COUPLING A-STAGE TECHNOLOGY with DISSOLVED AIR FLOTATION (DAF) for INCREASED ORGANICS REMOVAL and COMPACT SLUDGE PRODUCTION at PILOT SCALE Aantal Woorden: 22531

COUPLING A-STAGE TECHNOLOGY WITH DISSOLVED AIR FLOTATION (DAF) FOR INCREASED ORGANICS REMOVAL AND COMPACT SLUDGE PRODUCTION AT PILOT SCALE Aantal woorden: 22531

Stijn Decru Stamnummer: 01202687

Promotor: Prof. Dr. ir. Korneel Rabaey Prof. Dr. ir. Bart De Gusseme Copromotor: dr. ir. Jo de Vrieze, ir. Cristina Cagnetta

Masterproef voorgelegd voor het behalen van de graad Master of Science in de bio- ingenieurswetenschappen: milieutechnologie

Academiejaar: 2016 - 2017

“The author and the promoter give the permission to use this thesis for consultation and to copy parts of it for personal use. Every other use is subject to the copyright laws, more specifically the source must be extensively specified when using results from this thesis.”

“De auteur en de promotor geven de toelating deze scriptie voor consultatie beschikbaar te stellen en delen ervan te kopiëren voor persoonlijk gebruik. Elk ander gebruik valt onder de beperkingen van het auteursrecht, in het bijzonder met betrekking tot de verplichting de bron te vermelden bij het aanhalen van resultaten uit deze scriptie.”

Gent, juni 2017

De promotoren, De auteur,

Prof. dr. ir. Korneel Rabaey Stijn Decru

Prof. dr. ir. Bart De Gusseme

Dr. ir. Jo de Vrieze

Cristina Cagnetta

Acknowledgements

The year that lies behind me is one in which I learned a tremendous amount of things. The experience I gained during the HRAS-DAF pilot work told me that you should never expect anything to work properly from the beginning and that really anything can fail. The problem solving was challenging but kept everyone closely involved to the project. Working in the lab gave me insights in the chemical analyses that are necessary to run a biotechnological plant.

First of all I would like to thank Cristina for her enthusiasm and patience she showed towards me. She was always available for questions and if something wasn’t clear from the first time, she would eagerly explain it to me twice. She steered me in the good way and gave me the critical insights that were necessary to build the story of my thesis. I am grateful to Jo Devrieze for the advice he gave me on the anaerobic digestion of A-stage sludge and for reading my thesis. He was also the one who supplied me with vacutainers without which I couldn’t have brought a single gas sample safely back to the faculty for analysis.

I remember feeling like an engineer for the first time when I attended Professor Rabaey courses on environmental technology during the third year. It was during these courses that he awakened my interest in biotechnological processes for wastewater treatment and made me choose one of his topics. His advice was always to the point and helped a lot to keep the focus and define the paths to take. My first encounter with professor De Gusseme was during the course ‘microbial resource recovery’. I admired his practical approach and strong link with the industry and his involvement was an enriching complement to the story of the thesis.

To the people from Aquafin. Thank you Bart & Francis for sustaining the HRAS-DAF pilot together with me and answering my questions and Marjolein for guiding this project. Special thanks to Francis for bringing samples several times from Aartselaar to the lab in Gent. Thank you Marc, Solo (Souleymane), Kris (Christiaan) and Eric who were there to come up with fast and practical solutions if there was a mechanical problem with the HRAS-DAF pilot.

To the people from Nijhuis for making this thesis possible by providing the DAF unit and doing necessary adaptions to make everything work. iii

Further, I would like to thank everyone from CMET. The people I could chat or have fun with during my work in the lab, have a coffee together or provide me with answers to my questions.

My fellow leaders at the youth movement (chiro), I would like to thank you for letting me do one extra year, knowing that I would be less closely involved because of the thesis. Once again we had fantastic moments, and you tolerated the many times I was absent. It’s like Timmy Simons once said, it’s better to quit to late than to early.

Finally, to my parents, sister and smaller brother, I am grateful that you pushed me through the last weeks of the thesis. You supported me when I needed it and gave me courage to go further.

Abstract

Current wastewater treatment processes such as the conventional activated sludge process (CAS) focus solely on the removal of organics and nutrients and are therefore not sustainable. The organics present in the wastewater possesses an abundant resource potential and the focus should shift from solely environmental protection to the recovery of energy and materials.

High rate activated sludge (HRAS) systems such as the A-stage or high rate contact stabilisation (HiCS) process are able to capture a significant fraction of these organics and transfer them to the sludge. Although some energy is lost via microbial respiration, most of it is captured in the sludge. Part of the energy stored in the sludge can be recovered under the form of biogas (CH4) via anaerobic digestion (AD). The high rate activated sludge is more biodegradable and conversion efficiencies during AD are up to 2.5 times higher than for excess sludge produced during CAS. Unfortunately, the high food to microorganism ratios specific to these high rate systems lead to a poor settling sludge.

Solid/liquid separation of HRAS via conventional settling generates diluted sludge (± 10 g L- 1). Further thickening is thus required prior to AD. Moreover, a considerable part of the organics (up to 50 %) are not separated in the settler and leave with the effluent. In the view of optimal energy recovery, a more efficient technique for solid/liquid separation is needed. During this thesis dissolved air flotation (DAF) was applied for solid/liquid separation and it was shown that the HRAS-DAF combination is feasible at pilot scale but needs close monitoring.

The HRAS-DAF pilot was able to obtain high organics removal and produce concentrated sludge (21 – 47 g COD L-1). The sludge was anaerobically digested to recover the energy present in the sludge under the form of biogas. When a single polymer was used to assist flocculation, conversion efficiencies to methane were 58 – 68 %. This is comparable to a conventional HRAS where settling is used for the solids separation. When a dual polymer system was used conversion efficiencies were lower, 40 – 42 %. This was likely due to phosphorous or trace elements depletion or to the polymers interfering with hydrolysis and digestion of sludge during AD.

Nederlandse samenvatting

De huidige technieken voor waterzuivering zoals het actief slib proces (CAS) richten zich enkel op de verwijdering van organisch materiaal en nutriënten en zijn daarom niet duurzaam. De aanwezige organische stoffen in het afvalwater beschikken over een overvloedig potentieel voor grondstof recuperatie en de focus moet overgaan van enkel milieubescherming naar het terugwinnen van energie en materialen.

Hoog belast geactiveerd slib (HRAS) systemen, zoals de A-trap (A-stage) of het hoog belast contact stabilisatie proces (HiCS), kunnen een significante fractie van deze organische stoffen opnemen en overbrengen naar het slib. Hoewel sommige energie via microbiële respiratie verloren gaat, wordt het grootste deel in het slib vastgelegd. Een deel van de energie die in het slib wordt opgeslagen, kan vervolgens worden teruggewonnen onder de vorm van biogas (CH4) via anaerobe vergisting. Het hoog belaste geactiveerd slib is beter biologisch afbreekbaar en de omzettingsefficiënties tijdens anaerobe vergisting zijn maximaal 2,5 keer hoger dan voor slib geproduceerd tijdens CAS. Helaas leiden de hoge ‘feed tot micro-organismen’-verhoudingen (F/M) die specifiek zijn voor deze hoog belaste systemen tot een slecht bezinkbaar slib.

Vast/vloeistof scheiding van HRAS slib via conventionele sedimentatie zorgt voor laag geconcentreerd slib (± 10 g L-1). Verdere verdikking is dus vereist voorafgaand aan anaerobe vergisting (AD). Bovendien wordt een aanzienlijk deel van de organische stoffen (tot 50%) niet gescheiden in de bezinkingstank maar meegenomen in het effluent. Met het oog op optimale energieherwinning is er een efficiëntere techniek voor vast/vloeistof scheiding nodig. Tijdens deze thesis werd opgeloste lucht flotatie (DAF) toegepast voor vast/vloeistof scheiding en het bleek dat de HRAS-DAF combinatie haalbaar is op pilootschaal mits nauwkeurige opvolging.

De HRAS-DAF piloot kon hoge organische verwijdering verkrijgen en geconcentreerd slib produceren (21 - 47 g COD L-1). Het slib werd anaeroob vergist om de aanwezige energie in het slib in de vorm van biogas te recupereren. Wanneer enkel anionisch polymeer werd gebruikt voor flocculatie, was de conversie efficiëntie naar methaan 58 – 68%. Dit is vergelijkbaar met een conventionele HRAS, waar bezinking gebruikt werd voor de vaste-stofafscheiding. Wanneer een dubbel polymeer systeem werd gebruikt, was de conversie-efficiëntie lager, 40 – 42%. Dit was waarschijnlijk te wijten aan de uitputting van fosfor of sporenelementen of aan de polymeren die interfereren met hydrolyse en vertering van slib tijdens AD. vi

Table of contents

Acknowledgements ...... iii

Abstract ...... v

Nederlandse samenvatting ...... vi

Table of contents ...... vii

List of abbreviations ...... xi

PART 1: LITERATURE REVIEW ...... 1

1 Municipal wastewater treatment ...... 2

Characteristics of Municipal wastewater ...... 2

Conventional Activated Sludge (CAS) process ...... 3

Drawbacks related with the CAS process ...... 4

Achieving sustainability through energy recovery ...... 5

2 High rate activated sludge processes ...... 7

The A/B-process ...... 7

Benefits and drawbacks from A-stage technology ...... 9

The high rate contact stabilisation process ...... 10

3 Coagulation and flocculation ...... 11

4 Dissolved air flotation (DAF) ...... 13

Working principle ...... 13

Applications of DAF at MWTP’s ...... 15

5 Anaerobic digestion of high rate sludge ...... 16

The role of anaerobic digestion at the modern WWTP ...... 16

General process principles ...... 17

6 Research questions ...... 19 vii

PART 2: MATERIALS & METHODS ...... 20

1 Pilot scale HRAS-DAF set-up ...... 21

Configuration 1: A-stage ...... 24

1.1.1 Treatment 1: dosing of FeCl3 and A130hp ...... 24

1.1.2 Treatment 2: dosing of FeCl3, C492 and A130hp ...... 24

Configuration 2: HiCS operation (treatment 3) ...... 25

MLSS characteristics ...... 25

Digesters ...... 26

Sampling ...... 27

2 Jar testing ...... 28

Experiment 1: Optimal polymer selection for flotation ...... 28

Experiment 2: Optimal flocculant dosage (T1) ...... 28

Experiment 3: Optimal flocculant dosage (T2) ...... 29

Experiment 4: Optimal coagulant & flocculant dosage (T3)...... 29

3 BMP tests ...... 29

4 Fed batch digesters ...... 32

5 Chemical analyses ...... 32

COD analysis ...... 32

Solids analysis ...... 33

VFA analysis ...... 33

Gas analysis ...... 34

Anion analysis ...... 34

Nitrogen analysis ...... 34

viii

PART 3: RESULTS ...... 35

1 Pilot scale HRAS-DAF ...... 36

Organics removal ...... 36

1.1.1 A-stage-DAF T1 ...... 38

1.1.2 A-stage-DAF T2 ...... 38

1.1.3 HiCS-DAF T3 ...... 38

DAF separation efficiency ...... 39

1.2.1 A-stage-DAF T1 ...... 41

1.2.2 A-stage-DAF T2 ...... 41

1.2.3 HiCS-DAF T3 ...... 41

1.2.4 Effluent characteristics ...... 41

Sludge characteristics ...... 44

COD balance ...... 46

Digester performance ...... 47

2 Jar tests ...... 48

Experiment 1: optimal polymer selection for flotation ...... 48

Experiment 2: optimal flocculant dosing (T1) ...... 48

Experiment 3: optimal flocculant dosing (T2) ...... 49

Experiment 4: optimal FeCl3 and C492 dosing (T3) ...... 51

3 BMP tests ...... 52

BMP T1 ...... 52

BMP T2 ...... 52

BMP T3 ...... 53

4 Fed batch digesters ...... 54

PART 4: DISCUSSION ...... 55

1 Pilot scale HRAS-DAF ...... 56

Organics removal ...... 57

1.1.1 A-stage-DAF T1 ...... 57

1.1.2 A-stage-DAF T2 ...... 59

1.1.3 HiCS-DAF T3 ...... 60

DAF performance ...... 60

1.2.1 A-stage-DAF T1 ...... 61

1.2.2 A-stage-DAF T2 ...... 61

1.2.3 HiCS-DAF T3 ...... 61

1.2.4 Effluent characteristics ...... 61

Comparison between HRAS-DAF and HRAS-settling ...... 62

Sludge characteristics ...... 64

COD balance ...... 65

2 Jar tests ...... 66

Experiment 1: optimal polymer selection for flotation ...... 66

Experiment 2: optimal flocculant dosing (T1) ...... 66

Experiment 3: optimal flocculant dosing (T2) ...... 66

Experiment 4: optimal FeCl3 and C492 dosing (T3) ...... 67

3 BMP tests ...... 67

BMP T1 ...... 67

BMP T2 ...... 67

BMP T3 ...... 69

4 Fed batch digesters ...... 69

5 General conclusions ...... 70

6 Future work and perspectives ...... 72

PART 5: REFERENCE LIST ...... 74

List of abbreviations

A/B Adsorption / Bio-oxidation

AD Anaerobic Digestion

BOD Biological Oxygen Demand

CAS Conventional Activated Sludge

cCOD colloidal COD

CD Charge Density

CEPT Chemically Enhanced Primary Treatment

CHP Combined Heat & Power

COD Chemical Oxygen Demand

DO Dissolved Oxygen

EPS Exo Polymeric Substances

F/M Food to Micro-organism

HRAS High Rate Activated Sludge

HRT Hydraulic Retention Time

MLSS Mixed Liquor Suspended Solids

MO Micro Organism

MW Molecular Weight

PAC Poly Aluminium Chloride

pCOD particulate COD

PE People Equivalents

PFF Plug Flow Flocculator

RAS Return Activated Sludge

xi sCOD soluble COD

SRT Sludge Retention Time

SVI Sludge Volume Index

TKN Total Kjeldahl Nitrogen

TP Total Phosphorous

TSS Total Suspended Solids

VSS Volatile Suspended Solids

WAS Waste Activated Sludge

xii

PART 1: LITERATURE REVIEW

1 Municipal wastewater treatment

The United Nations World Water Development Report from 2015 states that global water use is increasing and potable water is becoming scarcer (Connor, 2015). Population growth, industrialisation and a higher consumption per capita are the main reasons for this. In the meantime, the production of wastewater is increasing. According to the 2030 Water Resources Group, the world will only have 60% of the water it needs by 2030 without significant global policy change (Boccaletti, 2009). In industrialised countries where there is enough potable water, there is a competition between the households and the industries. Also in Flanders, the high pressures on the groundwater reserves has led to stricter regulations for the industry and to incentives for water recovery from used waters (Verstraete et al., 2016).

Wastewater treatment consumes considerable amounts of energy and materials to meet discharge standards. Instead, wastewater should be looked at as a source of valuable resources. An important step towards self-sufficient wastewater treatment is controlling and managing the carbon flows to minimize oxidation and maximise sludge harvest (Rahman, 2016).

Characteristics of Municipal wastewater

Municipal wastewater discharged from houses, agriculture and industries contains contaminants that need to be removed. These contaminants are organics (carbohydrates, lipids, proteins), nutrients (nitrogen and phosphorous) and other chemicals. Medium strength wastewater contains about 500 mg L-1 chemical oxygen demand (COD), 40 mg L-1 total Kjeldahl nitrogen (TKN) and 8 mg L-1 total phosphorous (TP) (Tchobanoglous & Burton, 1991). The ratios of COD:N:P are thus in the order of 10:1:0.2.

The organic matter can be divided in a settable (i.e. particulate, pCOD) and non-settable fraction (i.e. soluble, sCOD and colloidal, cCOD). For municipal wastewater, these three fractions seem to be quite constant and are on average 45 %, 31 % and 24% respectively (Guellil et al., 2001).

Conventional Activated Sludge (CAS) process

Treatment of municipal wastewater is depending on microorganisms (MO) that are implemented in biotechnological processes. These processes are of great importance for human health, food provisioning and protection of natural ecosystems. One process in particular, the ‘activated sludge process’, is very widely spread (Modin et al., 2016).

“There is not a single process that has made such a tremendous impact on human health as the activated sludge process” (Rabaey, 2014)

Activated sludge systems for wastewater treatment consist of a biological aerated reactor with a microbial community of heterotrophic and autotrophic MO (Ekama & Wentzel, 2008). Municipal wastewater is brought into contact with this community and organic matter is removed. Part of the organics are transformed into CO2 and water to provide energy for catabolic processes (Henze et al., 2001). The other part is incorporated in the cell components of the MO’s or stored as intra- and extracellular energy source.

The activated sludge process is only a part of the whole municipal wastewater treatment process. It can be divided in three different stages: primary, secondary and tertiary treatment (Tchobanoglous & Burton, 1991). Before primary treatment, large particles ( > 1 cm) are removed by a grit and grease and fats can be skimmed off to protect downstream equipment.

Figure 1: Process configuration for a typical CAS treatment plant with primary settling

During primary treatment, the wastewater passes a primary settling tank in which 60 – 65 % of the suspended solids (TSS) are removed (Constantine et al., 2012), so the organic load on the activated sludge reactor is lowered and smaller reactor volumes and less aeration are required. This was an incentive to investigate chemically enhanced primary treatment (CEPT). Using chemical coagulants (i.e. iron chloride and aluminium sulphate), high removal efficiencies for

TSS (90 %) and TP (80 – 90 %) could be achieved but the technique has to cope with high chemical costs and excess sludge production (Xu et al., 2009).

Secondary treatment is typically a CAS system. Its main components are an aerated biological reactor, a sedimentation tank and a recycle system (Figure 1) for returning part of the solids/biomass back to the biological reactor (Ekama & Wentzel, 2008). Additional aeration and a sludge retention time (SRT) of 8 – 20 days are needed for bacterial nitrification (Loosdrecht et al., 1997). Operating at long SRT means that most biodegradable organics will be oxidised. The energy content of the organic carbon is lost rather than recovered (Verstraete -1 & Philips, 1998), and electrical energy (0.45 – 1.25 kWh kg O2) is needed for aeration (Rabaey, 2014). Additional organic carbon (e.g. methanol) is often added for bacterial denitrification in the anoxic part of the reactor.

The MO grow in flocs which allows solid/liquid separation by gravitational settling in the secondary settler. Part of the sludge from the settler is recirculated and the remainder is wasted. Exopolymeric substances (EPS) play a crucial role in the floc formation. They are formed by active cell secretion, cell lysis and adsorption processes. Sludge settles less well with increasing EPS content (Y. Liu & Fang, 2003).

Tertiary treatment is typically applied to meet effluent standards commissioned by the government. Residual suspended solids can be removed by ultrafiltration and also disinfection is sometimes applied (Tchobanoglous et al., 1998).

Drawbacks related with the CAS process

Large areas are needed for settling of primary and secondary sludge. The settling behaviour of activated sludge can be measured based on the sludge volume index (SVI). A good working secondary settler concentrates the incoming MLSS (1.5 – 3.5 g TSS L-1) to secondary sludge having a content of 5 – 15 g TSS L-1 (Higgins & Novak, 1997). Doing so 5 – 15 m2 secondary settling area per 1000 Person Equivalents (PE) is needed (Henze et al., 2008). The sludge needs to be further concentrated before anaerobic digestion. This is typically don by gravity thickening and the secondary sludge concentration is raised from 5 – 15 to 60 g TSS L-1 (EPA, 2003).

Current wastewater treatment processes do not focus on energy recovery, and are not sustainable (Verstraete et al., 2009). The CAS process consumes considerable amounts of fossil fuel-derived energy and chemicals. During oxidation, the chemical energy present as COD is lost as metabolic heat. According to Garrido (2013), the energy use for CAS treatment lies between 0.3 – 0.8 kWh m-³ of which most is needed for aeration (60 – 70 %) and pumping. In addition to the high energy use, the processes themselves also generate potent greenhouse gases such as CH4 and N2O (Sheik et al., 2014).

The energy potential of medium strength wastewater with a COD content of 500 mg L-1 is around 1.9 kWh m-³ (McCarty et al., 2011). This exceeds the energy needed for treatment by a factor four. A systematic evaluation of the energy content in wastewaters is needed since there is no straight correlation between the COD content and the energy content (Heidrich et al., 2010).

Achieving sustainability through energy recovery

The focus of municipal wastewater treatment has shifted from solely protection of downstream users to environmental protection. Currently it is considered as a source of three valuable types of resources: water, nutrients and energy (Mo & Zhang, 2012). Municipal wastewater is of specific interest due to its availability in urbanized regions.

Two main strategies to achieve higher sustainability during wastewater treatment are (1) energy efficiency improvement and (2) integrated resource recovery. Measures that reduce the energy consumption such as bubble aeration and aeration control systems are being implemented in wastewater treatment plants to improve the efficiency of energy usage (Frijns et al., 2013).

Methods for energy recovery are (1) on-site energy generation through combined heat and power (CHP) based on the organic load and/or thermal heat of the wastewater, (2) nutrient recycling which offsets the environmental load associated with producing the same amount of fertilizer and (3) water reuse (Mo & Zhang, 2013). On-site energy generation must deal with a high one-time investment cost in the order of 2000 € kW-1 installed power. High amounts of biogas are needed to cover this investment cost which is not in favour of smaller treatment plants.

Wastewater treatment plants are currently large energy consumers. About 3 % of electricity consumption in 2004 in US was used for wastewater treatment and this is similar for other developed countries (Cams, 2005). Achieving energy neutrality or even contributing energy to the society would be beneficial for multiple reasons (Constantine et al., 2012):

i. Energy produced by wastewater facilities is sustainable. ii. Achieving energy neutrality is important in terms of risk management to deal with fluctuating energy prices. iii. On-site energy production enables the WWTP’s to cope with power outages.

Both capturing the energy in the dissolved organics and meeting the effluent standards would lead to high energy saving (McCarty et al., 2011). Production of biogas is the main source of energy at a municipal CAS treatment plant. The biodegradable organics present in the sludge are converted to CH4 and CO2 and new biomass. Since direct anaerobic treatment of municipal wastewater is not applicable for low to medium strength wastewater, the recovery of chemical energy can be maximised by concentration of the organic carbon in the wastewater and maximised sludge digestion (Frijns et al., 2013).

Concentration of municipal wastewater followed by anaerobic digestion of organics and maximal reuse of the mineral nutrients and water is estimated to have a total cost of 0.9 € m-3 which is similar as for CAS treatment with little or no reuse. (Verstraete et al., 2009). Other techniques for energy recuperation from sludge like incineration, gasification and pyrolysis are less commonly used (Tyagi & Lo, 2013).

2 High rate activated sludge processes

High rate activated sludge processes (HRAS) are characterized by short sludge retention times (SRT) (< 1 day), low dissolved oxygen conditions (0.5 mg L-1) and a high food to microorganism (F/M) ratio (> 2 g BOD g-1 VSS day-1) (Smitshuijzen et al., 2016). These systems are able to maximise carbon harvesting from wastewater (Rahman, 2016). Most important process parameters are the HRT, SRT and DO. The organic matter is removed through non-oxidative processes rather than being oxidized to CO2. The major mechanisms for non-oxidative removal of carbon in the HRAS are assimilation, storage and sorption (Modin et al., 2016) and will further be referred to as ‘biosorption’. Applications of HRAS processes are hampered by poor settling properties of the high-rate sludge (Bisogni & Lawrence, 1971).

In the search for energy efficiency improvements at the wastewater facility, there is also renewed interest in employing two-stage biological treatment in which carbon and nitrogen removal are decoupled. Aeration energy savings could be around 15 – 20 % and 25 – 40 % more biogas could be produced when applying a two-stage treatment compared with the CAS (Constantine et al., 2012).

The A/B-process

Two stage processes with HRAS in the first step are referred to as the A/B-process. It is an old process that was reintroduced in the 70’s by Boehnke et al. (1998) as an answer to the higher treatment standards commissioned by the government and the desire for a cost-effective technology to reduce the micro pollutants and nutrients in the wastewater. A/B is an abbreviation for ‘Adsorption-Belebungsanlagen’ and is typically translated as ‘Adsorption- activated sludge’ or ‘Adsorption-Biooxidation system’. Several plants were installed in Germany and the Netherlands (de Graaff et al., 2016). Even more stringent effluent standards led to reconfiguration of existing A/B-processes for better nitrogen and phosphorous removal. The possibility to use anammox bacteria for autotrophic nitrogen removal combined with higher energy recovery renewed the interest in the A/B-process (Kartal et al., 2010).

The A/B process is a two-stage dual sludge system that consists of a highly loaded Adsorption- stage (A-stage) and a lowly loaded Bio-oxidation-stage (B-stage). In the A-stage the F/M ratio is high compared with the CAS system, 2 – 10 g BOD g-1 VSS day-1 and 0.3 – 0.6 g BOD g-1 VSS day-1 respectively. HRT is about 30 minutes and SRT is less than 1 day (Jimenez et al., 2015). The second stage, the B-stage, has a lower F/M ratio (less than 0.1 g BOD g-1 VSS day- 1) combined with higher SRT (8 – 20 days).

Figure 2: Process configuration of an A/B system, adapted from Modin et al. (2016)

The first stage, the A-stage, is the most innovative element in the process. No primary treatment (or CEPT) is applied in the A/B-process, so the A-stage can be seen as a biologically enhanced primary treatment (Constantine et al., 2012). Removal of the organic load in this stage happens fast and is based on physico-chemical processes. Colloidal and particulate matter can be removed by biosorption on the activated sludge flocs. Biosorption capacity increases with increasing SVI, and is negatively affected by long periods of anoxia (Pujol & Canler, 1992).

Next to the physico-chemical removal of organics, also intracellular storage occurs. Removal efficiencies for COD and BOD in existing A-B WWTPs are 50 – 60 % and 40 – 80 % respectively (de Graaff et al., 2016). Lower F/M and longer SRT values yielded sludge in which carbon storage was the main process for carbon capture (Lim et al., 2015).

Operational conditions under which the A-stage performs best were investigated by de Graaff et al. (2016). Based on results from four operational A/B wastewater treatment plants in the Netherlands, it was shown that the A-stage could more effectively remove COD than a primary settler when the influent contained a high fraction (> 25 %) of sCOD. Sludge production was -1 found to be maximal at an SRT of about 0.3 days, and sufficient aeration (> 2 mg O2 L ) seemed essential for a good sCOD conversion (de Graaff et al., 2016). Removal efficiencies of pCOD and cCOD increase linearly as the EPS production increases until an EPS production of 8 approximately 80 mg COD g-1 VSS (Jimenez et al., 2015). EPS production also increased at higher specific sCOD removal rates.

Microbial communities of A-stage reactors share more core genera with the influent municipal wastewater than CAS communities (Gonzalez-Martinez et al., 2016). The influent microbial community entering the A-stage only has a short time to shift and it leaves the bioreactor almost unchanged. The longer SRT in the CAS bioreactors impacts microbial community structure more profoundly.

In the B-stage, which resembles a CAS reactor, nitrification and denitrification and mineralisation of the residual COD occurs. Enough BOD leaves with the effluent of the A-stage to provide the denitrification in the B-stage. Another approach for nitrogen removal is autotrophic nitrogen removal from low strength wastewaters (< 100 mg N L-1) at low temperatures (< 20 °C) in the main stream via the cold anammox process. This process is still under investigation for application in the main stream (Hendrickx et al., 2012).

Benefits and drawbacks from A-stage technology

The A-stage technology is promising for achieving maximal energy recovery from wastewater with minimum energy expenditure. More COD is preserved in sludge and the sludge has a higher digestibility than CAS sludge (Bolzonella et al., 2005). The lower HRT requires smaller reactor volumes and lowers the systems footprint. The A-stage is able to handle shock loads and high variations in the influent and guarantees stable operation of the B-stage (Boehnke et al., 1998). The A-stage technology can be implemented in existing installations as was shown at the Strass wastewater treatment plant in Austria. This plant proves that it is possible to achieve energy neutrality. Indeed, 11% of the calorific energy in influent organics, i.e. 54 Wh PE-1, could be converted in electricity and this was sufficient to supply the energy needed for the entire plant (Wett et al., 2007).

The high loading of A-stage systems, which coincides with high F/M ratios ( 2 – 10 g BOD g-1 VSS d-1) leads to poor settling sludge with an SVI > 250 mL g-1 (Bisogni & Lawrence, 1971). Under these conditions low concentrated sludge (± 10 g COD L-1) is yielded and a large part of

9 the COD (40 – 50 %) leaves with the effluent (De Graaff & Roest, 2012). Other techniques should be looked at to separate the sludge from the water.

Since the A-stage only removes a minor part of the nutrients via the sludge, the remaining nutrients need to be removed in the B-stage. The A stage favours the processes in B stage by: buffering shock loads (thus creating a more stable influent for the B-stage) and increasing the relative abundance of easily degradable COD (Boehnke et al., 1998). Denitrification is possible in the A-stage but only if nitrates are present in the influent.

The high rate contact stabilisation process

High-rate contact stabilization (HiCS) was defined by Meerburg et al. (2015) as a high-rate system in a contact stabilization configuration, with a minimal sludge-specific loading rate of 2 g BOD g-1 VSS d-1 and a maximal SRT of 2 days. In the HiCS system, municipal wastewater is brought into contact with the activated sludge in a contact tank at low HRT (< 30 -1 min) and under moderate DO conditions (ca. 1 mg O2 L ) (Gujer & Jenkins, 1975). Removal of organics happens mainly through sorption on the activated sludge flocs.

The sludge flocs are then separated from the water in a secondary settler. Part of this sludge is wasted and part of it is sent to a stabilisation tank that is aerated to regenerate the sludge (Figure 3). The stabilizer stabilizes and oxidizes extracellular (particulate and colloidal) and intracellular (soluble) carbon from the returned activated sludge (RAS) which is rich in carbon (Rahman et al., 2016). This induces a feast and famine regime and improves the biosorption capacity and bioflocculation capacity of the sludge (Rahman et al., 2016). This way the adsorption area and the storage capacity of the RAS is increased (Meerburg et al., 2015).

Figure 3: Process configuration of a contact stabilisation system, adapted from Modin et al. (2016) 10

3 Coagulation and flocculation

An efficient solid/liquid separation between the sludge and the effluent from the A-stage is vital for good plant operation. The first step to achieve this separation is coagulation/flocculation which is a commonly used process in wastewater treatment (John Bratby, 2006). Coagulant and flocculant are added to the mixed liquor suspended solids (MLSS) to agglomerate the smaller sludge particles into larger flocs. The performance can be checked based on turbidity measurements expressed in nephelometric turbidity units (NTU) (Aziz et al., 2007).

The coagulation working principle is based on destabilization of stable suspensions. Sludge particles bear a negative surface charge which can be measured based on the zeta potential. These surface characteristics are strongly linked with the EPS content of the sludge (Y. Liu & Fang, 2003). Charged particles repel each other and cannot grow into a larger floc. Coagulation counters the negative surface charges of the sludge flocs by applying chemicals bearing positive charges (coagulants). Three types of inorganic coagulant are mainly used: iron chloride (FeCl3), aluminium chloride (AlCl3) and poly aluminium chloride (PAC) (John Bratby, 2006).

Flocculation is the subsequent process during which the destabilised particles grow into larger flocs. This can happen spontaneously (perikinetic flocculation) but will be enhanced by agitation (orthokinetic flocculation) (Verliefde, 2014). Anionic as well as cationic polyacrylamide polymers are being applied as flocculant to enhance floc formation by binding the flocs stronger together and allowing them to grow larger. The flocculant plays a role in dewatering the sludge and lowers the need for coagulant (Aguilar et al., 2002). The flocculant molecule consists of a polymer backbone with positively or negatively charged functional groups and its most important characteristics are molecular weight (MW) and charge density (CD) (John Bratby, 2006). The increase of the MW and CD will lead to better flocculation but is limited by decreasing solubility of the polymer and increasing viscosity of the polymer solution.

Two main mechanisms of flocculation in wastewater treatment are (Bolto & Gregory, 2007): i. Polymer bridging: the polymer is adsorbed on two or more particles thus holding them together. There should be enough unoccupied space on the particle for polymer attachment. ii. Charge neutralisation: particles are most often negatively charged in which case cationic polyelectrolytes are dosed to neutralize the charge. Their application leads to the formation of electrostatic patches on the particle surface.

Phosphorous present as orthophosphate is also removed during coagulation/flocculation. This means that the phosphate content in the wastewater should be accounted for when optimizing the coagulant dosage based on TSS concentration. Removal can happen the following ways (Aguilar et al., 2002):

i. Incorporation of the phosphate into the solids in suspension. ii. Removal as phosphate precipitate with the metal ion used as coagulant.

The mixing regime is important since coagulant needs to be uniformly distributed and the growing flocs need to ‘meet’ each other. Mixing intensity is expressed as the mean velocity gradient G (s-1). A rapid mixing phase should be followed by a slow mixing phase (Adachi, 1995). Too vigorous agitation will lead to flocs disruption. Next to basins with mixers, also plug flow flocculators (PFF) can be used. Velocity gradients in this setup are induced by the continuous flow reversal in the bends of the pipes.

4 Dissolved air flotation (DAF)

Dissolved air flotation (DAF) is a solid/liquid separation process that can be used as an alternative for settling. The suspended solids are separated from the water by producing low density bubble-floc aggregates that float to the top of the DAF unit. The DAF process is already used at large scale for drinking water treatment (Haarhoff, 2008) and to treat several industrial wastewaters (Poh et al., 2014). It is also used for sludge thickening and could increase the sludge concentration to 30 – 40 g TSS L-1 (De Rijk & den Blanken, 1994) or even to 60 g TSS L-1 (EPA, 2003). The main incentives to use DAF are the lower footprint, higher removal efficiencies and denser sludge production compared with conventional settling (Ødegaard, 2001).

Working principle

The DAF process takes place in a tank (Figure 4) which is divided in two zones by a baffle. In the first zone, the contact zone, air microbubbles (10 – 100 µm) are introduced in the influent and aggregate with the flocs (Agarwal et al., 2011). Turbulent as well as laminar flow conditions occur in the contact zone and it’s configuration is of great importance for the removal efficiency. The height of the contact zone should be higher than 0.8 m and the contact zone loading should not exceed 100 m h-1 to assure good mixing but meanwhile prevent floc disruption (Lundh et al., 2002). In the second zone, the separation zone, the bubble-floc aggregates are separated from the water phase by floating to the top where they form a floating layer. This layer is scraped off mechanically (Broeders, Menkveld, et al., 2014).

Part of the effluent, called subnatans, is recycled and brought to a pressure of 400 – 600 kPa in a pressurizing vessel called the saturator (J. Edzwald, 2010). Air is dissolved in this recycle water by adding pressurized air. The principle of microbubble generation is based on the higher solubility of air in water under high pressure, according to Henry’s law. When depressurized, the recycle flow becomes supersaturated with air and the microbubbles are formed.

Figure 4: DAF unit with a contact (left) and a separation (right) zone, adapted from Broeders, Menkveld, et al. (2014)

The recycle flow is most often described in terms of the recycle ratio R which equals the recycle flow QR divided by the influent flow Q, and is typically in the order of 10 – 20 % (Lundh et al., 2002). By depressurizing the recycle water through pressure reduction nozzles in the contact zone the microbubbles are generated with most bubbles ranging from 40 to 80 µm (Leppinen & Dalziel, 2004). The microbubbles give the water a milky appearance so that this water is often referred to as ‘white water’. Increasing saturator pressure yields finer micro-bubbles but their generation must cope with high energy consumption (0.020 – 0.040 kWh m-3) (Schofield, 2001). Efforts to reduce the energy consumption by oscillating the air stream are currently under investigation (Brittle et al., 2015).

Bubble-bubble and bubble-particle interactions are affected by several forces: London-van der Waals, electrostatic and hydrophobic interactions and hydrodynamic retardation. There are two ways for formation of bubble-particle aggregates (De Rijk & den Blanken, 1994):

i. Inclusion of the air bubble into the sludge floc ii. Adsorption of the bubbles on the outside of the flocs

A coagulation and flocculation treatment prior to the DAF affects it’s performance in a strong way. Without coagulation both bubbles and particles carry negative zeta potentials. Successful

14 bubble attachment or adhesion to particles requires charge reduction and also the creation of hydrophobic spots on particle surfaces (James K. Edzwald, 1995).

Rising velocities of the bubbles and bubble-floc aggregates under optimal conditions are about -1 20 m h (Haarhoff & Edzwald, 2004). The separation zone hydraulic loadings (푣ℎ푙) of conventional DAF units range from 5 to 15 m h-1 which still allows the aggregates to reach the top of the DAF unit. 퐴푠푧 is the footprint area of the separation zone.

푄 + 푄푅 푣ℎ푙 = 퐴푠푧 High-rate systems also exist with loadings from 15 to 30 m h-1. The hydraulic loading exceeds the rising velocities of the bubbles. These systems obtain a higher nominal surface area using lamella’s (James K Edzwald et al., 1999). In contrast, the surface overflow rate of a secondary clarifier is limited by the settling velocity and is in the order of 1.2 m h-1 . Compared to DAF with a an average surface loading of 12 m h-1 the footprint of a settler is a factor 10 higher and the retention time is brought down from several hours to about 30 minutes (Ding et al., 2015).

Applications of DAF at MWTP’s

A DAF was investigated at pilot scale in Finland as tertiary treatment of municipal wastewater (Koivunen & Heinonen‐Tanski, 2008). High removal efficiencies of enteric microbes (90 – 99 %), COD (55 – 81 %) and TP (29 – 39 %) were obtained at a PAC coagulant dose of 10 mg Al3+ L-1. The treatment of primary wastewater effluents was investigated to assess the performance of DAF in the treatment of by-pass wastewaters during WWTP overloading. The removal efficiency of COD and TSS was 47 % and 77 % respectively (Koivunen & Heinonen‐ Tanski, 2008).

In the captivator system® the DAF is used to provide solid/liquid separation after a biosorption process. The DAF functions as a combined liquid separator and sludge thickener (Ding et al., 2015).

5 Anaerobic digestion of high rate sludge

The role of anaerobic digestion at the modern WWTP

Sludge treatment is a major issue at municipal wastewater treatment plants and its disposal represents up to 50 % of the operating costs (Davis & Hall, 1997). Anaerobic digestion (AD) is a microbial technology that is often applied at WWTPs to simultaneously stabilize and minimize the amount of sludge that has to be disposed of (Appels et al., 2008). The AD process converts the organics in sludge into renewable energy in the form of biogas which is a mixture of 60 – 70 % CH4, 30 – 40 % CO2 and can contain traces of H2O, N2, H2S, NH3 and H2 (Shen et al., 2015). At WWTP level, the primary and secondary sludge are combined and thickened before entering the digesters (Figure 5). The supernatant from the digester is denitrified or sent back to the influent and the residual sludge is dewatered.

Figure 5: Flow chart of the anaerobic digestion process in municipal wastewater treatment systems (Appels et al., 2008)

The AD process is considered as an essential part of the modern WWTP as it removes most of the pathogens present in the sludge and reduces the sludge volume and odour problems, (Appels et al., 2008). There are some limitations as the organic fraction is only partially decomposed and reaction rates are quite slow. The process is complex and vulnerable to inhibition. Next, the biogas contains not only CH4 but also other constituents that need to be removed for applications like grit injection or car fuel (Persson et al., 2006).

Biogas can be upgraded and injected into the gas grit or used as vehicle fuel, but the main application of biogas worldwide is direct use through combined heat and power production via a combined heat and power (CHP) unit at the WWTP itself (Adachi, 1995). Upgrading the biogas implies removing CO2, H2S and H2O. Removal is mostly done via ad- or absorption processes. Also cryogenic condensation and membrane separation are possible techniques (Deublein & Steinhauser, 2011).

General process principles

During AD, which takes place in absence of oxygen, organics are transformed into digestate and biogas. This complex process can be divided in four phases (Batstone & Virdis, 2014) that are shown in Figure 6. These phases are (1) hydrolysis during which polymeric organic compounds (polysaccharides, proteins and fat) are hydrolyzed by extracellular enzymes into soluble organic substances; (2) acidogenesis, during which the smaller hydrolysis products are converted into hydrogen gas, acetate and other volatile fatty acids (VFA); (3) acetogenesis, where short-chain organic acids and alcohols produced by acidogenesis are metabolized into acetic acid and hydrogen by acetate forming bacteria; and (4) methanogenesis, during which biogas is produced from hydrogen, formate, and acetate by two groups of methanogenic archaea. Acetoclastic methanogens produce methane in syntrophic relation with acetate producing bacteria while hydrogenotrophic methanogens use hydrogen gas as electron donor and CO2 as electron acceptor.

Of these four steps, hydrolysis is considered as the rate limiting step due to the slow breakdown of polymeric substances (Tiehm et al., 2001).

Figure 6: The different microbial steps in the AD process (Appels et al., 2008)

Methanogens are sensitive to pH changes and thrive best between pH 6.5 and 7.2 (Boe & Angelidaki, 2006). Anaerobic digestion is carried out at a pH in this range. Typically AD of sludge is carried out at mesophilic temperatures (35 °C) so that more organic compounds are solubilized and the metabolic rates are increased compared to standard temperatures. Also pathogens are destroyed at a higher pace (Rehm & Winter, 1999). On the other hand ammonia and VFA toxicity increases with temperature and high fluctuations (> 1 °C day-1) should be avoided (Boe & Angelidaki, 2006).

Several compounds are known to inhibit the AD process and they can either be present in the substrate or be generated during the digestion process (Kroeker et al., 1979). Free ammonia

(NH3) damages the MO’s by passing through the cell membrane and causing proton imbalances (Chen et al., 2008). Higher temperatures and increasing pH values result in a shift from ionised + (NH4 ) to free (NH3) ammonia species, and this increases the toxicity.

Sulphide inhibition can have two reasons. First, there is competition for substrate between sulphate reducing bacteria (SRB) and the fermentative bacteria and methanogens. Second, there is sulphide toxicity which affects acetogens and methanogens the most (Boe & Angelidaki, 2006). Third, the formation of metal sulphides may lead to trace element deficiencies. Due to process imbalances, situations can occur wherein the methanogens are not able to convert the

VFA fast enough to CH4 and CO2. This results in an accumulation of VFA, and the pH decreases, so there is a shift to the non-dissociated form of the VFA. These can pass freely through the cell membrane and cause pH reduction in the cell.

The cell growth must compensate for the biomass loss when digestate is removed. Stable digestion can only be achieved at SRTs > 10 days (Miron et al., 2000). Typical loading rates are in the order of 1 – 2 kg VSS m-3 day-1 for WAS digestion and specific gas production of WAS decreases as the SRT in the CAS process is raised. Bolzonella et al. (2005) observed a -1 decreasing specific gas production from 0.18 to 0.07 m³ kg VSfed when the SRT in the CAS process was raised from 8 to 35 days. This stresses the potential of high rate systems like the A-stage or HiCS process to produce better digestible sludge.

Iron rich A-stage sludge has been used in co-digestion with kitchen waste and it improved the stability of the methane production (De Vrieze et al., 2013). The iron content could be correlated linearly with methanogenesis during single digestion of A-stage sludge (Cagnetta et al., 2016).

6 Research questions

In short the research questions are as follows: what are the maximum achievable removal efficiencies of a HRAS-DAF system for TSS, VSS and tCOD? How much of the influent organics can be captured in the sludge? How well concentrated is HRAS-DAF sludge? Is there an influence of flotation on the anaerobic digestion of HRAS-DAF sludge?

PART 2: MATERIALS & METHODS

1 Pilot scale HRAS-DAF set-up

A pilot-scale HRAS system, coupled with DAF separation in place of a settler, was built in collaboration with the companies Aquafin and Nijhuis Water Technologies. The pilot HRAS- DAF consisted of a contact tank with a working volume of 2 m3 (diameter 1.6 m) and a 0.66 m3 DAF unit for solids/liquid separation. Municipal wastewater (characteristics are given in Table 1) was collected from the WWTP of Aartselaar (Belgium) after grit removal and sand trap, filtered in a drum filter (1 mm pore size) to remove coarse particles (Figure 8, nr.1), and fed to the contact tank at a flowrate of 2 m³ h-1.

Sludge recirculation was set to keep the MLSS concentration in the A-stage contact tank at 1 g TSS L-1. This was done by manually changing the on/off time of the sludge return pump. The A-stage contact tank was equipped with a pH-probe (Hach, Germany), a DO-sensor (Optical Dissolved Oxygen Probe LDO® Model 2, Hach) and a TSS-sensor (Suspended Solids InSitu Sensor, Hach). The feeding of the wastewater was based on level control in the A-stage tank.

The MLSS were transferred from the A-stage tank to the DAF by means of a cavity pump. For optimal mixing, the coagulant was added via a peristaltic membrane pump directly before the cavity pump. The polymer(s) were dosed directly in the pipes used to transfer the MLSS from the contact tank to the DAF unit (further details in the next sections). The polymers were prepared as a 0.1 % solution and stored for 1 – 3 days.

The DAF unit (Figure 8, nr.4) had a volume of 0.66 m³ and an HRT of 20 – 25 min. The saturator pressure was 0.7 MPa and an air flow between 0.5 – 0.7 L min-1 was used. The water level in the DAF could be adjusted manually. A scraper removed the floating sludge, which was then pumped to a stirred collection tank. The scraper speed could be altered between 10 and 70 Hz and also the on/off time could be adjusted. From the collection tank (shown schematically on Figure 7), which was equipped with a DO, TSS and level sensor (Hach, Germany), the sludge was recirculated to the A-stage contact tank to maintain a solids concentration of 1 g L-1 or wasted. The DAF unit was equipped with an online turbidity sensor (SOLITAX sc200 Turbidity Analyzer, Hach, Germany) that was used to measure the turbidity

21 of the effluent. At the start of each treatment and also on day 64 of treatment 1, the collection tank was filled with fresh CAS sludge (8.5 ± 1.5 g TSS L-1) from the WWTP at Aartselaar.

Figure 7: Schematic representation of the HRAS-DAF pilot plant during A-stage-DAF operation

Figure 8: The components of the HRAS-DAF pilot: 1. Drum filter, 2. A-stage tank, 3. Plug flow flocculator, 4. DAF unit 22

An automated sand drain at the bottom of the DAF unit periodically removed settled particulates. During the operation of the HRAS-DAF pilot, there were two different system configurations, resulting in three different treatments (these are described further in Table 2). Treatment 1, 2 and 3 will be referred to as T1, T2 and T3, respectively.

Table 1: Overview of the influent characteristics during T1, T2 and T3 A-stage - DAF HiCS - DAF Treatment 1 Treatment 2 Treatment 3 Total chemical oxygen demand, tCOD (g L-1) 0.36 ± 0.12 0.16 ± 0.04 0.18 ± 0.05 Soluble chemical oxygen demand, sCOD (%) 28 31 33 Total suspended solids, TSS (g L-1) 0.17 ± 0.09 0.08 ± 0.02 0.08 ± 0.02 Volatile suspended solids, VSS (%) 81 85 87 Total Kjeldahl nitrogen, TKN (mgN L-1) 52 ± 11 40 ± 10 28 ± 10 Total ammonia nitrogen, TAN (mgN L-1) 32 ± 10 16 ± 4 22 ± 6 Phosphate, P (mg L-1) 1.9 ± 0.5 1.3 ± 0.7 1.3 ± 0.5

Some days of operation were confronted with foam formation in the A-stage tank (Figure 9) and in the DAF unit.

Figure 9: Foam formation in the A-stage tank

The overall A-stage-DAF removal efficiency 휂푝𝑖푙표푡 was calculated for TSS, VSS, tCOD, using the influent (summarized in Table 1) and the effluent concentrations.

퐶푒푓푓 휂푝𝑖푙표푡 = 1 − 퐶𝑖푛푓

The removal efficiency of the DAF itself 휂퐷퐴퐹 was calculated accordingly based on the effluent concentrations and the incoming MLSS concentrations.

퐶푒푓푓 휂퐷퐴퐹 = 1 − 퐶푀퐿푆푆

Configuration 1: A-stage

The HRAS-DAF pilot was operated in an A-stage configuration for 106 days during T1 and for 67 days during T2. In this configuration, the A-stage contact tank was mixed and aerated (DO set point of 3.5 mg L-1), while the collection tank was mixed but not aerated.

1.1.1 Treatment 1: dosing of FeCl3 and A130hp

-1 -1 Treatment 1 lasted 106 days. 50 mg L of FeCl3 (Kemira) and 3 mg L anionic polymer (A130hp, Kemira Superfloc) were dosed to assist coagulation and flocculation, respectively. The coagulant was prepared as a 10 % solution in a 0.2 m³ tank. During the first 64 days of operation the FeCl3 was dosed to the MLSS and was then pumped through 4.5 m pipes. Thereafter the polymer was added and the MLSS was transported through another 1.5 m of pipes before entering the DAF unit. Also, the pressurized water was added at the entrance of the DAF unit.

After 64 days of operation, a 17.5 m plug flow flocculator (PFF) was taken in operation to increase the flocculation time and enhance flocculation. The FeCl3 was added to the MLSS and pumped through pipes of a 4.5 m length, after which roughly half of the pressurized water was added to the pipes to increase flocs stability. Subsequently the MLSS was pumped through the 17.5 m PFF before entering the DAF unit. A130hp polymer was dosed in the PFF after 0.5 m.

1.1.2 Treatment 2: dosing of FeCl3, C492 and A130hp

-1 -1 Treatment 2 was operated for 67 days and 50 mg L of FeCl3 and a combination of 2 mg L cationic polymer (C492, Kemira) and 0.5 mg L-1 A130hp were added to assist coagulation and flocculation. The addition of FeCl3 and white water was performed as from day 65 to day 106

24 of T1. The C492 polymer was dosed in the PFF after 0.5 m. The A130hp polymer was dosed 0.5 m further.

Configuration 2: HiCS operation (treatment 3)

Before starting T3, the configuration of the pilot was changed from A-stage to high rate contact stabilisation (HiCS). The tank which was previously used as a collection tank for the sludge -1 was now aerated (DO set point of 3.5 mg O2 L ) and served as stabilization tank. The aerated A-stage contact tank was changed into an unaerated contact tank. The adaptations are schematically visualized in Figure 10.

Treatment 3 lasted 46 days. The use of the PFF and the dosing points of coagulant, white water -1 - and flocculant were the same as during T2. 30 mg L of FeCl3 and a combination of 1.5 mg L 1 cationic polymer (C492) and 0.5 mg L-1 anionic polymer (A130hp) polymer were added during T3.

Figure 10: Schematic representation of the pilot during HiCS operation

MLSS characteristics

The chemical dosing as well as the MLSS characteristics from each treatment are given in Table 3- 2. During T1 the molar ratio Fe:PO4 was 5.03. The T2 was characterized by the highest Fe:P ratio (7.52) since the same amount of FeCl3 was dosed as during T1 but the influent contained less phosphate

The F/M ratio of T2 and T3 were similar since the influent was similar. Although less iron was dosed during T3, the ash content was higher (lower VSS:TSS ratio). In any treatment, the molar ration of Fe to P was higher than 4.5. This is high compared to other A-stage systems where the Fe:P ratios are between 0.24 – 0.79 (De Graaff & Roest, 2012).

Table 2: Overview of the chemical dosing and MLSS characteristics during T1, T2 and T3

A-stage - DAF HiCS - DAF Treatment 1 Treatment 2 Treatment 3 Chemical dosing -1 -1 -1 FeCl3 50 mg L 50 mg L 30 mg L (10 %; 1 L h-1) (5 %; 2 L h-1) (3 %; 2 L h-1) Cationic polymer (C492, Kemira) - 2 mg L-1 1.5 mg L-1 (0.1 %; 4 L h-1) (0.1 %; 3 L h-1) Anionic polymer (A130hp, Kemira) 3 mg L-1 0.5 mg L-1 0.5 mg L-1 (0.1 %; 6 L h-1) (0.1 %; 1 L h-1) (0.1 %; 1 L h-1) MLSS characteristics Total suspended solids, TSS (g L-1) 1.08 ± 0.42 0.75 ± 0.23 1.17 ± 0.45 Volatile suspended solids, VSS (%) 56 61 45 Food to Microorganisms ratio, F/M 13.9 8.6 7.9 (kg COD kg-1 MLVSS d-1) mol Fe/ mol P 5.03 7.52 4.51

Digesters

Two 60 L pilot-scale digesters were operated at a HRT of 18 days. These were inoculated with digestate from a full-scale anaerobic digester from Aquafin, Leuven. The digesters were mixed mechanically, and operated at mesophilic conditions (35 °C). The sludge feeding was done continuously by a peristaltic pump with a flow rate of 170 mL h-1. The gas flow was measured by a MGC-10 Ritter® flow meter. Each of the two digesters had its own sludge buffer tank from which the sludge was fed. The buffer tanks were filled manually with fresh sludge on Monday, Wednesday and Friday. Each buffer tank was equipped with a mechanical mixer and a level sensor. For the A-stage-DAF configuration, A-sludge generated from the pilot was used solely to feed the digesters, and these were run as duplicates. For the HiCS-DAF configuration,

26 one of the digesters was fed with HiCS-sludge, while the second digester was fed with a mixture of WAS sludge (Aartselaar, Belgium) and HiCS-sludge at a TSS ratio of 3:7. Digestion of HRAS-sludge (A-stage or HiCS) generated during different treatments was carried out for 54 days (3 x 18 days) for each treatment.

Figure 11: The digesters at Aquafin, Aartselaar: 1. Buffer, 2. Digester, 3. Peristaltic pump, 4. Gas flow meter

Sampling

Samples from the influent wastewater (before and after the drum filter), MLSS, sludge and effluent taken collected three times per week, and analysed for TSS, VSS, tCOD and sCOD. Total Kjeldahl nitrogen (TKN) and Total ammonia nitrogen (TAN) were analysed every two weeks. Phosphate content was analysed for influent and effluent samples.

Gas samples from the digesters’ head spaces were collected twice a week using 10 mL vacutainers. The biogas composition was determined using a Compact GC. The digestate from both digesters was sampled on the same days as the gas samples. The pH, conductivity, TSS, VSS, tCOD and VFA content were determined for each sample.

2 Jar testing

Experiment 1: Optimal polymer selection for flotation

At the start of T1, jar tests were performed to select which polymer to use to assist flocculation. -1 1 L flasks were filled with 500 mL MLSS at the pilot installation and 50 mg L FeCl3 was added in combination with a cationic polyacrylamide (PAM) polymer (C492 or C494 or C496, Kemira) or an anionic PAM polymer (A130hp, Kemira). Polymer concentrations ranging from 0.5 to 4 mg L-1 were tested. The flasks were stirred for 10 minutes at 270 rpm using a magnetic stirrer. Then 200 mL white water was added to each flask and the flotation performance was evaluated visually.

The cationic polymers are characterized by a medium (C492, 105–106 g mol-1) to high (C494 and C496, >106 g mol-1) molecular weight. The charge density (CD) as well as the viscosity increases from low (C492, CD 20 %, 80 mPa.s for 0.1 % solutions) to medium (C494) and high (C496, 130 mPa.s for 0.1 % solutions) (Kemira, 2010). A130hp has a higher molecular weight (> 107 g mol-1) than the previous three cationic polymers, and has a CD of 30 %.

Experiment 2: Optimal flocculant dosage (T1)

The 1 L flasks were filled with 500 mL MLSS sampled at the HRAS-DAF pilot plant. FeCl3 (50 mg L-1) was added in combination with different concentrations of anionic polymer (A130hp) (0.75 – 6.5 mg L-1). The mixtures were stirred subsequently for a few seconds at 500 rpm. Then 200 mL of white water was added. After 20 minutes of flotation, a subnatans sample was taken from each beaker, and was analysed for TSS and for optical density (OD) at 610 nm.

Experiment 3: Optimal flocculant dosage (T2)

The 1 L flasks were filled with 500 mL MLSS sampled at the HRAS-DAF pilot plant. The -1 FeCl3 (50 mg L ) was added in combination with different concentrations of cationic (C492) (0.8 – 3.8 mg L-1) and anionic (A130hp) (0.1 – 0.9 mg L-1) polymers. The same procedure as in experiment 2 was followed.

Experiment 4: Optimal coagulant & flocculant dosage (T3)

1 L flasks were filled with 500 mL MLSS sampled at the HRAS-DAF pilot plant. A130hp -1 -1 polymer (1 mg L ) and different combinations of cationic C492 (2 – 3 mg L ) and FeCl3 (5 – 45 mg L-1) were added. The same procedure as in experiment 2 was followed.

3 BMP tests

Biochemical methane potential (BMP) tests were performed in batch reactors, each with a working volume of 80 mL and a headspace of 40 mL. The experiments were operated for 18 days at mesophilic conditions (35 °C). The inoculum was collected from a full-scale anaerobic digester in Leuven WWTP (Belgium) or from the pilot-scale digesters in Aartselaar. To each flask, first, a specific amount of inoculum was added to obtain a final VS concentration of 10 g VSS L-1. Second, A-sludge, A-sludge with addition of M9 minimal medium (modified with no glucose addition, Table 3) or A-sludge with addition of vitamins and trace elements (Table 4) was added (except for the negative controls) to obtain a substrate to inoculum ratio of 0.5 g COD g-1 VSS. Finally, tap water was added to acquire a total liquid volume of 80 mL in each bottle, irrespective of the selected inoculum or substrate. For each inoculum triplicate negative controls were run, containing the selected inoculum at a concentration of 10 g VSS L-1, to estimate endogenous methane production.

After inoculum and substrate addition, the serum flasks were sealed to avoid air intrusion, and connected to air-tight gas columns by means of an air-tight needle. These gas columns were placed in a water bath containing distilled water at pH < 4.3 to avoid CO2 in the biogas from dissolving. The serum flasks were incubated in a linear shaking water bath (Aqua 12 Plus,

Novolab, Geraardsbergen, Belgium). Volumetric biogas production was evaluated by means of water displacement in the gas columns. Biogas production was measured on daily basis for 7 days, until the biogas production was below 1 – 3 % of total production in all treatments for 3 consecutive days. Biogas volumes were reported at standard temperature (273 K) and pressure (101325 Pa) (STP). Biogas composition was evaluated at the end of the experiment. Methane yield was expressed as COD yield as the fraction of substrate COD converted to methane.

During T1, two BMP tests were performed on A-sludge from day 14 and 76 with inoculum from full scale anaerobic digesters from Leuven. Another BMP was performed on A-sludge from day 92, this time with inoculum from the pilot scale digesters in Aartselaar.

During T2, a BMP test was performed on A-sludge from day 54 with inoculum from Aartselaar. One additional BMP was performed on the A-sludge with addition of M9 minimal feeding medium. A second BMP was done on A-sludge with addition of vitamins. Last, a BMP in which a combination of M9 and vitamins was added, was performed.

During T3, BMP tests were performed on HiCS-sludge from day 21 and 38 with inoculum from the digesters in Aartselaar.

Table 3: Composition of M9 minimal medium

Component Volume %

M9 salts solution 20 Glucose (20%; Sigma-Aldrich)a 2

b MgSO4 (1 M; Fisher Scientific) 0.2

b CaCl2 (1 M; Fisher Scientific) 0.01

H2O 78 aFilter-sterilize and store at 4°C. bAutoclave and store at room temperature. cM9 salts solution composition Concentration (g L-1)

Na2HPO4.7H2O 64 KH2PO4 15 NaCl 2.5 NH4Cl, 5

Table 4: Composition of trace element and vitamin solution

Trace element solution Vitamin solution Nitrilotriacetic acid 1.50 g Biotin 2.00 mg

MgSO4.7H2O 3.00 g Folic acid 2.00 mg

MnSO4.H2O 0.50 g Pyridoxine-HCl 10.00 mg

NaCl 1.00 g Thiamine-HCl.2H2O 5.00 mg

FeSO4.7H2O 0.10 g Riboflavin 5.00 mg

CoSO4.7H2O 0.18 g Nicotinic acid 5.00 mg

CaCl2.2H2O 0.10 g D-Ca-pantothenate 5.00 mg

ZnSO4.7H2O 0.18 g Vitamin B12 0.10 mg

CuSO4.H2O 0.01 g p-Aminobenzoic acid 5.00 mg

KAl(SO4)2.12H2O 0.02 g Lipoic acid 5.00 mg

H3BO3 0.01 g Distilled water 1000 ml

Na2MoO4.2H2O 0.01 g

NiCl2.6H2O 0.03 g

Na2SeO3.5H2O 0.30 mg

Na2WO4.2H2O 0.40 mg Distilled water 1000 ml

4 Fed batch digesters

Two 2 L scotch bottles were filled both with 890 mL inoculum from a full-scale anaerobic digester in Leuven WWTP (Belgium) and 110 mL HRAS-sludge from the HRAS-DAF pilot in Aartselaar summing up to an operational volume of 1 L. They were fed in batch mode on Monday, Wednesday and Friday and operated at mesophilic conditions (35 °C). The inoculum had a VS content of 25 g L-1. To impose an SRT of 18 days, each Monday and Wednesday 110 mL of digestate was removed, and 110 mL fresh HRAS-sludge was added. On Friday 168 mL digestate was removed, and 168 mL fresh sludge was added. At day 55, 1 ml of the trace -1 elements and vitamins solutions (Table 4) was added together with 2.68 g L Na2HPO4 and -1 0.94 g L KH2PO4.. Gas production was followed by connecting each fed batch digester to a water column. Digestate was sampled and analysed for pH, conductivity and VFA content each time the digesters were fed.

5 Chemical analyses

COD analysis

The Chemical Oxygen Demand (COD) is the amount of oxygen required to oxidize organic carbon completely to CO2 by chemical means. Determination of total and/or soluble COD of aqueous (wastewater, effluent) or slurry-like (MLSS, sludge) samples quantifies the concentration of organic constituents.

The COD analysis on MLSS and sludge samples was performed according to Standard Methods for the examination of water and wastewater (5220-C, APHA, 1992). This is the classical oxidation method, using an excess of potassium dichromate (K2Cr2O7) in an acid environment and high temperature (148 °C) in the presence of silver sulphate (AgSO4) as catalyst. After oxidation, the COD is determined by a back titration in which the excess of unreduced K2Cr2O7 is titrated with an iron ammonium sulphate (FeNH3SO4) solution with ferroin as indicator.

Mercuric sulphate (Hg2SO4) is added in order to precipitate the chlorides to minimise their

32 interference. Samples with a COD content higher than 900 mg L-1 were diluted as the required concentration range of this method is between 100 and 900 mg L-1. The COD analysis of non-coloured samples (influent & effluent) was measured using Nanocolor® kits (CODE; Macherey-Nagel) of the type COD - 160 with an analysis range of 15 – 160 mg COD L-1. For determination of sCOD, the samples were centrifuged and filtered with 0.45 m pore diameter filters, and then also analysed using Nanocolor® kits. Starting from day 19 of T2, it was decided not to use the titration method anymore and use the Nanocolor® kits type COD - 1500 with an analysis range of 100 - 1500 mg COD L-1 for COD determination of the MLSS and sludge samples. In a comparison experiment, only minor differences (± 3 %) were observed between the titration and the colorimetric method.

Solids analysis

Total suspended solids (TSS) and volatile suspended solids (VSS) analysis was performed according Standard Methods 2540D and E (APHA, 1997). The TSS measurements included filtering the samples over a 0.45 µm filter, and drying them overnight at 105°C. The filters were weighed before and after. For VSS measurements, the dried filters were placed in a furnace (Nabertherm LE6/11/B150, Germany) at 550 °C for 1.5 h, and afterwards they were weighed again. The calculation of the TSS was done by distracting the clean filter weight from the dried filter weight and dividing by the applied sample volume. The VSS was calculated by distracting the muffled filter weight from the dried filter weight and dividing by the applied sample volume.

VFA analysis

The C2 – C8 fatty acids (including isoforms C4 – C6) were measured by gas chromatography (GC-2014, Shimadzu®, The Netherlands) with a DB-FFAP 123-3232 column (30m x 0.32 mm x 0.25 μm; Agilent, Belgium) and a flame ionization detector (FID).. The 2 mL liquid samples were conditioned with 0.5 mL concentrated sulphuric acid (H2SO4) and 0.4 g sodium chloride (NaCl) and 400 µL 2-methyl hexanoic acid as internal standard for quantification. Extraction of the VFA was done by adding 2 mL diethyl ether. The samples were than rotated (2 min) and centrifuged (3 min, 1000 g), and the diethyl ether phase was moved to a GC vial. The prepared sample (1 μL) was injected at 200 ºC with a split ratio of 60 and a purge flow of 3 mL min-1.

The oven temperature increased by 6 ºC min-1 from 110 ºC to 165 ºC where it was kept for 2 -1 min. The FID temperature was 220 ºC. The carrier gas was N2 at a flow rate of 2.49 mL min .

Gas analysis

The gas phase composition was analysed with a Compact GC (Global Analyser Solutions, Breda, The Netherlands), equipped with a Molsieve 5A pre-column and Porabond column

(CH4, O2, H2 and N2) and a Rt-Qbond pre-column and column (CO2, N2O and H2S). Concentrations of gases were determined by means of a thermal conductivity detector, which had a detection limit of 100 ppmv.

Anion analysis

3- Phosphate (PO4 ) in influent and effluent collected from the HRAS-DAF pilot plant was determined on a 761 Compact Ion Chromatograph (Metrohm, Switzerland), equipped with a conductivity detector.

Nitrogen analysis

Kjeldahl nitrogen (TKN) and total ammoniacal nitrogen (TAN) were analysed using steam distillation, according to standard methods (Greenberg et al., 1992). The TKN includes the organic nitrogen and TAN. The organic nitrogen can be determined by the subtraction of the TAN (total ammoniacal nitrogen) from the TKN. During Kjeldahl analysis, the distillation is preceded by a destruction phase. The organic nitrogen present in the sample is transformed into ammoniacal nitrogen (NH4)2SO4 by means of destruction at 380°C with sulphuric acid (H2SO4)

(98 %) and potassium and copper sulphate (K2SO4, CuSO4) as a catalyst. During steam distillation, 20 mL of sample (for TAN analysis) or destructed sample (for TKN analysis) is distilled, and the NH3 is captured in a boric acid indicator with an initial pH of 5.3. The NH3 that is captured in that acid solution (as (NH4)3BO3) was titrated with hydrochloric acid (HCl, 0.02N). The titration was carried out with a pH meter.

PART 3: RESULTS

1 Pilot scale HRAS-DAF

Results that were obtained during the three different HRAS-DAF pilot treatments will be presented in this chapter. On each time plot describing the HRAS-DAF pilot results, the transitions between the different treatments are indicated with black vertical dotted lines. Another grey vertical dotted line was added on day 65 to indicate the day the PFF was taken in use.

The primary goal of the HRAS-DAF pilot plant was to achieve high organics removal from the influent. Next, a good solid/liquid separation in the DAF unit was pursued as well for good effluent quality as for maximal preservation of the influent organics in the sludge. Since the DAF was desired to work as a combined settler and thickener, sludge concentration was analysed, and is also presented in this chapter. Finally, the results obtained from the pilot scale digesters are given.

Organics removal

The overall HRAS–DAF pilot removal efficiency 휂푝𝑖푙표푡 was calculated for TSS, VSS and tCOD , based on the influent and the effluent concentrations. The average removal efficiency for each treatment is given in Table 5. A timeline showing the removal during T1, T2 and T3 is shown in Figure 12 .

Table 5: Overview of the HRAS-DAF TSS, VSS and tCOD removal during T1, T2 and T3

Period TSS VSS tCOD

T1 (day 1 - day 64) 51 ± 17 % 53 ± 19 % 50 ± 13 %

T1 (day 65 - day 106) 77 ± 14 % 81 ± 10 % 62 ± 13 %

T2 71 ± 10 % 73 ± 10 % 58 ± 10 %

T3 67 ± 10 % 68 ± 9 % 54 ± 9 %

A-stage-DAF treatment 1 A-stage-DAF treatment 2 HiCS-DAF treatment 3

100

40 TSS

Removal efficiency (%) efficiency Removal VSS COD 20 TSS: 51 ± 17 % TSS: 77 ± 14 % TSS: 71 ± 10 % TSS: 67 ± 10 % VSS: 53 ± 20 % VSS: 81 ± 10 % VSS: 73 ± 10 % VSS: 68 ± 9 % tCOD: 50 ± 13 % tCOD: 62 ± 13 % tCOD: 58 ± 10 % tCOD: 54 ± 9 % 0 0 20 40 60 80 100 120 140 160 180 200 220 Time (days)

Figure 12: Timeline of the HRAS-DAF pilot removal efficiencies for TSS, VSS and tCOD during T1, T2 and T3. The grey dotted line at day 65 indicates the implementation of the plug flow flocculator.

1.1.1 A-stage-DAF T1

During T1 the influent wastewater had a COD concentration of 0.38 ± 0.12 g L-1 and could be classified as a medium-strength wastewater. The first 64 days of operation featured a short distance between polymer addition and the entrance of the DAF and the inherent short flocculation time could only support the growth of small flocs. The A-stage-DAF pilot removed 51 ± 17 % of the influent TSS, 53 ± 19 % of the VSS and 50 ± 13 % of the tCOD, and no standalone operation could be achieved for longer than 2 – 3 days. After day 64 and the implementation of the PFF, the removal was increased to 77 ± 14 % of the TSS, 81 ± 10 % of the VSS and 62 ± 13 % of the tCOD.

1.1.2 A-stage-DAF T2

During T2, the wastewater had a COD concentration of 0.18 ± 0.04 g L-1, and could be classified as a low-strength wastewater. The A-stage-DAF operation was stable, and the overall removal was quite constant. An average removal efficiency of 71 ± 10 % was obtained for the influent TSS, 73 ± 10 % of the VSS and 58 ± 10 % of the tCOD.

1.1.3 HiCS-DAF T3

During T3, the wastewater had a concentration of 0.16 ± 0.03 g COD L-1, and could be classified as a low-strength wastewater. The DAF operation was stable, and no process failures occurred during this treatment. Average removal was 67 ± 10 % of the influent TSS, 68 ± 9 % of the VSS and 54 ± 9 % of the tCOD.

DAF separation efficiency

To evaluate the performance of the DAF unit, the TSS, VSS and tCOD separation efficiency

휂퐷퐴퐹 was calculated based on the effluent concentrations and the incoming MLSS concentrations. The separation efficiencies of the DAF unit during each treatment are given in Table 6. A timeline showing the DAF unit separation efficiency during T1, T2 and T3 is shown in Figure 13 .

Table 6: Overview of the DAF unit TSS, VSS and tCOD separation during T1, T2 and T3

Period TSS VSS tCOD

T1 (day 1 - day 64) 90 ± 7 % 88 ± 7 % 83 ± 8 %

T1 (day 65 - day 106) 97 ± 2 % 96 ± 2 % 88 ± 7 %

T2 96 ± 3 % 95 ± 3 % 90 ± 3 %

T3 95 ± 4 % 93 ± 4 % 86 ± 8 %

The microbubble quality was evaluated visually each day by filling a bottle with pressurized water from the DAF. Good microbubble quality was assumed when the water had a milky appearance and when the bubbles stayed in suspension at least 2 minutes. Based on this determination it was decided whether or not to change the air flow (always between 0.5 – 1 L min-1) or to flush the pressurized water tubing to remove impurities from the recycle stream.

A-stage-DAF treatment 1 A-stage-DAF treatment 2 HiCS-DAF treatment 3

100

70 Removal efficiency (%) efficiency Removal TSS 60 VSS tCOD

50 0 20 40 60 80 100 120 140 160 180 200 220 Time (days)

Figure 13: Timeline of the DAF unit separation efficiencies for TSS, VSS and tCOD during T1, T2 and T3. The grey dotted line at day 65 indicates the implementation of the plug flow flocculator 40

1.2.1 A-stage-DAF T1

It can be observed from Figure 13 that the operation of the DAF unit was not stable before day 64. For several days, only 60 – 70 % tCOD removal was achieved. Average removal efficiencies during the first 64 days were 90 ± 7 % of the TSS, 88 ± 7 % of the VSS and 83 ± 8 % of the tCOD respectively. After day 64, the DAF performance became more stable, and the removal efficiency of TSS, VSS and tCOD was increased to 97 ± 2 %, 96 ± 2 % and 88 ± 7 %, respectively.

1.2.2 A-stage-DAF T2

The DAF unit operation was stable during the whole period of T2. Average separation efficiencies were 96 ± 3 % of the TSS, 95 ± 3 % of the VSS and 90 ± 3 % of the tCOD.

1.2.3 HiCS-DAF T3

The tCOD removal was low (± 70 %) on day 14, 15, 31 and 41. The main reason for the low performance of the DAF was the inferior microbubble quality at those days. The white water had no milky appearance and the bubbles were relatively large and didn’t stay in suspension for longer than a minute. Impurities in the recycle stream were presumably causing the inferior microbubble quality, and effluent results for these days were not considered. Average removal efficiencies were 95 ± 4 % of the TSS, 93 ± 4 % of the VSS and 86 ± 8 % of the tCOD.

1.2.4 Effluent characteristics

The average HRAS-DAF effluent composition for each treatment is included in

Table 7. During T1, the influent had a higher strength than during T2 and T3. The effluent produced during T1 (0.17 ± 0.06 g COD L-1) also had a higher strength compared to T2 (0.07 ± 0.03 g COD L-1) and T3 (0.09 ± 0.03 g COD L-1). This explains why the overall HRAS-DAF pilot organics removal remained roughly the same over the three treatments.

The VSS:TSS ratio was 75 % for T1, 82 % for T2 and 79 % for T3. This is slightly lower than for the influent. The sCOD:tCOD ratio was low during T1 (36%) but increased for T2 (56 %) and T3 (53 %).

Table 7: Overview of the effluent characteristics during T1, T2 and T3 A-stage - DAF HiCS - DAF

Treatment 1 Treatment 2 Treatment 3

Total suspended solids, TSS (g L-1) 0.07 ± 0.04 0.03 ± 0.01 0.04 ± 0.03

Volatile suspended solids, VSS (%) 75 82 79

Total chemical oxygen demand, tCOD (g L-1) 0.17 ± 0.06 0.07 ± 0.03 0.09 ± 0.03

Soluble chemical oxygen demand, sCOD (%) 36 56 53

Total Kjeldahl nitrogen, TKN (mgN L-1) 44 ± 9 42 ± 10 28 ± 12

Total ammonia nitrogen, TAN (mgN L-1) 27 ± 8 17 ± 1 22 ± 1

Phosphate, P (mg L-1) 1.2 ± 0.4 0.01 ± 0.03 0.2 ± 0.3

Sludge characteristics

The HRAS-DAF was expected to produce good digestible sludge with concentrations up to 60 g COD L-1. Table 8 contains the average sludge composition during the different treatments. Figure 14 displays how the sludge composition varied over the different treatments.

Table 8: Overview of the HRAS-DAF sludge composition (TSS, VSS, tCOD and sCOD) during T1, T2 and T3

Period TSS (g L-1) VSS (g L-1) tCOD (g L-1) sCOD (%)

T1 37 ± 8 25 ± 6 47 ± 9 0.8

T2 27 ± 5 16 ± 3 21 ± 6 1.1

T3 28 ± 5 27 ± 3 25 ± 6 1.2

The high concentration potential of the DAF unit was shown during the first 64 days of T1 as the average sludge concentration was 50 ± 8 g COD L-1. On two days (day 24 and 50) a concentration even higher than 60 g L-1 was observed. After day 64 until the end of T1, the concentration decreased to 39 ± 8 g COD L-1. In general the sludge from T1 had the following characteristics: 37 ± 8 g TSS L-1, 25 ± 6 g VSS L-1 and 47 ± 9 g COD L-1. Further, the T1 sludge was characterised by a VSS:TSS ratio of 69 % and a tCOD:TSS ratio of 1.28. The average SRT (1.11 d) and the aerobe SRT (0.40 d) were both lower than two days.

Throughout the whole operation of T2, the sludge concentration remained rather constant. The average composition was 27 ± 5 g TSS L-1, 16 ± 3 g VSS L-1and 21 ± 6 g COD L-1. Maximum sludge concentration measured during T2 was 32.2 gCOD L-1 (day 51). The sludge had a VSS:TSS ratio of 57 % and a tCOD:TSS ratio of 0.77. The SRT (1.39 d) and the aerobe SRT (0.39 d) were again lower than the maximum of two days.

The contact stabilisation sludge that was produced during T3 was similar to the sludge from T2. The highest sludge concentration during T3 was 33.6 g COD L-1 (day 36). Average composition was 28 ± 5 g TSS L-1, 17 ± 3 g VSS L-1 and 25 ± 6 g COD L-1.

A-stage-DAF treatment 1 A-stage-DAF treatment 2 HiCS-DAF treatment 3 70

TSS 50 VSS COD 40

Removal efficiency (%) efficiency Removal 20

0 0 20 40 60 80 100 120 140 160 180 200 220 Time (days)

Figure 14: Timeline of the sludge TSS, VSS and tCOD composition for T1, T2 and T3. The grey dotted line at day 65 indicates the implementation of the plug flow flocculator. 45

TSS VSS tCOD

60 )

1 50 - 40 30 20 10

TSS, VSS, tCOD (g VSS, L TSS, (g tCOD 0 Treatment 1 Treatment 2 Treatment 3 A-stage-DAF HiCS-DAF

Figure 15: Comparison of sludge TSS, VSS and tCOD for T1, T2 and T3

Sludge concentration was negatively affected by the amount of water that was scraped off together with the sludge. Minimising this amount of water could be done by lowering the scraper speed and lowering the level of the water in the DAF unit. Meanwhile, over accumulation of sludge at the DAF surface had to be avoided.

COD balance

In the search for optimal energy preservation, it was important that respiration of COD to CO2 was minimalised and that influent organics were preserved in the sludge. A COD balance (Figure 16) was made with data of the dates on which the pilot was running in a steady state and not recovering from a recent failure.

The influent COD has three possible routes to follow. It can end up in the effluent, in the waste sludge or be mineralised to CO2. For every treatment, the amount of COD that leaves with the effluent was around 40 %. There are large differences however in the proportion of removed COD that is mineralised. During the first 64 days of T1, 61 % of the COD that didn’t leave with the effluent was left in the sludge. The highest proportion of COD (84%) was captured in the sludge from day 64 until the end of T1. Sludge capture was lower for T2 (65 %) and T3 (72 %). Considering the quasi identical influent during T2 and T3, the HiCS process obtained a higher entrapment of organics.

Figure 16: COD balance for the three treatments. The proportion between the COD captured in the sludge and the mineralised COD is given in a pie chart next to each bar plot

Digester performance

The normalised methane content of the biogas for the different treatment is given in Figure 17.

Similar biogas (66 ± 1 % CH4) was produced in both digesters during T1. Higher percentages but also higher fluctuations of methane content were determined for the biogas during T2 (74

± 3 % CH4 for digester 1 and 75 ± 3 % CH4 for digester 2). Under the HiCS operation (T3) the digesters were fed differently. Digester 1 was fed HiCS sludge (72 ± 2 % CH4), while digester

2 (73 ± 4 % CH4) was fed a combination of CAS sludge and HiCS sludge in a 3:7 ratio on TSS basis. CAS sludge was co-digested with HiCS sludge as it was believed that the CAS sludge could provide the nutrients and trace elements that could be deficient in the HiCS sludge. The pH of both digesters (7.4 ± 0.1) remained stable throughout the entire experimental period and for both digesters, the conductivity varied between 5 – 10 mS cm-1.

The COD conversion efficiency during T1 was 30 – 40 %. The gas productions of T2 and T3 could not be obtained from Aquafin before the end of this thesis.

(%)

N 4 75

Normalised CH content Normalised methane Digester 1 Digester 2 Digester 1 Digester 2 Digester 1 Digester 2 A-stage sludge T1 A-stage sludge T2 HiCS sludge 30 % CAS + T3 70 % HiCS sludge

Figure 17: normalised methane content of the biogas during T1, T2 and T3

2 Jar tests

Experiment 1: optimal polymer selection for flotation

Before T1, an optimal polymer had to be selected for flotation. The C492 polymer was tested during the start-up, but no flotation occurred. Other polymers were tested via jar tests. These polymers were cationic C494 and C496 polymer and anionic A130hp polymer. Visually, A130hp polymer showed the best and fastest flotation, and, thus, it was selected over the C492, C494 and C496 polymers.

Experiment 2: optimal flocculant dosing (T1)

At the beginning of T1 (day 7 and 14), the optical density (OD) at 610 nm and TSS were determined on the subnatans from jar tests performed on MLSS. The FeCl3 concentration was kept constant at 50 mg L-1. A130hp polymer concentrations ranging from 0.75 to 6.5 mg L-1 were applied. An A130hp concentration of 3.5 mg L-1 resulted in the clearest effluent (OD of

0.3 and TSS 0.08 g L-1) and a TSS removal of 95 %. As can be seen in Figure 18 the TSS concentrations increases again for polymer dosages higher than 3.5 mg L-1.

Based on the advice from the experts from Nijhuis and via trial and error, a dosing of 3 mg L-1 A130hp was applied during T1. Under these conditions, the sludge floated and large flocs were formed after the addition of polymer. This jar test justified the dosing of A130hp

OD TSS 0.6 0.2

0.5 )

- 0.15 )

0.4 1 - 0.3 0.1

0.2 (g TSSL 0.05 OD at ( at OD nm 610 0.1 0 0 0.75 1.5 2.5 3.5 4.5 5.5 6.5 A130hp (mg L-1)

Figure 18: OD and TSS measurements on subnatans for constant FeCL3 (50 mg L-1) and different A130hp concentrations (n=2)

Experiment 3: optimal flocculant dosing (T2)

-1 -1 At a constant concentration of FeCl3 (50 mg L ) and C492 (2.3 mg L ), the A130hp polymer concentration was varied between 0.1 and 0.9 mg L-1 (Figure 19). The OD kept decreasing for higher polymer concentrations, while a minimum TSS concentration (0.045 g L-1) was observed for 0.5 mg L-1 A130hp. This concentration was selected to operate the pilot during T2.

OD TSS 0.25 0.1

) 0.2 0.08

) 1 0.15 0.06 -

0.1 0.04 TSS (g (g TSSL

OD at ( at OD nm 610 0.05 0.02

0 0 0.1 0.3 0.5 0.7 0.9 A130hp (mg L-1)

-1 Figure 19: OD and TSS measurements on subnatans for constant FeCl3 (50 mg L ) and C492 (2.3 mg L-1) and different A130hp (0.1 – 0.9 mg L-1) concentrations during T2

-1 -1 At a constant FeCl3 (50 mg L ) and an optimal A130hp (0.5 mg L ) concentration, the C492 polymer concentration was varied between 0.8 and 3.8 mg L-1 (Figure 20) The minimal OD was observed at 3.05 mg L-1 C492. The minimal TSS concentrations were measured at 0.8 and 3.8 mg L-1.

OD TSS 0.25 0.1

) 0.2 0.08

) 1 0.15 0.06 -

0.1 0.04 TSS (g (g TSSL

OD at ( at OD nm 610 0.05 0.02

0 0 0.8 1.55 2.3 3.05 3.8 C492 (mg L-1)

-1 Figure 20: OD and TSS measurements on subnatans for constant FeCl3 (50 mg L ) and A130hp (0.5 mg L-1) and different C492 (0.8 – 3.8 mg L-1) concentrations during T2

Experiment 4: optimal FeCl3 and C492 dosing (T3)

At an optimal A130hp (0.5 mg L-1) concentration and two different C492 polymer -1 -1 concentrations (2 – 3 mg L ) the FeCl3 concentration was varied between 5 and 45 mg L (Figure 21) The minimal TSS concentration was measured for 3 mg L-1 C492 and 35 mg L-1 -1 FeCl3. An increase in TSS concentration in the subnatans could be seen from 5 to 25 mg L -1 FeCl3 after which a minimum was found at 35 mg L .

C492 2 mg L-1 C492 3 mg L-1

0.06

0.05 1)

- 0.04 0.03

0.02 TSS (mg L (mg TSS 0.01 0 0 10 20 30 40 50 -1 FeCl3 (mg L ) Figure 21: TSS measurements on subnatans for constant A130hp (0.5 mg L-1) and different C492 (2 – 3 mg L-1) and FeCl3 (5 – 45 mg L-1) concentrations during T3

3 BMP tests

During the three different treatments, BMP tests were performed on the sludge from the pilot to determine the biodegradability and the potential methane yield. The sludge characteristics were given earlier this chapter. The BMP tests were carried out until the volume of biogas produced did not increase for 2 consecutive sampling points.

BMP T1

The results for the BMP tests on the sludge that was obtained during T1 are shown in Figure 22. The COD conversion efficiency to methane was between 58 and 68 %. Conversion efficiencies were similar to these reported for conventional A-stage systems where settling is used for solid/liquid separation (De Vrieze et al., 2013). Whether the inoculum was from the full scale digester in Leuven of form the pilot scale digesters in Aartselaar had no influence on the BMP results.

100 4 67

) 80 fed

1 68 - 58

60 COD

CH4 40

20 (% (% COD

Conversion efficiency CH to efficiency Conversion 0 Day 14 Day 76 Day 92 Inoculum Leuven Inoculum Aartselaar

Figure 22: Conversion efficiencies for the BMP tests during T1

BMP T2

A BMP test was performed on A-sludge from day 54 of T2 with inoculum from Aartselaar. A low COD conversion to methane (24 ± 2 %) was observed. 52

The BMP tests with addition of M9 and/or vitamins or trace elements were performed to investigate if the low conversion efficiencies to CH4 for sludge generated during T2 were due to a lack of phosphorous and/or nutrients. Conversion efficiencies were similar for the control and the test where M9 was added, at 24 % and 28 %, respectively. When trace elements and vitamins were added, in combination with M9 or solely, conversion efficiencies were higher, at 37 % and 36 %, respectively (Figure 23).

50 4 37 36

) 40 28

fed 1 - 24

30 COD

CH4 20

10 (% (% COD

Conversion efficiency CH to efficiency Conversion 0 BMP BMP M9 BMP M9 + BMP tap water + vitamin vitamin

Figure 23: Conversion efficiencies for the BMP tests during treatment 2

It was investigated if the polymer or the iron were inhibiting the AD by doing a BMP test on

MLSS where neither FeCl3 nor polymers were added yet (except the one coming from the recirculation). Further, to test whether the incoming wastewater contained compounds that could inhibit AD of the A-sludge generated during the second treatment of the A-stage-DAF system, a BMP test was performed on waste activity sludge (WAS) generated in a CAS system in Aartselaar treating the same wastewater. The COD to CH4 conversion efficiencies for MLSS and WAS were 26 ± 1 % and 23 ± 3 %. Respectively.

BMP T3

Poor COD conversion to methane (± 42 %) was observed for the sludge yielded during T3 (Figure 24). The COD conversion to methane was higher compared to T2 but lower compared to T1

50 42 41

)

fed 1 - 40

COD 30 CH4 20

(% (% COD

4 Conversion efficiency to efficiency Conversion

CH 0 Day 21 Day 38

Figure 24: Conversion efficiencies for the BMP tests during treatment 3

4 Fed batch digesters

During the first 18 days (= 1 SRT) a COD to methane conversion efficiency between 40 – 70 % was observed (Figure 25). Afterwards, the conversion went down to about 15 %. On day 55 there was an addition of trace elements and phosphorous salts leading the a minimal recovery of the methanogenesis (± 20 % conversion). On day 62, there was a switch from A-stage-DAF sludge to HiCS-DAF sludge.

100 4 A-stage-DAF (T2) HiCS-DAF (T3)

) 80

fed

1 -

60 COD

CH4 40

20 (% (% COD 0 Conversion efficiency CH to efficiency Conversion 0 20 40 60 80 100 120 Time (days)

Figure 25: Time plot showing the conversion efficacy of the fed batch digesters. The dotted line on day 55 indicates the addition of phosphorous salts.

PART 4: DISCUSSION

1 Pilot scale HRAS-DAF

The HRAS-DAF pilot was designed to resemble the A-stage of an A/B-process in terms of SRT, HRT and DO. A DAF was used with the objective to obtain better solid-liquid separation, higher organics removal and more compact sludge in comparison with conventional settling. The A/B-process is a dual stage wastewater treatment system in which the first stage is a high rate reactor designed for carbon and phosphorous removal carbon removal whereas the second stage serves for removal of nitrogen and residual COD. The high rate operation of the first stage, ensures that the influent sCOD, cCOD and pCOD are concentrated to the sludge with minimal energy input and at a small footprint (Jimenez et al., 2015).. .The dual stage operation of the A/B-process makes it possible to optimise the organics and nitrogen removal separately. By incorporation in the high rate sludge, part of the nutrients are already removed in the first stage (Jetten et al., 1997)

There are different scenarios for nutrient removal in the B-stage and these are determined by the effluent characteristics from the A-stage. A/B-systems were described where enough BOD (100 – 150 mg L-1) remains in the A-stage effluent to have a BOD:N ratio of 4 or higher (Boehnke et al., 1998). A BOD:N ratio of 4 is considered the minimum to support conventional nitrification/denitrification (Tchobanoglous & Burton, 1991). The HRAS stabilises shock loads for the B-stage and the HRAS effluent organics are more biodegradable compared to CAS effluent organics. It was shown that recalcitrant compounds are disproportionally more extracted in HRAS systems compared to CAS because of the adsorption processes (Sorensen et al., 1994). This effect is even increased by the high sludge production in the first stage.

For effluent from the first stage having a BOD:N ratio lower than 4, complete nitrification/denitrification in the B-stage is not possible without the addition of an external carbon source (Loosdrecht et al., 1997). A second option for nitrogen removal in the B-stage is via the partial nitrification/annamox process (Kartal et al., 2010). However this process is still under lab scale investigation for application on cold (< 20 °C) and diluted streams (< 100 mg N L-1) (Hendrickx et al., 2012).

A second goal of the HRAS-DAF pilot was to produce more concentrated sludge compared to conventional A-stage systems with a settler where sludge is concentrated to ± 10 g COD L-1 56

(De Graaff & Roest, 2012). Producing concentrated sludge should be possible due to the coupling with DAF. It has already been proven on lab scale that an A-stage-DAF system was able to produce sludge twice as concentrated (16 – 20 g COD L-1) (De Saedeleer, 2016). The higher concentration ability of the DAF makes the use of a thickener before AD unnecessary. In this perspective, the DAF can be seen as a combined settler and thickener. The higher energy use of a DAF compared to a settler sets the higher operating costs for DAF solid/liquid separation.

Organics removal

The organics removal in the HRAS-DAF pilot plant during the three treatments will be discussed in the following sections. Three processes regulate the organics removal:

1. Biological growth 2. Adsorption, coagulation and (bio)flocculation 3. Solid/liquid separation in the DAF

Fast biological growth occurs in both A-stage and HiCS process and part of the sCOD.is consumed or stored. During every treatment the ratio of sCOD:tCOD ration in the influent was higher than 25 %. Research done by de Graaff et al. (2016) shows that in this case an A-stage is better suited as primary treatment than a primary settler. In the HiCS system the sludge with the best biosorption capacity is selected via the feast and famine regime and should result in the highest pCOD and cCOD removal via biosorption. Bioflocculation is key for optimal sludge floc formation and chemicals are dosed to enhance the flocculation. The performance of the HRAS-DAF pilot is inherently related to the performance of the DAF unit. A good solid/liquid separation in the DAF unit strongly defines the overall removal efficiency of the HRAS-DAF pilot. The performance of the DAF unit itself relies on effective coagulation and flocculation and on the quality of the microbubbles. Both proved to be critical for stand-alone operation.

1.1.1 A-stage-DAF T1

Before day 64 and the implementation of the PFF, T1 faced several days of operation with low TSS removal (< 30 %) and also the overall solids removal was low (50 %). This had several causes. 57

First of all, on the days with low TSS removal the sludge flocs after coagulation and flocculation appeared to be small and weak (based on subjective visual criteria). Whereas the coagulant could be dosed just before the screw pump for optimal mixing, this could not be done for the flocculant as this would destroy the polymer structure (D. H. Liu & Liptak, 1999). It is likely that the flocculant was not enough dispersed into the MLSS to create stable flocs because of the short distance between the polymer addition and the entrance of the DAF. Unstable flocs are more prone to disruption when entering the turbulent contact zone of the DAF unit. The importance of stable flocs for flotation was shown by Klute et al. (1995) and flocs that are broken by shear do not regrow easily under less turbulent conditions (Yoon & Deng, 2004).

Secondly, it was assumed that the bubbles floc interaction was hampered due to inefficient charge neutralisation. Bubbles are hydrophobic and for bubble – floc adhesion it is needed that the negative charges on the particles are neutralised effectively via addition of FeCl3. The membrane pump that pumped the FeCl3 solution to the screw pump was working near its minimum flow rate.. This led to intermittent pulses being dosed instead of a more continuous flow of coagulant solution. This could be countered partly by operating the membrane pump in slow operation mode, which made the plunger movement smaller but faster. To tackle this problem, during T2 the concentration of the FeCl3 solution was decreased from 10 % to 5 %.

Thirdly, as a result of the inefficient solid/liquid separation, the sludge production on days with low TSS removal was not sufficient to keep the MLSS concentration in the A-stage tank at 1 g L-1 via sludge recirculation. This caused the MLSS concentration to drop below the desired concentration and that in turn led to an overdosing of coagulant and flocculant since the dosages were optimised for ± 1 g MLSS L-1. Overdosing of coagulant leads to charge inversion and re- stabilisation of the particles in suspension. Overdosing of A130hp polymer should be avoided since bridging flocculation is hampered. The optimal dosage for polymer bridging is reached when roughly half of the surface of the particle is covered with polymer as then the chance of bridge formation is at its largest (Bolto & Gregory, 2007). In addition, overdosing of polymer causes the flocs to become too heavy for flotation partially due to the high molecular weight of the polymers used to assist flocculation (particularly in case of A130hp with molecular weight (MW) higher than 107 g mol-1).

The performance of the A-stage-DAF system improved when the PFF was connected to the system and the polymers were dosed directly in the PFF. Organics removal was increased from ± 50 % for all three parameters (TSS, VSS and tCOD) during the first 64 days to 77 ± 14 % for TSS, 81 ± 10 % for VSS and 62 ± 13 % for tCOD from day 64 until day 106.

The improvement in solid/liquid separation was likely due to the implementation of the PFF, which increased turbulence and shear forces and resulted in the formation of stable sludge flocs more suitable for flotation. Under non-turbulent conditions the polymers in solution are randomly coiled, but when subjected to high shear, the polymer chains are considerably extended (Bolto & Gregory, 2007). Further, covalent bonds around the middle of the polymer chain can rupture, especially in the case of high MW polymers, leading to a reduction in MW (Scott et al., 1996). This causes a reduction in viscosity and is important for flocculation under turbulent conditions. The turbulent conditions in the PFF likely increased dispersion of coagulant and flocculant into the MLSS. In addition, roughly half of the pressurized white water was added directly in the PFF. This likely further increased turbulence, increased flocs buoyancy and led to stronger bubbles-floc adhesion. Finally, the PFF provided more time for flocculation and the dosage points of bubbles and flocculant could be optimised.

1.1.2 A-stage-DAF T2

During T2, FeCl3 was dosed with the same concentration as in T1, but it was chosen to change from using solely anionic A130hp polymer to a combination of cationic C492 and A130hp polymer to assist flocculation. This combination is called ‘dual polymer’ and the combination of a cationic flocculant and a high MW (> 107 g mol-1) anionic flocculant creates stronger and more compact flocs (Petzold et al., 2003). By using the dual polymer system, the microbubbles should be more firmly incorporated into the floc.

First the cationic polymer is dosed and it reduces the charge on the particles via charge neutralisation. Cationic patches arise according to the electrostatic patch mechanism: because of the higher charge density of cationic polymer compared to the surface of the particles, it is not possible that each surface charged site is neutralised by a cationic polymer segment (Bolto & Gregory, 2007). Further in the PFF the high MW anionic A130hp polymer is dosed and the cationic patches act as anchor points for anionic polymer binding. Since the number of patches is limited, the anionic polymer cannot bind several times to the same particle and will adopt a

59 more extended configuration, which enhances the flocculation via ion bridging. It was shown by Petzold et al. (2003) that the addition of cationic polymer followed by anionic polymer is the most effective order for flocculation. The dual polymer system assures a thoroughly neutralisation of the charge on the particles which improves the floc attachment to the hydrophobic flocs.

Due to an almost constant effluent concentrations in terms of TSS, VSS and tCOD, independently from the influent concentration, it was assumed that the system would be able to further improve removal efficiencies if medium-strength or high-strength wastewater would be treated.

1.1.3 HiCS-DAF T3

The HiCS process was chosen as earlier studies showed that HiCS could increase carbon capture compared to the A-stage (Meerburg et al., 2015). In this thesis similar removal efficiencies were obtained for the HiCS-DAF system as for the A-stage-DAF system. Nevertheless the amount of coagulant and flocculant dosed in HiCS-DAF configuration were lower than in the previous treatments. Therefore, the presence of an aerated stabilization tank seemed to enhance bioflocculation and floc formation, so that a lower amount of chemicals was necessary to obtain similar outputs. T3 was characterized by a lower F/M ratio (4.5) compared to T2 (7) and this increases the share of carbon storage for carbon capture (Lim et al., 2015).

DAF performance

Adequate aggregation of the particulates is crucial for efficient flotation (Klute et al., 1995). This stresses the importance of successful coagulation and flocculation. During operation, the DAF unit needed to be drained several times to remove settled sludge. To cope with this problem, the sand drain valve was set to open more regularly.

A sensitive parameter regarding the microbubble quality is the air flow (J Bratby & Marais, 1976). The optimal flow was between 0.5 – 0.7 L air min-1 and higher air flows did not yielded better flotation. Further, impurities in the recycle stream were harmful for the microbubble quality. The small tubing, specific to a pilot scale plant, clogged easily and had to be flushed several times a week. These clogging problems would less likely be encountered at larger scale. 60

It can be assumed that process stability would improve by installing a filter before the recycle tubing. This however comes along with an additional investment cost and higher operating costs due to a higher pressure drop and the need for filter backwashing/replacement.

The performance of the DAF unit during the different treatments will be discussed in the following sections.

1.2.1 A-stage-DAF T1

Via sludge recirculation, it was tried to maintain a MLSS concentration of 1 g TSS L-1 in the A-stage/contact tank but due to several reasons there was a wide spread in MLSS concentrations (1.08 ± 0.42 g TSS L-1). As mentioned before MLSS dropped when problems with the DAF occurred. Sludge production was too low to assure enough recirculation to the A-stage tank and so the MLSS concentration went down several times during T1. Next, the sludge recirculation was fixed via an on/off setting of the screw pump. When the sludge characteristics changed but recirculation remained the same, this influenced the MLSS concentration. The WWTP receives wastewater from a combined sewer system so rain events diluted the wastewater and this also had an impact on the MLSS concentration. Al these factors led to suboptimal dosing of coagulant and flocculant and thus to suboptimal flotation.

1.2.2 A-stage-DAF T2

During T2, not enough solids could be recirculated to keep the MLSS at the desired concentration of 1 g L-1 (0.75 ± 0.23 g L-1). Nevertheless, the separation efficiency was excellent for TSS and VSS (> 95 %) and tCOD (90 %).

1.2.3 HiCS-DAF T3

The conversion from A-stage to HiCS was not a problem for flotation. Similar separation efficiencies as during the two A-stage-DAF treatments were obtained.

1.2.4 Effluent characteristics

The effluent characteristics are discussed to show to which extent the organics could be removed by the HRAS-DAF systems. The HRAS-DAF pilot performance was the main factor defining the effluent characteristics, but also the influent strength influenced the effluent

61 concentration as the influent was less concentrated (but not significantly) during T2 and T3 compared to T1. The influent vs effluent strength is compared in Figure 26.

TSS VSS tCOD

0.5 1)

- 0.4

0.3

0.2

0.1 TSS, VSS, tCOD (g VSS, L (g TSS, tCOD

0.0 Influent Effluent Influent Effluent Influent Effluent Treatment 1 Treatment 2 Treatment 3 A-stage-DAF HiCS-DAF

Figure 26: Comparison of influent and effluent TSS, VSS and tCOD for T1, T2 and T3. The error bars show 1 standard deviation.

Different DAF operation strategies can be chosen in function of the desired effluent. In this thesis, optimal organics removal was pursued through optimization of coagulant and flocculant addition. Under these circumstances, the BOD:N ratios are low and the effluent cannot support 3- complete denitrification in the B-stage. Further, due to the high Fe:PO4 ratios (4 – 7) almost 3- no PO4 was left in the effluent which is detrimental for biological growth in the B-stage 3- 3- (especially the vulnerable nitrifiers). At a Fe:PO4 ratio of 1, about 80 % of PO4 is removed (Broeders, Menkveld, et al., 2014). A phosphate concentration of 1.5 mg L-1 is advised for good microbial growth. This could be achieved by applying a strategy in which the dosing of FeCl3 is controlled by on online monitoring of the effluent phosphate concentration. In contrast, the DAF could also be operated to have an effluent with a BOD:N ratio that allows denitrification in the B-stage (Broeders, Menkveld, et al., 2014).

Comparison between HRAS-DAF and HRAS-settling

To evaluate the HRAS-DAF systems on their solid/liquid separation efficiency and their sludge production and to evaluate whether it could be useful to implement DAF in existing A/B-

62 processes, the HRAS-DAF system performances obtained during this thesis are compared with four operational A/B systems in the Netherlands (De Graaff & Roest, 2012). The removal of TSS and tCOD for four full scale A-stage systems in the Netherlands where settling is used for solid/liquid separation is given in Table 9.

Table 9: Removal of TSS and tCOD for four full scale A-stages in the Netherlands (De Graaff & Roest, 2012)

WWTP TSS tCOD

Nieuwveer, Breda 59% 53%

Dokhaven, Rotterdam 68% 74%

Utrecht - 60%

Garmerwolde, Groningen 67% 55%

The tCOD removal for the HRAS-DAF systems was 50 – 62 % and this is comparable to what is achieved at the large scale A-stages in the Netherlands . Only the A-stage at Dokhaven achieved a considerably higher tCOD removal (74 %).

The TSS removal obtained via conventional settling can be compared to the TSS separation efficiency obtained with the DAF. Whereas the settlers were able to remove 59 – 68 % of the solids, the HRAS-DAF in this thesis could remove 90 – 97 % of the solids. This shows that DAF is a better solid/liquid separation technique. The lower solid/liquid separation efficiency of the settler can be contributed to the high F/M ratio (Bisogni & Lawrence, 1971). On the other hand, this does not seem to influence the flotation potential.

The best flotation results were obtained using the dual polymer system. Apart from selection the optimal combinations of coagulant and flocculant, also the costs of these chemicals should be accounted for. Market prices are €4 kg-1 polymer (Broeders, Menkveld, et al., 2014) and -1 FeCl3 comes at a cost of 0.25 – 0.3 € kg (Yonge, 2012). Both the efficacy of flotation and the cost of coagulant and flocculant use should be taken into consideration. A higher flotation efficacy must outweigh the higher costs related an increased chemical use. Further the influence of the coagulant and flocculant on the downstream AD cannot be neglected.

DAF The different energy consuming components of the HRAS-DAF installation are:

1. The influent pump 2. The mixers for the making of the polymer solutions 3. The pumps for the dosing of coagulant and flocculant 4. The pumps for removal of floated sludge and for the sand drain 5. The scraper system 6. The recirculation pump 7. The compressor The recirculation pump is the largest energy consumer and the total energy use of a large scale DAF installation sums up to ± 0.033 kWh m-3 (Broeders, Flameling, et al., 2014). The energy consumption of a primary settler for running influent and recirculation pumps is considerably lower and amounts to 1% of the total energy use at a WWTP (Wendland, 2005). Although a DAF unit has larger operational costs compared to a settler, it saves the capital expenditures of a thickener and saves a considerable amount of space due to its ten times lower footprint.

Sludge characteristics

Sewage sludge is a complex heterogeneous mixture of active and dead biomass, undigested organics, plant residues, oils, or faecal material, inorganic materials and moisture (Tyagi & Lo, 2013). Nearly 10 million tons of dry sludge are produced in the EU each year (Commission, 2010). Sludge treatment is expensive (€35 m-³ sludge, 22% solids) (Broeders, Menkveld, et al., 2014) and can amount up to 50 % of the operating costs of a WWTP. The better digestion properties of HRAS sludge should result in lower overall sludge production compared to single stage nutrient removal processes with primary sedimentation (Loosdrecht et al., 1997).

The HRAS-DAF sludge concentration was between 47 (T1) and 21 (T2) g COD L-1 which is high compared to a normal A-stage-settler in which sludge is only concentrated to ± 10 gCOD L-1 (De Graaff & Roest, 2012). The lower organic loading during T2 and T3 supported less biomass growth and thus less sludge production. This can explain why the HRAS-DAF sludge was less concentrated compared to T1. Another reason could be that a considerable amount of water was scraped off together with the sludge. This is mainly determined by the level in the DAF and the scraper speed.

The digestibility of the HRAS-DAF sludge was tested on pilot as well as on lab scale and these results will be discussed further.

COD balance

The ratio of COD that left via the effluent was similar for all treatments (± 40 %). The proportion of removed COD that was transferred to the sludge lay between 61 – 84 %. To compare with, the COD balance for a conventional CAS system (Ding et al., 2015) is shown in Figure 27. In the CAS system, 59 % of the removed COD ends up in the sludge. In this thesis a higher proportional COD capture was achieved using the HRAS-DAF processes. High DAF separation efficiency and a medium influent strength resulted in the highest COD capture (84 %) during the last part of A-stage-DAF (T1) operation. Probably because of the lower influent strength during A-stage-DAF (T2) and a higher SRT (1.39 d vs 1.11 d) compared to T1, only 65 % of the removed COD ended up in the sludge. Being confronted to a similar strength wastewater as T2, the HiCS-DAF (T3) performed slightly better with 72 % of the organics being sent to the sludge. The higher biosorption capacity of the HiCS sludge might be an explanation.

Figure 27: COD balance for a conventional CAS system. From the COD that is not left via the effluent, the proportion is given between COD in the sludge and COD that is mineralised. 65

2 Jar tests

Experiment 1: optimal polymer selection for flotation

The optimal polymer for achieving the best flotation was selected visually. A130hp polymer was seen to have the best flocculation and flotation abilities although it has a higher molecular weight than C492. The anionic nature of the polymer sets the mechanism of flocculation to be via ion bridging. The coagulant Fe3+ ions are bound on the negative charges of the flocs and the negative charged groups of the anionic polymer link to the positively charged Fe3+ ions. This way the anionic polymer can form a bridge between different flocs.

Experiment 2: optimal flocculant dosing (T1)

A dosage of 3 mg L-1 A130hp polymer was selected for the A-stage-DAF operation during T1 and the jar tests confirmed that this is the optimal dose. This is similar to a pilot study wherein DAF was investigated as wastewater pre-treatment technique and 1.8 – 2.8 mg polymer L-1 was dosed for maximal organics removal (Broeders, Menkveld, et al., 2014).

The average MLSS concentration was 1.08 g L-1 resulting in 2.8 mg polymer g-1 TSS being dosed. This is high compared to settling where 1 mg polymer g-1 TSS or less is applied (Bolto & Gregory, 2007). This test also showed that there is no benefit related to dosing more than 3 mg L-1 of the high MW A130hp polymer. Higher concentrations hamper bridging flocculation and the high MW of the polymer makes the flocs less susceptible for flotation.

Experiment 3: optimal flocculant dosing (T2)

In the first series of jar test, an optimal A130hp dosing of 0.5 mg L-1 was found. Further jar tests suggested that at this A130hp concentration in combination with either a low (0.8 mg L-1) or high (3.05 mg L-1) provided the best solid liquid separation. The low C492 concentration was attempted to run the A-stage-DAF pilot but no effective flotation was observed. So the C492 concentration was increased to 2 mg L-1. Although this is not the optimal concentration 66 as could be derived from the jar test, this concentration assured good flocculation and flotation at the A-stage-DAF pilot. This experiment showed that dosing more than 2 mg L-1 of C492 does not yield higher separation efficiencies for TSS. At a total dosage of 2.5 mg polymer L-1 during T2, with an average MLSS concentration of 0.75 g L-1, 3.3 mg polymer g-1 TSS was dosed, which is even higher compared to T1. From an economical viewpoint, the higher separation efficiency of the DAF should justify the high polymer consumption.

Experiment 4: optimal FeCl3 and C492 dosing (T3)

This experiment proved that is was advantageous to lower the FeCl3 concentration from 50 mg L-1 (as was applied during T1 & T2) to 35 mg L-1. Eventually 30 mg L-1 was applied to the HiCS-DAF pilot during T3. At a total dosage of 2 mg polymer L-1, with an average MLSS concentration of 1.17 g L-1, 1.7 mg polymer g-1 TSS was dosed. This was the lowest polymer dosing of the three treatments.

3 BMP tests

BMP T1

Conversion efficiencies to CH4 of 58 – 67 % were observed during BMP tests on A-stage-DAF sludge yielded during T1. This was comparable to A-sludge generated from a conventional high rate sludge where settling is used for solid/liquid separation (Bolzonella et al., 2005; De Vrieze et al., 2013).

BMP T2

Conversion efficiencies to CH4 (24 ± 2 %) indicated the lower (compared to T1) digestibility of the A-stage-DAF sludge yielded during T2. It was assumed that methanogenesis could be inhibited by a lack of phosphorous and other (micro)nutrients in the sludge, toxicity caused by components in the wastewater or a toxic effect of the coagulant or the flocculant.

The additional BMP that were performed showed that AD could not be restored to its original state by addition of M9 minimal feeding solution (containing phosphorous) or vitamins plus trace elements or a combination of these. Although there was an improvement for M9 addition, the positive effect on AD was the largest for addition of vitamins plus trace elements (39 % conversion).

The BMP that was performed on the WAS from Aartselaar had conversion efficiencies (23 %) that were comparable to other WAS (Bolzonella et al., 2005). This meant that the influent wastewater can be excluded from being a reason for the failing AD.

A possible explanation for the hampered methanogenesis is a toxic effect of the coagulant or flocculant. No toxic effect is to be expected from the FeCl3 coagulant since iron is an important trace element for methanogens and stabilizes the AD process (Zhang & Jahng, 2012). That is because iron acts as cofactor in several enzymatic reactions occurring during methanogenesis, and it disproportionally stimulates the methanogens over the fermentative bacteria thus enhancing AD.

No polymer toxicity data were found for bacteria nor archaea. Generally synthetic polymers tend not to be readily absorbed by organisms whereas the monomers are more toxic. Cationic polymers are rated more toxic to aquatic organisms than anionic polymers (Hamilton et al., 1994). Further, algae are sensitive to anionic polymers (order 1 mg L-1) because of the chelation of nutrient metal cations (Murgatroyd et al., 1996). The same mechanisms might hinder the methanogens. However no AD inhibition was seen during T1 when only anionic polymer was applied. Besides, no inhibition was seen neither for AD of high rate sludge that was treated with cationic polymer (De Saedeleer, 2016). The AD problems arose during T2 and T3 so it seems that the combination of cationic and anionic polymer is causing an obstacle for the methanogens.

It is hypothesized that first easily biodegradable organics are consumed, after which the PAMs are consumed by the MO’s. PAMs can both be used as a carbon or as a nitrogen source (Dai et al., 2014). When PAM is used an a N-source, this is accompanied by an accumulation of ammonium. No increase in ammonium was found which suggests that the PAMs were solely used as carbon source which in turn leads to accumulation of acryl monomer (Dai et al., 2014).

Eventually, when most of the PAM is used, the methanogenesis almost ceased. The reason for this could be an improved stability of the flocs could leading to a mass transfer resistance (Chu et al., 2003) so that the less biodegradable organics in the flocs are not degraded by the MO’s.

BMP T3

Conversion efficiencies to CH4 were 40 ± 2 % and thus lower compared to T1 of the HiCS- DAF configuration. The low conversion efficiency could be explained as for T2 by the combination of the anionic and cationic polymer.

4 Fed batch digesters

40 – 70 % conversion was obtained during the first SRT, but then dropped to 15 %. Addition of phosphorous salts on day 55 and micronutrients on day 57 only slightly improved the process (20 % conversion). It seemed that the nutrients in the inoculum were exhausted or no longer available for the MOs. With fresh inoculum, methanogenesis proceeded unhindered but after 18 – 20 days (equal to 1 SRT), the process was inhibited. It is likely that the HRAS-DAF sludge is the reason for the failing methanogenesis.

5 General conclusions

In this thesis, the feasibility of combining an A-stage or HiCS process at pilot scale for municipal wastewater treatment with DAF for solid-liquid separation was demonstrated. High removal efficiencies of TSS and VSS (> 70 %) and tCOD (58 – 62 %) were obtained during A- stage-DAF operation, even when confronted to low strength (< 0.2 g COD L-1) wastewater. During HiCS-DAF operation the removal of TSS (68 %), VSS (67 %) and tCOD (54 %) was comparable to the A-stage-DAF but nevertheless the HiCS performed better than the A-stage as less coagulant and polymer were dosed.

The high surface loadings (20 m h-1) determine the low footprint of the DAF installation compared to a settler (1.2 m h-1) and the ease of implementation make it a promising technique for the conversion or expansion of existing A/B-systems or the building of new plants. However the DAF process showed to be vulnerable to failure. In the view of standalone operation, there are some critical aspects that require intensive surveillance.

First of all, optimal coagulation and flocculation and good bubble floc attachment are crucial -1 for the flotation process. Using FeCl3 (30 – 50 mg L ) in combination with either solely anionic polymer (2.8 mg g-1 TSS) during T1 or a dual polymer system combining cationic and anionic polymer (1.7 – 3.3 mg g-1 TSS) during T2 and T3, flotation was proven to be successful. Introducing part of the microbubbles in the MLSS – coagulant – polymer mixture under turbulent conditions in a plug flow flocculator enhanced bubble – floc attachment and also flotation efficiency.

Second, the quality of the recycle stream water was crucial for qualitative microbubble generation since impurities disturb the generation of the microbubbles. A deteriorating effluent quality can lead to process failure.

This thesis showed that HRAS-DAF processes are capable of capturing a significant proportion (61 – 84 %) of the removed organics and diverting it to AD for biogas production. DAF proved to be an excellent alternative compared to settling in the view of energy recovery since less organics were washed out with the effluent and more concentrated sludge (21 – 47 g COD L-1) 70 was produced. The latter eliminates the need a of a thickener which further lowers the footprint and the capital expenditures of a WWTP.

The high rate operation of the HRAS-DAF pilot plant lead to the expectation for high digestibility of the HRAS-DAF sludge. High potential COD to CH4 conversion (58 – 67 %) was seen during BMP tests on A-stage-DAF sludge from T1 but AD was inhibited during T2 and T3 (24 – 42% conversion). The use of a combination of cationic and anionic polymer was assumed to be one of the main reasons why AD failed. Via addition of vitamins and trace elements it was possible to restore AD only partially.

6 Future work and perspectives

Further research of the HRAS-DAF system should focus on the biological activity and its role in uptake and release of sCOD. Next, the biosorption capacity of the HRAS-DAF sludge should be investigated and related to the biosorption capacity of activated sludge (40 – 100 mg COD g-1 TSS) (Guellil et al., 2001). It was shown that high rate activated sludge microbial communities differ significantly from conventional activated sludge communities (Gonzalez- Martinez et al., 2016). An interesting future research question is whether there is also a significant difference in bacterial community between floated high rate sludge and settled high rate sludge. This could be useful for inoculation of new HRAS-DAF systems.

Effective coagulation and flocculation is crucial for flotation. The dual polymer system (Petzold et al., 2003) that was applied during T2 and T3 guaranteed successful flotation, but still different polymer combinations with differing charge densities and molecular weights should be investigated. Further optimisation of the coagulation and flocculation could be achieved by dosing the polymer proportionally to the MLSS solids content, which varies with varying influent conditions. This makes sense since polymer bridging efficiency is at its maximum when half of the solids surface is occupied and the surface increases linearly with the solids content.

During this thesis, the HRAS-DAF sludge was seen not to be concentrated until is potential maximum of 60 g COD L-1. Further maximisation of the HRAS-DAF sludge concentration lowers the costs of sludge handling and reduces reactor volumes for AD. Based on TSS measurement in the small sludge storage basin of the DAF unit, the scraper speed and the water level in the DAF could be controlled to prevent large amounts of water being scraped off together with the sludge and thus diluting it.

The stability of the flocs might make it impossible for the MO’s to break down the organics in the flocs during AD. Sludge pre-treatment techniques before AD should be investigated to destabilize the flocs and make the highly biodegradable organics accessible for the MO’s. Further, the availability of nutrients and trace elements in the sludge should be investigated.

As the HRAS-DAF was built to resemble the A-stage of an A/B process, the focus should now be on the development of the B-stage and its coupling to the A-stage. The carbon limited effluent brings new challenges for nitrogen removal in the B-stage. There are two possible pathways that require investigation. Conventional nitrification/denitrification or partial nitrification/annamox via the cold annamox process (Kartal et al., 2010). Batch tests already were performed on denitrification in the effluent. The building of a continuous reactor should be the next step.

PART 5: REFERENCE LIST

Adachi, Y. (1995). Dynamic aspects of coagulation and flocculation. Advances in colloid and interface science, 56, 1-31. Agarwal, A., Ng, W. J., & Liu, Y. (2011). Principle and applications of microbubble and nanobubble technology for water treatment. Chemosphere, 84(9), 1175-1180. Aguilar, M., Saez, J., Llorens, M., Soler, A., & Ortuno, J. (2002). Nutrient removal and sludge production in the coagulation–flocculation process. Water research, 36(11), 2910-2919. Appels, L., Baeyens, J., Degrève, 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. Aziz, H. A., Alias, S., Adlan, M. N., Asaari, A., & Zahari, M. S. (2007). Colour removal from landfill leachate by coagulation and flocculation processes. Bioresource technology, 98(1), 218-220. Batstone, D. J., & Virdis, B. (2014). The role of anaerobic digestion in the emerging energy economy. Current opinion in biotechnology, 27, 142-149. Bisogni, J. J., & Lawrence, A. W. (1971). Relationships between biological solids retention time and settling characteristics of activated sludge. Water research, 5(9), 753-763. Boccaletti, G. (2009). Charting our water future. London: McKinsey. Boe, K., & Angelidaki, I. (2006). Online monitoring and control of the biogas process. Technical University of DenmarkDanmarks Tekniske Universitet, Department of Systems BiologyInstitut for Systembiologi. Boehnke, B., Schulze-Rettmer, R., & Zuckut, S. (1998). Cost-effective reduction of high- strength wastewater by adsorption-based activated sludge technology. J. Wat. Eng. Manage, 145(3), 1-34. Bolto, B., & Gregory, J. (2007). Organic polyelectrolytes in water treatment. Water research, 41(11), 2301-2324. Bolzonella, D., Pavan, P., Battistoni, P., & Cecchi, F. (2005). Mesophilic anaerobic digestion of waste activated sludge: influence of the solid retention time in the wastewater treatment process. Process biochemistry, 40(3), 1453-1460. Bratby, J. (2006). Coagulation and flocculation in water and wastewater treatment: IWA publishing. Bratby, J., & Marais, G. (1976). A guide for the design of dissolved-air (pressure) flotation systems for activated sludge processes. Water SA, 2(2), 87-100. Brittle, S., Desai, P., Ng, W. C., Dunbar, A., Howell, R., Tesař, V., & Zimmerman, W. B. (2015). Minimising microbubble size through oscillation frequency control. Chemical Engineering Research and Design, 104, 357-366. Broeders, E., Flameling, T., Kleinsman, S., Menkveld, H., van Nieuwenhuijzen, A., Schellekens, D., & Veldhoen, A. (2014). Dissolved Air Flotation (DAF) als voorbehandeling van communaal afvalwater: STOWA. Broeders, E., Menkveld, H., van Nieuwenhuijzen, A., & Veldhoen, A. (2014). DAF als voorbehandeling van communaal afvalwater demonstratieonderzoek RWZI Lienden: Stichting Toegepast Onderzoek Waterbeheer. Cagnetta, C., Coma, M., Vlaeminck, S. E., & Rabaey, K. (2016). Production of carboxylates from high rate activated sludge through fermentation. Bioresource technology, 217, 165-172.

Cams, K. (2005). Bringing Energy Efficiency to the Water Wastewater industry How Do We Get There ln: WEFI'EC Proceedings. Chen, Y., Cheng, J. J., & Creamer, K. S. (2008). Inhibition of anaerobic digestion process: a review. Bioresource technology, 99(10), 4044-4064. Chu, C., Lee, D., Chang, B.-V., You, C., Liao, C., & Tay, J. (2003). Anaerobic digestion of polyelectrolyte flocculated waste activated sludge. Chemosphere, 53(7), 757-764. Commission, E. (2010). Report from the Commission to the Council and the European Parliament on sustainability requirements for the use of solid and gaseous biomass sources in electricity, heating and cooling: Publications Office of the European Union. Connor, R. (2015). The United Nations world water development report 2015: water for a sustainable world (Vol. 1): UNESCO Publishing. Constantine, T., Houweling, D., & Kraemer, J. (2012). “Doing the Two-Step”–Reduced Energy Consumption Sparks Renewed Interest in Multistage Biological Treatment. Proceedings of the Water Environment Federation, 2012(10), 5771-5783. Dai, X., Luo, F., Yi, J., He, Q., & Dong, B. (2014). Biodegradation of polyacrylamide by anaerobic digestion under mesophilic condition and its performance in actual dewatered sludge system. Bioresource technology, 153, 55-61. Davis, R., & Hall, J. (1997). Production, treatment and disposal of wastewater sludge in Europe from a UK perspective. European water pollution control, 2(7), 9-17. De Graaff, M. S., & Roest, K. (2012). Inventarisatie van AB-systemen-optimale procescondities in de A-trap. KWR, Nieuwegein. de Graaff, M. S., van den Brand, T. P., Roest, K., Zandvoort, M. H., Duin, O., & van Loosdrecht, M. C. (2016). Full-scale highly-loaded wastewater treatment processes (A- Stage) to increase energy production from wastewater: performance and design guidelines. Environmental Engineering Science, 33(8), 571-577. De Rijk, S. E., & den Blanken, J. G. (1994). Bubble size in flotation thickening. Water research, 28(2), 465-473. De Saedeleer, I. (2016). Development and optimization of an A-stage–DAF system. De Vrieze, J., De Lathouwer, L., Verstraete, W., & Boon, N. (2013). High-rate iron-rich activated sludge as stabilizing agent for the anaerobic digestion of kitchen waste. Water research, 47(11), 3732-3741. Deublein, D., & Steinhauser, A. (2011). Biogas from waste and renewable resources: an introduction: John Wiley & Sons. Ding, H.-B., Doyle, M., Erdogan, A., Wikramanayake, R., & Gallagher, P. (2015). Innovative use of dissolved air flotation with biosorption as primary treatment to approach energy neutrality in WWTPs. Water Practice and Technology, 10(1), 133-142. Edzwald, J. (2010). Dissolved air flotation and me. Water research, 44(7), 2077-2106. Edzwald, J. K. (1995). Principles and applications of dissolved air flotation. Water Science and Technology, 31(3–4), 1-23. doi: http://dx.doi.org/10.1016/0273-1223(95)00200-7 Edzwald, J. K., Tobiason, J. E., Amato, T., & Maggi, L. J. (1999). Integrating high-rate DAF technology into plant design. American Water Works Association. Journal, 91(12), 41. Ekama, G., & Wentzel, M. (2008). Organic material removal. Biological Wastewater Treatment: Principles, Modelling and Design. Edited by M. Henze, MCM van Loosdrecht, GA Ekama and D. Brdjanovic. Published by IWA Publishing, London, UK, 53-86. EPA. (2003). Biosolids Technology Fact Sheet: Gravity Thickening: United States Environmental Protection Agency. Frijns, J., Hofman, J., & Nederlof, M. (2013). The potential of (waste) water as energy carrier. Energy Conversion and Management, 65, 357-363.

Garrido, J. M., Fdz-Polanco, M. , Fdz-Polanco, F. (2013). Working with energy and mass balances: a conceptual framework to understand the limits of municipal wastewater treatment. Water Sci. Technol. 67, 2294. Gonzalez-Martinez, A., Rodriguez-Sanchez, A., Lotti, T., Garcia-Ruiz, M.-J., Osorio, F., Gonzalez-Lopez, J., & van Loosdrecht, M. C. (2016). Comparison of bacterial communities of conventional and A-stage activated sludge systems. Scientific reports, 6. Greenberg, A., Clesceri, L., & Eaton, A. (1992). Standard methods for the examination of water and wastewater American Public Health Association (APHA). Washington, DC, USA, 4-77, 74-78 and 74-81. 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. Gujer, W., & Jenkins, D. (1975). The contact stabilization activated sludge process—oxygen utilization, sludge production and efficiency. Water research, 9(5-6), 553-560. Haarhoff, J. (2008). Dissolved air flotation: progress and prospects for drinking water treatment. Journal of Water Supply: Research and Technology-AQUA, 57(8), 555-567. Haarhoff, J., & Edzwald, J. K. (2004). Dissolved air flotation modelling: insights and shortcomings. Journal of Water Supply: Research and Technology-AQUA, 53(3), 127- 150. Hamilton, J. D., Reinert, K. H., & Freeman, M. B. (1994). Aquatic risk assessment of polymers. Environmental science & technology, 28(4), 186A-192A. Heidrich, E., Curtis, T., & Dolfing, J. (2010). Determination of the internal chemical energy of wastewater. Environmental science & technology, 45(2), 827-832. Hendrickx, T. L., Wang, Y., Kampman, C., Zeeman, G., Temmink, H., & Buisman, C. J. (2012). Autotrophic nitrogen removal from low strength waste water at low temperature. Water research, 46(7), 2187-2193. Henze, M., Harremoes, P., la Cour Jansen, J., & Arvin, E. (2001). Wastewater treatment: biological and chemical processes: Springer Science & Business Media. Henze, M., van Loosdrecht, M. C., Ekama, G. A., & Brdjanovic, D. (2008). Biological wastewater treatment: IWA publishing. Higgins, M. J., & Novak, J. T. (1997). Dewatering and Settling of Activated Sludges: The Case for Using Cation Analysis. Water Environment Research, 69(2), 225-232. Jetten, M. S., Horn, S. J., & van Loosdrecht, M. C. (1997). Towards a more sustainable municipal wastewater treatment system. Water Science and Technology, 35(9), 171- 180. Jimenez, J., Miller, M., Bott, C., Murthy, S., De Clippeleir, H., & Wett, B. (2015). High-rate activated sludge system for carbon management–Evaluation of crucial process mechanisms and design parameters. Water research, 87, 476-482. Kartal, B., Kuenen, J., & Van Loosdrecht, M. (2010). Sewage treatment with anammox. Science, 328(5979), 702-703. Kemira. (2010). Product Data Sheet Superfloc C-490 series cationic dry PAMs. Paper presented at the Product Data Sheet. http://www.strykerchem.com/docs/SuperflocC- 496TDS.pdf Klute, R., Langer, S., & Pfeifer, R. (1995). Optimization of coagulation processes prior to DAF. Water Science and Technology, 31(3-4), 59-62. Koivunen, J., & Heinonen‐Tanski, H. (2008). Dissolved air flotation (DAF) for primary and tertiary treatment of municipal wastewaters. Environmental technology, 29(1), 101-109. Kroeker, E., Schulte, D., Sparling, A., & Lapp, H. (1979). Anaerobic treatment process stability. Journal (Water Pollution Control Federation), 718-727.

Leppinen, D., & Dalziel, S. (2004). Bubble size distribution in dissolved air flotation tanks. Journal of Water Supply: Research and Technology-AQUA, 53(8), 531-543. Lim, C.-P., Zhang, S., Zhou, Y., & Ng, W. J. (2015). Enhanced carbon capture biosorption through process manipulation. Biochemical Engineering Journal, 93, 128-136. Liu, D. H., & Liptak, B. G. (1999). Environmental Engineers' Handbook CRC press. Liu, Y., & Fang, H. H. (2003). Influences of extracellular polymeric substances (EPS) on flocculation, settling, and dewatering of activated sludge. Loosdrecht, M. C. v., Kuba, T., Veldhuizen, H. M. v., Brandse, F. A., & Heijnen, J. J. (1997). Environmental impacts of nutrient removal processes: case study. Journal of environmental engineering, 123(1), 33-40. Lundh, M., Jönsson, L., & Dahlquist, J. (2002). The influence of contact zone configuration on the flow structure in a dissolved air flotation pilot plant. Water research, 36(6), 1585- 1595. McCarty, P. L., Bae, J., & Kim, J. (2011). Domestic wastewater treatment as a net energy producer–can this be achieved? Environmental science & technology, 45(17), 7100- 7106. Meerburg, F. A., Boon, N., Van Winckel, T., Vercamer, J. A., Nopens, I., & Vlaeminck, S. E. (2015). Toward energy-neutral wastewater treatment: A high-rate contact stabilization process to maximally recover sewage organics. Bioresource technology, 179, 373-381. Miron, Y., Zeeman, G., Van Lier, J. B., & Lettinga, G. (2000). The role of sludge retention time in the hydrolysis and acidification of lipids, carbohydrates and proteins during digestion of primary sludge in CSTR systems. Water research, 34(5), 1705-1713. Mo, W., & Zhang, Q. (2012). Can municipal wastewater treatment systems be carbon neutral? Journal of environmental management, 112, 360-367. Mo, W., & Zhang, Q. (2013). Energy–nutrients–water nexus: integrated resource recovery in municipal wastewater treatment plants. Journal of environmental management, 127, 255-267. Modin, O., Persson, F., Wilén, B.-M., & Hermansson, M. (2016). Nonoxidative removal of organics in the activated sludge process. Critical Reviews in Environmental Science and Technology, 46(7), 635-672. Murgatroyd, C., Barry, M., Bailey, K., & Whitehouse, P. (1996). A review of polyelectrolytes to identify priorities for EQS development. Environment Agency, Research and Development Technical Report P, 21. Ødegaard, H. (2001). The use of dissolved air flotation in municipal wastewater treatment. Water Science and Technology, 43(8), 75-81. Persson, M., Jönsson, O., & Wellinger, A. (2006). Biogas upgrading to vehicle fuel standards and grid injection. Paper presented at the IEA Bioenergy task. Petzold, G., Mende, M., Lunkwitz, K., Schwarz, S., & Buchhammer, H.-M. (2003). Higher efficiency in the flocculation of clay suspensions by using combinations of oppositely charged polyelectrolytes. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 218(1), 47-57. Poh, P., Ong, W. Y. J., Lau, E. V., & Chong, M. N. (2014). Investigation on micro-bubble flotation and coagulation for the treatment of anaerobically treated palm oil mill effluent (POME). Journal of Environmental Chemical Engineering, 2(2), 1174-1181. Pujol, R., & Canler, J. (1992). Biosorption and dynamics of bacterial populations in activated sludge. Water research, 26(2), 209-212. Rabaey, K. (2014). Biotechnological Processes In Environmental Sanitation 2014-2015. Rahman, A. (2016). Management of Bioflocculation Through High-rate Contact-Stabilization: A Promising Technology to Recover Carbon from Low-Strength Wastewater. Paper

presented at the WEF/IWA Nutrient Removal and Recovery 2016: Advances in process intensification, resource extraction, and reuse. Rahman, A., Meerburg, F. A., Ravadagundhi, S., Wett, B., Jimenez, J., Bott, C., . . . De Clippeleir, H. (2016). Bioflocculation management through high-rate contact- stabilization: A promising technology to recover organic carbon from low-strength wastewater. Water research, 104, 485-496. Rehm, H., & Winter, J. (1999). Vol. 11a: Environmental processes I: wastewater treatment: Weinheim [etc.]: Wiley-VCH. Schofield, T. (2001). Dissolved air flotation in drinking water production. Water Science and Technology, 43(8), 9-18. Scott, J., Fawell, P., Ralph, D., & Farrow, J. (1996). The shear degradation of high‐molecular‐ weight flocculant solutions. Journal of applied polymer science, 62(12), 2097-2106. Sheik, A. R., Muller, E. E., & Wilmes, P. (2014). A hundred years of activated sludge: time for a rethink. Frontiers in microbiology, 5, 47. Shen, Y., Linville, J. L., Urgun-Demirtas, M., Mintz, M. M., & Snyder, S. W. (2015). An overview of biogas production and utilization at full-scale wastewater treatment plants (WWTPs) in the United States: challenges and opportunities towards energy-neutral WWTPs. Renewable and Sustainable Energy Reviews, 50, 346-362. Smitshuijzen, J., Pérez, J., Duin, O., & van Loosdrecht, M. C. (2016). A simple model to describe the performance of highly-loaded aerobic COD removal reactors. Biochemical Engineering Journal, 112, 94-102. Sorensen, J., Thornberg, D. E., & Nielsen, M. K. (1994). Optimization of a nitrogen-removing biological wastewater treatment plant using on-line measurements. Water Environment Research, 66(3), 236-242. Tchobanoglous, G., & Burton, F. L. (1991). Wastewater engineering. Management, 7, 1-4. Tchobanoglous, G., Darby, J., Bourgeous, K., McArdle, J., Genest, P., & Tylla, M. (1998). Ultrafiltration as an advanced tertiary treatment process for municipal wastewater. Desalination, 119(1), 315-321. Tiehm, A., Nickel, K., Zellhorn, M., & Neis, U. (2001). Ultrasonic waste activated sludge disintegration for improving anaerobic stabilization. Water research, 35(8), 2003-2009. Tyagi, V. K., & Lo, S.-L. (2013). Sludge: A waste or renewable source for energy and resources recovery? Renewable and Sustainable Energy Reviews, 25, 708-728. Verliefde, A. (2014). Analysis and abatement of wastewater pollution. Ghent: Ghent University. Verstraete, W., Clauwaert, P., & Vlaeminck, S. E. (2016). Used water and nutrients: Recovery perspectives in a ‘panta rhei’context. Bioresource technology, 215, 199-208. Verstraete, W., & Philips, S. (1998). Nitrification-denitrification processes and technologies in new contexts. Environmental pollution, 102(1), 717-726. Verstraete, W., Van de Caveye, P., & Diamantis, V. (2009). Maximum use of resources present in domestic ‘‘used water”. Bioresource Technology 100, 5537-5545. Wendland, A. (2005). Operation costs of wastewater treatment plants. Retrieved from Wett, B., Buchauer, K., & Fimml, C. (2007). Energy self-sufficiency as a feasible concept for wastewater treatment systems. Paper presented at the IWA Leading Edge Technology Conference. Xu, G., Yan, Z., Wang, Y., & Wang, N. (2009). Recycle of Alum recovered from water treatment sludge in chemically enhanced primary treatment. Journal of Hazardous materials, 161(2), 663-669. Yonge, D. (2012). A comparison of aluminum and iron-based coagulants for treatment of surface water in Sarasota County, Florida.

Yoon, S.-Y., & Deng, Y. (2004). Flocculation and reflocculation of clay suspension by different polymer systems under turbulent conditions. Journal of colloid and Interface Science, 278(1), 139-145. Zhang, L., & Jahng, D. (2012). Long-term anaerobic digestion of food waste stabilized by trace elements. Waste Management, 32(8), 1509-1515.