Up-concentration techniques for zero-waste water ...

1 downloads 0 Views 26MB Size Report
Willy Verstraete. (LabMET) and prof. Ludo Diels (VITO), for giving me the opportunity to make this thesis and for their guidance and encouragement. I would also ...
Faculty of Bioscience Engineering

Up-concentration techniques for zero-waste water treatment by Dennis Cardoen

Promotors: Prof. Dr. Ir. W. Verstraete, Prof. Dr. L. Diels Tutors: Ir. B. Bundervoet, MSc. B. Eggermont

Masters dissertation submitted in partial fulfillment of the requirements for the degree of Master of environmental sanitation Academic year 2010-2011

1

Copyright The author and promoters authorize consultation and partial reproduction of this thesis for personal use only. Any other reproduction or use is subject to copyright protection. Citation should clearly mention reference of this work. Gent, August 2011

The promoters, Prof. dr. ir. Willy Verstraete

Prof. dr. Ludo Diels

The author, Dennis Cardoen

Acknowledgements The experimental work reported in this thesis was carried out at the Laboratory of Microbial Ecology and Technology (LabMET) of Ghent University. I would like to express my gratitude to my promotors, prof. Willy Verstraete (LabMET) and prof. Ludo Diels (VITO), for giving me the opportunity to make this thesis and for their guidance and encouragement. I would also like to thank prof. Verstraete for his inspiring teaching. I am very grateful and deeply indebted to my tutors Bert Bundervoet and Bram Eggermont for their patient and kind guidance and support throughout the 6 months spent at LabMET. I would like to thank the staff LabMET and at the Centre for Environmental Sanitation (CES) for their kind help, and in particular Rita for showing me the ropes in the lab. Thanks also to Leonie Hartog and Harry Vriendt at Waterschap Brabantse Delta for introducing me to the Breda WWTP and providing sludge samples, to Robben and Mariane van Wambeke at Avecom for their help with using the FMX, and to Maryam Hakimhashemi at the Particle and Interface Technology group for letting me use the RO setup. I sincerely thank my family for their love and support. Finally, I want to thank Poonam, for supporting, encouraging and inspiring me.

4

Abstract Recovery of resources from wastewater is a step waiting to be taken on the road towards greater sustainability. Not only the water content of wastewater streams is to be recovered, but attention must also be paid to the recovery of the energy and nutrients present in it. In this context, the crucial technology for energy recovery from liquid waste streams is anaerobic digestion. The main problem with energy recovery from municipal wastewaters by means of anaerobic digestion, though, is its dilute nature, which makes the process inefficient. There is therefore a real advantage to be gained from the development of separation processes which can upconcentrate municipal wastewater in a thorough but also, and crucially, energyefficient manner. Such an upconcentration would then allow for economical and sustainable energy recovery from the produced concentrate through anaerobic digestion and for reuse of the clarified water fraction, either directly or after another treatment step such as reverse osmosis. This thesis deals with the search for an effective upconcentration technique. The techniques which were investigated are chemically enhanced primary treatment, centrifugation, primary membrane filtration and so-called bio-adsorption or bioflocculation in a highly loaded membrane bioreactor. In addition, the efficiency of the anaerobic digestion of concentrates produced by upconcentration techniques was looked into, in order to be able to get an insight into the energy efficiency of the combination of an upconcentration technique and anaerobic digestion. Digestion tests were performed upon concentrates obtained by means of bio-adsorption and centrifugation. Three different digester types were tested in order to investigate which can achieve the best possible energy recovery: completely mixed tank reactors, anaerobic sequencing batch reactors and a temperature phased reactor. 5

It was found that the most effective up-concentration techniques were chemically enhanced primary treatment and bio-adsorption in a highly loaded MBR, but the goal of achieving an energy positive treatment train could not be met. The anaerobic sequencing batch reactor was found to be a suitable reactor for digesting the concentrate produced by bio-adsorption processes, allowing for a 50% reduction in reactor volume compared to completely mixed reactors. Temperature phased digestion did not add any significant benefits.

Contents Preface

4

Abstract

5

Table of contents

7

List of abbreviations

11

List of figures

14

List of tables

18

1 Literature review 1.1 The twin problems of water scarcity and wastewater management . 1.2 Resources present in municipal wastewater . . . . . . . . . . . . . . 1.3 The current municipal wastewater treatment paradigm . . . . . . . 1.4 The zero-waste water treatment concept . . . . . . . . . . . . . . . 1.5 Energy recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5.1 Anaerobic digestion . . . . . . . . . . . . . . . . . . . . . . . 1.6 Water recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6.1 Membrane filtration . . . . . . . . . . . . . . . . . . . . . . 1.7 Nutrient recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.7.1 Nitrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.7.2 Phosphorus . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.8 Approaches towards maximal recovery of resources . . . . . . . . . 1.8.1 Source separation of municipal wastewater . . . . . . . . . . 1.8.2 Municipal wastewater as a matrix for dilution of organic wastes 1.8.3 Up-concentration of municipal wastewater at the WWTP level

19 19 22 26 28 31 31 38 38 39 39 41 41 42 42 43

7

1.9

Up-concentration techniques . . . . . . . . . . . 1.9.1 Chemically enhanced primary treatment 1.9.2 Centrifugation . . . . . . . . . . . . . . . 1.9.3 Primary membrane filtration . . . . . . . 1.9.4 Bio-adsorption and -flocculation . . . . . 1.10 Research question . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

2 Materials and methods 2.1 Analytical techniques . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1 Total solids, total suspended solids, volatilizable solids and volatilizable suspended solids . . . . . . . . . . . . . . . . 2.1.2 Chemical oxygen demand . . . . . . . . . . . . . . . . . . 2.1.3 Total ammonium/ammonia nitrogen . . . . . . . . . . . . 2.1.4 Kjeldahl nitrogen . . . . . . . . . . . . . . . . . . . . . . . 2.1.5 Total and soluble phosphorus . . . . . . . . . . . . . . . . 2.1.6 Volatile fatty acids . . . . . . . . . . . . . . . . . . . . . . 2.1.7 pH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.8 Alkalinity and Ripley index . . . . . . . . . . . . . . . . . 2.1.9 Turbidity . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.10 Nitrate, nitrite and orthophosphate . . . . . . . . . . . . . 2.1.11 Sludge volume index . . . . . . . . . . . . . . . . . . . . . 2.1.12 Methane concentration . . . . . . . . . . . . . . . . . . . . 2.1.13 Capillary suction time and Kozeny constant . . . . . . . . 2.2 Experimental setups . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Wastewater . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2 Chemically enhanced primary treatment . . . . . . . . . . 2.2.3 Centrifugation . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.4 FMX ultrafiltration . . . . . . . . . . . . . . . . . . . . . . 2.2.5 Conventional biological adsorption/flocculation . . . . . . 2.2.6 Biological adsorption/flocculation in membrane bioreactor 2.2.7 Reverse Osmosis . . . . . . . . . . . . . . . . . . . . . . . 2.2.8 Anaerobic digestion batch tests . . . . . . . . . . . . . . .

. . . . . .

45 45 48 49 50 56

57 . 57 . . . . . . . . . . . . . . . . . . . . . .

57 58 58 59 59 59 60 60 60 60 60 61 61 61 61 61 63 65 67 67 69 69

3 Results 72 3.1 Chemically enhanced primary treatment . . . . . . . . . . . . . . . 72 3.1.1 flocculant selection . . . . . . . . . . . . . . . . . . . . . . . 72

3.2 3.3

3.4 3.5

3.6

3.1.2 Coagulant dose selection . . . . . . . . . . . 3.1.3 Upconcentration test . . . . . . . . . . . . . Centrifugation . . . . . . . . . . . . . . . . . . . . . Primary membrane filtration . . . . . . . . . . . . . 3.3.1 Clean water flux . . . . . . . . . . . . . . . 3.3.2 Pressure selection test . . . . . . . . . . . . 3.3.3 Upconcentration test . . . . . . . . . . . . . 3.3.4 Intermittent filtration . . . . . . . . . . . . . Bio-adsorption and sedimentation . . . . . . . . . . Bio-adsorption in a membrane bioreactor . . . . . . 3.5.1 Loading rate and biomass concentration . . 3.5.2 Flux and fouling remediation . . . . . . . . 3.5.3 COD removal . . . . . . . . . . . . . . . . . 3.5.4 Nitrogen removal . . . . . . . . . . . . . . . 3.5.5 Phosphorus removal . . . . . . . . . . . . . 3.5.6 Recovery . . . . . . . . . . . . . . . . . . . . 3.5.7 Reverse osmosis treatment of MBR efffluent Anaerobic digestion . . . . . . . . . . . . . . . . . . 3.6.1 Adsorptive sludge . . . . . . . . . . . . . . . 3.6.2 Centrifuged sludge . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

4 Discussion 4.1 The benchmark: conventional activated sludge treatment 4.1.1 Energy recovery in CAS . . . . . . . . . . . . . . 4.1.2 Costs and water recovery from CAS effluent . . . 4.2 Chemically enhanced primary treatment . . . . . . . . . 4.2.1 Energy recovery in CEPT . . . . . . . . . . . . . 4.2.2 Water recovery from CEPT eflluent . . . . . . . . 4.2.3 Costs of CEPT . . . . . . . . . . . . . . . . . . . 4.3 Centrifugation . . . . . . . . . . . . . . . . . . . . . . . . 4.3.1 Energy efficiency . . . . . . . . . . . . . . . . . . 4.3.2 Costs . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Primary membrane filtration with FMX . . . . . . . . . 4.4.1 Energy efficiency . . . . . . . . . . . . . . . . . . 4.4.2 Costs . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

74 76 79 81 81 82 83 87 88 89 89 90 91 92 94 95 96 99 99 105

. . . . . . . . . . . . .

109 . 109 . 109 . 116 . 116 . 116 . 120 . 120 . 120 . 120 . 121 . 121 . 121 . 124

4.5

4.6

Bio-adsorption in an MBR . . . . . . . . . 4.5.1 SRT = 1.2 d . . . . . . . . . . . . . 4.5.2 SRT = 2.8 d . . . . . . . . . . . . . Digester configuration . . . . . . . . . . . 4.6.1 Anaerobic sequencing batch reactor 4.6.2 Temperature phased digestion . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

124 124 130 132 132 132

5 Conclusions 134 5.1 Recommendations for future research . . . . . . . . . . . . . . . . . 134

Bibliography

135

List of abbreviations A/B

Adsorption/bio-aeration (Boehnke) process

AD

Anaerobic digestion

AnMBR

Anaerobic membrane bioreactor

AnSBR

Anaerobic sequencing batch reactor

BDL

Below detection limit

CAR

Contact adsorption regeneration process

CDW

Cell dry weight

CWF

Clean water flux

CSTR

Continuous stirred tank reactor

bCOD

Biodegradable chemical oxygen demand

BOD

Biological oxygen demand

CAS

Conventional activated sludge process

CAR

Contact adsorption regeneration process

CDW

Cell dry weight

CEPT

Chemically enhanced primary treatment

CHP

Combined heat and power generation a.k.a. cogeneration

CODp

Particulate chemical oxygen demand

CODs

Dissolved chemical oxygen demand

CODt

Total chemical oxygen demand

EPS

Extracellular polymeric substances

GHG

Greenhouse gas

HRT

Hydraulic residence time 11

IE

Inhabitant equivalents

K

Kozeny constant

Kj-N

Kjeldahl nitrogen

MBR

Membrane bioreactor

MLVSS

Mixed liquor volatilizable suspended solids

Nt

Total nitrogen

NSF

Natural stable fertilizer

NTU

Nephelometric turbidity units

OLR

Volumetric organic loading rate

PE

Poly-electrolyte

Ps

Dissolved phosphorus

Pss

Phosphorus associated to suspended solids

Pt

Total phosphorus

RE

Removal efficiency

RO

Reverse osmosis

SE

Separation efficiency

SRT

Sludge residence time

SVI

Sludge volume index

TAN

Total ammoniacal nitrogen

TMP

Transmembrane pressure

TS

Total solids

TSS

Total suspended solids

UF

Ultrafiltration

UASB

Upflow anaerobic sludge blanket

VFA

Volatile fatty acids

VS

Volatilizable solids

VSS

Volatilizable suspended solids

WWTP

Wastewater treatment plant

TPAD

Temperature phased anaerobic digestion 12

13

List of Figures 1.1 1.2 1.3 1.4 1.5 1.6

Number of people expected to be living in water-stressed areas in 2030 by country type. . . . . . . . . . . . . . . . . . . . . . . . . . The redesigned water cycle: short-cycling of resources to achieve recovery (Verstraete & Vlaeminckx (2011)). . . . . . . . . . . . . . Simplified outline of the conversion steps in AD. . . . . . . . . . . . Integration of OLAND for nitrogen removal in WWTPs. . . . . . . Process scheme for maximal recovery of resources from municipal wastewater. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Some options for recovering high quality water from municipal wastewater using membrane technology. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

45 49

CEPT setup: mixers. . . . . . . . . . . . . . . . . . . CEPT setup: sedimentation. . . . . . . . . . . . . . . Westfalia RTC centrifuge unit. . . . . . . . . . . . . . Conventional cross-flow filtration and FMX filtration. FMX filtration unit. . . . . . . . . . . . . . . . . . . Half open circuit for up-concentration tests. . . . . . Conventional activated sludge setup. . . . . . . . . . Membrane bioreactor setup. . . . . . . . . . . . . . . AD batch reactors. . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

62 63 64 65 66 66 67 68 70

3.1

Selection of coagulant dose prior to flocculant addition. Setted wastewater samples after coagulant addition. . . . . . . . . . . . . Flocculant selection. . . . . . . . . . . . . . . . . . . . . . . . . . Flocculant selection. COD of supernatant. . . . . . . . . . . . . . Effect of coagulant dose on the turbidity of settled wastewater. . .

. . . .

73 73 73 74

14

. . . . . . . . .

29 32 40

2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9

3.2 3.3 3.4

. . . . . . . . .

20

3.5 3.6 3.7 3.8

3.9

3.10

3.11

3.12 3.13

3.14 3.15 3.16 3.17 3.18 3.19

Effect of coagulant dose on the COD of settled wastewater. Flocculant dose was 1 mg PE/L. . . . . . . . . . . . . . . . . . . . . . . Effect of coagulant dose on the PO4 3 − concentration of settled wastewater. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Evolution of supernatant (top layer) turbidity with settling time for primary settling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . The recovery of clean water volume in the effluent and of COD, Kj-N and Pt in the concentrate after primary sedimentation, resp. CEPT treatment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . CODt removal efficiency as a function of the particulate fraction of the COD, achieved by centrifugation at 1000 L/h of municipal wastewaters of varying quality. . . . . . . . . . . . . . . . . . . . . CODt removal efficiency as a function of the influent CODt, achieved by centrifugation at 1000 L/h of municipal wastewaters of varying quality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CODt, CODp, Kj-N and Pt removal efficiency achieved by centrifugation at 1000 L/h of municipal wastewater, with and without the use of FeCl3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Clean water flux as a function of trans-membrane pressure, with and without vortex generation. . . . . . . . . . . . . . . . . . . . . Wastewater filtration permeate flux as a function of time during half-open circuit operation with vortex generation, for membrane cut-offs 0.5 and 1 µm. . . . . . . . . . . . . . . . . . . . . . . . . . Fraction of resources recovered by FMX filtration as a function of the amount of used energy per volume of influent. . . . . . . . . . . Fraction of resources recovered by FMX filtration as a function of the amount of used energy per volume of influent. . . . . . . . . . . Recovery of resources by FMX upconcentration at a consumed energy of 31 and 46 kWh/m3 for cut-off = 1 µm, resp. 0.5 µm. . . . . Flux evolution over time during intermittent vortex generation. . . CODt and CODs removal by bio-adsorption reactor with sedimentation. SRT = 1 d. . . . . . . . . . . . . . . . . . . . . . . . . . . . Evolution of applied SRT and subsequent HRT and permeate flux in the MBR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15

75 76 77

78

80

80

81 82

84 85 86 87 88 89 90

3.20 VSS in mixed liquor and concentrate and volumetric loading rate (Bv ) and sludge loading rate (Bx ) of bio-adsorption MBR operated on municipal wastewater at SRT ≈ 1.2 and 2.8 d. . . . . . . . . . . 3.21 Effect of relaxation and membrane washing on the permeate flux. (SRT = 2.8d.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.22 COD removal by bio-adsorption MBR operated at SRT ≈ 1.2 and 2.8 d. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.23 Nitrogen removal by bio-adsorption MBR operated at SRT ≈ 1.2 and 2.8 d. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.24 Fractions of organic, ammoniacal, nitrate and nitrite nitrogen in influent and effluent at SRT = 1.2 and 2.8 d. . . . . . . . . . . . . . 3.25 Phosphorus removal by bio-adsorption MBR operated at SRT ≈ 1.2 and 2.8 d. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.26 COD mass balance at SRT = 1.2 and 2.8 d. . . . . . . . . . . . . . 3.27 Recovery of resources by AMBR treatment at SRT = 1.2 and 2.8 d followed by a thickening of the concentrate by sedimentation. . . . . 3.28 Permeate flux and recovery as a function of TMP for reverse osmosis treatment of MBR effluent. . . . . . . . . . . . . . . . . . . . . . . . 3.29 Comparison of the specific methane productions of the different reactor set-ups during digestion of adsorptive sludge. . . . . . . . . . 3.30 Specific methane production of AnSBR reactor digesting adsorptive sludge, compared to a CSTR operating at the same HRT and OLR. 3.31 Fraction of total effluent load of TS, COD, N, resp. P, present in the overflow effluent, resp. wasted sludge of an AnSBR digesting adsorptive sludge at mesophilic conditions. . . . . . . . . . . . . . . 3.32 VFA evolution in the two TPAD reactors. . . . . . . . . . . . . . . 3.33 Effect of pH control in the thermophilic TPAD reactor on the combined specific methane production of the TPAD reactors and on the specific methane production and pH of the thermophilic TPAD reactor itself. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.34 Organic loading rate, HRT and resulting specific methane production by a mesophilic and thermophilic CSTR reactor digesting concentrate obtained by centrifuging municipal wastewater. . . . . . . .

16

91 92 93 93 94 95 96 97 98 101 103

104 106

106

108

4.1

Process layout of conventional activated sludge system with anaerobic digestion of excess sludge. . . . . . . . . . . . . . . . . . . . . . 111

17

List of Tables 1.1

Potential recovery of resources from municipal wastewater. . . . . . 22

3.1 3.2 3.3

Effluent quality after primary sedimentation and CEPT treatment Effluent quality after FMX filtration of municipal wastewater. . . bio-adsorption and sedimentation: influent and effluent quality and removal efficiency. . . . . . . . . . . . . . . . . . . . . . . . . . . . Influent, effluent and removal efficiency by RO treatment of bioadsorption MBR effluent. . . . . . . . . . . . . . . . . . . . . . . . Operational conditions (HRT, SRT, OLR), pH, VFA alkalinity (Ripley index), and the timespan of the test runs performing anaerobic digestion of adsorptive sludge. . . . . . . . . . . . . . . . . . . . . Methane content of biogas and specific methane production rates of reactors performing anaerobic digestion of adsorptive sludge. . . . Influent (’In’) and effluent (’Out’) qualities and removal efficiencies (’RE’) during anaerobic digestion of adsorptive sludge. . . . . . .

3.4 3.5

3.6 3.7

metabolism metabolism metabolism metabolism metabolism metabolism

of of of of of of

. 79 . 83 . 88 . 98

. 100 . 100 . 102

4.1 4.2 4.3 4.4 4.5 4.6

Energy Energy Energy Energy Energy Energy

the conventional activated sludge system. . . CEPT + AD. . . . . . . . . . . . . . . . . . Centrifugation + AD. . . . . . . . . . . . . . FMX filtration + AD. . . . . . . . . . . . . . MBR + AD with SRT = 1.2 d in the MBR. . MBR + AD with SRT = 2.8 d in the MBR. .

5.1

Overall and technical energy efficiency of the upconcentration techniques in combination with AD, as well as their estimated OPEX and an assessment about the suitability of their effluent for water recovery via UF+RO or only RO. . . . . . . . . . . . . . . . . . . . 135

18

114 119 122 125 129 131

Chapter 1

Literature review 1.1

The twin problems of water scarcity and wastewater management

The quality of water is important for the well-being of the environment, society and the economy. Access to clean water and adequate sanitation are crucial and interlinked prerequisites for protecting the health of individuals and communities (UNEP (2011)). Because of this, wastewater management plays a crucial role in achieving sustainable development (Corcoran et al. (2010)). Unfortunately, the world is facing the twin problems of growing water scarcity and an increase in unregulated or illegal discharge of contaminated water into the environment (Corcoran et al. (2010)). The OECD estimates that by 2030, due to population growth, increasing living standards, over-exploitation, water pollution, ecosystems degradation and climate change, nearly half the world’s population will be living under conditions of severe water stress, which it defines as a situation in which ratio of local water use to the renewable supply exceeds 40% (OECD (2009)). In fact, unless improvements in the efficiency of water use are made at a much faster rate than is currently being achieved, the total global water demand is projected to overshoot supply by 40% by that time (Water Resources Group (2009)). The BRIC countries (Brazil, Russia, China and India) are expected to bear the heaviest burden (see Figure 1.1). This looming water crisis poses a fundamental threat to many rapidly developing countries’ aspirations of lifting billions of people out of poverty. Increases in supply

19

Figure 1.1: Number of people expected to be living in water-stressed areas in 2030 by country type. OECD: 34 member states of the Organisation for Economic Co-operation and Development, i.e. industrialized nations (OECD (2009)). BRIC: Brazil, Russia, India and China, i.e. rapidly emerging economies. RoW: Rest of the world. .

through such measures as the construction of dams and reservoirs, implementation of inter-basin transfers, desalination plants and increased recycling by means of the current conventional technologies are expected to be able to fill 40 per cent of the supply-demand gap, but are not without their disadvantages. Dams and reservoirs and interbasin transfers are controversial in terms of social and/or ecological disruption. Other water supply measures are typically reliant upon the use of increased amounts of energy and, as a result, the costs of water provision are on the rise in most regions where there is physical water scarcity (UNEP (2011)). Along with crucial investments in water infrastructure and water-policy reform, there is, therefore, a real need for the development and implementation of new, more sustainable technologies for water supply and management that can cover the remaining 60 per cent in an energy-efficient and locally affordable manner (UNEP (2011)). Meanwhile, driven by the growing global and urban populations and by increases in living standards, wastewater production is on the rise. Wastewater contaminates ecosystems, threatens food security and access to safe drinking and bathing water (Corcoran et al. (2010)). For example in India, discharge of inadequately 20

treated or untreated municipal wastewater has resulted in the significant contamination of 75 % of all surface waters (GOI (2008)). Whilst wastewater production is on the rise in many parts of the world, the wastewater infrastructure and management systems are inadequate, both in the developing countries as well as in the industrialized world. In many parts of the developing world, treatment systems are either overwhelmed, poorly functioning or non-existent (Corcoran et al. (2010); Leblanc et al. (2008); Kumar (2009)). One of the biggest opportunities to make the transition to a greener economy in these parts of the world lies in investing in the provision of adequate water and - perhaps even more pressing - sanitation and treatment services to the poor, as 39% of the world population still does not have access to improved sanitation (Hutton & Bartram (2008); WHO/UNICEF (2010)). UNEP has stated that failure to do so quickly and sufficiently will exacerbate the current situation in which millions of people die every year due to water-related diseases and will cause further losses in biodiversity and ecosystem resilience (Corcoran et al. (2010)). In the industrialized countries on the other hand, despite having achieved much in terms of effluent quality, the realization is growing that the current wastewater treatment infrastructure poses an unsustainable burden, because of its energy consumption and carbon footprint as well as because of the wastage of potentially recoverable resources which it entails (Verstraete & Vlaeminckx (2011); Verstraete et al. (2009); Muga & Mihelcic (2008)). Therefore, societies must come to consider wastewater as a part of the solution to the problem of scarcity of water in particular and natural resources in general. Public wastewater infrastructure can be one of the determining factors in the sustainable development of societies and in their future use of water, energy and materials, as wastewater actually is a potential source from which clean water, energy, nutrients and carbon can and should be recovered in an economical manner (Verstraete et al. (2009); Angenent et al. (2004); Sutton et al. (2011); Leblanc et al. (2008)).

21

Table 1.1: Potential recovery of resources from municipal wastewater. Prices are based on the market value of comparable products. (Verstraete et al. (2009)).

Potential recovery

Per m3 sewage

2009 market prices

Total per m3 sewage (e)

Water Nitrogen Methanea Organic fertilizerb Phosphorus

1 m3 0.05 kg 0.14 m3 0.10 kg 0.01 kg

e0.25/m3 e0.22/kg3 e0.34/m3 CH4 e0.20/kg e0.70/kg

0.25 0.01 0.05 0.02 0.01

Total

0.35

a

Based on 80% recovery of organic matter in the form of biogas and 0.35 m3 /kg COD removed.

b

Based on 20% organic matter remaining after AD. The price is based on the agricultural value of

organics.

1.2

Resources present in municipal wastewater

Municipal wastewater typically consists of the following components: water; nontoxic organic compounds; nitrogen and phosphorous components; pathogens and other microbiological pollutants; toxic inorganic and organic pollutants such as heavy metals, polycyclic aromatic hydrocarbons (PAHs), pesticides, linear-alkylsulfonates, etc; and inorganic non-toxic compounds such as silicates, aluminates, and calcium- and magnesium-containing compounds (Rulkens (2008)). It can also have a substantial heat content. Of these, the water, energy (in the form of chemical bonds and heat) and nutrients (N, P and C) can be considered valuable resources, the recovery of which can be desirable. Table 1.1 lists the potential recovery of resources from municipal wastewater, not including heat recovery, under the assumption that the chemical energy is recovered in the form of methane through anaerobic digestion (AD) and that the remaining organic carbon after such treatment is used as an organic fertilizer (Verstraete et al. (2009)). Particulate and dissolved components

The distribution of contaminants between settleable (>105 nm), colloidal (1-105 nm) and dissolved ( 20 to 25 µm (Hjorth et al. (2010)). Pretreatment with coagulants and/or flocculants can increase the particle size and thus improve the removal efficiency of fine colloidal material in the centrifuge (Woon & Leung (1998)). On the other hand, it has been found that pretreatment of sewage with a coagulant dose up 100 mg FeCl3 /L, and 1 mg cationic flocculant followed by centrifugation at 2000G could only achieve maximally 10% better recovery of COD than centrifugation without coagulants and flocculants, which achieved some 52% COD recovery (Van Wesenbeeck (2010)). The same author did find that, in order to reach a phosphorus removal of 85%, by combining coagulation with centrifugation, a four times lower dose of FeCl3 could be applied compared to conventional CEPT treatment with sedimentation.

48

Figure 1.6: Some options for recovering high quality water from municipal wastewater using membrane technology (Diamantis et al. (2009)).Top: activated sludge pretreatment. Middle: membrane bioreactor pretreatment. Bottom: primary membrane filtration (no pretreatment).

1.9.3

Primary membrane filtration

Biological or physico-chemical pretreatment can be left out of the process train entirely and membrane filtration applied directly to raw wastewater (or primary effluent) as the main up-concentration technique (see Figure 1.6), in which case one speaks of direct or primary membrane filtration (Ravazzini et al. (2005); Diamantis et al. (2010)). It has the advantage that it could provide reusable water (e.g. for irrigation) in one step and can be operated discontinuously and therefore it is expected that this approach will be best applicable in decentralized, smallscale reuse schemes (Ravazzini et al. (2005)). The reason why primary membrane filtration is not yet considered feasible on a larger scale are the substantial fouling problems. The deposition of substances (suspended or dissolved) on or in the membrane causes the water flux through the membrane to decrease and increases the operating and capital cost of membrane filtration by increasing the frequency of membrane backwashing and chemical cleaning. Using flat sheet membranes with a cut-off of 0.4 µm, a permeate flux of 7 L/(m2 ·h), a COD removal of 85% and a SS removal of 100% has been reported, at a cost of 2.1 e/m3 (Diamantis et al. (2009, 2010)). Using a vortex-generating rotating membrane module known as FMX, (Van Wesenbeeck (2010)) reported a COD removal of 87% at a cost of 2.6 e/m3 . As such, primary membrane filtration has thus far remained much more expensive than a CAS + UF/RO treatment (approximately 1e/m3 , (Van Houtte & Verbauwhede (2008))).

49

Several strategies for mitigating membrane fouling have been attempted, such as cross-flow operation, vibrating modules, rotating modules, spacers, inserts and other turbulence promotors, air scouring and backwashing ((Diamantis et al. (2010)) and references therein). In this thesis, a roting membrane module with vortex generating structures, known as FMX, was used. In addition to the operating mode, the membrane material is important too, with a need for narrow pore size distributions and fouling-resistant (hydrophilic) surface chemistries in order to achieve high fluxes (Shannon et al. (2008)). In this respect, electrospun nanofibrous membranes which can be produced with a well-defined fibre diameter and porosity can outperform conventional porous membranes such as those created by the phase immersion process (Yoon et al. (2006)). 1.9.4

Bio-adsorption and -flocculation

The upconcentration techniques mentioned so far are limited mostly to the recovery of particulate matter but are not able to deal succesfully with dissolved COD. An alternative approach towards upconcentration which could also handle dissolved components to a certain extent could be to utilize adsorption, i.e. the passive uptake of pollutants from aqueous solutions by adsorbent materials, which can then be easily separated from the aqueous solution, a method which is typically used for heavy metal or toxic organic compound removal from (industrial) wastewaters (Babel & Kurniawan (2003); Forgacs et al. (2004)). A well known adsorbent is activated carbon, but in the context of municipal wastewater upconcentration, cheaper and more practical adsorbent would be required (Aksu (2005)). Examples of low-cost adsorbents are waste products such as bagasse pith, rice husk, fly ash and carbonized sewage sludge, but their adsorption capacities are typically limited and non-regenerable (Aksu (2005)). Alternatively, so-called biological adsorption, i.e. the physical uptake of pollutants by microbial cells or flocs - as opposed to biodegradation through aerobic or anaerobic metabolic pathways - can be considered (Aksu (2005)). This concept originates from the observation that COD removal in activated sludge reactors takes place as a result of three distinguishable processes: the physicochemical adhesion of organic compounds, both soluble and particulate, onto the sludge flocs; the extracellular hydrolysis of adsorbed particulate and complex soluble substrate, and the metabolisation (towards cell synthesis and respiration) 50

proper of the soluble BOD inside the cells (Tan & Chua (1997); Zhao et al. (2001); Riffat & Dague (1995)). The first occurs on a significanlty shorter timescale than the latter two (typically 10 to 30 min), which can be exploited by separating sludge from the mixed liquor after a short contact time, in effect working at low SRT and HRT. It is exptected that extracellular polymeric substances (EPS) excreted by the sludge cells play a significant role in the observed physico-chemical COD uptake, with biopolymers on the sludge surface acting as flocculants that bind colloidal and suspended matter to the sludge flocs (Sheng et al. (2010)). This relatively fast physico-chemical uptake step of specifically colloidal and suspended matter is usually referred to as biological flocculation, with the physico-chemical uptake of soluble COD being referred to as biological adsorption (Huang & Li (2000); Leal et al. (2010); Akanyeti et al. (2010)). For simplicity’s sake, we will refer to physicochemical adhesion of pollutants to sludge flocs as bio-adsorption, regardless of the size of the pollutant. The adsorptive ability of sludge flocs can be advantageous to the cells because colloidal and soluble but slowly biodegradable COD require extracellular breakdown by exoenzymes into soluble, low molecular weight compounds prior to their transport into the cells for it to be metabolized for energy production and cell growth. Such extracellular breakdown can occur more efficiently when those compounds are adsorbed onto sludge flocs (Guellil et al. (2001)). (Guellil et al. (2001)) reported that in some 20 to 40 mintues, some 45% of the non-settleable COD mostly colloidal matter - of a municipal wastewater could be adsorbed by activated sludge, which was reported to have a biosorption capacity of 40-100 mg COD/(g TSS), and that bio-adsorption was more efficient when the sludge cells were subjected to a high loading of organic matter. This ability of activated sludge to rapidly adsorb BOD is exploited in several wastewater treatment processes: the well-known contact stabilization process (Magara et al. (1976)), a two-step activated sludge processes with separate sludge recycling loops known as the Adsorption/Bio-oxidation (A/B) or Boehnke process (Bohnke (1985)) and the so-called Contact Adsorption Regeneration Process (Liu et al. (2009)). A hybrid treatment combining chemical coagulants and biosorption known as Chemical Biological Flocculation has also been described (Zhang et al. (2007)). 51

The A/B process

The A/B process comprises a small first reactor which has an extremely high sludge loading rate Bx (some 3 to 15 kg bCOD/(kg MLVSS · d)) and has a high rate of sludge withdrawal (SRT of 0.1 to 0.5 d) (Bohnke (1985); Versprille et al. (1985); Bohnke et al. (1998); Salome & Eggers (1997)). Such a first reactor is typically able to remove some 45 to 65% of BOD of which an estimated 60 to 70% is removed through adsorption to the sludge flocs and subsequent wasting of the energy-rich sludge, thus preventing excessive aerobic mineralization from occurring (Gao & Cardoen (2011)). We will refer to this type of sludge as adsorptive sludge. As an advantage over purely physico-chemical up-concentration, the young, fast growing biomass in the bio-adsorption process can remove some soluble COD, although it has been reported that the very low SRT serves to select bacteria which are able to metabolize only a limited fraction (some 50%) of (what is normally considered as) readily biodegradable COD (Haider et al. (2000, 2003)). The young and genetically even bacterial community is relatively adept at neutralizing pH variations in the influent and well resistant to changes in DO, load and influent toxicity (Bohnke (1985)). The adsorptive sludge is also reported to settle well (SVI < 100 mg/L), but it has been suggested however that SRT in itself affects bioflocculation, with a minimum SRT of 2 days required to allow sufficient EPS production to occur and flocs to form (Bohnke et al. (1998); Sheintuch et al. (1986)). The limited oxygen demand and the passing on of adsorptive sludge - which is in effect a mixture of young cells and adsorbed and settled primary sludge - to the AD, can result in good energy recovery and in some cases even an energy-positive wastewater treatment (Wett et al. (2007)). In the A/B system, the second reactor operates under a relatively low sludge loading rate (Bx ≈ 0.1 kg bCOD/kg MLVSS·d); it’s function is to oxidate the remaining organics and perform nitrogen removal (Bohnke et al. (1998); Gao & Cardoen (2011)). Contact adsorption regeneration process

It has been argued that the adsorptive activated sludge step used in the A/B process can be optimalized by intensively aerating the fraction of recycled sludge prior to bringing it in to contact with the influent, as kinetic theory suggests that rapid removal of colloidal COD will occur better when the biomass is in an

52

endogenous state (Bunch & Griffin (1987)). The aeration serves to regenerate the adsorptive and flocculating capacity of the flocs by allowing the recycled sludge to biodegrade any bio-adsorbed matter and produce EPS before coming again into contact with fresh influent, and create a selective pressure towards bacteria which use extensive adsorption as part of their survival strategy (Huang & Li (2000)). The CAR process thus employs a mixing flocculation tank where influent is mixed with recycled sludge (recycle rates of 15% to 40% have been used) coming from a sludge aeration tank where it had been thoroughly aerated for some 30 minutes to 2 h (Huang & Li (2000); Liu et al. (2009)). After the adsorption/flocculation in the mixing tank, the mixed liquor is allowed to settle in a sedimentation tank. The CAR process has been reported to remove some 50% to 80% of COD. It’s claim to better cost-effectiveness lies in the fact that only the recycled fraction of the sludge is aerated, as opposed to the whole mixed liquor in conventional CAS; the energy requirement for aeration was calculated to be down to 9% of that of CAS (Huang & Li (2000)). Bio-adsorption in a membrane bioreactor

Conventional membrane bioreactors (MBR) have become an established alternative to CAS. In the immersed-type MBR, the membrane is immersed in the aerated reactor and is operated using a relatively small underpressure (0.2-0.6 bar), resulting in relatively small fluxes (10 to 35 L/(m2 · h)) and thus a large required membrane area, but relatively smaller power costs than external cross-flow MBR systems (some 0.3 kWh/m3 , comparable to CAS), also because no recirculation pumping is required and because the aeration performs the double function of providing oxygen to the biomass as well as scouring the membrane (Van der Meeren (2010)). MBRs are relatively expensive to install and operate but their ability to retain all suspended solids (including most viruses and bacteria), allows for either direct reuse (e.g. for flushing purposes) or further treatment with reverse osmosis to produce drinkable water (Melin et al. (2006); Comerton et al. (2005)). MBR treatment is only slightly more expensive than activated sludge treatment (both approximately 0.6 e/m3 ), resulting in an approximate cost of 0.8-0.9 e/m3 for an MBR+RO system, compared to some 1.0 e/m3 for a CAS+UF+RO system (Diamantis et al. (2010)). MBRs allow to retain a large biomass concentration in the reactor, and thus allow 53

to operate at very low sludge loading rates, leading to the production of very little excess sludge (typically less than 0.1 kg TS/kg COD removed) (Ng & Hermanowicz (2005)). Although this approach has been popular because of the high costs associated with sludge treatment, it is not desirable if one wishes to use MBR filtration as an up-concentration technique within the zero-waste concept, as the excessive mineralization and the production of a sludge which is hard to digest anaerobically will lead to a negative energy balance. An alternative approach is therefore to operate the MBR at a high volumetric loading rate (and thus also having a very small footprint), in analogy to first reactor in an A/B system, and to waste a large fraction of the energy-rich, adsorptive sludge towards AD (Leal et al. (2010); Akanyeti et al. (2010)). To achieve an as concentrated adsorptive sludge as possible, the ratio SRT/HRT should be as high as possible in such an MBR, whilst keeping the SRT low enough to avoid excessive mineralization. Indeed, SRT/HRT would be equal to the concentration factor (i.e. the ratio of the COD of the concentrate and that of the influent), were it not for mineralization and incomplete membrane retention of COD (Akanyeti et al. (2010)). The advantage over a comparable treatment with a settler is that MBRs can be operated with few concerns about sludge settleability and that the MBR can achieve much better removal. The effluent is free of suspended solids, whereas that of the first reactor in an A/B system is typically still quite turbid. There is no selective pressure towards flocculating microorganisms as they are no longer subjected to settling, and as a result MBR systems have been reported to have a more dispersed microbial community, which is thought to contribute positively towards better CODs removal at extremely short SRT compared to a system utilizing sedimentation which has a more flocculated sludge (Ng & Hermanowicz (2005)). The creation of a fouling biofilm layer on the membrane surface is also thought to contribute positively to the removal of CODs, with a 60% difference observed between the CODs in the mixed liquor and that in the permeate (Akanyeti et al. (2010)). Fouling of the membrane however puts a limit to the maximum flux which can be achieved and as such it puts a lower limit to the HRT in an MBR, and thus to the maximum obtainable concentration of the concentrate (Akanyeti et al. (2010)). The combination of non-flocculating microorganisms and reduced EPS production at SRTs less than 2.5 d contributed to poorly sludge settling in the MBR (SVI > 500 mg/L) (Ng & Hermanowicz (2005)), which can be problematic for the further 54

thickening of the sludge prior to AD. Working at SRTs as low as 0.25 d, an MBR operating on a completely soluble synthetic substrate was able to achieve 98% COD removal, some 10% to 20% better than a comparable setup utilizing sedimentation (Ng & Hermanowicz (2005)). In a setup treating grey water, at SRTs of 0.2 to 1.0, an HRT of approx 1.8 h, a 78% COD removal, with 61%, resp. 55% of the influent COD being retained in the concentrate was achieved (Leal et al. (2010)). Using municipal wastewater, an MBR using a PVDF membrane (0.1 µm cut-off) and operating at 0.25, resp. 1 d SRT and 1.2 h HRT achieved fluxes of 9.3, resp. 7.8 L/(m2 · h) and COD removal of 77%, resp. 87%, with 54%, resp. 33% being retained in the concentrate, i.e. the wasted sludge (Akanyeti et al. (2010)). A highly loaded MBR system treating effluent from the primary treatment of municipal wastewater experienced a 20-fold increase in fouling rate over a five-fold decrease in SRT from 10 to 2 d and a ceasure of nitrification at SRT lower than 3 d (Trussell et al. (2006)). It was observed that the extent of flocculation was considerably higher at an SRT of 1 d than at SRTs of 0.5 or 0.25 d. The methane production potential of the concentrate was observed to be 64 %, allowing for an overall energy recovery of influent COD to methane of 35%. With regards to nutrients, some 40% of P was retained in the concentrate, while another 40% remained in the permeate and some 20% presumably remained in the fouling layer. Nitrification was observed to be absent and some 15 to 25 % of total N in the influent ended up in the concentrate, which suggests the ability of the sludge to adsorb NH+ 4 . (Nielsen (1996)) also reported the ability of activated sludge to adsorb some 0.5 mg NH+ 4 -N/g SS. Reverse osmosis

Reverse osmosis (RO) is a pressure-driven membrane filtration process using semipermeable membranes with a pore radius smaller than 1 nm, which is able to filter out dissolved components (van Voorthuizen et al. (2005)). The filtration occurs via two mechanisms: electrostatic exclusion and size exclusion. As a result, (hydrated) ions, especially those with high ionic charge, and all but the smallest molecules are rejected by the membrane. Most often, RO membranes are made of cellulose acetate, polyamide or polysulfone/polyamide thin-film composites (Van der Meeren (2010)). Reverse osmosis is widely used for seawater desalination to produce drinkable water, but this requires high pressures (50 bar) and thus high energy costs. 55

When applied for water reclamation from wastewater treatment effluent, typical pressures required to reach the same recovery are about half of those required for desalination, and thus the energy requirement is much less (Van der Meeren (2010)). In addition to serving as a final treatment step towards recovering clean water from treated wastewater as permeate, RO could be used to recover nutrients from treated wastewater in the concentrate, to be used as a liquid fertilizer. Results with anaerobically treated blackwater showed recoveries for ammonium between 80% and 90% and for phosphate above 90% (van Voorthuizen et al. (2005)).

1.10

Research question

The main objectives of this study were to assess and optimize the removal and resource (water, COD, N, P) recovery efficiency of 3 upconcentration techniques - chemically enhanced primary treatment, centrifugation, primary membrane filtration and bio-adsorption - applied to municipal sewage, and the performance of the subsequent anaerobic digestion of concentrates produced through such upconcentration techniques. The energy and cost efficiency of the upconcentration and anaerobic digestion combination will be compared to the current benchmark, conventional activated sludge treatment with anaerobic digestion of secondary activated sludge.

56

Chapter 2

Materials and methods 2.1 2.1.1

Analytical techniques Total solids, total suspended solids, volatilizable solids and volatilizable suspended solids

Total solids (TS) were determinated by drying the sample in a crucible for 24h at a temperature of 105◦ C. Volatilizable solids (VS) were determined through a subsequent ashing of the dry solids for 2h at a temperature of 600◦ C. The difference in mass between the dry solids and the ashes was taken as the mass of the volatilizable solids in the sample. For samples containing a relatively high amount of suspendid solids, total suspended solids (TSS) and volatilizable suspended solids (VSS) were determined in the same manner, but after the following pretreatment. Samples were centrifuged for 10’ at 10000 rpm, the supernatant was removed and replaced with MilliQ water, afer which the sample was mixed, centrifuged again, the supernatant removed, and the residu transferred to the crucible. For samples containing a low amount of suspended solids, the TSS content was determined by filtering a known amount of sample over a 0.45 µm filter and drying and weighing the filter at 105◦ C for 24 h.

57

2.1.2

Chemical oxygen demand

Determination of the total chemical oxygen demand (COD) of sludge samples was performed using the classical method of oxidation by potassium dichromate. 10 mL of potassium dichromate (0,25 N K2 Cr2 O7 ), 30 mL of sulfuric acid-silver sulphate (as a catalyst) solution and some HgSO4 (to prevent chloride interference) were added to 20 mL of the sludge sample. After a destruction period of 2 hours at a temperature of 153◦ C, the solution was titrated against an 0.0625 N Fe(NH4 )2 (SO4 )2 solution with ferroine as indicator. This method allowed COD determination withing a range of 100-800 mg COD/L. Appropriate dilutions were made when necessary. For samples with a COD between 10 and 100 mg/L, K2 Cr2 O7 with a concentration of 0,025 N was used. Soluble COD determination was performed in the same manner, after centrifuging (10000 rpm for 10’) and filtering (pore size 0.45 µm) the sample. The COD content of water samples was analyzed using Hanna Instruments cuvette test kits (HI 93754A-25 LR 0-150 mg/L, HI 93754B-25 MR 0-1500 mg/L) and colorimetry (DR/700, Hach Company). To determine the soluble COD, the samples were first filtered over a filter with pore size 0.45 µm. 2.1.3

Total ammonium/ammonia nitrogen

The total ammonium/ammonia nitrogen (TAN) of samples containing more than 2 mg TAN/L was determined through distillation. 0.4 g of MgO was added to 10 mL of sample to convert all ammonium to ammonia, which was subsequently distilled out (Vapodest 30s, Gerhardt) and condensed into a boric acid indicator solution. Back-titration of the boric acid indicator solution with 0.02 M HCl to its original pH of 5.3 allows for determination of the ammonium content. For samples containing less than 2 mg TAN/L, the Nessler colorimetric method was used. 1 mL of Nessler reagens (Potassium tetraiodomercurate(II) (K2[HgI4]) in a potassium hydroxide solution) was added to 50 mL of the (diluted) sample, along with 1 mL potassium sodium tartrate solution (to prevent interference). After 10’, the colour intensity of the resulting yellow complex (relative to that of a blank) was determined colorometrically at a wavelength of 425 nm (WPA-Lightwave II

58

UV/ Visible spectrophotometer, Biochrom) . 2.1.4

Kjeldahl nitrogen

The Kjeldahl nitrogen (encompassing both the organically bound nitrogen and ammonium/ammonia nitrogen) content of samples was deternmined by destructing the sample at 400 ◦ C in the presence of concentrated sulfuric acid, cupper sulfate (as a catalyst) and potassium sulfate (to increase the boiling point), causing all organically bound nitrogen to be converted to ammonium. Next, sodium hydroxide was added to the sample to convert all ammonium to ammonia. The remainder of the procedure was as for TAN-determination. 2.1.5

Total and soluble phosphorus

Total phosphorus determination was performed by means of colorimetry. Samples were first ashed as per the VS determination method, after which 10 mL of 1N nitric acid was added. The nitric acid solution was filtered over a paper filter and diluted to 50 mL. 0,5 mL of the solution was further diluted to 3mL, after which of 0.5 mL of Scheel I and Scheel II reagents were added. After mixing and a reaction time of 10‘, 1 mL of Scheel III was added, again followed by mixing and a reaction time of 10‘, after which colorimetric determination of the phosphorus content took place. To determine the amount of phosphorus bound to the suspended solids (PSS ) of a sample, the same method was applied on the ashes of the suspended solids. Soluble phosphorus was determined in much the same manner, after filtering the original sample over a 0.45 µm filter. Ashing of the sample was only performed if the sample had colour. 2.1.6

Volatile fatty acids

Determination of volatile fatty acid (VFA) - i.e. acetate, propionate, isobutyrate, butyrate, isovalerate, valerate, isocaproate and caproate - was based on ether extraction and gas chromotography. 0.5 mL of sulphuric acid, 0.4 g of sodium chloride, 0.4 mL of an internal VFA standard and 2 mL of diethyl ether were added to 2 mL of sample. The mixture was mixed for 2’ and centrifuged for 3’ at 3000 59

rpm. The etheric layer was removed and analysed through gas chromotography with flame ionisation detector (GC 2014, Shimadzu) using an Alltech EC-1000 column. VFA concentrations between 0 and 10 g/L could be determined through this method. 2.1.7

pH

pH measurements were performed with a Consort SP10B pH electrode connected to a Consort C532 multimeter analyser. 2.1.8

Alkalinity and Ripley index

Titration of the sample with 0.35 N sulfuric acid down to pH 4.3 allows to calculate the total alkalinity. Titration from pH 5.75 down to pH 4.3 allows to calculate the intermediate alkalinity, which serves as a measure for the VFA alkalinity. The alkalinities are calculated as: A · 0.35 · 50000 , V with A the volume of acid (mL) and V the sample volume (mL). The Ripley index (Ripley et al. (1986)) is defined as the ratio of the intermediate alkalinity and the total alkalinity. Alkalinity (mg CaCO3 /L) =

2.1.9

Turbidity

The turbidity of a sample was determined by measuring scattered light intensity with a nephelometer, using a a formazine solution and distilled water as standards. 2.1.10

Nitrate, nitrite and orthophosphate

− 3− Nitrate (NO− 3 ), nitrite (NO2 ) and orthophosphate (PO4 ) concentrations were analyzed through ion chromatography (IC 761 Compact,Metrohm, with an AS9HCcolume and a Metrosep A 4/5 guard column). The non-coloured sample was filtered over 0.45 µm filter. A further inline filtration of 0.2 µm filter was done in the IC.

2.1.11

Sludge volume index

Sludge volume index (SVI) was determined by allowing 1 L of sludge to settle in an Imhoff cone for 30’ (unless otherwise mentioned). The ratio of the volume of 60

settled sludge (mL/L) and the TSS of the sludge (g/L) is the SVI (mL/g). 2.1.12

Methane concentration

The methane and carbon dioxide content of biogas was determined using gas chromotography with a thermal conductivity detector (Shimadzu GC-14B, equipped with a Hayesep Q 80-100 kolom, and a Shimadzu C-R8A integrator) and corrected to take into account the headspace volume. 2.1.13

Capillary suction time and Kozeny constant

The capillary suction time (CST) of sludge samples was determined by measuring the rate at which the capillary suction pressure generated by a standard filter paper could suck water out of a sludge sample contained within a open-ended cilinder resting on the centre of the filter paper. The time taken for the water front to pass between two electrodes positioned on top of the filter paper and at a standard distance from each other constitutes the CST. From the CST, the Kozeny constant (K), an index for the dewaterability of sludge, can be calculated according to: K = 10−4 ·

2.2 2.2.1

TS CST

Experimental setups Wastewater

The wastewater used in the up-concentration tests was obtained from Ossemeersen WWTP (Aquafin, Gent). Unless otherwise mentioned, the wastewater was collected at the point where it enters the WWTP, ie prior to pre-treatment. 2.2.2

Chemically enhanced primary treatment

Flocculant selection

A preliminary test was performed for flocculant selection. Degritted domestic wastewater was obtained from Ossemeersen WWTP, Gent. FeCl3 ·6H2 O was used as the coagulant, of which a series of doses (0, 20, 40, 80, 150, 200, 250, 400 and 500 mg FeCl3 /L) was added to 100 mL samples who were then mixed at approximately 100 rpm for 10’. pH was measured and if necessary corrected with NaOH 61

Figure 2.1: CEPT setup: mixers.

to be above 6.5. The optimal coagulant dose was visually selected on the basis of the clarity of the liquid phase. Next, 800 mL wastewater samples were dosed with the optimal coagulant dose, mixed at 100 rpm for 10’, after which four different flocculants were tested: Catflock P1556 (medium cationic), Zetag 72 (weak cationic), Betz 1100P (weak anionic) and M156 (strong anionic), at two different doses: 0,5 mg/L and 1 mg/L. The samples were mixed at 150 rpm for 15”, then 30 rpm for 10’. A visual check for floc formation was performed. The flocs were allowed to settle for 30’, after which the supernatant was sampled and analyzed for CODt content to determine the optimal flocculant type. Coagulant dose selection

Both FeCl3 ·6H2 O and Al2 (SO4 )3 ·12H2 O were used as coagulant. Doses of 0, 20, 40, 60, 80, 100, 150, 200 mg FeCl3 /L and 21, 41, 62, 82 and 103 mg Al2 (SO4 )3 /L were added to Ossemeersen wastewater samples (800 mL) and rapidly mixed at 150 rpm for 15’. pH was checked and if necessary corrected with NaOH to be above 6.5. The wastewater was allowed to settle for 30’, after which samples (normal and filtered over 0.45 µm) of the supernatant were taken. The turbidity of the supernatant was analyzed on the spot. Next, 1 mg/L of Betz 1100P (weak anionic) flocculant was added, mixing performed at 150 rpm for 15”, then 30 rpm for 10’, after which the wastewater was allowed to settle for 30’. Similar as before, the supernatant was sampled and turbidity determined on the spot. All samples were

62

Figure 2.2: CEPT setup: sedimentation.

frozen and analzed the following day for CODt , CODs and PO3− 4 . Up-concentration test

Having selected coagulant dose and flocculant type and dose, up-concentration tests were performed. For FeCl3 , a dose 100 mg FeCl3 /L was added to 800 mL of wastewater. For Al2 (SO4 )3 , a dose of 103 mg Al2 (SO4 )3 /L was used. For each coagulant, up-concentration was tested both without and with flocculant (Betz weak anionic at a dose of 1 mg/L), the latter in triplicate. Mixing regimes were performed as described in the previous section. As a blank treatment, simple sedimentation was performed (also in triplicate). The settling for all the treatments was analyzed as follows: after 5, 30 and 60’, the volume of settled sludge was recorded and the turbidity of the supernatant analyzed. After 60’ of settling, samples were taken of supernatant and sludge. The supernatant was analyzed for CODt , CODs , TSS, TAN, Pt and PO3− 4 . The sludge was analyzed for CODt , TS, VS, Kj-N and Pt . 2.2.3

Centrifugation

In the experiments an industrial grade, vertical type centrifuge (Westfalia Seperator RTC) which operates in continuous mode was used (see figure 2.3). It could generate a G-force of 11.7 at a rotation rate of 9210 rpm. The centrifuge chamber volume was 3.0 L and the clarification surface was 10.2 m2 . The concentrate was collected in a 1.5 L chamber. To collect the concentrate from the chamber, rinsing water had to be used, causing a dilution of the solutes into a 1.5 L solution. An additional test was performed in which FeCl3 coagulant was added. 400 L vessels of raw wastewater were dosed with FeCl3 /L), stirred vigorously for 10’ and 63

Figure 2.3: Westfalia RTC centrifuge unit.

64

(a) Cross-flow

(b) FMX

Figure 2.4: Conventional cross-flow filtration and FMX filtration.

gently for another 10’ before submitting the wastewater to centrifugation. 2.2.4

FMX ultrafiltration

The FMX unit is a membrane filtration module produced by Filmax co. that uses stacked circular flat sheet membranes interspersed with vortex generators that are rotated by means of a central driving shaft and as such generate Karman vortices. These vortices maintain turbulent flows near the membrane surface which disrupt the boundary layers built by the foulants (see figure 2.4). The FMX filtration unit is depicted in figure 2.5. For the primary membrane filtration experiments, an FMX KPT20 type was used, equipped with a single membrane with a surface area of 0.877 m2 . Rotation speed was 450 rpm, with a concurrent energy consumption of 5.32 kW. The feed pump was able to deliver a constant feed flow rate up to pressures of approximately 3 bar. In full-scale application, membranes can be stacked in the FMX module to give a total membrane surface of 3.15 m2 . Two membranes were tested. Both were polyamide-6 fuzzy electrospun nanofibre membranes, produced by Enanics. The first membrane had a cut-off of 0.5 µm, the second 1 µm (fibre size 130 nm, fibre density 2 g/m2 ). 65

Figure 2.5: FMX filtration unit.

Figure 2.6: Half open circuit for up-concentration tests.

A clean water flux test was performed at the start of each experiment. Both membranes were subjected to wastewater filtration tests in a closed circuit (recirculation of permeate and concentrate to the feed) in which the flux was measured in function of the pressure, in order to select the best membrane (on the basis of the highest obtainable flux) and to determine the optimal transmembrane pressure. The actual up-concentration tests were performed in a half-open circuit, i.e. with recirculation of the concentrate to the feed (see Figure 2.6). During the up-concentration tests, regular flux measurements were taken, temperature and pH of the feed were monitored and samples of the permeate and concentrate were collected and analyzed for CODt , CODs , TAN, Kj-N, Pt , Ps .

66

Figure 2.7: Conventional activated sludge setup.

2.2.5

Conventional biological adsorption/flocculation

A conventional activated sludge set-up was constructed (see figure 2.7). The reactor was inoculated with activated sludge from the Ossemeersen WWTP and run at an SRT of 14 days before starting the experiment. Both the aeration basin and the sedimentation cone had a volume of 1 L. The influent flow rate was 1.45 L/h. For practical reasons, the concentrate was withdrawn from the CSTR tank, at a rate of 40 mL per hour. As such, the reactor was operated at an HRT (in the aeration tank) of 0.7 h and an SRT of 1 d. The HRT in the sedimentation tank was 0.7 h. The dissolved oxygen level in the aeration basin was maintained at 3.5 mg O2 /L. The supplied air-liquid ratio was 8:1. The set-up was run for 20 consecutive days. Measurements commenced after 3 SRTs had passed in order to allow the activated sludge to become adapted to the 1 d SRT. Influent and effluent samples were taken at regular intervals and analyzed for CODt , CODs , TAN and Kj-N. 2.2.6

Biological adsorption/flocculation in membrane bioreactor

A Kubota-type immersed MBR was used (see figure 2.8). The aeration basin had a volume of 7 L and contained three membrane frames, to give a total membrane area of 0.3 m2 , or 43 m2 /m3 . The membranes were made 67

Figure 2.8: Membrane bioreactor setup.

of chlorinated polyethylene with a pore diameter of 0.4 µm. The reactor was inoculated with activated sludge from Ossemeersen WWTP. The wastewater was stored in 500 L vessels and replaced approximately every 4 days. It was mixed to prevent sedimentation and provide an influent of relatively constant quality to the MBR. Filtration of the influent occurred by a applying an underpressure of approximately 0.2 bar. 10 L of air per minute was introduced through bubble aeration at the bottom of the MBR to scour the membrane and provide oxygen to the biomass. To further counter fouling, a filtration/relaxation regime of 8/2’ was used, with an additional 10’ of relaxation every hour. Flux measurements were performed daily. The membrane setup was non-backwashable. Following an initial period of 6 days wherein no cleaning of the membranes was done, they were removed and rinsed with tapwater manually once per day for the remainder of the test. The washing water was sampled to determine the COD content of the fouling cake. HRT was determined by the flux through the membrane and amounted to 1.6±0.2 h. Instantaneous samples of influent and permeate were taken daily and analyzed − 3− for CODt , CODs , TAN, Kj-N, NO− 3 , NO2 and PO4 . TSS was determined once. The MBR was operated at SRT = 1.2 d for a period of 12 days and at SRT = 2.8

68

days for a period of 15 days.. The concentrate was obtained by pumping mixed liquor from the aeration basin into a receptacle. TSS and VSS of both the mixed liquor in the aeration tank and of that in the concentrate receptacle were analyzed daily, and the Pss occasionally. The SVI of the concentrate was measured daily to asses how well it could be further up-concentrated by means of sedimentation. The supernatant and settled concentrate sludge were analyzed four times for solids content, COD, N and P. At the end of the experiment (at SRT = 2.8 d), a coagulation/flocculation for enhanced sedimentation of the concentrate was attempted, as well as to gain some insight into the bioflocculation process. Three treatments were tested: 1. A dose of 100 mg FeCl3 /L and 1 mg/L of Betz 1100P (weak anionic) flocculant. 2. A dose of 103 mg Al2 (SO4 )3 /L and 1 mg/L of Betz 1100P (weak anionic) flocculant. 3. A dose of 1 mg/L of Betz 1100P (weak anionic) flocculant, without coagulant addition.

2.2.7

Reverse Osmosis

Permeate from the MBR was subjected to RO treatment. An RO test unit with a total membrane area 0.36 m2 and a maximum TMP of 10 bar was used. Conduc− 3 tivity, CODt , NO− 3 , NO2 , PO4 − and TAN of the permeate and concentrate were analyzed. 2.2.8

Anaerobic digestion batch tests

Anaerobic digestion tests were executed in 1L erlenmeyers, connected to biogas collection columns. The test were executed at both mesophilic (34◦ C) and thermophilic conditions (54 ◦ C). Inoculation was done with mesophilic and thermophilic anaerobic digester sludge obtained from Aquafin, resp. Avecom. A dose of 4,2 g/L NaHCO3 as pH buffer was administered at the start of the experiments. Feeding was performed 3 times per week, during which pH was monitored. The organic loading rate (kg COD/m3 d) was increased gradually at start-up. Gas composition analysis was performed (and corrected for headspace volume) once per week, as was the analysis of TS, VS, TSS, VSS, CODt , CODs , VFA, Kj-N, 69

Figure 2.9: AD batch reactors.

NH4 -N and Pt . Ps , Ripley index were analyzed occasionally. CST was determined at the end of the experiments. Substrates

Two substrates were used for the anaerobic digestion tests were thickened activated sludge from the adsorption/bioflocculation reactor (further on referred to as ’adsorptive sludge’) of a full scale WWTP of the Waterschap Brabantse Delta in Breda, the Netherlands, which operates according to the AB principle. Sludge was collected on four different occasions and kept in cold storage during the tests. The second type of feed was centrifuge concentrate. Concentrates of varying solids contents, obtained by centrifuging wastewater from Ossemeersen WWTP, were used. Completely mixed batch tests

For the CSTR batch tests, tests were performed on: 1. The adsorptive activated sludge from Breda WWTP, at HRT (=SRT) of 15 and 30 days 2. Centrifuge concentrate, for which the HRT was adapted to accomodate the varying COD content of the concentrates obtained from different centrifugation tests, as will be described in the Results chapter. The reactor content was well mixed prior to removing effluent from the reactor and feeding it. The reactor content was again well mixed after feeding. The reactors contained 1L of sludge to minimize headspace.

70

Temperature phased AD batch tests

In the TPAD tests, adsorptive sludge was again used as the feed. It was preheated for half an hour at 54 ◦ C before being fed to the thermophilic reactor, which contained 0.5 L of sludge. The effluent from the thermophilic reactor was used to immediately feed the mesophilic reactor, which contained 1 L of sludge. Both reactors were operated in fully mixed mode, as described in the previous paragraph. The HRT (=SRT) in the thermophilic reactor and mesophilic reactor was 4 days, resp. 15 days. The biogas from both reactors was collected and analyzed seperately. In the first stage of this test, no pH correction was performed. In the second stage, the pH in the thermophilic reactor was maintained above 6.8 by adding NaOH. Anaerobic sequencing batch tests

In the ASBR tests, adsorptive sludge was used. The feeding regime was as follows: 1. Decanting of all the supernatant in the reactor into an Imhoff cone. 2. After a sedimentation period of 30’, a certain volume of supernatant - so that the HRT would be 15 days - was decanted from the Imhoff cone to serve as the effluent. 3. In case insufficient supernatant was available to attain the required HRT, the remaining effluent volume was made up by removing settled sludge from the reactor, after having mixed it well. 4. The reactor was fed with adsorption/bioflocculation activated sludge and mixed well. No mixing was performed in between feedings. In case the effluent consisted of both supernatant (’referred to as overflow’) and settled sludge, both fractions of the analyzed seperately.

71

Chapter 3

Results 3.1 3.1.1

Chemically enhanced primary treatment flocculant selection

The preliminary test for flocculant selection assessed the suitability of four different polyelectrolyte flocculants: Catflock P1556, a medium cationic PE; Zetag 72, a weak cationic PE; Betz 1100P, a weak anionic PE and M156, a strong anionic PE. The coagulant dose for this test was selected based on visual inspection with the criterium being the lowest dose which resulted in a clear liquid in between the visible suspended solids, regardless of whether these settled or not (Figure 3.1). On this basis, a dosage of 40 mg FeCl3 /L was selected (sample 3 in Figure 3.1). Upon addition of this dose, the pH had only dropped from 7.5 to 7.2 so no addition of NaOH was required. Using the selected coagulant dose, addition of 0.5 mg/L of the different flocculants resulted in insufficient flocculation, whereas acceptable floc formation was achieved at a dose of 1 mg PE/L (Figure 3.2). No significant differences of the supernatant COD between the treatments with only coagulant and with the different flocculants was detected, all achieving 80±10 mg/L, whereas sedimentation with no coagulant or flocculant resulted in a supernatant containing 145 mg COD/L (Figure 3.3). Based on floc size and settling speed, the weak anionic Betz 1100P was selected as the most suitable flocculant.

72

Figure 3.1: Selection of coagulant dose prior to flocculant addition. Setted wastewater samples after coagulant addition. Samples 2, 3, 4, 5 and 6 represent FeCl3 doses of 20, 40, 80, 150 and 200 mg FeCl3 respectively.

Figure 3.2: Flocculant selection. Settled wastewater after coagulant and 1 mg/L flocculant addition. From lef to right are shown: control 1 (no coagulant or flocculant), control 2 (no flocculant), medium cationic, weak cationic, weak anionic and strong anionic PE.

Figure 3.3: Flocculant selection. COD of the supernatant of settled wastewater after coagulant and flocculant addition.

73

Figure 3.4: Effect of coagulant dose on the turbidity of settled wastewater. Flocculant dose if applied was 1 mg PE/L.

3.1.2

Coagulant dose selection

The wastewater used had a CODs/CODt ratio of 25%. The effect of FeCl3 and Al2 (SO4 )3 dosage on the turbidity and on the COD of the settled wastewater is shown in Figure 3.4, resp. Figure 3.5. The control treatment, i.e. primary sedimentation, resulted in a turbidity decrease of 24%, from 101 NTU to 77 NTU, and a COD removal of 35%, from an initial COD of 480 mg/L. For both FeCl3 and Al2 (SO4 )3 , the decrease of both turbidity and COD with increasing coagulant dose slowed down significantly above 100 mg/L, which corresponds to 0.62 mmol Fe/L and 0.60 mmol Al/L. For both coagulants, the achieved turbidity removal at these dosages was sufficient to reach 15 NTU in the supernatant, which would render it suitable to conventional crossflow UF treatment. For these reasons, these dosages were selected as optimal. The turbidity removal efficiency at these selected dosages, was 87% (down to 13 NTU) and COD removal was 73% (down to 130 mg COD/L from the initial 480 mg COD/L) for FeCl3 + PE and 84% (down to 15 NTU), respectively 78% (down to 106 mg COD/L) for Al2 (SO4 )3 + PE. As such, more than triple the turbidity removal and more than double the COD removal was achieved compared to primary settlement. Almost complete turbidity removal (99%) was only reached at a dose of 200 mg FeCl3 . On average, FeCl3 + PE treatment resulted in 7±2% 74

Figure 3.5: Effect of coagulant dose on the COD of settled wastewater. Flocculant dose was 1 mg PE/L.

better turbidity removal than Al2 (SO4 )3 + PE treatment. The opposite was true for COD removal, with Al2 (SO4 )3 + PE outperforming FeCl3 + PE by 10±3%. A small part of the removed COD was soluble COD, but CODs removal increased much less slowly with coagulant dose than total COD removal did (not shown). CODs removal reached a maximum of 22% (from 126 mg/L to 98 mg/L) at a dose of 80 mg FeCl3 /L for FeCl3 + PE treatment and reached 15% (from 126 mg/L to 107 mg/L) at a dose of 100 mg Al2 (SO4 )3 /L for Al2 (SO4 )3 + PE treatment. Increasing the coagulant doses further did not significantly improve the CODs removal. At the dose of 100 mg FeCl3 , 91% of particulate COD was removed whereas at 200 mg FeCl3 , 97% of the particulate was removed. At the selected doses, the COD of the effluent thus consisted mostly of CODs. The above illustrates the maximum achievable COD removal of CEPT treatment for a wastewater of this composition, which was calculated at approximately 80%. The addition of PE flocculant (1 mg/L) on average improved the turbidity removal by 5±1%, but it’s main contribution lay in speeding up the settling process, with sedimentation being completed after 5’ in contrast to the treatment with coagulants only, which needed up to 30’ to settle. Compared to primary settling, the use of flocculant without coagulant improved COD removal by only 10%. 75

Figure 3.6: Effect of coagulant dose on the PO4 3 − concentration of settled wastewater. Flocculant dose if applied was 1 mg PE/L.

PO4 3− removal was complete (dropping below the detection limit of 0.2 mg P/L) from a dose of 60 mg FeCl3 /L upward and from a dose of 62 mg Al2 (SO4 )3 /L upward, the initial concentration having been 5.0 ± 0.2 mg/L (Figure 3.6). The pH remained above 6.5 for all dosages of both FeCl3 and Al2 (SO4 )3 up to 100 mg/L, the initial pH of the wastewater having been 7.3. 3.1.3

Upconcentration test

The CEPT upconcentration tests were performed with the previously selected doses of 100 mg/L of both FeCl3 and Al2 (SO4 )3 coagulant and the weak anionic PE flocculant Betz 1100P at a dose of 1 mg PE/L. The control was simple primary sedimentation. Coagulant only and coagulant plus flocculant tests were performed. The pH was 6.6, resp 6.9 after addition of FeCl3 , resp. Al2 (SO4 )3 , down from 7.4. The wastewater had a slightly different composition than in the previous test, with a CODs/CODt ratio of 20%, respectively 14% for the tests with primary sedimentation and FeCl3 , respectively Al2 (SO4 )3 . Floating sludge was not observed after any of the treatments. The speed of sedimentation of the different treatments is reflected in Figure 3.7. The results show that the top water layer reaches a steady turbidity level within 5’ after CEPT treatment, whereas the turbidity dur76

Figure 3.7: Evolution of supernatant (top layer) turbidity with settling time for primary settling, treatment with 100 mg FeCl3 /L + 1 mg PE/L and treatment with 100 mg Al2 (SO4 )3 + 1 mg PE/L.

ing primary treatment could not be said to have reached a steady state after 60’. The volume of concentrate produced became steady after 30’ for primary treatment and was relatively small (8.3±0.4 mL/L) but was highly concentrated (26.5 g COD/L), whereas the CEPT treatment resulted in larger volumes of less concentrated sludge: 39±1 mL/L with a concentration of 9.5 mg COD/L for treatment with FeCl3 + PE and 36±1 mL/L with a concentration of 11.1 mg COD/L for treatment with Al2 (SO4 )3 + PE. The FeCl3 + PE treatment concentrate was found to compress during the sedimentation period: from 53±1 mL/L after 5’ to 39±1 mL/L after 60’, whereas no significant change in the volume of the concentrate formed by Al2 (SO4 )3 + PE treatment was observed after 5’. The VS/TS ratio, i.e. the organic fraction, of the FeCl3 + PE, resp.Al2 (SO4 )3 + PE concentrates was 62, resp. 66%, which was lower than that of primary concentrate (73%) for all treatments,. In the raw wastewater, the VS/TS had been 50%. The COD/VS ratio of the concentrates was 1.5 to 1.6 g COD/(g VS). The recovery of resources (water, COD, Kj-N and Pt ) achieved by primary sedimentation and CEPT treatment is shown in Figure 3.8. The separation by CEPT treatment resulted in a clear water fraction whose volume was 96% of the initial wastewater volume. The recovery of COD into the concentrate, which in the case

77

Figure 3.8: The recovery of clean water volume in the effluent and of COD, Kj-N and Pt in the concentrate after primary sedimentation, resp. CEPT treatment with 100 mg FeCl3 /L, 100 mg FeCl3 /L + 1 mg PE/L, 100 mg Al2 (SO4 )3 and 100 mg Al2 (SO4 )3 + 1 mg PE/L. The numbers above the bars represent the concentrations of COD (in g/L), Kj-N (mg N/L) and Pt (mg P/L) of the obtained concentrates.

of CEPT treatment equals the removal efficiency of the process, was 84±4% for FeCl3 + PE and 87 ±1% for Al2 (SO4 )3 + PE treatment, about twice as much as was recovered through primary sedimentation. The use of PE improved the COD recovery by 6 to 10%. The quality of the influent and treatment effluents is shown in Table 3.1. The effluent contained 58, 56, resp. 204 mg COD/L from FeCl3 + PE, Al2 (SO4 )3 + PE, resp. primary sedimentation treatment. CODs removal was 20%, 30%, resp. 4%. 31±2%, resp. 44±1% of Kjeldahl nitrogen was recovered by FeCl3 + PE, resp. Al2 (SO4 )3 + PE treatment, whereas primary treatment only recovered 13±1%. The effluent contained 43, 34, resp 19 mg Kj-N/L, of which approximately 22 mg NH+ 4 -N/L, which is not exprected to have been removed in any significant amount (Table 3.1). Nitrate was not significantly removed either. The total phosphorus content of the initial wastewater was 8.8 mg P/L, of which 4.0 mg P/L was dissolved PO3− (Table 3.1). 46±17%, resp. 61±1% of total 4 phosphorus was recovered by FeCl3 + PE, resp. Al2 (SO4 )3 + PE treatment, 78

Table 3.1: Effluent quality after primary sedimentation and CEPT treatment

CODt

CODs

Kj-N

NO− 3

Pt

PO3− 4

SO2− 4

(mg/L)

(mg/L)

(mg N/L)

(mg N/L)

(mg P/L)

(mg P/L)

(mg SO2− 4 /L)

Influent Prim. sedim. FeCl3 FeCl3 + PE

315 204 90 58

64 61 51 51

49 43 37 34

3.5 2.8 2.9 2.9

8.8 7.5 1.3 0.6

4.0 3.9 BDL BDL

41 41 41 38

Influent Al2 (SO4 )3 Al2 (SO4 )3 + PE

420 76 56

60 44 40

34 18 19

2.9 2.9 2.9

9.0 4.1 3.7

4.0 BDL BDL

43 95 77

whereas primary treatment only recovered 16±1%. Dissolved PO3− 4 was completely removed by CEPT treatments. The use of 100 mg Al2 (SO4 )3 /L increased the dissolved sulphate content from 8.6 to 15.4 mg SO2− 4 .

3.2

Centrifugation

Centrifuge treatment of municipal wastewaters of different strengths was performed at different speeds, ranging from 200 to 1000 L/h. No clear influence of the speed on the removal/recovery efficiency could be discerned. Nor did the treatment speed significanty affect the energy consumption. The COD removal efficieny of centrifugation at 1000 L/h showed a linear dependance on the fraction of particulate COD (Figure 3.9), illustrating the inability of centrifuge to remove small and dissolved compounds. In general, for the wastewaters tested, those with higher total COD concentrations tended to also have a higher fraction of particulate COD, and as such higher removal efficiencies were achieved for more concentrated wastewaters (Figure 3.10). On average, CODp removal was 86±11%, indicating that a fraction of presumably colloidal particulate material could not be removed. The effluent quality was fairly constant, at 97±22 mg CODt/L, of which 65±21% was dissolved COD. In an attempt to improve the removal of colloidal particles by neutralizing their surface charge, FeCl3 coagulant was added prior to centrifugation. The effect on the removal efficiency of CODp was however negative, with a CODp removal ef79

Figure 3.9: CODt removal efficiency as a function of the particulate fraction of the COD, achieved by centrifugation at 1000 L/h of municipal wastewaters of varying quality. The removal efficiency was more or less proportional to the fraction of particulate COD (R2 =0.81).

Figure 3.10: CODt removal efficiency as a function of the influent CODt, achieved by centrifugation at 1000 L/h of municipal wastewaters of varying quality.

80

Figure 3.11: CODt, CODp, Kj-N and Pt removal efficiency achieved by centrifugation at 1000 L/h of municipal wastewater, with and without the use of FeCl3 . The coagulant dose represented here was 150 mg FeCl3 /L. The higher removal of CODt in the case of treatment with coagulant is not due to the effect of the coagulant, but rather due to a higher proportion of particulate COD in the influent (CODp/CODt= 96%).

ficiency of 57% at a dose of 30 mg FeCl3 /L, 68% at 40 mg FeCl3 /L, and only reaching a removal efficiency similar to the one achieved without coagulants at a dose of 150 mg FeCl3 /L. No significant effect could be observed on the removal efficiency of Kjeldahl nitrogen, which was 17±7%, the effluent thus still containing 26.7 mg Kj-N/L. This was not unexpected, since 76% of the Kj-N in the influent was present as dissolved NH+ 4. Centrifugation allowed to produce very highly concentrated concentrates, the most concentrated of those produced in the tests containing 126 g COD/L, 5.1 g Kj-N/L, 0.87 g Pt /L and a solids (TS) content of 7.3%.

3.3 3.3.1

Primary membrane filtration Clean water flux

The clean water flux of the polyamide-6 fuzzy electrospun nanofibre membranes was measured in the FMX unit. The 1 µm cut-off membrane achieved a maxi-

81

Figure 3.12: Clean water flux as a function of trans-membrane pressure, with and without vortex generation. Membrane cut-off: 1 µm.

mum clean water flux of 1520 L/(m2 ·h) at a trans-membrane pressure (TMP) of 2 bar when the FMX was operated with the vortex generator switched on (Figure 3.12). Increasing TMP above 2 bar did not improve the clean water flux. When operating the unit without vortex generation, the clean water flux was found to increase linearly (R2 =0.99) with TMP within the tested range of TMPs (1 to 4 bar), achieving a clean water flux of 1465 L/(m2 ·h) at 4 bar (Figure 3.12). As such, it appeared that the effect of vortex generation is not limited to fouling control but, within the tested pressure range, also positively affected the flux of clean water, increasing it by some 40% at 2 bar. The clean water flux of the 0.5 µm cut-off membrane was substantially higher, peaking at 7870 L/(m2 ·h) at 2 bar TMP when vortex generation was in operation (not shown). This was presumably due to a higher porosity of this membrane, which according to its producer was between 80 and 90%. The clean water flux of the 1µm membrane was tested again after having performed upconcentration tests with 61 L municipal wastewater (see following sections). It was found that at 3 bar TMP, the clean water flux had declined by 90%, indicating substantial membrane fouling. 3.3.2

Pressure selection test

Wastewater was filtered in a closed circuit setup (i.e. recycling of both permeate and rejectate) to asses the optimal TMP. At the maximum achievable TMP (5 82

Table 3.2: Effluent quality after FMX filtration of municipal wastewater.

cut-off

CODt

CODs

Kj-N

TAN

Pt

PO3− 4 -P

(µm)

(mg/L)

(mg/L)

(mg N/L)

(mg N/L)

(mg P/L)

(mg P/L)

61±2 55±3 10±5%

37±3 27±2 26±5%

29±2 27±1 7±2%

4.7±0.5 2.4±0.6 53±13%

3.0±0.1 2.2±0.3 25±9%

0.5

influent effluent removal

99±8 35±5 65±5%

1

influent effluent removal

224±10 70±4 67±4%

bar), the maximum flux had not yet been reached (not shown). Due to influent pump malfunction occurring at TMP > 3 bar, and under the consideration that operation at maximal flux would speed up irreversible fouling of the membranes, a TMP of 3 bar was selected to perform the upconcentration tests. 3.3.3

Upconcentration test

Upconcentration tests were performed in a half-open circuit, i.e. with the recycling of the rejectate, to achieve an as highly concentrated concentrate as possible. The characteristics of the influent, effluent and removal efficiencies (calculated on the basis of the permeate quality) are listed in Table 3.2. It can be observed that the COD removal efficiencies of both membranes were not significantly different, despite the difference in cut-off. This is presumably due to the difference in influent composition, as the influent during treatment with the 0.5 µm membrane had a very low COD concentration and can be thus be expected to have had a relatively high dissolved COD fraction (see Section 3.2), which would not have been removed to a high extent by the membrane. The COD and P in the effluent of the 1 µm cut-off membrane consisted for 80, resp. 90% of dissolved compounds. An apparently significant removal of dissolved PO3− 4 can nevertheless be remarked upon. Almost all of the nitrogen in the effluent consisted of dissolved ammoniacal nitrogen. The permeate fluxes achieved were 44±3 and 24 ±2 L/(m2 ·h), for the membranes with cut-off 1 µm, resp. 0.5 µm, which is 3%, resp. 1% of the initial clean water

83

Figure 3.13: Wastewater filtration permeate flux as a function of time during halfopen circuit operation with vortex generation, for membrane cut-offs 0.5 and 1 µm.

fluxes and - in the case of the 1 µm membrane - some 30% of the clean water flux achieved post upconcentration test (Figure 3.13). This suggests that, roughly speaking, concentration-polarization reduced the flux to an estimated third of the clean water flux and that fouling inside the membrane caused another estimated ten-fold drop in flux. As fouling was occurring during continuous operation, during the first hour of operation the flux through the membrane with the smaller cut-off could be seen to drop some 30% from its initial value of 40 L/(m2 ·h) and a further 10% until the end of the test after 28 h, despite the fact that the concentrate never achieved a very high strength (see further). The flux through the membrane with the larger cut-off, on the other hand, did not change significantly through the 15 h of operation, although this may partly be due to the membrane having already undergone some fouling prior to this test. The progression of the separation process is illustrated in figures 3.14 and 3.15 by plotting the achieved recovery (of the volume of water in the form of permeate and of the mass of COD, N and P in the concentrate) as a function of the energy which would be consumed by the filtration process in order to reach that level of recovery if the FMX unit would be outfitted with the maximum possible membrane area. The energy consumption of the FMX unit is calculated based upon the measured

84

Figure 3.14: Fraction of resources recovered by FMX filtration as a function of the amount of used energy per volume of influent. Numbers to the right: concentrations of COD in the influent, resp. in the concentrate, obtained after separating the influent into filtered water and a concentrate whose volumes were 81%, resp. 19% of the initial influent volume (61 L). Membrane cut-off = 0.5 µm. TMP = 3 bar.

power of 4940 W. It is assumed that the addition of extra membrane modules, up to a maximum membrane area of 3.15 m2 would not significantly alter the power consumption of the FMX unit operating at full capacity. It can be seen that 31 kWh/m3 was required to achieve a permeate volume which was 96% of the initial wastewater volume, whereas already 46 kWh/m3 was required to reach a water recovery of only 81%, during filtration at membrane cut-off of 1 µm, resp. 0.5 µm. The concentration of COD in the concentrate at 0.5 µm cut-off membrane was very low (0.3 g COD/L), mostly due to a low strength influent with a high fraction of dissolved COD. It must be oberved that the mass balance could not be closed to satisfaction, as the difference in removal efficiency (based on the permeate quality, see Table 3.2) and recovery (based on the concentrate quality, see figures 3.14 and 3.15) illustrates. The presence of some of the contaminants in the fouling layer can only partly explain this. The achieved recovery of resources in the concentrate produced by the FMX filtration is summarized in Figure 3.16. It can be noted that the concentrates did 85

Figure 3.15: Fraction of resources recovered by FMX filtration as a function of the amount of used energy per volume of influent. Data in the box: influent quality. The numbers to the right: concentrations of COD, P and Kj-N in the concentrate, obtained after separating the influent into filtered water and a concentrate whose volumes were 96%, resp. 4% (3.3 L) of the initial influent volume (61 L). Membrane cut-off = 1 µm. TMP = 3 bar.

86

Figure 3.16: Recovery of resources by FMX upconcentration at a consumed energy of 31 and 46 kWh/m3 for cut-off = 1 µm, resp. 0.5 µm. TMP = 3 bar. Numbers above the bars represent the concentrations of COD (g/L), CODp (g/L), Kj-N (mg N/L), Norg (mg N/L), Pt (mg P/L) and PSS (mg P/L) in the obtained concentrates.

not have a high concentration of COD, N or P, despite the considerable amount of energy expended in producing them. The COD, N and P was upconcentrated by factors 7, 3 and 3.5, respectively. 3.3.4

Intermittent filtration

In an attempt to optimize the energy efficiency of the filtration process, an intermittent mode of vortex generation was attempted, with a 5’ on/5’ off cycle (Figure 3.17). The unit was operated in a closed circuit set-up for this test. A sharp decrease of 50% in the permeate flux was observed during the first two cycles, after which the flux stabilized and did not significantly recover during the intervals when the vortex generation was functioning, not even during an extended (25’) interval. This indicates that the quick flux decline in this intermittent mode of operation was due to irreversible fouling of the membrane which could not be remediated by the generation of vortices. The steady flux which was achieved during intermittent operation (39±5 L/(m2 ·h)), was however only 16% lower than the flux achieved with continuous vortex operation during the upconcentration test.

87

Figure 3.17: Flux evolution over time during intermittent vortex generation (5’ on/5’ off ). Membrane cut-off = 1 µm, TMP = 2 bar. Table 3.3: bio-adsorption and sedimentation: influent and effluent quality and removal efficiency.

influent effluent removal

3.4

CODt

CODs

Kj-N

TAN

(mg/L)

(mg/L)

(mg N/L)

(mg N/L)

175±29 86±9 50±7%

67±10 60±10 10±4%

36±2 26±4 27±8%

26±3 20±2 23±8%

Bio-adsorption and sedimentation

A small indicative test was performed to get an indication of the performance of the bio-adsorption process at lab scale. Run at an HRT of 0.7 h and an SRT of 1 d, the bio-adsorption set-up with sedimentation achieved the removal efficiencies listed in Table 3.3. Total COD removal was 50±7%. This value is on the low side of the COD removal typically observed in full-scale reactors, which may partly be due to edge effects in the reactor, since a significant amount of biomass tended to cling to the reactor vessel walls rather than remain in suspension. A small CODs removal of 10±4% can be remarked upon, which is only some 5% better than what was observed 88

Figure 3.18: CODt and CODs removal by bio-adsorption reactor with sedimentation. SRT = 1 d.

during primary sedimentation at an equivalent HRT (see Table 3.1). Particulate COD removal was 76±10%. The day by day evolution of influent and effluent COD is shown in Figure 3.18. It indicates that peaks in load can be balanced out by this process, resulting in a relatively constant effluent quality. A relatively high dissolved TAN removal can be remarked upon (23±8%). Nitrification is not plausible at such a low SRT, suggesting that some NH+ 4 was removed through adsorption by the sludge.

3.5 3.5.1

Bio-adsorption in a membrane bioreactor Loading rate and biomass concentration

The reactor was first operated at SRT = 1.2 ± 0.3 d and subsequently at SRT = 2.8 ± 0.6 d (Figure 3.19). The volumetric and organic loading rates (Bv and Bx ) to which the reactor were submitted are shown in Figure 3.20. It can be observed that at SRT = 1.2 d, the organic loading rate reacted quickly to the variations in volumetric loading rate (which themselves were due to the variations in influent strength, see also Figure 3.22), whereas at SRT = 2.8 d, Bx was less sensitive to changes in Bv , as the biomass concentration in the reactor was able to increase in response to increased volumetric loadings. At SRT = 1.2, average Bv was 2.4±0.9 89

Figure 3.19: Evolution of applied SRT and subsequent HRT and permeate flux in the MBR. The dotted line marks the start of the daily membrane washing regime.

kg COD/(m3 ·d) (or 2.1 ±0.8 kg bCOD/(m3 ·d)) and Bx was 14±7 kg COD/(kg MLVSS·d) (or 13 ±6 kg bCOD/(kg MLVSS·d)). The MLVSS at SRT = 1.2 d appeared to be gradually decreasing, down to approximately 1 g MLVSS/L at the end of the run. At SRT = 2.8 d, average Bv was 3.0±1.2 kg COD/(m3 ·d) (or 2.7 ±0.1.1 kg bCOD/(m3 ·d)) and Bx was 12±4 kg COD/(kg MLVSS·d) (or 11 ±3 kg bCOD/(kg MLVSS·d)). At SRT = 2.8 d, the variations of the MLVSS to the volumetric loading rate resulted in it varying between 2 and 5 g MLVSS/L. 3.5.2

Flux and fouling remediation

To counter fouling, a regime of 8’ filtration and 2’ relaxation was applied, with an additional 10’ relaxation every hour, thus performing filtration for a total of 40’/h. This was found to perform slightly better than a regime of 5’/5’, filtering a total 30’/h. Course bubbling aeration at 10L/min occurred continuously. Excessive foaming occurred when higher aerations rates were attempted. After 6 days, a membrane washing regime was started to further remedy fouling. The membranes were removed from the reactor and rinsed with tap water with the use of a soft sponge. The effect of these measures on the flux is illustrated in Figure 3.21. Measured at the end of a one day period of filtration without membrane washing, over a one hour period following a 10’ relaxation, the average flux over 90

(a) VSS

(b) Bv and Bx

Figure 3.20: VSS in mixed liquor and concentrate and volumetric loading rate (Bv ) and sludge loading rate (Bx ) of bio-adsorption MBR operated on municipal wastewater at SRT ≈ 1.2 and 2.8 d.

one filtration cycle of 10’ was seen to have dropped from 19 to 16 L/(m2 ·h). The membrane washing restored the flux back to 21 L/(m2 ·h). During the period when no membrane washing was applied and SRT = 1.2 d, the flux could be seen to drop gradually during 6 days by 45% from a daily average of 20 L/(m2 ·h) to 11 L/(m2 ·h) (Figure 3.19). After intiating a daily membrane cleaning regime, the daily flux average recovered to 16±1 L/(m2 ·h). When SRT was 2.8 d and membrane washing was applied, the flux could be seen to drop during 16 days by 30% from an initial daily average of 17 L/(m2 ·h) to 12±2 L/(m2 ·h). Average HRT, which depends directly on the flux, was 1.4±0.1 h at SRT = 1.2 d and 1.6±0.2 h at SRT = 2.8 d (Figure 3.19).

3.5.3

COD removal

The removal of CODt and CODs is shown in Figure 3.22. It can be observed that despite the variations in the influent strength, a relatively constant permeate COD quality of 43±8 mg COD/L was obtained, regardless of the SRT. Over the whole run, the COD removal efficiency at SRT = 1.2 d was 74% and was 80% at SRT = 2.8 d. Considering the membrane cut-off of 0.4 µm, it is not surprising that all of the COD in the permeate was dissolved COD. It can be remarked that at SRT = 1.2 d, the CODs in the effluent was higher than that in the influent, 91

Figure 3.21: Effect of relaxation and membrane washing on the permeate flux. (SRT = 2.8d.)

indicating that solubilization of particulate COD occurring in the reactor at a rate which was faster than the metabolization of CODs by the biomass. The opposite was observed at SRT = 2.8 d, where CODs removal could be observed, but apparently only during periods of biomass concentration growth in response to increased volumetric loading (figures 3.20 and 3.22). 3.5.4

Nitrogen removal

The removal of nitrogen is illustrated in Figure 3.23. It can be observed that despite the variations in the influent strength, a relatively constant permeate quality of 30±3 mg total N/L was obtained, regardless of the SRT. Considering the membrane cut-off of 0.4 µm, it is not surprising that almost all of the nitrogen in the effluent was present as either ammoniacal nitrogen, nitrate or nitrite, all dissolved forms of nitrogen. Over the whole run, the removal of total nitrogen at SRT = 1.2 d was 24% of the combined influent load and was only 4% at SRT = 2.8. The ammonification of organic nitrogen retained by the membrane to dissolved TAN can be assumed to have lead to the low total nitrogen removal at this SRT. The nitrogen fractions in the influent and permeate are shown in Figure 3.24. Significant nitrification can be observed to have occurred at SRT = 2.8 d, contributing to a daily average TAN removal of 83±15% and, equally, a total removal 92

(a) CODt

(b) CODs

Figure 3.22: COD removal by bio-adsorption MBR operated at SRT ≈ 1.2 and 2.8 d.

(a) Total N

(b) TAN

(c) TON

Figure 3.23: Nitrogen removal by bio-adsorption MBR operated at SRT ≈ 1.2 and 2.8 d. Total nitrogen includes ammoniacal nitrogen (TAN), organic nitrogen, nitrates and nitrites.

93

Figure 3.24: Fractions of organic, ammoniacal, nitrate and nitrite nitrogen in influent and effluent at SRT = 1.2 and 2.8 d.

of 85% of the TAN load over the whole test run. It could be calculated that at SRT = 2.8 d, 85% of the Kj-N influent load ultimately got converted (through − ammonification and nitrification) to nitrite and nitrate, with an NO− 2 -N:NO3 -N ratio in the permeate of 0.9:1, indicating that nitrification was by no means complete. At the shorter sludge residence time, it was surprising that approximately 60% of the TAN influent load was still removed, and 30% of the Kj-N influent load − got converted to oxidized nitrogen. The NO− 2 -N:NO3 -N ratio of the permeate was higher still, at 1.7:1. 3.5.5

Phosphorus removal

The removal of phosphorus is illustrated in Figure 3.25. It can be observed that at SRT = 1.2 d, despite the PO3− 4 concentration in the influent being relatively constant, the variations in the phophorus content of the effluent followed those of the influent, indicating that a substantial amount of phosphorus may have been present in another soluble form that was not retained by the membrane. At SRT = 2.8 d, the effluent had a relatively constant phosphorus content of 2.9±0.2 mg P/L. At both SRTs, the phosphorus removal efficiency was 40%, both on a daily average (43±16%) as well as based on the total phosphorus load over the test run. It appears likely that the phosphorus removal is mostly due to retention by membrane (cake), since no significant phosphorus removal is observed in bio94

(b) PO3− 4

(a) Total P

Figure 3.25: Phosphorus removal by bio-adsorption MBR operated at SRT ≈ 1.2 and 2.8 d.

adsorption units functioning with a sedimentation tank. 3.5.6

Recovery

A concentrate was produced by wasting of the mixed liquor from the reactor. This concentrate could be further upconcentrated by thickening through sedimentation, with an SVI of 97±21 after 30’ sedimentation. Sludge bulking was observed during thickening on 3 out of 20 test days. There were thus five fates which COD could undergo: into the permeate, the fouling cake, the thickened concentrate (or occasionaly a bulking sludge layer, not considered here), the supernatant of the thickened concentrate or mineralized into CO2 . For simplicity’s purpose, the COD of the fouling cake (which was 13, resp 12% of the total COD at SRT = 1.2, resp 2.8 d) is included into the recovery of thickened concentrate under the assumption that the COD of in the washing water could be recovered in a full-scale set-up. The amount of mineralization was calculated as the COD fraction required to close the mass balance, which is shown in Figure 3.26. It can be seen that at SRT = 1.2 d, with a 58% COD recovery the MBR achieved only a slightly higher COD recovery than primary sedimentation (44 %, at 1 h HRT), despite the influents in both cases having comparable particulate COD fractions (84, resp. 80%). The improved COD removal of the MBR at this SRT is thus mainly due to mineralization taking place (16% of COD). At SRT

95

Figure 3.26: COD mass balance at SRT = 1.2 and 2.8 d.

= 2.8 d, despite the good effluent quality, the COD recovery was only 33%, due to extensive mineralization (44%). It must be noted that the particulate fraction of the influent during this test at SRT = 2.8 d (65%) was less than in the other test, which may have aggravated mineralization. Figure 3.27 summarizes the recovery achieved by the MBR treatment followed by a 30’ thickening of the concentrate, compared to the recovery that was achieved by primary sedimentation. At SRT = 2.8 d, thickened concentrates were produced which contained 12.8±0.4 g COD/L, 814±102 mg N/L and 159±34 mg P/L. Prior to thickening, these concentrates contained 2.6±1.2 g COD/L (which can be considered insufficient for economic AD), 218±136 mg N/ and 51±22 mg P/L. The COD of the concentrate produced at SRT = 1.2 d was not significantly different than that at SRT = 2.8 d. 3.5.7

Reverse osmosis treatment of MBR efffluent

Since the effluent of the bio-adsorption MBR is free of any turbidity, it was possible to submit it to reverse osmosis. Effluent produced when the MBR was operating at SRT = 2.8 d, was used. The RO set-up was unfortunately not able to operate at TMPs higher than 10 bar, which limited the recovery (I.e. the ratio of permeate flow to feed flow) to to maximum 7% (Figure 3.28). The permeate flux reached 9 L/(m2 ·h) at that TMP. An idea of the quality of the RO permeate can nevertheless 96

Figure 3.27: Recovery of resources by AMBR treatment at SRT = 1.2 and 2.8 d followed by a thickening of the concentrate by sedimentation, compared to the recovery achieved by primary sedimentation. Numbers above the bars represent average concentrations of COD (g/L), total N (mg N/L) and total P (mg P/L) that were obtained in the thickened concentrates.

97

Figure 3.28: Permeate flux and recovery as a function of TMP for reverse osmosis treatment of MBR effluent. Table 3.4: Influent, effluent and removal efficiency by RO treatment of bioadsorption MBR effluent. The last column lists the EU drinking water standards.

NH+ 4 CODt Cl− NO− 3 NO− 2 PO3− 4 SO2− 4

(mg (mg (mg (mg (mg (mg (mg

N/L) COD/L) Cl/L) N/L) N/L) P/L) SO2− 4 )

Influent

Permeate

Removal efficiency (%)

DW standard

3.4-3.6 32-36 86-103 17-20 0.6-0.7 2.2 58-66

0.2 56 96 94 93 100 97

0.5 5 250 50 0.5 250

be gained by looking at the effluent produced at TMP = 10 bar (Table 3.4). It can be seen that for all the parameters tested, the permeate met the drinking water quality norms set out in EU directive 98/83/EC.

98

3.6 3.6.1

Anaerobic digestion Adsorptive sludge

Adsorptive sludge was obtained from the Breda WWTP, which operates according to the A/B principle. Sludge was taken on three occasions from the sludge recycle line of the A-reactor, which operates at an SRT of 0.7 d, an HRT of of 1.1 h, a Bx of 2.1 kg bCOD/(kg MLSS·d) and a Bv of 6.2 kg bCOD/m3 . The reactor is dosed with FeSO4 for phosphorus removal and as a result, the sludge can be expected to contain a maximum of 3% Fe on a dry weight basis (assuming all dosed iron is precipitated) in addition to the iron originating from the influent itself. The sludge was thickened before being fed to the anaerobic reactors. The characteristics of the thickened sludge can be found in Table 3.7. Three different types of reactors were tested: continuously stirred tank reactors (CSTR), anaerobic sequencing batch reactors (AnSBR), both at mesophilic as well as thermophilic conditions, and a temperature phased two-phase reactor (TPAD) consisting of a thermophilic pretreatment reactor and a mesophilic main reactor. The conditions under which the reactors were operated are listed in Table 3.5. Note that for the completely mixed reactor types, the SRT is by definition equal to the HRT. Note also that a ripley index smaller than 0.30 is an indication that the digester is operating optimally (Smits (2009)). The specific methane production rates are listed in Table 3.6 and compared in Figure 3.29. Overall, high methane production rates were achieved. A maximum of 301 mL CH4 /(g COD fed) was attained by the CSTR reactors operating at 30 d HRT under mesophilic conditions, with no significant difference between mesophilic and thermophilic digestion. This amounts to approximately 85% of what is normally considered the theoretical maximum of 350 mL CH4 /g COD for sludge digestion. The percentage of methane in the biogas was high overall. The methane content of the biogas produced in thermophilic reactors was observed to contain some 5% more methane than that of their mesophilic counterparts, but only for those reactors with an SRT of 30 d or higher. The effluent qualities and removal efficiencies are listed in Table 3.7. Ammonium concentrations were always around 1 g N/L, a level to which methanogens are thought to be able to adapt without inhibition. 99

Table 3.5: Operational conditions (HRT, SRT, OLR), pH, VFA alkalinity (Ripley index), and the timespan of the test runs performing anaerobic digestion of adsorptive sludge. For the TPAD reactor, entries containing two numbers refer to the thermophilic and mesophilic reactor separately. ’M’: mesophilic conditions. ’T’: thermophilic conditions. ’pH’: pH controlled by NaOH addition. ’no pH’: no pH control.

CSTR CSTR AnSBR TPAD

M T M T M T no pH pH

HRT

SRT

OLR

(d)

(d)

(g COD/(L·d))

30 30 15 15 15 15 4, 15 4, 15

30 30 15 15 53 38 4, 15 4, 15

1.2 1.2 2.4 2.4 2.4 2.4 2.0 2.0

pH

Ripley

run (d)

7.2 7.4 7.2 7.6 7.2 7.4 5.9, 7.1 7.0, 7.2

0.28 0.29 0.32 0.32 0.35 0.31 0.92, 0.31 0.66, 0.27

116 116 32 32 102 102 91 26

Table 3.6: Methane content of biogas and specific methane production rates of reactors performing anaerobic digestion of adsorptive sludge. For the TPAD reactor, entries containing two numbers refer to the thermophilic and mesophilic reactor separately. ’M’: mesophilic conditions. ’T’: thermophilic conditions. ’pH’: pH controlled by NaOH addition. ’no pH’: no pH control.

HRT

CH4

CH4 prod.

(%)

(mL CH4 /g COD)

74±1% 81±1% 77±3% 78±3% 75±2% 83±5% 75±3% 76±3%

301±21 314±16 220±25 208±18 282±39 279±38 219±60 297±38

(d)

CSTR CSTR AnSBR TPAD

M T M T M T no pH pH

30 30 15 15 15 15 4, 15 4, 15

100

Figure 3.29: Comparison of the specific methane productions of the different reactor set-ups during digestion of adsorptive sludge. ’M’: mesophilic conditions. ’T’: thermophilic conditions. ’pH’: pH controlled by NaOH addition. ’no pH’: no pH control.

None of the stabilized sludges had a more than average dewaterability (defined at K = 1·105 kg/(m3 ·s)), with the best dewaterability observed in the sludges produced by the AnSBR and the worst by that of the TPAD. Anaerobic sequencing batch reactor

Iin Figure 3.30, the daily methane production rate of the AnSBR reactors is compared to those of CSTR reactors operating at the same OLR and HRT. With an average methane production of 277±37, the mesophilic AnSBR performed 25% better than the CSTR and almost reached the methane production of a CSTR operating at half its OLR and twice its HRT (Figure 3.29). The thermophilic AnSBR performed slightly worse than the mesophilic one. In general, methane production rates in AnSBRs where less stable than in CSTRs. The total removal efficiency (taking into account both overflow effluent and the wasted sludge) of the mesophilic AnSBR reactors was 69% for VS and 70% for COD, which also is comparable to the performance of a CSTR reactor operating at an HRT twice as large (or in other words, twice as big). The fractionation of 101

Table 3.7: Influent (’In’) and effluent (’Out’) qualities and removal efficiencies (’RE’) during anaerobic digestion of adsorptive sludge. Note also that the KjN content of the sludge was on average 2.1±0.2 g N/L. ’M’: mesophilic conditions. ’T’: thermophilic conditions. ’pH’: pH controlled by NaOH addition. ’no pH’: no pH control. For the AnSBR reactor, the COD, VS, VSS and TAN effluent data refer to the overflow effluent, whereas K refers to the wasted excess sludge.

In

CODt

VS

VSS

TAN

K

(g/L)

(g/L)

(g/L)

(mg N/L)

(kg/(m3 ·s))

36 - 40

24 - 26

21 - 24

313 - 427

CSTR

M, 30 d

Out RE

12±2 68%

8.5±0.9 66±3%

7.6±0.1 66%

1066±40

4.7·10−6

CSTR

T, 30 d

Out RE

13±2 64%

9.9±0.2 61±3%

8.2±0.2 62%

1099±20

3.1 · 10−6

CSTR

M, 15 d

Out RE

14.9±0.9 59%

10.6±0.4 55±2%

10.1±0.4 52%

916±80

-

CSTR

T, 15 d

Out RE

19±3 47%

12.7±0.7 46±2%

10.6±0.3 50%

1186±80

5.5 · 10−6

AnSBR

M

Out RE

2.2±0.2 94%

1.9±0.2 88±2%

1.0±0.2 95%

1130±60

1.5 · 10−5

AnSBR

T

Out RE

4.3±0.8 89%

3.0±0.3 85±2%

1.0±0.1 95%

994±40

8.6 · 10−6

TPAD

no pH

Out RE

13±3 63%

9±1 62±4%

7.8±0.2 65%

935±90

-

TPAD

pH

Out RE

12±2 70%

8.3±0.4 68±1%

7.6±0.2 68%

1164±70

2.7 · 10−6

102

Figure 3.30: Specific methane production of AnSBR reactor digesting adsorptive sludge, compared to a CSTR operating at the same HRT and OLR.

the contaminants into the overflow effluent and the wasted sludge and the resulting concentrations in those two fractions are shown in Figure 3.31. The major part of solids and COD leaving the reactor do so via the wasted sludge. Nitrogen, composed of some 75% dissolved ammonium, mostly leaves the reactor via the overflow. Some 70% of the phosphorus entering the reactor leaves it as part of the wasted sludge. The retention of sludge in the AnSBR reactors allowed them to reach a MLVSS concentration of 26±2 g MLVSS/L and 23±2 g MLVSS in mesophilic, resp. thermophilic conditions, compared to the MLVSS concentration in CSTR reactors (operated at similar OLR and HRT) of only 10.1±0.3, resp. 10.6±0.4 g MLVSS/L. During the start-up of the mesophilic reactor, inert suspended solids accumulated in the reactor up to a concentration of 31 g/L. After regular wasting of excess sludge was commenced and a steady state reached, the concentration of inert suspended solids dropped down to 20±1. In the thermophilic reactor, the inert solids reached approximately 20 g/L during start-up before wasting of the sludge had to be initiated, indicating worse (inorganic) solids retention in thermophilic conditions. The sludge retention obviously also lead to a higher SRT in the AnSBRs (Table 103

Figure 3.31: Fraction of total effluent load of TS, COD, N, resp. P, present in the overflow effluent, resp. wasted sludge of an AnSBR digesting adsorptive sludge at mesophilic conditions. The numbers inside the bars represent the average concentrations (g/L) of TS, COD, N and P found in the overflow effluent and wasted sludge.

3.5). The SRT was calculated as the ratio of the amount of VSS in the reactor and the amount of VSS leaving the reactor per day, either via the supernatant or the wasted sludge (Wang et al. (2009)). At steady state conditions, it reached 53, resp. 38 days in the mesophilic, resp. thermophilic reactor. The comparatively worse sludge retention in thermophilic conditions also manifested itself by more severe foaming, which lead to occasional malfunctioning of the reactor. Temperature phased reactor

The pH of the adsorptive sludge being fed to the reactors was 5.4±0.2. Without pH control, the thermophilic pretreatment reactor of the TPAD had a pH of only 5.9. The high VFA accumulation in this reactor was reflected by a high ripley index of 9.2 (Table 3.5) and is illustrated in Figure 3.32, with total VFA concentrations of 7±1 g/L on average and a high acetate concentration of 2.7±0.7 g/L, which is indicative of a failing methanogenesis. This failing methanogenesis is shown in Figure 3.33, where it can be seen that the first reactor produced only 13±5 mL CH4 /g COD, which is only 6% of the methane produced by both reactors combined, which was 196 ± 58 mL CH4 /g COD. As such, TPAD produced 104

slightly less methane than a CSTR operated with an HRT equal to that of the mesophilic TPAD reactor (Table 3.6). The mesophilic reactor contained a total VFA concentration of only 0.2 g/L on average, with no propionate present at any time, conditions which can be considered conducive to methanogenesis. However, the ripley index for this reactor, when measured, was only 0.31, indicating an only fairly well operating methanogenic community. It was observed that at times the VFA concentration in the reactor collapsed, indicating unstable conditions for the methanogenic community. This is probably what lead to the large variations in specific methane production observed in Figure 3.33. When the pH in the pretreatment reactor was maintained above 6.8 by dosing NaOH at a dose of 0.4 mg NaOH per gram of sludge wet weight fed to the reactor (or 14 g NaOH/(g VS fed)), the conditions for methanogenesis in the first reactor improved, with a ripley index of 0.66 and dropping acetate levels (Figure 3.32). Indeed, the methane production in the first reactor picked up to 66±34 mL CH4 /g COD. The average methane production by the mesophilic reactor was comparable to the production when pH control was applied, but arguably the main improvement resulting from the dosing of NaOH was that lead to more stable conditions in the mesophilic reactor, which lead to a more stable methane production there and, on average, a 44% higher total methane production. The TPAD reactor thus achieved a methane production rate only slightly less than that produced by a CSTR operated at HRT 30 d, whereas the HRT of the total TPAD was only 19 d. 3.6.2

Centrifuged sludge

Sludges produced during the centrifugation tests were digested in CSTR reactors under mesophilic as well as thermophilic conditions. Due to varying influent strengths and the limited amount of sludge which could be produced from each centrifugation tests, it was not possible to maintain steady conditions (OLR, HRT) in the reactors for as long as would be desirable. The specific gas production (mL CH4 /g COD fed) of the reactors is shown in Figure 3.34. For both mesophilic and thermophilic conditions, the methane production rates varied between 190±50 mL CH4 /(g COD) when OLR was 2.0 g COD/(L·d) and 400±40 mL CH4 /(g COD) when OLR was 0.4 g COD/(L·d). For thermophilic conditions, the range was between 200±90 and, resp., 430±110 mL CH4 /(g COD). Although no significant difference in the average specific methane production rates could be observed 105

Figure 3.32: VFA evolution in the two TPAD reactors. The dotted line marks the start of pH control.

Figure 3.33: Effect of pH control in the thermophilic TPAD reactor on the combined specific methane production of the TPAD reactors and on the specific methane production and pH of the thermophilic TPAD reactor itself.

106

when comparing mesophilic and thermophilic conditions, it was apparent that the methane production rate in the former was more stable. During one period of the test run, iron-dosed sludge was fed to the reactors (Figure 3.34). This sludge was obtained by dosing 1600 L of municipal wastewater with 90 g FeCl3 and subsequent centrifuging. During this period, OLR was 1.1 g COD/(L·d), HRT was 45 d, VS removal was 56%, resp. 51%, COD removal was 61, resp. 55%, and the methane production rate was seen to drop to 260, resp 225 mL CH4 /g COD in mesophilic, resp thermophilic conditions. These levels are comparable to the methane production observed when digesting undosed sludge at a similar OLR but three times lower HRT. This indicates that iron may have inhibited methane production, most severely in the case of the thermophilic reactor, but the test run was not long enough to be able to state this conclusively. When the feeding iron-dosed sludge was stopped and undosed sludge was again fed to reactor (albeit it an OLR of only 0.4 g COD/(L·d)), the methane production was seen to increase rapidly.

107

Figure 3.34: Organic loading rate, HRT and resulting specific methane production by a mesophilic and thermophilic CSTR reactor digesting concentrate obtained by centrifuging municipal wastewater. ’FeCl3’ refers to a test period during which iron-dosed sludge was fed to the reactor.

108

Chapter 4

Discussion 4.1

The benchmark: conventional activated sludge treatment

The aim of the zero waste water treatment approach with upconcentration at the WWTP level is to achieve a resource recovery that significantly improves upon the current treatment approaches. It is therefore useful to analyze the efficiency of the what can be considered the benchmark treatment process, which is conventional activated sludge treatment with anaerobic digestion of the secondary sludge. Focus will in particular be put upon the energy efficiency. 4.1.1

Energy recovery in CAS

Sludge digestion is the process which allows for energy recovery from CAS sludge. As mentioned in Section 1.3, the tendency to use extended aeration CAS treatment has led to the production of smaller quantities of worse digestible sludges. Because of the high costs of sludge management, the production of smaller quantities of excess sludge has often been regarded as a positive aspect. From an energy point of view however, it simply means that a large fraction of the energy embodied in the wastewater has been lost as heat because of excessive endogenous respiration by the aerobic biomass. To get an idea of the extent of this, a CAS system without primary sedimentation and an SRT of 25 days can be compared to one with primary sedimentation and an SRT of 5 days in the secondary treatment: the former yields only some 14.5 kg VSS/(IE·year), whereas the latter yields a 109

total of 19.4 kg VSS/(IE·year), of which 9.2 kg VSS/(IE·year) is primary sludge (Diamantis et al. (2011)). The fact that the excess sludge is harder to digest anaerobically is a more obvious drawback in terms of energy recovery. It is caused by the fact that the digestion of municipal wastewater sludges is typically limited by hydrolysis, and that the most important factor determining the hydrolizability of secondary sludge is the residence time of the sludge (i.e. the sludge age) in the CAS system (Carrere et al. (2010); Batstone et al. (2008b)). Indeed, typically only 35 to 60 % of the organic solids (VS) in secondary sludges from CAS systems that operate at an SRT of 25 days can get hydrolized (Batstone et al. (2008b); Verstraete (2011)). In addition, the rate at which hydrolysis of the sludge occurs is slower for older sludges (Batstone et al. (2008b)). Because of these limitations, it is observed that standard digestion technology tends to achieve a conversion efficiency of organics to methane of only 15% (at CAS SRT of some 45 d) to 30% (at CAS SRT of some 15 d) (Batstone (2010)). Others have reported that even the digestion of typical mixtures of primary and secondary sludge can achieve maximum 40% COD conversion to methane (STOWA (2005)). This is much less than for primary sludge in itself, which is highly putrescible (hydrolysis of the particulates in primary sludge occurs about 20 times faster than for secondary sludge (Verstraete (2011))) and thus typically has an anaerobic degradability of 60% to 80% at mesophilic conditions (Mahmoud et al. (2004)). The latter phenomenon can be modelled as a function of the activated sludge age (θX ). The anaerobically biodegradable fraction (D) of the sludge (under mesopilic conditions) is modelled as (Verstraete (2011)): D = (1 + (1 − d) · b · θX )−1

(4.1)

with d the degradility of young active sludge (typically 0.73) and b the decay coefficient of the sludge (typically in the order of 0.05 d−1 and slightly lower for extended aeration systems). According to equation 4.1, a sludge which has a θX = 25 days is only 55% biodegradable in anaerobic digestion, whereas a sludge which has a θX = 5 days is almost 70% biodegradable. Combined, these two factors mean that less energy can be recovered in the form of 110

Figure 4.1: Process layout of conventional activated sludge system with anaerobic digestion of excess sludge.

biogas. Indeed, a quick comparison between the two systems cited as an example shows that, under the assumption that 30% of the VS in the 5 day old and 22% of the 25 day old activated sludges are converted to methane during AD and that 70% of the primary sludge VS gets converted, in the system with primary treatment and a young secondary sludge, (9.2 · 70%)+(10.2 · 30%)= 9.5 kg VSS/(IE · year) gets converted into 4.7 m3 CH4 /(IE · year), whilst in the system with secondary treatment only, (14.5 · 22%) = 3.2 kg VSS/(IE · year) gets converted to 1.6 m3 CH4 /(IE · year), or about three times less. Furthermore, the amount of stabilized sludge remaining after digestion (and thus needing further management) is even slightly less in the former case (19.2 - 9.5 ≈ 10 kg VSS/(IE · year) and 14.5 - 3.2 ≈ 11 kg VSS/(IE · year), respectively), despite the fact that initially more aerobic sludge was produced. As mentioned in section 1.2, the lower potential methane producion per IE in CAS systems with extended aeration implies that anaerobic digestion with CHP combustion is only attractive for large WWTPs. Conventional CHP engines typically cannot effectively operate below an output of 0.35 MW (Batstone et al. (2008a)). 111

As methane has an energy content of 32 MJ/m3 , this implies that at least 3.45·105 m3 /y of methane must be produced at the plant. As a result, CAS plants with no primary treatment and a sludge age of 25 days, must treat at least 3.45·105 m3 /y · (1.6 m3 /(IE·year))−1 ≈ 215,000 IE, or receive additional external sludge, in order to be able to effectively convert the produced methane to energy in a CHP engine. In comparison, plants with primary treatment and a sludge age of 5 days that have sizes down to some 75,000 IE treated can generate enough methane to effectively utilize conventional CHP. To further elaborate the energetic efficiency of a CAS system with no primary treatment, we can consider a load of 500 g COD (1 m3 of wastewater, containing also 40 g Kj-N) entering the plant. Assuming that the wastewater has a chemical energy content of 17.9 kJ/g COD (Heidrich et al. (2011)), this amounts to 8950 kJ. If one assumes there is a 90% COD removal efficiency, and that the removable compounds have an average energy content of 15 kJ/g COD (see Section 1.2), then this implies that 500 g COD · 90% · 15 kJ/g COD = 6750 kJ is removed, either via mineralization (in which case the energy is released as heat) or biomass growth (in which case the energy remains contained within biomass), and thus that 8950 - 6750 = 2200 kJ is released via the effluent. For a typical CAS plant with extended aeration and nitrification/pre-denitrification (SRT = 15 d), the typical excess sludge yield is 0.3 g CDW/(g COD removed) (Verstraete (2010)). The typical heat of combustion of bacterial cells is 22 kJ/g CDW (Powers et al. (1973)), which means that 6.6 kJ/(g COD removed) gets embodied in excess sludge, or in total 500 g COD · 90% · 6.6 kJ/(g COD removed) = 2970 kJ. Thus, 6750 - 2970 = 3780 kJ gets released as heat. This heats up both the water and the excess sludge, with about 0.9◦ C (specific heat of water = 4.18 kJ/(◦ C · L)). A mass of 500 g COD · 90% · 0.3 g CDW/(g COD removed) = 135 g CDW of excess sludge goes to the digester. Bacterial biomass has a COD content of 1.42 g COD/(g CDW) (Marais & Ekama (1976)), which means that 135 g CDW · 1.42 g COD/(g CDW) = 192 g COD goes into the digester, or some 40% of the original COD. In the digester, at 34◦ C, we assume that 30% of the (secondary) sludge COD gets converted to methane (Bolzonella et al. (2009)). 0.35 L of CH4 is produced when 112

1 g of COD is removed (Verstraete (2011)), so 192 g COD · 30% · 0.35 L/(g COD removed) = 20.2 L of methane gets produced. When this methane is combusted, releasing 32 MJ/(m3 CH4 ), a total of 646 kJ can be set free in the form of heat (but preferably in the form of electricity). However, part of this energy must be used to heat the reactor, since only some 4.6% of the chemical energy content of the anaerobically degraded COD gets converted to heat (Gallert & Winter (2005)), i.c. 45 kJ. Assuming that the (thickened) sludge has a dry weight percentage of 7%, 1.9 L of thickened sludge must be heated. A typical median temperature for wastewater in a western European WWTP is 17◦ C (Verstraete & Vlaeminckx (2011); Wett et al. (2007)). Thus, to heat the sludge to 34◦ C, 135 kJ is required, leaving 646 - (135 - 45) = 556 kJ of net recoverable energy (assuming perfect isolation of the reactor vessel). The stabilized sludge, finally, still contains 192 g COD · 70% = 134 g COD, or 2077 kJ embodied energy. To accomplish the COD and N removal, aeration must be provided. To achieve 90% COD removal, the heterotrophic bacteria require 500 g COD · 90% · 0.6 g O2 /(g COD removed) = 270 g O2 (Verstraete & Vlaeminckx (2011)). Assuming an 80% Kj-N removal, the autotrophic nitrifiers require 4.33 g O2 /(g N) * (40 g available N - 0.05 g N immobilized/(g CDW) · 135 g CDW - 20% · 40 g N) = 109 g O2 (Verstraete (2010)). Part (maximum 62.5%) of the nitrification oxygen demand can be recovered through denitrification. For a typical pre-denitrification efficiency of 40%, 109 g O2 · 62.5% · 40% = 27 g O2 is recovered. Thus, there is a total oxygen demand of 270 + 109 - 27 = 352 g O2 . Because of limited oxygen transfer from the air bubbles to the bacteria, typically twice as much aeration is provided than what is demanded by the biomass: i.c. 704 g O2 . A typical energetic efficiency of an aerator is 1.5 kg O2 /kWh (although efficiencies up to 5 kg O2 /kWh are possible (SenterNovem (2006))) or 0.42 g O2 /kJ, which means that 704 g O2 / (0.42 g O2 /kJ) = 1680 kJ of electrical energy is required for aeration (minimum 504 kJ in case of high-efficiency aerators). Aeration typically consumes some 60% of the energy consumption of a CAS WWTP (Zessner et al. (2010); SenterNovem (2006)), thus an estimated total of 2790 kJ of electrical energy is consumed. We can now summarize the chemical energy metabolism of the combined CAS and AD system (see Table 4.1) and can conclude that only 6% of the total energy which enters the system (including the chemical energy in the raw sewage, the 113

Table 4.1: Energy metabolism of the conventional activated sludge system. Values listed apply to 1 m3 of municipal wastewater containing 500 g COD and 40 g kj-N. Shown is the efficiency of conversion of chemical energy into useful energy in a CAS system with AD for renergy recovery, and the efficiency of conversion of total energy (heat+chemical) to useful energy for a system in a temperate climate utilizing both AD and heat pumps for energy recovery.

Energy Input 1. Raw sewage COD Used 2a. Aeration (standard efficiency) 2b. Aeration (high efficiency) 3. Heating AD reactor 4. Other operations Recovered 5. Methane Output 6. Effluent heat 7. Effluent COD 8. Stabilized sludge

g COD

kJ

500

8950 1680 504 135 1110 556

50 134

3780 2200 2077

Scenario with standard efficiency aeration & without heat recovery: Energy consumed ( = 1 + 2a + 3 + 4 - 8)

9798

Energy recovered ( = 5)

556

→ (CAS + AD) overall efficiency = 5.7% Scenario with high efficiency aeration & without heat recovery: Energy consumed ( = 1 + 2b + 3 + 4 - 8)

8622

Energy recovered ( = 5)

556

→ (CAS + AD) overall efficiency = 6.4% Scenario with standard aeration efficiency & effluent heat recovery Energy consumed ( = 1 + 2a + 3 + 4 – 8 + 25.1 MJ)

34985

Energy recovered ( = 5 + 16.7 MJ)

17276

→ (CAS + AD + heat recovery) overall efficiency = 49.5%

114

electricity consumption for plant operation, but substracting the chemical energy still contained in the stabilized sludge, as that can be used in a further stage, e.g. during incineration) can be recovered in the form of methane. We refer to this measure of the energy efficiency as the overall energy efficiency. If one only takes into account the energy input for operation of the plant and the energy output from methane combustion, and not the energy contained in the influent, effluent or stabilized sludge, an energy efficiency of 20% is arrived at. We refer to this more conventional manner of measuring the energy efficiency as the technical energy efficiency. Note also that 134 g of COD in the stabilized sludge amounts to 94 g VS remaining as stabilized sludge - or assuming a VS/TS ratio of 0.75 (Alphenaar et al. (1992)) - 125 g TS. It should also be noted that some factors have not been taken into account, such a the heat losses from the digester (can be minimized by proper isolation, and potential heating saving by performing heat exchange between digester effluent and influent. Especially for countries with a cold climate, it can be instructive to also take the heat content of the wastewater into account, and consider the use of heat pumps for extracting heat energy from the effluent (Thornberg & Johansen (2010)). If one again assumes a median temperature of the WWTP effluent of 17◦ C (Verstraete & Vlaeminckx (2011)) - of which 0.9 ◦ C is due to heat released during aerobic respiration - and a median ambient temperature of 11◦ C, it can be calculated that the influent has an extractable heat energy of 25.1 MJ/m3 . It has been reported that in similar circumstances, heat pumps can produce hot water of 90◦ C at a coefficient of performance (i.e. the ratio of recovered heat to consumed work) of at least 3 (Thornberg & Johansen (2010)). That means that heat pumps would be able to extract a netto amount of 25.1 MJ/m3 · 2/3 = 16.7 MJ/m3 . The total (i.e. heat + chemical) energy conversion efficiency of the combined CAS, AD and heat pump is then a much higher 57% (see Table 4.1). This illustrates that, although the chemical energy contained in wastewater is in theory more than what is currently being spent to run CAS systems, there is - at least in temperate climates - a potentially much more significant energy recovery achievable from the heat energy in municipal wastewater. It must be stressed however that this recovered energy comes in the form of heat, which has a low market value compared to e.g. the electrical energy which can be generated during combustion of methane in a CHP engine. 115

4.1.2

Costs and water recovery from CAS effluent

Adding on tertiary treatment behind the conventional treatment, as discussed in Section 1.4, does not cause a significant increase in the energy impact of the plant (0.16 kg CO2 GHG emissions in addition to 0.83 kg CO2 (Pasqualino et al. (2011))). In terms of costs, a UF + RO system can have a combined capital and operational cost (CAPEX+OPEX) of 0.33 EUR per m3 of secondary effluent treated within the European context (Verstraete & Vlaeminckx (2011)). Even in the case where the reclaimed water is used to recharge an aquifer used for drinking water production (so called indirect reclamation) - as is for example done in Oostduinkerke, Belgium - the overall cost to produce drinking water from secondary effluent is only 0.54 EUR/ m3 (Van Houtte & Verbauwhede (2008))). The CAS system itself has a typical CAPEX+OPEX of some 0.6 EUR/m3 (Verstraete & Vlaeminckx (2011)). To compare, a survey in 14 countries found that municipal water prices vary between 0.5 EUR/m3 (in the US) and 1.6 EUR/m3 (Clark (2011)). It is thus not unlikely that the reclaimed water can be valued high enough to make the extra investment in tertiary treatment a break-even operation. If the water can be valued at around 1.1 EUR/m3 , even the whole wastewater treatment can turn from a costly enterprise into one that can breaks even.

4.2 4.2.1

Chemically enhanced primary treatment Energy recovery in CEPT

It was shown that CEPT is able to recover 84%, resp. 87% of the influent COD when using a dose of 100 mg/L coagulant - FeCl3 , resp. Al2 (SO4 )3 - and 1 mg/L weak anionic polyelectrolyte (Figure 3.8). Considering that Al2 (SO4 )3 has a greater footprint and toxicity and did not significantly outperform FeCl3 in COD recovery, and taking into account that we have no data available on the digestibility of aluminum-dosed CEPT sludge, we will retain FeCl3 as the coagulant of choice for further analysis. It can be assumed that iron has a less negative effect on digestion than aluminum, as it tends to precipitate with sulfide, a compound which is toxic to methanogens and stimulates competition from sulphate reducing bacteria (Van Wesenbeeck (2010)). To treat 1 m3 of influent (as before, we assume 500 g COD/m3 and 40 g Kj-N/m3 ), 116

100 g of coagulant and 1 g of PE are thus required. Considering the energy used in producing these chemicals (see Section 1.9.1), the energy consumption of CEPT treatment amounts to 120 kJ/m3 for the coagulant, 45 kJ/m3 for the flocculant and 47 kJ/m3 for mixing. The volume of CEPT sludge produced was 4% of the influent volume, with a solids content of 1% and, arguably, a sufficiently high COD content (9.5 g/L) to not require further thickening prior to digestion. Thus, 40 L of sludge is fed to the AD reactor, containing 84% · 500 g COD/m3 = 420 g COD. Considering the (tentative) result that AD of iron-dosed sludge was most severely inhibited in thermophilic conditions, mesophilic digestion is selected. This means that 40 L of sludge has to be heated from 17◦ C to 34◦ C, requiring 2842 kJ. When considering the efficiency of AD of this sludge, we must take into consideration the Fe content of the sludge. Unfortunately, the literature is not clear on this. If we assume that practically all Fe dosed remains in the sludge, we get that 0.88 g Fe/(L sludge) is present, or 9% Fe on a dry weight basis. This is about 3 times more than the iron content of the adsorptive sludge on which most of the digestion tests were performed. It is however more similar to the expected iron content (6.5% Fe on dry weight basis) of the sludge that was produced by centrifuging chemically dosed wastewater (Section 3.2). It is therefore more suitable to base our estimation of the methane production potential of this sludge on the results obtained from AD tests on iron-dosed centrifuge sludge (Section 3.6.2). For that type of sludge, a methane production rate of 260 mL CH4 /(g COD) was observed. We thus obtain 420 g COD · 260 mL CH4 /(g COD) = 109.2 L CH4 . When combusted, this methane releases 3494 kJ. The COD removal during digestion was 61%, thus 164 g COD remains present in the stabilized sludge. It is now instructive to pay attention to an apparent discrepancy between the observed COD removal during AD of this kind of sludge, 61%, and the observed methane production rate, which was 74% of what is considered the theoretical maximum of 350 mL CH4 /g COD. It is important to realize that this theoretical maximum applies only to the digestion of carbohydrate compounds (Verstraete (2011)). Municipal wastewater, however, contains a chemical energy content of 17.9 kJ/g COD, which is significantly higher than that of a purely carbohydrate wastewater at 14.7 kJ/g COD (Section 1.2). It is therefore clear that the value of 350 mL CH4 /g COD does not apply to sludges produced by a physico-chemical 117

upconcentration of municipal wastewater such as CEPT, since these methods can not be assumed to preferentially recover only carbohydrate compounds from the wastewater. Based on the experimental results, the expected amount of methane produced during the anaerobic destruction of 1 g of COD from CEPT sludge is (260 mL CH4 /g COD) / 61% = 426 mL CH4 / g COD, rather than the 350 mL CH4 /g COD in purely carbohydrate sludge. That this reasoning is plausible can be seen by comparing the ratio of these methane production rates for CEPT sludge and carbohydrate feed, with the ratio of the chemical energy contents per COD of municipal wastewater and carbohydrates, which should be equal for the argument to be correct. And indeed, these ratios are respectively 426/350 = 1.22 and 17.9/14.7 = 1.22. It thus becomes apparent that physico-chemical upconcentration processes have a distinct advantage over purely biological processes, in that the sludge they produce contains more chemical energy per g of COD than biomass does, which is reflected in a higher maximum achievable specific methane production rate. Returning to the energy analysis of CEPT treatment, it can now be seen that the energy content of the stabilized sludge is 164 g COD · 17.9 kJ/g COD = 2936 kJ. With this final piece of data, the energy efficiency of the CEPT process combined with anaerobic digestion can now be calculated, and is summarized in Table 4.2. The overall energy efficiency of the whole process - taking into account the chemical energy in influent, effluent and stabilized sludge - is 34%. When only taking into account the energy used for operation of the plant and the energy recovered by methane combustion, the technical energy efficiency is 84%. Note that this could be further improved upon if an energetically efficient thickening of the CEPT sludge were possible, which would reduce the substantial heating requirements for the AD reactor (see further). It must be noted that no nitrogen removal or recovery is included in this consideration. Phosphorus recovery into the sludge on the other hand, was shown to be excellent (average 70%) during CEPT. Note also that the amount of stabilized sludge, when expressed as VS is 95 g VS, which is the same as after CAS+AD. There will however be a substantially higher fraction of inert solids in the stabilized sludge than after CAS+AD, since some 34 g of Fe (or better e.g. 66 g Fe(OH)3 ) can be expected to still be in the stabilized sludge.

118

Table 4.2: Energy metabolism of CEPT + AD. Values listed apply to 1 m3 of municipal wastewater containing 500 g COD. Shown is the efficiency of conversion of chemical energy into useful energy in a CEPT system with AD for energy recovery.

Energy Input 1. Raw sewage COD Used 2. Coagulant production 3. Polyelectrolyte production 3. Mixing 4. Heating reactor 5. Other* Recovered 6. Methane Output 7. Effluent COD 8. Stabilized sludge

g COD

kJ

500

8950 120 45 47 2842 1110 3494

80 164

1432 2936

Energy consumed ( = 1 + 2 + 3 + 4 + 5 - 8)

10178

Energy recovered ( = 6)

3494

→ (CEPT + AD) overall efficiency = 34% * The same value as for CAS is assumed

119

4.2.2

Water recovery from CEPT eflluent

Since the effluent of CEPT treatment was shown to have a turbidity below 15 NTU, it is suitable for treatment with UF, which can then be followed on with RO. It has been reported that this could work at least as well as UF+RO treatment of CAS effluent and would thus not come at a higher cost (Even-Ezra et al. (2011)). This is in part thanks to the complete removal of phosphates by CEPT, the absence of soluble microbial products and the decrease in pH, all of which contribute to comparatively less fouling of the RO membrane. RO would also largely retain the nitrogen which CEPT treatment could not remove or recover. 4.2.3

Costs of CEPT

At a cost of 0.54 e/kg FeCl3 , 3e/kg PE, the chemicals used in CEPT, at the selected doses, costs 0.054 e/m3 , resp. 0.003 e/m3 for coagulant and PE respectively. The sludge treatment costs, at 0.47 e/kg DW (Verstraete & Vlaeminckx (2011)), amount to some 0.1 e/m3 . Electricity cost, at 0.12 e/kWh, for mixing amounts to 0.0016 e/m3 . After using the heat produced from methane combustion in a CHP unit for heating the reactor, at least 0.18 kWhel /m3 is left over. This can be used to cover the assumed energy consumption of the ’other operations’. This leaves an electricity cost of 0.13 kWh/m3 · 0.12 e/kWh = 0.015 e/m3 . The balance of OPEX thus is an expected cost of about 0.17 e/m3 .

4.3

Centrifugation

4.3.1

Energy efficiency

It was found that centrifugation removed 86% of particulate COD. For a typical CODp/CODt ratio of 70%, the CODt removal amounts to 60%. We can analyze the energy efficiency of the centrifuge + AD treatment of municipal wastewater. We look again at 1m3 of wastewater containing 500 g COD and 40 g Kj-N. The tested centrifuge had an energy consumption of 1 kWh/m3 = 3600 kJ/m3 , but centrifuges are available which use only half of that. Highly concentrated sludge could be produced, with a total solids content of 7.3 % the highest achieved in our tests. For 1 m3 of influent centrifuged, only 4 L of such a concentrate is produced. This means that only 284 kJ is required to heat the sludge during digestion. It was found that when digesting this kind of sludge in a mesophilic CSTR at an OLR of 120

2.0 g COD/(L·d) and an HRT (= SRT) of 45 d, a methane production of 190 mL CH4 /(g COD) was achieved. The amount of methane produced would thus be 57 L, which would produce 1824 kJ when combusted. 166 g COD would leave the digester as stabilized effluent, embodying some 2975 kJ. The energy efficiency of the CEPT process combined with anaerobic digestion can now be calculated, and is summarized in Table 4.3. The overall energy efficiency of the whole process taking into account the chemical energy in influent, effluent and stabilized sludge is 20%. When only taking into account the energy used for operation of the plant and the energy recovered by methane combustion, the technical energy efficiency is 57%. Expressed as COD, the amount of stabilized sludge leaving the reactor is the same as was the case for CEPT, but due to its high solids content, the volume will be substantially less. Clearly though, the removal efficiency, both for COD, N and, presumably, P, of the centrifuge is not high enough to produce an effluent of satisfactory quality for either releasing into the environment or treating it with RO. The centrifuge is therefore more suitable for thickening concentrates produced by other upconcentration techniques instead of directly applying it to municipal wastewater. Finally, it was found that the addition of coagulants to municipal wastewater prior to centrifugation was not found to be effective. 4.3.2

Costs

After combustion of the methane in a CHP with efficiency 31%, 1259 kJth /m3 or 0.35 kWhth /m3 is produced, which is more than sufficient to heat the anaerobic digester. 0.16 kWhel /m3 is also produced, whilst the centrifuge and ’other operations’ consume 0.8 kWh/m3 . Sludge treatment costs can be assumed to amount to approximately 0.08 e/m3 . This makes an OPEX balance of approximately 0.16 e/m3 .

4.4 4.4.1

Primary membrane filtration with FMX Energy efficiency

Due to excessive fouling during FMX filtration of municipal wastewater, the fluxes through the membrane with cut-off with 0,5 µm, resp. 1 µm were prohibitively small, resulting in an excessive energy consumption of 46 kWh/m3 = 166 MJ/m3 , resp. 31 kWh/m3 = 112 MJ/m3 , which is far more than is present in munici121

Table 4.3: Energy metabolism of Centrifugation + AD. Values listed apply to 1 m3 of municipal wastewater containing 500 g COD. Shown is the efficiency of conversion of chemical energy into useful energy by a centrifuge with AD for energy recovery.

Energy Input 1. Raw sewage COD Used 2. Centrifugation 3. Heating AD reactor 4. Other operations Recovered 5. Methane Output 6. Effluent COD 7. Stabilized sludge

g COD

kJ

500

8950 1800 284 1110 1824

200 166

3580 2975

Energy consumed ( = 1 + 2 + 3 + 4 - 7)

9169

Energy recovered ( = 5)

1824

→ (Centrifuge + AD) overall efficiency = 19.9% * The same value as for CAS is assumed

122

pal wastewater in terms of chemical energy (9 MJ/m3 for a wastewater with 500 g COD/m3 ). When using intermittent vortex generation it was observed that the flux was only 15% lower than during continuous vortex generation, while it reduces the energy consumption of the FMX unit by half. Even if one were to operate with intermittent vortex generation and as such cut the energy consumption for filtration in half, the FMX unit as used would nevertheless still be completely uneconomical for the upconcentration of municipal wastewater. Larger units do exist with total membrane areas of 92.55 m2 and power consumption 3.6 times as high as the unit used in the experiment. Assuming such a unit operating under intermittent vortex generation, an energy consumption of 8044 kJ would still be required to upconcentrate 1 m3 of wastewater up to a permeate recovery of 96%, using the membrane with cut-off 1 µm. The FMX had a COD recovery of only 39% for this membrane. This means that if 1 m3 of influent containing 500 g COD were upconcentrated, 40 L of concentrate containing 195 g COD would be produced, thus having a concentration of 4.9 g COD/L. Digestion tests with such a type of concentrate were not performed. (Van Wesenbeeck (2010)) did however perform digestion tests with FMX concentrate that had a COD concentration of 7.8 g COD/L. Operating at an OLR of 0.4 g COD/(L·d) and an HRT = SRT of 20 d, a biogas production of 0.175 mL/(g COD fed) was achieved, but the methane content of this biogas was not reported. If we assume that the methane content would be similar to the one measured during our experiment with the digestion of centrifuged sludge, i.e. 80%, then we arrive at a specific methane production of 140 mL CH4 /g COD fed. This would imply a COD destruction of 33% during AD. For the situation under consideration, all this would mean that 140 mL CH4 /g COD · 195 g COD = 27.3 L of methane would be produced, releasing 874 kJ upon combustion. And that 131 g COD, or 2345 kJ would remain as stabilized sludge. The energy balance is shown in Figure 4.4. The overall energy efficiency of the whole process - taking into account the chemical energy in influent, effluent and stabilized sludge - is only 4.6%. When only taking into account the energy used for operation of the plant and the energy recovered by methane combustion, the technical energy efficiency is 7.2%. This configuration thus performs even worse than CAS in terms of energy efficiency, and its effluent quality is not better than CAS 123

either, with a COD removal of only 70% and a nitrogen removal of only 26%. It is not even sure that the permeate would be suitable for RO treatment, considering its high COD content. Furthermore, there would be the recurring cost of membrane replacement to be taken into account. In addition, the concentrate produced was not sufficiently concentrated and would ideally receive further thickening prior to digestion. It can thus be concluded that FMX treatment of raw wastewater was uneconomical. The FMX filtration may be more useful in thickening concentrates obtained by other techniques, since it was observed that the flux did not seem to get significantly worse when the feed to the membrane became progressively more concentrated. 4.4.2

Costs

With a CHP unit producing electricity from the methane, a balance of 2.5 kWhel /m3 spent is arrived at. The 0.17 kWhth /m3 produced is not enough to meet the heating requirements of the digester: a shortfall of 0.62 kWhth /m3 is found. The sludge treatment cost amounts to about 0.03 e/m3 . If one assumes a membrane replacement cost of 0.1 e/m3 , an estimated OPEX cost of approximately 0.41 e/m3 is arrived at.

4.5 4.5.1

Bio-adsorption in an MBR SRT = 1.2 d

The MBR operated at SRT 1.2 d achieved a COD removal of 74% and water volume recovery of 94%. For every m3 (as always assuming it contains 500 g of COD and 40 g Kj-N), 130 g COD = 2327 kJ would thus be lost via the effluent. The COD recovery (after 30’ sedimentation of the concentrate) was 58%. As such, per m3 influent, 500 · 0.58 = 290 g COD is recovered. The settled concentrate contained 12.8 g COD/L, which means that there is 22.7 L of settled sludge. Since mesophilic and thermophilic digestion of adsorptive sludge did not perform significantly different, mesophilic digestion can be selected as the most energy efficient. To heat the concentrate to 34◦ C, 1610 kJ is required. When digesting adsorptive sludge from Breda WWTP, the highest methane production rate achieved during mesophilic digestion was 301 mL CH4 /g COD. Note that this sludge had a sludge age of 0.7 days, relatively close but smaller than the sludge age of the MBR. Due 124

Table 4.4: Energy metabolism of FMX filtration + AD. Values listed apply to 1 m3 of municipal wastewater containing 500 g COD. Shown is the efficiency of conversion of chemical energy into useful energy by a full-scale FMX system operating with intermittent vortex generation and fitted out with an Enanics electrospun nanofibre membrane with cut-off 1µm, and with AD for energy recovery.

Energy Input 1. Raw sewage COD Used 2. Filtration 3. Pressurization* 4. Heating reactor 5. Other** Recovered 6. Methane Output 7. Effluent COD 8. Stabilized sludge

g COD

kJ

500

8950 8044 235 2842 1110 874

165 131

Energy consumed ( = 1 + 2 + 3 + 4 + 5 - 8) Energy recovered ( = 6)

2953 2345 18836 874

→ (FMX + AD) overall efficiency = 4.6% ** Pressurize from 1 to 3 bar at 85% efficiency * The same value as for CAS is assumed

125

to the limitations of flux, it was not possible to obtain a smaller HRT and as such an SRT had to be chosen that was higher than optimal, in order to still achieve sufficient upconcentration. The amount of methane produced per m3 of influent is 87 L, which releases 2793 kJ upon combustion. The COD removal during digestion was 68% which is slightly better than what was reported by (Akanyeti et al. (2010)) for a similar sludge (sludge age 1 d). There thus remains 93 g COD in the stabilized sludge. Note that the adsorptive sludge from Breda WWTP was dosed with Fe, while this was not true for the sludge in the MBR. It is not clear what the effect of this iron has been on the digestion of adsorptive sludge. Based on the COD removal and the specific methane production, it can be seen that 301/0.68 = 443±22 mL of CH4 was produced per g of COD destroyed, more or less equal to the value found in Section 4.2.1. This appears remarkably high at first sight. Analysis of the performance of the Breda WWTP has indicated that an estimated 60% of the COD removal in the bio-adsorption reactor there occurs through adsorption of contaminants to sludge flocs (Gao & Cardoen (2011)). The remaining 40 % is metabolized by bacterial biomass, with a sludge yield of 0.4 g CDW/(g COD removed). Bacterial biomass has a COD content of 1.42 g COD/(g CDW) (Marais & Ekama (1976)). This implies that for every gram of COD removed in the A-reactor, 0.4 g CDW/(g COD removed) · 40% · 1.42 g COD/(g CDW) = 0.27 COD becomes embodied in bacterial biomass, while 0.60 g COD gets adsorbed. Thus, adsorptive sludge is expected to consist of 0.6/(0.6+0.27)= 69% adsorbed COD and 0.27/(0.6+0.27) = 31% COD embodied in bacterial biomass. The former fraction is expected to have an energy content of 17.9 kJ/g COD, while the latter has an embodied energy content of 22 kJ/(g CDW) · 1.42 g COD/(g CDW) = 15.5 kJ/g COD (Powers et al. (1973)). Thus, every g COD in adsorptive sludge is expected to have an energy content of 69% · 17.9 kJ + 0.31 · 15.5 kJ = 17.2 kJ. This is very close to the energy content of raw wastewater (17.9 kJ/g COD), which goes a long way to explain why the amount of methane produced by the destruction of 1 g of COD in adsorptive sludge is substantially higher than what is common for the digestion of pure bacterial biomass such as secondary sludge. In fact, it is as high as the methane production per (g destroyed COD) for primary sludge which was found in Section 4.2.1. The explanation could perhaps be 126

that the anaerobic biomass preferentially digests the adsorbed matter in adsorptive sludge rather than the bacterial cells, since the latter are in a less reduced state than the raw adsorbed contaminants and thus less energy can be gained from metabolizing their biomass. In addition, they are presumably also harder to hydrolyze. Given that the COD removal during digestion was observed to be 68%, it can be assumed that the majority of the COD in the stabilized sludge leaving the digester is in the form of bacterial biomass, as most of the adsorbed contaminants would have been digested. This means that the stabilized sludge is expected to have an energy content of 15.5 kJ/g COD. Returning back to the analysis of the energy efficiency of the MBR system, we can now see that the 93 g COD in the stabilized sludge represents 93 g COD · 15.5 kJ/g COD = 1442 kJ. Simillar to Section 4.1.1, the heat released into the effluent as a result of the mineralization of some of the removed COD can be calculated. Based on the assumption that 40% of the removal is through bacterial metabolism and that the sludge yield is 0,4 g CDW/g COD, it can be calculated that for every m3 influent, 918 kJ gets released into the water (which heats it up by only 0.2 ◦ C). Note that we are able to check whether our assumption that the COD mass balance could be closed by assuming the missing fraction as being the result of mineralization, is true or not (Section 3.5.6). If 40% of the COD removal occurs through metabolic activity, then 40% · 74% = 30% of the influent COD is removed through metabolic activity. Given that aerobic metabolic pathways convert 50% of the used COD to biomass and mineralize the remaining 50% (Gallert & Winter (2005)), we find that the mineralization should be 30% · 50% = 15%, which is more or less exactly what was needed to close the COD mass balance (Figure 3.26). The water volume recovery was 94%. Thus, to create the underpressure of 0.2 bar and assuming a pump efficiency of 85%, the amount of energy used per m3 influent was: 0,2 bar · 10L/m · 0.94 / 0.85 = 22 kJ. The oxygen requirement of the biomass was actually quite limited: a maximum of 500 g COD/(m3 influent) · 74% · 0.6 g O2 /(g COD removed) = 222 g O2 /(m3 influent) was required for COD metabolization. In addition 30% Kj-N removal via 127

nitrification was observed. This would have consumed 4.33 g O2 /g N · 40 g N/(m3 influent) · 30% = 52 g O2 /(m3 influent) maximum, since the nitrification was not complete. Considering the flux was 15.7 L/(m2 ) and the water volume recovery 96%, it can be calculated that the biomass required 1.3 g O2 /h. In fact, this is a strong overestimation, since only some 40% of the COD removal is expected to occur via bacterial metabolism, while the remainder is through adsorption, which does not require oxygen. A more realistic oxygen requirement is therefore 0,7 g O2 /h. Taking a standard OC/load of 2, the oxygen requirement from the aeration was thus approximately 1.4 g O2 /h. As said, 20 L air/min was supplied to the reactor, which amounts to 180 g O2 /h, much more than required for bacterial metabolism. The large air supply however mainly served to keep the biomass in suspension and scour the membrane. Assuming a standard aeration efficiency of 1.5 kg O2 /kWh, it can be calculated that the aeration consumes 0.114 kW. To treat 1 m3 of influent, this means that 84 MJ of aeration energy was consumed. Obviously this was excessive. In fact, it is plausible that the high aeration rate create too turbulent conditions for optimal bioflocculation to be able to take place. (Judd (2006)) reported that the energy requirement for scouring MBR membranes is only around 600 kJ/m3 influent. This would still supply some 1.3 g O2 /h to the reactor, or just less than what is needed. Roughly speaking, 1000 kJ/m3 should however be enough to supply the biomass with the required oxygen for metabolism. We will therefore accept 1000 kJ/m3 as the energy consumption due to aeration. We can now calculate the energy efficiency of the whole process, which is done in Table 4.5. The overall energy efficiency of the whole process - taking into account the chemical energy in influent, effluent and stabilized sludge - is 25%. When only taking into account the energy used for operation of the plant and the energy recovered by methane combustion, the technical energy efficiency is 75%. Note however that these figures are heavily dependent on the assumption that 1000 kJ/m3 spent on aeration is sufficient to keep the sludge in suspension and supplied with sufficient oxygen and to scour the membrane enough to maintain the measured fluxes. Costs

With an aeration cost of 0.15 e/kWh (Verstraete & Vlaeminckx (2011)), a cost of 0.04 e/m3 for aeration is arrrived at. For the remaining operations combined, an 128

Table 4.5: Energy metabolism of MBR + AD with SRT = 1.2 d in the MBR. Values listed apply to 1 m3 of municipal wastewater containing 500 g COD. Shown is the efficiency of conversion of chemical energy into useful energy by an MBR with AD for energy recovery.

Energy Input 1. Raw sewage COD Used 2. Aeration 3. Pressurization* 4. Heating reactor 5. Other** Recovered 6. Methane Output 7. Effluent heat 8. Effluent COD 9. Stabilized sludge

g COD

kJ

500

8950 1000 22 1610 1110 2793

130 93

918 2327 1442

Energy consumed ( = 1 + 2 + 3 + 4 + 5 - 9)

11250

Energy recovered ( = 6)

2793

→ (MBR*** + AD) overall efficiency = 25% * 0.2 bar underpressure at 85% efficiency ** The same value as for CAS is assumed *** SRT = 1.2 d

129

electrical energy balance of -0.08 kWh/m3 is found, or 0.01 e/m3 . The heat produced in the CHP is just about enough to meet the digester heating requirements. Sludge treatment amounts to an estimated 0.06 e/m3 , making a total OPEX of 0.11 e/m3 . 4.5.2

SRT = 2.8 d

At SRT = 2.8 d, given the high level of mineralization that was calculated from the COD mass balance, i.e. 44%, and the fact that COD removal (80%) was less than twice as high, it is unlikely that any significant removal occurred through adsorption of contaminants to the sludge flocs. Based on this assumption and using the results about COD recovery (33%), water volume recovery (96%), flux (14.2 L/(m2 ·h)), etc. similar calculations as in Section 4.5.1 allow to calculate the energy balance (Table 4.6). The increased nitrification and COD removal through metabolic pathways requires extra aeration compared to the MBR with SRT = 1.2 d, and as a result, some 1600 kJ/m3 is required to perform aeration. The more extensive mineralization results in less COD going into the anaerobic digester. A specific methane production rate of 210 mL CH4 /g COD, with a COD removal of 60% was assumed. The overall energy efficiency of the whole process - taking into account the chemical energy in influent, effluent and stabilized sludge - is 10%. When only taking into account the energy used for operation of the plant and the energy recovered by methane combustion, the technical energy efficiency is 30%. Note again that these figures are heavily dependent on the assumption that 1600 kJ/m3 spent on aeration is sufficient to keep the sludge in suspension and supplied with sufficient oxygen and to scour the membrane enough to maintain the measured fluxes. In any case, it is clear that operating at SRT = 2.8 is much less energetically efficient than at SRT = 1.2 d, due to the excessive mineralization. It’s only advantage is the more extensive nitrification, which renders the effluent better fit for RO treatment, since nitrates and nitrites are often better retained by RO membranes than ammoniacal nitrogen. The RO test performed on the effluent of the MBR demonstrated the good removal efficiency which is achievable, allowing for the production of water which met the drinking water criteria for the parameters tested. Unfortunately it was not possible 130

Table 4.6: Energy metabolism of MBR + AD with SRT = 2.8 d in the MBR. Values listed apply to 1 m3 of municipal wastewater containing 500 g COD. Shown is the efficiency of conversion of chemical energy into useful energy by an MBR with AD for energy recovery.

Energy Input 1. Raw sewage COD Used 2. Aeration 3. Pressurization* 4. Heating reactor 5. Other** Recovered 6. Methane Output 7. Effluent heat 8. Effluent COD 9. Stabilized sludge

g COD

kJ

500

8950 1600 22 969 1110 1109

130 66

2480 1790 1023

Energy consumed ( = 1 + 2 + 3 + 4 + 5 - 9)

11628

Energy recovered ( = 6)

1109

→ (MBR*** + AD) Overall efficiency = 10% * 0.2 bar underpressure at 85% efficiency ** The same value as for CAS is assumed *** SRT = 2.8 d

131

to reliably assess the optimal pressure for RO treatment of the effluent, and thus no reliable assessment of the energy cost can be made. Costs

The thermal energy produced by CHP with an electrical efficiency of 31% is almost enough to meet the digester heating requirements. The electricity costs are calculated to be 0.09 e/m3 . Sludge treatment amounts to approximately 0.06 e/m3 . This makes a total OPEX of 0.15 e/m3 .

4.6 4.6.1

Digester configuration Anaerobic sequencing batch reactor

The anaerobic sequencing batch reactors treating adsorptive sludge worked well, reaching methane production rates equivalent to those of CSTR reactor which are double as big, thus potentially allowing savings on CAPEX (which is about 500 e/m3 reactor volume). It furthermore separated the effluent into a nitrogen rich liquid effluent and a high solids sludge effluent which contained most of the COD and P. In general, though, they appeared to be less stable than CSTRs, in particular at thermophilic conditions. In terms of average methane production, no significant difference between mesophilic and thermophilic conditions could be observed. 4.6.2

Temperature phased digestion

Considering the earlier remarks about the composition of adsorptive sludge as consisting of both adsorbed compounds and young bacterial cells, a temperature phased approach seemed a suitable option, with the aim to get a better hydrolysis of the bacterial cells in the first reactor. This did not appear to help though, as the methane production of the TPAD was about the same as that of a CSTR operating at an HRT equal to that in the mesophilic section of the TPAD. It could be that the microbial community satisfied itself with the adsorbed compounds and left the aerobic bacterial cells untouched because of their lower energy content. In any case, it was observed that the thermophilic reactor was quite unstable and thus affected the OLR of the methanogenic reactor, which may have hampered its

132

productivity. It is therefore conceivable that operating two methanogenic reactors in sequence might be more succesfull, as a microbial community might develop in the second reactor for which the aerobic bacterial biomass is an attractive energy source when a substantial amount of the adsorbed compounds is removed in the first reactor. Therefore, NaOH addition was performed to bring the pH up to methanogenfriendly levels. This worked to an extent, as biogas was produced in the second reactor and the total biogas production picked up and became more stable. Nevertheless, the TPAD still did not manage to perform significantly better than the AnSBR with a smaller volume, despite the extra cost of NaOH addition.

133

Chapter 5

Conclusions The results of the previous chapters are summarized in Table 5.1. ’Overall energy efficiency’ refers to the energy balance which takes into account the chemical energy which is present in the wastewater and which we ought to be attempt to exploit if we are serious about moving towards a zero-waste attitude with regard to wastewater treatment, whereas the ’technical energy efficiency’ refers to the balance between energy spent on operating the treatment and the energy recovered via AD, which is the more traditional way of looking at energy efficiency. Overall, the CEPT and MBR treatment at short SRT stand out as the most costeffective treatments, with the lowest OPEX, best energy efficiency and an effluent quality which is fit for further upgrading towards high quality water. There is certainly potential in the idea, but it would appear that an approach is needed in which the short time scale at which adsorption occurs can be more efficiently exploited. In an MBR setup, this was not completely possible, as the HRT could not be brought down because of flux limitation. Furthermore, there was no way of actually wasting sludge immediately after it having adsorbed COD.

5.1

Recommendations for future research

An approach that could perhaps be envisioned and researched in the future is a type of aerobic sequential batch reactor in which a cycle consists of: inflow of influent and mixing of the mixed liquor for about 10’ to allow bio-adsorption to happen, next a sedimentation, followed by an ample wasting of sludge which is rich in adsorbed matter at this moment, next an extended aeration to remove sub134

Table 5.1: Overall (’OE’) and technical (’TE’) energy efficiency of the upconcentration techniques in combination with AD, as well as their estimated OPEX and an assessment about the suitability of their effluent for water recovery via UF+RO (at an additional cost of 0.33 e/m3 ) or only RO.

CAS CEPT Centrifuge FMX-UF MBR @ SRT 1.2 d MBR @ SRT 2.8 d

OE %

TE %

OPEX e/m3

UF+RO?

RO?

6 34 20 5 25 10

20 84 57 7 75 30

0.6* 0.17 0.16 0.41 0.11 0.15

yes yes no -

no no no no? yes yes

* OPEX+CAPEX

strate, allow the biomass to grow and enter a famine mode, a sedimentation with decantation, and then back to the start of the cycle, with the bacterial biomass in an endogenous state and eager for a new round of adsorption. The MBR approach itself can certainly still be optimized along similar lines, perhaps with varying aeration cycles and well-selected sludge wasting periods to achieve better bioflocculation and timely removal of energy-rich sludge. Utilizing cross-flow filtration instead of submerged membranes after bio-adsorption could also be interesting in terms of reducing aeration requirements and preventing membrane fouling.

135

Bibliography Aivasidis, A., & Diamantis, V. I. (2005). Biochemical reaction engineering and process development in anaerobic wastewater treatment, vol. 92 of Advances in Biochemical Engineering / Biotechnology, (pp. 49–76). Berlin: Springer-Verlag Berlin. Akanyeti, I., Temmink, H., Remy, M., & Zwijnenburg, A. (2010). Feasibility of bioflocculation in a high-loaded membrane bioreactor for improved energy recovery from sewage. Water Science and Technology, 61 (6), 1433–1439. Aksu, Z. (2005). Application of biosorption for the removal of organic pollutants: A review. Process Biochemistry, 40 (3-4), 997–1026. Alphenaar, P. A., Perez, M. C., Berkel, W. J. H., & Lettinga, G. (1992). Determination of the permeability and porosity of anaerobic sludge granules by size exclusion chromatography. Applied Microbiology and Biotechnology, 36 (6), 795–799. Angenent, L. T., Karim, K., Al-Dahhan, M. H., & Domiguez-Espinosa, R. (2004). Production of bioenergy and biochemicals from industrial and agricultural wastewater. Trends in Biotechnology, 22 (9), 477–485. Angenent, L. T., Sung, S., & Raskin, L. (2002). Methanogenic population dynamics during startup of a full-scale anaerobic sequencing batch reactor treating swine waste. Water Research, 36 (18), 4648–4654. Appels, L., Baeyens, J., Degreve, J., & Dewil, R. (2008). Principles and potential of the anaerobic digestion of waste-activated sludge. Progress in Energy and Combustion Science, 34 (6), 755–781. Aquafin (2010). Annual report. Tech. rep., Aquafin. 136

Babel, S., & Kurniawan, T. A. (2003). Low-cost adsorbents for heavy metals uptake from contaminated water: a review. Journal of Hazardous Materials, 97 (1-3), 219–243. Batstone, D. J. (2010). Approaching anaerobic digestion with a view to resource recovery. IWA yearbook 2010 , (pp. 43–45). Batstone, D. J., Darvodelsky, P., & Keller, J. (2008a). Trends in biosolids handling technologies: economics and environmental factors. Australian Water Association, (pp. 6–9). Batstone, D. J., Tait, S., & Starrenburg, D. (2008b). Estimation of hydrolysis parameters in full-scale anaerobic digesters. Biotechnology and bioengineering, 102 (5), 1513–1520. Battistoni, P., Fatone, F., Passacantando, D., & Bolzonella, D. (2007). Application of food waste disposers and alternate cycles process in small-decentralized towns: A case study. Water Research, 41 (4), 893–903. Bernet, N. (2010). Editorial. Reviews in environmental science and biotechnology, 9 (1), 1–2. Bohnke, B. (1985). Microbial reactions in the ab process. Acta Biotechnologica, 5 (1), 45–50. Bohnke, B., Schulze-Rettmer, R., & Zuckut, S. (1998). Cost-effective reduction of high-strength wastewater by adsorption-based activated sludge technology. Water Engineering and Management, December 1998 , 31–34. Bolzonella, D., Fatone, F., Di Fabio, S., & Cecchi, F. (2009). Mesophilic, thermophilic and temperature phased anaerobic digestion of waste activated sludge. Icheap-9: 9th International Conference on Chemical and Process Engineering, Pts 1-3 , 17 , 385–390. BP (2010). Statistical review of world energy. Tech. rep., British Petroleum. Brombach, H., Weiss, G., & Fuchs, S. (2005). A new database on urban runoff pollution: comparison of separate and combined sewer systems. Water Science and Technology, 51 (2), 119–128.

137

Bunch, B., & Griffin, D. (1987). Rapid removal of colloidal substrate from domestic wastewaters. Journal (Water Pollution Control Federation), 59 (11), 957–963. Cakir, F. Y., & Stenstrom, M. K. (2005). Greenhouse gas production: A comparison between aerobic and anaerobic wastewater treatment technology. Water Research, 39 (17), 4197–4203. Carballa, M., Omil, F., Ternes, T., & Lema, J. M. (2007). Fate of pharmaceutical and personal care products (ppcps) during anaerobic digestion of sewage sludge. Water Research, 41 (10), 2139–2150. Carrere, H., Dumas, C., Battimelli, A., Batstone, D. J., Delgenes, J. P., Steyer, J. P., & Ferrer, I. (2010). Pretreatment methods to improve sludge anaerobic degradability: A review. Journal of Hazardous Materials, 183 (1-3), 1–15. Chang, D., Hur, J. M., & Chung, T. H. (1994). Digestion of municipal sludge by anaerobic sequencing batch reactor. Water Science and Technology, 30 (12), 161–170. Clark, E. (2011). Water prices rising worldwide. Earth Policy Insitute, (May 16). Comerton, A. M., Andrews, R. C., & Bagley, D. M. (2005). Evaluation of an mbrro system to produce high quality reuse water: Microbial control, dbp formation and nitrate. Water Research, 39 (16), 3982–3990. Corcoran, E., Nelleman, E., Baker, R., Bos, R., Osborn, D., & Savelli, H. (2010). Sick water? the central role of wastewater management in sustainable development. Tech. rep., UNEP / UN-Habitat. Cordell, D. (2010). The Story of Phosphorus : Sustainability implications of global phosphorus scarcity for food security. Ph.D. thesis. Cote, P., Siverns, S., & Monti, S. (2005). Comparison of membrane-based solutions for water reclamaton and desalination. Desalination, 182 (1-3), 251–257. Crockett, J., & Muntisov, M. (1995). Potable water treatment by dissolved air flotation/filtration. Modern techniques in water and wastewater treatment. Melbourne.

138

Dague, R. R., Habben, C. E., & Pidaparti, S. R. (1992). Initial studies on the anaerobic sequencing batch reactor. Water Science and Technology, 26 (9-11), 2429–2432. De Clippeleir, H., Yan, X., Verstraete, W., & Vlaeminck, S. E. (2011). Oland is feasible to treat sewage-like nitrogen concentrations at low hydraulic residence times. Applied Microbiology and Biotechnology, 90 (1), 1537–1545. Dentel, S. K., & Gossett, J. M. (1982). Effect of chemical coagulation on anaerobic digestibility of organic materials. Water Research, 16 (5), 707–718. Deublein, D., & Steinhauser, A. (2008). Biogas from waste and renewable resources. Wiley-VCH. Dewettinck, T., Van Houtte, E., Geenens, D., Van Hege, K., & Verstraete, W. (2001). Haccp (hazard analysis and critical control points) to guarantee safe water reuse and drinking water production - a case study. Water Science and Technology, 43 (12), 31–38. Diamantis, V., Antoniou, I., Melidis, P., & Aivasidis, A. (2009). Direct membrane filtration of sewage using aerated flat-sheet membranes. In 11th international conference on environmental science and technology, (pp. 230–237). Diamantis, V., Melidis, P., Aivasidis, A., Verstraete, W., & Vlaeminckx, S. (2011). Efficiency and sustainability of urban wastewater treatment with maximum separation of the solid and liquid fraction. Comprehensive biotechnology, Under review . Diamantis, V. I., Antoniou, I., Athanasoulia, E., Melidis, P., & Aivasidis, A. (2010). Recovery of reusable water from sewage using aerated flat-sheet membranes. Water Science and Technology, 62 (12), 2769–2775. Dolnicar, S., & Schafer, A. (2006). Public perception of desalinated versus recycled water in australia. Dulekgurgen, E., Dogruel, S., Karahan, ., & Orhon, D. (2006). Size distribution of wastewater cod fractions as an index for biodegradability. Water Research, 40 (2), 273–282.

139

Durham, B., Bourbigot, M. M., & Pankratz, T. (2001). Membranes as pretreatment to desalination in wastewater reuse: operating experience in the municipal and industrial sectors. Desalination, 138 (1-3), 83–90. Even-Ezra, I., Beliavski, M., Tarre, S., Dosoretz, C., & Green, M. (2011). Chemical versus biological pretreatment for membrane filtration of domestic wastewater. Desalination, 272 (1-3), 85–89. Foresti, E., Zaiat, M., & Vallero, M. (2006). Anaerobic processes as the core technology for sustainable domestic wastewater treatment: consolidated applications, new trends, perspectives, and challenges. Reviews in environmental science and biotechnology, 5 , 3–19. Forgacs, E., Cserhati, T., & Oros, G. (2004). Removal of synthetic dyes from wastewaters: a review. Environment International , 30 (7), 953–971. Gallert, C., & Winter, J. (2005). Bacterial metabolism in wastewater treatment systems, chap. 1, (pp. 1–47). Weinheim: Wiley-VCH. Gao, C., & Cardoen, D. (2011). Case study: Nieuwveer wwtp. universiteit gent course report. Gavala, H. N., Yenal, U., Skiadas, I. V., Westermann, P., & Ahring, B. K. (2003). Mesophilic and thermophilic anaerobic digestion of primary and secondary sludge. effect of pre-treatment at elevated temperature. Water Research, 37 (19), 4561–4572. Ge, H. Q., Jensen, P. D., & Batstone, D. J. (2010). Pre-treatment mechanisms during thermophilic-mesophilic temperature phased anaerobic digestion of primary sludge. Water Research, 44 (1), 123–130. Ghyoot, W., & Verstraete, W. (1997). Anaerobic digestion of primary sludge from chemical pre-precipitation. Water Science and Technology, 36 (6-7), 357–365. GOI (2008). National urban sanitation policy of the government of india. Gomec, C. Y. (2010). High-rate anaerobic treatment of domestic wastewater at ambient operating temperatures: A review on benefits and drawbacks. Journal of Environmental Science and Health Part a-Toxic/Hazardous Substances and Environmental Engineering, 45 (10), 1169–1184. 140

Guellil, A., Thomas, F., Block, J. C., Bersillon, J. L., & Ginestet, P. (2001). Transfer of organic matter between wastewater and activated sludge flocs. Water Research, 35 (1), 143–150. Guida, M., Mattei, M., Della Rocca, C., Melluso, G., & Meric, S. (2007). Optimization of alum-coagulation/flocculation for cod and tss removal from five municipal wastewater. Desalination, 211 (1-3), 113–127. Gumbo, B. (2005). Short-cutting the phosphorus cycle in urban ecosystems. Ph.D. thesis. Gutowski, T. G. (2008). Thermodynamics and recycling, a review. 2008 Ieee International Symposium on Electronics and the Environment, (pp. 101–105). Haarhoff, J., & Van der Merwe, B. (1996). Twenty five years of wastewater reclamation in windhoek, namibia. Water Science and Technology, 33 , 25–35. Haider, S., Svardal, K., Vanrolleghem, P. A., & Kroiss, H. (2003). The effect of low sludge age on wastewater fractionation (s-s,s-i). Water Science and Technology, 47 (11), 203–209. Haider, S., Vanrolleghem, P. A., & Kroiss, H. (2000). Low sludge age and its consequences for metabolisation, storage and adsorption of readily biodegradable substrate. In 1st World Congress of the International Water Association. Hamilton, G., Arzbaecher, C., Ehrhard, R., & Murphy, J. (2009). Driving energy efficiency in the us water and wastewater industry by focusing on operating and maintenance cost reductions. Tech. rep. Han, Y., & Dague, R. R. (1995). Laboratory studies on the temperature-phased anaerobic digestion of domestic wastewater sludge. Water Environment Research, 69 (6), 1139–1143. Han, Y., Sung, S. W., & Dague, R. R. (1997). Temperature-phased anaerobic digestion of wastewater sludges. Water Science and Technology, 36 (6-7), 367– 374. Hartley, K., & Lant, P. (2006). Eliminating non-renewable co2 emissions from sewage treatment: An anaerobic migrating bed reactor pilot plant study. Biotechnology and bioengineering, 95 (3), 384–398. 141

Heffernan, B., van Lier, J. B., & van der Lubbe, J. (2011). Performance review of large scale up-flow anaerobic sludge blanket sewage treatment plants. Water Science and Technology, 63 (1), 100–107. Heidrich, E. S., Curtis, T. P., & Dolfing, J. (2011). Determination of the internal chemical energy of wastewater. Environmental Science and Technology, 45 (2), 827–832. Hjorth, M., Christensen, K. V., Christensen, M. L., & Sommer, S. G. (2010). Solidliquid separation of animal slurry in theory and practice. a review. Agronomy for Sustainable Development, 30 (1), 153–180. Huang, J., & Li, L. (2000). An innovative approach to maximize primary treatment performance. Water Science and Technology, 42 (12), 209–222. Hutton, G., & Bartram, J. (2008). Global costs of attaining the millennium development goal for water supply and sanitation. Bulletin of the World Health Organization, 86 , 1319. Jimenez, B., & Asano, T. (2008). Water reclamation and reuse around the world , (pp. 3–28). London: IWA Publishing. Jimenez, B., Chavez, A., Leyva, A., & Tchobanoglous, G. (2000). Sand and synthetic medium filtration of advanced primary treatment effluent from mexico city. Water Research, 34 (2), 473–480. Judd, S. (2006). The MBR book: Principles and applications of membrane bioreactors in water and wastewater applications. Elsevier. Kim, J., Kim, K., Ye, H., Lee, E., Shin, C., McCarty, P. L., & Bae, J. (2011). Anaerobic fluidized bed membrane bioreactor for wastewater treatment. Environmental Science and Technology, 45 (2), 576–581. Kujawa-Roeleveld, K., & Zeeman, G. (2006). Anaerobic treatment in decentralised and source-separation-based sanitation concepts. Reviews in environmental science and biotechnology, 5 , 115–119. Kumar, M. (2009). Reclamation and reuse of treated municipal wastewater: an option to mitigate water stress. Current Science, 96 (7), 886–889.

142

Laird, D. A., Brown, R. C., Amonette, J. E., & Lehmann, J. (2009). Review of the pyrolysis platform for coproducing bio-oil and biochar. Biofuels, Bioproducts and Biorefining, (pp. 547–62). Leal, L. H., Temmink, H., Zeeman, G., & Buisman, C. J. N. (2010). Bioflocculation of grey water for improved energy recovery within decentralized sanitation concepts. Bioresource Technology, 101 (23), 9065–9070. Leblanc, R., Matthews, P., & Richard, R. (2008). Global atlas of excreta, wastewater sludge, and biosolids management: moving forward the sustainable and welcome uses of a global resource. Tech. rep., UN-Habitat Greater Moncton sewerage commission. Lester, J. N., Soares, A., Martin, D., Harper, P., Jefferson, B., Brigg, J., Wood, E., & Cartmell, E. (2009). A novel approach to the anaerobic treatment of municipal wastewater in temperate climates through primary sludge fortification. Environmental Technology, 30 (10), 985–994. Lettinga, G., & Pol, L. W. H. (1991). Uasb-process design for various types of wastewaters. Water Science and Technology, 24 (8), 87–107. Levine, A., Tchobanoglous, G., & Asano, T. (1985). Characterization of the size distribution of contaminants in wastewater treatment and reuse implications. Journal Water Pollution Control Federation, 57 , 805–816. LFUW (2001). Energieoptimierung van klaranlagen - detailuntersuching von 21 anlagen. Tech. rep., Bundesministerium fur Land-und Forstwirtschaft Umwelt und Wasserwirtschaft. Liu, S.-G., Bing-Jie, N., Lin, W., Yong, T., & Han-Qing, Y. (2009). Contactadsorption-regeneration-stabilization process for the treatment of municipal wastewater. Journal of Water and Environmental Technology, (pp. 83–90). Lv, W., Schanbacher, F. L., & Yu, Z. (2010). Putting microbes to work in sequence: Recent advances in temperature-phased anaerobic digestion processes. Bioresource Technology, 101 (24), 9409–9414. Magara, Y., Nambu, S., & Utosawa, K. (1976). Biochemical and physicalproperties of an activated-sludge on settling characteristics. Water Research, 10 (1), 71–77. 143

Mahmoud, N., Zeeman, G., Gijzen, H., & Lettinga, G. (2004). Anaerobic stabilisation and conversion of biopolymers in primary sludge - effect of temperature and sludge retention time. Water Research, 38 (4), 983–991. Marais, G. V. R., & Ekama, G. A. (1976). The activated sludge process - part i steady state behaviour. Water SA, 2 (4), 163–200. Martin, M. A., De la Rubia, M. A., Martin, A., Borja, R., Montalvo, S., & Sanchez, E. (2010). Kinetic evaluation of the psychrophylic anaerobic digestion of synthetic domestic sewage using an upflow filter. Bioresource Technology, 101 (1), 131–137. Maurer, M., Schwegler, P., & Larsen, T. A. (2003). Nutrients in urine: energetic aspects of removal and recovery. Water Science and Technology, 48 (1), 37–46. McAdam, E. J., Luffler, D., Martin-Garcia, N., Eusebi, A. L., Lester, J. N., Jefferson, B., & Cartmell, E. (2011). Integrating anaerobic processes into wastewater treatment. Water Science and Technology, 63 (7), 1459–1466. Melin, T., Jefferson, B., Bixio, D., Thoeye, C., De Wilde, W., De Koning, J., van der Graaf, J., & Wintgens, T. (2006). Membrane bioreactor technology for wastewater treatment and reuse. Desalination, 187 (1-3), 271–282. Moller, H., Hansen, J., & Sorensen, C. (2007). Nutrient recovery by solid-liquid separation and methane productivity of solids. Transactions of the ASABE , 50 (1), 193–200. Muga, H. E., & Mihelcic, J. R. (2008). Sustainability of wastewater treatment technologies. Journal of Environmental Management, 88 (3), 437–447. Mulder, A. (2003). The quest for sustainable nitrogen removal technologies. Water Science and Technology, 48 (1), 67–75. Muller, E., & Kobel, B. (2004). Energetische bestandsaufnahme an klaranlagen in nordrhein-westfalen mit 30 millionen einwohnerwerten - energie-benchmarking und sparpotenziale. Korrespondenz Abwasser , 51 , 625–631. Ndegwa, P. M., Hamilton, D. W., Lalman, J. A., & Cumba, H. J. (2008). Effects of cycle-frequency and temperature on the performance of anaerobic sequencing batch reactors (asbrs) treating swine waste. Bioresource Technology, 99 (6), 1972–1980. 144

Neis, U., & Tiehm, A. (1997). Particle size analysis in primary and secondary waste water effluents. Water Science and Technology, 36 (4), 151–158. Ng, H. Y., & Hermanowicz, S. W. (2005). Membrane bioreactor operation at short solids retention times: performance and biomass characteristics. Water Research, 39 (6), 981–992. Nielsen, P. H. (1996). Adsorption of ammonium to activated sludge. Water Research, 30 (3), 762–764. Odegaard, H. (1998). Optimised particle separation in the primary step of wastewater treatment. Water Science and Technology, 37 (10), 43–53. Odegaard, H. (2001). The use of dissolved air flotation in municipal wastewater treatment. Water Science and Technology, 43 (8), 75–81. OECD (2009). Managing water for all: an oecd perspective on pricing and financing. Tech. rep., OECD. Oropeza, M. R., Cabirol, N., Ortega, S., Ortiz, L. P. C., & Noyola, A. (2001). Removal of fecal indicator organisms and parasites (fecal coliforms and helminth eggs) from municipal biologic sludge by anaerobic mesophilic and thermophilic digestion. Water Science and Technology, 44 (4), 97–101. Pasqualino, J. C., Meneses, M., & Castells, F. (2011). Life cycle assessment of urban wastewater reclamation and reuse alternatives. Journal of Industrial Ecology, 15 (1), 49–63. Pavan, P., Bolzonella, D., Battistoni, E., & Cecchi, F. (2007). Anaerobic codigestion of sludge with other organic wastes in small wastewater treatment plants: an economic considerations evaluation. Water Science and Technology, 56 (10), 45–53. Ponsa, S., Ferrer, I., Vazquez, F., & Font, X. (2008). Optimization of the hydrolytic-acidogenic anaerobic digestion stage (55 degrees c) of sewage sludge: Influence of ph and solid content. Water Research, 42 (14), 3972–3980. Powers, J. J., Howell, A. J., & Vacinek, S. J. (1973). Heat of combustion of cells of pseudomonas-fluorescens. Applied Microbiology, 25 (4), 689–690.

145

PRB (2009). World population data sheet. Tech. rep., Population Reference Bureau. PUB (2011). Public utilities board - newater. singapore, may 15 2011. Ragauskas, A. J., Williams, C. K., Davison, B. H., Britovsek, G., Cairney, J., Eckert, C. A., Frederick, W. J., Hallett, J. P., Leak, D. J., Liotta, C. L., Mielenz, J. R., Murphy, R., Templer, R., & Tschaplinski, T. (2006). The path forward for biofuels and biomaterials. Science, 311 (5760), 484–489. Raghupatti, U., Pancholy, A., Srivastav, V., Ahmed, M., & Nigam, A. (2005). Status of water supply, sanitation and solid waste management in urban areas. central public health and environmental engineering organisation, ministry of urban development, government of india. Ravazzini, A. M., van Nieuwenhuijzen, A. F., & van der Graaf, J. (2005). Direct ultrafiltration of municipal wastewater: comparison between filtration of raw sewage and primary clarifier effluent. Desalination, 178 (1-3), 51–62. Reinhardt, G., & Filmore, L. (2009). Energy opportunities in wastewater and biosolids. Tech. rep., Water environment research foundation. Riau, V., de la Rubia, M. A., & Perez, M. (2010). Temperature-phased anaerobic digestion (tpad) to obtain class a biosolids. a discontinuous study. Bioresource Technology, 101 (1), 65–70. Riffat, R., & Dague, R. R. (1995). Laboratory studies on the anaerobic biosorption process. Water Environment Research, 67 (7), 1104–1110. Ripley, L. E., Boyle, W. C., & Converse, J. C. (1986). Improved alkalimetric monitoring for anaerobic digestion of high-strength wastes. Journal WPCF , 58 (5), 406–411. Rohilla, S. (2010). Challenges in urban water and sewage management. In Decentralized wastewater treatment and reuse. Centre for Science and Environment, Delhi. Rojas, J. C. M. (2010). Anaerobic digestion of raw wastewater pre-treated by up-concentration. master thesis. universiteit gent.

146

Rulkens, W. (2008). Sewage sludge as a biomass resource for the production of energy: Overview and assessment of the various options. Energy and Fuels, 22 (1), 9–15. Salome, A., & Eggers, E. (1997). Ab-systemen, een inventarisatie. Tech. rep., STORA. Santos, A., & Judd, S. (2010). The fate of metals in wastewater treated by the activated sludge process and membrane bioreactors: A brief review. Journal of Environmental Monitoring, 12 (1), 110–118. Sato, N., Okubo, T., Onodera, T., Ohashi, A., & Harada, H. (2006). Prospects for a self-sustainable sewage treatment system: A case study on full-scale uasb system in india’s yamuna river basin. Journal of Environmental Management, 80 (3), 198–207. Seghezzo, L., Zeeman, G., van Lier, J. B., Hamelers, H. V. M., & Lettinga, G. (1998). A review: The anaerobic treatment of sewage in uasb and egsb reactors. Bioresource Technology, 65 (3), 175–190. SenterNovem (2006). Rioolwaterzuiveringsinrichtingen. Tech. rep., Ministerie van Volkshuisvesting, Ruimtelijke Ordening en Milieubeheer. Shannon, M. A., Bohn, P. W., Elimelech, M., Georgiadis, J. G., Marinas, B. J., & Mayes, A. M. (2008). Science and technology for water purification in the coming decades. Nature, 452 (7185), 301–310. Sheintuch, M., Lev, O., Einav, P., & Rubin, E. (1986). Role of exocellular polymer in the design of activated-sludge. Biotechnology and bioengineering, 28 (10), 1564–1576. Sheng, G. P., Yu, H. Q., & Li, X. Y. (2010). Extracellular polymeric substances (eps) of microbial aggregates in biological wastewater treatment systems: A review. Biotechnology Advances, 28 (6), 882–894. Shimamura, K., Ishikawa, H., Mizuoka, A., & Hirasawa, I. (2008). Development of a process for the recovery of phosphorus resource from digested sludge by crystallization technology. Water Science and Technology, 57 (3), 451–456.

147

Shizas, I., & Bagley, D. M. (2004). Experimental determination of energy content of unknown organics in municipal wastewater streams. Journal of Energy Engineering-Asce, 130 (2), 45–53. Siegrist, H., Salzgeber, D., Eugster, J., & Joss, A. (2008). Anammox brings wwtp closer to energy autarky due to increased biogas production and reduced aeration energy for n-removal. Water Science and Technology, 57 (3), 383–388. Smith, J. A., & Carliell-Marquet, C. M. (2008). The digestibility of iron-dosed activated sludge. Bioresource Technology, 99 (18), 8585–8592. Smits, M. (2009). Strategieen met betrekking tot de procesoptimalisatie van anaerobe vergisting. master thesis. universiteit gent. Stark, K. (2005). Phosphorus release and recovery from treated sludge. Ph.D. thesis. STOWA (2005). Slibketen studie. Tech. rep., STOWA. Suh, Y. J., & Rousseaux, P. (2002). An lca of alternative wastewater sludge treatment scenarios. Resources Conservation and Recycling, 35 (3), 191–200. Sung, S. W., & Dague, R. R. (1995). Laboratory studies on the anaerobic sequencing batch reactor. Water Environment Research, 67 (3), 294–301. Sutton, P. M., Melcer, H., Schraa, O. J., & Togna, A. P. (2011). Treating municipal wastewater with the goal of resource recovery. Water Science and Technology, 63 (1), 25–31. Tan, K. N., & Chua, H. (1997). Cod adsorption capacity of the activated sludge its determination and application in the activated sludge process. Environmental Monitoring and Assessment, 44 (1-3), 211–217. Thornberg, D., & Johansen, N. (2010). Heat recovery from wastewater effluent a feasibility study. Montreal, Canada. Tinker, P. B. (1977). Economy and chemistry of phosphorus. Nature, 270 (5633), 103–104. Tripathi, M. (2007). Life-cycle energy and emissions for municipal water and wastewater services: case studies of treatment plants in US . Ph.D. thesis. 148

Trussell, R. S., Merlo, R. P., Hermanowicz, S. W., & Jenkins, D. (2006). The effect of organic loading on process performance and membrane fouling in a submerged membrane bioreactor treating municipal wastewater. Water Research, 40 (18), 3479–3482. UNEP (2004). Freshwater in europe - facts, figures and maps. Tech. rep., Division of early warning and assesment office for Europe (DEWA-Europe) of UNEP, Geneva. UNEP (2011). Towards a green economy: Pathways to sustainable development and poverty eradication. Tech. rep., UNEP. USEPA (1996). Clean water needs survey report to congress. assesment of needs for publicly owned wastewater treatment facilities, correction of combined sewer overflows, and management of storm water and nonpoint source pollution in the united states. Tech. rep., USEPA. Van der Meeren, P. (2010). Membrane processes in environmental technology. Faculty of bioscience engineering, Ghent university. Van Houtte, E., & Verbauwhede, J. (2008). Operational experience with indirect potable reuse at the flemish coast. Desalination, 218 (1-3), 198–207. van Voorthuizen, E., Zwijnenburg, A., van der Meer, W., & Temmink, H. (2008). Biological black water treatment combined with membrane separation. Water Research, 42 (16), 4334–4340. van Voorthuizen, E. M., Zwijnenburg, A., & Wessling, M. (2005). Nutrient removal by nf and ro membranes in a decentralized sanitation system. Water Research, 39 (15), 3657–3667. Van Wesenbeeck, K. (2010). Opconcentratie van huishoudelijk afvalwater voor energierecuperatie. master thesis. universiteit gent. Vandenburgh, S. R., & Ellis, T. G. (2002). Effect of varying solids concentration and organic loading on the performance of temperature phased anaerobic digestion process. Water Environment Research, 74 (2), 142–148. Versprille, A. I., Zuurveen, B., & Stein, T. (1985). The a-b process - a novel 2 stage waste-water treatment system. Water Science and Technology, 17 (2-3), 235–246. 149

Verstraete, W. (2010). Biotechnological processes in environmental sanitation, lecture notes.. Faculty of bioengineering sciences, University of Ghent. Verstraete, W. (2011). Microbial reuse technology, lecture notes. Faculty of bioengineering sciences, University of Ghent. Verstraete, W., de Caveye, P. V., & Diamantis, V. (2009). Maximum use of resources present in domestic ”used water”. Bioresource Technology, 100 (23), 5537–5545. Verstraete, W., Morgan-Sagastume, F., Aiyuk, S., Waweru, M., Rabaey, K., & Lissens, G. (2005). Anaerobic digestion as a core technology in sustainable management of organic matter. Water Science and Technology, 52 (1-2), 59–66. Verstraete, W., & Vlaeminckx, S. (2011). Zerowastewater: Short-cycling of wastewater resources for sustainable cities of the future. International Journal of Sustainable Development and World Ecology, 18 (3), 253–264. Wang, Z. J., Wang, W., Zhang, X. H., & Zhang, G. M. (2009). Digestion of thermally hydrolyzed sewage sludge by anaerobic sequencing batch reactor. Journal of Hazardous Materials, 162 (2-3), 799–803. Wanner, O., Panagiotidis, V., Clavadetscher, P., & Siegrist, H. (2005). Effect of heat recovery from raw wastewater on nitrification and nitrogen removal in activated sludge plants. Water Research, 39 (19), 4725–4734. Watanabe, H., Kitamura, T., Ochi, S., & Ozaki, M. (1997). Inactivation of pathogenic bacteria under mesophilic and thermophilic conditions. Water Science and Technology, 36 (6-7), 25–32. Water Resources Group, . (2009). Charting our water future: Economic frameworks to inform decision making: The economics of water resources. Tech. rep., McKinsey and Company. WB (2011). Gni per capita. world bank report. may 14 2011. Wett, B., Buchauer, K., & Fimml, C. (2007). Energy self-sufficiency as a feasible concept for wastewater treatment systems. In IWA Leading Edge Technology Conference. WHO/UNICEF (2010). Progress on sanitation and drinking-water: 2010 update. 150

Woon, W., & Leung, F. (1998). Industrial centrifugation technology. McGraw Hill. Yoon, K., Kim, K., Wang, X., Fang, D., Hsiao, B. S., & Chu, B. (2006). High flux ultrafiltration membranes based on electrospun nanofibrous pan scaffolds and chitosan coating. Polymer , 47 (7), 2434–2441. Zabranska, J., Stepova, J., Wachtl, R., Jenicek, P., & Dohanyos, M. (2000). The activity of anaerobic biomass in thermophilic and mesophilic digesters at different loading rates. Water Science and Technology, 42 (9), 49–56. Zamalloa, C., Vulsteke, E., Albrecht, J., & Verstraete, W. (2011). The technoeconomic potential of renewable energy through the anaerobic digestion of microalgae. Bioresource Technology, 102 (2), 1149–1158. Zeeman, G., Kujawa, K., de Mes, T., Hernandez, L., de Graaff, M., Abu-Ghunmi, L., Mels, A., Meulman, B., Temmink, H., Buisman, C., van Lier, J., & Lettinga, G. (2008). Anaerobic treatment as a core technology for energy, nutrients and water recovery from source-separated domestic waste(water). Water Science and Technology, 57 (8), 1207–1212. Zessner, M., Lampert, C., Kroiss, H., & Lindtner, S. (2010). Cost comparison of wastewater in danubian countries. Water Science and Technology, 62 (2), 223–230. Zhang, Z.-b., Zhao, J.-f., Xia, S.-q., Liu, C.-q., & Kang, X.-s. (2007). Particle size distribution and removal by a chemical-biological flocculation process. Journal of Environmental Sciences, 19 (5), 559–563. Zhao, W., Ting, Y. P., Chen, J. P., Xing, C. H., & Shi, S. Q. (2001). Advanced primary treatment of waste water using a bio-flocculation-adsorption sedimentation process. Acta Biotechnologica, 20 (1), 53–64.

151