Volatile fatty acid production and application as external carbon source for denitrification

IN

DEGREE PROJECT ENVIRONMENTAL ENGINEERING, SECOND CYCLE, 30 CREDITS

,

STOCKHOLM SWEDEN 2020

Volatile fatty acid production and application as external carbon source for denitrification

CORA MICHELLE DÖHLER

Volatile fatty acid production and application as external carbon source for denitrification

CORA MICHELLE DÖHLER

Supervisor and Examiner

Professor Dr. ELZBIETA PLAZA

Supervisor and Examiner at TU Darmstadt Professor Dr. SUSANNE LACKNER

Degree Project in Environmental Engineering and Sustainable Infrastructure KTH Royal Institute of Technology

School of Architecture and Built Environment

Department of Sustainable Development, Environmental Science and Engineering SE-100 44 Stockholm, Sweden

© Cora Michelle Döhler 2020

Degree Project in Environmental Engineering and Sustainable Infrastructure Department of Sustainable Development, Environmental Science and Engineering School of Architecture and Built Environment

Royal Institute of Technology (KTH), SE-100 44 Stockholm, Sweden

Reference should be written as: Doehler, C.M. (2020) “Volatile fatty acid production and application as external carbon source for denitrification”

Sammanfattning

Det är möjligt att utveckla den nya generationen av avloppsreningsverk genom att ompröva avloppsreningsverk som resursanläggning. Därtill möjliggör det att uppnå miljömål som att minska koldioxidavtrycket och följa ökande utsläppskrav, t.ex. för kvävekoncentration, på ett mer hållbart sätt.

Denna forskningsstudie syftar till att analysera möjligheten att återcirkulera kol inom reningsverket i form av lättflyktiga fettsyror (engl. volatile fatty acids, VFAs), producerades genom samfermentering av primärslam och matavfall. Det erhållna fermenteringssubstratet används som extern kolkälla för att förbättra processeffektiviteten i en efterdenitrifikationsanläggning.

Två pilotskaliga fermenteringsreaktorer drevs i semikontinuerligt driftläge med endast en skillnad i pH.

Det var möjligt att utvärdera pH-påverkan på kolåtervinningsprocessen genom att kontrollera pH- värdet i en reaktor till pH 10, medan den andra reaktorn drevs utan pH-kontroll. På grund av reaktionerna som fortlöpte, justerade sig den icke-kontrollerade reaktorn själv till ett stabilt pH runt 5,4. Samfermenteringsprocessen övervakades genom veckoanalys av kemisk syreförbrukning i filtrerade prover (engl. soluble chemical oxygen demand, SCOD) och total mängd av VFAs (TVFA). Medan den alkaliska miljö i den första reaktorn gynnade en högre hydrolys av substratet, uppnådde den andra reaktorn en mer tydlig surgöring på grund av det lägre pH-värdet. Följaktligen innehåller SCOD i reaktorn som drivs utan pH-kontroll en större andel TVFA – 64 % av SCOD - jämfört med reaktorn som drivs vid pH 10, där TVFA utgör 40 % av SCOD.

Vidare analyserades den uppnådda fermenteringsgraden genom att beräkna nettoökningen av TVFA per gram VS, respektive VSS. En högre jäsningsgrad uppnåddes i sur miljö, vilket resulterade i en högre VFA-produktion jämfört med fermenteringsreaktorn som drevs vid pH 10. Därtill visade analys med gaskromtografi av de individuella VFA tydliga skillnader i sammansättning av substraten. Enligt rönen producerade reaktorn vid pH 10 mestadels ättiksyra (61 %) följt av propionsyra (18 %) och n-smörsyra (14 %). Däremot producerade fermenteringsreaktorn utan pH-kontroll mestadels n-kapronsyra (47 %) följt av ättiksyra (25 %) och n-smörsyra (16 %). Detta visar att trots att samma fermentationssubstrat användes för båda reaktorerna möjliggör den sura miljön i reaktorn utan pH-kontroll karboxylkedjeförlängningen från ättiksyra till n-kapronsyra.

Fermentationssubstraten av de två reaktorerna filtrerades, utspäddes till en koncentration av 5 g COD/L

och tillfördes som extern kolkälla, med ett kol/kväve-förhållande på 4,5, för att förbättra

denitrifikationen i två kontinuerliga drivna biofilmreaktorer med rörliga bärare (engl. moving bed

biofilm reactor, MBBR). En MBBR erhöll under hela experimentets gång den kolkälla som bildats under

alkaliska förhållanden och den andra MBBR:en erhöll motsvarande kolkälla som bildats i den sura

miljön i fermenteringsreaktorn utan pH-kontroll. Den maximala uppnådda denitrifikationskapaciteten

var ganska likartad för båda MBBR: 3,25 g NO

-N

/(m²·d) för den MBBR som opererades med den

alkaliska erhållen kolkällan och 3,38 g NO

-N

/(m²·d) för MBBR som erhöll den utspädda

fermenteringsvätskan bildad utan pH-kontroll. Emellertid uppnådde den MBBR som erhöll kolkällan

bildad i sura miljön en högre genomsnittlig denitrifikationskapacitet på 2.5 g NO

-N

/(m²·d) jämfört

med MBBR som fick kolkällan producerad genom fermentering vid pH 10 (1.8 g NO

-N

/(m²·d)). Den

lägre effektiviteten i den MBBR som fick den alkaliskt erhållna kolkällan orsakas av en ansamling av

NO

-N under denitrifikationsprocessen, vilket indikerar suboptimala förhållanden. Detta beror både på

sammansättningen av den tillförda kolkällan och ett högre totalt pH-värde under reduktionsprocessen,

vilket kan hämma fakultativa anaerober såsom denitrifierare. Trots det visar denna forskningsstudie att

båda de VFA-rika kolkällorna erhållna genom samfermentering av primärslam och matavfall är lämpliga

för att förbättra denitrifikationen av kommunalt avloppsvatten, varvid kolkällan som produceras genom

fermentering utan pH-kontroll uppnår en högre denitrifikationseffektivitet.

Zusammenfassung

Eine Neuinterpretation kommunaler Klärwerke als Rohstoff-Rückgewinnungsanlagen ermöglicht die Entwicklung der Kläranlagen der Zukunft. Umweltziele, wie die Reduktion des CO

-Fußabdrucks und die Einhaltung steigender Abwasserstandards im Hinblick auf die Stickstoffkonzentration können somit nachhaltiger erreicht werden. Diese Forschungsstudie zielt darauf ab, die Möglichkeit der Rückführung von Kohlenstoff in Kläranlagen in Form leichtflüchtiger Fettsäuren (engl. volatile fatty acids, VFAs) zu untersuchen. Diese VFAs werden durch Co-Fermentation von Primärschlamm und Lebensmittelabfäl- len erzeugt und als zusätzliche Kohlenstoffquelle einer nachgeschalteten Denitrifikation zugeführt, um die Prozesseffizienz zu steigern.

Zur Erzeugung der VFAs wurden zwei Fermentationsreaktoren halbkontinuierlich im Pilotmaßstab be- trieben, welche systematisch im pH-Wert variierten. Der Einfluss des pH-Wertes auf den Kohlenstoff- rückgewinnungsprozess konnte beurteilt werden, indem ein Reaktor auf pH 10 geregelt wurde, während dieser im zweiten Reaktor nicht beeinflusst wurde. In diesem stellte sich aufgrund ablaufender Reakti- onen ein stabiler pH-Wert um 5,4 ein. Der Co-Fermentationsprozess wurde durch wöchentliche Analyse des gelösten chemischen Sauerstoffbedarfs (engl. soluble chemical oxygen demand, SCOD) und der Ge- samtmenge an VFAs (TVFA) überwacht. Während die alkalischen Bedingungen in dem bei pH 10 be- triebenen Reaktor eine höhere Hydrolyse des Substrats ermöglichten, erreichte der zweite Reaktor auf- grund des niedrigeren pH-Werts eine stärkere Versäuerung. Folglich enthält der SCOD in dem Reaktor, der ohne pH-Regelung betrieben wurde, mit 64 % einen höheren Anteil an TVFA im Vergleich zu dem bei pH 10 betriebenen Reaktor mit 40 % TVFA.

Außerdem wurde der erreichte Fermentationsgrad durch Berechnung der Nettozunahme der TVFA pro Gramm flüchtige Feststoffe (VS) bzw. flüchtige suspendierte Feststoffe (VSS) erfasst. Ein höherer Fer- mentationsgrad konnte ohne pH-Regelung erzielt werden, welche eine höhere VFA-Ausbeute im Ver- gleich zur Fermentation bei pH 10 zeigt. Deutliche Unterschiede in der Zusammensetzung der gewon- nenen VFAs konnten durch Analyse mittels Gaschromatographie erfasst werden. Demzufolge entstand bei der Fermentation bei pH 10 hauptsächlich Essigsäure (61 %), gefolgt von Propionsäure (18 %) und n-Buttersäure (14 %). Im Gegensatz dazu, produzierte der Fermentationsreaktor ohne pH-Regelung überwiegend n-Capronsäure (47 %), gefolgt von Essigsäure (25 %) und n-Buttersäure (16 %). Trotz des gleichen Fermentationssubstrates, welches beiden Reaktoren zugeführt wurde, ermöglichen die sauren Bedingungen in dem Fermentationsreaktor ohne pH-Regelung, eine Verlängerung der Carbonsäureket- ten von Essigsäure zu n-Capronsäure.

Nach Filtration der in verschiedenen Milieus gewonnenen Fermentationssubstrate und Verdünnung auf

eine Konzentration von 5 g COD/L, wurden diese zwei im Pilotmaßstab kontinuierlich betriebenen

Fließbett-Biofilmreaktoren (engl. Moving bed biofilm reactor, MBBR) als zusätzliche Kohlenstoffquelle

zur Denitrifikation zugeführt. Über die gesamte Versuchsdauer wurden ein MBBR mit dem alkalisch

gewonnenen und der Andere mit dem im sauren Milieu erzeugten VFA-Mix betrieben. Das Kohlenstoff-

Stickstoff-Verhältnis (C/N Ratio) lag dabei bei 4,5. Beide MBBRs wiesen eine vergleichbare maximale

Denitrifikationsrate von 3,25 g NO

-N

/(m²·d) (VFAs pH 10) und 3,38 g NO

-N

/(m²·d) (VFAs pH un-

geregelt) auf. Der MBBR, welcher die im sauren Milieu rückgewonnene Kohlenstoffquelle erhielt, er-

reichte im Durchschnitt eine höhere Denitrifikationsrate von 2,5 g NO

-N

/(m²·d) als der MBBR, der

den bei pH 10 gewonnenen VFA-Mix erhielt (1,8 g NO

-N

/(m²·d)). Die im Vergleich geringere Effizi-

enz der alkalisch rückgewonnenen Kohlenstoffquelle wird durch eine NO

-N-Anreicherung während der

Denitrifikation verursacht, welche suboptimale Bedingungen während des Prozesses indiziert. Dies ist

sowohl auf die Zusammensetzung der zugeführten Kohlenstoffquelle, als auch auf einen insgesamt hö-

heren pH-Wert während des Reduktionsprozesses zurückzuführen, der fakultative Anaerobier, wie

bspw. Denitrifikanten, unterdrücken kann. Dessen ungeachtet zeigt diese Studie, dass beide durch Co-

Fermentation von Primärschlamm und Lebensmittelabfällen gewonnenen VFA-reichen Kohlenstoff-

quellen zur Verbesserung der Denitrifikation kommunalen Abwassers geeignet sind, wobei die durch

Fermentation ohne pH-Regelung erzeugte Kohlenstoffquelle eine höhere Effizienz aufweist.

Abstract

By rethinking wastewater treatment plants (WWTPs) as resource recovery facilities, it is possible to de- velop the next generation of WWTPs. Moreover, it allows to accomplish environmental goals, such as reducing the CO

footprint, and comply with increasing effluent standards regarding the concentration of nitrogen in a more sustainable way. This research study aims to analyse the possibility of recirculating carbon within WWTPs in form of volatile fatty acids (VFAs) produced by co-fermentation of primary sludge and food waste. The obtained fermentation liquid is utilised as carbon source to enhance the denitrification process in a post-anoxic denitrification plant setup.

Two pilot scale fermentation reactors were semi-continuously operated, systematically varying only in pH. By controlling one reactor to pH 10, while the second reactor was operated without pH control, it was possible to assess the influence of the pH on the carbon recovery process. Despite the pH not being controlled in the second fermentation reactor, it adjusted itself to a stable pH around 5.4. The co-fer- mentation process was monitored by weekly analysis of the SCOD and total amount of VFAs (TVFA).

While the alkaline conditions in the reactor operated at pH 10 allowed a higher hydrolysis of the sub- strate, the second reactor, operated without pH control, achieved a more distinct acidification, due to the lower pH. Consequently, the SCOD in the reactor operated without pH control contains a higher percentage of TVFA amounting to 64 % in comparison to the reactor operated at pH 10 with 40 % TVFA.

Furthermore, the achieved degree of fermentation was assessed by calculating the net increase of TVFA per gram of VS, respectively VSS. A higher degree of fermentation was achieved without pH control, resulting in a higher VFA yield compared to the fermentation reactor operated at pH 10. Moreover, anal- ysis of the individual VFAs by gas chromatography showed distinct differences in the composition of the fermentation liquids. According to the findings, the reactor operated at pH 10 produced mainly acetic acid (61 %), followed by propionic acid (18 %) and n-butyric acid (14 %). In contrast, the fermentation reactor operated without pH control produced mainly n-caproic acid (47 %), followed by acetic acid (25 %) and n-butyric acid (16 %). Despite the similar fermentation substrate supplied to both reactors, the acidic conditions in the reactor operated without pH control allowed carboxylic acid chain elongation from acetic acid to n-caproic acid, resulting in the main difference of the fermentation liquids.

The fermentation liquid of the two reactors was filtered, diluted to a concentration of 5 g COD/L and supplied as additional carbon source to enhance denitrification in two continuously operated pilot scale moving bed biofilm reactors (MBBR), applying a carbon-to-nitrogen ratio of 4.5. One of the denitrifica- tion MBBRs received the carbon recovered by fermentation at pH 10 as external carbon source, whereby the carbon source produced by fermentation without pH control was supplied to the other MBBR. The maximal achieved denitrification rate was quite similar for both MBBRs amounting to 3.25 g NO

- N

/(m²·d) for the MBBR receiving the carbon source recovered by co-fermentation at pH 10 and 3.38 g NO

-N

/(m²·d) for the MBBR receiving the VFA-mix obtained by co-fermentation without pH control. However, the MBBR provided with the carbon source recovered by co-fermentation under acidic conditions achieved a higher average denitrification rate of 2.5 g NO

-N

/(m²·d), compared with the MBBR receiving carbon produced by co-fermentation at pH 10 (1.8 g NO

-N

/(m²·d)). The lower efficiency of the MBBR supplied with additional carbon recovered by fermentation at pH 10 is caused by an accumulation of NO

-N during the denitrification process. This accumulation of NO

-N indicates suboptimal conditions, both due to the composition of the supplied carbon source and an overall higher pH during the denitrification process, which might supress facultative anaerobes, such as denitrifiers.

Nevertheless, this study shows that both VFA-rich carbon sources obtained by co-fermentation of pri- mary sludge and food waste are suitable to enhance denitrification of municipal wastewater, with the carbon source recovered by fermentation without pH control achieving a higher denitrification effi- ciency.

Keywords

VFAs, co-fermentation, primary sludge, food waste, denitrification, external carbon source, MBBR

Acknowledgements

This master thesis was conducted as part of the double degree programme in environmental engineering and sustainable infrastructure at the Royal Institute of Technology (KTH), Sweden, and the Technical University Darmstadt, Germany. The experimental studies were executed at the research facility Ham- marby Sjöstadsverk within the project Carbon Neutral Next Generation Wastewater Treatment Plants (CarbonNextGen). This project is a cooperation between the KTH Department of Sustainable Develop- ment, Environmental Science and Engineering, the KTH Department of Chemical Engineering and the Swedish Environmental Institute (IVL). Due to the scope of the project and the international coopera- tion of two universities, I would like to express my gratitude to my two supervisors professor Dr. Elzbieta Plaza and professor Dr. Susanne Lackner, who made this thesis project possible.

Furthermore, I would like to thank my co-supervisor Isaac Owusu-Agyeman, who introduced me to the project and helped with the coordination of the different experimental parts. My sincere gratitude goes out to PhD Andriy Malovanyy who helped me to set up the pilot scale denitrification MBBRs and shared his knowledge with me.

Moreover, I want to thank the amazing staff working at Hammarby Sjöstadsverk, especially Mayumi Narongin, Mila Harding, Jesper Karlson, Niclas Bornold and Andrea Carranza Muñoz who always took their time for me. Besides I would like to thank my amazing Swedish hostfamily Martin and Lisen Oliw and their beautiful children for their trust and moral support – without you my time in Sweden would not have been the same!

I want to express my sincere gratitude to my amazing family for the constant support and for raising me

to the person I am today! Also a big shout-out to all my friends for being part of my life! Lastly, I thank

my true companion Lars Laumeyer for always standing by my side - I am so lucky to have you in my life!

Table of contents

Table of contents ... IX Table of figures ... XI Table of tables ... XIII Acronyms and abbreviations ... XIV

1. Introduction ... 1

2. Aim and objectives ... 1

3. Theoretical background ... 2

3.1 Nitrogen in Wastewater ... 2

3.2 Denitrification ... 3

3.2.1 Process configurations of denitrification ... 3

3.2.2 Factors affecting the denitrification process ... 4

3.2.3 Efficiency of different carbon sources for denitrification ... 6

3.2.4 Volatile fatty acids as external carbon source... 7

3.2.5 Strategies to measure the denitrification activity ... 8

3.3 Anaerobic digestion ... 8

3.3.1 Hydrolysis ... 9

3.3.2 Acidogenesis ... 9

3.3.3 Acetogenesis ... 10

3.3.4 Methanogenesis ... 10

3.4 Production of volatile fatty acids (VFAs) ... 10

3.3.1 Factors influencing the production of VFAs ... 11

3.3.2 Nutrient release during the production of VFAs ... 14

3.4 Moving bed biofilm reactors ... 15

4. Materials and methods ... 16

4.1 Production of VFAs ... 16

4.1.1 Experimental setup ... 16

4.1.2 Substrates and inoculum ... 17

4.1.3 Experimental procedure ... 18

4.2 Utilisation of VFAs as external carbon source for denitrification ... 18

4.2.1 Experimental setup ... 18

4.2.2 Substrates ... 19

4.2.3 Experimental procedure ... 19

4.3 Analytical methods ...20

4.4 Calculations ... 22

4.4.1 Design and assessment of the denitrification pilot plant ... 22

4.4.2 Calculation of the amount of total, suspended and attached solids ... 23

4.4.3 Evaluation of the semi-continuous co-fermentation pilot reactors ... 24

5. Results and discussion ... 26

5.1 Production of VFAs ... 26

5.1.1 Influence of the pH on the total amount of VFAs produced ... 26

5.1.4 Production of biogas ... 36

5.2 Enhanced denitrification using waste-derived VFAs as external carbon source... 39

5.2.1 CSTR phase ... 39

5.2.2 Batch tests ... 45

5.2.3 Accumulation of nitrite-nitrogen during the denitrification process ... 47

5.2.4 Influence of VFAs on the denitrification rate ... 48

5.2.5 Impact of the external carbon sources on the ammonium-nitrogen and orthophosphate concentration ... 51

5.2.6 Evaluation of the developed biofilm ... 53

6. Conclusion ... 56

6.1 Findings from the co-fermentation pilot scale reactors ... 56

6.2 Findings from the enhanced denitrification pilot scale MBBRs... 56

7. Suggestions for further research... 58

References ... 59

Appendix ... 67

Table of figures

Figure 1: Flow schemes of different process configurations for denitrification: A: pre-anoxic

denitrification. B: post-anoxic denitrification. C: Low DO and cycling nitrification/denitrification

processes. (Adapted and modified from Tchobanoglous et al. (2014)) ... 4

Figure 2: Fractionation of the chemical oxygen demand (COD) in wastewater - VFAs are part of the

readily biodegradable (soluble) fraction (adapted and modified from Tchobanoglous et al. (2014)) ... 7

Figure 3: Flow scheme of the anaerobic digestion process. The blue box showing the favourable digestion

steps for the production of VFAs. (Adapted and modified from Tchobanoglous et al. (2014)) ... 9

Figure 4: Flow scheme of the semi-continuous pilot scale fermentation reactors ... 16

Figure 5: AnoxKaldnes K5 carrier before the beginning of the experiments ... 18

Figure 6: Flow scheme of the continuously stirred pilot scale denitrification MBBRs receiving VFAs

produced by co-fermentation as external carbon source... 19

Figure 7: Agilent Intuvo 9,000 Gas Chromatography System. Left: Operational setup; right: Open

system, showing the column ... 21

Figure 8: Variation of SCOD, TVFA and pH over time in the pilot scale co-fermentation reactor operated

at pH 10 (outliers marked transparent/blank)... 27

Figure 9: Comparison of the acidification yield, expressed as percentage of TVFA of the SCOD

concentration, achieved in the pilot scale co-fermentation reactor operated at pH 10 vs. without pH

control (outliers left blank) ... 27

Figure 10: Variation of SCOD, TVFA and pH over time in the pilot scale co-fermentation reactor operated

without pH control ... 28

Figure 11: Comparison of the achieved VFA yield in the pilot scale co-fermentation reactor operated at

pH 10 vs. without pH control. A: Relative VFA yield regarding VS, B: Relative VFA yield regarding VSS

(outliers marked transparent) ... 29

Figure 12: Composition of the VFAs in the co-fermentation liquid produced at pH 10 ... 32

Figure 13: Composition of the VFAs in the fermentation liquid produced without pH control ... 32

Figure 14: Comparison of the total nitrogen (TN) and ammonium-nitrogen (NH

-N) concentration in

the pilot scale co-fermentation reactors operated at pH 10 and without pH control ... 34

Figure 15: Comparison of the total phosphorus (TP) and orthophosphate (PO

-P) concentration in the

pilot scale co-fermentation reactors operated at pH 10 and without pH control ... 35

Figure 16: Measured biogas production during the co-fermentation of PS and FW in the pilot scale

reactor operated at pH 10 (outliers left blank) ... 37

Figure 17: A: Measured biogas production during the co-fermentation of PS and FW in the pilot scale

reactor operated without pH control, B: Percentage of CH

and CO

in the produced biogas ... 38

Figure 18: Nitrogen balance for the continuously stirred pilot scale MBBRs. A: MBBR A, receiving VFAs

produced by co-fermentation at pH 10, B: MBBR B, receiving VFAs produced by co-fermentation

without pH control ...40

Figure 19: Comparison of the achieved percental reduction of NO

-N

and SCOD as well as the amount

of SCOD per NO

-N

removed during the enhanced denitrification using diluted VFA-rich co-

fermentation liquid as external carbon source. Blank marks show process disruptions. A: MBBR A,

receiving VFAs produced by fermentation at pH 10, B: MBBR B, receiving VFAs produced by

fermentation without pH control, C: Comparison of SCOD per NO

-N

removed in MBBR A and MBBR

B ... 42

Figure 20: Comparison of the temperature and pH: A: Inflow from the buffer tank, B: Outflow MBBR A receiving VFAs produced at pH 10 as external carbon source, C: Outflow of MBBR B receiving VFAs produced without pH control ... 44 Figure 21: Comparison of the three conducted batch tests using undiluted fermentation liquid produced at pH 10 as external carbon source. A: Batch test 16.09.2019, B: Batch test 11.10.2019, C: Batch test 08.11.2019, D: Comparison of the NO

-N

concentration during the conducted batch tests ... 45 Figure 22: Comparison of the three conducted batch tests using undiluted fermentation liquid produced without pH control as external carbon source. A: Batch test 16.09.2019, B: Batch test 11.10.2019, C:

Batch test 08.11.2019, D: Comparison of the NO

-N

concentration during the conducted batch tests ... 46 Figure 23: Comparison of the achieved denitrification rate for the performed batch tests and the respective denitrification rate from the continuous operation ... 47 Figure 24: Comparison of the denitrification rate achieved by the pilot scale MBBRs receiving two different diluted VFA-rich co-fermentation liquids as external carbon source ... 49 Figure 25: Ammonium-nitrogen concentration in the pilot scale denitrification MBBRs’ outflow after dosing unrefined VFA-rich co-fermentation liquid produced at pH 10 and without pH control as external carbon source in relation to the inflowing concentration ... 52 Figure 26: Orthophosphate concentration in the pilot scale denitrification MBBRs’ outflow after dosing unrefined VFA-rich co-fermentation liquid produced at pH 10 and without pH control as external carbon source in relation to the inflowing concentration (outliers left blank) ... 53 Figure 27: Attached biofilm on K5 carriers after 8 weeks of the denitrification experiment: A: MBBR A, receiving VFA-mix produced by co-fermentation at pH 10, B: MBBR B, receiving VFA-mix produced by co-fermentation without pH control ... 54 Figure 28: Close-up images of the biofilm on AnoxKaldnes K5 carriers after 8 weeks of the denitrification experiment, made using a digital microscope Dino-Lite Premier with a magnification factor of 24.4. A:

MBBR A, receiving VFA-mix produced by co-fermentation at pH 10, B: MBBR B, receiving VFA-mix

produced by co-fermentation without pH control ... 55

Table of tables

Table 1: During nitrification and denitrification produced nitrogenous compounds (Gerardi, 2002) .... 3

Table 2: List of volatile fatty acids (VFAs) ... 11

Table 3: Chemical characteristics of the primary sludge (PS), food waste (FW) and feed stock (PS + FW)

used ... 17

Table 4: Feeding schedule of the VFA production reactors (adapted and modified from Bedaso (2019))

... 18

Table 5: Schedule of the parameter analysis during the conducted batch tests ...20

Table 6: Overview of the WTW Spectroquant® analysis kits used for the analysis of the wastewater

characteristics ... 21

Table 7: Conversion factors for the different VFAs to calculate the COD equivalent ... 24

Table 8: Overview of the VFA production using different substrates and operational conditions (adapted

and modified from Bedaso (2019)) ... 31

Table 9: Overview of the composition of the VFAs produced using different substrates and pH levels 33

Table 10: Comparison of the net release of NH

-N and PO

-P during the co-fermentation of PS and FW

at pH 10 and without pH control ... 36

Table 11: Comparison of the external carbon sources’ composition supplied to the two pilot scale

denitrification MBBRs. The shown values are calculated using the parameters’ concentration in the

initial fermentation broth and the applied dilution. ... 39

Table 12: Comparison of the total inorganic nitrogen removal rate in the two pilot scale MBBRs ... 43

Table 13: Overview of the denitrification rates achieved during the CSTR-phase in the two pilot scale

MBBRs using VFAs produced by co-fermentation at pH 10, respectively without pH control, as external

carbon source ... 48

Table 14: Overview of the achieved denitrification rates using different organic carbon sources at

different operating temperatures ... 50

Table 15: Comparison of the NH

-N and PO

-P concentration in the nitrified wastewater and the two

different external carbon sources used ... 51

Table 16: Assessment of the total attached solids (TAS) of the carriers of both pilot scale denitrification

MBBRs used after 8 weeks of operation ... 54

Acronyms and abbreviations

BNR Biological nutrient removal

C/N ratio Carbon-to-nitrogen ratio

CH

Methane

CO

Carbon dioxide

COD Chemical oxygen demand

CSTR Continuously stirred tank reactor

DO Dissolved oxygen

EBPR Enhanced biological phosphorus removal

Silox Decamethylcyclopentasiloxane (D

, C

H

O

Si

)

GC Gas chromatography

HRT Hydraulic retention time

HWWTP Henriksdal wastewater treatment plant

FW Food waste

MBBR Moving bed biofilm reactor

MPG Monopropylene glycol

NH

-N Ammonium-nitrogen

NO

-N Nitrite-nitrogen

NO

-N Nitrate-nitrogen

NUR Nitrate uptake rate

OLR Organic loading rate

Q Wastewater flow

PS Primary sludge

rpm Revolutions per minute

SCOD Soluble chemical oxygen demand

S

Nonbiodegradable chemical oxygen demand (soluble)

SRT Solids retention time

S

Readily biodegradable chemical oxygen demand (soluble)

PO

-P Orthophosphate as phosphorus

TAS Total attached solids

TCOD Total chemical oxygen demand

TIN Total inorganic nitrogen

TN Total nitrogen

TP Total phosphorus

TS Total solids

TSS Total suspended solids

TVFA Total amount of volatile fatty acids UASB reactor Upflow anaerobic sludge blanket reactor

VFAs Volatile fatty acids

VS Volatile solids

VSS Volatile suspended solids

WAS Waste activated sludge

WWTP Wastewater treatment plant

X

Slowly biodegradable chemical oxygen demand (particulate)

X

Nonbiodegradable chemical oxygen demand (particulate)

1. Introduction

During the last decades the awareness regarding environmental issues has increased on a multi-dis- ciplinary scope. Ambitious goals to reduce carbon emissions to slow down the anthropogenic climate change do not only affect industries but also the public sectors and therefore wastewater treatment facilities (Tchobanoglous, et al., 2014). Although there have been many proposals for new technolo- gies to reduce the carbon footprint of municipal wastewater treatment plants (WWTPs), the full scale implementation has not been successful yet.

Focusing on nutrient removal within WWTPs, the activated sludge process with denitrification and nitrification is the most energy intensive process amounting to 0.23 kWh/m³ wastewater treated.

This is largely because of the energy intensive aeration of the activated sludge, constituting over 55 % of a WWTP’s energy usage. (Tchobanoglous, et al., 2014)

Additionally, rising effluent standards require the supply of an external carbon source in the denitri- fication process to stay within the allowed thresholds. Commonly used external carbon sources to enhance the nitrate removal from the wastewater stream are crucial in terms of operational costs.

Hence the operational expenses can be reduced significantly by finding alternative ways to recycle carbon within the WWTP (Ucisik & Henze, 2008).

Volatile fatty acids (VFAs) are metabolic intermediates of the anaerobic digestion process, which is commonly used for resource recovery within WWTPs, producing biogas with ca. 60 – 70 % methane content, mostly used for heating and production of electricity (Kleerebezem, et al., 2015). However, VFAs are a readily biodegradable carbon source for denitrification. By co-digesting other waste prod- ucts, such as food wastes, the production of VFAs can be increased.

Furthermore, the regional carbon footprint is reduced by co-digestion as the release of carbon dioxide from an alternative waste decomposition is diminished (Tchobanoglous, et al., 2014). The produced VFAs can be used as readily biodegradable carbon source for denitrification, allowing a nutrient cy- cling within the WWTP. Hence, the operational expenses and the carbon footprint is reduced further, as the energy inputs to produce and transport alternatively used chemicals are no longer required (Tchobanoglous, et al., 2014).

2. Aim and objectives

The aim of the following research study is to analyse the possibility of closing the material loop for wastewater treatment plants focusing on carbon and therefore achieving a carbon neutral wastewater treatment. This objective shall be achieved by producing volatile fatty acids (VFAs) through co-fer- mentation of primary sludge (PS) and food waste (FW) and using them as a carbon source to enhance the denitrification process in a post-anoxic plant setup.

Therefore the following research objectives will be examined:

• The influence of the pH on the production of VFAs from co-digestion of PS and FW.

• Assessment and comparison of the effectiveness of the utilisation of VFAs produced at pH 10 and without pH control as external carbon source to enhance the denitrification of municipal wastewater.

• Investigation of the relative consumption of carbon source measured as soluble chemical ox- ygen demand (SCOD) to nitrate during the enhanced denitrification process.

• Measuring the denitrification rate of the biofilm with the supplied VFA-mix.

• Examining the possible increase of orthophosphates (PO

-P) and ammonium-nitrogen (NH

-

N) in the denitrification effluent due to high concentrations in the supplied VFA-mix.

3. Theoretical background

Wastewater is made up of different organic and inorganic components which can be measured to allow an assessment of the degree of pollution (Tchobanoglous, et al., 2014). The composition of the wastewater matrix depends on several factors, such as the regional collection of wastewater, the life- style of the people as well as climatic conditions (Valsami-Jones, 2004; Koppe & Stozek, 1999). There- fore the analysis of the wastewater components is needed to both choose the right treatment steps as well as to monitor and assess their efficiency. In a conventional wastewater treatment plant the wastewater is treated in three consecutive steps: physical, biological and chemical treatment. This study will focus on the biological treatment, as this is the relevant step for the effective removal of nitrogen which is an inorganic non-metallic component (Tchobanoglous, et al., 2014; EPA, 1993).

Accounting for ca. 78 % of all atmospheric gases, elemental nitrogen (N

) is the most abundant gas of the atmosphere. However, due to its very stable chemical constitution it cannot be utilised by most organisms in its elemental form. (Holmes, et al., 2019; Wright & Boorse, 2014) Similarly to phospho- rus, nitrogen is regarded as a limiting factor for plant growth and therefore the enrichment of nutri- ents from point sources, such as sewage treatment plants, needs to be minimised to prevent eutroph- ication (Wright & Boorse, 2014).

Although most of the nitrogen emitted to the Baltic Sea comes from agriculture, the effluent of WWTPs comes second. Therefore, the specific removal of nitrogen was introduced to Sweden’s WWTPs in 1985. According to Naturvårdsverket (2016), around 50 % of WWTPs had a specific setup for nitrogen removal in year 2016, removing on average 72 % of the total nitrogen (TN).

It is theoretically possible to remove more than the nowadays achieved 50 – 70 %, but this would sky rocket the energy consumption of WWTPs (Naturvårdsverket & Svenskt Vatten AB, 2013). Further- more, according to the Swedish environmental protection agency Naturvårdsverket and Svenskt Vat- ten (2013), the amount of external carbon needed to reach a threshold of 2 mg N/L would be equiva- lent to 20 – 30 % of the ethanol used for motor traffic in Sweden in 2013, considering all WWTPs bigger than 10,000 person equivalents.

3.1 Nitrogen in Wastewater

Wastewater contains nitrogen of several different oxidation states which can be changed by microor- ganisms. These changes can be either positive or negative, depending on the prevailing aerobic or anaerobic conditions. The most prevalent forms of nitrogen in wastewater are ammonia (NH

), am- monium (NH

), nitrite (NO

), nitrate (NO

) and nitrogen gas (N

). While the parameter total ni- trogen (TN) takes into account organic nitrogen, ammonia, nitrite and nitrate, the total inorganic nitrogen (TIN) only comprises ammonia, nitrite and nitrate. (Tchobanoglous, et al., 2014)

Microorganisms convert the organic nitrogen fraction in ammonium, hence untreated wastewater exhibits mainly ammonium-nitrogen with concentrations ranging around 12 – 45 mg NH

-N/L. The concentrations of nitrite- and nitrate-nitrogen in the influent are rather low, ranging from 0 mg/L to traces. (Tchobanoglous, et al., 2014) During the first step of the biological nutrient removal, the nitri- fication, ammonium-nitrogen is converted to nitrite-nitrogen and then to nitrate-nitrogen in an oxic environment, depicted in [1] (Gerardi, 2002).

NH NO NO

The actual nitrogen removal is achieved by denitrification under anoxic conditions, which is explained in detail in the following Chapter 3.2 Denitrification. In four consecutive reduction steps nitrate (NO

) or nitrite (NO

) are converted to nitric oxide (NO), nitrous oxide (NO

) and lastly dinitrogen (N

) as shown in Equation [2] with the four different enzymes required (Tchobanoglous, et al., 2014;

Ni, et al., 2016; Gerardi, 2002; Bitton, 2005). The resulting molecular nitrogen is an insoluble gas which escapes into the atmosphere, hence denitrification is classified as dissimilatory process (Gerardi, 2002).

NO NO NO N O N

Table 1 shows the nitrogenous compounds produced during the nitrification and denitrification of

wastewater (Gerardi, 2002).

The different nitrogen fractions in the effluent of a WWTP with biological nutrient removal amount to 0.7 – 3.0 mg NH

-N/L, 2 – 10 mg NO

-N/L and 0 mg/L or traces of NO

-N (Tchobanoglous, et al., 2014).

Table 1: During nitrification and denitrification produced nitrogenous compounds (Gerardi, 2002)

Nitrogenous compound Chemical formula Oxidation state of N

Nitrate ion NO3– + 5

Nitrite ion NO2– + 4

Nitric oxide NO +2

Nitroxyl NOH + 1

Molecular nitrogen N2 0

Hydroxylamine N2O – 1

Ammonia NH3 – 3

Ammonium ion NH4+ – 3

3.2 Denitrification

Denitrification is established subsequent to the nitrification to ultimately remove the previously formed nitrate from the wastewater (Ni, et al., 2016). Denitrification occurs under anoxic conditions, meaning that there is no molecular oxygen available (Gerardi, 2002; Lin, et al., 2009). During the process of denitrification soluble organic substrates are biologically oxidised using nitrate and/or ni- trite as the electron acceptor instead of oxygen, which is referred to as anoxic respiration (Gerardi, 2002).

To allow denitrification, the wastewater needs to contain soluble organic carbon that is readily avail- able as electron donor for the microorganisms. These substrates are available in wastewater to a cer- tain extent as biodegradable COD but can also derive from endogenous respiration or be supplied as external carbon source. (Bitton, 2005; Tchobanoglous, et al., 2014)

Although both chemolithoautotrophic and chemoorganoheterotrophic denitrifiers are known, deni- trification in WWTPs is almost entirely performed by heterotrophic bacteria. As most of these bacteria are facultative anaerobic, it is important to keep anoxic conditions or to implement an oxygen gradi- ent within the floc particles to allow the heterotrophic denitrification. (Ni, et al., 2016) Furthermore, one cannot necessarily assume that the denitrification pathway is completed if denitrification occurs, as significant changes in the operational conditions can disrupt the process. Additionally, some of the denitrifiers lack key enzymes to perform full denitrification. Despite denitrification being an integral part of the biological nitrogen removal, it can also contribute to operational problems if it occurs in the secondary clarifier or the sewer system. (Gerardi, 2002).

3.2.1 Process configurations of denitrification

Depending on the configuration of the biological nitrogen removal process it can be differentiated between three types (Tchobanoglous, et al., 2014):

1. Pre-anoxic denitrification processes 2. Post-anoxic denitrification processes

3. Low dissolved oxygen and cycling nitrification/denitrification processes

The difference of the pre-anoxic and post-anoxic denitrification process is depicted in Figure 1. More-

over, the differences of all three process configurations of denitrification shall be shortly presented.

Particular requirements regarding the operational conditions, especially in terms of aeration need to be met in all process configurations.

The pre-anoxic or substrate denitrification refers to a configuration of the biological nutrient removal (BNR) where the anoxic stage is prior to the aerated nitrification stage. Thereby the SCOD-rich wastewater inflow provides the required organic carbon source allowing the reduction of nitrate. The nitrate is internally recirculated from the aerobic nitrification tank.

The second process configuration is referred to as post-anoxic denitrification, positioning the anoxic tank subsequent to the aerobic nitrification tank. Hence, the SCOD has been already removed during the nitrification and is not available as substrate for the denitrifiers. This results in a significantly slower reduction of the nitrate, as the process depends solely on endogenous respiration. However, by adding an external carbon source the required SCOD can be supplied which increases the denitri- fication rate.

The third process configuration accomplishes both nitrification and denitrification in a single acti- vated sludge reactor, either by simultaneous nitrification-denitrification with a low amount of dis- solved oxygen (DO) or by cyclic nitrification-denitrification. In contrast to the previously presented process configurations there is no physical separation, which results in particular requirements re- garding the operational conditions, especially with regards to aeration. (Tchobanoglous, et al., 2014;

Ni, et al., 2016)

Figure 1: Flow schemes of different process configurations for denitrification: A: pre-anoxic denitrification. B: post- anoxic denitrification. C: Low DO and cycling nitrification/denitrification processes. (Adapted and modified from Tchobanoglous et al. (2014))

3.2.2 Factors affecting the denitrification process

Similar to nitrification there are several factors which affect the process of denitrification. The most critical factors are the presence of substrate in form of organic carbon and nitrate and the absence of free molecular oxygen. Furthermore, the origin of the carbon source, the carbon-to-nitrogen ratio (C/N ratio) and the active population of denitrifying bacteria influence the reduction reaction. Addi- tionally, parameters, such as pH, temperature and redox potential, influence the operational process.

(Gerardi, 2002; Ni, et al., 2016; Tchobanoglous, et al., 2014).

pH and alkalinity

The process of denitrification is relatively insensitive to variations of the pH and can even occur in acidic conditions (Lu, et al., 2014; Gerardi, 2002; Skiba, 2008; Lin, et al., 2009; Tchobanoglous, et al., 2014). Optimal pH conditions reported lie within the range of pH 7.0 – 9.0 (Tchobanoglous, et

A

B

al., 2014; Bitton, 2005; Gerardi, 2002; Lin, et al., 2009). However, according to Gerardi (2006), the activity of facultative anaerobes is depressed at pH values below 6.0 and above 8.0.

Accumulation of denitrification intermediates, such as NO

-N and NO, have been observed under suboptimal pH conditions (Lu, et al., 2014; Knowles, 1982). With typical pH conditions in WWTP between pH 7.0 – 8.0, it is not expected that the pH is a major concern regarding operational condi- tions for denitrification. (Ni, et al., 2016)

The process of denitrification is also important considering the return of approximately 50 % of the alkalinity lost during nitrification in form of hydroxyl ions (OH

) and carbon dioxide (CO

) (Gerardi, 2002; Tchobanoglous, et al., 2014). To be precise 3.57 g of alkalinity in form of CaCO

are produced per gram nitrate-nitrogen (NO

-N) reduced compared with 7.14 g CaCO

per g NH

-N oxidised during the nitrification process (Tchobanoglous, et al., 2014). Therefore, alkalinity and pH increase due to the denitrification process (Bitton, 2005; Tchobanoglous, et al., 2014).

Temperature

Denitrification can occur at a temperature range from 5 – 75°C (Lin, et al., 2009). The influence is more significant at lower temperatures (6 – 10°C) compared to medium temperatures (10 – 25°C) (Carrera, et al., 2004) Optimum conditions for nitrate removal via denitrification have been reported from various WWTPs within the range of 20 - 35°C (Dawson & Murphy, 1972; Knowles, 1982;

Tchobanoglous, et al., 2014).

Warmer temperatures do not only promote the activity and growth of the denitrifiers but also have a lower affinity for DO than colder wastewater. The denitrification rate decreases with decreasing tem- peratures and an inhibition of the process occurs at temperatures below 5°C. (Gerardi, 2002) Accord- ing to Lu et al. (2014), most denitrifiers are more sensitive to temperature variations than pH fluctu- ations.

Carbon-to-nitrogen ratio

The consumptive ratio or C/N ratio describes the amount of carbon in form of COD that is required as electron donor to reduce nitrate to organic nitrogen during the denitrification process. Therefore, the C/N ratio can be seen as the denitrification capacity of the wastewater or a supplied carbon source.

(Tchobanoglous, et al., 2014)

According to Barth et al. (1968), approximately 4 g biodegradable COD is required as electron donor to reduce one gram of nitrate, resulting in a C/N ratio of 4. Due to the limited amount of readily available organic carbon in wastewater, the complete removal of wastewater containing high nitrogen concentrations requires large amounts of an additional carbon source (Ni, et al., 2016;

Tchobanoglous, et al., 2014). In general, effluent requirements of less than 6 – 8 mg TN/L require an external carbon source (Tchobanoglous, et al., 2014). Several different chemicals are being used as external carbon sources, which will be discussed in depth in Chapter 3.2.3 Efficiency of different car- bon sources for denitrification.

Furthermore, the C/N ratio affects the denitrification efficiency, as it has an influence on the utilisa- tion pathway of the substrate. According to Ruiz et al. (2006), denitrification is the main substrate utilisation route for C/N ratios < 10, whereas the methanogenic pathway is favoured for high C/N ratios of 100. Moreover, C/N ratios of 1 resulted in a nitrite accumulation due to insufficient organic matter availability. At a C/N ratio of 5 nearly 100 % of nitrate could be removed. (Ruiz, et al., 2006) This has also been reported by Akunna et al. (1992) who found that methanogenesis dominates the substrate utilisation at C/N ratios > 53, while denitrification was dominant for C/N ratios < 8.86. C/N ratios of 8.86 – 53 showed the utilisation of the organic carbon for both denitrification and methan- ogenesis. However, due to the inhibitory effect of nitrate on the methane production, methanogenesis does not occur until denitrification is finished (Akunna, et al., 1992; Ruiz, et al., 2006).

Dissolved oxygen concentration

With denitrifiers being usually heterotrophic anaerobic bacteria, which are facultative anaerobes, aer-

obic respiration is preferred due to the higher energy yield: 686 kcal/molecule glucose from aerobic

respiration compared to anoxic respiration yielding 636 kcal/molecule glucose (Gerardi, 2002), re- spectively 570 kcal/molecule glucose (Bitton, 2005). Hence, denitrifiers only switch to anoxic respi- ration and thereby denitrification if the concentration of DO is very low (Zumft, 1997; Gerardi, 2002).

An inhibition of the denitrification can occur due to the repression of the nitrate reductase at DO concentrations above 0.2 mg/L (Tchobanoglous, et al., 2014). According to Gerardi (2002), the re- duction of nitrate and nitrite is inhibited if the oxygen concentration exceeds 1 mg/L (Gerardi, 2002) However, several studies also report rapid denitrification with DO concentrations exceeding 1 mg/L which can be explained by localised oxygen depletion within the activated sludge structure (Dawson

& Murphy, 1972; Knowles, 1982; Bitton, 2005). A dissolved oxygen gradient can be found in floc par- ticles of a size > 100 µm (Gerardi, 2002).

3.2.3 Efficiency of different carbon sources for denitrification

As discussed in Chapter 3.2.2 Factors affecting the denitrification process denitrifying bacteria re- quire a carbon source as electron donor during the reduction of NO

-N. One important factor regard- ing the sufficient supply of an organic carbon source has been introduced in form of the C/N ratio.

However, the carbon source in wastewater treatment systems can be usually classified into readily biodegradable substrate and slowly degradable substrate. To measure these different fractions the COD is usually applied which can be fractionated as depicted in Figure 2. Typically, the COD of the wastewater influent lies within the range of 300 – 1.000 mg O

/L (Koppe & Stozek, 1999;

Tchobanoglous, et al., 2014). The readily biodegradable substrate for denitrification is represented by the soluble fraction of the biodegradable COD (S

), whereas the particulate COD (X

) represents the slowly biodegradable substrate. Further, the readily biodegradable fraction of the COD can be subdi- vided into complex S

and VFAs. Additionally, the total COD (TCOD) comprises an inert fraction, which is referred to as nonbiodegradable COD, which can further be subdivided into a soluble (S

) and particulate (X

) fraction. (Tchobanoglous, et al., 2014)

Readily biodegradable substrate (S

)

The readily biodegradable substrate is presumably composed of simple, respectively low molecular soluble compounds, such as VFAs, characterised by fast uptake kinetics (Coen, et al., 1998). Hetero- trophic bacteria assimilate and mineralise the S

fraction of the COD, which makes it an important parameter to determine the success of wastewater treatment processes, such as denitrification (Myszograj, et al., 2017). This being said, the amount of S

has an immediate impact on the denitrifi- cation activity due to the fast availability of the substrate for denitrifiers.

Slowly biodegradable substrate (X

)

In comparison to S

, the slowly biodegradable substrate comprises particulate organic matter, which needs to by hydrolysed prior biodegradation (Coen, et al., 1998). Similar to the S

, the X

has a great influence on the dynamics regarding the activated sludge- and BNR processes (Myszograj, et al., 2017). The hydrolysis of the high-molecular compounds makes the substrate available for the micro- organisms and thereby enabling the substrate utilisation for denitrifiers.

Nonbiodegradable soluble COD (Si)

In contrast to the previously presented biodegradable fractions of the COD, the S

consists of inert

organic compounds, which are not taking part in the wastewater treatment process. This fraction can-

not be utilised by microorganisms and therefore is discharged with the effluent. (Myszograj, et al.,

2017)

Figure 2: Fractionation of the chemical oxygen demand (COD) in wastewater - VFAs are part of the readily biodegrada- ble (soluble) fraction (adapted and modified from Tchobanoglous et al. (2014))

The C/N ratio depends on the organic carbon source used. This can be explained by the biomass syn- thesis yield as well as the biodegradability, which also differ depending on the carbon source. An ex- ample is the 70 % lower C/N ratio of methanol compared to glucose, as a greater fraction of carbon source can be used for nitrate reduction when using methanol due to the lower synthesis yield.

(Tchobanoglous, et al., 2014)

The effectiveness of an external carbon source can be calculated as the inverse of the consumptive ratio, as depicted in Equation [3] (Tchobanoglous, et al., 2014).

1

E = C , = 2.86 1 − 1.42 Y

[3]

ECNO3 CR,NO3

YH

≙ effectiveness ratio

≙ consumptive ratio

≙ synthesis yield

[g COD/g NO3-N]

[g NO3-N /g COD]

[g VSS/g COD removed]

3.2.4 Volatile fatty acids as external carbon source

With VFAs being a readily biodegradable fraction of the SCOD, they are an effective source of external carbon for BNR. Therefore, it is possible to utilise the by fermentation produced VFAs as external carbon source for BNR processes, such as biological phosphorus and nitrogen removal. Focusing on the biological nitrogen removal, several studies compared the nutrient removal efficiency of waste- derived VFAs to commercially used products, such as methanol. It has been reported by several stud- ies that waste-derived VFAs achieved a better removal efficiency when compared with methanol (Elefsiniotis, et al., 2004; Liu, et al., 2016; Kim, et al., 2016). VFAs can be directly utilised by denitri- fiers, which causes a high removal efficiency compared to methanol, which needs to be oxidised to the correspondent VFA first (Elefsiniotis, et al., 2004). Further, a comparison of fermentation liquid pro- duced by FW fermentation with sodium acetate and glucose conducted by Zhang et al. (2016), showed that the fermentation liquid achieved a similar denitrification to sodium acetate, but better results of total nitrogen removal when compared to glucose.

In a study conducted by Elefsiniotis et al. (2004) fermentation liquid was utilised as external carbon source for denitrification and the consumption pattern of the different VFAs was analysed. According to their findings, using a C/N ratio > 2, acetic acid is the preferred VFA and consumed to 88 %, fol- lowed by n-butyric acid (73 %), propionic acid (69 %) and valeric acid (52 %) (Elefsiniotis, et al., 2004). Although this utilisation pattern was observed, it is not definitely known why certain VFAs are preferred over others. Two possible explanations were given, first, certain VFAs are easier to degrade on a biochemical level than others, which explains a lower consumption of VFAs needing a more com- plex metabolic pathway. The second explanation focuses on the availability of the utilised VFAs in the fermentation liquid, assuming that the VFAs of higher concentrations are consumed first.

(Elefsiniotis, et al., 2004) A subsequent study conducted by Elefsiniotis and Wareham (2007) con-

cluded that denitrifiers prefer acetic acid, followed by butyric acid and propionic acid, while valeric

acid is consumed only if a limitation of the other VFAs occurs.

3.2.5 Strategies to measure the denitrification activity

It is possible to assess the activity of a specific group of bacteria via batch activity tests which are performed under controlled conditions. This being said batch activity tests can be performed under aerobic, anoxic or anaerobic conditions, allowing the assessment of the kinetic rates for BNR pro- cesses, such as nitrification, denitrification and anammox. Batch activity tests for the assessment of process kinetics can be conducted using different tracking techniques, i.e. titrimetric, manometric and chemical tracking as well as respirometry. Titrimetric tracking applies pH-static titration to pro- cesses which relevantly affect the pH. Manometric tracking measures the pressure increment caused by the production of soluble gases of low solubility, such as N

, during a biological process. Respirom- etry can be used to assess aerobic processes that affect the DO concentration. Lastly, the chemical tracking method assesses the concentration of a certain substrate over time, such as NH

-N, NO

-N or NO

-N, depending on the assessed biological process. (Van Loosdrecht, et al., 2016)

Traditionally, the activity of denitrifying bacteria is assessed using the chemical tracking method with the nitrate uptake rate (NUR) test (Van Loosdrecht, et al., 2016). The NUR test allows the assessment of several parameters of interest simultaneously, such as nitrate utilisation rates, the consumption rate of organic matter as well as the yield coefficient of the anoxic biomass (Naidoo, et al., 1998;

Kujawa & Klapwijk, 1999). The assessment is conducted by taking samples throughout the batch test which are then analysed regarding the main substrates of interest, i.e. NO

-N, NO

-N and SCOD (Van Loosdrecht, et al., 2016).

Denitrification follows a zero order kinetic as long as there is enough substrate in form of COD and NO

-N. Hence, it is possible to calculate the specific denitrification rate from the linear parts of the process kinetic, using the concentration of volatile suspended solids (VSS) as reference for the bio- mass concentration (Naidoo, et al., 1998). In consequence of the design of the conducted studies as MBBR, an assessment of the biomass in form of the concentration of VSS was not easily possible, which is a known issue of biofilm reactors. As the performance of biofilm reactors depends primarily on the biofilm growth surface area, it is common to design and assess biofilm reactors based on a specific area removal rate expressed as gram of substrate removed per m² biofilm area per day (g/(m²·d)). (Ødegaard, et al., 2000)

3.3 Anaerobic digestion

Anaerobic digestion is the most frequently used stabilisation process used to treat primary and exces- sive sludge in WWTPs (Von Sperling, 2007). In this process microorganisms degrade organic matter in absence of dissolved oxygen and reduce the volatile solids by 35 – 60 %, depending on the opera- tional conditions (Bitton, 2005). Thereby the insoluble organic matter is converted into biogas which mainly consists of methane (CH

) and carbon dioxide (CO

) (Von Sperling, 2007; Bitton, 2005). Ac- cording to Tchobanoglous et al. (2014), the biogas of a stable anaerobic digestion process typically contains ca. 65 % CH

and ca. 35 % CO

. However, the full anaerobic digestion of organic matter can be described by equation [4] (Bitton, 2005).

Organic matter → CH + CO + H + NH + H S

The process is made up of four different consecutive stages which are hydrolysis, acidogenesis, aceto-

genesis and methanogenesis, displayed schematically in Figure 3 and explained in more detail in the

following paragraph. As this study focuses on the production of VFAs, the first three steps of the an-

aerobic digestion are favourable while the methanogenesis shall be avoided, to keep as many VFAs in

the digested residue as possible.

Figure 3: Flow scheme of the anaerobic digestion process. The blue box showing the favourable digestion steps for the production of VFAs. (Adapted and modified from Tchobanoglous et al. (2014))

3.3.1 Hydrolysis

The term hydrolysis refers to the addition of water to complex molecules by bacteria and therefore splitting unique chemical bonds, making them available for assimilation by microorganisms (Gerardi, 2006). Complex polymeric substrates can generally not be utilised by methanogenic and acetogenic bacteria directly, which makes a breakdown of these materials necessary. Consequently, the first step of the anaerobic digestion is the conversion of large and complex organic molecules, i.e. carbohy- drates, proteins and lipids, to soluble monomers by hydrolytic bacteria. (Gerardi, 2006; Mara &

Horan, 2003) Due to the large size of the substrate, the bacterial wall of the hydrolytic bacteria cannot be passed. However, extracellular hydrolytic enzymes are able to initiate the attack on these complex substrates. (Mara & Horan, 2003) These released enzymes include cellulose to convert carbohydrates to monosaccharides, proteinase to hydrolyse proteins to amino acids and lipase to transform lipids to long chain fatty acids (LCFA), among others (Henze, et al., 2011; Mara & Horan, 2003).

Both facultative and obligate anaerobic hydrolytic bacteria are present in the anaerobic digester with a quantity of 10

– 10

per millilitre (Mara & Horan, 2003). The process depends on different param- eters, such as temperature, pH, particle size, enzyme production, chemical structure of the substrate, diffusion and adsorption of the enzymes on the particular matter (Venkiteshwaran, et al., 2015). Due to the low biodegradability of the sludge from WWTPs, the hydrolysis is considered the rate-limiting step of the anaerobic digestion process (Carballa, et al., 2007; Henze, et al., 2011).

3.3.2 Acidogenesis

The second step of the anaerobic digestion comprises the acidogenesis, which is also referred to as

fermentation. During this process acidogenic bacteria utilise the substrate both as electron donors

and acceptors. (Tchobanoglous, et al., 2014; Venkiteshwaran, et al., 2015) Therefore, sugars, amino

acids and fatty acids are converted to organic acids (e.g. acetic, formic, lactic, butyric, succinic acids),

alcohols and ketones (e.g. ethanol, methanol, glycerol, acetone), acetate, CO

, hydrogen (H

) and wa-

ter (Gerardi, 2006; Bitton, 2005). As there are many different genera and species of acidogenic bac-

teria, the products vary with the type of microorganism as well as with the conditions, e.g. pH, redox

potential and temperature (Bitton, 2005; Mara & Horan, 2003). According to Venkiteshwaran et al.

(2015), the main product of the acidogenesis are VFAs, including acetic acid, propionic acid, butyric acid, iso-butyric acid, valeric acid and iso-valeric acid.

As acidogenesis is a rapid process, VFAs can accumulate in the reactor and inhibit the methanogenesis due to a drop in pH if the acid utilisation itself is inhibited or too slow (Venkiteshwaran, et al., 2015).

However, the facultative anaerobic acidogenic bacteria protect the sensitive methanogens by consum- ing traces of oxygen. With around 10

– 10

acidogenic bacteria per millilitre in the anaerobic digester, their cell count is lower compared to the hydrolytic bacteria. (Mara & Horan, 2003)

3.3.3 Acetogenesis

Acetogenic bacteria also belong to the fermentative bacteria, converting organic acids, alcohols and ketones to acetate, CO

and H

, of which acetate can be directly utilised by the methanogens (Gerardi, 2006; Tchobanoglous, et al., 2014).

The acetogenesis can be inhibited if the hydrogen pressure within the anaerobic digester is too high (Gerardi, 2006). If the partial hydrogen pressure exceeds a certain threshold, the formation of acetate is reduced and propionic acid, butyric acid as well as ethanol are produced instead (Bitton, 2005). As a consequence, acetogens rely on the hydrogen consumption by methanogens to keep the hydrogen pressure low enough (H

< 10

atm) (Tchobanoglous, et al., 2014; Mara & Horan, 2003). This mutu- alistic interaction between the two species is termed syntrophy (Mara & Horan, 2003).

3.3.4 Methanogenesis

The final step of the anaerobic digestion is performed by archaea known as methanogens, which can be classified in two groups. While the acetoclastic methanogens derive methane from breaking down acetate into CH

and CO

(Equation [5]), the hydrogenotrophic methanogens convert hydrogen (elec- tron donor) and CO

(electron acceptor) to CH

(Equation [6]). (Tchobanoglous, et al., 2014) Accord- ing to Tchobanoglous et al. (2014), methane is majorly produced by the acetoclastic methanogens with about 72 % of the total CH

produced.

Acetoclastic methanogens CH COOH → CH + CO

Hydrogenotrophic methanogens 4 H + CO → CH + 2 H O

There are two different temperature ranges for the methane-forming bacteria, differentiating between mesophilic (30 – 35°C) and thermophilic organisms (50 – 60°C). However, at a temperature between 40 – 50°C most of the methanogens are inhibited. Regarding the pH, an optimal range of 6.8 – 7.2 has been observed, with methanogens being sensitive to values below or above this range. (Gerardi, 2006) Therefore, a steady state operation of the digester is important, keeping the VFAs concentra- tion below 200 g/m³ and the pH ≥ 7 (Tchobanoglous, et al., 2014).

3.4 Production of volatile fatty acids (VFAs)

Volatile fatty acids (VFAs) are water-soluble short-chain fatty acids consisting of up to six carbon atoms, as presented in Table 2 (Fang, et al., 2019). They can be distilled at atmospheric pressure and the individual concentrations of many of the different VFAs can be determined via gas chromatog- raphy (GC), which is therefore a suitable control test for the anaerobic digestion (Baird, et al., 2017b).

As described in Chapter 3.3 Anaerobic digestion are VFAs important intermediates in the anaerobic digestion, but an accumulation exceeding 6.7 – 9 mol/m³ inhibits the methanogenesis (Appels, et al., 2008).

VFAs have a broad application in the chemical-, food- and pharmaceutical industry and are also used to produce biodegradable plastics (polyhydroxyalkanoates), biodiesel, biogas and to enhance BNR in wastewater streams (Atasoy, et al., 2018; Zacharof & Lovitt, 2013; Lee, et al., 2014). Nowadays VFAs are mostly produced by non-renewable petrochemical processes with severe environmental impacts.

Thus the production of VFAs using renewable resources has a promising future. (Atasoy, et al., 2018;

Lee, et al., 2014) Additionally, by decreasing the finally disposed sludge mass, the production of VFAs by anaerobic digestion at WWTPs can have a positive economic effect (Tchobanoglous, et al., 2014;

Von Sperling, 2007).