new
type of settler, where influent was introduced approximately one third of the distance

50
from the bottom of the plate, thus minimizing mixing between influent and settled
material.

Recently, an attempt has been made to use dimensional analysis in order to find the
optimum parameter values for solids removal with an inclined plate settler (Sarkar et al.,
2007) . The researchers obtained relationships which described the experimental results
for three different zones of operation (based on the ratio of Reynolds number to Froude
number), with good correlation fit coefficients (>0.93). However, there is lack of
consistency of units throughout the researchers’ calculations, which brings into question
the validity of their results. Additionally, the study used a monodisperse suspension of
sand and did not consider particle size.

A more recent Computational Fluid Dynamics (CFD) analysis by Salem et al. (2011)
investigated the impact of feeding inclined plate settlers using a nozzle system. The
researchers found this new inlet configuration led to better flow distribution over the
inclined plates with an associated improvement in separation efficiency; especially at
higher flowrates. Good agreement between the CFD model and experiment results was
also achieved. However, the work was based on the separation of crushed walnut
suspensions of uniform particle size, and is therefore of limited use in wastewater
treatment applications, where particle sizes vary greatly allow with other parameters
such as viscosity and density.

The studies have successfully produced models which adequately explain the produced
results. The studies have, however, generally dealt with model systems using discrete
particle sizes and little or no variation in the ‘stickiness’ or floc forming ability of the
solids. Therefore, these studies have not resulted in a comprehensive collection of design
equations, and design of inclined plate settlers remains based on experience from
industry.

51
2.7 Conclusion
The majority of the work carried out on the use of membrane aerated biofilms for
municipal wastewater treatment has focussed on the treatment of primary effluent. This
work builds on the existing body of research by investigating the treatment of secondary
effluent; additionally combining membrane aerated biofilms with inclined plate settlers
to create a new wastewater treatment technology.

This work will also explore the decolourisation of azo dye waste; something that has not
been reported previously using membrane aerated biofilms, but has been achieved with
other various biological processes.

52
3 Materials and Methods

The laboratory experimental work carried out during this project falls into two
categories: mass transfer experiments, where the oxygen transfer from air to water
through polymer membranes was characterised and compared for different membrane
materials, and membrane aerated bioreactor studies, where biofilms were grown on the
surface of polymer membranes and used for pollutant removal from wastewater.

The experimental set-ups and methods used for these laboratory studies are described
here in this Chapter. Pilot scale studies were also undertaken at municipal WwTWs; the
set-ups and methods for this work are described in Chapter 7.

3.1 Mass Transfer Studies

3.1.1 Membrane Materials

In this study, the oxygen transfer characteristics of two different membrane types are
investigated and compared: silicone rubber (polydimethylsiloxane), an example of a
dense membrane, and polyethersulphone (PES), an example of a microporous
membrane. The repeating structures of these two polymers are shown in Figure 3-1.

Figure 3-1: Structure of silicone rubber (left) and polyethersulphone polymers

The silicone rubber used was peristaltic pump tubing (Watson-Marlow Pumps, U.K.)
with an average tube bore of 1 mm and a wall thickness of 0.35 mm.

53
The PES membranes used were 0.2 µm (nominal pore size) microfiltration membranes
(A/G Technology Corporation, U.S.A.), with an external diameter of 1.5 mm, an average
wall thickness of 50 μm and a porosity of 30%.

3.1.2 Membrane Modules

In order to characterise the oxygen transfer properties of the membrane materials, it was
necessary to construct a membrane module. The modules were designed and constructed
using the same principles associated with that of a shell and tube heat exchanger – the
design of which allows high velocity of flow to be obtained, even at low levels of
volumetric liquid throughput, and therefore a good mixing regime is maintained
(Coulson et al., 1999b). This set-up is similar to that used in studies by Doig et al.
(1999) investigating diffusion rates of solutes in a liquid-liquid extraction operation with
silicone rubber membranes and by Yang and Cussler (1986) in an investigation into the
gas-liquid transfer characteristics of polypropylene hollow fibre membranes.

Figure 3-2: Silicone rubber membrane module

A cylindrical polycarbonate tube with an internal diameter of 25 mm was used to form

the shell, with the membrane fibres forming the tubes. The desired number (12) of fibres
were fed through a polycarbonate end plate, fitted inside the tube and secured and sealed
using an epoxy resin potting compound (RS Components, U.K.). The fibres were evenly
spaced around a central stainless steel support, which maintained a constant distance
between the two polycarbonate end plates during module construction.

54
The unit was made water and air tight by the use of modified George Fischer plumbing
fittings (RS Components, U.K.), allowing the module to be connected to compressed air
and water supply as required. This set-up essentially is that of a shell and tube mass
exchanger, with one shell pass and one tube pass. The unit was operated in counter-
current flow, as this mode of operation gives a higher mean transfer difference compared
to co-current flow (Coulson et al., 1999b).

3.1.3 Parameter measurement

3.1.3.1 Dissolved oxygen

Dissolved oxygen (DO) concentration and temperature were measured during each
experimental run using a Hach-Lange LDO dissolved oxygen meter (Isis Environmental,
U.K.) connected to a Hach-Lange sc100 Controller (Isis Environmental, U.K.).

The probe uses a membrane material whose response to incident laser light changes with
changing dissolved oxygen concentration; allowing DO concentration measuring
without any oxygen consumption, therefore giving more accurate measurements of DO
especially at low DO levels.

The probe is capable of measuring DO concentrations from 0-20 mg l-1 to a resolution of

±0.01 mg l-1 and temperatures of 0-50ºC to an accuracy of ±0.2ºC.

3.1.3.2 Air flowrate

Air flowrate was measured on a volumetric basis by use of two air flow control meters
(Key Instruments, U.S.A.), on the inlet and exhaust sides of the membrane module. The
exhaust side flow control meter controlled the flow of air through the membrane
module, whilst the inlet side was purely for measurement purposes. Two different ranges
of meters were used: capable of measuring 0 - 1.2 lpm or 0 - 5 lpm as required by
experimental conditions.

55
3.1.3.3 Air pressure
Inlet air pressure was controlled by a IMI Norgren (RS Components, U.K.) pressure
regular on the inlet side and measured by pressure gauges (RS Components, U.K.),
capable of measuring between 0 and 2.5 bar. A similar pressure gauge was used to
measure air pressure on the exhaust side. Compressed air was available from a service
line at a maximum pressure of approximately 6 barg.

3.1.4 Nitrogen diffusion

Prior to the start of each experimental run, dissolved oxygen was stripped from tap water
by sparging with nitrogen gas. Nitrogen gas (BOC, U.K.) was bubbled through the tank
using two Pyrex Grade 1 (pore index 90 – 150 μm) glass spargers (SciLabWare, U.K.).

3.1.5 Stirring equipment

To ensure the water in the sampling tank was well mixed, an IKA-Combimag Reo
magnetic stirrer (IKA Labortechnik, Germany) was employed in combination with a 25
mm x 5 mm x 5 mm triangular prism magnetic flea (Sigma Aldrich, U.K.). A rotational
speed of approximately 500 rpm was used, giving good mixing without any vortexing or
significant surface aeration occurring.

3.1.6 Pumping equipment

During each experimental run, the known volume of water was circulated from the
measurement tank, through the membrane module and returned to the measurement tank
by the action of a peristaltic pump.

The pump used was a Watson-Marlow 302S (Watson-Marlow, U.K.) capable of

operating at speeds between 0 and 65 rpm. The pump was used with an 8 mm internal
diameter tube, giving flowrates up to 550 ml min-1.

56
3.1.7 Datalogging
The sc100 controller is supplied with 2 analog (4-20 mA) outputs. Designed for use as a
control loop, these outputs were connected to a PC via an ADC-11 Data logger (Pico
Technology Limited, U.K.), and used for logging the DO and temperature data.

The 4-20 mA signal output is proportional to the DO concentration and temperature

reading detected by the probe, and can be converted back to original values using
calibration factors via the software provided with the datalogger.

During the mass transfer experiments, a data logging period of 10 seconds was used. For
the saturation concentration experiments, and when bulk dissolved oxygen in the
MABfR was monitored, a data logging period of 10 minutes was used – as in these
experiments the rate of change of oxygen concentration was much slower.

3.1.8 Experimental Set-up

The equipment described above was set-up up as shown in Figure 3-3 and Figure 3-4.

Figure 3-3: Schematic diagram of experimental set-up

57
 Figure 3-4: Water reservoir tank,
magnetic stirrer and Hach-Lange LDO
Meter

3.1.9 Experimental procedure

3.1.9.1 Aeration experiments

The oxygen mass transfer was investigated using a modified form of the method
described by the ASCE Standard (ASCE, 1992), and analysed using the method detailed
by Cussler (1997).

Two litres of tap water was placed in the holding tank, the lid fitted and the DO probe,
nitrogen gas spargers and inlet and outlet tubes put in position. The water was then
deoxygenated by bubbling nitrogen through the mixture under intense mixing until the
dissolved oxygen concentration dropped below 0.20 mg l-1.

Once this DO value was obtained the following procedures were carried out to conduct
the experiment:
 Impeller speed was reduced to the minimum value (~100 rpm)

58
  Compressed air was circulated through the tubes of the membrane
module by setting the inlet air pressure and exhaust air flowrate to the
desired values
 Water was circulated through the membrane module shell by operating
the pump at the desired setting
 Data was recorded by switching on the datalogger

The experiment then continued until a DO concentration of 6 mg l-1 was achieved.

3.1.9.2 Saturation Concentration Experiments

In order to use the analysis technique detailed by Cussler (1997) for ascertaining mass
transfer coefficients, saturation concentrations at experimental conditions must be
known.

To obtain the required saturation values, a modified form of the procedure previously
used for determining the temperature-saturation concentration relationship for surface
aeration – intended for calculating oxygen concentrations in water courses (Hendrickson
et al., 1960) – was used.

The work of Hendrickson et al. (1960) found that saturation concentration displays a
third order polynomial dependence on liquid temperature, giving a relationship of the
form:

[ DO]*  a  b1T  b2T 2  b3T 3 Equation 3-1

Where: [DO]* = saturation oxygen concentration at temperature T (mg l-1)

a = saturation oxygen concentration at 0 °C (mg l-1)
b1 = empirical coefficient (mg l-1°C-1)
b2 = empirical coefficient (mg l-1°C-2)
b3 = empirical coefficient (mg l-1°C-3)
T = temperature (°C)

59
As the enriching action of the membranes is dependent upon oxygen permeation rates in
the membrane material and gas side oxygen concentration, the saturation point must be
established for each of the membrane materials used in the study at each of the values of
inlet pressure used in experimental runs.

Ascertaining the constants for use in Equation 3-1 used a modified form of the set-up
described in Section 3.1.9. As much as was practicable of the experimental equipment
was placed in a Grant Instruments JB5 water bath (Grant Instruments, U.K.) and the
water bath temperature set for the desired value. In order to achieve temperatures below
ambient, copper coils through which coolant was circulated were placed in the water
bath. The coolant was circulated by Büchi CH 9230 recirculation chiller (Büchi,
Switzerland), capable of temperatures in the range -10 °C - +10 °C.

The membrane module was then operated with maximum water side flowrate until no
change in oxygen concentration was observed (typically after 24 hours). Due to heat
losses thermal equilibrium between the heating/cooling water and aeration water was not
achievable, but a stable approach temperature was obtained.

3.2 Membrane Aerated Biofilm Reactor Studies

Two MABfRs were operated in this study – both reactors used the same silicone rubber
as used in the aeration experiments as the membrane material. MABfR A was operated
with effluent designed to replicate ammoniacal nitrogen and organic carbon
concentrations in a non-consent compliant settler effluent in a munipical WwTW and
MABfR B used effluent designed to replicate effluent from the textile industry.

60
3.2.1 Reactor Set-ups

3.2.1.1 MABfR A
The equipment set-up used with MABfR A is shown in Figure 3-5.

Figure 3-5: Schematic diagram of MABfR A Set-up

The membrane module consisted of 5.0 m of silicone rubber membrane, split into two
approximately equal lengths wrapped around a PVC tube frame to give an effective total
membrane surface area of 0.024 m2, estimating that 10% of the available membrane area
is lost due to contact with the PVC frame.

The active volume (total volume minus volume of the membrane module) of the reactor
tank was 4.35 l, giving a specific surface area of 5.52 m2m-3.

Synthetic waste was delivered to the reactor by a Watson Marlow 101U/R peristaltic
pump (Watson Marlow, U.K.). The pump is capable of speeds between 2 and 32 rpm

61
and when fitted with a 3.2 mm internal diameter tube can deliver between 3.25 and 52
ml min-1. Effluent was allowed to overflow and collected for analysis.

The reactor was operated at ambient conditions with no temperature control employed.
Temperature was recorded during sample collection however, and was found to lie in the
range 17.1 – 21.8 °C during reactor operation.

3.2.1.2 MABfR B
The equipment set-up used with MABfR B is shown in Figure 3-6.

Recirculation pump

MABfR

Feed Tank
Overflow
Feed pump Holding Tank

PI PC

FI FC
Exhaust Air Compressed Air

Figure 3-6: Schematic diagram of MABfR B set-up

The membrane module consisted of lengths of silicone rubber wrapped around a PVC
frame. One fibre, 1.58 m in length was wrapped in the horizontal plain and a shorter,
0.68 m length was arranged in the vertical plain. Again allowing for a 10% loss of
membrane surface area due to contact with the frame, this gave an effective membrane
surface area of 0.00374 m2.

The active volume of the reactor, including the holding tank was 0.91 l, giving a specific
membrane surface area of 4.11 m2m-3. As with MABfR A, no temperature control was
employed.

62
Synthetic waste was delivered by a Watson Marlow 101U/R peristaltic pump (Watson
Marlow, U.K.), identical to that used in MABfR A. To prevent shortcutting and force
contact between influent and the membranes, the influent was introduced to the bottom
of the tank via a feed tube.

A recirulation pump provided mixing by returning liquor from the holding tank to the
feed tube of the reactor. Recirculation was carried out by a Watson Marlow 501U
peristaltic pump (Watson Marlow, U.K.), at a flowrate of 4.3 l h-1.

Effluent was allowed to overflow from the holding tank and collected for analysis.

3.2.2 Synthetic media

The composition of the synthetic wastewaters used in MABfR A and MABfR B are
detailed in Sections 3.2.2.1 and 3.2.2.2 respectively. All components were of analytical
grade.

3.2.2.1 MABfR A
The concentrations of the various components used in the synthetic waste used in the
MABfR A are given in Table 3-1:

63
 Table 3-1: Synthetic waste composition
Component Concentration
Soluble Starch 50 mg l-1
NH4Cl 30 mg l-1
CaCl2.2H2O 5 mg l-1
MgSO4.7H2O 4 mg l-1
NaCl 7 mg l-1
K2HPO4 28 mg l-1
Na2CO3 10 mg l-1
NaHCO3 20 mg l-1
Fe(III)-citrate 1 mg l-1
Trace elements

The media used in the MABfR was designed to replicate the ammonia and organic
carbon concentrations in non-compliant settler effluent (Irwin 21st April 2006, personal
correspondence). The recipe is based on that contained in OCED Guidelines (OECD,
1984), though for use in this project peptone and meat extract have been replaced by
soluble starch as the sole carbon source (so that soluble COD concentration can be
related to starch concentration) and ammonium chloride is used as the source of
ammonia as opposed to urea contained in the OCED guidelines.

Trace elements, required in order to ensure proper microbial growth, were supplied
based on a recipe included in Alef (1995).

To prepare the media detailed in Table 3-1, 50 mg l-1 of soluble starch was added to an
appropriate amount of distilled water. This solution was then autoclaved at 121 °C for 20
minutes in order to improve the stability of the media (Alef, 1995).

After sterilisation, 1 ml of each of three stock solutions was added to each litre of the
synthetic waste to give the media the composition detailed in Table 3-1. The
composition of the stock solutions is detailed in Table 3-2.

64
 Table 3-2: Composition of stock solutions used with MABfR A
Stock Solution 1 – Salts + Trace elements
NH4Cl 30 g l-1 MnCl2 0.01 g l-1 Na2MoO4 0.01 g l-1
CaCl2 4 g l-1 ZnCl2 0.01 g l-1 CoCl2 0.01 g l-1
MgSO4 2 g l-1 KBr 0.01 g l-1 Al2(SO4)3 0.01 g l-1
NaCl 7 g l-1 KI 0.01 g l-1 HBO3 0.01 g l-1
Fe(III)-citrate 1 g l-1 CuSO4 0.01 g l-1 EDTA 0.01 g l-1

Stock Solution 2 - Carbonate Stock Solution 3 - Phosphate

NaHCO3 20 g l-1 K2HPO4 28 g l-1
Na2CO3 10 g l-1

In the period where the effect of pollutant loading on the performance of the MABfR
was investigated, changes were made to the influent synthetic media. These changes are
discussed in the relevant section in Chapter 6.

3.2.2.2 MABfR B
The concentrations of the various components of the synthetic media used with MABfR
B during Run 1 are given in Table 3-3. The synthetic media was designed to replicate
the COD and azo dye concentrations present in effluent from a local carpet factory
(Wilkinson 2007).

65
 Table 3-3: Composition of synthetic dye waste used in MABfR B
Component Concentration
Acid Orange 7 20 mg l-1
Peptone 200 mg l-1
Sucrose 550 mg l-1
NH4Cl 350 mg l-1
MgSO4 50 mg l-1
Fe(III)Cl3 10 mg l-1
K2HPO4 30 mg l-1
NaHCO3 30 mg l-1
Trace elements

Trace elements, required for healthy microbial growth, were used in the same
concentration as with MABfR A and detailed in Table 3-2.

Changes were made to the composition of the synthetic media used with MABfR B for
runs 2 and 3. These changes are discussed in the relevant section of Chapter 8.

3.2.3 Seeding of reactors

The biomass used in this study were unknown mixed populations, obtained from three
sources:
a. Garden soil, in a way similar to that used by Lees (1951) for the first successful
isolation of nitrifying bacteria.
b. Activated sludge from a local municipal wastewater treatment works
(Newtownbreda WwTW, NI Water, U.K.). The wastewater treatment works was
achieving simultaneous organic carbon removal and nitrification at the time of
sludge sampling.
c. An experimental lab scale unit achieving azo dye decolourisation using
Shewanella sp. (Lipscomb et al., 2008).

66
Several grams of garden soil/activated sludge/dye degrading biofilm was collected and
added to 200 ml of the synthetic media detailed in Table 3-1 in a conical flask. The flask
was then placed in a water bath overnight at 37 °C and allowed to settle. The supernatant
was then used to seed the reactor.

3.3 Effluent characterisation

Samples of influent and effluent were regularly collected and immediately analysed for
pH (which is temperature dependant), and either analysed immediately, or stored in
accordance with Standard Methods until analysis was carried out (ASCE, 1992).

If analysis was to take place within 48 hours, the samples were refrigerated at 4 °C until
analysis took place; otherwise they were frozen at –17 °C, and defrosted overnight in a
refrigerator before analysis was carried out. Freezing of samples was only employed at
times when analysis could not be carried out (for example during lab holidays).

The analyses carried out on the influent and effluent from the MABfRs are summarized
in Table 3-4.

Table 3-4: Wastewater analyses

Reactor A Reactor B
Influent Effluent Influent Effluent
pH X X X X
COD X X X X
Ammoniacal Nitrogen X X X X
Nitrite X
Nitrate X X
Turbidity X X
Colour X X

67
3.3.1 pH
pH was measured with a Hanna Instruments 8424 pH meter (Hanna Instruments, U.K.)
at ambient temperature. The meter has a measurement range of -2.00 to 16.00, a
resolution of 0.01 pH units and an accuracy of ±0.01 pH units.

Accuracy of the meter was checked periodically by comparison with a pH 7.00 buffer
solution, and calibrations with pH 4.00, 7.00, 9.04 buffer solutions were carried out if
required.

3.3.2 Ammoniacal Nitrogen

Ammoniacal Nitrogen concentration was ascertained through use of Hach-Lange
sensIONTM2 fitted with Ammonia Ion Selective Electrode (Isis Environmental, U.K.),
capable of measuring NH4+-N concentrations between 0.05 and 14,000 mg l-1 over a
temperature range of 0 – 50 °C.

25 ml of sample was adjusted to high pH through the addition of one Hach-Lange

Ammonia Ionic Strength Adjustor (ISA) powder pillow (Isis Environmental, U.K.). The
ISA converts all ammoniacal nitrogen to free ammonia gas. The gas diffuses through the
electrode membrane and causes a pH change in the electrode solution. The magnitude of
the pH is proportional to the ammonia concentration in the solution being measured
(Hach Company, 2001). The probe was calibrated before each use with 10 mg l-1 and
100 mg l-1 NH4+-N standards.

3.3.3 Chemical Oxygen Demand

The Chemical Oxygen Demand (COD) was determined by a Hach-Lange DR2800
Spectrophotometer (Isis Environmental, U.K.) using the Reactor Digestion Method. Low
range (3-150 mg l-1) digestion vials were used with feed and effluent samples from
Reactor A and high range (20-1500 mg l-1) digestion vials used with feed and effluent
samples from Reactor B. Any turbidity (which would give a false high reading) was

68
removed by centrifugation at approximately 13,000 rpm for 6 minutes in a 1.5 ml
Eppendorf tubes (Premier Scientific Ltd., U.K.) with a MSE Micro Centaur centrifuge
(Anachem Scotlab, U.K.)

The method involves the oxidation of the sample at 150 °C for 2 hours by potassium
dichromate. As the oxidation progresses the orange dichromate ion (Cr2O72-) is itself
reduced to the green chromic ion (Cr3+) (Figure 3-7). Colorimetric determination of the
amount of Cr2O72- remaining or Cr3+ produced allows the COD of the sample to be
ascertained (Hach Company, 2007).

For MABfR A, where soluble starch is the only source of organic carbon, the value of
the COD can be directly related to starch concentration (dichromate does not oxidise
ammonia to nitrate).

Figure 3-7: Spent (left) and unused COD vials Figure 3-8: Fresh (left) and used nitrate vials

3.3.4 Nitrate
Nitrate concentration was established, again through use of a Hach-Lange DR 2800
Spectrophotometer (Isis Environmental, U.K.), using the chromotropic Acid Method
with Test ‘N Tube™ NitraVer® X Reagent Set. The method can accurately determine
nitrate concentrations between 0.2 and 30.0 mg l-1.

69
The method involves reaction of nitrate with chromotrophic acid under acid conditions
to yield a yellow product (Figure 3-8). The amount of product is ascertained by
absorbance at 410 nm and is directly related to nitrate concentration (Hach Company,
2007).

3.3.5 Nitrite
Nitrite concentration was determined by a Hach Lange DR 2800 Spectrophotometer (Isis
Environmental, U.K.) using the Diazotization Method with Test ’N Tube™ NitriVer®
Nitrite Reagent Vials. The method is capable of measuring NO2--N concentrations
between 0.003 to 0.500 mg l-1.

Nitrite concentration is determined by colorimetry at 507 nm following reaction of the

sample with excess sulfanilic acid to form an intermediate diazonium salt which then
couples with chromotropic acid to produce a pink coloured complex. The pink colour is
directly proportional to the amount of nitrite present (Hach Company, 2007).

3.3.6 Turbidity
Turbidity, as a guide to the amount of suspended biomass present in an effluent sample,
was obtained by a Hach-Lange 2100P Turbidimeter (Camlab, U.K.). The meter has a
measurement range of 0 – 1000 Nephelometric Turbidity Units (NTU) and a resolution
of 0.01 NTU and an accuracy of ±2 % of reading value.

3.3.7 Colour
Colour absorbance was determined using a Perkin-Elmer Lamba 9 UV-Visble
spectrophotometer (Perkin-Elmer, U.K.). Samples, which were centrifuged at 13,000
rpm for 6 minutes to remove any turbidity, were placed in polycarbonate cuvettes
(Premier Scientific, U.K.) and scanned from 700 nm to 300 nm, with the absorbance of
the samples normalized against the absorbance of a blank; a cuvette containing distilled
water (Figure 3-9).
70
 Figure 3-9: Spectrophotometer cuvettes

The concentration of coloured compounds contained in the samples can then be found
by identification and quantification of the compounds λmax, the wavelength at which
maximum absorbance is observed. Concentration can then be related to the maximum
absorbance via reference with a calibration curve, in accordance with Standard Methods
(Clesceri et al., 1998).

3.3.8 Biochemical Oxygen Demand (BOD5)

Biochemical Oxygen Demand was carried out on selected samples from the pilot trails
using the same method employed by Northern Ireland Water (Irwin 21st April 2006), a
modified form of that contained in Standard Methods (Clesceri et al., 1998).

In the method, a sample of wastewater is diluted with oxygen saturated water which has
been seeded with microorganisms and contains nutrients. The dissolved oxygen
concentration is measured and the sample is placed in a thermostatically controlled
incubator at 20 °C for 5 days, after which the DO concentration is measured again. The
BOD is determined from the change in sample DO in comparison to a blank sample
prepared in the same way.

Allyl thiourea is added to the samples in order to suppress nitrification during the
inhibition period. The result is therefore the BOD due to the presence of carbonaceous
rather than ammoniacal compounds.

71
3.4 Effluent disposal
All microbial biomass has the potential to be, or to become, pathogenic. As such, all
effluent was collected and treated with a disinfection solution (Virkon) before being
flushed to drain with excess water.

72
4 Mass Transfer Studies

In order to maximize the treatment potential of the MABfR, it was first necessary to
study the transfer of oxygen through polymer membranes. The data obtained through
these studies allowed choices to be made regarding the membrane type, membrane
arrangement and the various parameters (air flowrate, air inlet pressure, water flowrate)
in order to maximize the oxygen flux and therefore the aerobic treatment potential of the
MABfR.

4.1 Saturation concentration

In order to use the equations contained in Cussler (1997) for obtaining mass transfer
coefficients from experimental data, knowledge of the saturation concentration at the
experimental conditions is required.

Although temperature-concentration relationships for oxygen-water systems are

contained in literature (e.g. Hendrickson et al., 1960), they cannot be applied in this
situation. The saturation concentration in this case is not only a function of temperature,
but also of the oxygen concentration with which it is in equilibrium, which in turn
changes with both air-side gas pressure and membrane type.

The temperature-saturation concentration relationships for each value of inlet pressure

used in the investigation for both silicone rubber and polyethersulphone were therefore
found experimentally; thereby allowing Equation 3-1 to be used to find the oxygen
concentration at saturation.

[ DO]*  a  b1T  b2T 2  b3T 3 Equation 3-1

73
The saturation constants (for use in Equation 3-2) found in this study (using tap water)
for surface aeration were compared with those of Hendrickson et al. (1960), from a
study using distilled water (Table 4-1). The values obtained in this study were slightly
lower than those in the literature. This can be explained by the fact that gas solubility
decreases with increased dissolved salts concentration. This has been long accepted as a
cause of variation in oxygen levels in marine situations (Gerlach, 1994).

Table 4-1: Comparison of saturation constants

Hendrickson et al. (1960) This study
a 14.652 13.374
b1 -0.4102 -0.2264
b2 0.007910 0.002739
b3 -0.00007777 -0.00002525
R2 0.999 0.997

4.1.1 Effect of membrane type

The temperature-saturation concentration coefficients for use in Equation 3-1 for both
silicone rubber and polyethersulphone using an inlet pressure of 0.5 barg are shown in
Table 4-2.

Table 4-2: Saturation coefficients for 0.5 barg inlet pressure

Silicone Rubber Polyethersulphone
a 16.61 15.03
b1 -0.37 -0.33
b2 2.80 x10-3 1.53 x10-3
b3 17.3 x10-6 24.0 x10-6
R2 0.999 0.999

Values of the coefficients for all of the inlet pressures used in this study were all
ascertained and are contained in Appendix 1.

74
The data from Table 4-1 and Table 4-2 have been used to plot Figure 4-1 to illustrate the
effect of temperature on the saturation oxygen concentration for the experimental setup
used in this study.

25
Saturation oxygen conc. (mg l-1)

20
SR

15 PES

Control
10

0
0 5 10 15 20 25 30 35
Temperature (°C)

Figure 4-1: Temperature-saturation correlations (0.5 bar)

Figure 4-1 shows that the oxygen saturation concentration is greater when silicone
rubber is used compared to polyethersulphone at all temperatures. This is the result of a
higher interfacial concentration of oxygen due to higher oxygen permeability in silicone
rubber compared to polyethersulphone (Robb, 1968). This enrichment effect has been
previously noted as the cause of elevated saturation concentrations in work by Casey et
al. (1999).

The plots for silicone rubber and polyethersulphone are similar in shape, but different to
the control trace. The only factor affecting the saturation concentration in the control
situation is the solubility of oxygen in water at the experimental conditions. This is not
the case with diffusion through a membrane where temperature also effects the free
volume of the polymer material and hence the solubility and diffusion constants of the
system (Zhang & Cloud, 2006).

75
4.1.2 Effect of inlet pressure
The effect of inlet pressure on the saturation concentration at 20 °C is shown in Figure
4-2, using values calculated from the relationships obtained for each set of membrane
material and pressures.

18
Saturation oxygen conc. (mg l-1)

16
14
12
10
8 SR

6 PES
4
2
0
0 0.5 1 1.5 2 2.5
Inlet pressure (bar)

Figure 4-2: Effect of inlet pressure on saturation oxygen concentration (20 °C)

Linear regression of the information displayed in Figure 4-2 reveals proportionality

between inlet pressure and saturation oxygen concentration with both the silicone rubber
and polyethersulphone membrane modules. This result is expected as a linear increase
in pressure is accompanied by a linear increase in oxygen partial pressure.

Both of the relationships have correlation coefficients in excess of 0.95 and an intercept
of approximately 8.8 mg l-1, which is close to the saturation concentration of 9.7 mg l-1
predicted by the coefficients in Table 4-1. The error between the theoretical saturation
concentrations at zero inlet pressures and that obtained through surface aeration only is
partially explained by the pressure losses experienced as air flows through the module.
Use of an average pressure gives an intercept of approximately 9.1 mg l-1, with the
remainder of the error being attributed to error of experimentation.

76
4.2 Oxygen Transfer

4.2.1 General observations

4.2.1.1 Change of Dissolved Oxygen

The dissolved oxygen/time relationship for a typical aeration experiment is shown in
Figure 4-3. Similar plots were obtained in all aeration experiments.

7
Dissolved oxygen conc. (mg l-1)

0
0 10 20 30 40 50 60 70 80 90
time (min)

Figure 4-3: Change in dissolved oxygen over duration of mass transfer experiments (Silicone
Rubber, 0.5 bar inlet pressure, 2 lpm gas side flowrate, 550 ml min -1 water side flowrate).

Initially, the increase in dissolved oxygen concentration was slow as mass transfer was
limited by the oxygen concentration in the membrane material. Once membrane
saturation was achieved, aeration was controlled by diffusion away from the membrane
surface and slowed down as the water side concentration approached saturation.

4.1.1.1 Bubble formation

During the aeration experiments, a layer of bubbles formed on the membrane surface, as
also observed by Côté et al. (1988) and Casey et al. (1999), and as shown in Figure 4-4.
These bubbles gradually grew in size as the experiment progressed, until they reached a

77
critical size where the bubble was sufficiently buoyant to overcome surface tension
forces and escaped into the bulk flow.

Figure 4-4: Bubble formation on membrane surface (Silicone Rubber, 0.5 bar inlet pressure, 2 lpm
gas side flowrate, 550 ml min-1 water side flowrate)

Two possible explanations for the composition of these bubbles exist:

1. As part of the experimental procedure, oxygen was first stripped by passing a stream
of nitrogen bubbles through the water layer. This formed a saturated solution of nitrogen
and therefore any nitrogen passing through the membrane would be unable to enter the
bulk phase and would instead form bubbles on the membrane surface.

2. The bubbles are composed of oxygen and form on the membrane water interface
where oxygen concentrations exceed the saturation value – interface oxygen
concentrations of up to 100 mg l-1 have been achieved using silicone rubber membranes
(Casey et al., 1999).

It is likely that a combination of these two explanations is valid, and the bubbles are
composed of a mixture of oxygen and nitrogen gases.

Côté et al. (1989) attributed these nitrogen bubbles to an observed decrease in oxygen
mass transfer coefficient when air is used as the aeration gas. The researchers suggested

78
that these bubbles were responsible for stripping of oxygen at the membrane surface,
preventing diffused oxygen from entering the bulk fluid.

4.2.2 Calculation of average oxygen flux

Average oxygen flux was obtained from the experimental data using Equation 4-1:

( DODO5  DODO1 )
J
a(t DO5  t DO1 ) Equation 4-1

Where: 𝐽 ̅ = average oxygen flux (gO2m-2h-1)

tDO=5 = time at which dissolved oxygen = 5 mg l-1 (h)
tDO=1 = time at which dissolved oxygen = 1 mg l-1 (h)
a = specific surface area (m2m-3)

The average flux reported in this study was calculated between dissolved oxygen
concentrations of 1 mg l-1 and 5 mg l-1, chosen to exclude any instability during the early
stages of experimental runs and the effect of initial oxygen concentration, which varied
from experiment to experiment for practical reasons.

4.2.3 Calculation of mass transfer coefficient

The overall mass transfer coefficient, K, was calculated from experimental data using a
linearized form of Equation 2.5:

 C *  Ct 
ln  *
   Kat
 C  Equation 4-2

Where: C* = saturation oxygen concentration at experimental conditions (mg l-1)

Ct = oxygen concentration at time t (mg l-1)
a = specific surface area (m2m-3)

79
 t = time (s)
K = overall oxygen mass transfer coefficient (ms-1)

𝐶 ∗ −𝐶𝑡
A plot of ln ( ) versus t yields a straight line through the origin with a gradient
𝐶∗

of – 𝐾𝑎, as shown in Figure 4-5. This plot is typical of those obtained for all experiments
carried out in the investigation. Again the interval between DO = 1 mg l-1 and DO = 5
mg l-1 was used for comparison purposes.

time (s)
0 500 1000 1500 2000 2500 3000 3500 4000
0

-0.1

-0.2
ln ((C*-Ct)/C*)

-0.3

-0.4

-0.5

-0.6

-0.7

𝑪∗ −𝑪𝒕
Figure 4-5: Plot of 𝒍𝒏 ( ) versus t (Silicone Rubber, 0.5 bar inlet pressure, 2 lpm gas side
𝑪∗
flowrate, 550 ml min-1 water side flowrate)

4.3 Effect of air side flowrate

The effect of the air side flowrate on the oxygen transfer through the two membrane
materials was investigated at inlet flowrate values between 0.6 and 5 litres of air, an inlet
pressure of 0.5 bar and a water side flowrate of 550 ml min-1. Although not directly
measured, mass velocity, G, is the true independent variable being investigated in this
experimental work, as calculated by Equation 4-3 (Darby, 1996). Mass velocity is
related to pressure drop as shown in Equation 4-4, also given by Darby. Different air
flowrates were chosen for the two membrane modules in order to give a constant
pressure drop for comparison purposes.

80
 
m V Equation 4-3
G 
A NA

P12  P22  G 2 Equation 4-4

Where: V = Volumetric air flowrate (m3s-1)


m = Mass flowrate of air (kgs-1)
 = Air density (kgm-3)
A = Cross section of tube (m2)
N = Number of tubes
P1 = Inlet pressure (kgm-2)
P2 = Exhaust pressure (kgm-2)

a) Average Oxygen flux

The average fluxes obtained over the range of air side flowrates used in the investigation
are shown in Figure 4-6.

1.8
1.6
1.4
1.2
J (gO2m-2h-1)

1.0
0.8 SR

0.6 PES

0.4
0.2
0.0
0 1 2 3 4 5 6

Inlet Flowrate (lpm)

Figure 4-6: Effect of air side flowrate on average oxygen flux (0.5 bar inlet pressure, 550 ml min -1
water flowrate – minimum and maximum value error bar)

81
Figure 4-6 shows that air side flowrate had no significant effect on the oxygen flux
through either the silicone rubber or polyestersulfone membranes used in this study. Gas
flowrate is known to have a proportional effect on pressure drop (Darby, 1996), and
therefore, it may have been expected that increasing the flowrate would have a negative
effect on mass transfer, as, at higher flowrates, higher pressure losses would be
experienced, leading to a drop in the driving force. However, in practice, the pressure
drop along the length was negligible along the short length (0.19 m) of the polymer
tubes, and the total pressure drop (difference between inlet and outlet) was constant at a
level of approximately 0.05 bar throughout all experiments.

Additionally, Figure 4-6 does not reveal any evidence of inhibition of oxygen transfer
due to condensation of back diffused water vapour on the inner surface of the membrane
fibres. Fang et al. (2004), state that condensation is inevitable in hollow fibre
membranes; the mass transfer coefficient will be greater for water vapour than for gases
such as oxygen or nitrogen due to relative molecule sizes. They suggest however, that
the effects of condensation can be minimised by operating with sufficient gas flow to
enable condensate to be discharged from the module. The absence of condensation in
this study indicates that the range of flowrates used was sufficient to act as aeration and
sweep gas, preventing significant oxygen mass transfer inhibition.

82
b) Mass Transfer Coefficient
Figure 4-7 shows the effect of gas side inlet flowrate on overall mass transfer
coefficient.

70

60

50
K x106 (ms-1)

40
SR
30
PES
20

10

0
0 1 2 3 4 5 6

Inlet Flowrate (lpm)

Figure 4-7: Effect of air flowrate on overall mass transfer coefficient (0.5 bar inlet air pressure, 550
ml min-1 water flowrate – minimum and maximum value error bars)

Variation in the air inlet flowrate caused no discernable trend over the ranges tested on
the overall mass transfer coefficient of oxygen transfer to water through either silicone
rubber or polyethersulphone hollow fibre membranes. The value of the mass transfer
coefficient was constant at approximately 25x10-6 ms-1 for the polyethersulphone
membrane module and 58x10-6 ms-1 for the silicone rubber membrane module. The lack
of a significant effect on the overall mass transfer coefficient justified the omission of a
term describing the mass transfer resistance of the gas side boundary layer in Equation
2-1.

4.4 Effect of inlet air pressure

The effect of inlet air pressure on the oxygen transfer through the two membrane
materials was investigated using the pressure and air flowrate pairs shown in Table 4-3.

83
Volumetric air flowrate was varied in addition to inlet pressure to ensure the mass
velocity was constant in all experiments.

Table 4-3: Pressure flowrate pairs used in inlet air pressure investigation
Inlet pressure (bar) Inlet air flowrate (lpm)
0.20 5.0
0.25 4.0
0.40 2.5
0.50 2.0
0.80 1.25
1.0 1.0
2.0 0.5 (silicone rubber only)

a) Average Oxygen flux

The average oxygen obtained fluxes are shown in Figure 4-8.

2.5
y = 0.643x + 1.11
R² = 0.99
2
J (gO2m-2h-1)

SR
1.5 PES

1
y = 0.072x + 0.70
0.5 R² = 0.68

0
0 0.5 1 1.5 2 2.5
Pinlet (bar)

Figure 4-8: Effect of inlet pressure on average oxygen flux (2 lpm air flowrate, 550 ml min -1 water
flowrate, minimum and maximum value error bars)

Figure 4-8 shows a directly proportional relationship between inlet pressure and
obtained oxygen flux. This trend is expected as higher inlet pressure leads to a greater
gas side oxygen concentration and hence a larger driving force. Higher oxygen fluxes
are obtained with larger driving forces as shown by Equation 2-6.

84
The trace for polyethersulphone does not display as strong a relationship; the pressure
coefficient is much lower and the correlation coefficient is very poor. This indicates that
oxygen mass transfer is not limited by gas side oxygen concentration and instead is
limited by the oxygen concentration in the membrane itself.

A possible explanation for this is that the membrane material is hydrophilic; the
membrane pores are filled with water and limited by the diffusion rate of oxygen in the
water filled pores. As the diffusivity of oxygen is much slower in water than it is in air,
this results in slower mass transfer rates (Li et al., 2010). This explanation is discussed
further in Section 0c.

b) Mass Transfer Coefficient

The effect of inlet air pressure on calculated overall mass transfer coefficients is shown
in Figure 4-9.

70

60

50
Kx106 (ms-1)

40
SR
30
PES
20

10
0
0 0.5 1 1.5 2 2.5
Pinlet (bar)

Figure 4-9: Effect of inlet air pressure on mass transfer coefficient (2 lpm air flowrate, 550 ml/min
water flowrate, minimum and maximum value error bars)

No statistically significant effect of inlet air pressure on the mass transfer coefficient was
observed. A similar result was obtained by Côté et al. (1989), although a decrease in
mass transfer coefficient, attributed to increased significance of bubble formation, was
observed at oxygen partial pressures above the range included in this study.

85
4.5 Effect of water side flowrate
The effect of water side flowrate on the oxygen mass transfer characteristics of the two
membrane materials was investigated over the flowrate range 150-550 ml min-1.

a) Average Oxygen Flux

The average oxygen fluxes obtained over the water flowrate range are shown inFigure
4-10.

1.4

1.2

1.0
J (gO2m-2h-1)

0.8

0.6 SR
PES
0.4

0.2

0.0
0 100 200 300 400 500 600

Q (ml min-1)

Figure 4-10: Effect of water flowrate on average oxygen flux (0.5 bar air pressure, 2 lpm air
flowrate, minimum and maximum value error bars)

Higher oxygen fluxes were obtained at higher water flowrates. This result is to be
expected as it is in accordance with the well-established theory that the size of the
boundary layer, and hence the resistance to mass transfer, decreases with increasing
water velocity.

86
b) Overall mass transfer coefficient
The effect of inlet air pressure on calculated overall mass transfer coefficients is shown
in Figure 4-11.

60

50

40
K x106 (ms-1)

30
SR

20 PES

10

0
0 100 200 300 400 500 600
Q (ml min-1)

Figure 4-11: Effect of water flowrate on overall mass transfer coefficient (0.5 bar inlet air pressure,
2 lpm air flowrate, minimum and maximum value error bars)

Both membrane modules showed an increase in overall mass transfer coefficient with
increased water side flowrate. Again this is related to the decrease in the size of the
boundary layer on the shell side of the shell-and-tube mass exchanger.

c) Obtaining individual membrane mass transfer coefficients

The membrane mass transfer resistance can be calculated using Equation 4-5 for dense
silicone rubber, adapted from Côté et al. (1989), and Equation 4-6 for hydrophilic
microporous polyethersulphone (Vladisavljevic, 1999).

d 
d e ln  e 
Equation 4-5
1
  di 
KM 2 DM

1 (d  d i )
 e
KM 2 Dw  Equation 4-6

87
Where: KM = membrane mass transfer coefficient (ms-1)
DM = diffusivity of oxygen in membrane material (m2s-1)
Dw = diffusivity of oxygen in water (m2s-1)
de = external diameter of membrane fibre (m)
di = internal diameter of membrane fibre (m)
ε = membrane porosity
τ = membrane tortuosity

For calculating membrane resistance of hydrophobic membranes, the term D w is

replaced with Da, the diffusivity of oxygen in air.

Alternatively, having calculated overall oxygen mass transfer coefficients from the
experimental data, a Wilson plot can be constructed in order to obtain the membrane
mass transfer resistance. The Wilson plot, originally developed to obtain experimentally
heat transfer coefficients in heat exchangers (Fernandez-Seara et al., 2005), involves
plotting the inverse of the water side flowrate versus the inverse of the overall mass
transfer coefficient. This has been previously used in work by Vladisavljevic (1999) to
determine individual resistances in hollow fibre membrane contactor systems.

The y-intercept of the plot represents a theoretical point of infinite water flowrate where
the liquid film resistance, 1/KL, equals zero. The value of y-intercept is therefore equal
to the value of the membrane resistance, 1/KM.

88
 50
45
R² = 0.91
40
PES
35
K-1 x103(s/m)

30
25
R² = 0.95 SR
20
15
10
5
0
0 100000 200000 300000 400000 500000
Q-1 (s/m3)

Figure 4-12: ‘Wilson plot’, minimum and maximum value error bars

The membrane mass transfer resistances obtained from the Wilson plot and calculated
by Equation 4-5 and Equation 4-6 are presented in Table 4-4. The calculated values are
displayed in the middle column and the graphically obtained values in the right hand
column.

Table 4-4: Experimental and calculated membrane mass transfer coefficients

1/KM(calc) (m-1s) 1/KM(exp) (m-1s)
SR 33950 18610
PES 27030 32540

The 1/KM value for silicone rubber found experimentally is of the same order as the
calculated value and relates very closely to the value of 52.6 x10-6 m-1s calculated by
Côté et al. (1989), for the mass transfer coefficient for silicone rubber membranes of
similar thickness and oxygen permeability to that used in this study.

The value obtained for polyethersulphone is much higher than those obtained in
literature for hydrophobic microporous membranes; indeed Yang and Cussler (1986)
state that the membrane resistance of a hydrophobic material is negligible. This indicates

89
the membrane used in this study was hydrophilic, verifying the suggested explanation
for the observed results with variations in inlet pressure obtained with the
polyethersulphone membrane module. Table 4-5 compares the obtained values to those
obtained in literature.

Table 4-5: Comparison of membrane mass transfer coefficients

Membrane material 1/KM (m-1s)
This study PES 32540
Vladisavljevic (1999) hydrophilic PES 21200

The membrane mass transfer resistance is very similar to that found by Vladisavljevic,
with hydrophilic polyethersulphone membranes of similar dimensions to those used in
this study, supporting the assertion that the polyethersulphone membranes used in this
study were hydrophilic.

Using the information from the Wilson plot and the experimentally obtained values of
K, the liquid side mass transfer coefficient, KL, was obtained from Equation 4-7, a
rearranged form of Equation 2-1:

1
1 1 
K L     Equation 4-7
 K K M 

The liquid film resistance can then be plotted in its dimensionless form, the Sherwood
number, Sh, against the dimensionless Reynolds number, Re, on a logarithmic scale,
with the dimensionless numbers being calculated using Equation 4-8 and Equation 4-9.

Sh  K L d
Equation 4-8
D

90
 Qd
Re 
A(1   )  Equation 4-9

Where: d = characteristic length (m)

A = flow area (m2)
φ = membrane packing density
Q = volumetric flowrate of water (m3s-1)
μ = water viscosity (Pa.s)
ρ = water density (kgm-3)

3.5

3.0

2.5

2.0
log Sh

SR
1.5
PES
1.0

0.5

0.0
0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6
log Re

Figure 4-13: Sherwood number versus Reynolds number (0.5 bar, 2 lpm inlet flow, minimum and
maximum value error bars)

A least squares regression of the data in Figure 4-13 gives the following correlations
with correlation fit coefficients in excess of 0.97.

Table 4-6: Obtained Sherwood number/ Reynolds number relationships

Correlation Correlation coefficient
SR: Sh  543.3 Re1.07 0.97 Equation 4-10
PES: Sh  106.9 Re1.05 0.99 Equation 4-11

91
The exponent on Reynolds number is approximately equal to 1.00 for both the silicone
rubber and polyethersulphone membrane modules – explaining the pseudo-proportional
nature of Figure 4-11.

In this study, the effect of the Schmidt number, Sc, was not evaluated. A power
dependence of 0.33 is widely accepted in literature (Vladisavljevic, 1999, Yang &
Cussler, 1986, Ahmed & Semmens, 1996, Coté et al., 1989), and setting the exponent to
this value allows the development of the following relationships incorporating Sc:

SR: Sh  1.99 Re1.07 Sc 0.33 Equation 4-12

PES: Sh  0.96 Re1.05 Sc 0.33 Equation 4-13

The exponent on the Reynolds number is similar in both cases, reflecting the geometric
similarity of the two modules. The variation in the exponents may be related to the slight
difference in the voidage fraction caused by differences in fibre diameters. Table 4-7
compares the obtained relationships to those contained in literature.

Table 4-7: Mass transfer correlations for hollow fibre modules

Study Voidage Fraction Relationship
Schöner et al. (1998) 0.51 𝑆ℎ = 1.76𝑅𝑒 0.82 𝑆𝑐 0.33
This study 0.945 𝑆ℎ = 1.99𝑅𝑒 1.07 𝑆𝑐 0.33
This study 0.957 𝑆ℎ = 0.96𝑅𝑒 1.05 𝑆𝑐 0.33
Yang & Cussler (1986) 0.97 𝑑 0.93
𝑆ℎ = 1.25 (𝑅𝑒 𝑙 ) 𝑆𝑐 0.33

Li et al. (2010) 0.994 𝑆ℎ = 0.20𝑅𝑒 1.03 𝑆𝑐 0.33

The value of the coefficient is different for the two different membranes modules. Zheng
et al. (2005), link this to membrane porosity, γ, arguing that mass transfer in
microporous membranes takes place through the membrane pores only, and therefore the
pore area only should be considered in calculation of the membrane specific surface
area. Applying this analysis to the polyethersulphone membrane modules, using the

92
manufacturer’s supplied porosity yields Equation 4-14; a modified form of Equation
4-13.

Sh  3.20 Re1.05 Sc 0.33 Equation 4-14

The lack of agreement between this relationship and Equation 4-12 for the silicone
rubber membrane module implies that the assertion by Zheng et al. (2005) that mass
transfer takes place through the membrane pores only is not correct. A more likely
explanation is that mass transfer is a combination of both diffusion through the water
filled pores and a solution diffusion mechanism through the membrane wall
(Vladisavljevic, 1999).

Further comparisons between this study and others are difficult. An observation
previously made by Yang & Cussler (1986) - fluid flow through the shell of a shell-and-
tube heat or mass exchanger is extremely complex and no fundamental mathematical
description of its nature exists (Zheng et al., 2005).

As a result of this, the literature contains multiple methods of developing the

relationship involving different calculations of Reynolds number (i.e. different
characteristic length), different Reynolds number ranges, use of different dimensionless
groups, incorporating membrane fibre length (Yang & Cussler, 1986), incorporating
membrane porosity (Zheng et al., 2005), and incorporation of pressure (Ahmed et al.,
2004).

However, comparing the relationships obtained in this study to those contained in

literature for other parallel flow, hollow fibre membrane modules serves the purpose of
establishing the reliability of the data obtained. Although different relationships were
found, the differences can be explained by the difference in the membrane materials and
modules geometries used in the various investigations.

93
4.6 Conclusions

The oxygen mass transfer characteristics of silicone rubber and polyethersulphone

membranes were ascertained and compared at different values of inlet pressure, air
flowrate and water side flowrate.

Higher oxygen transfer rates were obtained at all parameter values with silicone rubber,
attributed to the high permeation rates of oxygen in silicone rubber and the hydrophilic
nature of the PES membranes used in this study.

Mass transfer coefficients and oxygen fluxes were found to increase with increasing
inlet pressure and increasing liquid side flowrate. Airside flowrate was not found to have
a statistically significant effect on mass transfer. These results were all corroborated by
theory and previous research, and justified some of the assumptions used in the data
analysis.

Relationships between the Sherwood number (dimensionless form of the liquid side
mass transfer coefficient) and the Reynolds number (dimensionless form of the liquid
side flowrate) were obtained, with good correlation coefficients, verifying the
experimental approach used.

The values of the constants in the Sherwood number – Reynolds number relationships
differed significantly from those contained in published research. These differences are
explainable by differences in experimental setup, membrane materials studied and data
analysis techniques.

94
5 Membrane Aerated Biofilm Reactor Studies – Part 1
Influence of inlet pressure on MABfR performance

In total, MABfR A was operated continuously and monitored for 18 months. This
chapter describes the first 190 days of operation, when the inlet pressure was varied
between 1.0 and 2.0 bar gauge and a hydraulic retention time (HRT) of 1 day was used.
This section examines the results obtained during this period, and uses the obtained
results to develop a method for obtaining oxygen flux from pollutant removal rates.

5.1 Reactor start-up

The reactor was filled with synthetic waste and inoculated with supernatant obtained
from garden soil in accordance with the procedure outlined in Section 3.2.5. With an
initial inlet pressure of 0.5 barg and HRT of 5 days, visible biomass was observed on the
membrane, membrane support and the walls of the tank within 5 days.

The HRT was then decreased stepwise over a period of 76 days using the values detailed
in Table 5-1, to avoid washout of biomass before membrane attachment occurred.
Throughout this period, the bulk pH was monitored and adjusted if needed via dropwise
addition of 0.1 mol l-1 NaOH solution to keep the pH in the range 6-7.5 and maintain
ideal growth conditions for the bacteria of interest in this study (Semmens et al., 2003).

Table 5-1: Conditions during start-up period

Day Inlet pressure Hydraulic retention time
(bar) (days)
1-16 0.5 5
17-36 0.5 3.5
37-55 0.5 2
56-76 0.5 1.5

95
At the end of the start-up period, the membrane support rack was removed from the
tank, all biomass removed from the walls of the tank, the membrane support rack
replaced and the tank filled with fresh synthetic waste. The HRT was also reduced to 1
day and the inlet pressure increased to 1 bar. Bulk dissolved oxygen was monitored
throughout this period and was observed to fall to below 0.20 mg l-1 after 8 days, after
which the COD and ammoniacal nitrogen removal was considered to be oxygen limited.

5.2 General observations

The COD, ammoniacal nitrogen and total nitrogen removal throughout the 190 days of
operation at 1 day hydraulic retention time are presented and discussed in sections
below. The conditions throughout the 190 days in question are summarised in Table 5-2
below:

Table 5-2: Conditions throughout inlet pressure investigation

Day Inlet pressure (bar) HRT (days)
1-17 1.0 1.0
18-42 1.1 1.0
43-55 1.2 1.0
56-92 1.3 1.0
93-140 1.4 1.0
141-153 1.6 1.0
154-190 2.0 1.0

5.2.1 pH
Influent and effluent pH throughout the 190 days of operation at a HRT of 1 day are
shown in Figure 5-1.

96
 9

7
pH

6 effluent
influent
5

3
0 50 100 150 200
day

Figure 5-1: Influent and effluent pH

A drop in pH was observed through the action of the biofilm. As shown by the half
equations for microbial processes in Chapter 1, consumption of basic carbonate and
ammonia, and production of acidic carbon dioxide and nitrate, occurs in the biofilm
leading to the change in pH. In order to maintain optimal conditions for biofilm growth,
pH was manually adjusted through dropwise addition of 0.1 M sodium hydroxide
solution if the value dropped below 5.0.

Variation in the pH of the influent media is caused by temperature, error in volume of

stock solutions used in preparation of influent media and changes in stock solution
concentration due to degradation etc.

The pH conditions throughout Runs 1-7 are summarized in Table 5-3.

97
 Table 5-3: Average and standard deviation pH
pH
Influent Effluent
Average 7.52 6.00
Standard Deviation 0.24 0.88

5.2.2 Nitrite concentration

Figure 5-2 shows the bulk nitrite concentration throughout the duration of the pressure
investigation.

1.0
bulk nitrite concentration (mg N l-1)

0.9
0.8
0.7
0.6
bulk nitrite
0.5
concentration
0.4
0.3
0.2
0.1
0.0
0 50 100 150 200
day

Figure 5-2: Bulk nitrite concentration (days 1-190)

The nitrite concentration was below the upper detection limit (0.500 mg l-1) at all stages
throughout the investigation, with an average concentration of 0.145 mg l-1 and a
standard deviation of 0.135 mg l-1.

The relatively low values obtained for bulk nitrite concentration (compared to influent
NH4+-N concentrations of approximately 9.5 mg l-1) implies that the majority of
ammoniacal nitrogen in the influent media was completely oxidized to nitrate.

98
This result is not unexpected and can be explained by examining the growth rates of the
two groups of bacteria responsible for nitrification. At ambient temperatures (at which
the investigation was carried out), the growth rate for the most common nitrite oxidising
bacteria, Nitrobacter, is significantly larger than the growth rate for the most common
ammonia oxidizing bacteria, Nitrosomonas (Hellinga et al., 1998). As such, it is
generally the oxidiation of ammonia to nitrite which is the rate controlling step of the
nitrification process, and nitrite accumulation is rare.

5.3 Stoichiometric model development

The half reactions for microbiological respiration given by McCarty (1975) can be
combined to describe the reactions taking place in the MABfR. Constructing an overall
stoichiometric reaction for a bacterial reaction involves combination of three oxidation
half reactions: one for the electron donor (Rd); one for the electron acceptor (Ra); and
one for bacterial cell assimilation (Rc). The overall reaction (R) is then obtained using
Equation 5-1 (McCarty, 1975).

R  Rd  f e Ra  f s Rc Equation 5-1

Where: fe = fraction of electron donor used for energy

fs = fraction of electron donor used for cell formation

In order to achieve a balanced equation:

fe  fs  1 Equation 5-2

fs and fe are functions of cell yield coefficient, cell decay rate, solids retention time
(sludge age) and the biodegradable fraction of microorganisms as described by the
relationship in Equation 5-3.

99
  f bt 
f s  ae 1  d s  Equation 5-3
 1  bt s 
Where: ae = cell yield coefficient
fd = biodegradable fraction of active microorganism
b = cell decay rate (day-1)
ts = solid retention time/sludge age (days)

The cell yield coefficient, ae, is representative of the fraction of electron donor used for
cell synthesis at zero sludge age, and can be calculated from thermodynamic
considerations or determined experimentally. Values of ae are available in literature, as
shown in Table 5-4:

Table 5-4: Values of ae used in model development (McCarty, 1975)

Reaction Value of ae
Aerobic heterotrophy 0.79
Nitrification 0.096
Denitrification 0.36

A value of 0.80 for the biodegradable fraction of biomass, fd, is acceptable for both
aerobic and anaerobic bacteria (McCarty, 1975). Cell decay rates are also available in
literature, as shown in Table 5-5:

Table 5-5: Cell decay rates (Manser et al., 2006)

Reaction Value of b (day-1)
Aerobic heterotrophy 0.28
Nitrification 0.15
Denitrification 0.033

With the values in Table 5-4 and Table 5-5, it is possible to use Equation 5-3 to calculate
the value of the fraction of electron donor, fs, for various values of sludge age, ts. The

100
obtained fs values are displayed in Figure 5-3. A logarithmic scale is used on the
horizontal axis.

1.0
0.9
0.8
0.7 Denitrifcation
0.6
Nitrification
0.5
fs

0.4 Aerobic
Heterotrophy
0.3
0.2
0.1
0.0
1 10 100 1000 10000
ts (days)

Figure 5-3: fs/sludge age relationships

It can be seen from Figure 5-3 that the value of fs reaches a steady value after a sludge
age of approximately 1000 days. Sludge age is not a true measure of biomass residence
time; rather it is the ratio of the mass of organisms in the reactor to the mass of
organisms removed each day, as calculated by Equation 5-4 (Tchobanoglous & Burton,
1991a).

Vr X
ts  Equation 5-4
QX e
Where: Vr = Reactor volume (l)
X= biomass concentration in reactor (mg l-1)
Q= flowrate (l day-1)
Xe = biomass concentration in effluent (mg l-1)

One of the major advantages of the MABfR is the high biomass retention, with biomass
concentrations of 14 gm-2 being observed by Brindle et al. (1998). In this study, the

101
biomass concentration in the effluent was very small, as evidenced by very low turbidity
throughout reactor operation. As the value of X is much larger than Xe, it is appropriate
to select the steady values for fs, as sludge age cannot be accurately measured without
the use of a destructive technique to obtain reactor biomass concentrations. A similar
assumption was made in biofilm modeling work by Shanahan & Semmens (2004).

Using the steady values from Figure 5-3 in combination with the half equations
contained in work by McCarty (1975); Equation 5-5, Equation 5-6 and Equation 5-7 can
be developed to describe the stoichiometry of the relevant microbial processes taking
place in Reactor A.

Aerobic Heterotrophy

CH 2 O  0.032HCO3  0.842O2  0.032 NH 4  Equation 5-5

0.032C5 H 7 O2 N  0.874CO2  0.968H 2 O

Nitrification
NH 4  0.031CO2  0.008HCO3  1.948O2  Equation 5-6

0.008C5 H 7 O 2 N  0.992 NO3  0.985H 2 O  1.985H 

Denitrification
NO3  1.329CH 2 O  H   Equation 5-7

0.014C5 H 7 O2 N  0.493N 2  1.781H 2 O  1.260CO2

Examination of Equation 5-6 and Equation 5-7 shows that, whilst nitrification produces
acidity (1.985 moles of H+ are produced per mole of NH4+ consumed), denitrification
reduces acidity (1 mole of H+ is consumed per mole of NO3- denitrified). This
observation explains the pH regulatory effect of simultaneous nitrification and
denitrification.

102
Conversion of the coefficients in Equation 5-5, Equation 5-6 and Equation 5-7 from a
molar to mass basis leads to the following ratios:

Table 5-6: Mass ratios of substrates involved in microbial reactions in the MABfR
Nitrification
NH4+-N O2 NO3--N Biomass
1g 4.45 g 0.99 g 0.113 g

Denitrification
NO3--N {CH2O} Biomass
1g 2.85 g 0.065 g

Aerobic heterotrophy
{CH2O} O2 Biomass
1g 2.07 g 0.121 g

The values in Table 5-6 are very similar to those published in the literature including the
value of 4.54 gO2 per gNH4+-N completely oxidised to nitrate contained in work by
Brindle et al. (1998); and the value of 4.57 gO2 per gNH4+-N used in a more recent study
by Hasar et al. (2008) in which biomass synthesis was considered negligible. The ratio
of 2.07 g of O2 per g of COD for aerobic heterotrophy is comparable to the 2.11 g of
oxygen per g of COD (presented by the authors as 0.473 g of COD per g of O2) used by
Shanahan & Semmens (2004), and the stoichiometric NO3--N:COD value for
denitrification is very similar to the value of 2.86 used by in a modelling study by
Matsumoto et al. (2007).

103
5.4 Pollutant removal

5.4.1 Removal Efficiency

The COD and ammoniacal nitrogen removal efficiencies throughout Runs 1-7 are shown
in Figure 5-3.

100
90
80
Removal efficiency (%)

70
60
COD
50
Ammonia
40
30
20
10
0
1.0 1.1 1.2 1.3 1.4 1.6 2.0
Inlet pressure (bar)

Figure 5-4: Pollutant removal efficiency in Runs 1-7 (standard error bars)

Figure 5-4 shows that, in general, higher pollutant removal efficiencies are achieved at
higher inlet pressures, attributable to greater availabilities of oxygen achieved at higher
pressures.

The absence of a discernible trend may be attributed to some of the experimental runs
being long enough for bulk oxygen levels to approach zero (allowing oxygen flux to be
calculated from pollutant removal), but not long enough for the biofilm to reach steady-
state. Due to their location within the biofilm, the increase in oxygen availability is first
exploited by nitrifying bacteria, leading to a disproportionately high ammoniacal
nitrogen removal. The reasons for this occurring are discussed in greater detail in
Section 5.5.

104
Additionally, the standard errors for COD removal are larger than those for ammoniacal
nitrogen removal. There are two possible explanations for this observation. As growth
rates for heterotrophic bacteria are larger than those for nitrifiers (Semmens et al., 2003),
respiration of any entrained biomass in the collected samples will have a greater effect
on COD concentration than ammoniacal nitrogen concentration.

Table 5-7: Comparison of pollutant removal efficiencies

Author HRT a % COD % NH4+-N
(day) (m2m-3) removal removal
Timberlake (1988) 1.1 0.19 55.9 70.4
Pankhania et al. (1994, 1999) 0.042 510 89 44.4
Yamagiwa & Ohkawa (1994) 0.5 24.6 >95 90
Semmens et al. (2003) 0.25 422 >95 90
This study 1 5.52 76.1 76.9

The removal efficiencies obtained in this study do not compare favourably with those
achieved in other studies where simultaneous COD and ammoniacal nitrogen removal
was obtained (Table 5-7). Some studies report ammoniacal nitrogen and COD removal
efficiencies of 90% and greater, but these studies used a different HRT, specific
membrane area and pollutant loadings.

The ammonia removal levels in this study compare favourably with other studies
treating wastewaters containing both COD and ammonia. Semmens et al. (2003),
achieved approximately 90% removal, but the very high level of removal achieved by
the researcher can be attributed to the very high specific membrane area used in their
study.

Pollutant removal efficiency is a useful measure in comparing the effect of different

parameter values within one pollutant treatment setup, but it is of limited use in making
comparison between different setups. A better comparison between studies can be made
by considering the pollutant removal rates on a specific membrane area basis.
105
5.4.2 Removal rates
The removal rates of COD and ammoniacal nitrogen on a specific membrane area basis
are shown at the different inlet pressures used in Figure 5-5.

12 6
COD removal rate (gCOD m-2day-1)

NH4+-N removal rate (gN m-2day-1)

10 5

8 4 COD
amm-N
6 3

4 2

2 1

0 0
0 0.5 1 1.5 2 2.5
Inlet pressure

Figure 5-5: Effect of inlet pressure on pollutant removal rates (standard error bars)

Figure 5-5 shows a general upward trend, with the highest pollutant removal rates being
achieved at the highest inlet pressure, where the availability of oxygen is greatest.

The removal rates achieved at 2.0 bar inlet pressure, are compared with those in
published literature in Table 5-8.

Table 5-8: Comparison of pollutant removal rates

Author HRT a rCOD ramm-N
(day) (m2m-3) (gCOD m-2day-1) (gN m-2day-1)
Timberlake (1988) 1.1 0.19 0.19 0.04
Pankhania et al. (1994, 1999) 0.042 510 15.08 n/a
Yamagiwa & Ohkawa (1994) 0.5 24.6 6.3 2.2
Semmens et al. (2003) 0.25 422 10 2
This study 1 5.52 10.2 1.32

106
When compared on a specific surface area reaction rate basis, the results obtained in this
study relate well with those contained in previously published research. Using the same
sample of publications as in Table 5-7, the COD and ammoniacal nitrogen removal rates
range from 0.19 – 15.1 g m-2day-1 and 0.04 – 2.2 g m-2day-1 respectively compared to the
10.2 gCOD m-2day-1 and 1.32 gN m-2day-1 obtained in this study.

Pankhania et al. (1994, 1999) do not give details of the ammoniacal nitrogen removal
rates in their study. The authors commented on the absence of nitrate being detected in
the effluent and attributed this to the low pH of operation which is likely to have been
inhibitory to nitrification. Additionally, the synthetic waste used in their study contained
a high COD:Amm-N ratio (~20) and at these conditions aerobic heterotrophs
significantly outcompete nitrifiers for available oxygen (discussed further in Chapter 6),
explaining the relatively high COD removal rate achieved.

The poor performance of the Timberlake (1988) study can be attributed to the low
pressure (0.2 barg) and hence low oxygen availability at the experimental conditions.

5.4.3 Total Nitrogen Removal

Nitrogen removal from wastewater is the result of two bacterial processes: assimilation
by heterotrophs and denitrification (Yamagiwa & Ohkawa, 1994). The extent of total
nitrogen removal achieved in Runs 1-7 is shown in Figure 5-6.

107
 100
90
80
70
% influent Total-N

60
50 Nit-N

40 Amm-N

30
20
10
0
1.0 1.1 1.2 1.3 1.4 1.6 2.0
inlet pressure (bar)

Figure 5-6: total nitrogen removal in Runs 1-7 (standard error bars)

The greatest extent of total nitrogen removal in Runs 1-7 was achieved in Run 7, where
the inlet pressure was highest. The greater availability of oxygen due to greater inlet
pressure leads to an increase in the nitrification rate. As a result, more nitrate was
available for denitrifiers, hence the greater total nitrogen removal is achieved.

The rate of denitrification can be calculated using effluent nitrate concentrations and the
relationship between ammoniacal nitrogen removal and nitrate production derived from
stoichiometry (Table 5-6). The denitrification rate is obtained using
Equation 5-8.

C nit N (inf)  0.99(C amm N (inf)  C amm N ( eff ) )  C nit N ( eff )

rdenitrification  Equation 5-8
a * HRT

Where: Cnit-N(inf) = Influent nitrate concentration (mg N l-1)

Cnit-N(eff) = Effluent nitrate concentration (mg N l-1)
Camm-N(inf) = Influent ammoniacal nitrogen concentration (mg N l-1)
Camm-N(eff) = Effluent ammoniacal nitrogen concentration (mg N l-1)

The denitrification rates at the different inlet pressures used in the investigation are
shown in Figure 5-7.

108
 0.6

denitrification rate (gNm-2day-1)

0.5

0.4

0.3

0.2

0.1

0.0
0 0.5 1 1.5 2 2.5

inlet pressure (bar)

Figure 5-7: Effect of inlet pressure on denitrification rate

In general the rate of denitrification increases with increasing inlet pressure, as shown by
the trend line, calculated ignoring rogue points at 1.1 and 1.6 bar inlet pressure. This
trend mirrors that obtained between ammoniacal nitrogen removal and inlet pressure and
implies that the denitrification rate is limited by nitrate availability in this situation.

The lower than expected denitrification rate obtained in Run 6 (1.6 bar inlet pressure)
may be attributed to the short duration of this experimental run. As stated previously, the
purpose of this series of experimental runs was to obtain an oxygen limited biofilm and
enough data points to ascertain oxygen flux from pollutant removal – this was achieved
in run 1.6 after only 10 days.

An increase in the inlet pressure leads to the extension of the aerobic zone closer to the
biofilm/bulk interface (Casey et al., 2000b). Any denitrifiers present before the inlet
pressure increase are inhibited from performing denitrification by high oxygen
concentrations, and this function of the biofilm is therefore lost. Due to the relatively
slow growth rate of denitrifying bacteria, the reappearance of significant amount of

109
denitrifying bacteria takes longer than the duration of Run 6, accounting for the lower
than expected denitrification rate observed in this period.

Denitrification rates in excess of those predicted by the trendline were also observed in
run 2 (1.1 bar inlet pressure); again this discrepancy can be attributed to the short
duration of the experimental run. As explained in Section 5.5, disproportionately large
increases in nitrification are observed following an increase in inlet pressure. In run 2,
the increase in inlet pressure did not lead to the loss of the denitrification function;
instead it was boosted by the availability of greater amounts of nitrate, resulting in the
high denitrification rate.

5.5 Response to increased inlet pressure

The bulk dissolved oxygen levels in the MABfR were monitored throughout periods in
which the pressure was increased. Two different types of responses were observed;
during the start-up period, increasing the inlet pressure would lead to an increase of the
bulk dissolved oxygen concentration as oxygen limitation was removed from the
biofilm.

Increasing the inlet pressure increases the availability of oxygen to the aerobic
heterotrophic and nitrifying bacteria present in the biofilm. However, as the growth rates
of aerobic heterotrophs are of the order of 7.3 day-1 compared to 0.6 day -1 for nitrifiers
(Semmens et al., 2003), it may be expected that it is aerobic heterotrophs that first
exploit the greater availability of oxygen. This was not found to be the case in this
investigation, as exemplified by Figure 5-8, which shows the COD and ammoniacal
nitrogen concentrations from day 142 to day 170. The inlet pressure was increased from
1.6 to 2.0 bar on day 154, marked by a dashed line.

110
 80 10

70 9

COD concentration (mgCOD l-1)

Amm-N concentration (mg N l-1)

8
60
7 inf COD
50 6 eff COD
40 5 inf amm-N

30 4
eff amm-N
3
20
2
10 1
0 -
140 145 150 155 160 165 170
day

Figure 5-8: Variation in COD and Ammoniacal nitrogen concentrations in response to changes in
inlet air pressure.

Effluent ammoniacal nitrogen concentrations show a significant decrease following the

pressure increment, whilst effluent COD concentrations initially show an increase before
gradually decreasing towards a steady state value.

This was also noted in work by Zhu (2008). Zhu carried out FISH analysis that verified
modelling results by Shanahan & Semmens (2004) which predicted that the highest
concentration of nitrifying bacteria is found at the membrane-biofilm interface.

As such, nitrifiers are ideally located to exploit the higher availability of oxygen and, in
the short term, a disproportionate increase in the ammoniacal nitrogen removal is
observed. The long term performance of the biofilm is controlled by influent
concentrations, however, as discussed further in Chapter 6.

111
5.6 Apparent oxygen flux
The biofilm was operated under oxygen limited conditions from the end of the start-up
period; all oxygen supplied from the membrane was consumed within the biofilm with
none being transferred into the liquid phase.

Using the equations developed in Section 5.3, the oxygen uptake rate (OUR) can be
calculated using the pollutant removal rates. The OUR can then be used to calculate the
oxygen flux (Casey et al., 1999).

5.6.1 Assumptions
In order to simplify the model the following assumptions are made:
1. All oxygen is supplied from the membrane.
The tank was fitted with a PVC lid in order to reduce air flowrate over the
surface of the water and minimise surface aeration. For the purposes of the
analysis presented here, surface aeration is considered negligible.

2. All oxygen supplied from the membrane is consumed within the biofilm.
The bulk oxygen concentration was periodically measured at values close to zero
throughout the period of operation in question; implying no oxygen from the
membrane reached the bulk fluid. Brindle et al. (1998) studied the oxygen
utilisation efficiency (OUE) of a nitrifying biofilm, and achieved 100% OUE
once the biofilm had reached maturity.

3. All ammonia is utilised aerobically ie. no significant amount of Anaerobic

ammonia oxidising bacteria (AnAOB) are present.
Although AnAOB have been successfully grown in biofilms (Tsushima et al.,
2007), they have extremely low growth rates and require highly specific
conditions in order to obtain any significant amounts of the bacteria. Such
conditions were not available in the reactor in this study. Additionally, work by
Lackner et al. (2008) has shown that ANAMMOX bacteria are significantly

112
 inhibited by the presence of heterotrophs, which predominate in the treatment of
wastewater.

4. All ammonia is completely oxidised to nitrate.

At ambient temperatures, the growth rate of ammonia oxidising bacteria (AOB)
is less than that of nitrite oxidizing bacteria (NOB) (Hellinga et al., 1998).
Therefore, ammonia oxidation is considered the rate controlling step of the
nitrification process. Throughout the experimental studies, the concentration of
nitrite was monitored. At no time was any nitrite accumulation noted (Section
5.2.2), with detected concentrations typically being less than 0.100 mg l-1. A
similar assumption was made by Shanahan & Semmens (2004) in a modelling
study of membrane aerated biofilms.

5. Aerobic heterotrophy and denitrification are the only processes consuming

soluble COD
Although it is possible that stratification can lead to the formation of anaerobic
niches within the biofilm; no characteristic noxious odours and/or black ferrous
sulfide deposits were observed during reactor operation, the absence of which
was the basis of Timberlake et al. (1988) asserting anaerobic activity was
negligible. This finding was supported in modelling work by Shanahan &
Semmens (2004), of membrane aerated biofilms in contact with similar
wastewaters.

6. Complete denitrification takes place

Nitrate is reduced to elemental nitrogen with no loss from the system of NO or
N2O gases. The stoichiometric amount of COD is also removed from the system
through the removal of nitrate. Again, this assumption was also made by
Shanahan & Semmens (2004).

113
5.6.2 Reaction scheme
A simplified reaction scheme is presented in Figure 5-9 for clarity.

Figure 5-9: Simplified reaction scheme

Using the mass ratios from Table 5-6 and the assumptions detailed in Section 5.6.1, the
following procedure is used in order to calculate the oxygen flux from pollutant
removal:

oxygen consumed in nitrification:

OURN  4.45 * ( NH 4  N ( feed )  NH 4  N (eff ) )

nitrate produced in nitrification:

NO3  N ( nitrification)  0.99 * ( NH 4  N ( feed )  NH 4  N (eff ) )

COD removed in denitrification:

COD( denitrification)  2.85 * ( NO3  N ( feed )  NO3  N ( nitrification)  NO3  N (eff ) )

114
COD removed by aerobic heterotrophy:
CODAH  COD feed  CODdenitrification  CODeff

oxygen consumed by aerobic heterotrophy:

OURAH  2.07 * CODAH

When the pollutant removal is calculated as a rate, as given by Equation 3-4, the
apparent oxygen flux can then be obtained using Equation 5-9:

J a  OURN  OURAH Equation 5-9

5.6.3 Effect of inlet pressure

Figure 5-10 illustrates the effect of inlet air pressure on apparent oxygen flux.

1.1

1.0
Apparent flux (gO2m-2h-1)

0.9

0.8

0.7

0.6
0.8 1.0 1.2 1.4 1.6 1.8 2.0

Inlet Pressure (bar)

Figure 5-10: Effect of inlet pressure on apparent oxygen flux (standard error bars)

115
Figure 5-10 shows the existence of a statistically significant relationship between inlet
pressure and apparent oxygen flux, albeit with a relatively poor correlation fit
coefficient. The poor fit and errors in average apparent oxygen flux can be explained by
the errors of analyses and the fact that a biofilm never truly reaches steady state (Casey
et al., 1999).

The maximum observed oxygen flux of approximately 1.0 gO2m-2h-1 at 2.0 barg inlet
pressure is much lower than the average values of approximately 2.4 gO2m-2h-1,
observed in mass transfer studies at similar inlet pressures with silicone rubber as the
aeration material. This occurs despite the fact that the oxygen fluxes in the presence of
the biofilm were obtained with a constant bulk dissolved oxygen concentration
approaching zero, rather than being averaged over the range of 1 – 5 mg l-1 during the
oxygenation experiments.

Comparisons between the obtained fluxes in the biofilm and the mass transfer
experiments are difficult due to the difference in the geometries of the experimental
setups, and the hydrodynamic conditions which result from this difference.

The arrangement of the membranes in the MABfR contained two relatively long tubes
with several bends, giving increased pressure drop between inlet and exhaust (Coulson
et al., 1999a) – the average pressure within the membrane and mass transfer driving
force was therefore lower.

The liquid side bulk turbulence was much lower – only moderate amounts of mixing
were provided by the impeller, giving added significance of the liquid side mass transfer
coefficient as described for low liquid side flowrates in Section 4.6.

Although the presence of a biofilm reduces the significance of the liquid side mass
transfer resistance, it itself represents a mass transfer resistance so that Equation 2-1 has
to be modified to allow for the presence of a biofilm:

116
 1 1 1 1 Equation 5-10
  
K KM KB KL
Where: KB = biofilm mass transfer coefficient (ms-1)

Although microbial respiration maintains the bulk oxygen concentration at values

approaching zero, thus preserving a high concentration difference, the oxygen diffusivity
in a biofilm is lower than in bulk solution (Ahmadi Motlagh et al., 2006). This slow
diffusion away from the membrane surface means that a biofilm will represent a greater
barrier to mass transfer than a liquid boundary layer of the same thickness.

Additionally, the presence of a biofilm on the membrane surface may affect the mass
transfer properties of the membrane material due to adsorption of CO2 and other
respiration products (Côté et al., 1989), increasing the magnitude of the membrane mass
transfer resistance.

Shanahan & Semmens (2006), carried out an investigation comparing oxygen transfer
with and without biofilm present, and also found a reduction in oxygen flux in areas
which experienced small boundary layers, as is expected in operation with hollow fibres,
in accordance with the results presented here.

Although the results presented here display proportionality between inlet pressure and
obtained oxygen flux in the presence of a biofilm, it is likely that this relationship does
not hold for all values of inlet pressure. In operation with high pressure, the membrane
will support a thicker, denser membrane. This thicker biofilm will represent a higher
barrier to mass transfer and, in accordance with Equation 5-10, a lower value of the
overall mass transfer coefficient.

Using linear regression, an empirical relationship (Equation 5-11) between apparent

oxygen flux and inlet pressure can be obtained from Figure 5-10 with a correlation fit
coefficient of 0.92:

117
 J  0.26Pinlet  0.50 Equation 5-11

The above relationship describes the situation in the reactor used in this study, but
cannot be used to predict the oxygen flux in other MABfRs. Using the average absolute
value for pressure in the membrane tubes yields the following relationship, with a
similar fit coefficient to Equation 5-11.

J  0.28Pav( abs)  0.22 Equation 5-12

Where:  P  Poutlet 
Pav( abs)   inlet   Patm Equation 5-13
 2 

Patm = atmospheric pressure (bar)

5.6.4 Mass transfer coefficient

The overall mass transfer coefficient can be found from Equation 5-14, a rearranged
form of Equation 2-6:

J
K  Equation 5-14
C

where: J = Average oxygen flux (gO2 m-2h-1)

K = Overall mass transfer coefficient (mh-1)
ΔC = Concentration difference across membrane (mg l-1)

Assuming that the dissolved oxygen concentration in the bulk phase of the reactor is
zero, the concentration difference can be assumed to be equal to the oxygen
concentration on the gas side of the membrane. As discussed in Section 2.1.1, this
concentration is difficult to accurately determine as diffusion of oxygen and nitrogen
from gas side to water side and back diffusion of water and respiration products (Côté et

118
al., 1988); combined with pressure losses due to friction (Darby, 1996) mean the
concentration changes along the length of the membrane tube.

However, a useful approximation can be made by considering only the pressure losses in
calculating the gas side oxygen concentration. During operation, the membrane fluxes
were of the order of 1 gO2m-2h-1 - small in comparison to the bulk gas mass flowrate
which was approximately 150 g h-1. Assuming, therefore, that the oxygen concentration
can be assessed at the average absolute pressure of the system as given by Equation
5-13, the oxygen concentration can be found using the combined gas law. This also
assumes the gas can be considered ideal as the conditions in question are not close to the
critical conditions (Darby, 1996).

The concentration difference can then be calculated using Equation 5-15 and Equation
5-16 with the values contained in Table 5-9, allowing the overall mass transfer
coefficient to be obtained from Equation 5-14.

PR TexpVm ( R )
Vm (exp)  Equation 5-15
TR Pav( abs)

f O2 M O2
C  Equation 5-16
Vm (exp)

119
 Table 5-9: Values used in calculation of gas side oxygen concentration
Parameter Symbol Value Source
Molar volume at reference Vm 22.41 l mol-1 Rogers & Mayhew (1994)
conditions
Pressure at reference PR 1 atm Rogers & Mayhew (1994)
conditions (1.01325 bar)
Temperature at reference TR 273.15 K (0 °C) Rogers & Mayhew (1994)
conditions
Fraction of oxygen in air fO2 0.2095 Rogers & Mayhew (1994)
Molar mass of oxygen MO2 32.00 g mol-1 Green & Perry (2008)
Temperature at Texp 293.15 K
experimental conditions (20 °C)
Average absolute pressure Pav(abs) Equation 5-13

The effect of inlet pressure on the overall oxygen mass transfer coefficient is shown in
Figure 5-11.

0.40

0.39
K (ms-1 x106)

0.38

0.37

0.36

0.35
0.8 1 1.2 1.4 1.6 1.8 2 2.2
Inlet pressure (barg)

Figure 5-11: Effect of inlet pressure on overall oxygen mass transfer coefficient

Figure 5-11 displays the existence of an inversely proportional relationship between inlet
pressure and overall oxygen mass transfer coefficient, with a correlation fit coefficient of

120
0.99. The proportionality is to be expected; it has already been shown that increased
inlet pressure leads to an increase in oxygen flux (Figure 5-10). This increased
availability oxygen means that more microbial biomass can be supported on the
membrane surface, which increases the resistance to oxygen transfer.

The results obtained from experimental oxygen fluxes and Equation 5-14, Equation 5-15
and Equation 5-16 are displayed and compared to the membrane resistance found in
Chapter 4 in Table 5-10. Units for mass transfer coefficients are given in hm-1 for
agreement with those given previously for oxygen flux.

Table 5-10: Mass transfer coefficients at different experimental conditions

Pin Pav(abs) ΔC K 1 1 1 1
+
(bar) (bar) (mg l-1) (ms-1 x106) 𝐾 𝐾𝑀 𝐾𝐿 𝐾𝐵
(ms x10-6)
-1
(ms-1 x10-6) (ms-1 x10-6)
1.0 1.96 540 39.6 2.53 0.0186 2.51
1.1 2.06 567 39.0 2.56 0.0186 2.54
1.2 2.16 595 38.5 2.59 0.0186 2.58
1.3 2.24 616 38.2 2.62 0.0186 2.60
1.4 2.34 643 37.8 2.65 0.0186 2.63
1.6 2.54 698 37.0 2.70 0.0186 2.68
2.0 2.91 801 35.9 2.79 0.0186 2.77

It can be seen that the combined resistance of the biofilm and liquid boundary layer is
much greater than the membrane resistance, with less than 1% of the resistance being
attributable to the presence of the membrane. This is in agreement with the work of
Picard et al. (2012) with neon diffusion through membrane aerated biofilms and found
the membrane contributed as little as 2% to the overall mass transfer resistance.

This finding has implications for the operation of an industrial MABfR; as the
membrane does not significantly control the oxygen transfer in the presence of a biofilm,
the oxygen permeability of a membrane material can be ignored during the membrane
selection process.

121
5.7 Conclusions
The pollutant removal performance of a lab scale MABfR was monitored over a period
of 190 days at various inlet air pressures. The pollutant removal rates were used to
develop a stoichiometric model from which the apparent oxygen flux, and therefore
mass transfer resistance of the biofilm, could be calculated.

The mass transfer resistance of the biofilm is much greater than that of the membrane
ascertained in Chapter 4. The implication of this is that the oxygen permeability of a
membrane material in a MABfR is not a major consideration in the design of a full scale
MABfR, such as the BioSettler.

122
6 Membrane Aerated Biofilm Reactor Studies – Part 2
Effect of pollutant loading

Chapter 5 assessed the performance of the MABfR though Runs 1-7, where inlet air
pressure was increased from 1.0 to 2.0 barg at a 1 day HRT, and developed a
stoichiometric model allowing the oxygen flux to be estimated from pollutant removal.
This chapter considers the effect of pollutant loading on the pollutant removal
performance of the MABfR.

Examined in this chapter is a period of 230 days during which the MABfR was operated
with a hydraulic retention time of 0.5 days and an inlet pressure of 3.0 barg. During this
period, the COD, ammoniacal nitrogen and nitrate concentration of the influent synthetic
wastewater was varied, and the effect of changing wastewater composition on reactor
performance studied.

Linear regression of the generated data was then employed to produce a series of
relationships which, when used in combination, describe the pollutant removal
performance of the MABfR and can also be used to predict the performance at
conditions outside of those employed in this study.

6.1 Pollutant removal

During the period in question, changes were made to the composition of the synthetic
wastewater fed to the MABfR to achieve variation in the COD, NO3--N loadings and
NH4+-N loadings. These changes were made in such a way as to produce five different
experimental conditions, as summarised in Table 6-1.

123
 Table 6-1: Pollutant loadings in Run 8-12
Run Days Influent COD Influent NH4+-N Influent NO3--N C:N ratio
loading loading loading
(gm-2day-1) (gm-2day-1) (gm-2day-1)
8 1-71 23.8±0.6 4.25±0.08 0.04* 5.6
9 72 – 99 25.6±0.8 3.66±0.13 2.00±0.07 7.0
10 100 - 115 25.7±1.1 2.59±0.01 1.59±0.03 9.9
11 116 – 150 42.4±0.6 2.93±0.06 1.91±0.04 14.4
12 151 – 230 25.1±0.5 2.89±0.10 2.01±0.06 9.1
*no nitrate was added to the feed during Run 8; periodic analysis of influent found nitrate
concentrations below the lower detection limit. For calculation purposes, the loading value
quoted corresponds to the lower detection limit of 0.1 mg l-1.

6.1.1 Chemical Oxygen Demand

The observed influent and effluent COD concentrations during the experimental period
in question are shown in Figure 6-1. For clarity, the vertical lines show the start of each
experimental run (the same device is also used in Figure 6-3 and Figure 6-5).

140

120

100
[COD] (mg l-1)

80
influent
60
effluent
40

20

0
0 50 100 150 200 250
day

Figure 6-1: Influent and Effluent COD concentrations in Runs 8-12

The data displayed in Figure 6-1 illustrates that the effluent COD concentration was
consistently lower than the influent concentration throughout the 230 days of operation

124
examined here. This indicates that successful treatment of this particular pollutant was
achieved in the MABfR.

The data is somewhat scattered, with the effluent data points being more scattered than
the influent data points, reflecting both the inherent inaccuracy of the COD analysis
method (Hach Company, 2007) and the dynamic nature of a biofilm (Khoyi &
Yaghmaei, 2005).

The average influent and effluent COD concentrations and percentage removals for
Runs 8-12 are illustrated in Figure 6-2.

160 80

140 70

120 60
[COD] (mg l-1)

100 50

% removal
Influent
80 40 Effluent
60 30 %removal

40 20

20 10

0 0
8 9 10 11 12
Run

Figure 6-2: Average COD concentrations and percentage COD removal (standard error bars)

Figure 6-2 shows percentage removal in excess of 40% with the highest percentage
removal achieved in Run 12, where 65.9% of influent COD was removed.

Comparisons between runs are difficult to make as COD is removed in the MABfR by
two different microbial processes: aerobic heterotrophy and denitrification. In addition
to limitations in the supply of oxygen, these processes both involve two substrates which
can be rate limiting themselves, meaning that the higher COD consumption rates are not
necessarily obtained at the highest COD loading rates. This is further discussed in
Section 6.2.
125
6.1.2 Ammoniacal nitrogen
The observed influent and effluent ammoniacal nitrogen concentrations during the 230
of operation examined in this chapter are illustrated in Figure 6-3 below.

14

12
[Amm-N] (mg l-1)

10

8
influent
6
effluent
4

-
0 50 100 150 200 250
day

Figure 6-3: Influent and effluent ammoniacal nitrogen concentrations during Runs 8-12

The data displayed in Figure 6-3 indicate that ammoniacal nitrogen removal was
consistently achieved throughout the 230 days of reactor operation. At all data points,
the effluent concentration is lower than the influent concentration, implying that
nitrification is taking place in the biofilm on the surface of the membrane.

In comparison to the COD data shown in Figure 6-1, the data are grouped closer
together. This reflects both the higher accuracy of the method used for measuring
ammoniacal nitrogen concentration (relative to the COD analytical method) and the
structure of membrane aerated biofilms.

As previously discussed, nitrifying bacteria are predominately found at the membrane-

biofilm interface (Shanahan & Semmens, 2004). In this location, they are able to take
advantage of the ‘safety in numbers’ aspect of existing in a biofilm and are more

126
protected from the influence of instability in the bulk liquid than the aerobic
heterotrophs which reside further from the membrane surface (Madigan & Martinko,
2006).

Figure 6-4 illustrates the average influent and effluent ammoniacal nitrogen
concentrations and percentage removals achieved in Runs 8 – 12.

18 60

15 50
[Amm-N] (mg l-1)

12 40
Influent

% removal
9 30 Effluent
%removal
6 20

3 10

0 0
8 9 10 11 12
Run
Figure 6-4: Average ammoniacal nitrogen concentrations and percentage ammoniacal nitrogen
removal in Runs 8 - 12(standard error bars)

Figure 6-4 shows that the highest percentage removal was obtained in Runs 8 and 9,
which are the runs with the highest ammoniacal nitrogen loading rate. Lower effluent
concentration was obtained in Runs 10 – 12, with effluent concentrations in the range
2.5 – 3.7 mg l-1 being achieved.

The higher removal obtained at the higher loading rates is to be expected. Unlike COD,
ammoniacal nitrogen is only consumed in significant quantities by one microbial process
in the MABfR. As nitrifying bacteria are predominately found close to the membrane
surface (Shanahan & Semmens, 2004), oxygen availability can be assumed not to be
limiting to nitrification. It is well established that growth rates of nitrifying bacteria,
and therefore ammoniacal nitrogen consumption rates, increase with increasing

127
ammoniacal nitrogen availability where ammoniacal nitrogen is the limiting substrate
(Shah & Coulman, 1978).

Variation in percentage removal amongst the three runs with similar ammoniacal
nitrogen loadings is due to competition for oxygen between nitrifiers and aerobic
heterotrophs (Zhang et al., 1995).

6.1.3 Total nitrogen

The observed total nitrogen concentrations throughout the 240 days examined in this
section are illustrated in Figure 6-5 below

18
16
14
[Tot-N] (mg l-1)

12
Influent
10
Effluent
8
6
4
2
0
0 50 100 150 200 250
day

Figure 6-5: Influent and effluent Total-N concentrations in Runs 8-12

128
Figure 6-5 shows, that at the vast majority of data points, total nitrogen concentration
was lower in the effluent compared to the influent, implying that total nitrogen removal
was successfully and consistently achieved during the 230 days of operation of the
MABfR. The handful of data points where effluent total nitrogen concentration was
higher than influent concentration can be attributed to human errors which occurred
during the sample collection and/or analysis.

The apparent instability in the total nitrogen removal can also be attributed to the
stratified structure of membrane aerated biofilms. Nitrogen is converted to the gas phase,
and therefore removed from the effluent via the process of denitrification. Previous
research has shown that denitrification occurs in the anoxic area of the biofilm furthest
from the membrane. In this location they are most likely to be sheared into the bulk
phase and are not protected from inhibitory compounds (such as elemental oxygen) by
the mass transfer resistance of the biofilm.

6.2 Removal rates

As discussed in Chapter 5, drawing comparisons with published research is difficult due
to the range of inlet pressures, hydraulic retention times and specific surface areas used.
Better comparisons are made using the removal rates. Table 6-2 gives the removal rates
obtained during Runs 8-12.

Table 6-2: COD and ammoniacal nitrogen removal rates in Runs 8-12 (standard errors)
Run COD:N ratio rCOD ramm-N
(gCOD m-2day-1) (gN m-2day-1)
8 5.6 13.8±1.0 1.92±0.08
9 7.0 14.6±1.8 1.64±0.06
10 9.9 11.0±1.8 1.29±0.02
11 14.4 21.5±1.7 1.59±0.06
12 9.1 16.9±1.1 1.83±0.14

129
Examination of the data in Table 6-2 reveals that the highest rate of COD removal was
obtained in Run 11, the run with the highest COD:N ratio. This can be attributed to the
higher growth rate, and therefore higher activity rate of bacteria which utilize COD for
respiration, in line with Monod kinetics (Shah & Coulman, 1978).

The highest rates of ammoniacal nitrogen removal were obtained in Runs 8 and 12.
These high rates can be attributed to the highest ammoniacal nitrogen loading in Run 8
(Shah & Coulman, 1978) and the low COD loading in Run 12 allowing aerobic nitrifiers
to compete for oxygen more favourably (Zhang et al., 1995).

Table 6-3 compares the results contained in literature to those generated by Run 8 of this
study. Run 8 is chosen as no nitrate was added to the influent synthetic wastewater in
this run, in common with the studies chosen for comparison.

Table 6-3: Comparison of pollutant removal rates (standard errors where shown)
Author COD:N ratio rCOD ramm-N
(gCOD m-2day-1) (gN m-2day-1)
Timberlake (1988)a 2.6 – 5.1 1.9 – 4.2 0.1 – 0.6
Pankhania et al. (1994, 1999) 14.0 – 21.3 15.1 n/a
Yamagiwa & Ohkawa (1994) 2.8 6.3 1.7 - 2.2
Semmens et al. (2003) 4.3 – 4.5 10 2
This study (Run 8) 5.6 13.8±1.0 1.92±0.08
a
The researcher used Total Organic Carbon as their measure rather than COD. TOC and COD
removal rates are assumed here to be equal for comparative purposes.

6.2.1 Oxygen consumption

The apparent oxygen flux was calculated from experimental pollutant removal rates,
using the same method as described in Chapter 5. The apparent oxygen flux throughout
the operational period in question is shown in Figure 6-6, alongside the predicted
oxygen flux as given by Equation 5-11 developed in Chapter 5.

130
 3.0

apparent oxygen flux (gO2 m-2h-1)

2.5

2.0
experimental
1.5 average
predicted
1.0

0.5

-
0 50 100 150 200 250
day

Figure 6-6: Apparent oxygen flux during Runs 8-12

In Figure 6-6 the red dashed line represents the oxygen flux as predicted by Equation 5-
11 developed in the previous chapter. The average oxygen fluxes obtained
experimentally and obtained through use of the previously developed relationships are
summarised in Table 6-4.

Table 6-4: Obtained and predicted average oxygen fluxes

Sample average Equation 5-11 Equation 5-12

Oxygen flux (gO2 m-2h-1) 1.15 1.28 1.27

Various reasons exist as to why some of the experimentally obtained oxygen fluxes are
higher than predicted, in addition to errors of analysis. Sloughing events, where areas of
biofilm become detached from the membrane, occur periodically (Chambless & Stewart,
2007). These detachment events leave areas of the membrane temporarily exposed;
greater oxygen transfer then occurs through the exposed membrane and is utilised by
both the biofilm and free swimming bacteria, contributing to pollutant removal.

Several studies (e.g. Pankhania et al., 1999) have reported a drop in performance
associated with the development of thick biofilms, and as such prevention of excessive

131
biofilm growth in membrane attached biofilm reactors has become the focus of several
researchers (e.g. Hwang et al., 2010).

Figure 6-6 shows no evidence of the existence of a relationship between apparent

oxygen flux changing influent composition over the duration of the 230 days in
question. The absence of a discernible trend implies that influent concentration has
negligible effect on the thickness and density of a biofilm associated with an aeration
membrane over the range of wastewater compositions used in this study.

The average oxygen fluxes obtained during each of the experimental runs in question in
this chapter are shown in Table 6-5 below. The average values were obtained using
median analysis as in Chapter 5 to allow for uneven data sizes.

Table 6-5: Average oxygen fluxes in runs 8-12

8 9 10 11 12
Run
Oxygen flux 1.22 1.28 1.62 1.85 1.28
(gO2 m-2h-1)

Performing an ANOVA on this data reveals that there is no statistically significant effect
of wastewater composition on the obtained oxygen flux. This is important to the model
development described in Section 6.3, as it justifies that oxygen flux can be related to
intra membrane pressure (Casey et al., 2000b), over the range of wastewater
compositions used in this study.

6.2.2 Relative oxygen consumption

In the MABfR, oxygen is consumed as a result of aerobic heterotrophy and aerobic

nitrification. Recent experimental (Meng et al., 2008) and modelling studies (Matsumoto
et al., 2007) based on the operation of a MABfR in plug flow mode, have proposed that

132
the relative consumption of oxygen in an oxygen limited situation is controlled by the
influent COD:N ratio.

Using the stoichiometric relationships developed in Chapter 5, the rates of aerobic

heterotrophy and nitrification can be estimated in the same way that was previously used
to obtain the oxygen uptake rate (OUR) and hence oxygen flux. This data is displayed in
Table 6-6.

Table 6-6: Rates of microbial processes and oxygen uptake (mean values, standard errors)
Run 8 9 10 11 12
rnit 1.92±0.07 1.64±0.07 1.28±0.21 1.59±0.21 1.83±0.25
g m-2day-1

OURnit 8.53±0.33 7.32±0.30 5.72±0.96 7.06±0.94 8.14±1.13

rAH 9.8±1.1 11.5±1.6 9.9±1.5 19.8±1.8 12.8±1.7
OURAH 20.3±2.2 23.5±3.4 20.4±3.1 40.9±3.7 26.6±3.6

From the data displayed in Table 6-6, the fraction of oxygen supplied utilised by aerobic
heterotrophs can be found using Equation 6-1.

OURAH
f AH 
(OURAH  OURnit ) Equation 6-1

Where: fAH = Fraction of oxygen utilized by aerobic heterotrophs

Table 6-7 displays the fraction of supplied oxygen at the five different COD:ammoniacal
nitrogen ratios used in this section of the investigation. The COD:ammoniacal nitrogen
ratio is calculated using Equation 6-2.

133
 [𝐶𝑂𝐷𝑖𝑛𝑓 ]
𝐶𝑂𝐷⁄𝑁𝑟𝑎𝑡𝑖𝑜 = Equation 6-2
[𝐴𝑚𝑚 − 𝑁𝑖𝑛𝑓 ]

Where: COD/Nratio = COD:ammoniacal nitrogen ratio

[CODinf] = Influent COD concentration (mg l-1)
[Amm-Ninf] = Influent Ammoniacal nitrogen concentration (mg l-1)

Table 6-7: Oxygen usage in Runs 8-12 (mean values)

Run COD/Nratio fAH fnit
8 5.6 0.70 0.30
9 7.0 0.76 0.24
10 9.9 0.78 0.22
11 14.4 0.85 0.15
12 9.1 0.77 0.23

The data presented in Table 6-7 and plotted in Figure 6-7, illustrates the relationship
between the fraction of supplied oxygen utilised by aerobic heterotrophs and COD:N
ratio.

0.9

0.8

0.7
f AH

0.6

0.5

0.4
5.0 7.5 10.0 12.5 15.0

COD:amm N ratio

Figure 6-7: fraction of supplied oxygen utilized by aerobic heterotrophs at different COD:Amm-N
ratios

134
The linear trend visible in Figure 6-7 suggests that the COD:N ratio also controls the
relative consumption of oxygen in a mixed flow reactor, as used in this study. A higher
fraction of oxygen is utilised by aerobic heterotrophs at higher COD:N, where their
higher growth rates allow them to outcompete nitrifiers for oxygen and space in the
biofilm structure (Zhang et al., 1995).

Linear regression of the data in Figure 6-7 yields the empirical relationship Equation 6-3
with a correlation fit coefficient of 0.92.

𝑓𝐴𝐻 = 0.016(𝐶𝑂𝐷⁄𝑁𝑟𝑎𝑡𝑖𝑜 ) + 0.63 Equation 6-3

Where: fAH = Fraction of supplied oxygen utilised by aerobic

heterotrophs
COD/N ratio = gCOD per g Amm-N in influent media

Although the high correlation coefficient implies that Equation 6-3 describes well the
fraction of oxygen used by the competing bacteria types over the range of COD/N ratios
used in the investigation, it is only a linear approximation based on the data generated by
this study. Microbial kinetics are complicated non-linear systems, and the relationship
presented in Equation 6-3 is an estimation of the performance of the MABfR over the
range of wastewaters used in this study.

With hindsight, the COD/N range used in the investigation should have been expanded
in order to obtain the limits of applicability of the developed relationship, and to
ascertain experimentally, if practicable, the loading ratio at which aerobic heterotrophs
will completely outcompete nitrifiers for oxygen. This however, was not possible given
the time constraints placed upon the project, and forms the basis of a section of
suggested future work.

135
6.2.3 Denitrification
The denitrification rates obtained in Runs 8-12 are shown in Table 6-8.

Table 6-8: Denitrification rates in runs 8-12 (mean values, standard errors shown)
Run 8 9 10 11 12
Denitrification rate 1.42±0.10 1.10±0.20 0.40±0.19 0.61±0.12 1.26±0.33
(gN m-2day-1)

Table 6-8 shows that the rate of denitrification rates achieved during stable operation
falls in the range 0.40 – 1.42 gN m-2day-1. This compares favourably with the limited
number of studies of denitrification in a MABfR contained in literature, as seen in Table
6-9.

Table 6-9: Comparison of denitrification rates

Author Denitrification rate Nitrification rate
gN m-2day-1 gN m-2day-1
Timberlake et al. (1988) 0.1 – 0.6 0.1 – 0.6
Semmens et al. (2003) 2.0 2.0
Satoh et al. (2004) 0.12 – 0.33 0.50
Downing & Nerenberg (2008a) 0.23 – 0.32 0.40 – 1.30
This study 0.40 – 1.42 1.28 – 1.92

In the majority of MABfR studies included in Table 6-9, including this study, the
denitrification rate approaches the nitrification rate. Denitrification was reported to
proceed at the same rate as nitrification by Timberlake et al. (1988) and Semmens et al.
(2003), in reactors with effluent COD/TOC concentrations in the range 30-50 mg l-1,
implying that denitrification is controlled by nitrate availability.

The denitrification rates observed in this study are higher than those reported by
Timberlake et al. (1988) and Downing & Nerenberg (2008a). Both studies reported that
denitrification proceeded at the same rate at which nitrate was produced via nitrification.

136
A higher rate (~2.0 gN/m2day) was reported by Semmens et al. (2003) in a reactor
operated with higher strength wastewater than used in this study.

As stated previously, there has been limited research undertaken into the factors
affecting denitrification in bacterial films. In addition to the studies discussed above, a
kinetic study using sequential aerobic and anoxic reactors (for nitrification and
denitrification, respectively), with denitrifying bacteria immobilised on support packing,
found that the highest denitrification rates were obtained at the highest ammoniacal
nitrogen loading rates (Dincer & Kargi, 2000). The authors attributed this to the greater
availability of NOx-N (nitrite and nitrate) to denitrifying bacteria in the anoxic reactor.
Organic carbon loading has been found to only control the denitrification rate in
situations when this pollutant loading was insufficient to obtain complete denitrification
(Downing & Nerenberg, 2008b).

Figure 6-8 shows the effect of available nitrate on the denitrification rates obtained
during runs 8-12 and runs 1-7, examined in the previous chapter. Available nitrate
concentrations are calculated from influent nitrate loading and nitrification rate, as based
on Equation 5-6 and shown below (Equation 6-4).

NO3  N available  LNO  N  0.992rnit Equation 6-4

3

Where: NO3--Navailable = Total available nitrate nitrogen (g m-2day-1)

LNO  N = Nitrate nitrogen loading (g m-2day-1)
3

rnit = Nitrification rate (gN m-2day-1)

137
 1.8

1.6

1.4
denitification rate (gN m-2day-1)

1.2

1.0 No nitrate
added
0.8 Nitrate added

0.6

0.4

0.2

0.0
1 2 3 4
available Nit-N (gNm-2day-1)

Figure 6-8: Variation of denitrification rates with nitrate availability (standard error bars)

Denitrification rates in the range 0.08 – 1.42 gNm-2day-1 were achieved. In general, the
higher rates were obtained when nitrate was added to the feed mixture and nitrate
availability was highest (Runs 9-12), but the highest rate was obtained in Run 8, when
the reactor was operated at 12 hour HRT but with no nitrate in the feed.

The data displayed in Figure 6-8 is grouped in two separate groups; the group on the left
corresponding to Runs 1-8, when no nitrate was added to the synthetic wastewater and
the second group corresponding to Runs 9-12 when nitrate was added to the influent.

Linear regression was used on each of these groups and in turn yielded Equation 6-5 and
Equation 6-6. This decision has neither basis in theory nor precedent in literature, but
was taken to fit the data generated by the investigation. The empirical relationships have
correlation coefficients of 0.91 and 0.85 respectively.

138
No nitrate added to feed: rden  1.15( NO3  N available)  0.92 Equation 6-5

Nitrate added to feed: rden  0.92( NO3  N available)  2.30 Equation 6-6

Where rden = Denitrification rate (gN m-2day-1)

Despite the relatively poor fit coefficients, Equation 6-5 and Equation 6-6 adequately
describes the data obtained from Runs 1-8 and 9-12 respectively. As with Equation 6-3,
it cannot be suggested that these equations hold for all values of available nitrate
loadings, but they are an adequate approximation of the system studied here.

The emergence of separate equations to describe the denitrification performance of the

MABfR in the two different circumstances above implies that there is a difference in the
availability of nitrate to denitrifiers between nitrate produced by nitrification and
commercially produced nitrate as used in the synthetic wastewater used in this study.

This difference is without precedent in the literature and there is no evidence in the
experimental results or in the literature to suggest that the addition of potassium ions (as
KNO3) has an inhibitory effect on denitrification. As removal of nutrients from WwTW
effluents becomes of greater importance as the Water Framework Directive is fully
implemented, further study of the factors controlling denitrification is likely to be
required.

6.3 Model development

Modelling of biofilms and biofilm systems has been a major focus of recent biofilm
research (e.g. Matsumoto et al., 2007, Lackner et al., 2008). Many of the models are
extremely complex, require specialised software to be computed (e.g AQUASIM), and,
although the modelling predictions are in very good agreement with experimental
results, their complexity limits their applicability in ‘real’ applications.

139
The model presented here uses a ‘black box’ approach; compromising the accuracy of
the model in favour of producing a model with applicability for a range of wastewater
compositions.

6.3.1 Assumptions

1. Oxygen transfer is controlled by gas side pressure.

Wastewater strength does have an effect on the structure and therefore oxygen
permeability of biofilms (Bishop et al., 1995). However, the effect is not
significant over the range of wastewater compositions considered here, as shown
by the results presented in Chapter 5. A similar conclusion was drawn by Casey
et al. (2000b) from their work with Xylene degradation in a MABfR.

2. Oxygen supplied by the membrane is completely consumed by the biofilm.

Dissolved oxygen concentrations of the bulk liquid were periodically measured
and monitored during changes to the experimental setup. During steady
operation, the bulk oxygen concentration showed little variation with values <0.1
mg l-1.

3. COD is consumed only by aerobic heterotrophy and denitrification.

Although it is possible for anaerobic niches to occur within the biofilm, no dark
precipitates or distinctive odours were observed throughout the operation of the
reactor, and, as such, any anaerobic activity can be considered negligible
(Timberlake et al., 1988).

4. Ammoniacal nitrogen is consumed only by aerobic autotrophic respiration

and cell synthesis in autotrophic respiration and aerobic heterotrophy.
Ammoniacal nitrogen removal by adsorption onto membrane tubing or the walls
of the tank is negligible.

140
5. Nitrogen for cell synthesis
Nitrogen required for cell synthesis is provided by ammoniacal nitrogen for both
nitrifiers and aerobic heterotrophy (McCarty, 1975). Nitrate is used both as an
oxygen source and nitrogen source by denitrifiers (Madigan & Martinko, 2006).

6. Biomass washout is negligible.

Biomass does leave the reactor in the effluent, as evidenced by the slightly turbid
nature of the effluent. However this turbidity was consistently low throughout
the operation of the reactor (<10 NTU), and therefore is indicative of a low solids
concentration in the effluent which can be considered negligible in comparison to
the biomass concentration on the surface of the membrane.

7. Biofilm is in a pseudo steady state

Although a biofilm is complex, dynamic system, with cell synthesis and death
constantly occurring, in an oxygen limited biofilm the growth rates are limited by
oxygen and substrate availability rather than biomass availability. As such,
mature populations are in a pseudo steady state where the net growth and net
death rates are equal.

8. Dead biomass forms soluble COD, Amm-N and detritus.

For every gram of biomass that dies, 0.2 g of detritus is formed, with 0.8 g being
made available in the liquid phase. This contributes a COD of 0.53 g and 0.12 g
of ammoniacal nitrogen. A similar assumption was made in the modelling work
of Shanahan & Semmens (2004).

141
6.3.2 Model inputs
Average oxygen flux:
J  0.28Pav( abs)  0.22 Equation 5-12

Fraction of oxygen consumption:

𝑓𝐴𝐻 = 0.016(𝐶𝑂𝐷⁄𝑁𝑟𝑎𝑡𝑖𝑜 ) + 0.63 Equation 6-3

𝑓𝑛𝑖𝑡 = 1 − 𝑓𝐴𝐻 Equation 6-7

Oxygen uptake rates:

OURAH  f AH J Equation 6-8

OURnit  f nit J Equation 6-9

Denitrification rate:
No nitrate added to feed: rden  1.15( NO3  N available)  0.92 Equation 6-5

Nitrate added to feed: rden  0.92( NO3  N available)  2.30 Equation 6-6

Pollutant loading rates are used for dimensional consistency with oxygen flux and are
calculated as detailed in Chapter 3 (Timberlake et al., 1988).

Cinf Equation 3-4

Ls 
a( HRT )

Reaction rates are then calculated from the equations above and mass ratios developed
from the stoichiometric relationships derived in Chapter 5. The relationships between
each reaction rate is shown in Table 6-10 (By convention, a species being consumed by
a reaction is designated with a minus sign).

142
From the calculated reaction rates, the effluent concentrations of each of chemical
species of interest can be calculated. These effluent ‘loadings’ can then be converted to
effluent concentrations using Equation 6-10 , a rearranged form of Equation 3-4

Ceff  Leff a(HRT ) Equation 6-10

143
 Table 6-10: Relationships used to calculate reaction rates in MABfR model
Species O2 {CH2O} NH4+-N NO3--N X (biomass)

Process
Aerobic  OURAH  0.483(OURAH )  0.0072(OURAH ) 0.0582(OURAH )
heterotrophy

Nitrification  OURnit  0.225(OURnit ) 0.223(OURnit ) 0.0254(OURnit )

Denitrification  2.85(rden )  rden 0.065(rden )

144
6.4 Model validity

6.4.1 Normalised standard deviation method

The validity of the developed model was assessed by using a modified form of the
normalised standard deviation (NSD) method introduced by O’Neill et al. (2009). The
NSD for the removal of each relevant component (COD, Ammoniacal Nitrogen and Total
Nitrogen) was calculated using Equation 6-11:

 n  C
  j ,exp,i  C j ,calc,i   
2


Equation 6-11
 C j ,exp,i  
 i 1   
NSD(%)  100
n 1
Where: n = number of experimental runs considered
Cj,exp,i = effluent concentration of component j ascertained experimentally
for experimental run i
Cj,calc,i = effluent concentration of component j calculated using the model
for experimental run i

The experimental and calculated effluent concentration values for COD, Ammoniacal
Nitrogen and Total Nitrogen are given in Table 6-11, Table 6-12 and Table 6-13
respectively; alongside the influent concentration and percentage error for comparative
purposes. The percentage errors are calculated using Equation 6-12:

Equation 6-12
c j ,exp,i  c j ,calc,i
error(%) 
c j ,inf,i
Cj,inf,i = influent concentration of component j for experimental run i

The influent, experimental effluent and calculated COD concentrations for Runs 1-12 are
shown in Table 6-11.

145
 Table 6-11: Experimental and calculated COD concentrations for Runs 1-12
Average COD concentration
Run Influent Experimental Calculated Error
(mg l-1) effluent (mg l-1) effluent (mg l-1) (%)
1 70 29 31 2.8
2 72 27 31 5.6
3 71 32 28 5.6
4 71 29 26 4.2
5 73 30 26 5.5
6 73 24 23 1.4
7 72 16 14 2.8
8 66 27 28 1.5
9 71 30 34 5.6
10 71 41 37 5.6
11 116 57 80 19.8
12 69 24 32 11.6
NSD (%) = 5.7

A significant error between the experimentally determined and calculated COD

concentration can be seen for Run 11 (19.8%), with the experimental value being
significantly lower than that which is predicted by the model. As seen in Table 6-4, the
apparent oxygen flux for this experimental run was significantly larger than that predicted
by Equation 5-12, which is used to obtain the oxygen flux as part of the model. The error
between the calculated and experimental COD removal can therefore be attributed to this
higher than normal oxygen availability. Considering the values for Run 11 as rogue and
omitting them when calculating the NSD gives a reduced value of 4.8% for COD
removal.

The influent, experimental effluent and calculated ammoniacal nitrogen concentrations

for Runs 1-12 are shown in Table 6-12.

146
 Table 6-12: Experimental and calculated Amm-N concentrations for Runs 1-12
Average Ammoniacal nitrogen concentration
Run Influent Experimental Calculated Error
(mg l-1) effluent (mg l-1) effluent (mg l-1) (%)
1 9.72 5.23 3.82 14.5
2 9.31 3.24 3.40 1.7
3 9.40 3.38 3.21 1.9
4 9.56 3.22 3.17 0.5
5 9.82 4.02 3.22 8.1
6 8.07 4.00 1.75 27.9
7 9.49 2.23 1.76 5.0
8 11.72 6.43 6.39 0.3
9 10.10 5.56 5.19 3.7
10 7.16 3.61 3.09 7.3
11 8.11 3.73 5.35 20.0
12 7.59 2.55 3.29 9.7
NSD (%) = 7.6

As with COD removal in Table 6-11, higher than predicted ammoniacal nitrogen removal
was obtained in Run 11. Again this can be attributed to the higher than expected oxygen
flux which was experienced during the duration of this experimental run. Omitting this
run in calculation of the NSD yields a NSD of 6.2%.

In Chapter 5, it was discussed how the nitrification and denitrification rates observed
during run 6 were not in accordance with the trend seen as inlet pressure was increased.
Although the cause of this was not known, the same cause can be attributed to the
significant error seen in Table 6-12 for Run 6. Also ignoring this run in calculation of the
NSD gives a further reduction in the NSD value to 5.0%.

147
The influent, experimental effluent and calculated total nitrogen concentrations for Runs
1-12 are shown in Table 6-13.

Table 6-13: Experimental and calculated Tot-N concentrations for Runs 1-12
Average Total Nitrogen concentration
Run Influent Experimental Calculated Error
(mg l-1) effluent (mg l-1) effluent (mg l-1) (%)
1 9.8 9.4 8.0 13.5
2 9.4 7.4 7.6 2.5
3 9.5 8.6 7.4 13.0
4 9.7 8.3 7.3 10.9
5 9.9 7.8 7.3 4.6
6 8.2 7.6 5.9 21.2
7 9.6 6.7 5.7 10.6
8 11.8 7.8 8.2 3.0
9 15.6 12.6 12.4 1.1
10 11.6 10.4 10.1 2.8
11 13.3 11.5 12.4 6.2
12 13.1 9.6 10.4 6.4
NSD (%) = 5.0

It can be seen from Table 6-13 that there is better agreement between experimental and
calculated total nitrogen concentrations for experimental runs 8-12 in comparison to runs
1-7. As discussed previously in Chapter 5, the duration of runs 1-7 was extended to
obtain oxygen limitation, not to achieve steady pollutant removal. This is likely to be the
cause of the greater error between calculated and experimental results.

As with ammoniacal nitrogen, considering the calculated results for Runs 6 and 10 to be
rogue reduces the NSD to 3.3%.

148
Table 6-14 summarises the percentage NSDs obtained from the model for each of the
experimental runs.
Table 6-14: Adjusted NSD (%)
Component Adjusted NSD (%)
Chemical Oxygen Demand 4.8
Ammoniacal Nitrogen 5.0
Total Nitrogen 3.3

As seen above, removing results which are considered erroneous means than an NSD of
5.0% or lower is obtained for each of COD, Ammoniacal nitrogen and Total nitrogen.
The authors that introduced this method considered an NSD of lower than or equal to
6.5% to be considered a good fit (O'Neill et al., 2009), which indicates that the model
presented here is a good description for the operation of the MABfR used in this study.

6.4.2 Sensitivity analysis

In addition to the use of the NSD method introduced by O’Neill et al. (2009), a
sensitivity analysis was also carried out. This facilitated the examination of predicted
effluent concentrations from the MABfR at conditions at which the unit was not operated
during the experimental work.

These predicted concentrations could then be compared to models in other published

work, allowing the model validity to be examined qualitatively in addition to the
quantitative model fit measurement obtained from the NSD method.

The effect of changing inlet pressure, influent COD, ammoniacal nitrogen and nitrate
nitrogen concentration was considered in the sensitivity analysis.

6.4.2.1 Effect of changing gas side pressure

The effect of changing inlet pressure (and hence oxygen flux) on effluent concentrations
of COD, ammoniacal nitrogen and nitrate nitrogen is shown in Figure 6-9. The effluent

149
concentrations are calculated from the presented model based on an influent containing
60 mg l-1 COD, 10 mg l-1 (as N) Ammoniacal nitrogen and 5 mg l-1 (as N) nitrate
nitrogen, and with the MABfR being operated using a 12 hour HRT.

10 50

effluent concentration (mgCOD l-1)

effluent concentration (mgN l-1)

8 40

Amm-N
6 30
Nit-N
COD
4 20

2 10

0 0
1 1.5 2 2.5 3 3.5 4 4.5 5
Average pressure (bar abs)

Figure 6-9: Sensitivity analysis for inlet pressure

As expected, increased inlet pressure leads to a decrease in the effluent concentration of

COD and ammoniacal nitrogen – the wastewater components which require oxygen for
removal. Effluent COD concentration decreases more strongly than effluent ammoniacal
nitrogen concentration as it is removed by both aerobic heterotrophy and denitrification.

Effluent concentration of nitrate nitrogen is relatively insensitive with effluent

concentrations varying only between 6.9 and 7.3 mg l-1. At higher inlet pressures, more
ammoniacal nitrogen is converted to nitrate nitrogen, but this higher availability of nitrate
increases the rate of nitrate consumption by denitrification.

This is of interest for design of a full scale unit as total nitrogen consents are introduced
in eutrophication sensitive areas as part of the full implementation of the Water
Framework Directive.

150
6.4.2.2 Effect of changing influent ammoniacal concentration
Figure 6-10 shows the effect on changing influent ammoniacal nitrogen concentration on
the calculated effluent concentrations of ammoniacal nitrogen, nitrate nitrogen and COD.
The effluent concentrations are calculated based on an inlet pressure of 3 bar, with
influent COD concentration of 60 mg l-1, influent nitrate nitrogen 5 mg l-1 and a 12 hour
HRT.

10 50

effluent concentration (mgCOD l-1)

effluent concentration (mgN l -1)

8 40
Amm-N
6 30 Nit-N
COD
4 20

2 10

0 0
5 7.5 10 12.5 15
Influent [Amm-N] (mg l-1)

Figure 6-10: Sensitivity analysis for influent ammoniacal nitrogen concentration

Influent ammoniacal nitrogen concentration has little or no effect on the effluent

concentration of COD or nitrate nitrogen, with effluent concentrations in the range 23 -
25 mgCOD l-1 and 7.0 – 7.2 mgN l-1.

Under increasing ammoniacal nitrogen loading, the shift in the COD:N ratio means that
more of the oxygen supplied by the membrane is consumed by nitrification (Equation 6-6
& Equation 6-7). This leads to an increase in the availability of nitrate nitrogen and an
associated increase in the rate of denitrification, which has a regulatory effect on the
effluent COD and nitrate nitrogen concentrations.

151
6.4.2.3 Effect of changing influent COD concentration
Figure 6-11 illustrates the effect of changing influent COD concentration on calculated
effluent concentrations of ammoniacal nitrogen, nitrate nitrogen and COD. The effluent
composition is calculated based on an inlet pressure of 3 bar, with influent concentrations
of 10 mg l-1 ammoniacal nitrogen and 5 mg l-1 nitrate nitrogen and a HRT of 12 hours.

8 160

effluent concentration (mgCOD l-1)

effluent concentration (mgN l-1)

6 120

Amm-N

4 80 Nit-N
COD

2 40

0 0
40 60 80 100 120 140 160
-1
influent [COD] (mg l )

Figure 6-11: Sensitivity analysis for influent COD concentration

The effluent concentrations of both COD and ammoniacal nitrogen increase with
increasing influent COD concentration. In the case of COD, this can be attributed to the
increased influent concentration, and for ammoniacal nitrogen it is due to less oxygen
being consumed by nitrifying bacteria in line with Equation 6-6 and Equation 6-7.

As with changing ammoniacal nitrogen, changing influent COD has little effect on the
effluent nitrate nitrogen concentration with calculated values obtained in the range 6.9 –
7.2 mg l-1. This is due to a combination of higher denitrification rates at low COD
loadings and less nitrification at higher COD loadings.

152
6.4.2.4 Effect of changing influent nitrate concentration
The effect of influent nitrate-nitrogen on the calculated effluent pollutant concentrations
is shown in Figure 6-12. The effluent concentrations are calculated based on an inlet
pressure of 3 bar, with influent COD concentration of 60 mg l-1, influent ammoniacal
nitrogen concentration of 5 mg l-1 and a 12 hour HRT.

8 40

effluent concentration (mgCOD l-1)

effluent concentration (mgN l-1)

6 30

Amm-N
4 20 Nit-N
COD

2 10

0 0
2 4 6 8 10
Influent [Nit-N] (mg l-1)

Figure 6-12: Sensitivity analysis for nitrate nitrogen

The effluent concentrations of both nitrate nitrogen and ammoniacal nitrogen are
insensitive to influent nitrate nitrogen concentrations, with only small increases visible
for both effluent concentrations. The fraction of available oxygen utilised by nitrification
is independent of nitrate concentration (Equation 6-3), therefore ammoniacal nitrogen
concentration is unaffected.

In accordance with Equation 6-8, the denitrification rate increases with increasing
available nitrate concentration, with a proportionality of 0.92; accounting for the almost
horizontal trend in effluent nitrate concentrations predicted by the model. A decrease in
the effluent COD concentration is observed, as 2.85 g of COD is consumed by
denitrification of each gram of nitrate nitrogen.

153
6.5 Conclusions
The MABfR was successfully operated for a period of 230 days at a HRT of 12 hours
using a range of wastewater compositions to examine the effect of pollutant loadings on
pollutant removal. The data generated allowed the formation of empirical equations
which adequately described the results obtained with the range of pollutant loadings used
here.

These empirical equations were utilised, in conjunction with established theory and other
published research to develop a simple model which predicts the performance of the
MABfR when operated over a range of inlet pressures and wastewater compositions.

However, the limit of operation over which the model is valid was not ascertained.
Establishing these limits are a suggestion for further work in this area.

154
7 Design and operation of a pilot-scale BioSettlerTM

Chapter 5 and Chapter 6 have demonstrated the ability of a MABfR to provide treatment
to wastewaters with compositions similar to those found entering a secondary settling
tank.

Using the data generated in Chapters 5 and 6, and drawing on best practice found in
industry, a 1.5 m3 pilot scale BioSettlerTM was designed and constructed, and operated at
two municipal WwTWs in Northern Ireland.

7.1 Pilot scale design

Inclined plate settlers were first mooted by Hazen in the early twentieth century (1904),
were introduced by the Swedish Company, Parkson Corporation, in the 1960s
(Hendricks, 2006) and have now entered widespread use. Despite this, the complexity of
the interactions between fluid and particles flows, and the large number of parameters
affecting these flows, mean that no definitive design equations are in existence.

In the absence of such design equations, the design presented here draws on the limited
published academic, industrial best practice and marketing literature, intended to be
compliant with the closest relevant design standard (BSi, 2002a).

Following consultation with local steel fabricators (Stoneyford Engineering, Lisburn,

U.K.), the CAD drawings shown in Figure 7-1 were prepared.

155
 Figure 7-1: CAD diagrams – cross sectional (left) and bird’s eye view

The prototype consists of three sections: an inlet zone, which acts in the same way as in a
stilling box in conventional settler; a plate pack, where the settlement and treatment takes
place; and a sludge collection zone, where settled sludge thickens and is removed from
the unit by the action of a peristaltic pump.

The tank was constructed by Stoneyford Engineering (U.K.) in 6 mm mild steel, and
protected by painting with suitable rust proof coating (Figure 7-2).

Figure 7-2: BioSettler prototype tank

156
Complete CAD diagrams and full details of the calculations are given in Appendix 3; a
summary of the key dimensions is given in Table 7-1.

Table 7-1: Summary of prototype dimensions

Number of plate packs 2
Number of plates per pack 11
Area of plates 0.5 m2
Plate angle 55° (above horizontal)
Settling volume 0.82 m3
Inlet zone volume 0.35 m3
Sludge collection volume 0.28 m3
Angle of sludge collection zone 20° (above horizontal)
Total tank volume 1.45 m3

7.1.1 Justification of design

7.1.1.1 Feed section

The purpose of the feed section in settling tanks is to dissipate the energy of the influent
wastewater and evenly distribute the incoming flow over the two plate packs.

This is achieved in the BioSettler through a stilling section (the triangular section seen on
the left hand side of the cross sectional CAD diagram in Figure 7-1), where the incoming
flow is slowed. From there, it entered a channel between the two plate packs and was
distributed between them, as illustrated in Figure 7-3.

157
 Figure 7-3: Inlet flow patterns

Pieces of PVC sheet were employed to force the influent to enter the plate in the lower
1/3 and prevent shortcutting.

7.1.1.2 Inclined plates

The selection of a plate spacing is a trade-off between solids removal and ease of
operation; smaller plate spacings yield higher solid removal efficiencies (due to a
reduction in the vertical distance a particle must fall to be removed from suspension), but
also cause clogging of the flow channel between the two plates. As a compromise, a plate
spacing of 1 - 3” (25 – 75 mm) is commonly employed (Metcalf & Eddy, 2003).

As illustrated in Figure 7-4, inclined plates lead to the formation of three discrete zones; a
sludge layer, a mixed layer and a clarified layer. Although using a larger spacing means a
longer residence time is required to achieve a specified degree of solids removal, this
demand is compatible with the biological treatment supplied by the BioSettler, which
improves with increasing HRT.

158
 Figure 7-4: Membrane arrangement in BioSettler

In the BioSettler, a plate spacing of 90 mm, slighter higher than the industry standard,
was used (Figure 7-5). This allowed the clarified layer to become large enough to
accommodate the membrane module whilst reducing the likelihood of the membrane
becoming fouled by solids.

Figure 7-5: BioSettler plate spacing

159
7.1.1.3 Sludge collection section
In small settling tanks, where no scraper or suction mechanism is employed, it is standard
industry practice to use tanks with floors angled at 50 - 60° above the horizontal (BSi,
2002a, BSi, 2002b). This allows sludge to thicken and effectively removed using gravity.

However, this was not possible in the design of the prototype BioSettler. Having a sludge
collection zone angled at 50° above the horizontal would lead to a tank which was
prohibitively high. Under health and safety regulations, a scaffold is required where work
is to take place above head height (HMG, 2005). Provision of a scaffold was not possible
within the project budget.

In order to meet the constraints of the budget, a tank with a floor sloped 20° above the
horizontal was used. This gave a tank height of approximately 1.6 m, preventing the need
for working above head height.

7.1.1.4 Effluent collection

Effluent was collected in a channel at the top of the tank, positioned between the two
plate packs and fitted with v-notch weirs (Figure 7-6).

Figure 7-6: V-notch weirs in BioSettler tank

160
Once filled with liquid, the tank was levelled, ensuring that overflow was equal from all
areas of the tank.

7.1.2 Membranes
In the experimental laboratory setups used in this project, long lengths of silicone rubber
were wound around suitable frames to act as the membrane aerators. This arrangement
was used due to simplicity of setups – only two hollow fibres had to be connected to the
compressed air supply.

This setup, however, has two main drawbacks. Due the relatively large lengths of
membranes required to obtain sufficient surface area to meet the oxygen transfer
requirements and the number of bends required in order to wrap the tubing around their
frame, the pressure drop was significant.

Additionally, using long lengths means that each individual hollow fibre has a greater
surface area available for back diffusion of water vapour compared with shorter lengths,
requiring greater air flowrate airflow to prevent condensation. This higher flowrate
requirement and pressure drop mean that this arrangement is not practicable due to high
operating costs.

Parallel flow, such as those used in the mass transfer experiments in this study, offers the
highest average concentration driving force, and as such is preferable in situations where
mass transfer is controlled by membrane mass transfer coefficient (Dindore et al., 2005)
.
The work presented in Chapter 5 ascertained that, in the presence of a membrane attached
biofilm, it is the biofilm which controls mass transfer. Cross-flow operation gives higher
shell side mass transfer coefficients compared to parallel flow (Vladisavljevic, 1999), and
therefore was chosen for this application to promote better substrate mixing between the
biofilm and bulk liquid, as illustrated in Figure 7-7.

161
 Figure 7-7: Cross flow operation

7.1.2.1 Membrane modules

Membranes were obtained from a hollow-fibre ultrafiltration membrane module (MWCO
150 kDa, Microdyn-Nadir GmbH, Wiesbaden, Germany) (Figure 7-8). The membrane
fibres were rendered hydrophobic in line with instructions from the supplier by flushing
out of the storage solution (1% aqueous formaldehyde solution) from the membranes and
then drying by passing air through the membranes (Steube 2009). The module was then
opened and the fibres used to form 20 membrane modules.

Figure 7-8: Daicen Membrane Module

162
To form each module, two lengths of 50 mm x 25 mm rigid PVC cable trunking (B&Q,
U.K.) were cut and one side drilled with 78 holes. The holes were arranged in alternate
rows of 3 and 2 in a staggered square arrangement with a nominal pitch of 10 mm.

The trunking was then secured 400 mm apart by 2 PVC struts and membrane fibres fed
through each hole as shown in Figure 7-9. The membranes were then sealed and secured
in position with an epoxy potting compound (R.S. Components, U.K.).

Figure 7-9: Membrane module construction Figure 7-10: Membrane sealing with epoxy
resin

The sections of cable trunking were then reformed and formed into boxes with two end
pieces (B&Q, U.K.), one of which was fitted with a standard airline push fitting (R.S.
Components, U.K.). These boxes were then sealed to make them air and water tight using
a combination of heat and TEC7® sealant (Contech Building Products, Ireland).

163
 Figure 7-11: Membrane modules in BioSettler unit

The membranes supplied a total aeration area of 2.036 m2. The membrane modules were
attached to the underside of the inclined plates with the membrane fibres parallel to the
plates as shown in Figure 7-11. This provided cross flow operation as the clarified
effluent rose in the channels between the plates.

7.1.3 Auxillary equipment

7.1.3.1 Pumps
Influent wastewater was provided to the unit by use of a Watson-Marlow 604U peristaltic
pump (Watson-Marlow Bredel, Falmouth, U.K.). Influent was drawn from the main
works at a flowrate of 4 m3day-1, giving a hydraulic retention time of approximately 6
hours.

A similar pump also controlled underflow from the bottom of the tank to facilitate solids
removal, with underflow rate determined by settled sludge volume.

164
7.1.3.2 Air blower
Air was provided to the lumen of the membranes by a Medo LA-120 Air Blower (Nitto-
Kohki Europe Ltd, Watford, U.K.). Designed for aeration of fish ponds, the unit provides
a constant air flowrate of 120 lpm of air at a pressure of 0.2 bar gauge.

7.1.3.3 Sampling equipment

Two ISCO 6700 automated water samplers were used to collect 24-hour composite
samples of the wastewaters which were influent and effluent to the BioSettler unit. 125ml
of both influent and effluent were collected at 90 minute intervals, giving a daily 2000 ml
sample representative of each day’s flow.

7.2 Trial locations

The unit was operated at two municipal WwTW owned and operated by Northern Ireland
Water.

7.2.1 Site 1
The first site was Newtownbreda WwTW, a 40,000 p.e. activated sludge plant in South
Belfast. The works produced fully nitrified and well settled effluent and was consent
compliant in 2009. Significant urban growth has taken place since the plant was last
upgraded in the 1980s and as such, the plant struggles with high hydraulic loading.
Capital works were carried out after this demonstration was completed at the site to
increase hydraulic capacity and add nutrient removal.

7.2.2 Site 2
The second site was a small works at Parkgate village, near Templepatrick in County
Antrim. The treatment consisted of a primary settler, trickling filter and final settling
tank. The works was borderline compliant with a 40:60 BOD:Suspended solids consent

165
and, as a result of these compliance issues, has now been decommissioned and replaced
with a sewage pumping station which transfers wastewater to a larger works nearby.

7.3 Trial results

7.3.1 Site 1
Several problems were experienced during operation at the first site. At Newtownbreda
WwTW, the aeration basins are split into three zones, with surface aerators and separated
with baffles. On leaving the aeration basin, the MLSS is divided, with a fraction passing
to the final settling tanks and the remainder being returned to the first of the three
aeration zones via a recycle line.

It is from this recycle line (Figure 7-12) that influent to the BioSettler was initially drawn.
However, in this location the feed intake became blocked with rags and other debris, and
led to no flow reaching the unit. In the diagram, the position of the BioSettler is marked
by the green rectangle, the first sample point by the red triangle and the second sample
point by the yellow star.

Figure 7-12: Layout of Newtownbreda WwTW (Googlemaps).

To avoid this blocking from occurring, the intake point was moved to the third of the
three aeration zones (Figure 7-13). The wastewater from this point has essentially the
same composition as that in the recycle line, and the turbulence caused by the surface
aerators helped clear any debris that built up on the influent intake pipe.

166
 Figure 7-13: Second influent intake point used at Site 1.

7.3.1.1 Membrane damage

As discussed in Section 7.1.1.3, the slope of the prototype tank was less than ideal in
order to comply with health and safety regulations. Whilst this did not interfere with the
clarification of the influent wastewater, it did limit the rate at which solids could be
removed from sludge collection section of the tank.

This complication led to long sludge retention times with the associated problem of rising
sludge, caused by bubbles of nitrogen gas formed by denitrification. As pockets of settled
sludge rose, they contacted the membrane modules and deposited large amounts of solids
on the surface of the membranes Figure 7-14.

167
 Figure 7-14: Sludge buildup on membranes Figure 7-15: Damage to membranes

The weight of the sludge which built up on the membrane surface caused membranes
breakages as shown in Figure 7-15. These membrane breakages led to the release of
bubbles with associated turbulence causing solids to carry over the v-notch weir.

These breakages meant that only periodic operation was possible. In order to facilitate
repairs, the plates were removed from the tank, broken membranes removed from the
membrane module and the module made airtight again by patching the hole with TEC7®
sealant.

7.3.1.2 Pollutant removal

Due to these problems, it was not possible to obtain a body of results. However, over the
course of the trial, 32 days of data was collected when operation was not hindered by the
issues described in Sections 7.3.1. This data is presented in average form in Table 7-2.

168
 Table 7-2: Pollutant removal obtained during Trial 1 (Standard errors shown)
Suspended COD Ammoniacal
Solids Nitrogen
Average influent concentration 3060±190 43.1±8.3 9.80±0.76
(mg l-1)
Average effluent concentration 22.0±3.0 15.2±2.5 8.05±0.71
(mg l-1)
Removal (%) 99.3 64.7 17.9

Whilst limited inferences can be made from this data due to the periodic nature of
operation during Trial 1, some conclusions can be drawn. Removal of COD and
ammoniacal nitrogen was achieved; demonstrating that membrane aerated biofilms can
obtain pollutant in ‘real-life’ situations.

Good solids removal was achieved with the BioSettler unit, with levels being reduced
from the concentrations found in aeration tanks to levels which are consent compliant.
This indicates that the process of bubblefree aeration does not interfere with the settling
process.

7.3.2 Trial 2
The unit was operated at Parkgate WwTW for six weeks over the summer of 2010 at the
same flowrates as at Newtownbreda WwTW. Removal of key pollutants was achieved,
and the experience from Trial 1 meant that the unit was operated without interruption for
42 days. Longer operation was not possible as the treatment works was replaced with a
sewage pumping station and flows were diverted.

To avoid operation being compromised by solids build-up in the BioSettler tank, as was
the case during the Newtownbreda trial, the effluent from the final settling tank was used
for BioSettler influent (Figure 7-16).

169
 Sample
Point

Figure 7-16: Sample location at Site 2

On day 36 of the trial, following a drop in performance, the liquid level in the unit was
reduced to remove excessive biomass from the surface of the membranes and deposited
solids from the surface of the inclined plates. In Figure 7-17, Figure 7-18 and Figure
7-19, this point where ‘backwashing’ took place is indicated by a vertical dashed line. For
clarity, a seven day moving average was used.

7.3.3 Solids Removal

Figure 7-17 shows the influent and effluent suspended solids concentrations during Trial
2.

170
 150

120
WwTW
Suspended Solids (mg l-1)

90 BioSettler

WwTW 7 Day
60 moving average
BioSettler 7 Day
30 moving average

0
0 10 20 30 40
Day
Figure 7-17: Solids concentrations during BioSettler trial 2

The effluent produced by the plant contained solids which were not settleable (verified by
the use of Imhoff cones). The BioSettler was therefore not able to achieve significant
additional solids removal in comparison to the plant. The data points where suspended
solids concentration is higher in the BioSettler than plant effluent can be attributed to
sloughing of biomass from the surface of the membranes into the bulk liquid.

7.3.4 Biochemical Oxygen Demand

Figure 7-18 shows the influent and effluent BOD concentrations obtained during Trial 2.
Variations in wastewater composition are linked to rainfall.

171
 140

120
WwTW
100
BioSettler
BOD (mg l-1)

80

60 WwTW 7-Day
moving average
40 BioSettler moving
average
20

0
0 10 20 30 40
day

Figure 7-18: BOD concentrations during BioSettler trial 2

Following commissioning of the BioSettler, a biofilm was quickly established on the

surface of the membranes and BOD removal was obtained. From Day 10 – 20
approximately 20% of the BOD contained in the influent wastewater was removed, at an
average rate of 33.1 gBODm-2day-1. After day 20, a reduction in performance was seen;
as biofilm maturity increases, so does the biofilm density. This denser biofilm slows the
diffusion rate of both oxygen and substrate, reducing pollutant removal rates (Matsumoto
et al., 2007).

Prior to ‘backwashing’, from approximately Day 31 to Day 37, the BioSettler average
BOD was higher than that of the plant effluent. This was limited by mass transfer due to
biofilm density. A similar effect was noted by Pankhania et al. (1999) and can be
attributed to a period when the rate of hydrolysis of wastewater constituents into a form
where they are more readily available to microorganisms is faster than the rate at which
they are utilised by the biofilm. Following backwashing, the BOD utilisation rate was
boosted and BOD was once again removed by the BioSettler.

The BOD loading at Parkgate WwTW was much higher than was considered in the
concept design; nonetheless, BOD removal rates of up to 96 gBODm-2day-1 were
obtained. This figure is far in excess of those reported for similar investigations (e.g.

172
Pankhania et al., 1999), and by previous laboratory work in this project. There are two
possible explanations; either greater oxygen transfer rates are obtained with the
membranes used in the pilot plant, or a significant amount of BOD removal is achieved
through the removal of suspended biomass. It is likely that both mechanisms contribute to
the high BOD removal rate.

7.3.5 Ammoniacal Nitrogen

The nitrification performance of the BioSettler was also detrimentally affected by the
high BOD loading due to the competition between heterotrophic bacteria and nitrifiers for
oxygen. Figure 7-19 shows the influent and effluent ammoniacal nitrogen concentrations
during Trial 2.

70

60
WwTW
50
Ammonia (mgN l-1)

BioSettler
40

30 WwTW 7 Day
moving average
20 BioSettler 7 Day
moving average
10

0
0 10 20 30 40
day

Figure 7-19: Ammoniacal nitrogen concentrations during BioSettler Trial 2

Stable nitrification was quickly established within the BioSettler and from Day 13 – 20
ammoniacal nitrogen was removed at an average rate of 13.4 gNm-2day-1 (giving
approximately a 20% reduction in ammonia concentrations). This value is also higher
than those contained in published research (e.g.Yamagiwa & Ohkawa, 1994), and, as
with BOD removal, it is likely that suspended biomass contributes to ammoniacal
nitrogen removal.

173
Ammoniacal nitrogen removal performance also declined after approximately Day 20 of
the trial. Due to slower decay rates, nitrifying bacteria are predominately found on the
membrane surface (Shanahan & Semmens, 2004). In this position they are starved of
substrate in thick biofilms; therefore ammonia removal declined faster than BOD
removal, which can occur in both aerobic and anoxic environments.

Heterotrophic bacteria have faster specific growth rates than nitrifying bacteria
(Shanahan & Semmens, 2004). In situations where their growth is not limited by BOD
availability, they can completely outcompete nitrifiers for space and oxygen in membrane
aerated biofilms (Zhang et al., 1995). For this reason, ammonia removal was not restored
by backwashing on Day 36.

7.4 Potential performance of scaled up system

7.4.1 Pollutant removal

Using the data obtained between day 10 and day 20, at which time the biofilm on the
membrane surface had reached maturity, but pollutant removal was not yet inhibited by
excessive biofilm growth, the average pollutant removal rates shown in Table 7-3 can be
calculated.

However, due to the poor settling characteristics of the wastewater at the Parkgate
WwTW, suspended biomass was present in the BioSettler effluent. This suspended
biomass had the potential to contribute towards pollutant removal. Allowing this
contribution would likely be small due to the anoxic nature of the bulk liquid. To allow
for this, only 50% of the obtained value is used for subsequent calculations. This leaves a
conservative estimate of the removal achieved by the presence of the membrane attached
biofilm.

174
 Table 7-3: Average pollutant removal rates
Obtained value 50% of obtained
value
Average BOD removal 33.1 16.5
rate (gBODm-3day-1)
Average NH4-N removal 12.5 6.3
rate (gNm-3day-1)

Practicalities of the manual manufacturing process using in the pilot scale BioSettler
meant that the membrane specific surface area (SSA) was limited to 2.04 m2m-3.
However, commercial membrane modules, such as those produced by Zena membranes
(Czech Republic – shown in Figure 7-20), with dimensions which would fit onto the back
of the BioSettler plates; contain 4 m2 of membrane surface area.

Figure 7-20: Zena membrane module

The nature of these modules, consisting of bunches of membrane fibres, means that
surface area will be lost as the membranes touch each other. Even if this means only 25%
of this area could be utilised for membrane attachment, the SSA of the pilot-scale
BioSettler would still be 20 m2m-3 and the potential removal rates would increase by an
order of magnitude, as shown in Table 7-4.

Table 7-4: Potential removal rates

Potential value
BOD removal rate (gBOD m-3day-1) 161.8
NH4-N removal rate (gN m-3day-1) 61.8

175
7.4.2 Energy consumption
Upgrade of a conventional settler to the BioSettler system will incur additional energy
costs in the supply of air to the lumen of the membranes. At Parkgate WwTW, the unit
was operated with excess air and specific energy consumption (estimated from the power
rating of the blower unit) of 0.78 kWh m-3 compared with 0.5 kWh m-3 for traditional
treatment systems (Kadar & Siboni, 1998).

However, air supply to the lumen provides both the required oxygen to the biofilm and
acts as a sweep gas, removing water vapour which diffuses in the opposite direction – i.e.
from the water side to the air side. It is likely that airflow used in the pilot unit was much
greater than needed through the BioSettler system.

Work investigating the oxygen mass transfer in the same polyethersulphone membranes
as in the pilot scale BioSettler without the presence of a biofilm has been carried out
(Doyle, 2011). Operating at an air flowrate per fibre of 12 times lower than that in the
BioSettler pilot unit, no inhibition in oxygen transfer by back diffusion of water vapour
was observed.

This implies that the same blower unit could be used to supply air to 12 times as much
membrane, and hence treat twelve times as much wastewater. If this flowrate could be
realised, the specific energy consumption would be reduced to a much more favourable
0.065 kWh m-3.

Additionally, as the presence of the biofilm acts as a barrier to mass transfer it may slow
the transfer of water vapour into the lumen side of the membrane. The laboratory
experiments discussed in Chapter 5 found the oxygen flux through the membrane in the
presence of a biofilm (with bulk DO approximately equal to zero) to be only 44% of the
average oxygen flux measured in Chapter 3 through clean membranes over the DO range
1 – 5 mg l-1.

176
If a similar effect was observed for the rate of transfer of water vapour, it could be
assumed that the same blower could be used for the aeration of another additional 2.25
times as much membrane and treat 2.25 as much wastewater, reducing the specific
energy consumption still further to a figure of 0.028 kWh m-3.

A better comparison is made between the aeration costs in a BioSettler and the energy
costs involved in pumping wastewater from non-consent-meeting works to alternative
locations for treatment. For example, following the closure of Parkgate WwTW,
wastewater is now pumped from Parkgate Village to the new 78,000 p.e. Antrim WwTW
via Templepatrick SPS – a distance of approximately 8.75 miles (14 km).

No data is available for the energy usage involved in pumping wastewater from Parkgate
SPS to where it undergoes treatment (Smyth 2011). However, an estimate can be made
from a study of the Oslo, Norway, wastewater treatment system (Venkatesh & Brattebo,
2011). Over the period 2000 – 2007, 46.34 GWh of energy were used in pumping of
1.01 Gm3 of wastewater. Assuming wastewater was pumped a distance of 10 km on
average, this gives a specific energy consumption of 0.46 kWh m-3km-1.

Applying this figure to the distance from pumping station to treatment works in the case
of Parkgate gives an estimated specific energy consumption of 0.064 kWh m -3. This is
comparable with the energy cost of the BioSettler and implies that the BioSettler concept
is energetically feasible, were upgrade possible with the installation of BioSettler system.

7.4.3 Energy consumption of unit pollutant removal

Assuming wastewater enters the treatment works at an average BOD of 200 mg l-1 and is
treated to an effluent concentration of 10 mg l-1, the specific BOD removal of the
activated sludge can be calculated on an energy basis from the specific energy
consumption on a volume basis. Using the figure reported by Kadar & Siboni (1998) –
0.5 kWh m-3 – yields a value of 0.38 gBOD kWh-1.

177
The same parameter can be calculated for the BioSettler. Using the BOD removal rate
reported in Table 7-3 (16.5 gBODm-2day-1), and a specific energy consumption of 0.028
kWh m-3 as calculated previously, a figure of 0.59 gBOD kWh-1 is obtained. This value is
55 % higher than that for activated sludge.

7.5 Conclusions
The performance of the BioSettler at two WwT sites has demonstrated the promise of the
technology, despite operation with wastewaters of less than ideal characteristics. The two
site trials have shown that the BioSettler concept is capable of simultaneous aerobic
biological treatment (treating BOD and nitrifying ammonia), denitrification and solids
removal.

During the initial trial (Site 1), improved suspended solids removal was demonstrated but
the BOD and ammonia levels in the (MLSS) influent were too low to show a
convincingly significant benefit due to treatment within the BioSettler.

At Site 2 the BOD and ammonia concentrations were very high, and, despite high
specific removal rates achieved by the BioSettler, successful operation of the technology
required control of biofilm growth. The variable performance of the BioSettler at Site 2
is adequately explained by competition between heterotrophs and nitrifiers and the
limitations of mass transfer in the thick biofilm developed under high BOD loading. A
weekly backwash is likely to be sufficient to control biomass overgrowth under
conditions of high BOD loading.

The BioSettler technology is a simple and attractive option for upgrade of plants,
promising minimal construction down time and reduced cost while maximising existing
assets. Further demonstration of the technology is required to assess the full potential
achievement of the promised benefits. This will allow optimisation of residence times
and energy consumption as well as further assessment of performance.

178
8 Treatment of azo dye waste in the Membrane Aerated
Biofilm Reactor – a feasibility study

The work reported in Chapters 4, 5 and 6 demonstrated the ability of membrane aeration
to supply sufficient oxygen to maintain a biofilm where both aerobic and anoxic
conditions exist.

Previous published research has demonstrated that these conditions are ideal for the
treatment of dye house waste. This chapter investigates the possibility of using the
MABfR to treat azo dye wastewater; which requires both aerobic and anoxic conditions
for complete pollutant removal.

8.1 Experimental conditions

A MABfR was constructed as described in Chapter 3, again using silicone rubber as the
aeration material. It was then operated for 121 days with a hydraulic retention time of 5
days, treating a synthetic dye house wastewater containing the azo dye Acid Orange 7
(AO7).

The investigation was carried out using three sets of experimental conditions. The aim of
each of these three experimental runs is given below:

Run 1: Establish AO7 reducing biofilm;

Run 2: Investigate effect of riboflavin addition as redox mediator to the
influent wastewater;
Run 3: Investigate effect of 20% reduction in total oxygen demand (TOD)
of influent wastewater.

The concentrations of those components of the synthetic wastewater, which varied in

order to achieve each of the goals listed above, are given in Table 8-1.

179
 Table 8-1: Component concentrations varied during dye degradation studies
Concentration (mg l-1)
Run Duration Sucrose Peptone Ammonium Riboflavin
(days) Chloride
1 1-58 550 200 350 -
2 58-75 550 200 350 10
3 75-121 440 160 280 10

8.2 General observations

8.2.1 Biofilm establishment

The reactor was seeded with a mixture of the supernatant from a soil suspension and the
effluent from an anaerobic/microaerobic column treating azo dye waste (Lipscomb et al.,
2008). This column had previously been seeded with Shrewanella sp., a bacteria shown
to be capable of azo bond cleavage (Pearce et al., 2006).

Figure 8-1 shows the presence of a thin biofilm on the surface of the membrane material,
and the turbid nature of the bulk liquid, after 12 days of reactor operation. This relatively
high turbidity can be attributed to the high COD of the influent media.

Figure 8-1: Biofilm development after 12 days of operation

180
8.2.2 Method of colour removal
Observations imply that removal of colour from the liquid bulk occurred in three ways:
(i) Adsorption onto tubing
(ii) Adsorption onto biomass
(iii) Azo bond cleavage

The adsorption of colour onto tubing and biomass was of greatest significance during
early operation of the reactor. Tubing was observed to quickly (within 48 hours) take on
an orange colour. Once this initial adsorption occurred, the adsorptive capacity of the
tubing was exhausted, and adsorption onto the tubing is considered to have had negligible
effect on colour removal.

During operation, biomass built up in the holding vessel, and was periodically removed
by filtering the contents through laboratory filter paper. The filtrate was then returned to
the system.

During early operation, biomass removed as the residue was orange in colour, implying
that colour removal during this phase of operation was achieved through adsorption onto
biomass. When the colour removal rates were at their greatest, the residue was grey in
colour, implying that once the bacteria in the biofilm had become adapted to the system,
colour removal was achieved through cleavage of the azo bond, rather than through
adsorption onto biomass (Figure 8-2).

181
 Figure 8-2: Removed biomass initially (left) and during greatest colour removal

8.2.3 pH
Similar changes to the influent and effluent pH were observed during the operation of the
reactor, as exemplified by the data from Run 2 shown in Figure 8-3.

5
pH

4 Influent

3 Effluent

0
60 70 80
day

Figure 8-3: Influent and effluent pH during Run 2

The pH of the influent wastewater displays significant change during run 2, with values
in the range 4.0 – 7.1 being observed during run 2. Furthermore, this variation can be
seen to be cyclical, with the influent pH dropping over the course of 3-4 day periods
which can be connected to the changing of the feed jar.

182
This drop in pH was accompanied by a sharp, pungent smell similar to that of acetic acid
and can possibly be attributed to acidity produced by fermentation taking place in the
feed jar. When compared with work in MABfR A, where very little variation was
observed in the pH of the influent, this drop highlights the difficulty in keeping high
strength synthetic wastewaters aseptic; even when correct procedures for preservation are
followed (Alef, 1995).

The pH of the effluent displays less variation, with pH values in the range 6.4 – 7.5 being
observed and the two lowest pH values being seen at times when the pH of the influent
was also at its lowest and 11 out of 15 observations being greater than 7.0. In general,
effluent pH values were higher than those of the influent as acidic fermentation products
were readily broke down in the presence of oxygen by the biofilm. Relatively stable
effluent pH values were seen due to the self-regulatory nature of the biological processes
taking place in the biofilm.

8.2.4 Ammoniacal nitrogen concentration

Figure 8-4 shows the influent and effluent ammoniacal nitrogen concentrations
throughout the dye degradation investigation. As in previous chapters, vertical lines are
used to indicate the beginning of each experimental run.

183
 200

160
[Amm-N] (mg l-1)

120
influent
effluent
80

40

0
0 25 50 75 100 125
day

Figure 8-4: Ammoniacal concentrations throughout dye investigation

With hindsight, the use of peptone as a source of ammoniacal nitrogen was a poor choice.
Peptone contains approximately 40% nitrogen by weight and is commonly included in
mineral media recipes for experimental degradation investigations as a source of
ammoniacal nitrogen (e.g. Saratale et al., 2009).

However, the hydrolysis of peptone to aqueous ammoniacal nitrogen does not go to

completition under the conditions under which the synthetic dye waste was prepared.
Instead, the peptone breaks down during the residence in the reactor, and the higher
effluent ammoniacal nitrogen concentration can be attributed to this phenomenon
(Lieben, 1943).

8.2.5 Nitrate concentration

No significant nitrate concentrations were observed in the effluent throughout the
operation of the MABfR. Three possible explanations for this observation are plausible.
Lipscomb et al. (2008) measured redox potentials during operation of the biological
column from which effluent was taken to seed the reactor used in this work. The

184
researchers found the highest colour removal was achieved at similar redox potentials to
those contained in the literature as being ideal for denitrification.

As such, any nitrate that was produced by nitrification in the aerobic zone could feasibly
be denitrified in the anoxic zone, leading to the low effluent nitrate concentrations
observed throughout the operation of this reactor.

Alternatively, the absence of significant nitrate detected in the bulk phase could be due to
insignificant nitrification taking place. Work by He & Bishop (1994) found the presence
of AO7 to have an inhibitory effect on nitrification, even when the dye concentration was
less than 5 mg l-1, with the activity of AOB being more sensitive to AO7 than NOB.

A third possibility is that the absence of significant nitrate in the effluent could be due to
aerobic heterotrophs outcompeting nitrifiers for space and oxygen in the biofilm (Zhang
et al., 1995). The influent COD concentrations used in this reactor were in the range 730
- 930 mgCOD l-1. At this COD concentration, the growth rate of aerobic heterotrophs is
very rapid, and prevents nitrifying bacteria from establishing themselves in the biofilm,
with no associated production of nitrate.

Without the use of complex molecular biological techniques which were not available for
use in this project, it is impossible to say which of these possible explanations is valid.

8.2.6 UV/vis spectrometry

A calibration was carried out by preparing a 125 mg l-1 stock solution of AO7, which was
diluted to give solutions with AO7 concentrations of 62.5 mg l-1, 31.3 mg l-1, 15.6 mg l-1,
20.0 mg l-1, 10.0 mg l-1, 5.0 mg l-1 and 2.5 mg l-1. The absorbance spectra of these
solutions over the range 400 – 550 nm is shown in Figure 8-5 below.

185
 3
62.5 mg l-1
2.5 31.3 mg l-1
15.6 mg l-1
Absorbance (AU)

2
20.0 mg l-1
1.5 10.0 mg l-1
5.0 mg l-1
1
2.5 mg l-1

0.5

0
400 450 500 550
wavelength (nm)

Figure 8-5: Absorbance spectra for standard AO7 solutions

Examination of the data generated by the absorbance spectroscopy reveals that the λmax,
the wavelength at which maximum absorbance occurs, is located at a value of 478 nm for
the AO7 used in this study. This value is similar to the 480 nm found by Ong et al. (Ong
et al., 2005) and 483 nm found by Coughlin et al. (Coughlin et al., 2002) in AO7
degradation studies.

Using the respective absorbencies of the standard solutions at 478 nm, the calibration
curve shown in Figure 8-6 can be drawn.

186
 70

60
AO7 concentration (mg l-1)

50

40

30

20

10

0
0 0.5 1 1.5 2 2.5 3
Absorbance at λmax (AU)

Figure 8-6: Calibration curve for AO7 concentration

Linear regression of the data shown in Figure 8-6 yields the relationship in Equation 8-1,
with a correlation fit coefficient of 0.989, which can then be used to relate the AO7
concentration in collected samples to the absorbance of the samples at 478 nm.

𝐶𝐴𝑂7 = 23.02𝐴478 Equation 8-1

Where: CAO7 = Acid orange 7 concentration (mg l-1)

A478 = Absorbance at 478 nm (AU)

When used to calculate influent AO7 concentrations, Equation 8-1 gives the data
displayed in Table 8-2.

Table 8-2: Average calculated AO7 concentrations in Runs 1-3 (standard errors shown)
Run 1 2 3
Average calculated 23.0±1.1 24.0±1.8 24.1±0.8
CAO7,inf

As seen in Table 8-2, the average calculated concentrations in Runs 1-3 are in the range
23.0 - 24.1 mg l-1, despite the influent media containing only 20 mg l-1 of AO7. This is

187
partly due to errors in the preparation of the influent solutions and concentration during
the sterilisation procedure, but is also attributable to the presence of Ferric Chloride and
Riboflavin in the influent media.

Ferric chloride and riboflavin dissolve in water to give pale yellow coloured solutions.
Figure 8-7 shows solutions of these components, at the concentrations used in the influent
media in Runs 2 and 3, alongside an AO7 solution for comparison.

Figure 8-7: Solutions of Acid Orange 7 (left), Fe (III) Chloride and Riboflavin (right) at the
concentration used in the influent media

As displayed in Figure 8-8, these compounds also absorb light in the same region in
which the λmax of AO7 is located. As shown by the red line in Figure 8-8, this gives
cumulative absorbance effects, meaning that the apparent AO7 concentration (obtained
by calculation using Equation 8-1) is higher than the true value.

188
 1.4

1.2

1
Feed Run 2 Day 2
Absorbance (AU)

0.8 20 mg l-1 AO7

10 mg l-1 Fe(III)Cl3
0.6
10 mg l-1Riboflavin
0.4

0.2

0
400 450 500 550 600
wavelength (nm)

Figure 8-8: Cumulative absorbance effects

This cumulative absorbance effect leads to the overestimation of the dye concentration as
exemplified by the data shown in Table 8-2. Although it causes an error in the
ascertaining of AO7 concentration, the rate of AO7 removal can still be found using
photospectrometry.

In this work, it is assumed that riboflavin acts purely as a redox mediator and that it is not
broken down by the action of the biofilm. As such, the absorbance attributable to
riboflavin at 478 nm will not differ in influent and effluent samples.

Although iron plays a key role in microbial life and as such will be taken up by the
biofilm (Madigan & Martinko, 2006), it is only present in the influent media in trace
amounts. At the concentration used, it does not give a visibly coloured solution (Figure
8-7) and gives an absorbance of only 0.015 AU at 478 nm. As such, the contribution of
Fe(III)Cl3 can be considered negligible.

As such, the changes in AO7 concentration can be considered reliable and used to
calculate AO7 removal rates.

189
8.2.6.1 λmax shift
Also visible in Figure 8-8 is a decrease in the wavelength at which maximum absorbance
is observed. This phenomenon can be attributed to two factors.

Firstly, as previously stated, riboflavin is a yellow solution (Figure 8-7), with a λmax of
approximately 450 nm (green line in Figure 8-8). As also seen in Figure 8-8, Riboflavin
also gives significant absorbance light in the range 450 – 475 nm, giving increased
absorbance over this range and contributing to the shift in λmax of the feed solution.

A shift in the λmax of azo dyes was also observed by Hanna (2005), in work using
dolomite for removal of dyes from wastewater via adsorption. The author attributed this
shift to the change of pH that occurred when wastewater was brought into contact with
the basic dolomite, with associated changing in structure of the dye alternating the
electron density around the chromophore. It is possible a similar effect was observed here
as the dye was partially broken and conditions in the reactor changed.

8.3 Colour removal

The fractional colour removal in Runs 1 – 3 is shown in Figure 8-9.

1.0

0.8

0.6
(Cinf-Ceff)/Cinf

0.4

0.2

0.0
0 20 40 60 80 100 120
day

Figure 8-9: Colour removal in Runs 1-3

190
Initially, colour removals were low, with only 15% removal being observed on Day 20,
when the first samples were taken for photospectrometric analysis. From this point, the
colour removal follows a general upwards trend as the slow growing anoxic bacteria
which are responsible for cleavage of the azo bond increase in numbers, reaching a
maximum of 0.87 fractional removal on day 69.

A drop in colour removal is seen around day 75, which coincides with the reduction in
the TOD of the influent media. The drop in TOD of the influent media will have had an
effect on the position of the anoxic zone. As a result, the bacteria which exploited the
niche for cleavage of the azo bond during Runs 1 and 2 no longer experience anoxic
conditions and the colour removal rate drops off.

After this drop, the colour removal increases again, with a maximum observed removal
of 0.90 on days 113 and 120. This recovery can be attributed to the azo bond cleaving
bacteria exploiting the new location of the anoxic zone in significant numbers.

8.3.1 AO7 Removal rate

After using median analysis to eliminate rogue data points, the obtained average AO7
removal rates are shown in Table 8-3 on a mass/membrane area basis, mass/reactor
volume basis and moles/reactor volume basis.

Table 8-3: AO7 removal rates in Runs 1-3

Run rAO7 (g m-2day-1) rAO7 (g m-3day-1) rAO7 (mmol m-3day-1)
1 0.38±0.02 1.57±0.10 4.81±0.30
2 1.01±0.07 4.16±0.30 12.76±0.93
3 0.85±0.05 3.49±0.20 10.70±0.60

Examination of the data in Table 8-3 reveals that the rate of AO7 removal was
significantly boosted by the addition of riboflavin to the influent media in Run 2, with the
AO7 removal rate 2.6 times higher in Run 2 compared to Run 1. This is consistent with

191
the findings of Van der Zee et al. (2003b), who observed approximately a five-fold
reduction in decolourisation time in a batch system.

However, this may be misleading. From examination of Figure 8-9, it could be argued
that the colour removal was still increasing, and that stable effluent concentrations had
not been reached at the end of Run 1. As such, the average removal rate calculated by
median analysis is lower than if the removals at the end of the experimental run was used,
and it can not be stated with certainty that the AO7 removal rate was boosted by the
addition of riboflavin.

It may be the case that riboflavin, or another effective redox mediator, was already being
naturally produced by the biofilm, and that colour removal was therefore not limited by
the presence of a redox mediator, but rather limited by the numbers of slow growing,
anoxic bacteria, who are responsible for the cleavage of the azo bond. It is for this reason
that the obtained removal rate is higher in Run 2, when the biofilm had reached maturity.

This uncertainty could have been avoided by continuing each experimental run until
stable values had been obtained, but this was not possible as this work took place at the
end of the research phase and was subject to time constraints.

The average removal rate in Run 3 was slightly lower than that obtained in Run 2. As
discussed above, this is due to a change in the location of the anoxic zone. This would not
had occurred were Run 3 extended in duration as the removals at the end of Run 3 were
the highest observed in the investigation.

The obtained removal rates cannot be directly compared with other published removal
rates contained in literature, as this is the believed to be the first use of a MABfR to
decolourise azo dyes. Instead, the results are compared on to alternative biofilm reactor
studies in Table 8-4.

192
 Table 8-4: Comparison of AO7 removal rates in literature
Author Method Maximum rAO7
This study MABfR 4.16 g m-3day-1
12.76 mmol m-3day-1
Coughlin et al. (2002) Rotating Drum Biofilm 1296 g m-3day-1
Ong et al. (2005) Packed Column Biofilm 4342 mmol m-3day-1

On first examination, the AO7 removal rates obtained in this study do not compare
favourably with those selected from the published research. The maximum rate of AO7
removal, obtained in Run 2 of the investigation, is three orders of magnitude lower than
that reported by both Coughlin et al. (2002) using a Rotating Drum Biofilm Reactor, and
Ong et al. (2005) using a Packed Column Biofilm reactor.

However, the specific surface area used here is significantly lower than what is possible
for a MABfR, being two orders of magnitude lower than that used by Pankhania (1994,
1999). If this specific membrane area, 515 m2m-3, is used in calculating the removal rate
on a volume basis, figures of 522 g m-2day-1 and 1490 mmol m-2day-1 are obtained and
the comparison with published research is much more favourable. Although still
approximately 50% lower than the rates contained in the literature, they are promising
given the early nature of this use of MABfR technology, and could possibly be increased
through optimisation of the process.

8.4 COD removal

The influent and effluent COD concentrations in Runs 1 - 3 are shown in Figure 8-10.

193
 1200

1000

800
[COD] (mg l-1)

influent
600
effluent

400

200

0
0 25 50 75 100 125
day

Figure 8-10: Influent and effluent COD concentrations during Runs 1 - 3

There is considerable variation in the influent COD concentration. This is due to the
difficulties in keeping high strength synthetic wastewaters aseptic, as discussed
previously in relation to the observed variation in the pH of the influent wastewater.

Less variation is seen in effluent concentrations, which are relatively well grouped in the
range 100 – 300 mg l-1. Effluent concentrations are more closely grouped in Runs 2 and
3, which can be attributed to the inherent stability of a more mature biofilm.

8.4.1 COD removal rate

The average COD removal rates obtained during the three experimental runs of the dye
MABfR is displayed in Table 8-5.

Table 8-5: COD removal rates in Runs 1 - 3

Run rCOD (gCOD m-2day-1)
1 29.4±1.4
2 37.9±1.4
3 21.7±1.7

194
On first examination of the data contained in Table 8-5, the addition of riboflavin in run 2
appears to have boosted COD removal, with the removal rate in Run 2 almost 30% higher
than that obtained in Run 1. However, this may also be due to the biofilm reaching
maturity during Run 2 and it cannot be concluded that riboflavin boosts COD removal
without further experimentation.

In comparison to those reported in the previous chapters, these removal rates are slightly
higher (the highest removal rate obtained in Chapter 6 was 21.5±1.7 gCOD m-2day-1).
This can be partly explained by Monod kinetics, as the COD loadings in this chapter were
in the range 28.4 – 46.4 gCOD m-2day-1 compared to 23.8 – 42.4 gCOD m-2day-1.

Additionally, as discussed in Section 8.2.5 above, the high COD concentrations in this
MABfR lead to the rapid growth of aerobic heterotrophs, preventing nitrifiers from
establishing themselves in the biofilm. As such, no oxygen is consumed by ammonia
oxidation, and all of the oxygen provided to the biofilm is utilized by aerobic
heterotrophs, with associated higher COD removal rates (Zhang et al., 1995).

The obtained COD removal rate in Run 3 was lower than that obtained in Run1 and Run
2. There are two possible explanations for this occurring. As observed in the studies with
MABfR A, the change in composition of the influent wastewater may have affected the
biofilm in such a way as to favour nitrification, and as such, less oxygen was available for
COD reduction. The difficulties in achieving complete hydrolysis of the nitrogen source
(described in Section 8.2.4) prevented this being qualified by effluent analysis.

Alternatively, as, in order to not disturb the microorganisms responsible for colour
removal which are located in the outer layer of the biofilm, the reactor was operated
without backwash. As the biofilm density is controlled by the strength of wastewater (Hu
et al., 2008), it is possible that the biofilm had grown to a sufficient thickness to act as a
barrier to the transfer of oxygen from the membrane into the biofilm and the transfer of
substrate into the biofilm.

195
In a similar way to that described in Chapter 7, and as also observed in pilot trials by
Pankhania et al. (1999) this mass transfer resistance leads to lower pollutant removal
rates, as evidenced here by the observed drop in COD removal rate.

8.5 Worked example

Using the obtained average AO7 removal rate obtained in this study, the membrane area
required to treat dyehouse effluent from a local carpet factory can be calculated (Ulster
Carpet Mills, Portadown, U.K.).

The factory currently discharges 1800 m3wk-1 of spent azo dye wastewater into the town
sewer (Wilkinson 2007), and has an azo dye concentration assumed here to be equivalent
in absorbance to the absorbance of a 20 mg l-1 of AO7.

There is currently no colour consent placed upon the carpet factory, but it is assumed for
the purposes of this calculation to be equivalent to the absorbance of a 5 mg l-1 solution
of AO7.

Flowrate: 𝑄 = 1800 𝑚3 𝑤𝑘 −1
𝑄 = 257 𝑚3 𝑑𝑎𝑦 −1

Required daily removal (loading basis): 𝐿𝐴𝑂7,𝑟𝑒𝑞 = (𝐶𝑖𝑛𝑓 − 𝐶𝑐𝑜𝑛 ) ∗ 𝑄

𝐿𝐴𝑂7,𝑟𝑒𝑞 = (20 − 5) ∗ 257
𝐿𝐴𝑂7,𝑟𝑒𝑞 = 3857 𝑔 𝑑𝑎𝑦 −1

Required membrane area: 𝑟𝐴𝑂7,𝑟𝑒𝑞

𝐴𝑚,𝑟𝑒𝑞 =
𝑟𝐴𝑂7
3857
𝐴𝑚,𝑟𝑒𝑞 =
1.01
𝐴𝑚,𝑟𝑒𝑞 = 3819 𝑚2

196
Total membrane area requirement – 𝐴𝑚,𝑟𝑒𝑞,𝑇 = 1.2 ∗ 𝐴𝑚,𝑟𝑒𝑞
giving 20% additional allowance: 𝐴𝑚,𝑟𝑒𝑞,𝑇 = 1.2 ∗ 3819
𝐴𝑚,𝑟𝑒𝑞,𝑇 = 4583 𝑚2

Using the specific surface area used in this investigation (4.4 m2m-3), the required
membrane area corresponds to a reactor volume of 1040 m3. This is unfeasibly large, but
the specific surface area used in this study is small compared to those reported in
literature. Using the highest specific area from literature for a MABfR, reported by
Pankhania et al. (1999, 1994) as 510 m2m-3, a reactor volume of 9.0 m3 is obtained.

However, although the reactor volume may be feasible in scale, the cost of membrane
may not be. The membrane unit purchased in order to produce the pilot scale BioSettler
system provided 7.2 m2 of membrane area and cost €1800. At these prices, the required
membrane area would cost more than €1M to purchase, and it is unlikely that the use of
MABfR would prove the most cost effective solution. Other techniques, like those
discussed in Section 8.3.1, are likely to be more financially viable.

8.6 Conclusions
This work has demonstrated that the cleavage of the azo bond, with associated colour
reduction, and simultaneous COD removal can take place in the MABfR under controlled
conditions – which has not previously been reported in literature.

Although demonstrated in published research using batch studies, the rate of azo dye
degradation was not shown to increase following the addition of riboflavin. Although not
conclusively shown here, it is believed that this result was seen as availability of
riboflavin, or another redox mediator, was not limiting the rate of azo bond cleavage.

COD loading must be matched to inlet pressure, and therefore supply rate of oxygen, in
order to achieve stable colour removal. Variations in oxygen demand of the influent
wastewater lead to a movement of the anoxic zone, and as the bacteria responsible for

197
cleavage of the azo bond are slow growing, the decolourising ability of the biofilm is
reduced.

Using the removal rates obtained in this work, the membrane area required in order to
successfully decolourise from a local dye house was calculated. Although the required
membrane area could be accommodated within a reactor size which could feasibly
installed at the carpet factory, the costs involved in the purchase of the membrane itself
were significant.

For this reason, it is unlikely that the MABfR will prove to be a feasible technology for
direct treatment of textile waste. However, given the expense and disposal issues
associated with chemical and physical methods, more research is required into biological
treatment options involving sequential anoxic and aerobic phases. Rotating Drum
Contactors and Packed Bed Reactors are currently suitable candidate technologies.

However, given further development and optimisation of the process, costs of a MABfR
could be reduced and removal rates increased to a point where MABfRs can be
considered an economically as well as technically viable technique for treatment of azo
dye wastewater.

198
9 Conclusions and Further Work

9.1 Summary of conclusions

The main conclusions that can be drawn from this work are:

Mass Transfer Studies

 The bubble free transfer of oxygen to water using tubular silicone rubber and
polyethersulphone membranes was explored. The effect of inlet air pressure, air
flowrate and water flowrate was investigated, allowing oxygen flux to be
measured and overall mass transfer coefficients to be evaluated.
 Obtained average oxygen fluxes ranged from 0.6 to 2.4 O2m-2h-1, with higher
fluxes being obtained with silicone rubber in all situations. Overall K values were
in the range 2.7x10-5 to 5.4x10-5 ms-1, similar to those observed by other
researchers.
 Wilson plots were used to obtain the individual KL and KM, allowing relationships
to be developed linking KL to Re, the dimensionless measure of water side
turbulence. The obtained empirical relationships were Sh = 1.99Re1.07Sc0.33 for the
silicone rubber membrane module and Sh = 0.96Re1.05Sc0.33 for the
polyethersulphone module.

Membrane Aeration Biofilm Studies

 Two lab scale MABfRs were designed, built and operated; one treating synthetic
secondary municipal WwTW effluent and one treating synthetic dyehouse
wastewater.
 For the municipal WwTW reactor, COD, Ammoniacal and Total-Nitrogen were
all successfully removed at a variety of inlet air pressures and wastewater
compositions, with removal rates as high as 13.8 g m-2day-1 , 1.92 g m-2day-1 and
1.42 g m-2day-1 respectively being achieved.
 The results were used to develop a simple model, allowing reactor performance to
be predicted outside of the range of wastewater compositions used in this study.

199
  The dyehouse wastewater reactor was used to successfully decolourise and
remove COD from synthetic dyehouse wastewater, with an AO7 removal rate of
1.01 g m-2day-1 being obtained. This is believed to be the first time azo dyes have
been broken down in a Membrane Aerated Biofilm.

Design and Operation of a pilot scale BioSettler

 Using the information obtained from laboratory studies and best practice from the
wastewater treatment industry, a pilot scale BioSettler unit was designed and
constructed, combining membrane aerated biofilms and inclined plate settlers into
a single innovative technology.
 The unit was operated and monitored at two NIW WwTWs and simultaneous
suspended soilds, BOD, Ammoniacal and Total Nitrogen removal was obtained –
thus proving the concept of the BioSettler.
 Removal rates up to 33.1 gm-2day-1 of BOD and 13.4 gm-2day-1 of ammoniacal
nitrogen were achieved; higher than those obtained in the lab scale studies.

9.2 Recommendations of further work

In Chapter 6, the effect of C:N ratio on the various biological processes taking place in
the biofilm was discussed. A clear relationship was established between the relative
amounts of oxygen being consumed by both aerobic heterotrophy and nitrification (and
therefore the relative rates of these processes) over the range of C:N ratios used in the
synthetics wastewaters used in this process. Expansion of this work to include C:N ratios
outside of those found here are required to establish whether the MABfR is a viable
technology for the treatment of wastewaters with high C:N ratios, such as those
originating from food processing premises contain high amounts of sugar, for example.

No clear relationship between the influent composition and rates of denitrification

obtained was established. A reason for this was not found in the literature. This paucity
can explained due to the absence of a research driver to explore this issue. In the
activated sludge process, it is simple to achieve complete denitrification by recycling

200
secondary effluent and mixing it with primary effluent. This ensures that the produced
nitrate sees high COD concentration which is utilised by denitrifiers. This configuration
is not possible in the BioSettler where concentrations of both organic carbon and nitrate
are low. Further research is required around this issue to develop a full understanding of
denitrification in the MABfR, allowing optimisation of Total Nitrogen removal in the
BioSettler.

The pilot scale studies described in Chapter 7 were successful in proving the concept of
the BioSettler. BOD, Suspended Solids, Ammoniacal and Total Nitrogen were all
successfully and simultaneous treated in the pilot scale unit using ‘real’ secondary
effluent from a municipal WwTW. However, the pilot scale studies also exposed two
major limitations of the unit – the removal of settled sludge from the bottom of the unit
was not effective, leading to the sludge rising and fouling and causing damage to the
membrane arrays, which were not strong enough to bear the weight of the sludge.

The issue with sludge removal is simply one of scale – a larger unit can be fitted with
conical hoppers from which sludge can easily be removed by gravity. Membranes with
higher tensile strength than the ones used in this work are required to prevent the second
issue from arising; either filtration membranes with a larger wall thickness or suitable
dense membranes are required. Overcoming these two problems is essential for the
BioSettler to establish itself as a wastewater treatment technology.

201
References

Acrivos, A. & Herbolzheimer, E. (1979), Enhanced sedimentation in settling tanks with

inclined walls; Journal of Fluid Mechanics 92 435-57.

Ahmadi Motlagh, A.R., Voller, V.R. & Semmens, M.J. (2006), Advective flow through
membrane-aerated biofilms: Modeling results; Journal of Membrane Science 273 1-
2 143-151.

Ahmed, T., Oakley, B.T., Semmens, M.J. & Gulliver, J.S. (1996), Nonlinear deflection of
polypropylene hollow fiber membranes in transverse flow; Water Research 30 2
431-439.

Ahmed, T. & Semmens, M.J. (1996), Use of transverse flow hollow fibers for bubbleless
membrane aeration; Water Research 30 2 440-446.

Ahmed, T., Semmens, M.J. & Voss, M.A. (2004), Oxygen transfer characteristics of
hollow-fiber, composite membranes; Advances in Environmental Research 8 3-4
637-646.

Aleboyeh, A., Olya, M.E. & Aleboyeh, H. (2009), Oxidative treatment of azo dyes in
aqueous solution by potassium permanganate; Journal of Hazardous Materials 162
2-3 1530-1535.

Alef, K. (1995), Nutrients, sterilization, aerobic and anaerobic culture techniques In:
Methods in Applied Soil Microbiology and Biochemistry, Academic Press Limited,
London.

ASCE (1992), ASCE Standard - Measurement of Oxygen Transfer in Clean Water,

ASCE/EWRI 2-06, ASCE, Reston, VA, U.S.A.

Baker, R.W., Cussler, E.L., Eykamp, W., Koros, W.J., Riley, R.L. & Strathmann, H.
(1991), Membrane Separation Systems: Recent Developments and Future
Directions, Noyes Data Corporation, Park Ridge, New Jersey, U.S.A.

Bishop, P.L., Zhang, T.C. & Fu, Y.-. (1995), Effects of biofilm structure, microbial
distributions and mass transport on biodegradation processes, Pergamon Press Inc,
Copenhagen, Den.

Bitton, G. (2005), Wastewater Microbiology, 3rd edn, John Wiley & Sons, New York,
USA.

Boycott, A.E. (1920), Sedimentation of Blood Corpuscles; Nature 104 2621 532.

202
Brindle, K., Stephenson, T. & Semmens, M.J. (1998), Nitrification and oxygen utilization
in a membrane aeration bioreactor; Journal of Membrane Science 144 1-2 197-209.

Brindle, K. & Stephenson, T. (1996), Nitrification in a bubbleless oxygen mass transfer

membrane bioreactor; Water Science and Technology 34 9 pt 5 261-267.

British Water (2013), Code of Practice: Flows and Loads 4 - Sizing Criteria, Treatment
Capacity for Sewage Treatment Systems.

Brown, D. & Labouruer, P. (1983), The aerobic biodegradability of primary aromatic

amines; Chemosphere 12 249-324.

BSi (2002a), British Standard BS EN 12255-4:2002, Wastewater treatments plants - Part

4: Primary Settlement.

BSi (2002b), British Standard BS EN 12255-6:2002, Wastewater treatments plants - Part

6: Activated sludge processes.

Casey, E., Glennon, B. & Hamer, G. (2000a), Biofilm development in a membrane-

aerated biofilm reactor: Effect of flow velocity on performance; Biotechnology and
Bioengineering 67 4 476-486.

Casey, E., Glennon, B. & Hamer, G. (2000b), Biofilm development in a membrane-

aerated biofilm reactor: Effect of intra-membrane oxygen pressure on performance;
Bioprocess and Biosystems Engineering 23 5 457-465.

Casey, E., Glennon, B. & Hamer, G. (1999), Oxygen mass transfer characteristics in a
membrane-aerated biofilm reactor; Journal of Engineering and Applied Science 62
2 183-192.

Chambless, J.D. & Stewart, P.S. (2007), A Three-Dimensional Computer Model Analysis
of Three Hypothetical Biofilm Detachment Mechanisms; Biotechnology and
Bioengineering 97 6 1573-1584.

Clesceri, L.S., Greenberg, A.E. & Eaton, A.D. (1998), Standard Methods for the
Examination of Water and Wastewater, 20th edn, American Public Health
Association, American Water Works Association & Water Environment Federation,
Baltimore, MD, U.S.A.

Cook, R.L. & Childress, J.J. (1978), Performance of Lamella Thickeners in Coal
Preparation Plants; Mining Engineering (Littleton, Colorado) 30 5 566-571.

Côté, P., Bersillon, J. & Huyard, A. (1989), Bubble-free aeration using membranes: mass
transfer analysis; Journal of Membrane Science 47 1-2 91-106.

203
Côté, P., Bersillon, J., Huyard, A. & Faup, G. (1988), Bubble-free aeration using
membranes: process analysis; Journal of the Water Pollution Control Federation 60
11 1986-1992.

Coughlin, M.F., Kinkle, B.K. & Bishop, P.L. (2003), High performance degradation of
azo dye Acid Orange 7 and sulfanilic acid in a laboratory scale reactor after seeding
with cultured bacterial strains; Water Research 37 11 2757-2763.

Coughlin, M.F., Kinkle, B.K. & Bishop, P.L. (2002), Degradation of acid orange 7 in an
aerobic biofilm; Chemosphere 46 1 11-19.

Coulson, J.M., Richardson, J.F., Backhurst, J.R. & Harker, J.H. (1999a), Flow in Pipes
and Channels In: Chemical Engineering: Volume 1, 6th edn, Butterworth
Heinemann, Oxford.

Coulson, J.M., Richardson, J.F., Backhurst, J.R. & Harker, J.H. (1999b), Heat Transfer
In: Chemical Engineering: Volume 1, 6th edn, Butterworth Heinemann, Oxford.

Cristovao, R.O., Tavares, A.P.M., Loureiro, J.M., Boaventura, R.A.R. & Macedo, E.A.
(2009), Treatment and kinetic modelling of a simulated dye house effluent by
enzymatic catalysis; Bioresource Technology 100 24 6236-6242.

Cussler, E.L. (1997), Membranes In: Diffusion, Mass Transfer in Fluid Systems, 2nd edn,
Cambridge University Press, Cambridge, U.K.

Darby, R. (1996), Compressible flows In: Chemical Engineering Fluid Mechanics,

Marcel Dekker Inc., New York.

DARD & DOE (2010), Nitrates Action Programme 2011 - 2014 & Phosphorous
Regulations, Northern Ireland Environment Agency, Lisburn, U.K.

Davies, L.C., Pedro, I.S., Novais, J.M. & Martins-Dias, S. (2006), Aerobic degradation of
acid orange 7 in a vertical-flow constructed wetland; Water Research 40 10 2055-
2063.

de Hoxar, D. (2000), Separator plates put sludge in a spin; Filtration and Separation 37 8
32-33.

DEFRA (2002), Sewage Treatment in the UK - UK Implementation of the EC Urban

Waste Water Treatment Directive, DEFRA Publications, London, UK.

Demir, A. (1995), Determination of settling efficiency and optimum plate angle for plated
settling tanks; Water Research 29 2 611-616.

Dincer, A.R. & Kargi, F. (2000), Kinetics of sequential nitrification and denitrification
processes; Enzyme and Microbial Technology 27 1-2 37-42.

204
Dindore, V.Y., Brilman, D.W.F. & Versteeg, G.F. (2005), Modelling of cross-flow
membrane contactors: physical mass transfer processes; Journal of Membrane
Science 251 1-2 209-222.

DOE NI & DARD NI (2004), Nitrates Directive - Second Consultation Paper: Proposal
for the Protection of Northern Ireland's Surface and Groundwaters.

Doig, S.D., Boam, A.T., Livingston, A.G. & Stuckey, D.C. (1999), Mass transfer of
hydrophobic solutes in solvent swollen silicone rubber membranes; Journal of
Membrane Science 154 1 127-140.

dos Santos, L.M.F. & Livingston, A.G. (1995), Membrane-attached biofilms for VOC
wastewater treatment. II: effect of biofilm thickness on performance; Biotechnology
and Bioengineering 47 1 90-95.

Downing, L.S. & Nerenberg, R. (2008a), Effect of bulk liquid BOD concentration on
activity and microbial community structure of a nitrifying, membrane-aerated
biofilm; Applied Microbiology and Biotechnology 81 1 153-162.

Downing, L.S. & Nerenberg, R. (2008b), Total nitrogen removal in a hybrid, membrane-
aerated activated sludge process; Water Research 42 14 3697-3708.

Doyle, C. (2011), Bubblefree membrane aeration using hydrophobic polyestersulphone

membranes, Queen's University, Belfast.

Dreszer, C., Flemming, H.-., Zwijnenburg, A., Kruithof, J.C. & Vrouwenvelder, J.S.
(2014), Impact of biofilm accumulation on transmembrane and feed channel
pressure drop: Effects of crossflow velocity, feed spacer and biodegradable nutrient;
Water Research 50 200-211.

Ergas, S.J. & Reuss, A.F. (2001), Hydrogenotrophic denitrification of drinking water
using a hollow fibre membrane bioreactor; Journal of Water Supply: Research and
Technology - AQUA 50 3 161-171.

Essila, N.J., Semmens, M.J. & Voller, V.R. (2000), Modeling biofilms on gas-permeable
supports: Concentration and activity profiles; Journal of Environmental Engineering
126 3 250-257.

European Council (2000), Directive 2000/60/EC of the European Parliament and of the
Council of 23 October 2000 establishing a framework for Community action in the
field of water policy .

European Council (1991a), Council Directive 91/271/EEC of 21 May 1991 concerning

urban waste-water treatment.

205
European Council (1991b), Council directive 91/676/EEC of 12 December 1991
concerning the protection of waters against pollution caused by nitrates from
agricultural sources.

Fang, Y., Novak, P.J., Hozalski, R.M., Cussler, E.L. & Semmens, M.J. (2004),
Condensation studies in gas permeable membranes; Journal of Membrane Science
231 1-2 47-55.

Fernandes, A., Morão, A., Magrinho, M., Lopes, A. & Gonçalves, I. (2004),
Electrochemical degradation of C. I. Acid Orange 7; Dyes and Pigments 61 3 287-
296.

Fernandez-Seara, J., Uhia, F.J., Sieres, J. & Campo, A. (2005), Experimental apparatus
for measuring heat transfer coefficients by the Wilson plot method; European
Journal of Physics 26 3 1-11.

Fivelman, Q. (2010), Stage Two Commerialisation Plan - BioSettler (Confidential), IP

Pragmatics, London, U.K.

Gerlach, S.A. (1994), Oxygen conditions improve when the salinity in the baltic sea
decreases; Marine Pollution Bulletin 28 7 413-416.

Gingell, R. & Walker, R. (1971), Mechanisms of azo reduction by Streptococcus faecalis

II, the role of soluble falvins; Xenobiotics 1 231-239.

Grady, C.P.L., Daigger, G.T. & Lim, H., C. (1999), Biological Wastewater Treatment,
Second Edition, Revised and Expanded edn, Marcel Dekker, Inc., New York, USA.

Gray, N.F. (2004), Biology of Wastewater Treatment, 2nd edn, Imperial College Press,
London, U.K.

Green, D.W. & Perry, R.H. (2008), Section 2: Physical and Chemical Data In: Perry's
Chemical Engineers' Handbook, 8th edn, McGraw-Hill.

Groom,E., Murray,S. and Ferguson,J. The Queen's University of Belfast. (2009),

Improvements Relating to Water Treatment PCT/GB2008/004193 .

Hach Company (2007), (June 2007-last update), DR 2800 Spectrophotometer Procedures

Manual. Available online:
http://www.hach.com/fmmimghach?/CODE%3ADOC022.53.00725-200719229|1
[2009, 03/02] .

Hach Company (2001), (August 2001-last update), sensION Ammonia Gas Combination
Electrode. Available online:
http://www.hach.com/fmmimghach?/CODE%3A5192788435%7C1 [2009, 03/02] .

206
Hanna, J. (2005), Industrial Wastewater Treatment using Dolomite and Dolomitic
Sorbents, Queen's University Belfast.

Hasar, H., Xia, S., Ahn, C.H. & Rittmann, B.E. (2008), Simultaneous removal of organic
matter and nitrogen compounds by an aerobic/anoxic membrane biofilm reactor;
Water Research, 42 15 4109-4116.

Hazen, A. (1904), On Sedimentation; Proceedings of the ASCE 53 45-88.

He, S., Xue, G. & Wang, B. (2009), Factors affecting simultaneous nitrification and de-
nitrification (SND) and its kinetics model in membrane bioreactor; Journal of
Hazardous Materials 168 2-3 704-710.

He, Y. & Bishop, P.L. (1994), Effect of Acid Orange 7 on nitrification process; Journal
of Environmental Engineering 120 1 108-121.

Healy, M.G. & O'Flynn, C.J. (2011), The performance of constructed wetlands treating
primary, secondary and dairy soiled water in Ireland (a review); Journal of
Environmental Management 92 10 2348-2354.

Hellinga, C., Schellen, A.A.J.C., Mulder, J.W., Van Loosdrecht, M.C.M. & Heijnen, J.J.
(1998), The SHARON process: An innovative method for nitrogen removal from
ammonium-rich waste water, Elsevier Sci Ltd, Kalmir, Sweden.

Hendricks, D.W. (2006), Sedimentation In: Water Treatment Unit Processes: Physical
and Chemical, CRC Press.

Hendrickson, E.R., Ingram, W.T., Churchill, M.A., Bogan, R.W., Hull, C.H., Faber,
H.A., Stone, R. & Nemerow, N.L. (1960), Solubility of Atmospheric Oxygen in
Water, Twenty-ninth Progress Report of the Committee on Sanitary Engineering
Research of the Sanitary Engineering Division; ASCE Proceedings: Journal of the
Sanitary Engineering Division41-53.

HMG (2005), Work at Height Regulations, U.K.

HMG (2003), The Water Environment (Water Framework Directive) Regulations

(Northern Ireland).

HMG (1999), The Water (Northern Ireland) Order 1999, U.K.

HMG (1995), The Urban Waste Water Treatment Regulations (Northern Ireland) 1995,
Northern Ireland.

Hu, S., Yang, F., Sun, C., Zhang, J. & Wang, T. (2008), Simultaneous removal of COD
and nitrogen using a novel carbon-membrane aerated biofilm reactor; Journal of
Environmental Sciences 20 2 142-148.

207
Humpal, G.J. & Chiesa, R. (1990), Considerations for selection, design, and operation of
inclined plate settlers for industrial wastewater treatment; Proceedings of the
Industrial Waste Conference 563.

Hwang, J.H., Cicek, N. & Oleszkiewicz, J.A. (2010), Achieving biofilm control in a
membrane biofilm reactor removing total nitrogen; Water Research 44 7 2283-2291.

Irwin, S. (21st April 2006), Final settler concentrations.

Janelt, G., Bolt, P., Gerbsch, N., Buchholz, R. & Cho, M.-. (1997), The lamellar settler -
a low-cost alternative for separating the micro-alga Chlorella vulgaris from a
cultivation broth? Applied Microbiology and Biotechnology 48 1 6-10.

Kadar, Y. & Siboni, G. (1998), Optimization of energy economy in the design and
operation of wastewater treatment plants.

Khoyi, M.R. & Yaghmaei, S. (2005), Simulation of competition between two

microorganisms in a biofilm reactor based on different growth models; Biochemical
Engineering Journal, 23 1 63-72.

Komorowska-Kaufman, M., Majcherek, H. & Klaczynski, E. (2006), Factors affecting

the biological nitrogen removal from wastewater; Process Biochemistry 41 5 1015-
1021.

Lackner, S., Terada, A. & Smets, B.F. (2008), Heterotrophic activity compromises
autotrophic nitrogen removal in membrane-aerated biofilms: Results of a modeling
study; Water Research 42 4-5 1102-1112.

Lee, K. & Rittmann, B.E. (2002), Applying a novel autohydrogenotrophic hollow-fiber

membrane biofilm reactor for denitrification of drinking water; Water Research 36 8
2040-2052.

Lees, H. (1951), Isolation of the Nitrifying Organisms from Soil; Nature 167 4244 355-
356.

Levine, W.G. (1991), Metabolism of azo dyes: implication for detoxication and
activation. Drug Metabolism Reviews 23 3-4 253-309.

Li, J., Zhu, L., Xu, Y. & Zhu, B. (2010), Oxygen transfer characteristics of hydrophilic
treated polypropylene hollow fiber membranes for bubbleless aeration; Journal of
Membrane Science 362 1-2 47-57.

Li, T., Liu, J., Bai, R. & Wong, F.S. (2008), Membrane-aerated biofilm reactor for the
treatment of acetonitrile wastewater; Environmental Science and Technology 42 6
2099-2104.

208
Lieben, F. (1943), On the hydrolysis of proteins and peptones at high temperatures and
on the catalytic effect of metal ions on the rate of hydrolysis; The Journal of
Biological Chemistry 151 1 117-121.

Lipscomb, D., Larkin, M., Irvine, H. & Allen, C. (2008), BIOCOL process for
remediation of dye house wastewaters.

Livingston, A.G., Arcangeli, J., Boam, A.T., Zhang, S., Marangon, M. & Freitas dos
Santos, L.M. (1998), Extractive membrane bioreactors for detoxification of chemical
industry wastes: Process development; Journal of Membrane Science 151 1 29-44.

Mace, G.R. & Laks, R. (1978), Developments in Gravity Sedimentation; Chemical

Engineering Progress 74 7 77-83.

Madigan, M.T. & Martinko, J.M. (2006), Microbial Ecology In: Brock Biology of
Microorganisms, 11th Edition edn, Pearson Education, Inc., Upper Saddle River,
New Jersey.

Manser, R., Gujer, W. & Siegrist, H. (2006), Decay processes of nitrifying bacteria in
biological wastewater treatment systems; Water Research 40 12 2416-2426.

Masschelein, W.J. (1992), Lamellar and Turbular Assisted Settling Processes In: Unit
Processes in Drinking Water Treatment, CRC Press, Boca Raton, Florida, U.S.A.

Matsumoto, S., Terada, A. & Tsuneda, S. (2007), Modeling of membrane-aerated

biofilm: Effects of C/N ratio, biofilm thickness and surface loading of oxygen on
feasibility of simultaneous nitrification and denitrification; Biochemical Engineering
Journal, 37 1 98-107.

McCarty, P.L. (1975), Stoichiometry of Biological Reactions; Progress in Water

Technology 7 1 157-172.

Meng, Q., Yang, F., Liu, L. & Meng, F. (2008), Effects of COD/N ratio and DO
concentration on simultaneous nitrification and denitrifcation in an airlift internal
circulation membrane bioreactor; Journal of Environmental Sciences 20 8 933-939.

Metcalf & Eddy, I. (2003), Wastewater engineering : treatment and reuse, 4th edn,
McGraw-Hill, Boston, London.

Metso Minerals (2006), , Inclined Plate Settlers. Available online:

http://www.metsominerals.com/inetMinerals/MaTobox7.nsf/DocsByID/909127C82
1746DDE41256B41004D7B71/$File/Inclined_Plate_Settler.pdf [2008, 04/04] .

Monroe, D. (2007), Looking for Chinks in the Armor of Bacterial Biofilms; Journal of the
Public Library of Science: Biology.

209
Morgenroth, E. (2008), Biofilm Reactors In: Biological Wastewater Treatment:
Principles, Modeling, and Design, IWA Publishing, London, U.K.

NI Water (2013), , About Us: Facts and Figures. Available online:

http://www.niwater.com/facts-and-figures/ [2014, April 22nd] .

NIEA (2014), Regulation of Water Utility Sector Discharges 2012, Northern Ireland
Environment Agency, Lisburn, U.K.

NIEA (2009), North Eastern River Basin Management Plan Summary, Northern Ireland
Environment Agency, Lisburn, U.K.

OECD (1984), OECD Guidelines for testing chemicals: 209 "Activated Sludge,
Respiration Inhibition Test".

O'Neill, R., Ahmad, M.N., Vanoye, L. & Aiouache, F. (2009), Kinetics of aqueous phase
dehydration of xylose into furfural catalyzed by ZSM-5 zeolite; Industrial and
Engineering Chemistry Research 48 9 4300-4306.

Ong, S., Toorisaka, E., Hirata, M. & Hano, T. (2005), Decolorization of azo dye (Orange
II) in a sequential UASB-SBR system; Separation and Purification Technology 42 3
297-302.

Pandey, A., Singh, P. & Iyengar, L. (2007), Bacterial decolorization and degradation of
azo dyes; International Biodeterioration and Biodegradation 59 2 73-84.

Pankhania, M., Brindle, K. & Stephenson, T. (1999), Membrane aeration bioreactors for
wastewater treatment: completely mixed and plug-flow operation; Chemical
Engineering Journal 73 2 131-136.

Pankhania, M., Stephenson, T. & Semmens, M.J. (1994), Hollow fibre bioreactor for
wastewater treatment using bubbleless membrane aeration; Water research 28 10
2233-2236.

Parkson (2010), , Lamella EcoFlow Inclined Plate Settler. Available online:

http://www.parkson.com/sites/default/files/documents/document-lamella-ecoflow-
brochure-482.pdf [2014, April 6th] .

Paul, E.L., Atiemo-Obeng, V. & Kresta, S.M. (2004), Handbook of Industrial Mixing:
Science and Practice, Wiley Interscience, Hoboken, NJ, U.S.A.

Pavasant, P., Freitas dos Santos, L.M., Pistikopoulos, E.N. & Livingston, A.G. (1996),
Prediction of optimal biofilm thickness for membrane-attached biofilms growing in
an extractive membrane bioreactor; Biotechnology and Bioengineering 52 3 373-
386.

210
Pearce, C.I., Christie, R., Boothman, C., Von Canstein, H., Guthrie, J.T. & Lloyd, J.R.
(2006), Reactive azo dye reduction by Shewanella strain J18 143; Biotechnology
and Bioengineering 95 4 692-703.

Peng, L., Chen, X., Xu, Y., Liu, Y., Gao, S. & Ni, B. (2015), Biodegradation of
pharmaceuticals in membrane aerated biofilm reactor for autotrophic nitrogen
removal: A model-based evaluation; Journal of Membrane Science 494 39-47.

Picard, C., Logette, S., Schrotter, J.C., Aimar, P. & Remigy, J.C. (2012), Mass transfer in
a membrane aerated biofilm; Water research 46 15 4761-4769.

Ponder, E. (1925), On Sedimentation and Rouleaux Formation; Quarterly Journal of

Experimental Physiology 15 235-252.

Probstein, R.F. & Hicks, R.E. (1978), Lamella Settlers: A New Operating Mode for Hign
Performance; Industrial Water Engineering 15 1 6-8.

Radcliff, R. & Zarnadze, A. (2004), Application of Membrane Technology to the

Production of Drinking Water; Water Conditioning and Purification 46 8 23-25.

Robb, W.L. (1968), Thin silicone membranes - their permeation properties and some
applications; Annals of the New York Academy of Science 146 1 119-137.

Rogers, G.F.C. & Mayhew, Y.R. (1994), Thermodynamic and Transport Properties of
Fluids (S.I. Units).

Rubino, V. (1996), Caught in the crossfire: wastewater plant retrofit; Pollution

Engineering 28 6 72-75.

Saleh, A.M. & Hamoda, M.F. (1999), Upgrading of secondary clarifiers by inclined plate
settlers; Water Science and Technology 40 7 141-149.

Salem, A.I., Okoth, G. & Thöming, J. (2011), An approach to improve the separation of
solid–liquid suspensions in inclined plate settlers: CFD simulation and experimental
validation; Water Research 45 11 3541-3549.

Saratale, R.G., Saratale, G.D., Kalyani, D.C., Chang, J.S. & Govindwar, S.P. (2009),
Enhanced decolorization and biodegradation of textile azo dye Scarlet R by using
developed microbial consortium-GR; Bioresource Technology 100 9 2493-2500.

Sarkar, S., Kamilya, D. & Mal, B.C. (2007), Effect of geometric and process variables on
the performance of inclined plate settlers in treating aquacultural waste; Water
Research 41 5 993-1000.

211
Satoh, H., Ono, H., Rulin, B., Kamo, J., Okabe, S. & Fukushi, K. (2004), Macroscale and
microscale analyses of nitrification and denitrification in biofilms attached on
membrane aerated biofilm reactors; Water Research 38 6 1633-1641.

Schneider, M., Reymond, F., Marison, I.W. & von Stockar, U. (1995), Bubble-free
oxygenation by means of hydrophobic porous membranes; Enzyme and Microbial
Technology 17 9 839.

Schöner, P., Plucinski, P., Nitsch, W. & Daiminger, U. (1998), Mass transfer in the shell
side of cross flow hollow fiber modules; Chemical Engineering Science 53 13 2319-
2326.

Semmens, M.J., Dahm, K., Shanahan, J. & Christianson, A. (2003), COD and nitrogen
removal by biofilms growing on gas permeable membranes; Water Research 37 18
4343-4350.

Shah, D.B. & Coulman, G.A. (1978), Kinetics of Nitrification and Denitrification
Reactions; Biotechnology and Bioengineering 20 1 43-72.

Shanahan, J.W. & Semmens, M.J. (2006), Influence of a nitrifying biofilm on local
oxygen fluxes across a micro-porous flat sheet membrane; Journal of Membrane
Science 277 1-2 65-74.

Shanahan, J.W. & Semmens, M.J. (2004), Multipopulation model of membrane-aerated

biofilms; Environmental Science and Technology 38 11 3176-3183.

Shaw, C.B., Carliell, C.M. & Wheatley, A.D. (2002), Anaerobic/aerobic treatment of
coloured textile effluents using sequencing batch reactors; Water Research 36 8
1993-2001.

Smyth, B. (2011), Pumping Energy Requirements.

Smyth, B., Crilly, A. & McDowell, K. (2013), Water efficiency as a means of reducing
carbon emissions in Northern Ireland (NI) water; Journal of Water Supply:
Research and Technology - AQUA 62 8 525-533.

Sponza, D.T. & Isik, M. (2002), Ultimate azo dye degradation in anaerobic/aerobic
sequential processes, IWA Publishing.

Stephenson, T., Judd, S., Jefferson, B. & Brindle, K. (2000), Membrane Bioreactors for
Wastewater Treatment, First Edition edn, IWA Publishing, London.

Steube, T. (2009), Manual for MOLSEP cartridge FS10 and FE10, Microdyn-Nadir,
Weisbaden, Germany.

Stewart, R. (2014), Nutrient consent in NI as enforced by NIEA.

212
Tan, N.C.G., Borger, A., Slender, P., Svitelskaya, A.V., Lettinga, G. & Field, J.A. (2000),
Degradation of azo dye Mordant Yellow 10 in a sequential anaerobic bioaugmented
aerobic bioreactor; Water Science and Technology 45 337-344.

Tchobanoglous, G. & Burton, F.L. (1991a), Biological Unit Processes In: Wastewater
Engineering: Treatment, Disposal and Reuse, 3rd edn, McGraw-Hill, New York,
U.S.A.

Tchobanoglous, G. & Burton, F.L. (1991b), Metcalf & Eddy's Wastewater Engineering:
Treatment, Disposal, Reuse, 3rd edn, McGraw-Hill, Singapore.

Tchobanoglous, G., Burton, F.L. & Stensel, H.D. (2004), Physical Unit Operations In:
Wastewater Engineering: Treatment and Reuse, Fourth edn, McGraw Hill,
Singapore.

Terada, A., Hibiya, K., Nagai, J., Tsuneda, S. & Hirata, A. (2003), Nitrogen removal
characteristics and biofilm analysis of a membrane-aerated biofilm reactor
applicable to high-strength nitrogenous wastewater treatment; Journal of Bioscience
and Bioengineering 95 2 170-178.

Thumheer, T., Koehler, T., Cook, A.M. & Leisinger, T. (1986), Orthanilic acid and
analogs as carbon sources for bacteria: groth physiology and enzymatic
desulphonation; Journal of General Microbiology 132 1215-1220.

Timberlake, D.L., Strand, S.E. & Williamson, K.J. (1988), Combined aerobic
heterotrophic oxidation, nitrification and denitrification in a permeable-support
biofilm; Water Research 22 12 1513-1517.

Tsushima, I., Ogasawara, Y., Shimokawa, M., Kindaichi, T. & Okabe, S. (2007),
Development of a super high-rate Anammox reactor and in situ analysis of biofilm
structure and function; Water Science and Technology 55 8 9-17.

US EPA (2012), Summaries of Water Pollution Reporting Categories (adapted from

Document no. EPA 841-R-12-104), US EPA.

Van Der Zee, F. P., Bisschops, I.A.E., Blanchard, V.G., Bouwman, R.H.M., Lettinga, G.
& Field, J.A. (2003a), The contribution of biotic and abiotic processes during azo
dye reduction in anaerobic sludge; Water Research 37 13 3098-3109.

Van Der Zee, F. P., Bisschops, I.A.E., Lettinga, G. & Field, J.A. (2003b), Activated
carbon as an electron acceptor and redox mediator during the anaerobic
biotransformation of azo dyes; Environmental Science and Technology 37 2 402-
408.

Van Der Zee, F.P. & Villaverde, S. (2005), Combined anaerobic-aerobic treatment of
azo dyes - A short review of bioreactor studies; Water Research 39 8 1425-1440.

213
Venkatesh, G. & Brattebo, H. (2011), Energy consumption, costs and environmental
impacts for urban water cycle services: Case study of Oslo (Norway); Energy 36 2
792-800.

Vladisavljevic, G.T. (1999), Use of polysulfone hollow fibers for bubbleless membrane
oxygenation/deoxygenation of water; Separation and Purification Technology 17 1
1-10.

Walker, G.M., Hansen, L., Hanna, J.-. & Allen, S.J. (2003), Kinetics of a reactive dye
adsorption onto dolomitic sorbents; Water Research 37 9 2081-2089.

Walter, B., Haase, C. & Rabiger, N. (2005), Combined nitrification/denitrification in a

membrane reactor; Water Research 39 13 2781-2788.

Wang, J., Liu, G., Lu, H., Jin, R., Zhou, J. & Lei, T. (2012), Biodegradation of Acid
Orange 7 and its auto-oxidative decolorization product in membrane-aerated
biofilm reactor; International Biodeterioration & Biodegradation 67 73-77.

Wenk, S.E. (1990), The theory, design and experience of Lamella Gravity Settlers in the
phosphate industry; Fertilizer Research 25 139-143.

Wilkinson, R. (2007), Effluent concentrations at Ulster Carpets Mills (personal

communication - email).

Yamagiwa, K. & Ohkawa, A. (1994), Simultaneous Organic Carbon Removal and

Nitrification by Biofilm formed on Oxygen Enrichment Membrane; Journal of
Chemical Engineering of Japan 27 5 638-643.

Yang, M. & Cussler, E.L. (1986), Designing Hollow Fiber Contactors; AICHE Journal
32 11 1910-1916.

Yasuda, H. & Lamaze, C.E. (1972), Transfer of gas to dissolved oxygen in water via
poroous and nonporous polymer membranes; Journal of Applied Polymer Science
16 595-601.

Zhang, T.C., Fu, Y. & Bishop, P.L. (1995), Competition for substrate and space in
biofilms; Water Environment Research 67 6 992-1003.

Zhang, H. & Cloud, A. (2006), The permeability characteristics of silicone rubber, Soc.
for the Advancement of Material and Process Engineering, Covina, CA 91724-3748,
United States.

Zheng, J., Dai, Z., Wong, F. & Xu, Z. (2005), Shell side mass transfer in a transverse
flow hollow fiber membrane contactor; Journal of Membrane Science 261 1-2 114-
120.

214
Zhu, I.X. (2008), Effect of Oxygen Partial Pressure and COD Loading on Biofilm
Performance in a Membrane Aerated Bioreactor, University of Toronto, Toronto,
Ontario, Canada.

215
Appendices

A1 Results summaries – Mass Transfer Studies

A1.1 Saturation coefficients for use in Equation 3-2

Inlet pressure Silicone Rubber Polyethersulphone

(barg)
0.2 a 16.03 15.24
b1 -0.36 -0.41
b2 1.81x10-3 5.62x10-3
b3 36.1x10-6 -53.7x10-6
0.25 a 16.12 14.56
b1 -0.53 -0.33
b2 12.0x10-3 2.91x10-3
b3 -0.13x10-3 -3.32x10-6
0.4 a 16.33 15.71
b1 -0.46 -0.49
b2 8.77x10-3 10.4x10-3
b3 -81.0 x10-6 -0.12x10-3
0.5 a 16.61 15.03
b1 -0.37 -0.33
b2 2.80 x10-3 1.53 x10-3
b3 17.3 x10-6 24.0 x10-6
0.8 a 17.80 15.08
b1 -0.48 -0.27
b2 8.00 x10-3 -0.25 x10-3
b3 -60.1 x10-6 36.2 x10-6
1.0 a 22.15 15.62
b1 -0.55 -0.29
b2 3.30 x10-3 -1.46 x10-3
b3 5.21 x10-6 99.0 x10-6
2.0 a 27.40
b1 -1.11
b2 33.9x10-3
b3 -0.40 x10-3

216
A1.2 Observed oxygen flux
Numbers in red type correspond to experiments where the regression coefficient in the
𝐶 ∗ −𝐶𝑡
plot of of ln ( ) versus t is poor (<0.98). These experiments are omitted from the
𝐶∗

calculation of the displayed averages.

A1.2.1 Effect of air side flowrate

Silicone rubber
Inlet Flowrate (lpm) 0.6 1 2 3 4 5
J1 1.49 1.38 1.55 1.41 1.53 1.38
gO2m-2h-1

J2 1.50 1.43 1.28 1.47 1.51 1.43

J3 1.45 1.45 1.46 1.52 1.45 1.43
Average 1.48 1.42 1.43 1.47 1.50 1.41

Polyethersulphone
Inlet Flowrate (lpm) 0.6 1 2 3 4 5
J1 0.59 0.58 0.53 0.55 0.56 0.54
0.55 0.51 0.49 0.56 0.55 0.55
gO2m-2h-1

J2
0.50 0.51 0.56 0.55 0.53 0.53
J3
0.53 0.53 0.52
J4
Average 0.56 0.52 0.53 0.55 0.55 0.53

217
Effect of inlet air pressure
Silicone rubber

Pressure (barg) 0.2 0.25 0.4 0.5 0.8 1 2

Flowrate (lpm) 5 4 2.5 2 1.25 1 0.5
1.29 1.25 1.38 1.46 1.61 1.84 2.43
J1
1.18 1.23 1.41 1.49 1.64 1.71 2.42
gO2m-2h-1

J2
1.25 1.22 1.53 1.46 1.57 1.69 2.38
J3
J4 1.40
average 1.24 1.23 1.44 1.47 1.61 1.75 2.41

Polyethersulphone
Pressure (barg) 0.2 0.25 0.4 0.5 0.8 1
Flowrate (lpm) 5 4 2.5 2 1.25 1
0.71 0.73 0.66 0.75 0.71 0.75
J1
0.71 0.74 0.74 0.75 0.78 0.79
J2
gO2m-2h-1

0.67 0.74 0.78 0.78 0.76 0.79

J3
J4
J5
average 0.70 0.74 0.72 0.76 0.75 0.78

A1.2.2 Effect of water side flowrate

Silicone rubber
Flowrate (ml min-1) 157.5 210.0 262.5 315.0 367.5 472.5 555.0
0.97 0.97 1.06 1.14 1.19 1.06 1.13
J1
0.90 0.95 1.06 1.10 1.15 1.14 1.25
J2
gO2m-2h-1

0.94 0.80 1.05 1.11 1.11 1.14 1.31

J3
J4 0.92 1.10 1.31
J5 1.14 1.26 1.27
average 0.94 0.95 1.06 1.12 1.12 1.15 1.25

218
Polyethersulphone
Flowrate (ml min-1) 157.5 210.0 262.5 315.0 367.5 420.0 472.5
0.50 0.50 0.51 0.49 0.57 0.61 0.56
J1
0.42 0.48 0.39 0.56 0.61 0.50 0.51
J2
gO2m-2h-1

0.44 0.40 0.52 0.45 0.44 0.42 0.46

J3
J4 0.40 0.43 0.46 0.47 0.49 0.57 0.47
J5 0.45 0.44 0.47 0.50 0.47
average 0.44 0.45 0.46 0.49 0.52 0.52 0.49

219
A2 Biofilm Reactor Studies MABfR A - Results summaries

A2.1 Inlet pressure studies (Chapter 5)

A2.1.1 COD

Influent
Inlet Pressure
1.0 1.1 1.2 1.3 1.4 1.6 2.0
(barg)
Influent COD (mg l-1)
Observation
1 72 69 68 72 75 76 71
2 68 74 70 70 73 70 72
3 70 76 76 72 72 76 72
4 64 70 73 72 72 73 73
5 78 71 68 71 72 72 72
Average 70 72 71 71 73 73 72
Standard Error 2.3 1.3 1.5 0.4 0.6 1.2 0.3

Effluent
Inlet Pressure
1.0 1.1 1.2 1.3 1.4 1.6 2.0
(barg)
Effluent COD (mg l-1)
Observation
1 40 30 41 27 31 18 14
2 28 28 42 31 39 24 17
3 24 28 27 28 27 26 18
4 21 28 24 30 27 22 14
5 34 23 25 31 24 29 16
Average 29 27 32 29 30 24 16
Standard Error 3.43 1.17 3.99 0.81 2.60 1.85 0.80

220
A2.1.2 Ammoniacal Nitrogen

Influent
Inlet Pressure
1.0 1.1 1.2 1.3 1.4 1.6 2.0
(barg)
Influent Ammonical Nitrogen (mg l-1)
Observation
1 9.87 9.35 9.56 9.52 9.77 9.62 9.84
2 9.62 9.27 9.07 9.51 9.88 9.72 9.78
3 9.77 9.14 9.45 9.48 9.76 9.6 9.6
4 9.56 9.35 9.46 9.65 9.9 9.77 9.21
5 9.77 9.45 9.45 9.62 9.77 9.6 9.03
Average 9.72 9.31 9.40 9.56 9.82 8.07 9.49
Standard Error 0.06 0.05 0.08 0.03 0.03 1.75 0.16

Effluent
Inlet Pressure
1.0 1.1 1.2 1.3 1.4 1.6 2.0
(barg)

Observation Effluent Ammonical Nitrogen (mg l-1)

1 6.85 3.11 3.06 3.53 3.57 3.97 2.54
2 4.49 3.21 2.91 3.51 3.81 4.42 2.25
3 3.92 3.35 2.31 2.92 3.85 4 2.04
4 5.82 3.52 3.21 3.31 4.43 3.77 2.29
5 5.07 3.01 5.43 2.84 4.42 3.83 2.04
Average 5.23 3.24 3.38 3.22 4.02 4.00 2.23
Standard Error 0.51 0.09 0.53 0.15 0.17 0.11 0.09

221
A2.1.3 Nitrate Nitrogen
Influent

Inlet Pressure
1.0 1.1 1.2 1.3 1.4 1.6 2.0
(barg)

Observation Influent Nitrate Nitrogen (mg l-1)

1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1
2 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1
3 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1
4 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1
5 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1

Effluent

Inlet Pressure
1.0 1.1 1.2 1.3 1.4 1.6 2.0
(barg)
Effluent Nitrate Nitrogen (mg l-1)
Observation
1 3.6 3.6 6.0 5.1 3.7 3.2 4.0
2 5.0 4.1 5.4 4.9 4.5 3.7 4.6
3 5.2 3.6 4.4 5.3 3.6 3.6 4.5
4 2.4 4.8 5.1 5.0 3.4 3.9 4.6
5 4.4 4.5 5.4 5.2 3.6 3.5 4.5
Average 4.1 4.1 5.3 5.1 3.8 3.6 4.4
Standard Error 0.5 0.2 0.3 0.1 0.2 0.1 0.1

222
A2.2 Pollutant loading studies (Chapter 6)
Run 8
Amm- Amm- COD COD Nit-N Nit-N
N inf N eff inf eff inf eff
-1 -1 -1 -1 -1
Observation mg l mg l mg l mg l mg l mg l-1
1 11.17 6.95 78 57 <0.1 1.5
2 12.3 7.08 60 21 <0.1 1.0
3 9.12 6.14 62 27 <0.1 1.5
4 11.99 6.28 63 18 <0.1 2.9
5 11.04 5.79 81 53 <0.1 1.5
6 12.33 5.57 61 19 <0.1 1.1
7 11.19 7.01 68 1 <0.1 3.8
8 12.24 6.55 76 11 <0.1 2.0
9 13.03 6.31 76 35 <0.1 1.5
10 13.03 6.34 76 38 <0.1 1.5
11 10.86 5.93 53 4 <0.1 1.5
12 10.86 5.55 53 14 <0.1 2.0
13 10.66 5.36 65 21 <0.1 0.9
14 8.13 5.52 54 20 <0.1 0.9
15 9.77 5.24 54 15 <0.1 1.6
16 10.74 4.86 60 30 <0.1 0.7
17 13.12 6.41 74 1 <0.1 2.4
18 12.56 7.8 65 7 <0.1 0.4
19 12.39 6.13 52 36 <0.1 0.7
20 12.16 8.22 80 38 <0.1 0.6
21 12.16 7.93 52 32 <0.1 3.1
22 12.16 8.21 63 52 <0.1 1.5
23 12.3 6.07 63 48 <0.1 0.8
24 13.12 7.62 58 42 <0.1 0.7
25 12.3 6.62 71 31 <0.1 1.7
26 11.83 6.28 72 32 <0.1 0.6
27 12.27 6.3 72 27 <0.1 0.7
28 12.27 6.18 71 34 <0.1 1.6
29 12.68 6.15 72 33 <0.1 0.5

223
Run 9
Amm- Amm- COD COD Nit-N Nit-N
N inf N eff inf eff inf eff
Observation mg l-1 mg l-1 mg l-1 mg l-1 mg l-1 mg l-1
1 10.3 6.25 67 12 5.5 6.8
2 10.3 5.66 58 10 5.8 8.4
3 10.62 6.22 70 46 6.0 8.1
4 9.91 5.45 71 20 4.9 8.6
5 9.51 4.92 81 44 6.0 7.9
6 9.38 5.58 66 35 6.0 5.6
7 10.59 4.99 77 37 5.4 5.0
8 10.16 5.4 76 39 4.5 5.4

Run 10

Amm- Amm- COD COD Nit-N Nit-N

N inf N eff inf eff inf eff
-1 -1 -1 -1 -1
Observation mg l mg l mg l mg l mg l mg l-1
1 7.26 3.7 63 23 5.4 7.5
2 7.18 3.8 75 50 5.0 6.4
3 7.14 3.42 73 39 4.9 7.2
4 7.08 3.52 63 30 4.6 7.9
5 7.11 3.7 75 50 6.5 4.8
6 7.2 3.52 77 52 0.0 7.1

224
Run 11

Amm- Amm- COD COD Nit-N Nit-N

N inf N eff inf eff inf eff
-1 -1 -1 -1 -1
Observation mg l mg l mg l mg l mg l mg l-1
1 8.4 3.14 122 35 4.8 7.8
2 8.33 3.49 119 55 5 7.6
3 8.18 3.96 113 62 5.1 7.4
4 8.59 4.39 113 58 5.3 6.8
5 8.81 4.49 127 77 5 8
6 7.78 3.81 117 65 5.9 7.5
7 7.48 2.98 108 40 5.1 10.2
8 8.49 3.22 115 52 5.4 9.8
9 6.62 3.45 116 77 5.3 8.6
10 7.45 3.71 115 60 4.8 8
11 8.38 4.37 117 17 5.2 7
12 9.08 3.94 126 61 4.6 5.9
13 7.82 3.56 106 83 5.8 7
14 7.65 3.61 125 62 5.9 6.9
15 8.05 3.41 120 77 5.4 6.7

225
Run 12

Amm- Amm- COD COD Nit-N Nit-N

N inf N eff inf eff inf eff
-1 -1 -1 -1 -1
Observation mg l mg l mg l mg l mg l mg l-1
1 7.13 3.8 62 43 4.5 7.3
2 5.79 1.858 81 29 5.2 6.2
3 6.92 5.09 78 38 6.3 6.7
4 7.71 4.07 69 37 6.6 10.1
5 7.67 2.92 66 10 5.9 10.5
6 8.1 0.1059 67 14 5.1 5.1
7 7.96 0.1091 62 8 5.1 4.9
8 9.46 2.42 66 9 5.3 5.3
9 8.25 3.87 67 37 4.1 12.6
10 6.89 1.742 77 65 4.8 0.4
11 5.02 1.986 67 6 6.1 12.4
12 8.57 1.986 74 43 6.1 10.0
13 8.28 4.16 66 6 6.1 8.0
14 8.38 4.8 80 48 4.9 5.4
15 9.28 4.38 66 32 6 6.9
16 9.44 1.094 70 49 5.8 4.2
17 9.86 0.292 62 21 6.6 5.7
18 8.95 6.37 66 15 5.4 8.7
19 8.88 2.51 65 17 5 4.9

226
A3 Pilot plant studies

A3.1 Tank dimensions

Plate packs
number of plate packs N 2
number of plates in
pack np 11
plate spacing ωp 0.11 m
plate thickness pt 0.01 m
plate angle α 55 ° 0.960 radians 1.570796
length of plates lp 1 m
breath of plates bp 0.5 m

area of plates Ap 0.5 m2

length of plate packs lpp 1.10 m

total settling area AT 5.74 m2

total settling volume VST 0.818 m3

total plate pack volume Vpp 0.90 m3

specific settling area

(volume) 3.97 m2m-3

specific settling area

(footprint) 2.86 m2m-2

Sludge collection zone

sludge collection
angle θ 20 ° 0.35 radians

height of sludge zone hsz 0.40 m

227
 volume of sludge
zone Vsz 0.26 m3

Inlet zone

feed channel
breadth bf 0.2 m

inlet zone breadth biz 1.2 m

volume of inlet zone Viz 0.28 m3

Tank dimensions

height of platepack hpp 0.82 m

length of platepack lpp 1.10 m

height of
sludgezone hsz 0.40 m

total tank volume VT 1.45 m3

footprint Af 2.01 m2

minimum height of note: no allowance for

tank hmin 1.22 m feet, over

minimum breadth
of tank bmin 1.2 m

228
A3.2 CAD diagrams

229
230
A3.3 Pilot Plant Studies – sample analysis

Raw data - plant effluent Raw data - BioSettler effluent

Amm Amm
flowrate COD BOD -N Nit-N SS pH Tot-N COD BOD -N Nit-N SS pH Tot-N
mg l-1
Day l day-1 mg l-1 mg l-1 mg l-1 mg l-1 mg l-1 mg l-1 mg l-1 mg l-1 mg l-1 mg l-1 mg l-1
1 4077
2 4077
3 4077
4 4077
5 4077
6 4077 563 55.7 0.9 70.9 7.47 56.6 483 45.5 0.7 92.5 7.56 46.2
7 4077 443 101 48 0.4 79 7.57 48.4 463 99 46.2 1 78 7.6 47.2
8 4077 261 40.7 0.6 54 7.7 41.3 187 30.7 1.3 38 7.72 32.0
9 4077 358 44 0.6 68 7.65 44.6 285 31.7 0.5 44 7.61 32.2
10 4077
11 4077
12 4077
13 4077 205 22 0.3 41 7.38 22.3 182 22.2 1.1 38 7.19 23.3
14 4077 236 51.2 22.8 0.3 39 7.34 23.1 189 37.4 27.5 0.2 31 7.5 27.7
15 4077 122 25.5 2.4 33 7.68 27.9 111 24.4 0.8 23 7.66 25.2
16 4077 148 22.2 1.3 27 7.36 23.5 129 26.8 1.1 18 7.7 27.9
17 4077 316 51.2 0.9 51 7.64 52.1 154 28.4 0.9 38 7.69 29.3
18 4077 537 58.5 0.7 94 7.71 59.2 283 37.9 1.3 62 7.69 39.2

231
19 4077 347 44.3 0.1 88 44.4 325 35.6 6.8 88 42.4
20 4077 278 83 33.6 0.1 89 33.7 128 49 18.11 6.8 62 24.9
21 4077 173 27.9 0.9 25 7.57 28.8 129 26.4 1.1 30 7.48 27.5
22 4077 197 32.6 1.2 49 7.54 33.8 142 27.5 1.1 29 7.53 28.6
23 4077 248 30.3 0.9 57 7.49 31.2 170 33.7 0.8 37 7.59 34.5
24 4077 264 30.7 0.9 58 7.42 31.6 251 31.8 0.6 61 7.49 32.4
25 4077 324 33 1 84 7.32 34.0 290 31.4 0.8 61 7.48 32.2
26 4077 309 43 0.7 66 7.36 43.7 235 41.8 0.5 65 7.54 42.3
27 4077 156 34.5 0.7 59 7.56 35.2 173 30.4 0.6 59 7.44 31.0
28 4077 132 51 22.8 2.4 43 7.21 25.2 158 62 29 0.7 53 7.59 29.7
29 4077 188 30 0.4 36 7.73 30.4 208 28.8 0.3 49 7.71 29.1
30 4077 222 31.3 1.3 23 7.77 32.6 264 29.2 0.2 39 7.74 29.4
31 4077 163 23.7 0.7 29 7.59 24.4 204 29.6 0.3 38 7.74 29.9
32 4077 178 23.8 2 27 7.65 25.8 214 25.8 0.5 42 7.78 26.3
33 4077 160 34.8 2.1 12 7.54 36.9 164 35.6 0.6 44 7.58 36.2
34 4077 172 34 4 14 7.56 38.0 148 34.8 0.4 52 7.64 35.2
35 4077 171 31.9 9.1 6 7.69 41.0 150 37.4 1.1 9 7.68 38.5
36 4077 158 33.6 1.3 37 7.82 34.9 297 23.6 1.2 391 7.41 24.8
37 4077 234 27.9 1.7 19 7.82 29.6 196 26.7 0.9 104 7.67 27.6
38 4077 309 36.3 1.1 58 7.69 37.4 154 28.1 0.6 38 7.71 28.7
39 4077 244 29.9 1.1 69 7.78 31.0 207 40.2 0.5 36 7.59 40.7
40 4077 103 10.42 2.3 24 7.18 12.7 105 11.3 0.9 36 7.64 12.2
41 4077 234 14.22 11.9 17 7.14 26.1 127 25.4 1.4 38 7.46 26.8
42 4077 116 4.1 19.19 14.1 27 7.3 33.3 79 10.8 18.2 2.6 20 7.32 20.8

232
A4 Biofilm Reactor Studies MABfR B - Results summaries
Run 1
Amm-N inf Amm-N eff COD inf COD eff
Observation mg l-1 mg l-1 mg l-1 mg l-1
1 116.9 35 990 248
2 116.9 55.6 960 194
3 116.9 63 975 195
4 116.9 64.6 955 148
5 116.9 66.5 938 211
6 116.9 63.5 925 170
7 122.7 92.9 977 286
8 122.7 95.9 958 355
9 122.7 102.1 948 289
10 122.7 113.2 972 189
11 135.8 106.4 965 178
12 133.7 118.3 965 218
13 133.7 111.9 965 499
14 133.7 137.4 883 295
15 131.8 137.7 835 312
16 109.9 135.5 590 315
17 109.9 141.3 - 249
18 132.4 171 1065 457
19 132.1 162.6 1010 766
20 121.3 171 950 335
21 116.9 130 1009 369
22 114.8 126.5 987 377
23 121.3 141.6 940 557
24 121.6 143.5 948 250
25 119.9 141.3 936 471
26 121.1 130.6 - -
27 108.6 121.6 899 518
28 113.5 128.5 886 239
29 113.5 130 - -
30 113.5 126.5 - -
31 110.9 140 634 210
32 110.9 148.6 850 258
33 102.3 153.8 750 217
34 98.6 144.9 704 289
35 120 151.6 1113 493

233
Run 2
Amm-N inf Amm-N eff COD inf COD eff
Observation mg l-1 mg l-1 mg l-1 mg l-1
1 145.2 181.6 1025 420
2 159.2 178.2 992 166
3 - - 1074 136
4 132.7 176.2 850 138
5 137.7 184.5 847 118
6 140.9 184.5 888 139
7 146.9 181.1 862 170
8 111.2 139.2 1013 134
9 111.7 150.3 1002 148
10 112.7 145.5 1005 149
11 113.9 152.1 934 205
12 131.2 167.32 843 780

Run 3
Amm-N inf Amm-N eff COD inf COD eff
Observation mg l-1 mg l-1 mg l-1 mg l-1
1 89.5 91.4 759 287
2 91.2 89.3 719 185
3 73.8 76 750 181
4 79.6 101.6 650 119
5 70.3 99.8 708 120
6 87.5 101.4 690 102
7 82.4 93.3 710 110
8 75 89.1 722 102
9 72.4 88.1 721 107
10 73.5 83.4 740 106
11 73.3 91.34 795 159
12 72.8 101.6 778 195
13 72.8 99.8 540 152
14 89.3 101.4 376 163
15 73.5 93.3 641 286
16 74.3 89.1 689 307
17 75.3 88.1 658 197

234
A5 Dissemination

Conferences

Murray, S., Allen, S.J. & Groom, E. (2008), Use of Membrane Aerated Biofilm Reactors
for Upgrading of Municipal Wastewater Treatment Works (Presentation), IWA,
Nanyang Technological University, Singapore.

Murray, S., Moroney, N., Allen, S.J. & Groom, R.E. (2008), Treatment of Azo Dye Waste
Using a Membrane Aerated Biofilm Reactor (Poster Presentation), Canadian
Society for Chemical Engineering, Ottawa, Canada.

Murray, S. & Groom, E. (2010), The BioSettlerTM: pilot-scale demonstration of a novel

wastewater treatment technology (Presentation and Conference Paper), Aqua
Enviro, Royal Armouries, Leeds, U.K.

Patent

Groom, E., Murray, S. & Ferguson, J. The Queen's University of Belfast, U.K. (2011),
Improvements relating to water treatment; US 2011000805, EP2222608,
CN101896434

235