The Anaerobic Digestion of Organic Solid Wastes of Variable ...

The Anaerobic Digestion of Organic Solid Wastes of Variable Composition

by

Nigel G. H. Guilford

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Graduate Department of Chemical Engineering and Applied Chemistry University of Toronto

© Copyright by Nigel G. H. Guilford 2017

ii

The Anaerobic Digestion of Organic Solid Wastes of Variable Composition

Nigel G. H. Guilford

Doctor of Philosophy

Graduate Department of Chemical Engineering and Applied Chemistry University of Toronto

2017

Abstract Every year in Canada approximately 8 million tonnes of organic solid waste is placed in landfills where it

decomposes anaerobically over decades, produces large volumes of leachate requiring treatment, and

releases 20 million tonnes of greenhouse gas emissions as CO2eq. Anaerobically digesting this waste

prior to landfill would obviously be beneficial, but this is difficult to achieve because solid waste is a

complex, heterogeneous and variable mixture, making any form of processing much more expensive

than landfill. This thesis investigates the capabilities of a new approach to the anaerobic digestion of

solid waste designed to overcome these obstacles. Most of the costly separation and pretreatment

steps common in European anaerobic digesters are eliminated. The waste remains stationary, the

leachate is recirculated through it, and the resulting digestate is aerobically cured. The biogas generated

is recovered for the generation of electricity or the production of renewable natural gas.

iii

A laboratory scale system comprising six sequentially batch fed leach beds and an upflow anaerobic

sludge blanket reactor was constructed, and operated continuously for 616 days. The feedstock

consisted of a mixture of cardboard, boxboard, newsprint, and fine paper, to which varying amounts of

food waste were added (from 0% to 29% on a COD basis). The digester accommodated these and other

changes without any signs of process upset or instability. It was found that the addition of food waste

increased biogas production from the fibre mixture from 101 L.kg-1CODfibreadded to 330 L. kg-1CODfibreadded

an increase of 225%. A substrate destruction efficiency of 65% (on a COD basis) and a methane yield of

225 L.kg-1 CODadded were achieved, at a solids retention time of 42 days. This performance was similar to

that of a CSTR digesting similar wastes. A financial analysis showed that the technology can be

competitive with landfill.

iv

Acknowledgements I have looked forward to this moment for a long time. To be able to reflect on the last 5 years, and to acknowledge all of those who have been a part of this journey, and who have helped me in so many diverse, and often unexpected, ways is a great privilege. I must begin of course with Prof. Elizabeth Edwards. She accepted me as a PhD student at a time in my life when most people are looking for simpler pleasures. To this day, I don’t really know why she agreed; perhaps it was my solemn undertaking to be a “low-maintenance student”, but whatever the reason I am eternally grateful. Elizabeth is one of those rare people who knows exactly when encouragement is needed, and never fails to provide it. On top of that our working relationship has always been easy and relaxed and a lot of fun.

To Professors Allen and Saville, Grant and Brad, my sincere thanks for your patience, wisdom and good humour. I always left my Committee Meetings with my head filled with ideas, and with renewed energy to pursue my work. And of course to Professors Tim Bender and Morton Barlaz I extend my thanks for reading this thesis, and participating in my defence and, in Professor Barlaz’s case, for coming such a long way to do so; thank you both. Three other professors made a material difference to my time here; Doug Reeve by making it OK to be an “old guy” with something to learn, Vlad Papangelakis for setting me straight at the very beginning about the challenges I would face, and Sasha Yakunin for his interest and encouragement throughout. A special thanks to my old friend, Professor Howard Goodfellow, who dropped by regularly to check up on me.

There is a handful of people without whose help my research would never have begun, let alone been completed, and to them I owe a particular debt of gratitude. Firstly, Graeme Norval for making available to me the help of an electronics wizard Glenn Wilson. I don’t know how Graeme knew I needed help, but I surely did and Glenn provided it. He was thorough, organized and methodical and assisted me in so many ways to build an experimental system that actually worked. And then of course there is Paul Jowlabar. I’ve never met anyone quite like Paul; he provided tools, materials, advice, free labour, food and above all his amazing good humour and patience. Paul’s teaching must surely be one of the most durable memories that every new Chem. Eng. grad carries with them when they leave. It’s a commonplace to refer to good people as irreplaceable when they are not; but in Paul’s case it’s true.

Thanks to Torsten Meyer and Abdul-Sattar Nizami who at different times, and in different ways, provided invaluable assistance. To Line Lomheim and Susie Susilawati, my eternal thanks for all your unstinting help in the lab and your forbearance when things “went wrong”. My special thanks to Susie for her first aid skills when a hole-saw ate a bit of my thumb late one Friday afternoon, and I passed out in her office. Pauline Martíni and Joan Chen made sure that I was able to navigate the mysteries of the grad office, and complete my degree requirements; thank you both so very much. To Sofia Bonilla, who brought sunshine into the lab every day (together with a little order and discipline); you were the best of lab-mates. Greg Brown and Andy Quaile shed light into corners of my research that would otherwise have remained impenetrable; thank you both.

It is now the turn of my six collaborators; those who worked with me, side-by-side in the lab, doing experimental work. In more or less chronological order. Corinne Bertoia learned how to use shop tools when we were building Daisy and became an expert in leak-testing anaerobic digesters. Kärt Kanger

v

came twice from Estonia to work in the Edwards lab and provided help and good humour during the testing times of Daisy’s startup, and who also set a standard of neatness in her lab book that will never be surpassed. Colin Harnadek performed analyses of cardboard and paper and food waste like a machine, and became a star at UNERD; Scott Mitchell spent the summer of 2016 doing more COD analyses than any normal human would complete in a lifetime, and brought a degree of sophistication of thought which was of great help to me, and will carry him far. A special thanks to Peter HyunWoo Lee who, as a brand-new MASc. student was thrown in at the deep-end, troubleshooting Daisy. As Peter’s experience grew so did his contribution to our collective research. His computer skills went a long way to compensate for my lack of them, and for that alone I would be extremely grateful, but Peter was responsible for making a substantial contribution to the vast amount of data that we generated. And finally there was Savia Gavazza, a visiting professor from Brazil, who decided that it would be a great idea to set up and run a large-scale, complicated experiment, to help us answer some of the more subtle questions about what goes on in Daisy – and it was a good idea. Thanks to you all, it was my great good fortune to work with all of you.

Next, my friends at the Miller Group; Mike Kopansky, John Tomory, Charlie Cassin and Kyle Schumacher; John and Charlie always made sure that I had a supply of waste materials to feed to a hungry Daisy. Not only did they prepare it they delivered it, always managing to fit my needs into their already full days. Kyle Schumacher designed, built, and ran a small-scale composting experiment, also while performing his normal duties at Miller. And my friend, Mike Kopansky, who always ensured that what was needed got done, and spent many hours thinking and talking about how to make my research commercially relevant. Sincere thanks to you all.

There is one acknowledgement and expression of deep gratitude, that I must make with a great deal of sadness; I knew the late Leo McArthur, CEO and owner of the Miller Group, for more than 20 years. He was a remarkable man, great to work for, and generous to a fault. When I told him that I wanted to leave Miller to pursue my PhD, he didn’t pretend to be pleased but he did agree to provide financial support for my research. For this act of generosity alone he has my eternal gratitude. Also, to my friend Blair MacArthur, Leo’s son and successor, thank you for your many kindnesses and continued friendship and support.

To all my friends and colleagues in Edlab and BioZone; thank you for your companionship, your good humour, conversation and all the many instances of help that you afforded me. This is a long list, but I think it reflects the extraordinary ethos of the Department - collaboration and mutual support, no matter who you are what you do; it is unlike anywhere I have ever worked.

Finally, and most importantly, I come to Irene, my life’s companion and wife of 40 years. Irene literally made this all possible by giving me unstinting encouragement from the very beginning, assuring me that I "could do it” even when I thought the evidence to the contrary was overwhelming. She steadied me when things got difficult, and we celebrated when things went well. For the last six months she has made sure that I had all the time that I needed to write this thesis. Thank you with all my heart - now it’s your turn.

vi

For Irene

and in memory of my wonderful parents

Dorothy and Gareth Guilford

vii

Table of Contents Abstract ......................................................................................................................................................... ii

Acknowledgements ...................................................................................................................................... iv

List of Plates ................................................................................................................................................. xii

List of Figures .............................................................................................................................................. xiii

List of Appendices ........................................................................................................................................ xv

Chapter 1. Introduction ................................................................................................................................. 1

1.1 An Enduring Obsession ........................................................................................................................ 1

1.2 Is there a practical, affordable, solution? ............................................................................................ 2

1.3 The Rationale for Anaerobic Digestion ................................................................................................ 5

1.3.1 Organic Solid Waste ..................................................................................................................... 5

1.3.2 The Case for Anaerobic Digestion ................................................................................................. 9

1.3.3 The Rationale for a New Technology .......................................................................................... 10

Chapter 2. Objectives of the Research Programme .................................................................................... 11

2.1 Starting Principles .............................................................................................................................. 11

2.2 Specific Objectives ............................................................................................................................. 12

2.3 Hypothesis ......................................................................................................................................... 12

Chapter 3. Thesis Outline ............................................................................................................................ 14

Chapter 4. Literature Review ....................................................................................................................... 15

4.1 What Kind of Anaerobic Digester? .................................................................................................... 16

4.2 Substrates, Bulking Agent and Digestate ........................................................................................... 18

4.2.1 Food Waste ................................................................................................................................ 19

4.2.2 Lignocellulosic Waste ................................................................................................................. 23

4.2.3 Co-digestion ............................................................................................................................... 26

4.3 Feedstock Preparation ...................................................................................................................... 27

4.3.1 Size Reduction in SS-AD .............................................................................................................. 27

4.3.2 Solids Content ............................................................................................................................ 28

4.3.3 Bulking Agent.............................................................................................................................. 29

4.3.4 Feedstock Pretreatment ............................................................................................................. 30

4.4 Digester Performance........................................................................................................................ 31

4.5 Aerobic Curing ................................................................................................................................... 34

4.6 Conclusions and Knowledge Gaps ..................................................................................................... 35

Chapter 5. Digester Design and Construction ............................................................................................. 38

viii

5.1 Design Basis ....................................................................................................................................... 38

5.1.1 Leach Beds .................................................................................................................................. 39

5.1.2 Upflow Anaerobic Sludge blanket (UASB) Reactor ..................................................................... 43

5.1.3 Leachate Tanks ........................................................................................................................... 45

5.1.4 Biogas Production ....................................................................................................................... 46

5.1.5 Aerobic Curing ............................................................................................................................ 47

5.2 Design and Construction ................................................................................................................... 47

5.2.1 Guiding Principles ....................................................................................................................... 48

5.2.2 Safety .......................................................................................................................................... 48

5.2.3 General Description .................................................................................................................... 49

5.2.4 Leach Beds .................................................................................................................................. 51

5.2.5 Upflow Anaerobic Sludge Blanket Reactor ................................................................................. 55

5.2.6 Leachate Tanks ........................................................................................................................... 57

5.2.7 Pumps and Valves ....................................................................................................................... 60

5.2.8 Biogas Measurement .................................................................................................................. 62

5.2.9 System Monitoring and Control ................................................................................................. 62

5.2.10 Sampling Locations ................................................................................................................... 66

5.2.11 Support Structure and Miscellaneous Equipment .................................................................... 66

Chapter 6. Analytical Methods .................................................................................................................... 68

6.1 Total Solids, Volatile Solids and Chemical Oxygen Demand .............................................................. 68

6.2 Elemental Analysis ............................................................................................................................. 72

6.3 Volatile Fatty Acids and Sulphate ...................................................................................................... 74

6.4 Alkalinity and pH ................................................................................................................................ 74

6.5 Biogas Analysis .................................................................................................................................. 75

Chapter 7. Commissioning and Startup ....................................................................................................... 76

7.1 Commissioning .................................................................................................................................. 76

7.2 Startup of Daisy with Synthetic Feed - Tanks and UASB operating .................................................... 79

7.3 Startup of Daisy - Tanks and Leach Beds operating ........................................................................... 82

7.4 Startup of Daisy – Entire System operating ....................................................................................... 83

7.5 Commissioning and Start-up Conclusions ......................................................................................... 98

Chapter 8. Operation of Daisy - Results ..................................................................................................... 100

8.1 Introduction..................................................................................................................................... 100

8.2 Overall Performance – Robustness and Stability ............................................................................. 103

ix

8.2.1 Substrate Destruction and Biogas Production .......................................................................... 103

8.2.2 Alkalinity Ratio and pH ............................................................................................................. 105

8.2.3 Wastewater and Inorganic Salts (including sulphate) ............................................................... 106

8.2.4 Chemical Oxygen Demand in Leachate .................................................................................... 108

8.2.5 Volatile Fatty Acid Concentration ............................................................................................. 108

8.3 Operation by Period - Synergy and the Bulking Agent Effect .......................................................... 111

8.3.1 Periods 2 and 3 - Seven Week SRT ........................................................................................... 114

8.3.2 Period 4 - Reduction in Food Waste Addition – Discovery of Synergy ...................................... 116

8.3.3 Period 5 – Raising Food Waste Addition; the Bulking Agent Effect .......................................... 118

8.3.4 Period 6 - High Levels of Food Waste Addition......................................................................... 121

8.3.5 Synergy ..................................................................................................................................... 123

8.4 Influence of Reactor Design ............................................................................................................ 127

Chapter 9. Supporting Experiments .......................................................................................................... 131

9.1 BMP experiments ............................................................................................................................ 131

9.1.1 Biochemical Methane Potential Test 2 – Substrate Digestibility and Synergy .......................... 131

9.1.2 Biochemical Methane Potential Test 3 – One Litre Bottles and Coarse Substrate ................... 143

9.2 Enzyme Activity Assays .................................................................................................................... 148

9.3 Composting of Digestate ................................................................................................................. 149

9.3.1 Aerobic Static Pile Design ......................................................................................................... 150

9.3.2 Description of Experiment ........................................................................................................ 150

9.3.3 Results ...................................................................................................................................... 151

9.3.4 Aerobic Curing Conclusions ...................................................................................................... 155

Chapter 10. Discussion, Commercial Implications, Conclusions and Recommendations .......................... 156

10.1 Discussion of Research Results ...................................................................................................... 156

10.1.1 Synergy ................................................................................................................................... 156

10.1.2 Aerobic Curing ........................................................................................................................ 159

10.1.3 Comparison of Daisy’s Performance to the Literature ........................................................... 159

10.1.4 Properties of Bulking Agent .................................................................................................... 161

10.1.5 Meeting the Objectives .......................................................................................................... 162

10.1.6 Testing the Hypothesis ........................................................................................................... 162

10.2 Commercial Implications ............................................................................................................... 164

10.3 Conclusions and Recommendations .............................................................................................. 166

10.3.1 Conclusions............................................................................................................................. 166

x

10.3.2 Recommendations for Further Research ................................................................................ 167

11. References ........................................................................................................................................... 171

xi

List of Tables Table 4.1 Comparison of the Properties of ss-FW and mr-OFMSW 20 Table 4.2 Theoretical BMP values compared to experimental values 21 Table 4.3 Elemental Composition of 3 Lignocellulosic Wastes 24 Table 4.4 Physicochemical Properties of 3 Lignocellulosic Wastes 24 Table 4.5 Biodegradability and Methane Yield Data of 3 Lignocellulosic Substrates 24 Table 4.6 Physical Properties and Biodegradability of 6 Paper and Cardboard Products 25 Table 4.7 Effect of Microbial Pretreatment on Lignocellulosic Fibres 31 Table 4.8 Biochemical Methane Potential L CH4.kg-1VSadded 33 Table 5.1 Leach Bed Digestion - Performance Data 41 Table 5.2 Leachate Recirculation Rates and Methane Yields 42 Table 5.3 Leach Bed and UASB Dimensions 44 Table 6.1 Comparison of COD Analysis Methods 71 Table 6.2 Coupon Placement in Leach Beds 72 Table 6.3 Synthetic Feed - pH and Alkalinity Day 56 Table 7.1 Permeability Tests

74 78

Table 7.2 Composition of Synthetic Feed 80 Table 7.3 Composition of Digester Feedstock – Period 1 82 Table 7.4 Biogas Methane Content 91 Table 7.5 Biogas Composition vs. Feeding Cycle 91 Table 7.6 Substrate Destruction Efficiency – Period 1 93 Table 7.7 COD Mass Balance at Start-up – Period 1 93 Table 8.1 Experimental Periods 101 Table 8.2 Daisy's Operating Conditions 102 Table 8.3 COD Substrate Destruction Efficiency by Period 104 Table 8.4 Stable Methane Production and Methane Yield 105 Table 8.5 Major Malfunctions and Maintenance Events 110 Table 8.6 Comparison of COD Analysis Methods 111 Table 8.7 Summary of Biogas and Methane Production (83 weeks) 111 Table 8.8 Elemental Analysis - Average of three samples 112 Table 8.9 Carbon Nitrogen Ratio 112 Table 8.10 Mass Balance by Period 113 Table 8.11 Leach Bed Serial Number Sequence 114 Table 8.12 Composition of Digester Feedstock – Periods 2, 3 and 5 114 Table 8.13 Composition of Digester Feedstock – Periods 4a, 4b and 4c 117 Table 8.14 Physical Properties of Bulking Agents BA4 and BA5 121 Table 8.15 Calculation of Synergy at 100% FW conversion 123 Table 9.1 Total Quantities of Substrates, Inoculum and Medium for BMPs 132 Table 9.2 BMP graphs - gradients mL/d/mgCODadded biogas at 310K 138 Table 9.3 Biochemical Methane Potential Test - Substrate Calculation 138 Table 9.4 BMP graphs – gradients ml/d/mgCOD as biogas at 310K 139 Table 9.5 Total Quantities of Substrates, Inoculum and Medium for Large BMPs 145 Table 9.6 Enzyme Activity Assays 149 Table 9.7 Properties of digestate, leaf and yard waste and mixture 153 Table 9.8 Sieve analysis of digestate mix and compost - percent minus 154 Table 10.1 Comparison of Daisy's performance with published data 160 Table 10.2 Financial Model - Major Assumptions 164 Table 10.3 Capital Cost Summary 165 Table 10.4 Year 10 Financial Projections 165

xii

List of Plates Plate 1. Daisy the Digester 50 Plate 2. Leach bed and solenoid valve 53 Plate 3. UASB Reactor 55 Plate 4. Heating tape installation on UASB 56 Plate 5. Tanks 1 and 2 plus Pumps 57 Plate 6. Leach beds, solenoid valves and leach bed gas manifold 61 Plate 7. Gas meters 1 and 2 and top of UASB 62 Plate 8. Control Panel - Valve sequencer feeding leach bed 6 64 Plate 9. Controls 64 Plate 10. Gas and liquid flows and sampling points 67 Plate 11. FW dried and broken up 69 Plate 12. Installing coupons for digestibility determination 71 Plate 13. Signs of partial leach bed flooding 86 Plate 14. Bulking agent a) BA4; b) BA5; c) BA6 120 Plate 15. BMP 3; Substrate bottles in water bath (rear); 144 Plate 16 BMP 3 Entire set-up with 11 bottles 144 Plate 17. Aerobic Static Pile Composter 1 151 Plate 18. Compost Feed and Product 1 152 Plate 19 Steam rising from active compost 153 Plate 20. Partially digested corn cob 158

xiii

List of Figures Figure 1.1 Schematic of the process (from the patents) .............................................................................. 3 Figure 1.3 Composition of IC&I waste (Ottawa 2007) .................................................................................. 8 Figure 1.2 Composition of MSW (Ontario 2004) .......................................................................................... 8 Figure 4.1 Simplified diagram of anaerobic digestion of organic waste .................................................... 15 Figure 4.2 Digester performance calculations from first principles. .......................................................... 33 Figure 5.1 Schematic flow diagram ............................................................................................................ 49 Figure 5.2 leach bed cross-section ............................................................................................................. 52 Figure 5.3 UASB cross-section.................................................................................................................... 55 Figure 5.4 Schematic of GLS separator in UASB ......................................................................................... 56 Figure 5.5 Tanks 1 and 2 ............................................................................................................................ 57 Figure 5.6 Leachate flows L/h - typical operating conditions ..................................................................... 59 Figure 5.7 Foam trap .................................................................................................................................. 62 Figure 6.1 TS/VS/COD methods – Food Waste .......................................................................................... 68 Figure 6.2 TS/VS/COD methods – Fibres and Bulking Agent ...................................................................... 69 Figure 6.3 TS/VS/COD methods - Digestate ............................................................................................... 70 Figure 7.1 Anaerobic digester temperature profile; Tank 1, Tank 2, UASB and 6 LBs ................................ 77 Figure 7.2 Waste permeability tests; FB alone, FB+BA, FB+BA+FW ........................................................... 79 Figure 7.4 Concentration of COD in UASB vs. time .................................................................................... 80 Figure 7.3 Biogas production with synthetic feed showing feeding events, amount added ...................... 80 Figure 7.5 Synthetic feed BMP results. ...................................................................................................... 81 Figure 7.6 UASB Upflow velocity – finally settled at 0.15 m/h ................................................................... 85 Figure 7.7 Leach bed permeability test – BA Batch 2 in S.017 BA Batch 1 in S.012 to S.016 .................... 87 Figure 7.8 pH and Alkalinity Ratio – recovery from initial failure followed by stable operation ................ 88 Figure 7.10 Recirculating inorganic salts – showing very little variation .................................................... 88 Figure 7.9 Leachate COD Concentration – sharp fall following restart ...................................................... 88 Figure 7.11 Biogas production 2015/05/19 to 2015/05/26 (GM1 – UASB, GM2 Tanks and LBs) .............. 90 Figure 7.12 Weekly biogas production – Period 1; 10 Wks of stable operation from Wk 6 to Wk 15 ....... 90 Figure 7.13 Sample calculation of COD destruction efficiency, with and without bulking agent ............... 92 Figure 7.14 COD balance week 6 to week 15 cumulative .......................................................................... 94 Figure 7.15 Example of Daily Log June 8th 2015; data output; samples; system adjustments ................... 96 Figure 7.16 Datalogger output, June 8, 2015 ............................................................................................. 97 Figure 8.1 Substrate destruction efficiency as TS, VS and COD vs. time (calculated without BA). ........... 103 Figure 8.2 Weekly biogas production showing periods of stable biogas production ............................... 104 Figure 8.3 Alkalinity ratio and pH vs. elapsed time; stability under changing levels of FW addition ....... 106 Figure 8.4 Concentration of inorganic salts in Daisy – following changes in CODFW addition .................. 106 Figure 8.7 Concentration of sulphate in Daisy – steadily rising at higher CODFW addition ....................... 107 Figure 8.6 Total VFA Concentration in Daisy expressed as mg/l COD, showing influence of CODFW ........ 108 Figure 8.5 Concentration of COD in Daisy. ............................................................................................... 109 Figure 8.8 Stoichiometry of digestion of the 83 week average substrate ................................................ 112 Figure 8.9 Weekly biogas production – Period 2; 7 wk. SRT .................................................................... 115 Figure 8.10 Weekly biogas production – Period 3; restart after 10d shutdown; return to 6 wk. SRT ...... 116 Figure 8.11 Weekly Biogas Production – Period 4: FW to zero in 3 steps; stable biogas production. ..... 117 Figure 8.12 Weekly biogas production – Period 5; return to CODFW17.2% .............................................. 119

xiv

Figure 8.13 Weekly biogas production – Period 6; CODFW to 21.7% then 29.3% in 2 steps ..................... 122 Figure 8.14 Destruction efficiency vs. digestion time – S.082 to S.088.................................................... 122 Figure 8.15 Synergistic biogas production from FB vs. FW addition ........................................................ 123 Figure 8.17 Specific Synergistic biogas production in L.kg-1VSadded calculated at 100% FW conv. ............ 124 Figure 8.18 Effect of FW addition on fibre digestibility – coupon data .................................................... 125 Figure 8.19 Concentration of individual VFAs vs. time. ........................................................................... 126 Figure 8.20 Effect of fresh waste on LB temperatures and biogas production ........................................ 128 Figure 8.21 Biogas production from a single leach bed vs. digestion time .............................................. 129 Figure 9.1 BMP2 – Controls; substrate alone and inoculum alone .......................................................... 134 Figure 9.2 BMP2 - CB, BB, NP and FP without FW, FW alone, BA alone; neg. control subtracted ........... 134 Figure 9.3 BMP2 Individual FB plus FW at 13% of CODTotal neg. con. subtracted ..................................... 135 Figure 9.4 Comparison of digestion of each fibre, with and without FW addition .................................. 137 Figure 9.5 Calculation of synergy – normalized for COD content ............................................................ 141 Figure 9.6 Synergy by fibre at 39d and 74d; normalized to constant COD ............................................. 142 Figure 9.7 Percent increase in biogas yield, as a result of FW addition vs time ....................................... 142 Figure 9.8 Large BMP experimental set-up .............................................................................................. 144 Figure 9.9 BMP 3 Negative (leachate) and positive (glucose) controls .................................................... 146 Figure 9.10 Large scale BMP 3; first 25 days showing initial acid inhibition in Experiment 2 FW + FB .... 146 Figure 9.11 Large-scale BMP to 112 days; acid inhibition reverses itself in Exp. 2 .................................. 147 Figure 9.12 Enzyme activity assays. Activity higher at high substrate destruction .................................. 148 Figure 9.13 Aerobic Static Pile Composter; a) plan view, b) side elevation ............................................. 150 Figure 10.1 Synergistic biogas - FW at 78% conv. vs FW at 100% conv. .................................................. 158 Figure 10.2 Fibre destruction efficiency. Shows synergistic effect of FW addition .................................. 159

xv

List of Appendices Appendix A - Equipment List .................................................................................................................... 177

Appendix B - Start-up ............................................................................................................................... 183

Appendix C - Analytical Procedures and Results ...................................................................................... 193

Appendix D - Operating Results ............................................................................................................... 196

Appendix E – Biogas Data ........................................................................................................................ 210

Appendix F – Biochemical Methane Potential Tests ................................................................................ 213

Appendix G Financial Projections ............................................................................................................. 220

xvi

Glossary of Terms and Abbreviations

Analest - Analytical Lab for Environmental Science Research and Training – University of Toronto ASP – aerated static pile (composting) BA – bulking agent BB – boxboard (e.g. cereal box) BMP – biochemical methane potential CB – cardboard (corrugated) COD – chemical oxygen demand CSTR – completely stirred tank reactor CSV – comma separated variable DG – digestate EBIT – earnings before interest and taxes EDF – electronic data file (number suffix) EPS – extracellular polymeric substance FB – fibres (mix of CB, BB, NP and FP) FP – fine paper (office paper) FW – food waste GC – gas chromatograph GM – gas meter (wet tip) HDPE – high-density polyethylene hp – horsepower HRT – hydraulic retention time IC&I – industrial, commercial and institutional waste ID – inside diameter IRR - Internal rate of return LCA - Lifecycle assessment LB - leach bed LT – leachate LYW – leaf and yard waste mr-OFMSW mechanically recovered MSW – municipal solid waste

NP – newsprint NPT – National pipe thread OD – outside diameter OLR – organic loading rate P1 – pump 1 P2 – pump 2 P3 – pump 3 PLC – programmable logic controller PP – polypropylene PVC – polyvinyl chloride RPM – revolutions per minute RTD – resistance temperature detector S001 to S088 - leach beds of waste, numbered sequentially SRT – solids retention time ss-OFMSW source separated organic fraction of municipal solid waste STP – standard temperature and pressure Tank 1 – UASB feed tank Tank 2 – leach bed feed tank TCD – thermal conductivity detector TS – total solids UASB – upflow anaerobic sludge blanket VG – valve on gas system VL – valve on liquid system VP – valve on pump VS – volatile solids VU – valve on UASB Vup – upflow velocity WAS – waste activated sludge WC – water column

1

Chapter 1. Introduction

1.1 An Enduring Obsession

My interest in the anaerobic digestion of solid waste began almost 30 years ago. In 1989 I was

president of Laidlaw Waste Systems Inc., the third-largest solid waste management company in North

America. We operated more than 40 landfills, some were equipped with landfill gas recovery systems,

but most were not. We also owned a subsidiary company, Laidlaw Gas Recovery Systems Inc. (LGRS),

headquartered in San Francisco. LGRS designed, built, owned and operated, seven landfill gas recovery

plants; one sold medium BTU gas to a neighbouring industrial customer, the remaining six generated

electricity for sale to local California utilities. This business was very successful and very profitable, but

the contrast between California, with its advantageous power purchase agreements for landfill gas

projects, and other jurisdictions across North America was stark. Value could not be extracted from

waste materials without some form of regulatory coercion or, as in this case, incentive.

Nevertheless, it seemed clear to me that organic waste did not belong in landfills where it generates

leachate, which has to be managed, and biogas which, in many cases, was not being managed. I

became interested in the prospect of using anaerobic digesters to process organic waste before it ever

got into a landfill, and to recover greater quantities of less-contaminated biogas from a smaller quantity

of waste, and sell the energy. I travelled to Ghent in Belgium to meet the principals of OWS, owners of

the DRANCO technology (an acronym for dry anaerobic composting). DRANCO is a vertical plug-flow

reactor operating at thermophilic temperatures (55°C). I was given a tour of the 10 m³ pilot plant, liked

what I saw, and we undertook to negotiate a five-year North American license agreement; I also agreed

to provide debt financing for 49% of the cost of their first commercial plant, to be built in Brecht,

Belgium. License in hand, we sought business opportunities. Unfortunately, in the end, we were unable

to convince any North American customers to pay the premium necessary to anaerobically digest food

waste, and the unsubsidized energy sales were never going to be sufficient to make it financially viable.

Eventually the license lapsed, but the Brecht plant was a success, underwent an expansion and

refurbishment a number of years ago, and continues to operate to this day. This experience taught me

many things, but perhaps the most important was the vital role that regulations and incentives play in

the adoption of waste processing technologies. My enthusiasm had clouded my business judgment.

My next encounter with anaerobic digestion occurred in the late 1990s when, now working as an

independent consultant, I was asked by TD Capital to evaluate the technical and commercial prospects

of a proposed anaerobic digester for commercial waste, to be constructed as a merchant plant (that is

2

one with no long-term contracts) in Newmarket, Ontario. It was to employ the BET technology from

Germany, a CSTR-based system. My report contained three principal conclusions: the technology was

ill-chosen, the demand for the service was nonexistent, and the principals had no experience in the

waste management industry, deficiencies that would cost them dearly. TD Capital declined to finance

the project, but it did get built and the “opportunity” came around again.

In October 2002, I was commissioned by Waste Services Inc. to provide a technical and financial

evaluation of the very same anaerobic digestion plant. The lenders, CIBC, had foreclosed on their $25

million loan, spent another million dollars to “fix” the plant (we called it perfuming the pig), and were

looking for a buyer. My report recommended that, under no circumstances, should any time or money

be wasted pursuing this project; it was technically flawed beyond fixing and, even if it were fixable, had

no real prospects of being a commercial success.

It wasn’t difficult to write a condemnatory report, but the experience prompted me, and my partner

Ron Poland, to ask ourselves this: in the face of uncertain government policies and the associated

economic challenges, is there a plausible technological approach to the digestion of solid organic waste

that would be sufficiently robust, versatile, and affordable to become a commercial success? Can this

riddle be solved? We set out to try.

1.2 Is there a practical, affordable, solution?

Adding a third partner, Brian Forrestal, the three of us, based solely on our collective 90 plus years of

experience in the solid waste management industry, designed a system that we believed would meet

the objectives. We obtained patents in Canada and the United States (Forrestal et al., 2006a; b).

My M.Eng. thesis, A New Technology for the Anaerobic Digestion of Organic Waste, (Guilford, 2009) is

divided into three parts; a detailed analysis of the role and importance of government regulations and

incentives, a description of the technology, and a detailed financial analysis of a full-scale plant. The

essential elements of the technology are described in the summary below, which was abstracted from

my thesis with minor edits; a simple process diagram is also included, Figure 1.1.

“The process is essentially that of a hybrid sequencing batch reactor with a UASB secondary reactor

(Lissens et al., 2001). It was specifically designed to take advantage of the many benefits of anaerobic

digestion while eliminating the drawbacks of commercially available systems: the result is a process with

significantly lower capital and operating costs than competing technologies. Key elements of the design

are as follows;

The organic waste remains stationary and leachate is recirculated through the waste mass

3

The solids retention time is measured in the “months-to-years” range

Because the waste is stationary, the process is largely indifferent to inhomogeneity of the

feedstock and the presence of foreign objects (though these may affect the rate of digestion).

It incorporates the economies of scale and proven anaerobic degradation processes that occur in

a landfill, or bioreactor landfill,

Yet it has the continuous treatment capability and gas recovery potential of commercial

anaerobic digesters.

The only significant operational disadvantage is that such a design configuration requires much more

space than a commercial anaerobic digestion plant of equivalent capacity. For example, a 60,000 tonne

per year facility, on a greenfield site, would require about seven hectares of land for the process plant

itself and the subsequent open-air curing of the digestate. It is a system for the continuous

decomposition of organic waste to produce biogas and compost. The system has a primary reactor

consisting of multiple reactor zones, all but two filled with waste and sealed.

Figure 1.1 Schematic of the process (from the patents)

Fresh organic waste is placed on a daily basis into the active, partially-filled, zone of the primary reactor.

Prior to placement, the waste is passed through a mechanical device (an agricultural feed processor was

successfully tested) which opens the plastic bags of organic waste and blends the waste with a bulking

agent (coarse wood chips were tested) without reducing particle size. A 50:50 w/w blend yielded a bulk

density of 750kg/m3 and a porosity of 0.52. During pretreatment and placement, the waste is constantly

exposed to air. Once placed in the primary reactor zone organic waste remains undisturbed until it is

excavated following treatment. Organic waste in each zone is allowed to decompose anaerobically until

gas production is essentially complete, at which point the process is switched to aerobic decomposition

followed by excavation of raw compost. A secondary upflow anaerobic sludge blanket (UASB) reactor

anaerobically digests the organic content of leachate generated in the primary reactor to produce spent

UASB Anaerobic Digester

Reactor Zone 2

(Aerobic)

Reactor Zone 3

(Anaerobic)

Reactor Zone 5

(Anaerobic)

Reactor Zone 6

(Anaerobic)

Reactor Zone 4

(Anaerobic)

Air/O2

Gas Plant

Composting

Excavate Reactor Zone 1 and return to Filling and repeat cycle

4

liquor and biogas. The UASB effluent is recirculated to the primary reactors, which thus act as leach beds,

to enhance anaerobic decomposition. The biogas harnessed from both reactors is collected and treated,

used as fuel for the operation of the organic treatment facility, or for sale as medium Btu gas or electricity.

The organic waste decomposition process is controlled within the primary reactor such that anaerobic

digestion is optimized between the primary and secondary reactors.

All the individual unit operations are simple, reliable and proven. All have been used successfully many

times in other waste management operations and none is being asked to operate outside its normal

operating parameters:

Receipt and placement of waste and application of cover are essentially the same as used in a

conventional landfill but without the need for compaction.

The primary reactor is almost identical in construction to a bioreactor landfill cell and operates as

a leach bed.

The secondary reactor is proven technology from the wastewater treatment industry.

The collection of gas and conversion into useable energy is similar to the system used for landfill

gas recovery.

The aerobic curing process is similar to that used in windrow composting as is the final screening

to recover compost overs and contaminants such as plastic bags.”

Our approach was to design a system that would process the waste, largely as-received and with little

pre-treatment, using proven unit operations. In contrast, the more conventional European technologies

such as Dranco (vertical plug-flow), Kompogas (horizontal plug-flow) and Valorga (CSTR) (Lissens et al.,

2001), are all based around the idea that the waste can be effectively, and affordably, altered to suit

the process, through a combination of separation at source and pretreatment at the plant. Based on all

of the above, we were able to convince ourselves that the process would “work”; the initial financial

model, based on very conservative assumptions (also included in my M. Eng. Thesis) showed that the

technology was close to being competitive with landfill.

The technology (referred to hereinafter as the BioPower process) is licensed to Miller Waste Systems

Inc. and is incorporated in a planned, large-scale, privately-financed waste diversion project in the City

of Ottawa. The Environmental Assessment for the project, the result of seven years’ work, was

approved by Ontario’s Minister of Environment and Climate Change on May 17th 2017. The project

must now proceed through several other stages of approval before construction can begin.

5

Nevertheless, beyond our practical experience as garbage men, our confidence had no real scientific

underpinnings, nor had we examined the rationale behind the choice of AD in the first place, or the

driving forces, beyond financial return, that would turn it into an attractive alternative for managing

organic solid waste. It became my mission to try and change this.

In 2011 I applied to study for my Ph.D in the Department of Chemical Engineering and Applied

Chemistry with the aim of making this technology the focal point of my research. My initial benchmarks

for the performance of the technology, and for my research, were the ones described above: versatility,

robustness, and affordability. These criteria (particularly the first two) are too subjective to serve as

testable hypotheses, but we shall see how they evolved as the work progressed and the capabilities of

the process became clearer.

1.3 The Rationale for Anaerobic Digestion

To begin my research, it was necessary to take a long step backwards, adopt a broader perspective, and

ask two fundamental questions: why is anaerobic digestion a preferred method for treating organic

solid waste, and why do we need a new technology? By answering these questions, it was my goal to

justify the choice of anaerobic digestion, explain why this particular technology is appropriate, describe

its advantages and commercial prospects, and provide suitable criteria against which to judge its

performance.

1.3.1 Organic Solid Waste

What is organic solid waste, where does it come from and how is it managed today? Solid waste comes

from the homes we live in, the places where we work, the institutions we attend, and the commercial

and industrial enterprises that are a central part of our economic life. The proper management of solid

waste is the responsibility of each of us individually and all of us collectively. The generation and

management of solid waste are directly influenced by a unique combination of forces; social,

governmental, technical and economic. Social because the nature of society is a determinant of what

we consume and throw away; governmental because the improper handling of waste leads to

environmental damage from releases of leachate, and of greenhouse gases, and because the desire to

recycle more, and dispose of less, requires government intervention; technical because the handling of

solid waste for any purpose other than transportation and landfill disposal is complex and difficult; and

economic because meeting the technical challenges of waste processing and recycling is expensive.

6

What is it about waste that makes it so difficult and costly to process? In my experience, there are four

principal reasons. Firstly, it is a complex mixture of all the things that we, as individuals and as a society,

no longer want - it is plentiful and extremely heterogeneous; secondly, its composition is variable -

from place to place, from time to time and with the seasons; thirdly, beyond the enormously wide

range of things that one might expect to find in garbage, there is always the “foreign object” – a brake

drum in the green bin is a real-life example – that will wreak havoc with processing equipment. Fourthly

and lastly, the nature and composition of waste itself actually evolves over time (Wilson, 1976) – for

example, the growth in consumer packaging is a post-war development which profoundly altered the

solid waste stream. Collectively, these factors present unique problems for the engineer trying to

design a solid waste processing system.

Broadly speaking there are five approaches to the management of solid waste; the last four are

applicable to organic waste, and it is on these that we shall concentrate:

Recycling

Composting

Anaerobic digestion

Energy from waste (combustion)

Landfill

Landfill is the universal omnivore; it is versatile, it is robust, it is unaffected by any of the complications

associated with the heterogeneity and physical properties of the waste, and is therefore relatively

inexpensive. It is readily apparent that any form of waste processing quickly adds complexity and cost.

Thus diversion from landfill will only occur when regulations are put in place to require it. Such

regulations, either explicitly or implicitly, drive up the costs of the options deemed undesirable

(obviously landfill but sometimes combustion too) and make affordable the options that are preferred,

such as recycling, composting, anaerobic digestion and energy from waste. My M.Eng. thesis contains a

detailed examination of the way government intervention affects, often profoundly, the economics of

solid waste management (Guilford, 2009). In particular, it describes the overarching objectives of the

EU Landfill Directive (European Union, 1999) which was specifically designed to drive organic waste out

of landfills. Thus the EU and its member states have, over many years, imposed regulations and strict

standards on how organic waste is managed; as a result, expensive waste management technologies,

like anaerobic digestion (AD) and mechanical biological treatment (MBT), have become necessary and

therefore, by definition, affordable. Consequently, many European nations have adopted various forms

of AD as important components of their modern waste management systems (De Baere, 2000; De

7

Baere, 2008; De Baere L., 2013). By comparison with the European Union, the regulations that have

governed organic waste diversion in North America in general, and the Province of Ontario in particular,

are less effective, which explains why advanced organic waste processing, particularly anaerobic

digestion, is very common in Europe and very rare in North America. This may finally be about to

change in Ontario with the enactment of the Waste Diversion Transition Act (Ontario, 2016), but no

regulations under the act have yet been promulgated, so the future remains uncertain.

How big is the problem? In Canada we still dispose of 25,000,000 tonnes of solid waste per year (after

recycling), all but a very small amount of it going to landfill. In Ontario about 76% of the solid waste

produced (before recycling) still goes to landfill (and most Provinces have similar records) (Canada,

2015; Government of Canada, 2010). Landfills generate leachate to be treated and biogas to be

collected (or released into the environment). Organic waste, as much as 40% of the total going to

landfill (Government of Ontario, 2004), is the principal source of this leachate and biogas, and its

diversion from disposal would meet the dual objectives of environmental protection and renewable

energy generation, defined as the production of energy from renewable resources that can replenish

themselves as rapidly as they are consumed, in this example the resource is biomass.

To define the nature and scope of the task in more precise terms, it is necessary to go beyond the big

picture just provided, and define the nature and sources of solid waste in general, and the organic solid

waste component in particular. Waste management terminology is inexact and varies significantly from

one political jurisdiction to another. In North America we define the sources of solid waste as:

MSW - municipal solid waste (originating from residences, both single family homes

and apartments)

IC&I - industrial, commercial and institutional waste

C&D construction and demolition waste

Such definitions are useful for large-scale planning purposes but not for designing waste processing

systems. Organic waste is present in the first two but not the third, which mostly consists of roof

shingles, brick, concrete and wood, with minor amounts of cardboard from the construction

component. The composition of residential solid waste, presented in Figure 1.2, comes from a study by

Mohareb (2008) and was obtained by hand-sorting garbage into separate piles and weighing each pile.

The composition of commercial and industrial waste comes from a report published by the City of

Ottawa (City of Ottawa, 2007) and is presented in Figure 1.3.

8

They give a clear picture of the difference between MSW and IC&I waste; note the differences in the

make-up and quantity of the broadly-defined organic waste component, which includes food waste,

paper and cardboard and leaf and yard waste. Both figures show the composition of the waste before

waste diversion.

Residential waste represents 37% of the total solid waste produced, and IC&I plus C&D make up the

remaining 63%. In 2006, Canadians diverted 22% of the total waste stream from landfill; by 2010 this

had grown to 24% (Government of Canada, 2010). Organic waste (broadly defined) constitutes about

62% of both MSW and IC&I waste, and is the single largest component by far. But their respective

proportions of food waste to paper fibres are vastly different. MSW consists of 38% food waste + leaf

24%

38%5%

3%

4%

26%

Paper, cardboard Organics (food & yard) Glass Metal Plastic Other

Figure 1.2 Composition of MSW (Ontario 2004)

42%

14%5%

10%

10%

19%

Paper, Cardboard Organics (food) Glass Metal Plastic Other

Figure 1.3 Composition of IC&I waste (Ottawa 2007)

9

and yard waste, plus 24% paper and cardboard, while IC&I waste consists of 14% food waste plus 47%

paper and cardboard. These figures serve as a guideline only, and should not be taken too literally for

planning purposes. They are subject to variability caused by methodology, which is anything but

standardized, the local balance of single-family/multifamily residences, the nature of local industry and

commerce, and the season of the year in which the samples were taken.

Some paper and cardboard can be, and is, recycled. Though more can be done, much of it is too

contaminated to be recovered economically, and therefore requires treatment as organic waste. In

many cities in Canada, food waste from single-family homes is separately collected and mostly

composted, but separate collection of food waste from multi-family homes (apartment buildings) is

very rare; it must therefore be disposed of without processing. Likewise, very little of the organic waste

from commercial and industrial sources is recovered, largely for reasons of cost; landfill is cheaper. It is

hard to estimate precisely how much organic waste, from all sources is being landfilled, but a Canada-

wide figure of 8,000,000 tonnes per year is probably conservative. This has implications for government

policy, the development of processing technologies, the economics of organic waste processing and the

generation of renewable energy.

1.3.2 The Case for Anaerobic Digestion

Fugitive releases of landfill gas, containing approximately 50% CH4, are a major source of greenhouse

gas emissions; 20 million tonnes per year as CO2eq. (Canada, 2013). Organic waste can be buried,

burned, composted or digested; how do these alternatives compare from an environmental,

particularly greenhouse gas emissions, perspective? Combinations of these options have been

compared by several authors using a lifecycle analysis approach. Bernstad and Jansen (2011) compared

incineration, composting, and anaerobic digestion of household food waste and concluded that AD has

the lowest GHG emissions. A direct comparison of landfill gas to energy with anaerobic digestion (with

electricity production) by Sanscartier et al.(2012), showed that under all circumstances AD was

environmentally preferred, but required financial subsidies. More recently, Hodge et. al. (2016)

conducted a very thorough LCA analysis of the management of food waste from industrial, commercial

and institutional sources in the United States, and concluded that, from a GHG perspective, anaerobic

digestion combined with landfill was the leading alternative in terms of global warming potential

reduction. They also noted that regulations governing the management of organic waste were

beginning to emerge in individual states and cities in the US. So the argument that the diversion of

organic waste from landfill is highly desirable is very strong. But the question still remains, how can this

be done economically within the current North American regulatory and economic framework?

10

1.3.3 The Rationale for a New Technology

The traditional chemical engineering approach to process design is driven by four objectives;

maximizing throughput, reducing hydraulic and solids retention times, reducing capital and operating

costs, and minimizing space requirements. Without costly source separation or extensive pretreatment,

this approach to anaerobic digestion will fail, and it is because of this that the proposed design is

different. It is premised on the idea of accepting slower reaction rates, longer retention times (in the

weeks to months range) and larger physical size, in exchange for simplicity of design and operation and

a larger footprint, with lower capital and operating costs than for conventional CSTR or plug flow

digesters of similar capacity.

There is also another, more subtle, aspect to the design. An anaerobic digester fulfils three purposes; it

reduces waste volume, it generates renewable energy and it produces a stabilized digestate when

combined with aerobic curing, but these purposes are not perfectly congruent. There will be

circumstances, related for example to process design or to substrate composition, in which stabilization

takes precedence over energy production or vice versa. From a waste manager’s perspective, volume

reduction and stabilization of the waste will always take priority over maximizing energy recovery.

There are two reasons for this; firstly, volume reduction and waste stabilization are the primary

objectives but, more importantly, it is also how most of the revenue is generated. The financial analysis

included in my M.Eng. thesis (Guilford, 2009) showed that only 30% of the gross revenue would be

generated from the sale of the recovered energy and 10% from the sale of compost, with 60% coming

from the tipping fee. It would take a profound increase in the value of renewable energy, or the

introduction of new regulations and incentives, to alter this circumstance because tipping fees are

market-specific, commercially competitive, and highly dependent on transportation cost. For these

reasons the engineering approach to this project is driven by the requirement to stabilize as much

organic waste as possible, and not by the desire to generate as much energy as possible, so whenever a

conflict between these two objectives arises waste stabilization prevails.

The proposed approach using a series of sequentially batch-fed leach bed reactors linked to a UASB (a

form of solid-state anaerobic digestion or SS-AD) has been successfully demonstrated at a laboratory

scale digesting grass silage (Nizami & Murphy, 2011). I have found no published research on the use of

this method to digest heterogeneous mixed solid waste of variable composition, or attempts to study

changing waste composition as a process variable. This is the starting point for the design of the

research programme.

11

Chapter 2. Objectives of the Research Programme

2.1 Starting Principles

Successful application of this approach to digesting heterogeneous and variable feedstocks can be

hypothesized by applying a specific set of propositions;

a) The range of organic wastes reliably treatable by AD, at an affordable cost, can be greatly

expanded beyond those originating from single-family homes, to include waste from multi-

family buildings and commercial/industrial/institutional sources.

b) The primary objective is to stabilize organic waste for environmental reasons, and the

secondary objective is to recover renewable energy for sale.

c) A reasonable proxy for the organic fraction of IC&I solid waste would comprise a mixture of

food waste, corrugated cardboard, boxboard, newsprint and fine paper, plus leaf and yard

waste.

d) The feedstock will receive as little pretreatment as possible.

e) By keeping the feedstock stationary and recirculating the liquids, the costs associated with

materials handling and pretreatment, and with equipment failure caused by the presence of

unwanted objects, can be dramatically reduced.

f) Compared to conventional AD (e.g. completely stirred tank and plug flow reactors) the mixing

of liquids and solids will be relatively inefficient; the reaction rate will be slower; the solids

retention time will be greater, the reactor volume larger, and more space will be required for a

plant of given throughput.

g) This trade-off will be economically advantageous.

h) Long-term process stability can be achieved by using multiple leach beds (charged with a blend

of organic waste and bulking agent, prepared with the minimum of pre-processing), with each

leach bed managed independently according to the properties of its contents and stage in the

digestion process, and all connected to a single UASB.

i) The composition of the waste can differ from one leach bed to another without causing process

instability.

j) Given the right biochemical and microbiological conditions, organic waste will digest, even if it

is heterogeneous and variable, and even if it contains foreign objects and undigestible waste

(below concentrations toxic to the process).

k) As the microbial community is converting the substrate, the community itself is evolving,

making it more, or less, resilient to sudden changes in the make-up of the substrate or the

12

presence of inhibitors; the effects of changes in the substrate under these circumstances are

not fully understood. By gaining an understanding the factors that enhance resilience and why,

it may be possible to maintain process stability when substrate changes occur.

l) Renewable energy in the form of biogas can be generated in commercial quantities.

m) The digested contents of each leach bed will require aerobic treatment (composting) as a final

step. The composted digestate may be too contaminated to be suitable for land application

which, while desirable, is less important than removing organics from landfill and generating

renewable energy.

Collectively, these propositions are different from those used as the basis of traditional AD and form

the starting point for setting out the objectives and planning the PhD research programme. The specific

objectives which follow are influenced, in part, by the Literature Review presented in Chapter 4.

2.2 Specific Objectives

1) Design, build, and commission a laboratory scale version of the BioPower process

2) Test and troubleshoot the initial operation to confirm operability and create a baseline for

subsequent experiments

3) Develop analytical methodology and a sampling plan compatible with the operation of the

digester that will enable mass and energy balance calculations and performance monitoring

and benchmarking to other systems

4) Operate the reactor under a variety of test conditions to measure efficiency in terms of biogas

yield, COD destruction and solids stabilization, as well as robustness and stability, starting with

examining the effects of changes in substrate composition

5) Use the performance data obtained to refine economic predictions for a full scale system

2.3 Hypothesis

The hypothesis that the research programme has been designed to test is as follows:

An anaerobic digester, comprising multiple sequentially-fed leach beds feeding a single UASB is capable of successfully processing mixed organic wastes, consisting of varying proportions of cellulosic wastes (paper, cardboard, boxboard) and food waste, to produce commercial quantities of biogas and a stabilized digestate.

This can be accomplished by managing the leach beds independently of one another, balancing leachate flow between recirculation to the leach beds and delivery to the UASB, allowing adequate solids retention time to achieve the maximum practical conversion of biomass to biogas, and by aerobically curing the digestate. Digestion rate and yield can be enhanced by managing the proportion of food waste to fibres; the production of waste water (except for water-of-saturation in the digestate) can be

13

minimized or even eliminated; process stability can be maintained, despite feedstock variability, by controlling two principal variables; C:N ratio and alkalinity.

14

Chapter 3. Thesis Outline

The body of the thesis is laid out as follows. The literature review is presented in Chapter 4 and

concludes with a summary and a statement of the knowledge gaps. Chapter 5 describes the process of

system design through to the completion of the construction of the lab scale digester. A description of

the analytical methods follows in Chapter 6. Commissioning and startup are covered in Chapter 7;

startup is divided into three segments and concludes with 10 weeks of stable operation, creating a

baseline for the experiments that follow. Chapter 8 is a description of digester operations through 83

weeks, divided into six operating periods each devoted to a particular aspect of the research, for

example changes in feedstock composition. To complement the data obtained from the operation of

the digester, three supporting experiments were conducted, related to substrate digestibility and

aerobic curing; these are described in Chapter 9. Results, discussion, conclusions and recommendations

are presented in Chapter 10. The appendices are in two parts; additional information in the form of

figures and tables, plus large data files in electronic form to be made available through TSpace at:

https://tspace.library.utoronto.ca/handle/1807/9945

15

Chapter 4. Literature Review

Anaerobic digestion is a natural process in which organic matter decomposes in the absence of oxygen

to produce biogas (roughly equal quantities of methane and carbon dioxide) and digestate containing

inorganic matter and residual un-decomposed organic compounds. Anaerobic digestion takes place in

four sequential steps, hydrolysis, acidogenesis, acetogenesis, and methanogenesis, each the work of a

different family of microbes; a simplified schematic is shown in Figure 4.1. It is in fact a much more

complex process than this diagram would suggest, and a microbiologist would find it woefully

oversimplified.

But from a humble waste management engineer’s perspective it is very useful. The intermediate

products generated at each step of the process serve as the substrates for the next; as long as this all

stays in balance the process will remain stable. With easily digested substrates, methanogenesis is

usually the rate-determining step, so if the concentration of volatile fatty acids (VFAs) rises and the

methanogens are unable to consume them as rapidly as they are produced, they will accumulate, the

pH will drop, and the process will be inhibited, either temporarily or permanently.

Anaerobic digesters are relatively common at wastewater treatment plants, especially in Europe. But

more recently, anaerobic digestion technologies have been developed specifically for the purpose of

Hydrolysis by

Hydrolytic Bacteria

Complex Polymers Carbohydrates, Fats, Proteins, Lignocellulose

Monomers Fatty acids, Sugars, Amino-acids

Hydrogen, Carbon Dioxide H2, CO2

Volatile fatty acids Propionic acid, Butryric acid,

Biogas CO2 + CH4

Acetic Acid CH3COOH

Acidogensis by

Fermentative Bacteria

Methanogenesis by

carbon dioxide reducing hydrogen oxidizing acetogens and

Acetaticlastic methanogens

Acetogenesis by

Hydrogen oxidizing acetogens Hydrogen-producing Acetogenic bacteria

Figure 4.1 Simplified diagram of anaerobic digestion of organic waste

16

treating organic solid waste. This occured in response to environmental regulations, particularly in

Europe, requiring the diversion of organic wastes from landfill. Regulations provide an economic driving

force for technical development in the field, a subject that I studied for my M.Eng. (Guilford, 2009).

The application of anaerobic digestion to organic solid waste brings with it greater complexities and

engineering challenges than does the treatment of wastewaters streams. The research performed on

the anaerobic digestion of solid waste is the subject of this literature review, which focuses on the solid

state anaerobic digestion (SS-AD) of food waste, cardboard, boxboard (e.g. cereal boxes), newsprint,

fine paper and wood chips.

4.1 What Kind of Anaerobic Digester?

To choose a basic design concept for an anaerobic digester intended for the treatment of solid waste,

three binary decisions have to be made;

o “wet” or “dry” (an operational distinction, generally taken to mean below 12% or above 20% total solids),

o single-stage or two-stage, and

o mesophilic (37°C) or thermophilic (55°C).

Nizami & Murphy (2010) reviewed the pros and cons of “wet” versus “dry”. The “wet” process operates

from 2-12% solids; the “dry” process operates from 20 - 50% solids. The dry process requires simpler

pretreatment and consumes less energy, but mixing of substrates and materials handling are more

complex. The wet process despite a longer operating history (from wastewater treatment) has

shortcomings; high water consumption and sensitivity to shock-loadings among them. In many respects

this choice must be made based upon the properties of the substrates.

A single-stage digester is obviously less complex and less expensive to build, but the two-stage

alternative has, at least in theory, certain advantages. In particular by operating the first stage at lower

pH, substrate hydrolysis can be enhanced; the two-stage process is also said to be more robust and less

susceptible to failure (Nizami et al., 2010). The theory behind two-stage digestion is a simple one;

hydrolysis of the solid substrate and acidogenesis take place in the first stage, then the leachate

containing VFAs and other soluble intermediate products is fed to a methanogenic reactor where they

undergo acetogenesis and methanogenesis by a consortium of archaea and bacteria, see Figure 4.1. In

practice it is not so clear-cut; biogas can be generated in the first reactor and in the leachate storage

tanks, as well as the methanogenic reactor; also the role and behaviour of the first reactor can change

with the age of waste that it contains. Nevertheless, two-stage digestion of many substrates, including

17

organic solid waste, has been shown to have important advantages over single-stage systems, for

example minimal feedstock preparation and digester stability. The literature shows that the single-

stage digester of choice is generally a CSTR, and that two-stage digesters comprise either a leach bed or

a CSTR coupled to a methanogenic reactor, commonly a UASB but sometimes an anaerobic filter (Mata-

Alvarez et al., 2000). Some work has been performed using two CSTRs in series; in one example Banks

et al. (1998) used a two-stage CSTR, digesting a mixture of lignocellulosic wastes, to demonstrate

enhanced process stability, when compared to a single CSTR, by flushing VFAs from the first reactor to

the second. Also, two CSTRs operating in series, digesting highly degradable food waste, showed

enhanced process stability (Pavan et al., 2000). Chugh et al. (1999) tested a pair of leach beds operating

in series, digesting MSW; the system was set up with fresh waste in the first leach bed and stabilized

waste in the second. The result was accelerated substrate degradation compared to a single-stage LB

under similar conditions.

In a 20 day side-by-side comparison of a single leach bed alone with a leach bed coupled to a UASB,

digesting presorted MSW, the two-stage digester gave a 30% greater methane yield and 2.5x the

volatile solids reduction (O'Keefe & Chynoweth, 2000). In a side-by-side comparison of a single-stage

CSTR to a two-stage reactor comprising 6 Leach beds and a UASB digesting grass silage, Nizami &

Murphy (2010) showed that, while the CSTR gave higher yields (451 L CH4 kg-1VSadded in 50d vs. 341 L

CH4 kg-1VSadded in 30d, the two-stage reactor could operate at higher loading rates, was more versatile,

and purported to be less expensive at commercial scale. In a very recent paper, Di Maria (2017)

obtained similar results from single-stage digestion of OFMSW; a CSTR gave yields of 320

L.CH4.kg-1VSadded, compared to a LB which gave 250 L CH4 kg-1VSadded; they also confirmed the benefits of

leachate recirculation to improve the digestion efficiency of SS-AD. Siegert & Banks (2005) showed that

the separation of the process into two stages makes it possible to minimize VFA accumulation and

prevent acidification from inhibiting hydrolysis in the leach bed first stage, while maintaining a higher

pH in the UASB second stage. Nizami and Murphy (2011a), using grass silage as substrate, maintained a

pH 5.9 in the leach beds and pH 7.4 in the UASB, creating favourable conditions for hydrolysis and

acidogenesis in stage 1 followed by acetogenesis and methanogenesis in stage 2.

A side-by-side comparison of mesophilic and thermophilic digestion of cardboard and paper, in a 150

day experiment by Qu et al. (2009), showed that higher methane yields were generated in the

mesophilic system. For the digestion of food waste, mesophilic digestion was shown to be more stable

than thermophilic (Yirong C., 2013). Mata-Alvarez et al.(2000) in their review paper, found that the

benefits of thermophilic digestion depended on the substrate, and that sometimes these were offset by

18

higher energy demands. Kim (2012) compared the hydrolysis of organic wastes under thermophilic and

mesophilic conditions in sequencing batch reactors, finding that protein hydrolysis at thermophilic

temperatures was four times faster than at mesophilic.

De Baere (2000) reported on the prevalence of various commercial reactor designs across Europe with

the following results; Europe is 38% “wet” and 62% “dry”; 93% single stage and 7% two stage; 67%

mesophilic and 33% thermophilic. The dominance of single stage digestion is presumably because,

when the feedstock has been extensively pre-treated, two-stage no longer offers sufficient advantages

to justify the cost.

Summary (Digester design concept) Both wet and dry anaerobic digestion technologies are proven, both in the laboratory and on a

commercial scale. The choice of which to use is largely dependent on the physical properties of the substrates, and the amount of pretreatment given to the waste.

A lot of laboratory research has been devoted to two-stage versus single-stage anaerobic digestion, partly because two-stage digestion appears to offer advantages for solid substrates. However, single-stage digestion is by far the more common in commercial applications, for reasons of economics and simplicity.

Far less research has been conducted on direct comparisons of mesophilic and thermophilic digestion and the results are somewhat inconclusive. Nevertheless, both work successfully at a commercial scale. Thermophilic is also a greater challenge in cold climates but does a better job of killing pathogens, a necessary step that can also be accomplished during aerobic curing to create a compost product.

4.2 Substrates, Bulking Agent and Digestate

As we have seen, organic solid waste from municipal and commercial sources consists primarily of food

waste and various lignocellulosic fibres such as cardboard, boxboard, newspaper etc., plus leaf and yard

waste. From the perspective of their behaviour in the AD process and their biodegradability, substrates

can be divided into two separate groups; firstly FW alone, secondly wood-derived fibres collectively

(FB) and unprocessed wood, the bulking agent (BA).

Food waste is one of the most extensively studied substrates for anaerobic digestion; it is produced in

large quantities and, improperly managed, it presents an environmental threat; but it is a readily-

digestible source of renewable energy. Lignocellulosic wastes are produced in even larger quantities,

with much greater renewable energy potential (per dry tonne), but they digest more slowly and,

arguably, are less of an environmental threat as a result. This section reviews the relationship between

the digestion process and the substrates and their properties under the following headings:

o Food Waste o Lignocellulosic waste

19

o Co-digestion o Feedstock preparation o Size reduction o Digester solids content o Bulking agent in anaerobic digestion o Digester performance – predictions from the literature o Aerobic curing – the final step

4.2.1 Food Waste

Food waste contains many things, fruits and vegetables, meats, dairy products, baked goods, peelings

and trimmings from fresh waste, plate scrapings, bones, shells, paper towels and much more besides.

Furthermore, composition depends on the season of the year, methods of collection, the socio-

economic circumstances of the people generating it, and subsequent preprocessing. Five descriptive

terms are common in the literature (Guilford, 2009; Zhang et al., 2008);

Source separated organic fraction of municipal solid waste (ss-OFMSW) Mechanically recovered organic fraction of municipal solid waste (mr-OFMSW) Fruit and vegetable waste (FVW) Canteen or cafeteria waste (a more restricted, less contaminated, version of ss-OFMSW) Source-separated organics (SSO)

This last term is a North American version of ss-OFMSW. These are all operating definitions of

convenience and cannot be relied upon for anything beyond general guidance. There are probably as

many definitions of food waste (or indeed any other organic waste) as there are laboratories studying

it, and this makes inter-laboratory comparison of research very difficult. Often, but not always, the

situation is made clearer by the inclusion of an analysis of the food waste.

The example shown in Table 4.1 is a very detailed comparison of the properties of ss-OFMSW and ms-

OFMSW from Zhang et al. (2012a); it illustrates how different source-separated and mechanically-

separated waste streams can be. Zhang used the data to estimate the biochemical methane potential

(BMP) of the two waste samples using three different methods, biochemical composition, the Buswell

equation, and a carbon mass balance. The biochemical composition was derived from the work of

Angelidaki & Ellegaard (2003), who calculated the theoretical methane yields of several substrates. It

appears that Zhang et al. (2012a) appropriated these data by assuming that lignocellulosic products

could be treated as carbohydrates, though this is not explained in their work.

The Buswell equation method (Banks et al., 2012; Nizami et al., 2011) is based on elemental analysis

and stoichiometry; the result is the maximum possible yield of methane, assuming 100% chemical

20

oxygen demand (COD) conversion. The carbon mass balance was based on the measured destruction of

volatile solids (VS) and biogas composition, which explains why it is closest to the measured values.

Table 4.1 Comparison of the Properties of ss-FW and mr-OFMSW

ss-MSW mr-OFMSW

General

Total solids (TS) 24 53

Volatile solids (VS) (% of TS) 91 64

Biochemical composition (g/kg) (VS basis)

Carbohydrates 453 340

Lipids 151 69

Crude proteins 235 130

Hemicellulose 38 52

Cellulose 50 252

Lignin 17 184

Nutrients and potentially toxic elements (TS basis)

TKN (g.kg-1) 34.2 13.9

Total phosphorus (g.kg-1) 5.41 2.17

Total potassium (g.kg-1) 14.3 4.26

Cd (mg.kg-1) <1.0 1.50

Cr (mg.kg-1) 29.0 263

Cu (mg.kg-1) 7.2 107

Hg (mg.kg-1) <0.010 0.179

Ni (mg.kg-1) 7.0 97.0

Pb (mg.kg-1) <10 162

Zn (mg.kg-1) 33 259

Elemental composition) (TS basis) (TS basis) (VS basis)

C (%) 47.6 33.0 53.6

H (%) 7.0 4.8 7.80

O (%) 33.3 22.2 36.1

N (%) 3.4 1.3 2.14

S (%) 0.15 0.25 0.41 Adapted from (Zhang et al., 2012a)

The two substrates were digested in CSTRs, in a side-by-side experiment, which ran for 284 days. The

BMP data are compared to the experimental digester performance data in Table 4.2. These results

illustrate three important points, the marked difference in waste properties and digester behaviour

21

between source-separated waste and mechanically separated waste, the challenges to maintaining

digester stability with FW alone, and the value of analytical data to predict digester behaviour.

Table 4.2 Theoretical BMP values compared to experimental values

Data Source ss-FW

CH4 mL/gVS

mr-OFWSW

CH4 mL/gVS

mr-OFWSW

Exper. as percent

theor. (Buswell)

Experimental BMP value (80d) 445 344 63

Experimental semi-continuous trial (284d) 425a 304b 56

BMP biochemical composition 494 401

BMP Buswell equation 547 557

BMP carbon mass balance 467 349

Adapted from (Zhang et al., 2012a) afailed at 180d bran to 284d

Zhang et al. (2012a) is one of many papers published by Charles Banks’ laboratory at the University of

Southampton. The work performed in this lab is generally of a very high quality; it employs the same

equipment and relies on the same basic analytical data, ensuring internal consistency, however

comparison with other labs is much more uncertain and must be made with care.

Most of the research on the digestion of food waste has been conducted in single-stage CSTRs, or a

two-stage single leach bed plus UASB; while there are suggestions in the literature that two-stage

digestion is inherently more stable than single-stage (Nizami & Murphy, 2010; Pavan et al., 2000; Shin

et al., 2001; Wang et al., 2003), it is not a panacea; feedstock, digester design and process operation are

all important; this will be dealt with further in Chapter 5 Digester Design and Construction.

The digestion of food waste alone is subject to instability for many reasons (Chen et al., 2008;

Kayhanian, 1999; Zhang & Banks, 2013; Zhang et al., 2012a), but two stand out. Firstly, FW degrades

easily, quickly producing high concentrations of VFAs and reducing pH; secondly the high nitrogen

content of food waste leads to high ammonia concentrations. Both of these conditions can lead to

methanogen inhibition and process failure. A critical factor affecting food waste digestion is finding a

way to achieve the engineer’s objectives of high rate and yield without inducing the inhibitory effects of

VFAs and ammonia.

VFA Accumulation If the methanogens are unable to consume a surge of VFAs, and the alkalinity of the digester is

insufficient to neutralize the acidity which ensues, the pH will fall, the methanogens will be inhibited,

and the reactor will halt and may fail altogether. This behaviour has been observed in both single stage

22

CSTRs (Zhang et al., 2012a), and two-stage digesters employing leach beds and a USAB (Browne, 2013;

Browne et al., 2013); it can occur at any time in a long-running experiment, sometimes prompted by

operational changes (Climenhaga & Banks, 2008a; Climenhaga & Banks, 2008b) but sometimes for no

apparent reason (Jiang et al., 2012; Zhang & Banks, 2013; Zhang et al., 2012a). The wash-out of crucial

trace elements, particularly Se, Co and W, which are specifically required by certain enzymes for

methanogenesis, is one contributing cause. Comparing reactors with and without trace element

supplements showed that the former remain stable much longer – 520d vs. 120d. (Banks et al., 2012;

Jiang et al., 2012). In a parallel study, Zhang et al. (2012b) maintained stability by adding a mixture of

CB and BB to the food waste, instead of adding trace elements; as a result a reactor facing incipient

failure was restored to stability, though this did not happen immediately, taking about 90 days in all. It

has also been shown that lengthening the hydraulic retention time (HRT) to reduce loss of trace

elements by wash-out can prevent, or greatly postpone, reactor failure (Climenhaga & Banks, 2008a;

Nagao et al., 2012). Conversely, and somewhat paradoxically, three articles show that flushing

accumulated VFAs out of a leach bed reactor to a UASB in a two-stage digester, stabilizes digester

operation (Climenhaga & Banks, 2008b; Han & Shin, 2004; Shin et al., 2001); note that the SRT of these

experiments was between 25 and 30 days. There is another perspective; Stroot et al. (2001) conducted

an extensive study of the single-stage anaerobic digestion of a synthetic OFMSW under mesophilic

conditions, operated semi-continuously with daily wasting and feeding, using either cattle manure or

waste activated sludge as inoculum. The variables studied were organic loading rates and mixing

practices. They determined that waste activated sludge was superior to cattle manure, that minimal

mixing led to higher yields than continuous mixing – especially at higher OLRs, and that unstable,

continuously mixed, reactors could be stabilized by shifting to minimal mixing. The continuously mixed

reactors consistently showed a propionate buildup. Two possible explanations for this behaviour were

proposed, physical disruption of the syntrophic relationship between hydrolytic/acidogenic bacteria

and methanogenic archaea, or overloading the methanogens with high concentrations of hydrolysis

and fermentation products, resulting from continuous mixing.

Inhibition of methanogens at a low pH is always preceded by a sharp rise in VFA concentration.

However, if the system is well buffered, the pH remains high and the VFAs are consumed. The best

indicator of impending reactor failure is the alkalinity ratio, used in one form or another by most of the

researchers cited in this section. It is a titrimetric method and there are several versions; the most

commonly used appears to be that of Ripley which, for reasons explained in Section 6.4, I adopted for

use in this study.

23

Ammonia Accumulation Food waste has a relatively high nitrogen content, which originates from leafy vegetables and the

proteins in meats and leads to a low C:N ratio. If the amount of nitrogen present in the digester is

greater than the metabolic requirements of the microbial community for sustenance and growth, the

excess will form ammonia which is inhibitory to methanogens, particularly in the undissociated NH3

form present at higher pH; Kayhanian (1999) reported inhibition at total ammonia concentrations of

1200mg.L-1 with a free NH3 concentration of 45 mg.L-1, at pH7.2. This condition generally occurs when

the C:N ratio is <15; for safety it should be >20 (Igoni et al., 2008; Wu et al., 2010; Yadvika, 2004). For

example, in a side-by-side experiment by Zhang et al. (2012a), mr-OFMSW with a C:N ratio of 25:1

remained stable, but SS-OFMSW with a C:N ratio of 14:1 failed. There are only three practical ways of

maintaining digester stability in this situation; ensuring that the C:N ratio is always high enough by

adding a carbon source, diluting the contents of the reactor with water if ammonia levels rise to

inhibitory concentrations (Kayhanian, 1999), and operating with a long SRT (Bhattacharya & Parkin,

1989; Chen et al., 2008). As we shall see later, the C:N ratio in my experiments never fell below 45:1

and was usually much higher; as a result the reactor never experienced ammonia inhibition.

Summary (Food waste) Food waste must be analyzed thoroughly and its properties defined well; definitions of

convenience cannot be relied upon. Rapid initial digestion and surges in VFA concentration must be expected along with the potential

for inhibition; this can be managed by allowing longer HRT and SRT and by including a high rate reactor.

pH and alkalinity are indicators of incipient instability or inhibition. Ammonia inhibition is a distinct possibility when digesting FW alone; stability is best achieved by

maintaining the C:N ratio above 20 to 1 and operating with a long SRT. Food waste digestibility will be between 60% and 80% depending on its make-up, the content of

inorganics, and proportion of lignocellulosic materials (see Section 4.4 for further discussion). A digester that is operable over a wide range of HRTs and SRT’s has greater operating flexibility. Mixing and stirring should be kept to a practical minimum.

4.2.2 Lignocellulosic Waste

Data on composition and properties of typical fibres are presented in summary tables derived from

several sources. Table 4.3 shows the elemental analysis for fine paper (FP), newsprint (NP) and

cardboard (CB); these are the data from several publications, abstracted from Zhou et al. (2016) and

averaged.

24

Table 4.4 shows a typical example of the physicochemical properties of the same three fibres. All of

these materials consist of a combination of cellulose, hemicellulose and lignin, not all of which is

convertible to biogas. Lignin itself is very

resistant to digestion, but it is also bound to the

cellulose and hemicellulose, protecting them

from attack by microbes and enzymes; as a

result, cellulose and hemicellulose hydrolyze

slowly. All of these lignocellulosic fibres are derived from wood pulp, and the pulping process has a

direct influence on their

biodegradability and the methane yield

which they generate. Newsprint (NP) is

made from mechanically ground wood

without the addition of chemicals to

remove lignin. Both cardboard (CB) and

boxboard (BB) are made from

unbleached Kraft paper. The Kraft

process involves chemical treatment to partially remove lignin. Fine paper (FP) is made from bleached

Kraft paper and has virtually no lignin. The biodegradability and methane yields of these materials,

obtained from three publications, are presented in Table 4.5. Although the data are incomplete, and

the absolute values for biodegradability and methane yield vary from one source to another, all show

the same descending order of biodegradability, FP > CB > NP; this ranking may be related to the pulping

process used in their manufacture, and their resultant biochemical makeup.

Abstracted from (Buffiere et al., 2008b; Eleazer et al., 1997; Owens & Chynoweth, 1993; Yuan et al., 2012; Yuan et al., 2014) ND - not determined

Pommier (2010) created a blend of lignocellulosic fibres, consisting of a mixture of office paper,

newspaper, corrugated carton, cardboard and magazine stock, to simulate MSW. The physical

Table 4.3 Elemental Composition of 3 Lignocellulosic Wastes

Substrates C H O N C:N Ratio

Fine paper 44.76 5.29 49.81 0.10 448

Newspaper 46.70 5.23 48.03 0.17 275

Cardboard 47.31 6.15 46.00 0.33 143

Abstracted from (Zhou et al., 2014)

Table 4.4 Physicochemical Properties of 3 Lignocellulosic Wastes

Parameter Fine Paper (FP) Newsprint (NP) Cardboard (CB)

TS (%) 95.3 93.2 95.4

VS (% TS) 98.6 96.1 87.2

Ash (% TS) 1.4 3.9 12.8

Lignin (% TS) 1.4 23.4 17.8

Cellulose (% TS) 84.9 68.5 56.9

Hemicellulose (% TS) 12.3 13.1 10.7

COD g/g TS 1.07 1.21 1.10

Adapted from Yuan et al. (2012)

4.5 Biodegradability and Methane Yield Data of 3 Lignocellulosic Substrates

Eleazer Owens and

Chynoweth

Buffiere Yuan

Substrates % VS

destr.

CH4

mL/gVS

% VS

destr.

CH4

mL/gVS

% VS

destr.

CH4

mL/gVS

% VS

destr.

CH4

mL/gVS

Fine paper (FP) 54.6 288 ND 369 67.9 351.4 ND 208

Cardboard (CB) 54.4 155 ND 278 50.1 239.8 ND 96

Newsprint (NP) 31.1 74 ND 100 ND ND ND 75

25

properties of each component were measured and biodegradability determined. Also, specific biogas

production was measured for the model substrate. The results are summarized in Table 4.6. Pommier

(2010) and Buffiere (2008b) derived inverse linear relationships between lignin content and

biodegradability, and Eleazer (1997) a direct linear relationship between content of cellulose +

hemicellulose and methane yield; two sides of the same coin. At 0.53, 0.66 and 0.49 respectively their

regression coefficients are not impressive but the trends are consistent and the rankings of

biodegradability always in the same descending order FP > CB > NP with destruction efficiencies from

20% (NP, Table 4.6) and 68% (FP, Table 4.5).

Table 4.6 Physical Properties and Biodegradability of 6 Paper and Cardboard Products

Substrates % In mixture VS (%) COD (g.g-1TS) % VS destr. Biogas mL.g-1VS

Office paper 9 79.6 1.07 58 ND

Newspaper 24 91.6 1.22 20 ND

Corrugated Carton 21 83.2 1.08 45 ND

Cardboard 13 93.7 1.12 54 ND

Magazine 32 72.7 0.82 28 ND

Model substrate 100 82.6 1.03 43 141

Adapted from (Pommier et al., 2010) ND – not determined

In certain types of solid state anaerobic digestion, structural material, such as woodchips or shredded

wood, is added to food waste as a bulking agent to maintain permeability and porosity within the

digester as the waste decomposes. As a lignocellulosic waste, its biochemical and physical properties

must be understood because some degree of digestion of the bulking agent must be expected.

Published information on the anaerobic biodegradability of wood and wood chips is very limited. In

fact, the biodegradability has to be inferred from data on similar materials, such as tree and shrub

branches found in leaf-and-yard waste. Eleazer (1997) measured 28% VS destruction of coarsely

shredded branches in a 600 day BMP test. Owens & Chynoweth (1993), in a 75 day experiment,

determined the BMP of branches ground to <1.53 mm to be 134 mL CH4.gVS-1added, slightly higher than

that of NP. Trzcinski & Stuckey (2012) found the BMP of garden waste to be 114 mL CH4.gVS-1added in a

240 day experiment. A bulking agent consisting of shredded ash wood would thus be expected to make

some contribution to the biogas produced, but probably with a lower yield than reported for garden

waste.

In a review paper on the solid-state anaerobic digestion of lignocellulosic biomass, Ge et al. (2016)

reported that the use of inoculum containing a high population of hydrolytic bacteria improved the

26

degradation of cellulosic and hemicellulosic substrates. Xu et al. (2013) identified dairy waste effluents

in particular as being the most beneficial in this respect.

Summary (Lignocellulosic wastes) Expect slow rates of digestion and low yields and, as a consequence, a long SRT. Neither VFAs nor ammonia are expected to be present in sufficient quantities to cause digester

instability. There is no published information on the properties of coated cardboard, referred to in North

America as boxboard, or its biodegradability. An educated guess would suggest that BB would behave similarly to CB but with lower yields because of the coatings.

Fibre biodegradability is ranked FP > CB > NP. Digestibility ranges from 30% to 70% depending on the fibre; the main determinant is lignin

content. Hydrolytic bacteria in the inoculum are beneficial for the hydrolysis of lignocellulosic wastes. The C:N ratio of any combination of fibres will be very high, but the addition of FW will lower

the ratio in proportion to the amount of food waste added. Substrates should be analyzed to permit calculation of BMPtheor. for both reactor design

purposes and performance measurement. 4.2.3 Co-digestion

Co-digestion is the combined digestion of two or more substrates for the purpose of improving digester

performance, defined in terms of yield, rate and stability. Depending upon the substrates chosen and

reactor conditions, co-digestion can improve all three. In a review paper, Mata-Alvarez et al. (2011)

highlight the two most active areas of co-digestion research, based on an analysis of the number of

publications in the literature; they are sewage sludge plus organic fraction of municipal solid waste

(OFMSW), followed by co-digestion in the agricultural sector – farm wastes and energy crops. Among

the citations are eighteen references to co-digestion of OFMSW and sewage sludge with other

substrates, only one of which, pulp and paper mill sludge, is a cellulosic material (Zhang et al., 2008). In

a more recent review paper, Esposito et al. (2012) cite numerous examples of substrate combinations

but, once again, only one involves co-digestion of OFMSW with cellulosic waste - in this case biosludge -

in a paper by PoggiVaraldo et al. (1997) and the reported results are of limited value to this study.

Thus the situation respecting the co-digestion of the two major components of MSW, food waste and

cellulosic fibres, is rather strange. Usually they are generated together, sometimes they are separated

prior to treatment or disposal, but more often they are collected and processed or disposed of

together. Yet research on their co-digestion is basically nonexistent. The only significant exceptions are

studies of anaerobic decomposition in simulated landfills (Eleazer et al., 1997; Ledakowicz & Kaczorek,

2004; Yazdani, 2010), but these are of limited help because the experiments were not designed to

27

assess the contributions of the individual co-substrates, or their respective influence on one another in

the digestion process. Finally, from a commercial perspective, co-digestion has achieved limited

success. De Baere (2013) reported that only 11% of installed anaerobic digestion capacity in Europe

employs some form of co-digestion, most of which is concentrated in the agri-industrial sector.

Summary (Co-digestion) Interest in the co-digestion of organic wastes has grown significantly in recent years, as

evidenced by the increase in the number of research publications. OFMSW and sewage sludge are two of the most popular substrates for co-digestion. Paper and cardboard and other wood-pulp-derived substrates have received little or no

attention in this respect. I can find no publications examining the co-digestion of FW and FB for any purpose other than

improving stability of FW digestion.

4.3 Feedstock Preparation

Given the physical properties of organic solid waste, some form of feedstock preparation is generally

required prior to anaerobic digestion. It can involve size reduction, water addition, the use of a bulking

agent, or a pretreatment step (or some combination of these) to improve the biodegradability of the

substrates. Each of these is discussed in what follows.

4.3.1 Size Reduction in SS-AD

There are three principal reasons for size-reducing feedstock in SS-AD; to improve biodegradability

(both rate and yield), to reduce channeling in leach bed reactors, and to improve uniformity for

analytical purposes. Chynoweth et al. (1993) measured the Biochemical Methane Potential (BMP) of

MSW which had been ground to various particle sizes, and found that the yield changed little, but that

the rate of digestion of the most finely ground MSW increased by about 50%. Palmowski & Muller

(2000) determined that the effect of size reduction depended on the substrate; vegetables and meat

showed no change, while lignocellulosic materials - sunflower seeds, hay and leaves - generated 15 to

20% more biogas after comminution. Zhang & Banks (2013) found that size reduction of OFMSW made

no difference to yield but increased the rate of digestion; there was however a side effect, the more

finely ground material led to acidification and reactor failure. Motte et al. (2014) used wheat straw as a

model substrate for lignocellulosic biomass, ground to three different degrees of size reduction, fine,

medium and coarse with median particle sizes of 0.11 mm and 0.67 mm and 1.45 mm respectively. The

most finely ground material showed a substantial increase in solubility, but the BMP test underwent

rapid acidification and failure. The medium and coarsely ground materials behaved similarly to one

another in the BMP test, remaining stable throughout, though the coarsely ground material gave a 14%

28

higher yield. Chugh et al. (1999) performed large-scale (200 L) leach bed experiments on unsorted,

coarsely shredded, MSW (average particle size 10 cm) and also experienced acidification, particularly

from propionate, but maintained digester stability by flushing the VFAs to a second leach bed

containing stabilized waste. Pommier et al.(2010) compared the BMPs of two different samples of a

model substrate (see Table 4.6), one with an average particle size of 20 mm2 and the other <1 mm2. The

ultimate yields were almost identical, but the finer particles digested at a slightly higher rate; it was

concluded that size reduction was not worth the money and energy expended.

Summary (Size reduction) Size reduction can improve the biodegradability of lignocellulosic substrates, affecting rate

more so than yield, but the results are different for different substrates. Excessively fine material can undergo rapid hydrolysis, produce high VFA concentrations,

leading to methanogenic inhibition and failure. The size reduction of FW is not beneficial. Since size reduction is energy-intensive and costly, it is very hard to justify. Supports the principle of minimal pretreatment (Chapter 2); size reduction should be limited to

creating uniformity, rather than enhancing biodegradability.

4.3.2 Solids Content

The performance of a solid state anaerobic digester (SS-AD) - that is one that does not employ a stirred

tank reactor - is greatly influenced by the total solids content of the feedstock and the digestate.

Pommier et al. (2007) simulated the digestion of paper and cardboard under landfill-like conditions in a

400 day BMP experiment; the TS content ranged from 60% to 33 %; the latter representing landfill field

capacity. Water content was found to have a great influence on digestion; higher levels of moisture

(>65%) increased the microbial growth rate, the bioavailability of the substrate, and the yield.

Fernandez et al. (2008) digested OFMSW in 1.7L CSTRs at 20% and 30% total solids for 100 days under

mesophilic conditions. At 30% solids the start of methanogenesis was delayed by 36 days vs 15 days

and the methane yield was 21% lower. Le Hyaric et al. (2011) took a different approach by spiking

digestate with propionate and measuring the methane produced. The digestate came from a

commercial scale mesophilic digester processing residual solid waste (after removal of recyclables and

easily digested organics such as food waste); the experiments were conducted at 18%, 20%, 25% and

35% TS. Specific methane activity decreased almost fourfold over this range. Abbassi-Guendouz et al.

(2012) evaluated the effect of moisture content on the digestion of shredded cardboard (<2 mm) in a

298 day BMP experiment under mesophilic conditions. Methane yield declined slightly when TS

declined from 10% to 25%. At 30% TS two of the replicates performed similarly to 25% TS; the other

two replicates exhibited inhibition of methane production, as was the case for all four replicates at 35%

29

TS. This behaviour was attributed to mass transfer limitations, derived from Anaerobic Digestion Model

No. 1. Motte (2013) studied the dynamic effect of TS content, low substrate/inoculum ratio and particle

size on SS-AD, identifying optimum conditions as 20% TS with coarsely ground substrate (median

particle size 1.45 mm). Xu et al. (2014a) developed a mass diffusion based model of the effect of total

solids content on SS-AD, and applied it to their own experimental results and those of Abbassi-

Guendouz et al. (2012) and Motte (2013). In all cases maximum methane production was found

between 18 and 22% TS.

Summary (Solids content) Total solids content in SS-AD matters a great deal. It would appear that at lower TS hydrolysis becomes substrate limited, but above a certain TS

content mass transfer limitations start to inhibit hydrolysis and digester performance declines. The combined results of several research programmes, all designed and executed somewhat

differently, suggest an optimum TS content of between 18% and 30%, with a preponderance around 20%.

To the extent possible, a target of 20% TS should be used for the experimental programme.

4.3.3 Bulking Agent

Loss of porosity and permeability are a threat to long-term stability in any solid-state process (aerobic

or anaerobic) designed to decompose substrates containing food waste. For aerobic composting, FW is

always blended with shredded leaf and yard waste to preserve stability and enhance the process

(Pognani et al., 2012). For SS-AD it is equally important, though very little is published on the subject.

Xu et al.(2011) compared five different materials to optimize the hydrolysis of food waste in leach beds,

and concluded that woodchips, added at about a 5.3:1 ratio (substrateTS:BATS) was the most beneficial.

Murto (2013) conducted a side-by-side comparison of the digestion of OFMSW, with and without the

addition of bulking agent (BA). The experiment was carried out in a pair of two-stage anaerobic

digesters, each comprising a leach bed reactor and an anaerobic filter. The experiment was on a large

scale; both leach beds contained 3 tonnes of waste, one had 600 kg of BA added at a 1.6:1 ratio (TS/TS);

the BA consisted of chipped leaf and yard waste (-10 cm). The difference in performance between the

two was striking; the cumulative methane production from the leach bed with bulking agent climbed

steadily until the experiment was terminated at 74 days. The other system, in effect, failed at 55 days.

Although an inorganic bulking agent would simplify substrate biodegradability measurements, it is not a

practical choice at full scale. Leaf and yard waste is a practical choice operationally but its composition

varies significantly with the seasons, unnecessarily adding another variable to a controlled experiment.

30

A predictable, readily available, organic material of low biodegradability would be ideal; for example

ground ash wood is plentiful in southern Ontario because of the emerald ash borer infestation.

Summary (Bulking agent) Stable, long-term SS-AD requires that the waste mass remain porous and permeable. This is best accomplished by the addition of bulking agent. Waste to bulking agent addition rate ranging from 5.3:1 to 1.6:1 (substrateTS:BATS). Ground ash wood is preferred based on performance, consistency and availability.

4.3.4 Feedstock Pretreatment

Because one of the starting principles is to minimize pre-treatment to contain costs (Chapter 2), this

section concentrates on simple, inexpensive forma of pretreatment, and is thus limited in scope. In a

recent review paper on SS-AD, Ge et al. (2016) identify crop residues, OFMSW, forestry waste and

energy crops as major potential sources of renewable energy through anaerobic digestion. They

describe a wide range of pretreatment options and their applicability to 25 different lignocellulosic

substrates, and discuss various operational strategies for enhancing digester performance. Of the

pretreatment methods reviewed - dilute acid, steam explosion, alkaline treatment, hot water

treatment, wet oxidation and biological treatment, the observed improvements in digestibility varied

considerably. For example, alkaline pretreatment was more effective than steam explosion or hot

water in improving methane yield from wheat straw. But the authors concluded that comparisons

between treatment methods could not be made reliably, and that more research was required. Given

the number of substrates studied, and the specificity of the treatments employed, it is impossible to

determine what effects these forms of pretreatment might have on the digestibility of organic solid

waste. For reasons of practicality and economics, only the biological option, specifically the treatment

with hydrolytic microbes, might be beneficial if it can be done inexpensively. For example, Yuan et al.

(2012) studied the microbial pretreatment of CB, FP and NP with a partially characterized consortium

(comprising Clostridium, Pseudoxanthomonas, Brevibacillus, and Bordetella) to enhance their

biodegradability. Following initial incubation of the consortium, the substrate was treated for 14 days

at 50°C. Table 4.7 summarizes the results, which show that both the rate and yield of methane

production were improved for all the substrates, and the improvements were substantial. If

pretreatment of this sort could be achieved with a naturally occurring consortium it is possible that the

improvements in biodegradability might be sufficient to justify an added process step.

Following this line of thought, it is important to note that FW and MSW already undergo a form of

pretreatment that is neither measured nor controlled. The elapsed time between waste generation in

the home or commercial building and the start of treatment, can be anywhere from 1 to 2 weeks, a

31

period during which the waste spends time under the kitchen sink, in the garage, at the curb, in the

truck where it is partially compacted, and on the tipping floor at its destination.

Table 4.7 Effect of Microbial Pretreatment on Lignocellulosic Fibres

Treated Untreated

Substrate CH4 mL.g-1VS (55 d) CH4 mL.g-1VS (12 d) CH4 mL.g-1VS (55 d) CH4 mL.g-1VS (12 d)

Fine paper 287 228 208 93

Cardboard 231 163 96 41

Newsprint 192 154 75 31

Reproduced from (Yuan et al., 2012)

What happens to it during this time depends greatly on the season of the year. In winter, once it is out

of the house, it is frozen and essentially inactive. In the heat of summer it begins to degrade

immediately and, by the time source separated FW reaches its destination, large quantities of leachate

have drained from the waste and into the storage tank beneath the truck body. To some unknown

degree, this will affect its biodegradability and probably its BMP. I have been unable to find any

published work that investigated this behaviour.

Summary (feedstock pre-treatment to enhance digestibility) Research into substrate pretreatment to enhance the digestion rates and yields has largely

been devoted to individual substrates such as crop residues as opposed to solid waste. A passive biological pretreatment step might be feasible for solid waste if it could be done

inexpensively. During the elapsed time between generation and treatment, organic waste will begin biological

degradation, the extent of which will depend upon prevailing temperatures. If this step were better understood it might be possible to harness it as an inexpensive form of

pretreatment.

4.4 Digester Performance

Beyond reviewing the literature on the anaerobic digestion of solid waste, it is possible to go one step

further and use the published data to calculate how a lab scale digester, processing a solid waste

mixture, might perform and use the result as a guide for experimental design. Because of its

completeness, the data provided in Table 4.1 (Zhang et al., 2012a) was used as the basis for the

calculations. Of the two waste streams compared in this work, mechanically recovered organic fraction

of municipal solid waste (mr-OFMSW), essentially a mixture of food waste and fibres of unknown

proportions, more closely resembles the composition of the feedstock contemplated for my research

programme. All the calculations were performed using the elemental analysis provided by Zhang,

converted to a VS basis.

32

Four calculations generated the key data to measure digester performance; the calculations are

presented in Fig. 4.2.

a) The stoichiometric formula of mr-OFMSW, from the elemental analysis, was found to be

C 29.4 H51.3 0 14.9N.

b) Stoichiometric calculations were performed using the appropriate organic half reactions, the

fraction of electrons to cell synthesis (fs) equal to 0.08 for low-yield anaerobic processes, and

C5H7O2N as the formula for new cells (Rittmann & McCarty, 2001). They showed that, at 70% VS

destruction, mr-OFMSW would produce 377 LCH4.kg-1VSadded. By comparison, Zhang’s measured

BMP was 364 LCH4.kg-1VSadded; this suggests that 70%VS destruction is a good working estimate.

c) The energy balance was calculated using a modified form of Dulong’s law; the modifications

were based upon an extensive analysis of the elemental composition and calorific value of

OFMSW (Komilis et al., 2012); this source was chosen specifically because the work

concentrated on substrates of interest. The energy content of the VS destroyed (16.00 MJ)

balanced with the energy content of the CH4 + biomass produced (15.66 MJ), to within 2%.

d) The rate calculations were based on a pseudo-parallel first-order kinetic model and rate

constants developed by Zhang (2012a):

𝑦 = 𝑦 1 − 𝑃𝑒 − (1 − 𝑃)e (4.1)

Where y = cumulative methane yield at time t, ym = ultimate methane yield, k1 and k2 are the first-

order rate constants for the proportion of readily and less-readily degradable material respectively, and

P is the proportion of readily degradable material. By using this expression it was possible to estimate

the short-term and longer-term biogas generation rate from mr-OFMSW. The rate calculations indicate

initial biogas production rates of 8L to 11 L.h-1.kg-1VSadded in the first hour, 5.8 to 8.2 L.h-1.kg-1VSadded in

the first 24h, and an ultimate yield of 343 to 349 L.kg-1VSadded after 42 days; in each case, the spread

represents the difference between P = 0.3 and P = 0.5; the higher rate corresponding to the higher

fraction of readily degradable waste. The calculated and experimental yields from published sources

are summarized in Table 4.8.

33

Table 4.8 Biochemical Methane Potential L CH4.kg-1VSadded

Calculated from equation (1)

P = proportion of readily degradable material Experimental Results

P = 0.3

mr-OFMSW

P = 0.5

mr-OFMSW

Zhang

mr-OFMSW

Zhang

ss-MSW

Bolzonella

ss-OFMSW

Abassi-Guendouz

Cardboard

Buffiere

Cardboard

343 349 344 445 400 176 240

a) Stoichiometry Based on data from Table 4.2.1 the stoichiometric equation for the digestion of mr-OFMSW is: 7.3C 29.4 H51.3 0 14.9N (656.5)+63H2O (18) = 4C5H7O2N (113)+115CH4 (16)+77CO2 (44) +3NH4

+ (18)+3HCO3- (61)

(MWT in parentheses) b) Methane Yield

i. Converting the stoichiometric equation to mass in g; 4792g C 29.4 H51.3 0 14.9N + 1134g H2O = 452g C5H7O2N + 1840g CH4 + 3388g CO2 + 54g NH4

+ + 182g HCO3-

ii. Converted to 1kg of VS basis gives: 1000 + 237 = 94.3 + 384 + 716 + 11.3 + 38.0

1237g ≃ 1244g

iii. CH4 = 16/22.412 L.g-1 → 384g/0.714 g.L-1 = 538L CH4 or, at 70% VS conversion, 377L CH4, iv. CO2 = 44/22.412 L.g-1 → 716g/1.96 g.L-1 = 365l CO2 or, at 70% VS conversion, 256L CO2. v. Methane content = 538/(538 + 365) x 100 = 59.6% CH4.

Theoretical yield = 538l CH4 and 903L biogas per kgVSadded At 70% VS conversion yield = 377L CH4 and 632L biogas per kgVSadded

c) Energy Balance Using modified Dulong formula (Komilis et al., 2012) Energy content:

i. mr-OFMSW = (341.9C+1272H-149.6O)/1000 = (341.9x53.6+1272x7.8-149.6x36.1)/1000 = 22.85MJ.kgVS-1 ii. Biomass = (341.9x53.0+1272x6.2-149.6x28.3)/1000 = 21.77MJ.kgVS-1

iii. CH4 = 37.78MJ.m-3 iv. Biogas @ 59.6% CH4 = 22.5 MJ.m-3 v. 632l biogas = 14.22MJ

vi. 94.3 x 0.7g biomass created = 1.44MJ vii. Total e) + f) = 15.66MJ

viii. 0.7kgVSdestroyed = 16.00MJ d) Methane Production Rate (from BMP)

𝑦 = 𝑦 1 − 𝑃𝑒 − (1 − 𝑃)e

ym = 364 L.kgVS-1added, P = 0.32 or 0.3, k1 = .77 and k2 = 0.05

Calculation performed at P = 0.3 and P = 0.5 and methane converted to biogas yield at 59.6% CH4 gives: i. 𝒚 = 7.8L biogas at 1 hour, 140L at 1 day and 343L at 42d (L.kgVS-1

added at P = 0.3) ii. 𝒚 = 11.2L biogas at 1 hour, 196L at 1 day and 349L at 42d (L.kgVS-1

added at P = 0.5)

Figure 4.2 Digester performance calculations from first principles. a) Stoichiometry of mr-OFMSW digestion; b) Theoretical methane yield at 70% VS conversion; c) Energy balance: COD destroyed vs. (CODout biomass + CODout CH4 ) d) Estimates of biogas production at 1h, 24h and 42d.

34

Summary (Digester performance) Based on published data and these calculations of methane yield, the experimental digester

would be expected to produce between 100 and 450 L CH4.kg-1VSadded, depending upon the mixture of substrates fed.

In a batch-fed system, the rate of biogas production will rise quickly, immediately after fresh waste is added, and then steadily decline in a manner determined by the fraction of readily-digested waste added.

The initial rate of biogas production could reach 11 L.h-1.kg-1VSadded and, making allowances for potential errors in this estimate, provision should be made to accommodate a peak biogas production rate of 15 L.h-1.kg-1VSadded.

The methane content of the biogas produced, depending on the specific mix of substrates, could reach 60%.

Approximately 65% of the VS fed to the reactor (349 L.kg-1VSadded/377 L.kg-1VSadded*70%) can be expected to be converted to biogas.

4.5 Aerobic Curing

The anaerobic digestate will not be biologically stable, may still contain pathogens, and may be

odorous; aerobic curing therefore is the necessary final step of the process (Brinton, 2006; Pognani et

al., 2012; Yazdani et al., 2012). Pognani et al. (2012) present a complete mass balance of the combined

anaerobic/anaerobic treatment plant for OFMSW, based on one year of operation of the DRANCO

technology. The digestate was mixed with wood chips and aerobically composted in a tunnel system for

one week, under forced air conditions. At the end of this phase the partly composted digestate was

placed in piles, which were turned twice per day for a further 1 to 2 weeks. The final compost product

was screened and the recovered bulking agent returned to the process; the end-product was sterile and

suitable for land application.

Yazdani (2012) describes an anaerobic/aerobic landfill-based digester, using 1.7 tonnes of yard waste as

the substrate, ultimately producing a finished compost. The process was carried out in three stages;

filling (93 days), anaerobic digestion (366 days) and aerobic curing (191 days). During the curing process

methane continued to be produced because not all the waste mass had turned aerobic, but most of the

methane was removed by the biofilter. At the end of the aerobic curing, the composted product was

recovered and tested and found suitable for land application; one of the recommendations for future

design and operation was to shorten the aerobic curing period.

Summary (Aerobic curing) Aerobic curing (composting) of digestate is required. Composting can be carried out in situ or ex situ or some combination of both. The planned use of bulking agent to aid anaerobic digestion will also assist in the aerobic

curing.

35

Recovery and reuse of bulking agent is feasible. A good quality compost requires clean feedstock. Composting is a logical complement to anaerobic digestion.

4.6 Conclusions and Knowledge Gaps

The findings of the literature review confirm that it should be possible, within limits yet-to-be-defined,

to digest mixed wastes comprising food waste plus lignocellulosic fibres in varying proportions. The

food waste portion will digest rapidly and the fibres less rapidly; surges in biogas from freshly-fed waste

(particularly food waste) are expected, and could rise as high as 11L.h-1.kg-1VSadded as the easily-digested

FW is converted; this will then subside as the more recalcitrant substrates are digested. Digestibility of

the waste will range from 30% to 80% but probably, on average, be around 65%, depending on the mix.

Management of the C:N ratio, alkalinity and pH will reduce the risk of process instability or inhibition.

Some size reduction of the substrates will be necessary to improve uniformity. Maintaining a high

moisture content within the waste mass, which must remain permeable, is critical; a total solids

content between 20% and 25% would appear to be a good place to start. A final composting step to

cure the digestate is necessary to achieve stability and sterilization and produce a usable product.

In the course of this review, several notable gaps in the literature were identified. First and foremost no

extensive studies were found that were specifically designed to test the co-digestion of food waste with

lignocellulosic fibres arising from the paper and cardboard products in the solid waste stream.

Secondly, no attempt has been made to study the stability and performance of an anaerobic digester

treating any waste streams of variable composition over an extended period of time. As a consequence,

a digester specifically intended for this purpose has not been built, and a suitable design concept has

yet to be defined. While management of the C:N ratio and digester pH and alkalinity are proven tools to

manage digester stability, their suitability in a system with variable feedstock composition is unknown.

Given the heterogeneous nature of the wastes to be studied, and the probable presence of foreign

matter, the preferred approach is solid state anaerobic digestion (SS-AD), whereby the waste remains

stationary and has leachate percolate through it. The system would take the form of a two-stage

anaerobic digester, comprising sequentially fed leach beds plus a UASB operating at mesophilic

temperatures, to be followed by an aerobic curing step. This arrangement would conform to the

concept of the BioPower process.

A combination of cardboard, boxboard, newsprint and fine paper (CB, BB, NP and FP) collectively fibre

(FB), plus food waste (FW), will serve as a suitable proxy for the organic fraction of commercial solid

waste. Bulking agent in the form of woodchips will be required to ensure continued permeability in the

36

leach beds. A portion of the FW will digest very rapidly; the balance will digest more slowly. The CB, BB,

NP and FP, and the BA will all be relatively slow to digest, depending upon their respective lignin

contents.

FW alone has a low C:N ratio and is always susceptible to process instabilities caused by high VFAs and

resultant low pH, or high ammonium, or alternate between these conditions. The presence of FB will

raise the C:N ratio and thereby stabilize the digester.

Co-digestion of two or more substrates has been shown to enhance digester performance; digestion

rates and yields can be increased and stability enhanced. FW has been widely studied as a co-substrate

but usually with waste-water sludges or crop residues. The co-digestion of FW with the fibres found in

solid waste, under conditions of varying FW addition, is essentially unexplored.

Some feedstock preparation will be necessary. Following the principles established in Section 2.1, this

will be kept to a minimum and consist primarily of pre-sorting the FW to remove undigestible

contaminants, plus some size reduction of the FB.

Digester performance, measured as the rate and yield of methane production, is expected to vary

considerably depending upon the level of FW addition. A range of 100 to 450 L CH4.kg-1VSadded is

expected. After fresh substrate has been introduced a rapid rise in biogas production, to perhaps

11 L.h-1.kg-1VSadded, is expected.

Digester stability can be achieved by monitoring and controlling the C:N ratio, pH, alkalinity ratio, VFA

concentration and ammonia concentration. The suitability of these methods is well-documented for a

digester that has only been fed with food waste of consistent composition. But their effectiveness for

monitoring and controlling the digestion of variable feedstocks is unknown.

The behaviour of the microbial community may be able to provide insight into the process, but how

that knowledge can be developed and applied in a practical sense, has yet to be determined.

Thus there remain several important knowledge gaps; questions which the literature cannot fully

answer, specifically:

1. How does the C:N ratio vary in response to the range of waste compositions contemplated by

this research and what does this imply for process performance and stability?

2. How coarse can the substrates be before channelling begins in the leach beds?

3. How does the preparation of the bulking agent affect digester performance?

37

4. What is the effect on process stability of short-term (week to week) and longer term (month to

month) changes in the proportions of FW and FB?

5. To what extent does the addition of FW enhance the digestibility of recalcitrant substrates such

as FB’s, compared to a baseline without FW?

6. What are the benefits of using sequentially-fed leach beds compared to other AD technologies;

can they be quantified?

7. Are pH and alkalinity, VFA concentration and C:N ratio adequate tools to control the AD process

in this application, and how can they be used to greatest advantage?

8. How tolerant is the UASB of the properties and variability of the leachate fed to it, and what are

the preferred operating conditions of OLR, HRT and Vup?

9. What proportion of the biogas is generated in the leach beds and how does the gas production

vary with time as the digestion process progresses?

10. How does the microbial population respond to changes in feedstock composition, does it

become more diverse and robust?

11. How effective is aerobic curing of the digestate; how long does it take, and what is the quality

of the product?

12. What are the key operating variables to be monitored and controlled, and what are their

preferred operating ranges in a digester processing mixed solid wastes?

13. In what ways and to what extent can the experimental results be used to design a pilot-scale or

full-scale digester?

These are the questions which this research programme sets out to investigate and answer, and since

part of the proposed research involves simulating the real-world variability of solid waste, devising a

reliable design and operating plan to control the process will be particularly important.

38

Chapter 5. Digester Design and Construction

In this Chapter a design concept is described with reference to the BioPower process, then refined,

major component by major component, with further reference to the literature. The final result is a

refined design concept ready for procurement and construction.

5.1 Design Basis

The relationships among biochemistry, process operating conditions, feedstock properties, and digester

performance are at the heart of the engineer’s challenge when designing, building, and operating an

anaerobic digester for solid waste. Simply put, the objectives are to achieve rapid substrate destruction

and methane formation, with the minimum amount of energy input, and without causing reactor

instability. This will be a constant theme throughout this work. The wastes that concern us in this

project are of two types, food waste and lignocellulosic fibres. The lignocellulosic fibres, namely

cardboard (CB), boxboard (BB), newsprint (NP), and fine paper (FP) - collectively called fibres (FB) are

derived from wood, the food waste comes from a green bin program. It was decided to use a bulking

agent (BA) in the form of woodchips to enhance the permeability and porosity of the substrates. For

convenience, the BA is included in this section as a substrate. The rationale and methodology for the

use of BA is covered in Section 4.3.3. The principal objective was to determine what is known about the

intrinsic properties of the chosen substrates, individually and collectively, and how they behave in the

anaerobic digestion process. One of the underlying principles of the BioPower process is the

employment of solid state anaerobic digestion (SS-AD) to avoid these complications and costs, so the

“wet” versus “dry” decision has, in effect, already been made.

Given the versatility and apparent advantages of two-stage digestion, particularly with more easily

digested waste (such as food waste), I chose two-stage digestion, using leach beds for the first stage

and a UASB for the second stage. The choice of a UASB over other methanogenic reactors, such as

anaerobic filters or CSTRs, was for reasons of practicality, simplicity and cost. The final design made it

possible to operate the digester with or without the UASB though, in practice, it remained in operation

throughout.

I also selected mesophilic temperatures, partly for practical reasons associated with heating a lab scale

reactor, but also because the literature did not provide strong arguments for choosing thermophilic. A

comparison of mesophilic and thermophilic digestion with substrates containing cellulose and

hemicellulose found that, while the reaction rate at thermophilic temperatures was greater, the

digester was prone to surges of VFAs resulting in reactor instability (Shi et al., 2013).

39

There was one additional constraint which affected the design of the experimental digester. Because

the system would be producing methane gas and trace amounts of hydrogen sulfide, it had to be

constructed in a walk-in fume hood. This imposed a practical upper limit on the volumetric capacity of

the system, and had a great influence on the ultimate layout.

Only certain aspects of reactor design are discussed in the literature. These are principally the capacity,

dimensions, and operating conditions of leach beds and of UASBs. In the initial stages of planning, I

received fortuitous assistance from Abdul-Sattar Nizami, a postdoctoral fellow in the Department of

Department of Chemical Engineering and Applied Chemistry. Nizami’s PhD research and associated

publications centred around the SS-AD of grass silage in a two-stage digester consisting of six

sequentially-fed leach beds and a UASB. In several discussions we reviewed his design, the logic behind

it, the operational difficulties he had encountered, and how he dealt with them; he provided me with

engineering drawings of his system. We also held a video conference call with James Browne who had

adapted Nizami’s digester to process source-separated food waste. Browne had run into numerous

difficulties; low yield (compared to the grass silage for which the system had been designed), a large

accumulation of ammonium ions, made worse by rising pH (ammonia being more inhibitory than

ammonium), an increasing recirculating load of inorganic salts, foaming in the UASB and an

accumulation of fines plugging the leach beds. The C:N ratio, at 14:1, was low and this certainly

contributed to the problems encountered. A review of the published papers by Browne & Murphy

(2013; 2014) describe some of these difficulties, but not in the graphic detail of our conference call.

I adopted the design of Nizami et al.(2011) as a template, to be altered and refined, as appropriate, to

suit my substrates, and the results of a further review of the literature dealing with other examples of

this kind of digester configuration.

5.1.1 Leach Beds

The purpose of the leach bed is to provide intimate contact between the solid substrates and the

leachate, which contains nutrients and the microbial community; typically the goal is to effect

hydrolysis of the substrates followed by acidogenesis in the leach bed. The design has to achieve three

things; even distribution of the leachate over the entire cross section, a long pathway for leachate

percolation, and the ability to retain as many of the fines as is practical, all of this to be accomplished

while minimizing wall effects.

The literature contains few examples of the use of multiple leach beds for the digestion of solid waste.

Chugh et al. (1999) used a combination of two leach beds operating in series, one with fresh waste, one

40

with stabilized waste, to accelerate the digestion of MSW; they achieved 75% degradation of the

rapidly biodegradable fraction within two months. In an earlier paper, O'Keefe et al. (1993) also

digesting MSW, operated three leach beds labelled “new”, “mature”, and “old”. The mature leach bed

recirculated leachate only within itself, while the new and old were interconnected and recirculated

leachate between them; 49.7% of VS was degraded in 42 days. To digest food waste, Han (2004)

operated four leach beds in parallel connected to a UASB, achieving 80.5% VS reduction in two days;

the experiment ran for 110 days. Using the same experimental system, Shin (2001) achieved 85% VS

reduction in 10 days in an experiment that ran for 150 days. In neither case did they experience any

process instability. Yet Browne et al., also digesting food waste alone in a similar system, ran into

serious instability problems; the difference would appear to be that Han and Shin diluted the leachate

in the system every 10 days to lower the ammonia concentration, while Browne did not. This further

highlights the challenges of digesting food waste by itself, as discussed in Section 4.2.1. A summary of

reactor volumes, biogas yields, and SRTs from leach bed digestion experiments performed on organic

waste and related substrates is presented in Table 5.1. The clear outlier of this group is the work of

Eleazer et al. (1997) whose aim was to study anaerobic decomposition of organic waste under

simulated landfill conditions, using well digested solid waste as the inoculum. Of the remaining studies,

Murto et al.’s (2013) substrate was the recalcitrant portion of OFMSW, containing mostly lignocellulosic

materials.

Based on all these data, I decided on an initial SRT of 42 days. Using the conceptual design of six leach

beds, each would have sufficient capacity to hold a mixture of saturated substrates weighing

approximately 6 kg (about 8.5 L), for a total of 36 kg wet weight at about 30% TS and an aggregate

working volume of about 50 L. This arrangement would require the removal and replacement of one

leach bed per week, which I thought operationally manageable. It also meant that each charge of fresh

waste would represent only 16% of the total mass within the digester, the balance providing

operational buffering, presumably enhancing digester stability. Also, with careful design of the piping

and valves, operation with fewer leach beds would be feasible, if this were necessary or desirable.

To meet the objective of evenly distributing the leachate over the cross-section of a leach bed, a

delivery system was required. The only information I could find in the literature was the description of

gravity-drained leachate cups (Nizami et al.,(2011).

41

Table 5.1 Leach Bed Digestion - Performance Data Substrate Reactor Leach bed

working capacity (L)

SRT (d)

Biogas yield L CH4.kgVS-1

added Comments Source

Maize Leach bed + UASB

4.0 28 540 (Cysneiros et al., 2012; Cysneiros et al., 2011)

Food waste Leach beds + UASB

17.0 24 340 (Browne et al., 2013)

Dry fraction OFMSW

Leach bed + UASB

5300 74 233a Mostly lignocellulo

sic

(Murto et al., 2013)

Food waste Leach beds + UASB

35.0 10 270 (Han et al., 2002)

MSW Multiple Leach beds

400b 35 270 Bekon system

(Qian et al., 2016)

Grass silage Leach beds + UASB

17.0 30 - 42

341 optimized (Nizami & Murphy, 2011)

Food waste Leach bed 2.0 180 320 Landfill simulation

(Eleazer et al., 1997) Newsprint Leach bed 2.0 430 75.4 Cardboard Leach bed 2.0 470 155 Fine paper Leach bed 2.0 680 288 MSW Leach bed +

UASB 14.0 20 85 Landfill

simulation (O'Keefe &

Chynoweth, 2000) aestimated from kgTS-1added. bm3.

When these are full, a level sensor activates a solenoid valve causing them to drain into the leach bed

through a sprinkler system. This arrangement was plagued with problems, mostly caused by the

presence of suspended particulate matter that both fouled the sensor and partially plugged the

sprinkler system. Direct injection appeared to offer greater reliability than gravity delivery; a closed

loop, to constantly circulate leachate through a manifold above the leach beds and back to the tank,

was designed. Pneumatic solenoid valves, opening automatically on a signal from a programmable logic

controller (PLC), would redirect the flow to each leach bed in turn. Thus the potential for plugging

would be greatly reduced and constant volumes of leachate always delivered. A method to ensure the

even distribution of leachate required some thought and experimentation. All methods employing any

form of nozzle or orifice with leachate under pressure, would operate differently at different flowrates

and be subject to plugging and failure. Relying on drainage through a plate with holes in it would be

subject to plugging, and also to uneven settlement of the substrate, resulting in maldistribution of the

leachate. The method chosen uses a 4mm thick non-woven geotextile (TenCate Mirafi 1600N). In a

simple bench test, 500 mL per minute of water was delivered onto a 15 cm geotextile disc; the water

was found to be reliably distributed to the perimeter before completely draining through. Once fresh

waste was loaded into a leach bed, a geotextile disc was placed on top covering about 80% of the cross-

sectional area. This technique was used throughout and worked very well.

42

Suspended fine particles are present in the leachate drained from the leach beds but also in the

leachate delivered. Various forms of filtration media have been used in previous work; stainless steel

mesh, (Chugh et al., 1999; Cysneiros et al., 2012; Nizami et al., 2011) perforated plate (Jagadabhi et al.,

2011; O'Keefe & Chynoweth, 2000), and gravel layer (Xu et al., 2014b). I chose a woven geotextile

(TenCate Mirafi 400) resting on a No. 6 mesh stainless steel support (for strength). This arrangement

proved very simple to use and easy to clean. The choice of geotextile mesh size is a compromise

between maintaining flow and allowing some fines to pass on the one hand, and prevent plugging on

the other; this arrangement never plugged.

The performance of the leach beds depends on the volume and frequency of leachate recirculation, in

relation to the mass and volume of substrate present. The objective is to achieve intimate contact

between the leachate, which carries with it the microbial consortium and enzymes, and the substrate.

The importance of maintaining a sufficiently high moisture content was discussed in Section 4.3.2, and

this is directly dependent upon leachate recirculation. In some published examples recirculation is

continuous (Yesil et al., 2014), but more often it is intermittent (Murto et al., 2013; Nizami et al., 2011).

In reviewing the published data I calculated recirculation rates L.d-1.Lwaste and L.d-1kg.VS-1added. The

results are summarized in Table 5.2. Example 4 is of particular interest; firstly the digester is similar in

Table 5.2 Leachate Recirculation Rates and Methane Yields

Substrate Recirculation Rate Days Methane Efficiency Source

L.d-1.l-1waste L.d-1.kg-1VS 60 L CH4.kg-1VSadded

% VS

Destr.

1 MSW 0.11 0.43 76 180 (Chugh et al., 1999)

2 Dry fraction of

OFMSW 4.5 20.0 15 98 (Murto et al., 2013)

3a OFMSW 1.6 13.1 30b n.a. (Yesil et al., 2014)

4 Grass silage

1.0 17.5 30b 310 (Nizami & Murphy,

2011) 5.8 120 16 341

5 Food waste 0.43 5.4 17 234 (Xu et al., 2014c)

6

Grass silage

8.8 83

24 - 32

66.3

(Xie et al., 2012)

5.9 67 64.8

4.4 41 62.1

a continuous recirculation – others intermittent b sequentially fed leach beds – experiment ran for 60d.

configuration to my proposed arrangement, secondly, the results show only a 10% improvement in

methane yield in exchange for a sevenfold increase in recirculation rate; this hardly seems like a good

43

bargain; it is operationally impractical in the lab and financially burdensome on a commercial scale. In

Example 6, also a study of the effects of leachate recirculation rate on the hydrolysis of grass silage, the

recirculation rates are similar to those of Example 4, but also show that a doubling of the recirculation

rate yields only a 7% increase in VS destruction. Example 2 is from a large-scale experiment (3 tonnes of

substrate) using wood chips as a bulking agent. In general, higher recirculation rates seem to produce a

higher methane yield, and greater VS destruction. The review of solids content (Section 4.3.2) showed

that when the moisture content is too high, digester performance deteriorates. There is also the

practical question of how the actual substrate used in the proposed experiments would behave; for

example how much moisture would a leach bed of solid waste hold, and how quickly would it drain? I

decided to design the system with a lot of flexibility and provide for a recirculation rate ranging from

5 to 20 L.d-1.kgVS-1added. The frequency and duration of leachate recirculation would be determined

experimentally.

Summary (Leach bed design) the digester will have six leach beds (plus a seventh to simplify removal and replacement) each leach bed will have a working volume of about 8.5 L. the waste will be supported on a No. 6 mesh stainless steel screen mounted in the bottom. the loss of fines will be minimized by a woven geotextile filter sitting on the screen. a nonwoven geotextile will be placed on top of the waste to distribute the incoming leachate

over the cross-section of the leach bed. initial leachate recirculation rate will be between 5 and 20 L.d-1.kgVS-1

added. the frequency and duration of leachate delivery to the leach beds will be determined

experimentally. the question of whether or not to provide external heat for the leach beds, and if so how,

remained unresolved at this point.

5.1.2 Upflow Anaerobic Sludge blanket (UASB) Reactor

Mechanically, the UASB reactor is a simple piece of equipment with no moving parts (other than a feed

pump). It is capable of converting soluble COD at a high organic loading rate (OLR), and does not

require a lot of space. It does require supplementary heating and thermostatic control to maintain a

steady operating temperature of about 37°C. The logic behind the choice of OLR, HRT and Vup is

thoroughly explained in the literature (Campos & Anderson, 1992; Goncalves et al., 1994; Lettinga &

Pol, 1991); all of them are older publications from the early days of UASB development. More

specifically, a review paper on the application of UASBs for waste water treatment by Latif et al. (2011)

provides a summary of published information on the design and operation of UASBs in many different

applications. OLRs range from 0.77 to 30 gCOD.L-1.d-1, HRTs from 2.4 to 480 h and upflow velocities (Vup)

from 0.0018 to 2.0 m.h-1. These are very wide ranges, but by eliminating designs which lacked

44

supporting data, or did not work well, and by examining the specific applications, it was possible to

narrow the ranges substantially; OLRs from 3 to 14 gCOD.L-1.d-1, HRTs from 5.0 to 48 h and Vup from 0.1

to 0.6 m.h-1. These were adopted as the design operating ranges. For comparison purposes,

experiments with grass silage were conducted using an OLR of 6 gCOD.L-1.d-1, an HRT of 7.5 h, and a Vup

of 0.1 m.h-1 (Nizami et al., 2011).

Table 5.3 summarizes the most useful published information on the capacity and aspect ratios of leach

beds and UASBs, and the volumetric relationship between them. Reactor dimensions vary widely and

analyzing this information is made difficult by the complete lack of explanation in the literature on how

the chosen dimensions were arrived at. Though one thing is certain; Jagadabhi et al.’s (2011) UASB was

greatly undersized for FW digestion; it was unable to cope with the OLR, and had to be uncoupled from

the LB. For my proposed digester, comprising six leach beds of 50 L working volume in total, using a

Table 5.3 Leach Bed and UASB Dimensions

Substrates Leach beds UASB Vol. ratio Source

d(cm) h(cm) v(L) # h/d d(cm) h(cm) v(L) h/d LB/UASB

1 MSW 20 44 14 1 2.2 15 29 5 1.93 2.8 (O'Keefe & Chynoweth, 2000)

2 MSW 200 2 (Lai et al., 2001)

3 ss-FW 4.5 1 3 1.5 (Wang et al., 2003)

4 Grass silage 38 15 16 6 0.4 23 75 31 3.3 3.1 (Nizami et al., 2011; Xu et al., 2011)

5 FW 4.6 1 10 1.8

(Jagadabhi et al., 2011; Xu et al., 2011)

6 Crop

residue and FW

10 1 0.5 1.0 20.0 (Jagadabhi et al., 2011)

7 Maize 13 30 3.5 1 2.3 6.2 50 1.5 8.0 2.7

(Cysneiros et al., 2012; Cysneiros et al., 2011)

8 Grass silage 9.2 30 2.0 1 3.3 (Xie et al., 2012)

9 Landfill leachate

15.6 26 5 1.67 (Kennedy & Lentz, 2000)

10 FW (cafeteria)

19 38 8.0 4 2.0 *20 and *32

*80 and *20

41 1.9 0.8 (Shin et al., 2001)

11 FW + FB 20 26 8.5 6 1.3 25 55 27 2.2 1.9 This study

*lower and upper sections of different diameters

45

LB:UASB volumetric ratio of 0.8 (Example 10) gives a UASB volume of 63 L, and using a LB:UASB

volumetric ratio of 3.1 (example 4) gives a UASB volume of 16 L. The aspect ratio (h:d) of UASB reactors

shows a range of 1.7:1 to 8:1. Meyer (2013) used 6.8:1 with a working capacity of 5.4 litres, and Nizami

et al. (2011) used 3:1 with a working capacity of 31.4 L.; the most relevant data found in the literature

are shown in Table 5.3. It is possible that the specific dimensions are much less important than the

combination of OLR, HRT and Vup. Accordingly, the UASB was designed and built in two sections of

unequal length, so that it could be operated either as a single combined unit, or as one of two smaller

units, with approximate capacities of 20 L, 30 L or 60 L. This combination allowed operation at a

LB/UASB ratio from 0.8 to 3.1, covering the entire range in Table 5.3 (with the exception of Example 6

which was included as an example of a poor design). Furthermore, by selecting the appropriate

peristaltic pump and tubing, the HRT and Vup could be set anywhere within the preferred ranges.

To operate the digester as a closed system, the UASB effluent must be discharged to the storage

tank(s). Table 5.3, Example 11, shows the dimensions finally used.

Summary (UASB design) the UASB will be constructed in two sections of unequal length with a maximum combined

capacity of approximately 60 L. it will be operable at capacities of about 20 L, 30 L or 60 L depending on the configuration

chosen. a peristaltic pump and tubing will be chosen to achieve an HRT from 5.0 to 48 h and a Vup from

0.1 to 0.6 m.h-1. leachate inflow will be distributed with a No. 6 mesh stainless steel screen mounted the

bottom of the UASB. the UASB will be supplied with external heat and thermostatic control.

5.1.3 Leachate Tanks

Leachate storage in a heated, thermostatically controlled tank is required to;

supply leachate and heat, to the leach beds.

receive the leachate drained from the leach beds and

supply leachate to the UASB.

maintain hydraulic and thermal balance.

The required capacity for this storage is a function of the volume of substrate in the leach beds, the

recirculation rate, the leachate retention capacity of the waste, and the required flow rate to the UASB.

In particular, intermittent delivery of leachate to the leach beds draws down the contents of the

storage tank, and is only replenished when the leachate percolates through the waste and returns to

the tank; during this period the UASB must not be starved of flow. There is very little information in the

46

literature to guide the decision on tank capacity. Nizami et al (2011) operated with a 40 L tank,

representing 31% of the combined volume of a 31 L UASB and 96 L of waste in the leach beds. Murto et

al. (2013) operated with a 700 L tank representing 9% of the combined volume of a 2.6 m³ anaerobic

reactor and 5.3 m³ of waste in the leach bed. Based on the available information, I concluded that

leachate storage capacity of about 40 L would be required. The question of whether this should be

provided using one or more tanks was unresolved at this point. There is one other factor to be

considered; a UASB works very well for soluble COD, but less well for suspended COD (Lettinga & Pol,

1991). Suspended matter is certain to be present in the storage tanks, and it would be desirable to

install a weir to restrict the flow of particulate matter to the UASB, but then what steps would be

needed to prevent sludge accumulation in the tank(s)?

Summary (Leachate tanks) leachate storage of about 40 L is required. it could be a single tank or perhaps two tanks. the tank(s) must be heated and thermostatically regulated. it would be advantageous to install a weir or clarifier in the storage tank. some mechanism for preventing sludge accumulation would then be required.

5.1.4 Biogas Production

Nizami et al. (2010) only measured biogas from the UASB, not from the leach beds, apparently because

virtually all the biogas was produced in the UASB. Accordingly, the original intention was to install a

single gas meter at the outlet from the UASB and connect in the headspace from the tank(s) and the

leach beds to capture any biogas that they might contribute, and to maintain hydraulic balance.

However, a further review of the limited information available in the literature suggested that the

principal source of the biogas would depend upon digester conditions, principally pH, and might in fact

come primarily from the leach beds. Murto et al. (2004) digesting mr-OFMSW ran two experiments,

each for 74 days, each with a combined single leach bed and methane reactor; bulking agent was added

to the first, but not to the second. With bulking agent added, 75% of the biogas came from the leach

bed; without bulking agent 23% came from the leach bed and the digester produced 43% less biogas in

total. Shin et al. (2001) digesting food waste, acknowledged that biogas is produced in the leach beds,

and made an aggregate measurement of the total biogas produced. Chugh et al. (1999), digesting

MSW, operated two leach beds in series, one of fresh waste and the other of stabilized waste. Fresh

waste leachate had a starting pH of 4.2, but 10 days of leachate recirculation through stabilized waste

caused the pH to rise to 6.5. At this point the fresh waste leach bed was uncoupled from the stabilized

waste leach bed, and leachate was recirculated from and to the fresh waste leach bed. During the initial

10-day period the majority of the biogas was generated from the stabilized waste. Thereafter it was

47

generated from the fresh waste. It appeared that the source of biogas generation might vary with

experimental conditions so the decision was made to devote one gas meter to the UASB and install a

second to measure gas from the balance of the system, otherwise important information about the

behaviour of the digestion process was going to be permanently lost.

Summary (Biogas measurement) biogas can be generated in any part of the system. the source depends upon prevailing digester conditions. it probably also depends upon the substrates. two gas meters are required to avoid loss of valuable data.

5.1.5 Aerobic Curing

The practical value of aerobic curing as part of a commercial operation has been recognized (Brinton,

2006). One example of combined anaerobic and aerobic processing, at a semi-commercial scale, used

an anaerobic digestion period of 366 days, followed by an aerobic curing period of 191 days to treat

yard waste (Yazdani et al., 2012).

The original plan for the experimental programme was to cure the digestate in the seventh, separate,

leach bed with an initial SRT of one week, as suggested by Pognani et al. (2012). In practice the quantity

of waste was found to be too small and heat losses too great to do this successfully. To solve this, all

the digestate retrieved over the course of the experiment (about 300 kg) was stored at 4°C then

aerobically cured in a single experiment conducted at Miller Waste Systems, using an aerated static pile

system designed and built for the purpose (Chapter 9).

5.2 Design and Construction

This section describes in detail the design and construction of the experimental anaerobic digester,

which was given the sobriquet Daisy, after Daisy the Cow (Jones, 2013). The original Daisy is a stuffed

specimen of a breed of long-horned Gloucester cattle, and is currently resident in the Gloucester Folk

Museum. She comes from the same breed used by Edward Jenner to develop the smallpox vaccine in

1796. Daisy was stuffed in 1935, by the same firm of taxidermists used by Charles Darwin for his

specimens. Her pedigree is impeccable.

The design of every component in Daisy was an iterative process; most were sketched in several

versions, and many factors had to be taken into consideration before final decisions were made. Only

the most important steps of this process are included in what follows.

48

5.2.1 Guiding Principles

The guiding criteria for the design are as follows

safe operation under all circumstances

flexibility and versatility of operation

ease of feeding

the capacity to hold as much waste as physical constraints allow

24 hour operation, seven days per week, most of the time unattended

automated only to the extent necessary for safe continuous operation

ability to handle the surges of biogas expected after fresh waste is added

The following principal topics are covered:

leach beds: number, volume, dimensions

leachate delivery system and volumes

leachate storage: number size and configuration of storage tanks

UASB: volume, dimensions

pumps: type, number and capacity

heaters: capacity and type

control system and datalogging

biogas measurement

5.2.2 Safety

Safety considerations were always the first priority in the design, construction and operation of Daisy. A

risk analysis was conducted prior to construction, and is attached in Appendix B. The maximum

concentrations of methane and hydrogen sulfide that would be produced were estimated using

published data and conservative assumptions. The only risk identified which could not be engineered

around was a general power outage or the potential failure of the fume hood fan, which could result in

elevated concentrations of CH4 or H2S. In either circumstance several hours would elapse before

hazardous conditions were reached, based on the conservative assumption that there would be no

natural draft in the fume hood. Draft tests were performed with the fume hood fan switched off, using

dry ice; these showed that there was sufficient natural draft to maintain in-flow to the fume hood in

the event of a fan failure or power outage. For greater security an H2S monitor was installed in the lab.

49

A spill containment tray was designed, and constructed from polycarbonate plastic, to cover the entire

floor of the walk-in fume hood; its capacity was calculated at 110% of the total volumetric capacity of

the system. The tubing in the peristaltic pumps was repositioned at the three week mark, and replaced

every six weeks, throughout the operating period. Electrical equipment was located outside the fume

hood to the greatest extent possible. Electrical connections within the fume hood were all taped and

shielded from splashed liquids. Hot electrical terminals in the control panel were also shielded to

prevent accidental electrical shocks.

The datalogger was set up so that it could be monitored remotely through an Internet connection, and

this I did at least twice daily for the entire operating period. On four or five occasions, when

information from the data logger indicated process malfunction, I was able to drive to the University

and restore normal operation within an hour. A WebCam was mounted inside the fume hood pointed

directly at the floor of the spill containment tray; other than a small water leak during commissioning,

Daisy operated without any process leaks.

5.2.3 General Description

The major components of Daisy consist of seven Leach beds, one aerobic and six anaerobic, six

automatic three-way valves, one UASB, two tanks, two gas meters, and three pumps. All these

components and their interconnections are shown in Figure 5.1.

P2

Gas Meter 1

Aerobic Curing

Tank 2

LB 2 Anaerobic

Week 5

LB 5 Anaerobic

Week 2

LB 6 Anaerobic

Week 1

LB 4 Anaerobic

Week 3

Air Gas Meter 2

Waste Water

Biogas

Leachate

Leachate

UASB effluent

LB 3 Anaerobic

Week 4

U A S B

Tank 1

Treatment and Disposal Land Application

Screened Solid Digestate

Renewable energy

P1

P3

Automatic 3-way valves

Figure 5.1 Schematic flow diagram

50

The six operating leach beds contain waste that varies in age from six weeks (LB1) to one week (LB6) as

indicated by the shading. As shown, assuming a six week SRT, LB1 is ready for removal and replacement

with a fresh leach bed.

The system operates as follows. Pump three (P3) transfers leachate from Tank 2 (T2) to the leach bed

feed manifold and back to T2 in a continuous loop. The programmable logic controller (PLC, Plate 10)

actuates each three-way valve in sequence, redirecting leachate flow to each leach bed in turn, for a

pre-set time. The programme incorporates a short time delay between the closure of one valve and the

opening of the next to prevent overlap. The timing and duration of leachate delivery can be changed by

uploading different programs to the PLC. The system operates continuously. Leachate percolates

through the waste mass in each leach bed, and flows to a common LB manifold, and thence directly

into T1.

Pump one (P1) transfers leachate from T1 to the UASB at a predetermined fixed rate. In the UASB the

leachate passes through a suspended bed of granular sludge to the overflow point near the top, where

it exits as treated effluent and drains by gravity back to T2. From T2 the leachate is transferred once

again to the leach beds, and the closed loop is complete.

Biogas is produced in the UASB, in the leach beds, and in the tanks. Biogas from the UASB exits through

gas meter 1 (GM1) and from the leach beds and tanks combined through GM2; this arrangement was

used throughout except for brief periods when particular experiments, or GM calibration, required

temporary changes. The gas meters are simple wet tip devices that measure biogas in 100 ml

increments. The hydraulic

behaviour of the leach beds and

the tanks are intimately

connected. The transfer of

leachate from T2 to the leach

beds is balanced by an exchange

of headspace volume. As a

result, the biogas produced in

both must be measured

collectively. Separate

measurement of biogas

production in the leach beds is

very difficult, and requires Plate 1 Daisy the Digester – Overall view showing main components

51

temporary isolation of individual leach beds from the rest of the system; this is discussed more fully in

Chapter 8. The same conditions do not prevail in the UASB where the liquid level is constant and biogas

production is connected to the UASB headspace only.

The entire system runs automatically; temperatures in all six leach beds, T1 and T2, the UASB and GM2

are measured with three-wire resistance temperature detectors (RTDs), and recorded in the data logger

every 15 minutes. The RTD was chosen for its accuracy and durability, and for its flexibility – each is

installed through a compression fitting which allows the depth of the detector to be adjusted. Each tip

of GM1 and GM2 is recorded, and summed every five minutes, and every hour, by the datalogger.

Figure 5.1 depicts the original design. Two significant changes were made as a result of initial operating

experience. Firstly, as noted previously, it was found that aerobically curing the digestate in a leach bed

simply did not work because composting temperatures could not be achieved within the relatively

small mass of material present. Secondly, wastewater storage was not required. It was fully expected

that, with the successive addition of fresh waste and the water that it contained, wastewater would

have to be drained from the system from time to time. This did not happen; throughout the 616 days of

continuous operation Daisy remained in almost perfect hydraulic balance, requiring no net addition of

water or withdrawal of leachate. Once this balance was confirmed, the wastewater tank was removed

from the system. Daisy the Digester in her final form is shown in Plate 1. The following sections describe

each of the major components in greater detail.

5.2.4 Leach Beds

To arrive at the right specifications and configuration of the leach beds required many design iterations.

The final design is shown in Figure 5.2. The body of the leach bed, the flat top and the conical bottom

were fabricated from 1/2 inch PVC pipe (20 cm ID) and 1/2 inch PVC sheet. The top and bottom flanges

were each drilled with eight 1/4 inch bolt holes. Each flange had a machined groove to hold a Viton o-

ring. Three solvent-welded supports lugs were installed at the bottom to support the stainless steel

screen and geotextile. The leachate outlet was tapped for a ¾” NPT leachate drain pipe. The top was

drilled to accommodate three bulkhead fittings for gas venting, leachate delivery, and the resistance

temperature device (RTD). The gas venting, leachate delivery, and leachate drain pipe fittings are all

quick disconnects, equipped with a shut off valve in each half. This prevented the ingress of air during

connection and disconnection, and permits pressure testing. The RTD connection is a simple three

pronged plug. The lid bolts are fitted with wing nuts for ease of removal and replacement. Plate 2

shows an installed leach bed and its associated three-way automatic valve.

52

Heating tape

RTD

Bulkhead fittings

Leachate in from T2

Biogas out

Leachate out to T1

Bolts with wing-nuts Viton O - ring

PVC vessel with solvent-welded flanges

Woven geotextile

Viton O - Ring

Leachate drainage compartment

Stainless steel screen: 6 mesh

32cm

20cm

Figure 5.2 leach bed cross-section

53

The most important factors that contributed to the design were as

follows:

a) Dimensions and aspect ratio Nizami et al. (2010) used an aspect ratio of 0.4 - very wide and

flat as discussed previously. Although this had the advantage of

taking up less vertical space in the fume hood, it exacerbated the

problem of leachate distribution across the surface of the waste.

Since overall diameter was constrained by the available space in

the fume hood, I finally decided upon a ratio of 1.6 resulting in

dimensions of 20 cm d x 32 cm h to give a total volume of 10 L,

and a working volume of approximately 8.5 L. Diameter at the

flange was 27 cm.

b) Suspended particulate, fines and plugging This was the subject of several discussions. Nizami et al (2011)

and Browne et al. (Browne et al., 2013), treating grass silage and

food waste respectively, experienced serious plugging problems

with their two-stage digester system. After much thought, it

became apparent that:

fines would be produced in significant but unknown quantities.

they should be kept out of the UASB as far as possible they should not be

allowed to accumulate in the tanks - as far as possible.

they should not be allowed to settle in the piping system - as far as possible.

and that the best way to accomplish all this was to:

keep the leachate moving at all times

use the leach beds as a filter

incorporate a clarifier in the leachate tank (thank you Torsten Meyer)

flush the leachate lines regularly by temporarily raising pump speeds.

For the LB to function as a filter, the contents must of course remain permeable; the addition of

BA and the installation of geotextiles top and bottom helped achieve this. It was thought at the

time that eventually the LB might plug anyway, so I always installed a 6 mm ID leachate bypass

tube within the leach bed against the wall (thank you Elizabeth), protruding about 3 cm above

the upper geotextile, and held in place by the waste. Slight flooding occurred on only a handful

of occasions, and the bypass tube served its purpose.

Plate 2. Leach bed and solenoid valve

54

c) Heating Three options for providing external heat to the leach beds were considered; a heated

enclosure, a water jacket, or heating tape. The fourth alternative was to rely upon the heat

capacity of the recirculated leachate to raise LB temperature to about 35° C. After performing

some heat loss calculations, I decided on the simple option, insulate the leach beds with

Reflectix - a bubble pack product with a reflective coating, set the temperature of T2 to 39°C

(40°C at start up then subsequently lowered), and rely on frequent leachate recirculation to

maintain LB temperature. Daisy ran this way for six months with reasonable success, but I then

installed heating tape with manual temperature control on all the leach beds to raise

temperatures slightly and improve temperature stability. This was accomplished without

requiring a shutdown. Heat tape installation is described in Section 5.1.5 below.

d) Leachate delivery and distribution I sketched six different designs in an attempt to devise a plug-free system that would operate

effectively at any leachate flow rate. Ultimately, I chose the simple approach of nonwoven

geotextile as described in Chapter 4 (Section 4.5.2). This worked amazingly well; no dry spots

were ever found in the digestate. However, some of the early leach beds showed signs of

channeling and uneven settlement; though I think this was more a consequence of poor waste

placement during filling by a novice operator, rather than malfunction of the geotextile.

e) Removal and re-installation These operations have four distinct stages; filling a leach bed with fresh waste, removing a

leach bed at the conclusion of the digestion period, emptying and cleaning it, and installing the

freshly filled leach bed. To make all this as simple as possible the leach beds were designed with

the following features: lids that could be removed and replaced simply and quickly,

components that are easily removable for cleaning, o-rings to provide a seal, wing nuts that

need only be hand-tightened, bulkhead fittings that are easily replaceable in the event of

failure, and the quick disconnects mentioned previously. Lastly, when full, the leach bed could

not be so heavy and bulky that it could not be lifted in and out safely. In the event, a full leach

bed weighed about 12.5 kg, and its disconnection, physical removal, and the installation of a

replacement, took less than two minutes in total.

55

5.2.5 Upflow Anaerobic Sludge Blanket Reactor

As described in Chapter 4 (Section 4.5.2), the UASB was

built in two sections of unequal length; the upper section is

55 cm in height, the lower section 75 cm to give a

combined total overall height of 1.3 m. In practice, Daisy

was operated throughout with a 75 cm UASB, with a 25 cm

ID and a working volume of 27 L. Figure 5.3 shows the UASB

in this configuration and Plate 3 shows the UASB installed.

In all respects, the fabrication methods used for the UASB

were identical to those for the leach beds. The fittings

however are different. All connections are permanent (no

quick disconnects), the top and bottom sections are shown

bolted in place; wing nuts are not

required. The influent passes

through a backflow preventer to

stop discharge of the UASB

contents in the event of peristaltic

tubing failure at the inlet. The

bottom geotextile was attached

beneath the stainless steel mesh to

act as a flow distributor. Eventually

this became plugged, the geotextile

was removed, and the UASB was

returned to operation using only

the stainless steel mesh as a flow

distributor; it performed well. As

the leachate approaches the

overflow it is deflected inwards to

Plate 3. UASB Reactor

75cm

Backflow preventer

Baffle

Biogas out to GM 1

Leachate from T1

Woven geotextile beneath 6 mesh stainless-steel screen

Gas/liquid/solid separator

Effluent to T2

Heating

Sample ports

25cm

Figure 5.3 UASB cross-section

56

the gas liquid solid separator (GLS) by a baffle mounted on the wall. Inside the GLS, biogas bubbles

detach from the sludge granules and exit through the top, and thence to GM 1 (see Plate 1).

The particles descend back into the sludge bed and the liquid exits through the side of the reactor and

drains by gravity to T2; the system was constructed so that the effluent could be redirected to T1, but

Daisy was only operated this way to facilitate T2 removal for maintenance. The headspace above the

liquid in the UASB is separately vented to allow biogas

to exit through the top of the reactor and join the main

gas pipe; this prevents pressure buildup and hydraulic

disturbance in the UASB headspace above the GLS.

Five liquid/sludge sample ports were installed between

the inlet and the overflow. Heating tape was installed

along the length of the UASB. For clarity the RTD is

shown rotated 90° around the vertical axis from its

actual position. Plate 4 shows the finished heating tape

installation on the lower section of the UASB. Exactly

the same technique was used on the leach beds and on

the tanks. Each vessel was wrapped with 2 inch

aluminum tape. It covers the entire cylindrical surface

of the leach beds and UASB and all four sides of both

tanks to improve heat conduction across the surface of the vessels (the bottom of the tanks are

insulated with styrofoam). The heating tape was

also covered entirely with the aluminum tape to

hold it in place, ensure intimate contact with the

vessel wall and provide heat conduction from the

outside surface of the heating tape. Reflectix

insulation was then applied. There were fewer

constraints on the design of the UASB than on

that of the leach beds. As discussed in Section

4.5.2. by constructing the reactor in two sections,

and ensuring that the UASB feed pump had a

wide enough flow rate range, all practical

combinations of HRT and OLR and Vup could be

accomodated.

UASB effluent

Baffle

Biogas out

Figure 5.4 Schematic of GLS separator in UASB

Plate 4. Heating tape installation on UASB; sample points visible

57

The GLS separator was fabricated from a modified polypropylene funnel cemented into a female NPT

PVC fitting. The baffle located below the GLS is removable; it was constructed of two narrow sections of

25cm ID PVC pipe (2.5 cm and 1.0 cm in width respectively). Each has a section of its circumference

removed, such that it can be compressed and hold itself in place within the UASB like a piston ring. The

narrower section is installed within the wider section, to create a stepped profile (see Figure 5.4).

5.2.6 Leachate Tanks

The essential functions of leachate storage are set out in Section 4.5.3, including a total capacity of

around 40 L. In addition, one of the decisions from the lengthy deliberations about leach bed design

and function, was to include some form of clarifier ahead of the UASB, and to send the fines back to the

leach beds to filter them out (Section 5.2.4). The other constraint on the choices of tank configuration

*Can discharge to either tank

UASB FEED TANK LEACH BED FEED TANK

RTD

Clarifier To leach bed manifold

Return from leach bed manifold *UASB effluent

T1 – T2 exchange

T1 – T2 exchange

T1 – T2 balance

pipe

Sight glass (connected to gas system)

Biogas Biogas

Leachate drained from LBs

*UASB effluent

To UASB

Heating tape

RTD

Figure 5.5 Tanks 1 and 2

Plate 5. Tanks 1 and 2 plus Pumps 1, 2 and 3

58

was the matter of space; a single, low-profile, 40 L tank of appropriate dimensions was hard to find; I

also developed an operational preference for having two tanks to create some redundancy in the event

one had to be removed for service. The final choice was for two interconnected 20 L HDPE tanks, each

with a working volume of 17.5 L. Figure 5.5 shows Tanks 1 and 2 with their fittings and

interconnections. Plate 5 shows Tanks 1 and 2 plus the three pumps.

a) Design elements common to both tanks The tanks were identical when delivered. Each was modified to suit its specific purpose by cutting holes

and installing fittings as required. There were several common elements of design; both were fitted

with an RTD mounted through a bulkhead fitting in the roof of the tank. Both came equipped with a

large screw-on lid which was supposed to be gas tight. However, it proved hard to seal reliably. After

using silicone grease, then a gasket, then both (each of which worked for a while) I finally settled on

silicone sealant which was completely successful. More torque was required to remove the lid, and all

the old sealant had to be thoroughly removed, but this was an infrequent task and the inconvenience

was trivial.

Initially, a 100W immersion heater was installed through a bulkhead fitting in the lower side-wall of the

right-hand side of each tank. After about 30 days of operation, Tank 1 heater suffered a complete

failure. Suspended matter settled on the top of the heater element, lowering its efficiency and making

it work harder. Eventually silt, which had collected in the annulus between the heater element and the

bulkhead fitting, became baked in place by the overworking heater. This prevented any cooling of the

fitting, which promptly softened and began to leak. The system was shut down, the tanks removed, and

the immersion heaters replaced with heating tape. The heating tape was installed as described in

Section 5.2.5, and the tanks re-insulated. This design was much less responsive to demands for heat,

but more reliable and a lot safer. As part of the heater replacement modification, both tanks were

equipped with sight glasses to monitor liquid levels. The sight glasses are connected through lower

side-wall openings with bulkhead fittings on the right-hand side of each tank. At their open ends, the

sight glasses are connected to the gas manifold system to maintain hydraulic balance and prevent

blow-out. Biogas is vented through the top of both tanks to a joint manifold that, in turn, is connected

to the gas monitoring system.

Pump 2 (P2) transfers leachate at a low flow rate, between the two tanks, to help maintain hydraulic

balance. The initial plan was to install a level switch in T1 to turn P2 on and off as required. However, it

was feared that sludge would foul the level switch and compromise leachate flow to the UASB, so

instead a balance pipe was installed at the required leachate level of T1, permitting leachate levels to

59

equalize. Experience showed that operating with the flow from T2 to T1 at about 100 ml per minute

ensures that the clarifier never runs dry. Eventually this became standard operating practice. Once per

week however, all three pumps were run at full speed for 5 minutes to clear all the lines, flush the

piping, and reduce the buildup of solids. In the final step of this process, P2 is run at full speed in the

reverse direction, in an attempt to transfer accumulated solids from the bottom of T1 to T2 and from

there to the leach beds for filtration.

Both tanks have a roof-mounted aperture and valve for the purpose of introducing liquids. Most

commonly this was used to allow the addition of water to replace samples drawn from Daisy, and to

return the accumulated leachate that drained from fully digested leach beds after their removal from

the system.

a) Leach bed feed tank (Tank 1) As leachate drains from the six leach beds it flows into a common manifold that discharges into T1, near

the bottom right front. The conditions in T1 are fairly quiescent; under the operating conditions

adopted the T1 HRT is about 1.5 hours (depending on the T1/T2 exchange flowrate). The clarifier

consists of a section of 6 cm ID PVC pipe connected to a sump beneath T1. Clarified leachate is drawn

from the sump by P1 and delivered to the bottom of the UASB. Leachate is circulated between the two

tanks using a second transfer point located at the bottom left rear of T1. With the chosen leachate

recirculation rate through the leach beds, the flow rate of leachate draining to T1 exceeds the flowrate

to the UASB. This causes the level in T1 to rise, and the excess leachate to flow through the balance

P2

Tank 2

LB 2 LB 5 LB 6

LB 4

LB 3

UASB

Tank 1

P1

P3

7.2 l/h

6.0 l/h

30.6 l/h

1.1 l/h

LB 1

24.0 l/h

1.1 l/h 1.1 l/h 1.1 l/h 1.1 l/h 1.1 l/h

6.6 l/h

5.4 l/h 7.2 l/h

Figure 5.6 Leachate flows L/h - typical operating conditions

60

pipe to T2. The balance pipe is positioned to ensure that the leachate level in T1 is 2 cm higher than the

lip of the clarifier.

Tank 1 is also equipped with a return pipe for UASB effluent that discharges at the bottom right rear.

This was used only rarely to assist with maintenance.

b) Leach bed feed tank (Tank 2) Tank 2 serves as the reservoir from which the leach beds are fed. Pump 3 draws leachate from a

transfer point located at the bottom right front of the Tank and pumps it through a half inch pipe to the

leach bed manifold above LB1 on the left-hand side of Daisy. It flows through the manifold to the down-

comer on the right-hand side and back to T2 at the bottom right rear. The effluent from the UASB is

discharged to the bottom left rear of T2. The pickup point for leachate transfer to T1 is located left of

centre at mid-depth of the tank, and exits through a bulkhead fitting in the roof of the tank. Tank 2 also

receives overflow from T1, through the balance pipe, during periods of peak leachate drainage from the

leach beds. The combined flows of the entire system, from T2 through P3 back to T2, from T2 through

P3 and through the leach beds to T1, from T1 through P1 to the UASB to T2, from T2 through P2 to T1,

and the overflow from T1 to T2, are shown in Figure 5.6, a simplified schematic of liquid flows only.

5.2.7 Pumps and Valves

a) Pumps

Daisy is equipped with four identical peristaltic pumps, three active plus one spare. Flow requirements

were estimated (see Figure 5.6), and the pump specifications chosen such that the pumps would not be

required to run at more than about 60% of their rated maximum speed of 300 rpm. Norprene A-60-G

peristaltic tubing was chosen because of its long life. Published test data showed a mean time to failure

of 1000 hours at 600 RPM with a 3-roller pump head. In practice, and for greater safety and security,

the tubing was changed every 1000 hours but repositioned at 500 hours, and the maximum continuous

pump speed used was about 170 RPM. Three sizes of tubing were used; 1/8 ID, 3/16 inch ID, and 1/4 ID

for P1, P2 and P3 respectively. I experienced one pump head failure near the end of the experiment.

Daisy’s design also has built-in redundancy; it can operate for extended periods using P1 and P3 alone,

and for short periods with a single pump alternating between the duties of P1 and P3.

b) Valves Each leach bed was equipped with an automatic, pneumatically-actuated, three-way PVC valve (see

Figure 5.6). The leachate manifold comprises all 6 three-way valves connected in series; in the

normally-closed position the flow is straight through the valves, returning the leachate directly to T2. As

61

each valve is actuated in turn, the flow is redirected downward into the corresponding leach bed for a

predetermined time controlled by the PLC (see Section 5.2.9). Plate 6 shows 5 of the 6 valves with their

corresponding leach beds; the leach bed gas manifold is in the foreground.

There are, in total, 100 manual valves (including all shutoff valves and sampling points). They are

divided into 5 groups; quick disconnects, leachate, biogas, UASB, and pump; they are listed, coded, and

described in Appendix A. Every component is labelled to correspond to the equipment list. The quick

disconnects (VQ) are fitted only to the leach beds and are described in Section 5.2.4. The leachate

valves (VL) are primarily for the

purpose of taking samples, which can

be drawn from the outlet of each

leach bed and the outlet from both

tanks. They also have a secondary

purpose, allowing leachate flows to

be redirected with temporary tubing

for maintenance purposes, or to

permit occasional measurements of

Daisy’s behaviour, for example leach

bed permeability. The biogas valves (VG) allow gas flow from various components of the system to be

directed to one or other of the gas meters. Also, a tee-fitting with a butyl rubber septum in the side arm

was installed in the gas outlet from each leach bed, and the gas inlet to each gas meter, to permit the

withdrawal of biogas samples for GC analysis. The UASB valves (VU) permit sampling of the inflow of

leachate and the outflow of effluent and of the contents of the UASB (leachate and sludge) at different

depths within the reactor. Lastly, each pump is equipped with two valves (VP), one at the inlet and one

at the outlet, which can be closed to isolate the pumps for service or calibration.

All manual valves are made of PVC, the quick disconnects are made of PP, and all the interconnecting

piping and associated fittings are also made of PVC. The leachate drain lines are of 3/4 inch pipe, and

the leachate lines and the gas lines are of 1/2 inch pipe. All liquid lines, with the exception of the leach

bed drain lines, are insulated.

Plate 6. Leach beds, 3-way solenoid valves and leach bed gas manifold

62

5.2.8 Biogas Measurement

Daisy was set up with two identical wet-tip gas meters as shown in Plate 7.

Gas meter number one (GM1) was connected exclusively to the UASB. GM2

was connected exclusively to a combination

of all six leach beds and both tanks. There

were exceptions to this arrangement; gas

flows were sometimes redirected for

recalibration purposes, or to conduct

specific experiments on biogas production

from leach beds. Each tip of the gas meter

represents 100 ml of biogas at atmospheric

pressure and gas meter temperature.

Corrections to STP were made based on the

daily average temperature measured in GM2

and on the daily average barometric pressure, measured on the roof of

the Wallberg building by Prof. Greg Evans’ group. A one-litre foam trap

(Figure 5.7) was installed in each gas line immediately upstream of the

gas meter to prevent the foam which anaerobic digesters sometimes generate from entering the

meter. Digester conditions were such that no foam was ever produced, so the traps never served their

intended purpose. They did however prove useful to intercept water that was occasionally syphoned

out of the gas meters as a result of pressure drops within Daisy, usually as the results of operator error.

Under normal conditions, each gas system operated at the slightly elevated pressure of 12 cm WC

(water column), the depth of the water in the gas meters. Two one-litre Tedlar bags (not shown) were

installed, one on a tee-branch off each gas line. Under normal operating conditions these bags are full

and under pressure. When liquid is withdrawn from Daisy the gas pressure drops; when a larger volume

is withdrawn the bags deflate, preventing creation of a partial vacuum. These worked very reliably and

were a great help.

5.2.9 System Monitoring and Control

The control system manages all of the automated functions of Daisy, and records biogas production

and system temperatures. Digester safety was a priority throughout. The controls have six main

functions:

Plate 7. Gas meters 1 and 2 and top of UASB

Gas out

Gas in

Figure 5.7 Foam trap

63

Valve sequencer – consisting of a PLC and six solenoid valves, each with a manual override

switch, the valve sequencer controls the opening and closing of the automated valves.

Signal transmitter - transmitter modules mounted in the transmitter rack to receive, process,

and distribute signals from the resistance temperature detectors (RTD) and the gas meters to

the data logger and to the temperature controllers.

Temperature controllers - three programmable temperature controllers, fed by a temperature

signal from the UASB and from each tank via the transmitter rack, activate solid-state relays to

deliver power to the heaters.

Data logger - the data logger records the tips of both gas meters and the temperatures from 10

different locations within Daisy.

E-stop - the Big Red Button permits emergency shutdown of the entire system. For added

safety, the power supply to the pumps, the leach bed heaters, and the air compressor are all

routed through the control panel, so that these components too are controlled by the E-Stop.

Internet connection – a specially assigned URL permits remote monitoring and downloading

from the data logger but, for security reasons, does not allow system control.

The following functions are manually controlled:

Pump speed and direction

Leach bed heaters

All other valves

Plate 8 shows the outside of the control panel during operation. Plate 9 shows the inside of the control

panel with all major components labelled.

a) Valve Sequencer and PLC The PLC controls the operation of the automatic valves, and the warning lights on the face of the

control panel display the system status (Plate 9). Pressing the START button activates the valve

sequencer which functions as follows:

on starting, there is a hold time (set as desired).

after the hold time expires, the first solenoid, for LB1, is energized and a dedicated timer is

started (time set as desired) .

when LB1 timer expires the LB1 solenoid is de-energized and its timer reset. There is a six

second delay (time set as desired) between the closure of LB1 solenoid and the opening of LB2

solenoid.

64

The cycle continues for all six leach

beds.

When the LB6 timer times out, the

sequence returns to the beginning

and the ‘hold’ timer is started again.

If the valve sequencer STOP button is

pressed, the sequence stops, all

timers are reset, and all solenoid

outputs de-energized.

When re-started, the sequence

begins with the ‘hold’ duration. Each

output, when energized, turns on an

indicator light on the front panel to

show the solenoid is energized.

To provide operating flexibility, each solenoid valve can be controlled manually by a three-position

switch on the control panel. During normal operation the switches are set to ‘auto”; in the ‘closed’

position the LB is by-passed; in the ‘manual’ position the solenoid is activated and the leachate flow

redirected for as long as the switch remains in that position.

Transmitters The transmitters receive a millivolt signal from the RTDs and the gas meters and convert it to a 1 to 5 V

DC/4 to 20 mA DC output to the data logger, and to the temperature controllers. The transmitter and

Heater relays

PLC

Heater controls

Datalogger

Plate 9. Controls - inside view

Solenoids

Transmitter rack

Plate 8. Control Panel - Valve sequencer feeding leach bed 6

65

transmitter rack are used components obtained by Glenn Wilson, free of charge, from Xerox

Corporation. The transmitters were modified with new resistors, and calibrated.

b) Temperature controllers Three automatic temperature controllers, one for each tank and one for the UASB, were installed in the

front of the control panel. Based on the RTD signals and the set-point, the controllers activate solid-

state relays which turn the power to the heaters on and off. The controllers themselves are

programmable; Initially they were set up using the ‘autotune’ feature that monitors the behaviour of

the temperatures being controlled and automatically sets the on-off cycle time and PID parameters

(proportional band, integral setting and the derivative or rate), then switches itself out of autotune.

Collectively these settings narrow the band within which the temperature fluctuates.

c) Data Logger The data logger was a dataTakr DT80 supplied by Dycor Technologies who also wrote the software for

this particular application and provided technical support free of charge. The system records

temperatures from 10 locations every 15 minutes, and gas meter tips every five minutes (plus hourly

totals) and stores all data in CSV format. Weijun Gao wrote a program to process downloaded data,

convert it to Excel format, and display the results in both tabular form, and graphically as follows;

Leach bed temperatures UASB, T1 and T2 temperatures GM2 temperature Hourly gas production – GM1, GM2 and total Cumulative gas production GM1, GM2 and total.

Downloading and conversion can be carried out for whatever periods of time one chooses, but was

always done daily. The Internet connection was used first thing every morning, to download results for

the previous 24 hours, examine them, link them to the daily log sheet and store them. Periodically, data

was also downloaded for more extended periods to view longer-term trends. The Internet connection

also provided the capability to monitor operations, which I did every morning and every night (and

often many times in between). With experience, it was possible to analyze and diagnose operating

anomalies very readily, usually from temperature patterns but also from gas measurements. On a

handful of occasions, this required a drive to the lab late at night to correct a problem. This obviated

much subsequent grief.

66

d) Control system reliability Overall, the system was extremely reliable, with only two non-trivial problems. At startup Daisy

experienced a series of solenoid failures, rapidly using up the spares - which also failed. The supplier

replaced them all, plus I bought additional spares; the problem never recurred.

Although the immersion heater failure was dealt with as described above, another heater problem,

which proved very intractable, arose in June 2015 when the Tank 1 heater began cutting out, seemingly

randomly, causing system temperatures to drop. Trying to understand and troubleshoot this problem

consumed an enormous amount of my time, and that of Dan Tomchyshyn and Glenn Wilson; every

component in the control circuit was either replaced or tested and found to be sound; the controller

itself was made in China, and Omega, the supplier of the controller, was unable to help. The specific

cause remains unknown, but it was finally established that this was not a random event; it occurred

under specific conditions of falling temperature in Tank 1, and appeared to be intrinsic to the controller

itself. Ultimately, in early August 2015, at the suggestion of Torsten Meyer, the problem was solved by

raising the T1 RTD by about 5 cm, out of what I presumed was a cool mixing zone near the bottom of

the tank. The problem never recurred.

5.2.10 Sampling Locations

Plate 11 shows the eight locations for sampling biogas, the eight locations for sampling leachate and

the five locations for sampling the UASB contents (sludge and leachate).

5.2.11 Support Structure and Miscellaneous Equipment

Daisy’s components are all mounted on Dexion frames with support platforms fabricated from 1 inch

clear acrylic. All this material, together with all manner of other bits and pieces, was made available at

no charge from the Unit Operations lab courtesy of Paul Jowlabar. Paul was also endlessly patient and

extremely generous with his time, providing valuable assistance and guidance on matters of design and

fabrication.

An air compressor with an accumulator tank provides the motive power for the automatic valves. An

argon cylinder was installed to flush oxygen out of the system, in particular from freshly filled leach

beds, and to pressure test the leach beds and other system components. A Variac power supply

controls the power input to all six leach bed heaters, each of which is equipped with an individual

dimmer switch. Using this combination, power to the heaters can be controlled collectively and

individually.

67

Pu

mp

s

UA

SB

fe

ed

ta

nk

L

ea

ch

be

d f

ee

d t

an

k

Ga

s m

ete

r 1

UA

SB

Ga

s m

ete

r 2

Ta

nk

s +

LB

s UA

SB

Le

ac

h b

ed

s 1

- 6

UA

SB

F

eed

Lea

ch b

ed

Fee

d

UA

SB

E

fflue

nt

Lea

chat

e

sam

plin

g G

as m

eter

1

Gas

met

er 2

G

as

sam

plin

g

Le

ac

h b

ed

7

Off

lin

e

Plat

e 10

. Gas

and

Liq

uid

Flow

s an

d Sa

mpl

ing

Poin

ts

UAS

B sa

mpl

ing

68

Chapter 6. Analytical Methods

6.1 Total Solids, Volatile Solids and Chemical Oxygen Demand

The measurement of TS, VS and COD of leachate samples, granular sludge samples, all the substrates,

the bulking agent (BA), and the digestate (DG) are critical to the determination of substrate destruction

efficiency, calculation of Daisy’s mass balance, and assessment of internal behaviour, including the

concentration of inorganic matter in Daisy’s

recirculating leachate. The goal was simple;

achieve consistent, accurate and reproducible

analytical results. Considerable time and

effort were expended to achieve this. The

analyzes themselves were performed using

standard procedures (APHA, 1992). Leachate

samples presented no particular problems;

note that VS and COD were measured as

totals, rather than as separate solids and

filtrates.

However, the preparation of the solid

substrates required some method

development to reduce particle size and

homogenize the samples. Two different

methods were developed and compared. The

first was based on pulping and subsequent

dilution - the “wet” method - in which all the

substrates are prepared in a similar manner

to the food waste (FW) which is wet to begin

Dry

Dry then suspend in

water

FW hand-sorted 250g + 250g water

DWet Blend 1 - step

50g pulp to 1L with water

Blend in 3 steps

Refrigerate 2hr

Pipette

10ml x 6 TS (12h)/VS

Pipette Pipette

2.5ml x 6 COD

2.5ml x 6 COD

100g Pulp

Dry 48h

Wiley Mill -40 mesh

25ml pulp in 1L with water

500mg in 1L water

Figure 6.1 TS/VS/COD methods – Food Waste

69

with. The second was based on drying and grinding

the samples to a powder, followed by subsequent

suspension in water - the “dry” method. Three

variants of the methods were developed to suit the

properties of the solids. A summary of the procedures

follows and the details are provided in Appendix C.

Figure 6.1 shows the FW methods, Figure 6.2 shows

the fibre methods and Figure 6.3 shows the DG

methods.

a) Food Waste

Figure 6.1. Food waste first had to be hand-sorted to

remove bones, pieces of wood, metal and pottery,

and plastic bags, string, and other ‘windables’ to

avoid damaging the blender. Regardless of whether

the “wet” or “dry” method was to be used, the next

step involved blending to a pulp to create

homogeneity; a 50:50 mixture with water (250 g of

each) was blended in a single cycle at low speed. The

resulting pulp was divided as shown. First the “wet”

method; 50 g of the pulp was diluted to 1 L and

blended again in three steps of increasing speed. This

created a fine pulp but raised the temperature to

60°C, so the sample was covered, and refrigerated for

two hours to prevent evaporative losses during subsequent

pipetting. Six 10 mL replicate samples were then pipetted for

TS/VS measurement; 25 mL of this pulp was further diluted to 1 L,

and finally six 2.5 mL replicate samples were pipetted directly to

Hach tubes for COD analysis. Second the “dry” method; 100 g of

the initial pulp was dried for 48 hours at 105°C; the resulting

“biscuit” was broken into pieces (Plate 11) then ground to - 40

mesh (- 0.4 mm) in a Wiley mill. A 500 mg sample of this material

was dried again, reweighed, then suspended in a litre of water;

finally, six 2.5 mL replicate samples were pipetted directly to Hach tubes for COD analysis.

Plate 11. FW dried and broken up

CB, BB, NP, FP, BA as-received

2.5ml x 6 COD

Dry then suspend in

water

500mg in 1L water

Pipette

D Wet Dry

6g sample in 1L water

Blend in 3 steps

Refrigerate 2hr

Pipette

10ml x 6 TS(12h)/VS

Pipette

2.5ml x 6 COD

25g sample

Dry 2h

Wiley Mill -60 mesh

-40 mesh NP

10ml pulp in 200mL water

4g sample

TS(2h)/VS

Pipette

Figure 6.2 TS/VS/COD methods – Fibres and Bulking Agent

70

b) Fibres Figure 6.2. No initial sorting was required

except for the removal of odd bits of plastic.

For the “wet” method, 6 g of dry sample were

blended in 1 L of water; the subsequent steps

were similar to those used for food waste. For

the dry method, triplicate samples of about 4 g

each (depending on the bulk density of the

fibre) were analyzed directly for TS and VS. A

25 g sample was dried for two hours then

ground in the Wiley mill to - 60 mesh

(-0.25 mm), except for NP to - 40 mesh (to

prevent screen plugging). These finely ground

samples were then suspended in water and

analyzed in the same way as the dried FW.

c) Digestate Figure 6.3. The samples have a high (78%)

moisture content. For the “wet” method, a 26g

sample of digestate was mixed in 1 L of water,

blended in three steps and then followed the

same steps as the fibres, with an additional

dilution step for the TS/VS samples. For the

“dry” method, the procedures are similar to

those for fibre samples, except that the initial

samples were of larger size because of their high moisture content, and their drying times

correspondingly longer.

d) Comparison of Methods Table 6.1 shows a comparison of COD measurements using both “wet” and “dry” methods; it also

includes published data and calculated values derived from elemental analyzes of three sets of milled

samples of substrates prepared using the “dry” method (see Section 6.2). The “wet” method gives

higher values than the “dry” method for all the substrates and for the DG; but for the BA the “dry”

method was slightly higher. The literature sources and the calculated values all confirm the accuracy of

the “dry” method, particularly for CB, BB and NP. The error analysis (± one STDEV) shows the “wet”

Dry D Wet

26g sample in 1L water

Blend in 3 steps

Refrigerate 2hr

Pipette

10ml x 6 TS(12h)/VS

Pipette

1.5ml x 6 COD

25g sample

10ml pulp in 200mL water

20g sample

Pipette

Digestate as-sampled

25ml pulp in 500mL water

Pipette

TS(48h)/VS Dry 48h

Pipette

1.5ml x 6 COD

Wiley Mill -40 mesh

Dry then Suspend in

water

500mg in 1L water

Figure 6.3 TS/VS/COD methods - Digestate

71

method to be less precise. The potential for errors in the “wet” method could be observed in the

numerous pipetting steps required. Despite all the energy input from the blender, larger particles were

still present and could be seen during pipetting; occasional plugging of the cut-tip pipette was

unavoidable. In all respects, the “dry” method proved simpler, and less subject to random error.

Table 6.1 Comparison of COD Analysis Methods COD g/gTS

Sample Calculated from Elemental Analysis Literature Measured 1 2 3 Avg. STDEV Dry1 Dry2 STDEV Dry Error Wet STDEV

Cardboard (CB) 1.16 1.14 1.10 1.13 0.03 1.08 1.10 0.02 1.14 0.05 1.33 0.12

Boxboard (BB) 1.12 1.08 1.09 1.10 0.02 1.12 - - 1.13 0.01 1.28 0.12

Newsprint (NP) 1.34 1.28 1.27 1.30 0.04 1.22 1.21 0.03 1.33 0.02 1.43 0.15

Fine paper (FP) 1.09 1.03 1.05 1.05 0.03 1.07 1.07 0.02 1.00 0.03 1.05 0.15

Food waste (FW) 1.53 1.38 1.48 1.46 0.07 - - 1.30 0.11 1.37 0.12

Bulking agent (BA) 1.27 1.26 1.31 1.28 0.02 - - 1.32 0.05 1.27 0.14

Digestate (DG)

1.19 0.05 1.21 0.10

The data presented here represent the analysis of a large number of samples over a period of many

months, and led to the conclusion that the dry method was preferred for the reasons outlined.

Ultimately this conclusion was confirmed by the mass balance calculations presented later. However, I

was not able to determine the cause of the systematically higher results obtained using the “wet”

method, other than to say that it is highly unlikely that pipetting errors could have created the

systematic errors observed. All of the results presented in this thesis are based upon data obtained

using the “dry” method (unless otherwise noted).

e) Digestion of individual substrates – using coupons While it was possible to measure the

TS/VS/COD content of the individual

substrates going in, it was impossible

to separate them, or form any

opinion about their digestibility on

the way out; only the overall

digestibility of the aggregate waste

mass could be determined. To

remedy this a very simple analytical

procedure, the coupon method, was

developed. Each leach bed of fresh

waste received stainless steel tea

1(Pommier et al., 2010) 2(Yuan et al., 2014)

Plate 12. Installing triplicate coupons for digestibility determination

72

balls (coupons) containing samples of individual substrates or bulking agent on a rotating schedule

summarized in Table 6.2. Two sets of three 2.5 cm tea balls, each set comprising triplicate weighed

quantities of one specific substrate, were inserted

into the first, second, fourth and fifth leach beds in

a six-week sequence. For example, in Week 2 three

coupons of BB were placed at a depth of 25 cm and

three coupons of FP at a depth of 20 cm (see Plate

12). In weeks three and six a single 5 cm coupon,

filled with BA, was placed in the middle of the leach

bed. The larger size was necessary because the low digestibility of the wood chips required larger

samples and BA consists of relatively large particles. Installing three of these in a single leach bed would

have taken up too much space and disrupted hydraulic flow patterns. The bottom half of the BA

coupons was lined with a glass fibre filter paper to retain fines and prevent false results.

In this manner the contents of each coupon were exposed to the same digestion conditions as the rest

of the feedstock, but kept separate from it, enabling individual destruction efficiencies to be measured.

Another advantage of this procedure was that the substrates were being digested in exactly the same

form as they were present in the bulk mass of waste; that is without any further size reduction.

At the end of the digestion period, the coupons were carefully removed, weighed wet, dried and TS

measured. To measure VS, it was necessary to remove the dried digested material from the coupon and

re-weigh it into an aluminum dish. Repeated trips to the furnace at 550° C would soon have destroyed

the tea balls. Attempts were made to find materials other than stainless steel tea balls for use as

coupons to minimize flow anomalies; for example durable fabrics, whose shape would conform to the

waste as it was digested. However, none could be found that would function as required, and survive

the trip through the drying oven.

6.2 Elemental Analysis

Three complete sets of samples, each comprising CB, BB, NP, FP, BA, FW, and DG, were analyzed for C,

H, and N content. The analyzes were performed by ANALEST, a commercial lab in the Department of

Chemistry.

The equipment used was a Thermo Flash 2000 CHN Analyzer, which is based on the principle of

dynamic flash combustion, and uses a single reaction tube (combustion and reduction function

combined). Grade 5.0 He is the carrier, with a flow rate of 140 mL/min.

Table 6.2 Coupon Placement in Leach Beds Week Size Coupon Depth

25 cm 20 cm 1 2.5 cm CB NP 2 2.5 cm BB FP 3 5.0 cm BA

4 2.5 cm NP CB 5 2.5 cm FP BB 6 5.0 cm BA

73

Oxygen is used with a flow rate of 250 mL/min for 5 seconds to combust the samples. The column is

PTFE, (2 m (l) x 5 mm (ID) housed in a chamber with the TCD, at 75°C. Samples are weighed into a tin

capsule using a microbalance with a readability of 1 µg, and dropped into the oxidation/reduction

reactor housed in a furnace at 950°C. Oxygen is introduced and the temperature is further elevated to

1800°C by the exothermic reaction between oxygen and tin. All substances are converted into

elemental gases and then reduced in the reactor. The gases are then separated on a column and

measured with a thermal conductivity detector. The percentages of C, H and N in the sample are then

calculated using the peak areas of the chromatograms and the calibration curves for C, H and N

obtained using an acetanilide standard. Percent O was then calculated by difference. Sample sizes

ranged from 0.6 to 1.1 mg.

With the exception of the FW, the nitrogen content of the samples was right at the detection limit of

the Thermo Flash analyzer. Consequently, total Kjeldahl nitrogen (TKN) analyzes were also performed in

Professor Diosady’s lab in the Department of Chemical Engineering and Applied Chemistry. It is these

results that appear in Table 5.1. The procedure uses a Büchi 425 digester and a Büchi K-350 distillation

unit. A sample (approximately 1 g) is weighed on weighing paper and the sample plus weighing paper

added to the digestion tube. Similarly, a blank is prepared using only the weighing paper. Four Kjeldahl

tablets and 25 mL of concentrated H2SO4 are added to each tube and digested in the Büchi 425

digester. After one hour of digestion the tubes are allowed to cool; 50 mL of water is added slowly and

gently mixed.

The sample tube is inserted into the Büchi K-350 distillation unit. The distillate is condensed and

collected in an Erlenmeyer flask containing 60 mL of 4% boric acid and the indicator. A 32% NaOH

solution (about 75 mL) is added to the sample tube until the mixture turns alkaline. The condensed

distillate, sample and blank, are then titrated with standard H2SO4 solution.

The nitrogen content is calculated as follows:

%N = [14 x n x (V-Vo)]/W x 100%

n = normality of H2SO4

V = Vol. H2SO4titrated for sample (ml)

Vo = Vol. of H2SO4 titrated for blank (ml)

W = sample weight (mg)

The samples analyzed were taken in April, May, and August of 2016, and were prepared using the “dry”

method described above. The results are summarized in Table 8.8, and the raw data are presented in

74

Appendix C. They are remarkable for their consistency; the greatest variability is seen in the FW, which

is intrinsic to the material itself and to be expected. Each analysis, using both methods, was carried out

in duplicate.

6.3 Volatile Fatty Acids and Sulphate

VFAs and sulphate were analyzed using a Dionex ICS-2100 ion chromatograph. The column was an RFIC

Ionpac with an AG18 guard column/AS18 analytical column. The runtime was 23 minutes per sample.

The eluent was a KOH solution at a flow rate of 1 mL/min. Using the gradient method, the eluent

concentration was increased from 2 mM to 32 mM over the first 15 minutes, and remained at 32 mM

for 3 min then reduced to 2mM for the remaining 5 min. Suppressor values were 80 mA. Standards

were prepared at 5, 10, 50, 100, 200, and 500 µM, by making serial dilutions of 10 mM solutions of

sodium acetate sodium propionate and sodium butyrate, with 18 MΩ Milli-Q water.

Leachate samples (10 mL) were taken from Daisy (contemporaneously with separate samples obtained

for analysis of TS, VS and COD, and pH and alkalinity) and immediately filtered with a 0.22 µm syringe

filter, and stored at -20° C. After initial trials, samples were diluted 5X, 10X, 20 X, and 50X as required,

to fall within the calibration range. A six point calibration was included in every run; two blanks were

run prior to and following each set of standard samples. The concentration of each VFA and of sulphate

was calculated in µM, by the software in instrument, using the calibration data from the run.

All of the VFA and sulphate analyses were performed by Masters student Peter HyunWoo Lee (thank

you Peter).

6.4 Alkalinity and pH

Leachate samples were analyzed regularly for pH and alkalinity ratio. The latter is calculated as the ratio

of intermediate alkalinity (IA) to partial alkalinity (PA). Two methods were compared during start up

with synthetic feed (see Chapter 7 Section 7.3.2). The method of Ripley et al. (1986) is based on a two-

step titration; step 1

from starting pH to

pH 5.75, and step 2

from pH 5.75 to pH

4.3. The second

method, of uncertain origin but believed to come from the USEPA, is based on a titration from starting

pH to pH 4.0, followed by boiling to evaporate VFAs, then a back-titration to pH 7.0. Each method was

Table 6.3 – Synthetic Feed - pH and Alkalinity Day 56 pH PA IA IA/PA

USEPA fresh sample 7.01 1330 360 0.32 USEPA centrif. Sample 7.32 1220 510 0.44 Ripley fresh sample 7.01 1110 510 0.46 Ripley centrif. Sample 7.32 1190 470 0.39

75

tested on raw leachate and centrifuged leachate. In both methods titration was carried out with 0.1N

H2SO4 using a pH meter. The results, comparing both methods, are shown in Table 6.3.

The Ripley method is much less involved, did not give materially different results, and so became the

method of choice for all subsequent tests (without filtration or centrifugation of the sample). According

to Ripley, the ideal range for stable operation is an alkalinity ratio < 0.5 and preferably <0.4. The

usefulness of this method was demonstrated by Zhang et al. (2013); the IA/PA started to rise ̴20d

before the pH started to fall leading to reactor failure.

6.5 Biogas Analysis

Samples of biogas were analyzed using a Hewlett Packard 5890 equipped with CTR I column and a

thermal conductivity detector (TCD). Helium was used as the carrier gas at a column pressure of 180

kPa. The injector and detector were both set at 200˚C. The oven temperature was operated

isothermally at 50˚C and the runtime was set at 11 minutes. Methane and CO2 standards were used to

prepare calibration curves. Serum bottles were filled with high-purity standard gases by bubbling on the

water; the bottles were closed with butyl stoppers; the calibration curves were created by withdrawing

five aliquots of standard gas, 200 µL, 150 µL, 100 µL, 50 µL, and 25 µL, and injecting them into the GC.

CO2 was eluted at about 0.45 min. and methane was eluted at about 8 min. The results were used to

generate a calibration curve for each gas. The GC was calibrated three times during the course of the

research programme.

Samples of biogas, 200 μL, were extracted from Daisy using a 500 μL gas-tight syringe, and injected into

the GC. The CH4 and CO2 content were determined from the slope of the calibration curves.

76

Chapter 7. Commissioning and Startup

Commissioning was carried out in three stages, leak testing – static and dynamic, calibration of gas

meters and pumps, and finally operation of the entire system with water only. Start-up was also

conducted in three stages between December 3rd, 2014 and July 5th 2015 as follows;

Synthetic feed - tanks and UASB operating

Solid waste - tanks and leach beds operating

Solid waste - entire system operating

By following this stepwise approach, problems could be identified and fixed and safe operating

procedures developed. By the time commissioning and start-up were complete, Daisy was fully

functional and operating reliably 24 hours per day, having completed 10 consecutive weeks of

consistent operation.

7.1 Commissioning

Leak Testing The system was leak-tested at 2 psi using compressed air, first in individual sections then in its entirety.

The tanks and the UASB were then filled with water and the system tested again. During this process,

some of the fittings, particularly bulkhead fittings, were changed or modified. The entire system was

made gas-tight and water-tight. A safety inspection was carried out by Prof. Brad Saville on August 7th,

2014; two conditions were attached to his approval-to-operate; installation of a spill alarm in the fume

hood and an H2S alarm in the lab. Both were completed promptly.

Calibration The peristaltic pumps were calibrated with three different sizes of tubing, against no head and against a

1.3 m head (the height of the leach bed manifold) – no difference was found. Calibration was checked

from time to time during the course of operation. Most particularly, calibration tests were run just prior

to replacing the peristaltic pump tubing; the calibration was found to be unchanged.

The gas meters were calibrated at 100 ml per tip; initially this was done using the 140 mL syringe device

supplied by the manufacturer. Subsequent calibrations were carried out using a purpose-built

calibration device, designed to simulate normal operation. It consisted of a 10 L airtight plastic tank,

into which water was pumped using one of the peristaltic pumps; the displaced air was discharged

through each gas meter in turn, at a flow rate approximating that of biogas production. In this way, the

gas meter remained pressurized throughout calibration, and I could record the volume of 20 tips in a

77

period of 10 minutes under conditions that simulated normal operation. The gas meters held their

calibration very well.

System Test With Daisy still filled with water, the pumps, valve sequencer, and heaters were turned on, and the

system tested for performance and stability. Vessels reached their set-point temperatures of 37°C (T1

and UASB) and 40°C (T2) in approximately 3 hours; after some initial troubleshooting and adjustments

to the controls, the system was operated continuously for two weeks, during which time everything ran

smoothly; temperatures remained very stable with no leaks or other equipment malfunctions. Figure

7.1 shows the recorded temperatures between hour 90 and hour 100, out of 115 hours of continuous

stable operation. The periodicity of the temperature follows that of the valve sequencer.

Field Capacity and Permeability Tests A series of tests was conducted to get a preliminary idea of how the waste would behave in the leach

beds; determine how much water would be required to saturate the fibres, measure the bulk density of

the waste plus water mixture and the permeability of that mixture within the leach bed. The waste was

prepared in three stages and permeability measured at each stage as follows; Test 1: a blend of

shredded fibres, Test 2: fibres plus bulking agent and Test 3: fibres plus bulking agent plus food waste.

This last combination was chosen to approximate the average composition of organic waste from

commercial sources. Water was added in Tests 1 and 2 to achieve saturation before placement in the

0

5

10

15

20

25

30

35

40

45

88 90 92 94 96 98 100 102

Tem

pera

ture

°C

Elapsed Time hrs

LB1LB2LB3LB4LB5LB6

Tank 2Tank 1UASB TopUASB Bot.

Figure 7.1 Anaerobic digester temperature profile; Tank 1, Tank 2, UASB and 6 LBs, operating with water only

78

leach beds. Test 3 was a continuation of Test 2, with food waste added. The frozen food waste released

some free water when thawed, so no further water was added to the waste in Test 3. The prepared

waste was placed in a single leach bed, lightly packed down, and the headspace above the waste

measured. Two paper towels (unbleached paper) were placed on top of the waste, followed by a disc of

non-woven geotextile, equal in diameter to the leach bed. The purpose of this combination was to

disperse the added leachate across the top of the waste before it drained into, and through, the waste.

The paper towels were not used after this first trial. The leach bed was then closed, weighed, installed,

and allowed to drain for one hour to release any surplus water; the drained water was measured, and

subtracted from the water added, to calculate the saturation capacity of the waste. Water (535 ml) was

then introduced into the leach bed by setting the three-way valve to “Manual” for one minute; the

volume collected at the bottom of the leach bed was measured at intervals. At the conclusion of the

tests, the waste was allowed to drain for 24 hours, the leach bed weighed again, the headspace

measured, and the total weight and density of saturated waste determined. The waste composition is

shown in Table 7.1 and the results of the permeability tests in Figure 7.2. No settlement of the waste

was observed.

Several observations were made and conclusions drawn. The fibres and woodchips require a lot of

water to reach saturation, and this must be added before the waste is introduced to the digester – both

to ensure proper mixing, and to avoid

excessive drawdown of the leachate in

Tank 2. Assuming that the food waste is

about 70% moisture (a good figure

from the literature), the final moisture

content of the waste mixture was 72%,

66% and 68% in Tests 1, 2 and 3

respectively. This indicated that, for a

solids retention time of 42 days, about

3.7 litres of free water per week would be generated, and require removal from the system but, in

practice, none was. The permeability tests show that fibre alone drained very rapidly, and all the added

water was drained within 60 minutes; fibre plus woodchips had a lower initial permeability but still

drained all the water within 70 minutes; the addition of food waste flattened the curve further, and at

60 minutes the waste still retained 28% of the added water. Nevertheless, twenty four hours later, all

the water had drained (not shown). This prolonged retention of water was thought to be favourable for

Table 7.1 Permeability Tests Test 1 Test 2 Test 3

Newsprint 0.10 kg 0.10 kg 0.10 kg Cardboard 0.44 kg 0.44 kg 0.44 kg Boxboard 0.35 kg 0.35 kg 0.35 kg Fine paper 0.20 kg 0.20 kg 0.20 kg Woodchips

0.50 kg 0.50 kg

Food waste

0.73 kg Water added 2.74 kg 3.17 kg 3.07 kg Total mass 3.83 kg 4.76 kg 5.29 kg

Volume 5.65 l 7.70 l 8.48 l Density 0.68 kg/l 0.62 kg/l 0.62 kg/l

79

hydrolysis, but there remained a concern about how the permeability might change as digestion

progressed, and how effective the bulking agent would be.

Summary (Commissioning) All the sub-systems and the digester as a whole were made leak-free with both air and water

under conditions of elevated pressure. The peristaltic pumps and the gas meters were successfully calibrated. The system as a whole was filled with water, brought to operating temperature, and run for

115 hours without incident. The permeability of the proposed mixture of substrates and bulking agent was tested and

found to have a beneficial effect by slowing initial leachate flow and slightly delaying drainage, both helpful for leachate substrate contact.

7.2 Startup of Daisy with Synthetic Feed - Tanks and UASB operating

The first stage of startup was carried out with a synthetic feed, using the tanks and UASB only, with the

leach beds empty but the valve sequencer operating, so the leachate flowed through the leach beds

directly to Tank 1. The inoculum in the UASB consisted of sludge from Maryland Farms (MF), an on-farm

digester in Ontario, licensed to process food waste from commercial sources but only processing farm

waste at the time the samples were taken. Daisy was operated in this mode continuously for 59 days.

Biogas composition was measured and BMP tests were run. She was then decommissioned, cleaned,

and thoroughly leak-tested. The general conclusion from this phase was that all the components of the

system performed more or less as planned and that startup with synthetic feed could begin.

0

100

200

300

400

500

600

0 20 40 60 80

Leac

hate

dra

ined

mL

Elapsed time min.

Fibre + woodchips

Fibre + woodchips + food waste

Fibre alone

Figure 7.2 Waste permeability tests; FB alone, FB+BA, FB+BA+FW

80

The composition of the synthetic feed is shown

in Table 7.2. Consisting of a mixture of soluble

starch, carboxymethyl cellulose (CMC), peptone

and lipids, it was adapted from Guerrero et al.

(2009a) and formulated to approximate the

composition, on a source-of-COD basis (i.e

carbohydrates, cellulose, protein and lipids) of a typical solid waste feed. Thirteen litres were prepared

in one-litre batches using a

blender. The end result was

a solution/suspension of the

components which was

introduced into Tank 2 as a

concentrate. Daisy’s heaters

were turned on and brought

to 37OC; the valve sequencer

was started, then feed was

introduced into Tank 2 from

time to time (over a period

of 59 days). Biogas

Table 7.2 Composition of Synthetic Feed On a COD basis:

40% starch 40% carboxymethyl cellulose 10% peptone 10% vegetable oil

Concentrated solution ~ 50 gCOD per litre

0.0

2.0

4.0

6.0

8.0

0 5 10 15 20 25 30 35 40 45 50 55

Biog

as P

rodu

ctio

n (li

tres

/day

)

Elapsed time - days

Feeding

GM1 (UASB)

GM2 (Tank1and Tank 2)

Total

g COD added44.630.3 10.3 49.6 49.6 81.0

Figure 7.3 Biogas production with synthetic feed showing feeding events, amount added, UASB, Tanks 1+2 + 6 LBS

0

1

2

3

10 15 20 25 30 35 40 45 50 55 60

COD

(g/L

)

Elapsed time - days

UASB inflow

UASB outflow

Figure 7.4 Concentration of COD in UASB vs. time; concentration grows because carboxy-methyl cellulose v. slow to convert

81

production was recorded. Samples of leachate were taken at intervals of 3 to 5 days and analyzed for

TS/VS/COD using standard methods. Towards the end of the trial period, samples of leachate were

analyzed for pH and alkalinity for the purposes of method comparison (see Table 6.3). Biogas was

sampled at each gas meter and its composition analyzed using GC-TCD. Four litres of MF sludge at 72

gCOD.L-1, was introduced to the UASB.

Biogas production for the entire 59 day period is shown in Fig. 7.3 which also shows how much feed

was added and when. The COD concentration in the leachate for the same period is presented in Fig.

7.4 and shows a rise in COD concentration to a plateau, then another rise throughout the test period.

The total biogas yield was 84.9L at 53.2% CH4 and 297K.

A BMP test (BMP 1) was performed in triplicate on six samples and two controls; each of the four

synthetic feed components, plus the whole feed, were tested with MF sludge, and the whole feed

tested with an alternative inoculum, Tembec granules, from an anaerobic digester treating pulp mill

effluents located at Temiskaming. The details of this test are provided in Appendix E.

Two principal conclusions were drawn from Daisy’s operating data; addition of fresh feed created an

immediate spike in biogas production which quickly subsided, and the amount of gas produced was less

than would be expected if most of the feed added had been consumed. It is believed that the starch

was being rapidly consumed but that the CMC was, for reasons unknown, not being digested, leading to

the steady rise in COD concentration. The results of the BMP test (Figure 7.5) confirmed this conclusion,

showing that even after 65 days only 23% of the CMC was digested but 95% of the starch and 86% of

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

0 10 20 30 40 50 60 70

Biog

as p

rodu

ced

mL

(at 2

97K)

Elapsed Time - days

Tembec + SF

MF sludge +SF

MF sludge +starchMF sludge +CMCMF sludge +peptoneMF sludge +lipidsTembec

MF sludge

Max.

Figure 7.5 Synthetic feed BMP results. Tembec granules + SF; MF sludge +SF, +starch, +CMC, +peptone, +lipids. Tembec granules alone, MF sludge alone.

82

the peptone had been digested (based on the theoretical maximum biogas production of 76 mL for all

samples). The MF sludge showed lag times of 10 to 20 days, depending on the substrate. The Tembec

granules showed no lag time, a much higher initial rate, and a more typical curve.

Summary (Start-up, synthetic feed) Tembec granules are a superior inoculum to Maryland Farms sludge. Ripley’s procedure is the preferred method for measuring alkalinity. GC-TCD worked well for biogas analysis. The synthetic feed was suitable for the purposes of initial start-up. Datalogging of temperatures and gas production was reliable. Daisy performed as planned, all system components functioned well and no operating

problems were experienced.

7.3 Startup of Daisy - Tanks and Leach Beds operating

To start up the leach beds, three were charged simultaneously with a mixture of FW, FB, and BA plus 90

g of Tembec granules in each leach bed as inoculum. The feedstock composition is shown in Table 7.3.

This mixture approximates the average composition of commercial organic waste in the City of Ottawa

(MacViro Consultants, 2007). The amount of water added, comprising 62.6% of the total, is typical of

“field capacity” (saturation) of solid waste within a landfill. The food waste came from the green bin

program operated by Miller Waste Systems for the Region of Durham and during the course of the

research programme Miller delivered two shipments. Both were collected early in the month of May;

the date was chosen so that it was warm enough to successfully sort, coarsely shred (shear shredder

with 4 inch knives) and repackage the bulk material, but cool enough to minimize premature

degradation. The food waste was stored at -20°C. The fibres and bulking agent were also supplied by

Miller; they came from the York Region Materials Recovery Facility operated by Miller. Lastly, the

bulking agent was ash wood ground by Miller using a commercial Roto-Chopper.

Table 7.3 – Composition of Digester Feedstock per Leach Bed– Period 1

Category Component Code Quantity g. Percent of TS Percent of VS Percent of COD Fibre Cardboard CB 440 33.0 33.5 32.5 Boxboard BB 350 28.1 27.3 27.5 Newsprint NP 110 8.3 9.2 9.6 Fine paper FP 200 15.3 14.5 13.2 Sub-total fibre FB 1,100 Food waste Green bin waste FW 675 15.3 15.4 17.2 Bulking agent Ash wood chips BA 500 Water 3,800 Total 6,075

83

All three leach beds were started simultaneously on February 17th 2015 (the UASB was idle); since

inoculum had been added to each leach bed, biogas production began quickly, but also ended very

quickly. By February 20th all biogas production had ceased. By February 24th the alkalinity ratio had risen

to 3.29 and the pH had dropped to 5.5. Sodium bicarbonate addition returned the pH to 6.90 but the

alkalinity ratio remained very high at 2.14 indicating incipient reactor failure. The system had essentially

failed because of a lack of inoculum; it is believed that Daisy was either producing fatty acids far more

quickly than she could consume them, or the food waste contained more soluble COD than the

inoculum could convert. At this point the immersion heater in Tank 2 suffered a progressive failure as a

result of fouling; the system was immediately shut down for repair. External heat tapes were installed

in place of immersion heaters; these were tested with water in the system and found to be satisfactory;

the rate of heating was significantly lower than with the immersion heaters but, since the objective is

temperature stability within a narrow range, this was of no concern. More details can be found in

Section 5.1.3. The modifications took four weeks to complete. During the shut-down the waste was left

sealed in leach beds 1, 2 and 3.

Summary (Start-up, solid waste) The alkalinity ratio responded to the process upset as intended, but under the prevailing

conditions and given the trial and error nature of the operation at this stage, there was no feasible way to prevent reactor failure.

The leach beds worked well, in that the recirculating leachate led to hydrolysis of the feed and a build-up of soluble/suspended COD in the system.

Direct heating does not work in this environment because suspended particulate matter leads to fouling of the heater elements and failure of the bulkhead fitting.

7.4 Startup of Daisy – Entire System operating

This phase of the start-up process ran from March 23, 2015 to July 5, 2015 (104 days) and was carried

out with ten specific objectives:

1. Test each component of the system under typical operating conditions for an extended

period

2. Test the system as a whole

3. Establish feeding protocols

4. Establish digestate removal and sampling protocols

5. Establish sampling and analytical procedures; location, sample size and frequency

6. Identify and correct any defects

7. Learn the interrelationships among variables

84

8. Refine operating practices

9. Gather as much data as possible, and complete a preliminary assessment of performance

10. Establish a baseline of digester performance from which to plan and conduct the balance of

the research programme.

Process operation – Period 1 The original three leach beds, designated S.001, S.002 and S.003 (all the leach beds were numbered

sequentially in this manner) were reactivated on March 23, 2015 (Day 0) after a 28 day hiatus. Details

of the contents and history of S.001 to S.087 can be found in EDF5. The leachate, which had been

stored at 4°C during the shut-down, was returned to the Tanks and brought to operating temperature.

The headspace of the entire system was flushed with 20% CO2/80% N2; argon was used subsequently

because it is heavier than air and better able to maintain anaerobic conditions in open vessels.

Operating days were deemed to begin at 12.00.00 midnight and end at 11.59.59 pm; Daisy ran

throughout on eastern daylight time. On Day 1, leach beds 1, 2 and 3, plus Tanks 1 and 2, were

operating. Because at the time of the shut-down the pH of the leachate was low, an aliquot was titrated

to pH 7 to determine buffering requirements and 90 mL of saturated NaHCO3 solution was added to

each tank. On Day 2 the UASB was charged with 12 L of Tembec granules, topped up with DI water, and

the heater turned on. P1 started delivering leachate to the UASB at 13.50h.

From this point on feedstock components (FW, FB and BA) were sampled and analyzed periodically for

TS, VS and COD; FW was analyzed weekly or bi-weekly, and FB and BA approximately monthly. For

several months only the “wet” method was used. When it was superseded by the “dry” method, DG

samples were re-analyzed for TS/VS/COD.

The three remaining leach beds were filled and installed at one-week intervals, using the same

feedstock composition (Table 7.3). The fibres were weighed dry and mixed in two 20 L pails, water was

added and mixed with the fibres. The frozen waste was thawed at 4oC over the weekend then foreign

matter (e.g. broken glass and pottery, large bones, bits of metal, plastic and “windables”) were

removed by hand. The prepared food waste was mixed with water at a 2:1 ratio by weight, processed in

the blender at low speed, then added to, and mixed with, the fibres using large stainless steel spoons.

The mixture was then ladled into the leach bed and gently levelled and packed down by hand. The

geotextile was installed on the top and the lid and its o-ring secured in place. A pointed stainless steel

‘prodder’ was used to create a pilot hole in the waste, and the RTD inserted to its fullest extent. The

leach bed was then connected to the manometer and pressure tested at about 50 cm WC using argon.

All the fittings were checked with “snoop”; if leaks were found the fittings were reseated and tightened

85

and the process repeated until the leach bed was gas tight. The headspace was then flushed with argon

for five minutes and the leach bed re-pressurized to 50cm WC and installed in Daisy; The shut off valves

in the quick disconnects made it possible to do this without loss of pressure; the gas pressure in the

GM1 and GM2 systems was then measured (12 cm WC was normal). This leach bed pressurization step

was first adopted on May 11, 2015 (S.010). Prior to that, each newly-installed leach bed had caused the

internal pressure within Daisy to drop and gas measurement to briefly cease until working pressure was

re-established.

From Jun 12 2015 onward (S:015), the pre-sorted food waste was no longer blended before being

mixed with the fibres, to determine if blending had any effect on its digestibility; none was observed,

and biogas production remained unchanged.

The PLC was programmed to open each valve for one minute and six seconds, close, pause for six

seconds, then open the next valve, and continue in this way until the cycle was complete. This

sequence was repeated every 30 minutes. P3, feeding the leach beds, was set at 167 rpm, or 510

mL/min. Thus each leach bed received 560 mL of leachate twice per hour. The speed of P1 governs Vup

in the UASB. It was tested over a

range of settings, as shown in

Figure 7.6. Further adjustments

were made later, and the final

setting of 0.15 m/hr was arrived at

in Sep 2015. Changes in Vup were

found to have little influence on

UASB performance measured as

biogas production in L.h-1.

Inevitably, some of the granular sludge did find its way out of the UASB and into the tanks and leach

beds and thus served as inoculum throughout Daisy. P2 operation was varied through the startup

process, both in speed and direction, to find the most suitable arrangement to keep the system in

balance. Thereafter, in normal operation, P2 ran at 100 mL/min, pumping from T2 to T1.

The heaters were set as follows: T1 at 37°C, T2 at 39°C and the UASB at 37°C. The elevated temperature

of T2, the leach bed feed tank, was chosen to help ensure that the leach bed temperatures were in the

desired operating range of 35-37°C since, at this point, the leach beds were not externally heated.

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

0 10 20 30 40 50 60 70 80

Upf

low

Vel

ocity

m/h

r

Elapsed time in Days

Figure 7.6 UASB Upflow velocity – finally settled at 0.15 m/h

86

The SRT was set at 42 days. At the end of the digestion period the spent leach bed was drained in situ

for about 30 minutes, with its valve sequencer switch in the closed position. It was then removed,

weighed, and placed on a stand to drain ex situ for 24 hours. A freshly filled leach bed was immediately

installed in its place.

Digestate sampling and analysis The day following the leach bed exchange, the spent leach bed was weighed for a second time, then

opened. The freeboard was measured to determine the degree of settling; the general appearance of

the digestate (DG), for example uneven settling, any indication of flooding, or any sign of leachate

channeling (Plate 13), was recorded on the Leach Bed Checklist (Appendix B). The leach bed was

emptied progressively; four samples of DG were taken for TS/VS analysis from four different depths (-

13cm, -20cm, -25cm and -28cm) and one sample (-25cm) for COD analysis. A 400 g sample of DG was

put in a freezer bag and stored at -20°C. The remaining DG was stored in a large bag at +4°C for

subsequent aerobic curing tests (see Chapter 9, Section 9.3). Finally, the leach bed was dismantled,

thoroughly washed, reassembled and left on the stand to dry.

The drained leachate was saved and returned to Daisy. Initially

the drained leachate was stored in the fridge and added to the

next fresh leach bed the following week. From December 1st,

2015 onwards (S.030), drained leachate was returned directly

to T1. As a general rule, larger volumes of liquid (leachate or

water) were introduced to the tanks in increments of 150 mL or

less to minimize hydraulic disturbance.

From beginning to end, the process of draining and removing a

leach bed, then filling and installing its replacement, takes

about 2 hours in total and entails about 50 individual steps;

these were incorporated into the Leach Bed Checklist (Appendix B). The physical swapping of the leach

beds takes less than two minutes.

Leach Bed Permeability The drainage properties of the leach beds were measured by discharging 500ml of leachate into each

leach bed in turn (under normal operating conditions) and measuring the rate at which it drained, using

a stopwatch and measuring cylinder. The leachate withdrawn was returned to the system. A true

permeability test would measure flow rate through the leach bed under conditions of constant head. In

Daisy, this is impractical. The function of the bulking agent, in this case woodchips, is to enhance the

Plate 13. Signs of partial leach bed flooding

Overflow tube

Digestate deposits on the walls Geotextile

87

permeability of the material in the leach beds to maintain drainage and prevent plugging. To minimize

the variability of Daisy’s feedstock ground ash wood was chosen because it is in plentiful supply. It is

also important that the bulking agent not be so coarse as to cause channeling through the beds. The

material originally supplied by Miller,

and used without further size

reduction, worked very well. A second

shipment appeared to be too coarse

and its use in one leach bed (S.017)

proved it to be so; temperatures in

this bed were always a degree or two

lower than the others, and a

permeability test of all the leach beds

was carried out in an attempt to

measure the effect. Fig. 7.7 shows the

permeability curves for all six leach

beds measured on July 6th 2015; S.017 drained very rapidly, and though the other five leach beds were

nominally the same, S.012 was quite different with a very long lag time until leachate began to to drain

(14 minutes), and only managed to drain all the leachate added by the 30 minute mark, just before the

next cycle began. This behaviour suggests incipient plugging, and the flat temperature profile for this

leach bed, with only small variations, is consistent with this hypothesis. When S.012 was removed and

opened, it was clear that it had partially flooded and the liquid had risen to within 2cm of the overflow,

further confirming these conclusions (Plate 13). The bulking agent did its job, but the leach beds must

be carefully watched to make sure that they don’t plug, and that channeling does not occur. It must be

pointed out here that a subsequent, much more exhaustive, attempt to use this same technique to

quantify another, more subtle, BA behaviour problem was completely unsuccessful (see Chapter 8,

Section 8.3.3).

Before further use (in S.018, S.019 and S.020), the second shipment of woodchips was ground up in the

blender; a third shipment BA3 proved to be suitable as-received.

Figure 7.7 Leach bed permeability test – BA Batch 2 in S.017 BA Batch 1 in S.012 to S.016

88

Leachate sampling and analysis Leachate samples were taken initially

from valve VP04 and subsequently from

VU15 (the outlets from T1 and T2

respectively). Initially this was carried out

daily, and the samples analyzed for pH

and alkalinity (Figure 7.8). The graphs

show the decline in alkalinity ratio, and

the rise in pH, after the restart on March

24th (Day 1), followed by very stable

performance. The alkalinity ratio rises

slightly with the addition of fresh waste

and then declines again within a day, in a

fashion consistent with the rise and fall in

COD concentration and biogas production.

Once digester stability had been

established, sampling frequency was

reduced to four times per week -

immediately before replacing a leach bed

(initially Monday morning, subsequently

Tuesday morning), late afternoon on the

same day, early the following morning, and

lastly on Friday morning, and analyzed for pH, alkalinity, TS, VS, and COD. This schedule was developed

to track the short-term changes in COD that

occur when fresh feedstock is added (Figure

7.9), and likewise the alkalinity ratio. The

concentration of COD in the leachate

responds directly to the addition of fresh

waste; immediately before a new leach bed is

added, the concentration is at a low point and

reaches a high point one day later; the short-

term pattern of biogas production tracks this

very closely. The TS/VS results were also used

6.306.406.506.606.706.806.907.007.107.207.30

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

0 50 100 150

pH

Alka

linity

Rat

io

Elapsed time - Days

Alkalinity Ratio pH

Figure 7.8 pH and Alkalinity Ratio – recovery from initial failure followed by stable operation

0

1

2

3

4

5

6

7

8

9

0 50 100

COD

Conc

entr

atio

n (g

/L)

Elapsed time - Days

VP04

VU15

high after fresh waste added

low before fresh waste added

excess COD being consumed in UASB -see peak in GM1 curve Fig.6.12

Figure 7.9 Leachate COD Concentration – sharp fall following restart; Daisy recovering from failure event

0.00

1.00

2.00

3.00

4.00

5.00

6.00

0 10 20 30 40 50 60 70

Inor

gani

c sa

lts g

/L

Elapsed time - Days

Figure 7.10 Recirculating inorganic salts – showing very little variation

89

to measure the recirculating load of inorganic salts within Daisy to ensure that there is no long-term

build-up (Figure 7.10). The results shown lie in a fairly narrow range from which, as we shall see, they

did not deviate greatly throughout the entire research programme, except for a slow steady rise during

the last 100 d.

Aerobic Composting the initial plan to use the seventh leach bed for aerobic composting of the digestate failed because the

quantity, about 5 kg wet weight, was too small to maintain the required temperature of 50°C. As a

consequence, all the digestate was stored at 4°C, and accumulated until we had sufficient material for

one larger scale compost experiment to be conducted by Miller Waste Systems (see Chapter 9, Section

9.3).

Data Logging The Datalogger was initially set up to record temperature every 15 minutes in seven leach beds (six

anaerobic and one aerobic), T1 and T2 and the UASB. When it became clear that aerobic curing in the

seventh leach bed was not a practical proposition this RTD was relocated to GM2 to record the

temperature at which biogas was being measured, for the purposes of correction to STP. The tip-count

of GM1 and GM2 was recorded every 5 minutes, and the hourly total recorded automatically. The data

were downloaded daily as a single comma separate variable (CSV) file, and converted to an Excel file

using a macro which tabulated data summaries and created five 24 hour graphs:

Leach bed temperatures

T1, T2 and UASB temperatures

GM2 temperature

Hourly gas production - GM1, GM2, and the total

Cumulative gas production - GM1, GM2, and the total.

The macro was also used for downloading data over more extended periods of time.

Biogas measurement, sampling and analysis After about four weeks of operation the process stabilized, based on biogas production. Figure 7.11

shows biogas production in L.hr-1 for an eight-day period from May 19th to May 26th, (Day 50 to Day 57)

which includes two leach bed replacement events. The shape of the biogas production curve, which it

soon became clear is very typical, shows an immediate response to the fresh waste – particularly the

easily-digested food waste.

The peak is reached within 7 hours of feeding but takes about another 40 hours to subside. Production

slowly declined to about 2.2 L.hr-1 before the next addition of fresh feed. Total gas production from

90

May 19th to May 26th was 451 L (at STP). Based on the assumption that the peak is primarily generated

from rapidly-digesting food waste, and the baseline gas production, presumably from the more slowly

degrading lignocellulosic wastes, is 2.2 L.hr-1, then an estimated 340 L..wk-1 (75.5%) comes from ligno-

cellulosic waste and 111 L .wk-1 (24.5%) comes from food waste. This suggested that there are two

overlapping, and presumably complementary, digestion processes taking place.

The differences in the gas production curves that were observed later in the research programme

largely related to the height and breadth of the initial peak, and the level to which biogas production

descended before the next leach bed replacement.

0

100

200

300

400

500

600

700

0 2 4 6 8 10 12 14 16

Biog

as p

rodu

ctio

n l/w

eek

at 2

97K

Total GM1 GM2

10 weeks' operation at 6 wk SRTAccumulated COD consumed

Last of original leach beds replaced

Figure 7.12 Weekly biogas production – Period 1; 10 Wks of stable operation from Wk 6 to Wk 15 inclusive. Initial peak from conversion of COD from first 3 LBs, accumulated during shutdown

Elapsed time - weeks

0123456789

10

Biog

as l

itres

/hr

Elapsed Time - Hours

GM1 GM2 Total

Figure 7.11 Biogas production 2015/05/19 to 2015/05/26 (GM1 – UASB, GM2 Tanks and LBs) Two feeding events 7d apart; shows sharp rise and slow decline in biogas production; consumption of fresh FW

91

Biogas production for the entire start-up period is summarized on a weekly basis in Figure 7.12. During

the first 5 weeks, Daisy was consuming the accumulated COD from the initial, failed, start-up and this

can be seen in the relatively high biogas output in GM1. By the beginning of May (Week 6), Daisy had

begun operating at a remarkably steady rate of about 460 litres of biogas per week (at STP). For the

next 10 weeks methane production was 245.7±8.7 L.wk-1 (STP); error equals one standard deviation.

The stabilty and consistency of the process are apparent from Figures 7.11 and 7.12; this formed a solid

base for the next phase of the research programme.

Samples of biogas were taken twice per week from each of six leach bed sampling locations (VG13 to

VG18), and two gas meters (VG23 and VG24), on Tuesdays and Fridays between 16 June and 31 July

2015. They were analyzed for methane and carbon dioxide using a GC-TCD; the results are shown in

Table 7.4. It was observed that the methane content in GM1 (UASB) was always higher than that in

GM2 (Leach beds and Tanks), and that the average methane content of all the leach beds was

essentially the same as that of GM2. This was not unexpected, since all the leach beds share a common

gas manifold and all discharge through GM2.

From this point on gas sampling and analysis

was limited to GM1 and GM2 because the

hydraulics of Daisy made it impossible to

obtain truly representative samples of the

biogas from individual leach beds. This is

discussed further in Chapter 8, Section 8.4.

The composition of biogas varied through

the digestion cycle, with higher methane

content immediately after fresh waste was

added (Table 7.5). The causes of the

variations are presumed to be related to the

high energy content of readily digested food waste components but this is not fully understood. The

long-term weighted average methane content of the biogas was determined to be 52.4%. During Period

1 (Weeks 6 to 15 inclusive

from May 3rd to the end of

start-up on July 5th) Daisy

produced 4,656 L of biogas,

or 2,457 L of methane, at

STP. Expressing this in terms

Table 7.4 Biogas Percent Methane Content

Date Day GM1 GM2 GM2 Avg GM2 STDEV n 2015/06/16 85 60.8 49.5 50.2 0.75 6

2015/06/19 88 53.2 49.0 48.4 1.3 6

2015/06/23 92 58.2 49.8 50.5 1.5 6

2015/06/26 95 54.1 48.5 49.1 0.74 6

2015/06/29 98 57.8 46.2 46.7 1.7 6

2015/07/03 102 51.9 46.1 47.4 1.1 6

2015/07/07 106 58.7 53.7 55.2 1.9 6

2015/07/14 113 52.5 50.6 51.3 0.52 6

2015/07/11 110 63.0 50.7 48.7 2.6 6

2015/07/17 116 55.2 53.4 53.6 0.57 4

2015/07/21 120 65.3 53.9 51.7 2.1 6

2015/07/28 127 55.1 46.5 46.1 1.4 6

2015/07/31 130 54.0 50.0 50.1 0.48 6

AVG

56.9 49.8 49.9

Table 7.5 - Biogas Composition vs. Feeding Cycle Date GM1 GM2

%CH4 %CO2 %CH4 %CO2

Feed Jul 14 2015 63.0 35.7 50.7 43.6 Jul 17 2015 55.2 42.8 53.4 44.1

Feed Jul 21 2015 65.3 35.7 53.9 46.5 Jul 28 2015 55.1 35.1 46.5 42.6

92

of COD, at 350 mL of methane per 1.0 g COD destroyed, shows that Daisy converted 7,020 g of COD to

methane.

Feedstock destruction efficiency and COD mass balance It was decided from the beginning that the measurement of Daisy’s performance would be based

primarily on COD measurements. The analytical task to do this is more challenging and time-consuming

than the more usual VS measurements, but the calculation of the mass balance is more robust and

useful because it is essentially on the basis of electron equivalents and does not rely upon knowing the

oxidation state of the substrates. Nevertheless, destruction as TS and VS were measured throughout

for the purposes of comparing the results of this study with the published literature which more

commonly uses TS/VS rather than COD measurements. The substrate destruction efficiency was

determined as mass, volatile solids and COD on a weekly basis from the analysis of the digestate. It was

calculated in two ways, with and without bulking agent. The former is the more accurate and simpler to

calculate, but the latter measures substrate destruction, one of the principal objectives of the research,

and is thus important. Because a small amount of the bulking agent is digested, and contributes to

biogas production, the destruction efficiency of the bulking agent alone must be determined in order to

calculate substrate destruction efficiency separately. Data from the coupon technique (Section 6.1e)

was used to make this determination; it showed that, depending upon operating conditions, the

destruction of bulking agent ranged from 2% to 10% and averaged approximately 7%. The two

calculations, with and without bulking agent, are explained in Figure 7.13 using the digestate from

S.012 as an example. The coupons were first used in S.015, so the value for Y came from subsequent

analytical results. Table 7.6 shows the destruction efficiency results for Period 1, calculated both with

Using S.012 as the example:

Method 1 – Including BA

Destruction Efficiency = (CODFW+FB+BA – CODDG)/CODFW+FW+BA*100%

Destruction Efficiency = (2096 – 1180)/2096*100 = 43.7%

Method 2 – Excluding BA

Destruction Efficiency = (CODFW+FB – (CODDG – [CODBA (100 – y)]/100))/CODFW+FB*100%

Destruction Efficiency = (1452 – (1180 – [645(100-7.04)/100]))/1452*100% = 60.0%

Y = BA destruction efficiency = 7.04% in this example

Figure 7.13 Sample calculation of COD destruction efficiency, with and without bulking agent

93

and without bulking agent. These data formed the baseline against which to compare Daisy’s

performance under different operating conditions and feedstock composition.

To assess Daisy’s performance, and the dependability of the analytical data, a mass balance calculation

was performed for Period 1; the results are presented in Table 7.7.

Table 7.6 Substrate Destruction Efficiency – Period 1

Serial No.

Percent Destruction Efficiency

Mass VS COD

With BA Without BA With BA Without BA With BA Without BA

S.008 41.9 55.7 46.1 63.1 43.5 59.7

S.009 37.4 49.5 41.6 56.6 39.2 53.5

S.010 35.5 46.8 39.6 53.8 37.3 50.8

S.011 38.1 50.4 42.0 57.3 39.3 53.6

S.012a 40.5 53.9 46.3 63.4 43.7 60.0

S.013 42.3 56.3 47.1 64.6 44.6 61.3

S.014 36.8 48.7 42.2 57.6 39.5 53.9

S.015b 36.1 47.7 40.7 55.3 37.9 51.5

S.016b 34.9 46.0 39.7 53.9 36.8 50.0

S.017b 30.4 39.7 34.9 47.1 31.8 42.9

Average 37.4 49.5 42.0 57.3 39.4 53.7 aSee calculation in Figure 5.3.14 b the FW in these Leach beds was not blended

Table 7.7 COD Mass Balance at Start-up – Period 1

Week Serial No.

COD g Methane

(a)

Substr. in

(b) DG out

(c) Destr. (d) conv. cells

(e) Conv. CH4 (f) Vol. out L (g) as COD out g.

Mass Bal. %

wkly cum. wkly cum. wkly cum. wkly cum. wkly cum.

6 S.008 2100 1180 913 910 73.0 73.0 840 840 246 246 702 702 83.5

7 S.009 2100 1280 822 1740 65.8 139 756 1600 253 498 721 1420 95.4

8 S.010 2100 1310 783 2520 62.6 201 720 2320 248 746 708 2130 98.3

9 S.011 2100 1270 824 3340 65.9 267 758 3070 237 983 677 2810 89.3

10 S.012 2100 1180 917 4260 73.4 341 844 3920 254 1240 726 3530 86.1

11 S.013 2100 1160 936 5190 74.8 416 861 4780 232 1470 664 4200 77.2

12 S.014 2100 1270 828 6020 66.3 482 762 5540 243 1710 695 4890 91.1

13 S.015p 2100 1300 794 6820 63.5 545 731 6270 238 1950 680 5570 93.1

14 S.016p 2100 1330 772 7590 61.8 607 710 6980 261 2210 747 6320 105.1

15 S.017p 2100 1430 668 8260 53.5 661 615 7600 245 2460 700 7020 113.9

Totals 20970 12710 8260 660 7596 2460 7020 92.4

c = a - b d = c*0.08 e = c - d g = f/0.35l.gCOD-1 Mass bal. = 7020/7596*100% = 92.4 %

mass balance = g/e*100% = 92.4% p the FW in these leach beds was not blended

94

The principal conclusion of these calculations is that, over a 10 week period, assuming 8% of the COD

destroyed was converted to new cells (Rittmann & McCarty, 2001), 92.4% of the remaining COD

destroyed could be accounted for as methane produced. These results were also plotted in Figure 7.14,

which shows:

gCODdestroyed (c)

gCODdestroyed converted to gCODnew cells (d)

methane produced converted to gCODmethane (g)

gCODproducts (e) [=(d) + (g)]

The trendlines for (c) and (g) are plotted; in a perfect mass balance the ratio of their gradients, c/g = 1;

in practice c/g = 767/834 = 0.92 or 92.0% compared to 92.4% in Table 7.7.

One feature of the way Daisy operates complicates the reconcilliation of biogas produced and substrate

destroyed; biogas is measured essentially continuously and represents the aggregate output from six

leach beds, two tanks and one USB, measured by two gas meters. Conversely, the measurements of

digestion efficiency, and COD consumed, are single weekly figures derived from the analysis of the

contents of a single leach bed. It is therefore not possible to make a direct correlation between the

biogas output of a single week with the COD destroyed within the leach bed removed at the end of that

same week. Putting it differently, at any one moment six leach beds, each at a different stage of

digestion, are contributing to the biogas being measured. This discrepancy varies with changes in

Graph (e) equationy = 767x - 3760

R² = 1.00

0.0

1.0

2.0

3.0

4.0

5.0

6.0

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

6 8 10 12 14 16

Met

hane

m3

Cum

ulat

ive

COD

gram

s

Elapsed time - weeks

(c) g COD destr.

e) g COD destr. conv. to methane =c*0.92

(g) g COD out conv. from L methane =f*1000/0.35

(d) g COD destr. conv. to cells =c*0.08

(f) Litres methane prod.

c

d

e

f

g

Mass bal. = 767/834 = 92.1%

Figure 7.14 COD balance week 6 to week 15 cumulative; COD products (methane +biomass) e) ∕COD destroyed c) *100 = % mass balance

Graph c) equation Y = 834x – 4090 R2 = 1.00

95

operating conditions, including FW addition. This only really matters when looking for short term

relationships; under steady operating conditions it is not material.

Wastewater Daisy produces no free wastewater. The only water leaving the digester is the water within the

digestate; approximately 3.8L per week, and as water vapour in the biogas amounting to about 2.5% by

volume (140 g over the 15 week start-up period). When liquid samples are withdrawn they are replaced

in like quantity with de-ionized water. Daily records of leachate withdrawal and replacement were

maintained throughout. Tests were conducted on the feedstock and the digestate to determine if the

lack of free waste water was causing an increase in the recirculating load of inorganic material in

solution or suspension (Figure 7.10). The initial concentration, when Daisy was restarted, was greater

than 5.0 g.L-1. It rapidly dropped to 3.49 g.L-1, and for the next 60 days remained between 2.99 and 3.64

g.L-1. The total inorganic content of the feedstock in each leach bed was determined to be 160 g, and

the same determination on four samples of digestate averaged 171 g, thus the inorganic components

of the feedstock/digestate, in and out, were in balance.

Other observations and operating challenges Figure 7.15 shows Daisy’s Daily Log for June 8 2015, and Figure 7.16 shows the data logger output for

the same day, converted using the macro and reached through the data file hyperlink. This particular

date was chosen because it shows the effects on performance of a leach bed of fresh waste inserted at

16.10 h. Figure 7.16 a) shows the leach bed temperatures; the fresh leach bed (LL02, S.014) can be seen

rising to meet the others, all of which can be seen to be rising slightly. Figure 6.16 b) shows the tank

temperatures rising also. The slightly spiked lines in these two graphs track the valve sequencer cycle

and represent real temperature fluctuations. Biogas production begins to rise within an hour of

installation of the fresh leach bed - Figure 6.16 c) - peaks about five hours later, then begins its slow

decline. During this phase the percentage of biogas produced in the UASB climbs sharply, reflecting the

surge in soluble COD associated with the fresh food waste. Lastly, Figure 6.16 d) shows cumulative gas

production, and linear trend lines with poor regression coefficients. On the other six days of the week

the regression coefficients are generally 0.97 or better; the hourly average biogas generation rates

from GM1 and GM2, and the total, were always recorded in the Daily Log.

Daisy is simple to run; three pumps and six automatic three-way valves are the only moving parts and

all worked flawlessly through commissioning and startup. When it became apparent that a pump

failure could be a serious setback to the research programme, a spare pump was purchased as a back-

up. Other idiosyncrasies were discovered, largely relating to hydraulic behaviour. For example, because

the system is always under slight pressure (12 cm of water – the head in the gas meters), a new leach

96

bed, if it is not first pressurized, will cause an immediate drop in system pressure. This can then deflate

the gas bag and syphon water out of the gas meters into the foam traps, preventing gas from exiting

the digester and leachate from entering the leach beds; this effectively shuts Daisy down. The relatively

low temperature of a fresh leach bed and its contents also contributes to this behaviour.

Accordingly, a procedure was developed to pressurize new leach beds to 50 cm wc, using argon, before

installing them; this permanently resolved the problem. A little bit of trial and error was needed to

arrive at the right amount of pressurization; too little and the system pressure fell, too much and the

gas meter tipped prematurely. As Figure 7.16 c) shows, gas production remained smooth following the

Date: Jun 8 2015 1. Datalogger a) Data download Y/N:Y ..\..\Data File 1 - Data Taker\06 2015 Excel Format\Jun 8 2015 099017_JOB1_20150609T065811.xls b) Comments on data output 2.43 l/h Total 2.11 l/h GM2 Re-seat RTD fitting on LL02 Done 0.33 l/h GM1 2. Other Data 3. Samples taken Material Code Location Date/Time Quantity (ml or g) Preserve Analyze for* Comments LT-VU15-08/06/15-14.00-15 15ml Fridge TS/VS/COD LT-VU15-08/06/15-14.00-50 50ml pH/alk LT-VL05-08/06/2015-10.00-15 15ml Fridge TS/VS/COD 4. Material added or removed (other than samples) Material Code (plus A or R) Location Time Quantity (ml or g)** Comments Water added T2 14.05h 80ml Replaces samples

5. System adjustments Comments Removed LL02 (LB07, S.008) Installed LL02 (LB07, S.014) 6. General Comments Tried to sort out the problem with T1 heater but to no avail. Checked connectivity in heater cable - OK reprogrammed to match T2 exactly - OK(?). Sent note to Omega

Figure 7.15 Example of Daily Log June 8th 2015; data output; samples; system adjustments

97

installation of the fresh leach bed; the problem had been solved by June 8, 2015.

25

35

45

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Tem

pera

ture

Deg

. C

Elapsed Time Hrs

a) Leach Bed Temperatures2015-06-08

LB1 (degC) LB2 (degC) LB3 (degC)

LB4 (degC) LB5 (degC) LB6 (degC)

25

30

35

40

45

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Tem

pera

ture

Deg

. C

Elapsed Time Hrs

b) Tank 1, Tank 2 and UASB Temperatures2015-06-08

Tank1 (degC) Tank2 (degC) UASB Bottom (degC)

0

2

4

6

8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Biog

as l

itres

/hr

Elapsed Time Hrs

c) Biogas Production2015-06-08

GM1 GM2 Total

y = 0.33x - 1.09R² = 0.80

y = 2.11x - 4.54R² = 0.94

y = 2.44x - 5.64R² = 0.92

-20

0

20

40

60

80

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Biog

as P

rodu

ced

litre

s

Elapsed Time Hrs

d) Cumulative Biogas Production2015-06-08

GM1 GM2 Total

Figure 7.16 Datalogger output, June 8, 2015; a) LB temps; b)Tanks and UASB temps; c) biogas hourly; d) biogas cumulative GM1, GM2 and Total. Fresh leach bed installed at 14:30h. All raw data can be found in EDF1

98

EDF2 contains a complete set of Daily Logs and EDF1 contains all the raw data from the datalogger.

The hydraulic behaviour also makes it very difficult to measure biogas production from the leach beds,

either individually or collectively. To make this measurement it is necessary to separate the leach-bed

headspace from the tank headspace. When this was tried it resulted in rapid process upset in the form

of water syphoning, and had to be quickly reversed. Because the leachate is pumped to the leach beds

from Tank 2 and returns from the leach beds to Tank 1, the leachate withdrawn from T2 causes the

transfer of an equal volume of gas from the leach-bed headspace to the tank headspace to

compensate. The dynamics of this arrangement move larger volumes of biogas back and forth, in a

period of a few minutes, than the system generates during the same period, making biogas

measurement extremely difficult.

Much later in the research programme I did have limited success measuring gas production from

individual leach beds, but this could only be accomplished by cutting off leachate feed for the duration

of the test, which was generally about four hours, and isolating the leach bed entirely except for the gas

line to GM1 to prevent transfer of biogas through the waste mass. This is described more fully in

Chapter 8, Section 8.4. The control system, with the exception of the heater problem, worked perfectly,

as did the gas meters. On July 5th 2015 (day 104) after 70 days of very stable operation the startup

phase, Period 1, was determined to be complete.

7.5 Commissioning and Start-up Conclusions

The commissioning and start-up process accomplished many things and these are summarized below;

but two overarching conclusions can now be reached; firstly, this configuration of digester is capable of

reliably digesting a mixed organic waste stream for an extended period of time (15 weeks) without

process upset; Daisy worked. Secondly, the experience and data gained during the commissioning

period established a series of operational sampling and analytical procedures, and a performance

baseline against which to compare future results.

Summary Period 1, operation with solid waste in the final phase of startup, ran for 104 days. It began with 10 specific objectives; all of which were met. Biogas production at 509.5±3.5% L.wk-1 (297K) and methane production at 245.7±3.5% L.wk-1

(STP) was very stable. Methane production was unaffected when FW blending was eliminated. The following questions were answered:

o pressure loss in the system following leach bed installation was solved by pressurization o a leak testing procedure for leach beds was established

99

o a technique for measuring leach bed permeability in situ was devised o a technique for measuring substrate digestibility was developed o a recurring heater problem was finally solved.

The following problems were deferred: o small-scale aerobic curing did not work; a separate subsequent experiment with

accumulated digestate was devised. o attempts to measure biogas from the leach beds alone caused serious hydraulic

imbalance and failed; data gathered in subsequent experiments suggest that gas production occurs in all leach beds after fresh waste added.

o Daisy was found to have one operational drawback; the lack of ability to test substrate destruction efficiency versus SRT. This was eventually solved, but only in a single experiment at the very end (Week 88), when Daisy was shut down and all six leach beds, at different stages of digestion, were removed at the same time.

The following data management practices were established: o a step-by-step procedure for removing and replacing leach beds. o a complete record of the setup and contents of each leach bed. o a daily electronic log of all operating activities with a hyperlink to the daily download

from the data logger. o a continuous weekly record of biogas production. o electronic data files for all analyzes performed.

The following variables were set: o Vup o temperatures o SRT o settings for the data logger

The following knowledge was gained: o the profile of hourly gas production was established. o COD concentration correlates with the feeding cycle. o the use of the alkalinity ratio is a valuable indicator of digester stability. o the correlation between substrate destruction and biogas production was clearly

demonstrated. o a method to calculate substrate destruction efficiency, both with and without bulking

agent, was established. o Daisy produces no free wastewater o further refinement of the procedures used for feeding, digestate removal, sampling,

and analysis is possible.

Commissioning and startup went as smoothly as one could have hoped and Daisy performed as well as,

or better than, anticipated; the groundwork was laid for the main part of the research programme –

testing the hypothesis.

100

Chapter 8. Operation of Daisy - Results

8.1 Introduction

The experimental programme was planned to test the hypothesis by measuring the effects of changes

in feedstock composition and operating variables on the performance and stability of Daisy.

The initial plan was to test five variables in the following order:

a) solids retention time

b) quantity of food waste in the substrate (expressed as %CODFW)

c) leachate recirculation through the leach beds - volume and frequency

d) bulking agent addition rate

e) UASB functionality; how does it perform and is it needed

The measurement tools to be used were biogas production and composition, substrate destruction,

leachate properties – specifically pH and alkalinity, COD content and VFA concentration, and C:N ratio.

For reasons that will become clear, the programme ultimately concentrated on variables a), b), and a

variant of d).

Daisy was operated for 88 consecutive weeks, divided into six operating periods (including Period 1 –

Startup described in Chapter 7); data from Weeks 1 to 5 inclusive were omitted from long-term

calculations because of irregularities in the feeding and operating plan related to start-up. Table 8.1

summarizes the dates and durations (in weeks and days) of the six operating Periods; it also includes

the Serial numbers of the leach beds in place at the time, and a brief description of the objective for

each Period. Table 8.2 describes operating conditions prevailing in each operating Period. The principal

changes were the amount of food waste added to the fibres during Periods 4, 5 and 6, and changes in

bulking agent during Period 5.

All of the assessments of Daisy’s performance, and the major conclusions drawn, rest on three separate

sets of data:

COD measurements of substrate and digestate; these were carried out in sextuplicates rather than the usual triplicates because of the heterogeneity of the substrates.

Biogas volumes; these were converted to methane at STP using gas analysis data and temperature and pressure measurements. The average of 40 separate measurements of biogas composition was used.

101

Elemental analyzes of all substrates (duplicate tests of three separate sets of samples) were used to verify the COD methodology and results, calculate the stoichiometry of the process and predict the methane content of the biogas.

Together, these form the basis of the discussion presented in Section 8.3. The culmination of all of

these sets of data is the 83 week, COD-based, mass balance of Daisy, which shows that 100.8% of the

COD destroyed can be accounted for as methane produced plus new cell growth. The latter was

calculated as 8% of COD destroyed based on Rittmann & McCarty (2001). To the best of my ability to

determine from the literature, an AD experiment of this duration has never been accompanied by a

COD-based mass balance. Given the duration of the experiment, and the diversity of the measurements

used to obtain this result, I would argue that the mass balance lends credibility to the results and

conclusions of this thesis. The bulk of the performance data, except BMPs, synergy, permeability and

enzyme activity are summarized in EDF4.

Table 8.1 Experimental Periods

Period Dates (inclusive) Weeks Days Sub

periods LB Serial No. Description Comments

1 Mar 24 - Jul 5 2015 1 to 15 1 to 104 2 S.001 to S.017 Commissioning Data from

weeks 1 to 5 excluded

2 Jul 5 to Sep 6 2015 16 to 24 105 to 167 none S.018 to S.024 7 wk. SRT

3 Sep 7 to Oct 26 2015

25 to 31 168 to 217 none S.025 to S.031 Return to 6wk. SRT

4 Oct 27 2015 to Feb 29 2016

32 to 49 218 to 343 3 S.032 to S.048 CODFW 17% to 0% in 3 steps

Synergy observed

5 Mar 1 to Aug 22

2016 50 to 74 344 to 518 4 S.049 to S.073 Return to

CODFW 17% in 1 step.

Bulking agent effect

6 Aug 23 to Nov 28

2016 85 to 88 519 to 616 2 S.074 to S.087

CODFW to 55% in 2 steps

The results are presented and discussed from three different perspectives in three different

subsections, each one of which of tells a different part of the story.

Firstly the long view, looking at the experiment in its entirety, presents an assessment of

process robustness and stability under changing conditions, both deliberate and accidental,

using several different sets of data.

Secondly, a closer examination of performance Period by Period provides an analysis of

lengthening the SRT to 7 weeks, the synergistic effect of food waste addition on the digestion

102

of fibres, and an analysis of unexpected changes in performance apparently related to the

properties of the bulking agent.

Thirdly, an examination of short-term trends – hourly and daily – explains how Daisy responded

to short-term changes in operating conditions and defines some of the benefits of the

sequentially batch-fed design concept.

Throughout the research programme five supporting experiments were conducted to gain a deeper

understanding of Daisy’s performance and the mechanisms behind the observed behaviour; these

were:

leach bed permeability tests

leach bed biogas production tests

biochemical methane potential tests (BMPs)

enzyme assays

aerobic curing of digestate

Table 8.2 Daisy's Operating Conditions

Period CODFW %

total COD

UASB Vup m.h-1

Pump Settings rpm/ml.min-1

Temperature settings °C

Bulking Agent Batch

1 2 3 T1 T2 UASB

1 17.2 0.1/0.15 125/120 55/100 167/510 37 39 37 1 & 2

2 17.2 0.15 125/120 55/100 167/510 37 39 37 2 & 3

3 17.2 0.15 125/120 55/100 167/510 37 39 37 3

4a 12.9 0.15 125/120 55/100 167/510 37 39 37 4

4b 7.9 0.15 125/120 55/100 167/510 37 39 37 4

4c 0 0.15 125/120 55/100 167/510 37 39 37 4

5a 17.2 0.15 125/120 55/100 167/510 37 39 37 4

5b 17.2 0.15 125/120 55/100 167/510 37 39 37 5

5c 17.2 0.15 125/120 55/100 167/510 37 39 37 4

5d 17.2 0.15 125/120 55/100 167/510 37 39 37 6

6a 21.7 0.15 125/120 55/100 167/510 37 39 37 7

6b 29.3 0.15 125/120 55/100 167/510 37 39 37 8

The first two supporting experiments are incorporated in this chapter, in Section 8.3. The last three are

described in Chapter 9. The discussion of results, presented in Chapter 10, interprets the operating data

with the help of the supporting experiments. The graphs and tables are all colour-coded by period for

ease of reference.

103

8.2 Overall Performance – Robustness and Stability

Long-term performance was tracked with seven different measurements, substrate destruction (as

mass, VS and COD), biogas production, pH and alkalinity and concentrations of total inorganic salts,

sulphate, COD, and VFAs within Daisy. The data are first described, then discussed in the context of

stability and robustness.

8.2.1 Substrate Destruction and Biogas Production

In Figure 8.1, substrate destruction is expressed on the basis of TS, VS, and COD. The VS and COD

measurements correspond quite closely; TS destruction efficiency is lower because the inorganic

matter present in the waste is also present in the digestate. COD destruction efficiency is used

throughout for calculating the mass balance. The VS destruction efficiency is used primarily to make

comparisons with work published by others.

Figure 8.2 presents weekly biogas production expressed as L.wk-1 at 297K. By comparing Figures 8.1 and

8.2 it is clear that destruction efficiency and biogas production follow similar paths. The

correspondence is not exact because there are always six leach beds contributing to the biogas

measured in any one week, making it hard to achieve exact correlation between substrate destruction

and biogas production. This can be illustrated by two apparent anomalies; the sharp drop in gas

production around Week 25 (Figure 8.2) occurred during a 10-day period during which the heaters and

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60 70 80 90

Dest

ruct

ion

Rate

[%]

Leach Bed Serial Number (over 88 weeks)

TS Destruction VS Destruction COD Destruction

P1 P2 P3 P4 P5 P6

Figure 8.1 Substrate destruction efficiency as TS, VS and COD vs. time (calculated without BA). Destruction efficiency ranges from 18 to 70% on a COD basis and rises and falls with CODFW addition

104

valve sequencer were turned off and Daisy was not fed, but there was no corresponding decline in

substrate destruction during that same period (Figure 8.1). When Daisy was turned back on, the biogas

production rate rose beyond where it had been previously, but there was no corresponding rise in

substrate destruction; the six week average biogas production through this period was 483 L.wk-1, only

5% lower than in Period 1.

Table 8.3 COD Substrate Destruction Efficiency by Period

Operating Data Mass VS COD

Period Weeks Serial Nos. With BA Without BA With BA Without BA With BA Without BA

1 6 - 15 S.008 - S.017 37.4 49.5 42.0 57.3 39.4 53.7

2 16 - 24 S.018 - S.024 37.1 49.0 42.7 58.2 39.2 53.5

3 25 - 31 S.025 - S.030 37.2 49.2 43.4 59.3 39.7 54.2

4a 32 - 37 S.030 - S.035 27.8 36.0 32.2 43.2 29.1 38.9

4b 38 - 44 S.036 - S.042 21.1 28.7 23.7 33.2 22.6 31.8

4c 44 - 49 S.043 - S.048 13.4 17.5 14.8 20.0 13.7 18.6

5a 50 - 57 S.049 - S.056 36.4 48.0 39.1 53.1 37.9 51.7

5b 58 - 63 S.057 - S.062 32.0 41.9 34.1 45.9 32.5 43.8

5c 64 - 70 S.063 - S.068 32.7 43.0 36.1 48.7 32.5 43.9

5d 71 - 74 S.069 - S.073 36.8 48.6 41.3 56.3 38.9 53.0

6a 74 - 80 S.074 - S.079 41.1 54.2 46.8 63.5 41.3 56.0

6b 81 - 88 S.080 - S.087 48.3 61.8 52.9 69.4 49.9 65.3

Totals 6 - 88 S.008 - S.087 33.7 44.3 37.7 51.0 34.9 47.2

Table 8.3 is a summary, by Period, of substrate destruction efficiency expressed as mass, VS and COD,

both with and without bulking agent. The average destruction efficiencies (corresponding to periods of

stable biogas production at each level of CODFW addition) range from 18.6% to 65% (calculated without

0

100

200

300

400

500

600

700

800

900

0 10 20 30 40 50 60 70 80 90

Biog

as P

rodu

ctio

n L/

wk.

at 2

97K

Elapsed Time - weeks

513L/wk.

797L/wk.

620L/wk.Stable Biogas Production

509L/wk 414L/wk.

287L/wk.

132L/wk.

Figure 8.2 Weekly biogas production showing periods of stable biogas production at different levels of CODFW addition.

509L/wk

105

BA). Table 8.4 is a summary, by period, of methane production as L.wk-1, LCH4. kg-1CODadded and LCH4.

kg-1VSadded (all based on calculations without BA); the COD data are also presented as a percentage of

theoretical, which ranged between 15.1% at 0% CODFW (Period 4c) and 64.4% at 29.3% CODFW (Period

6b). These data are discussed again in Section 8.3.

8.2.2 Alkalinity Ratio and pH

The alkalinity ratio and pH for the entire duration of the experiment are shown in Figure 8.3. The x-axis

in this graph, and three of the next four, is in units of days to accommodate the frequency of sampling

and measurement. This graphs tell a story of long-term stability. The low pH and high alkalinity ratio

measured when Daisy was restarted at day zero (Period 1, Figure 7.8) were partially corrected by

sodium bicarbonate addition, but needed no further intervention from that point on. Apart from a

number of brief spikes in the alkalinity ratio between Days 150 and 200, it remained below 0.6 and

became progressively more stable. After Day 343 the pH rose slightly from about 6.9 to about 7.2 as a

result of increasing food waste addition. The short-term fluctuations are discussed further in Section

8.4.

Table 8.4 Stable Methane Production and Methane Yield Period Weeks Methane Production (at STP) CODadd.

kg.wk-1 LCH4.

kg-1 CODadded COD conv.

% theor.

VSadded kg.wk-1

LCH4. kg-1VSadded

Vol. L.wk-1.

AVG STDEV %STDEV n

1 6 - 15 246 8.7 3.5 10 1.450 169 48.3 1.120 219

2 15 - 23 194 33.4 17.2 6 1.450 13409 38.3 1.120 173

3 25 - 30 268 30.6 11.4 6 1.450 185 52.8 1.120 239

4a 32 - 25 198 6.1 3.1 4 1.380 143 40.9 1.070 184

4b 41 - 44 138 3.4 2.5 4 1.310 105 30.1 1.020 135

4c 46 - 49 63 3.9 6.2 4 1.200 52.7 15.1 0.950 678

5a 50 - 57 235 25.7 10.9 5 1.450 162 46.2 1.120 210

5b 58 - 63 244 29.7 12.2 6 1.450 168 47.9 1.120 217

5c 64 - 69 224 21.4 9.6 6 1.450 154 44.1 1.120 200

5d 70 - 74 245 19.0 7.7 5 1.450 169 48.2 1.120 2197

6a 75- 80 290 21.0 7.2 6 1.540 189 54.0 1.180 246

6b 82 - 87 384 7.1 1.8 6 1.700 225 64.4 1.300 296

COD conv. % theor. = LCH4 kg-1.CODadded/350*100

106

8.2.3 Wastewater and Inorganic Salts (including sulphate)

Throughout Period 1, leachate levels within Daisy remained stable, and there was no requirement to

drain wastewater. This behaviour continued through Periods 2 and 3, and it became clear that the

water added to the waste before feeding, water removed with the digestate as water of saturation, and

water consumed, were in near perfect balance. This raised the question of the inorganic salts

circulating within Daisy, and whether their concentration might rise to inhibitory levels. Figure 8.4 was

plotted using data from the ash measurements from the VS analyzes.

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

0 100 200 300 400 500 600

Conc

entr

atio

n of

inor

gani

c sa

lts g

/L

Elapsed Time - days

P1 P2 P3 P4 P5 P6

Figure 8.4 Concentration of inorganic salts in Daisy – following changes in CODFW addition

5.5

6.0

6.5

7.0

7.5

8.0

0.0

0.20.40.60.8

1.01.21.41.6

1.82.0

0 100 200 300 400 500 600

pH

Alka

linity

Rat

io (I

A/PA

)

Elapsed Time - days

Alkalinity Ratio (IA/PA) pH

P4 P5 P6P1 P2 P3

Figure 8.3 Alkalinity ratio and pH vs. elapsed time; stability under changing levels of FW addition

107

These data tell us four things; firstly, the concentration rises and falls with the amount of food waste

added; secondly, the lower the concentration the smaller the short-term fluctuations; thirdly, the range

within which the fluctuations occur is rather narrow - between 2.0 and 3.5 g.L-1, fourthly, in Period 6 the

trend was slightly upward.

Figure 8.7 shows the concentration of sulphate within Daisy (note the different scale on the x-axis).

Once again the pattern of short-term highs and lows associated with the addition and consumption of

fresh waste, especially food waste, can be seen. The upward trend observed in the concentration of

salts beyond Day 400 (Figure 8.4) is more clearly defined. From about Day 440 onwards, at higher

concentrations of food waste addition, the sulphate concentration starts to build up. During this period,

the weekly low concentration rises steadily from about 600 mg.L-1 to about 1300 mg.L-1, which

coincides with an increase of about 800 mg.L-1 in the concentration of inorganic salts during the same

period. It seems possible that continued operation at high food waste addition rates may result in a

continuing rise in sulphate concentration, perhaps eventually causing increased activity by sulphate

reducing bacteria (SRB) and a fall in methane production. Sulphate reducing bacteria (SRBs) would

normally be expected to consume this sulphate; the reason this is not occurring is unknown. Three

explanations are possible; an absence of SRBs, oxidation of sulphide in the samples prior to analysis, or

some form of inhibition. The first should become clearer when further amplicon sequencing data

become available. The second requires further analytical work. The third is speculative at present.

0

200

400

600

800

1000

1200

1400

1600

1800

240 290 340 390 440 490 540 590

Tota

l Sul

phat

e Co

ncen

trat

ion

mg/

L

Elapsed Time - days

CODFW 7.9%

COD

FW12

.9% CODFW

Zero%CODFW 17.2%BA5

CODFW 17.2%BA4

CODFW 17.2%BA6

CODFW 17.2%BA4

CODFW 29.3%

CODFW 21.7%

Figure 8.7 Concentration of sulphate in Daisy – steadily rising at higher CODFW addition

108

8.2.4 Chemical Oxygen Demand in Leachate

The COD content of the leachate was sampled and analyzed four times per week. Figure 8.5 shows the

weekly minimum and the weekly maximum. The former occurred immediately before installation of a

fresh leach bed, and the latter about six hours later. The spread between the two grows larger at higher

levels of food waste addition; at zero food waste (around Weeks 40 to 47), the difference between the

two is very small. Over the entire 83 weeks the weekly low concentration ranges between 1.0 g.L-1 and

1.7 g.L-1 and the high concentration between 1.2 g.L-1 and 2.5 g.L-1 with a short-lived spike to 3 g.L-1.

8.2.5 Volatile Fatty Acid Concentration

VFA measurements began at Day 240. Acetate, propionate and butyrate were measured as described in

Chapter 6. For the purpose of this assessment of Daisy’s stability the individual concentrations were

converted to COD and summed. The results are presented in Figure 8.6. Once again, the results show

the effects of food waste addition. Total VFA’s rise and fall with the weekly cycle of feeding but steadily

decline with the reduction in food waste between Day 240 and Day 290. Then, when food waste drops

to zero, the concentration of CODVFA falls to about 50 mg.L-1. When food waste addition recommences

at Day 340, VFAs rise once again and the weekly cycle resumes. The trend here is important; although

CODVFA concentration responds to food waste addition, there is no sign of any consistent buildup of

VFAs within Daisy, nor any sign of VFA inhibition. Throughout the 83 week period, Daisy remained

completely stable, as evidenced by all the data presented above. During this time, food waste addition

0

100

200

300

400

500

600

700

240 290 340 390 440 490 540 590 640

Tota

l VFA

Con

cent

ratio

n m

g/L

as C

OD

Elapsed Time - days

CODFW 7.9%

COD

FW 1

2.9%

CODFW Zero%

CODFW 17.2%BA5

CODFW 17.2%BA4

CODFW17.2%BA6

CODFW 17.2%BA4

CODFW 29.3%

CODFW 21.7%

Figure 8.6 Total VFA Concentration in Daisy expressed as mg/l COD, showing influence of CODFW addition and perhaps of the variation in properties of BA from shipment to shipment

109

varied from CODFW0% to CODFW29.3%. Mostly the changes were made incrementally but in one case, at

week 49, the increase was more dramatic, going from CODFW0% to CODFW17.2% in a single step.

These data, expressed as individual concentrations of acetate, propionate and butyrate, are discussed

further in Section 8.3.5.

To this point, the discussion has described, in broad terms, the way in which Daisy’s operation

remained stable despite numerous changes in operating conditions, particularly food waste addition.

There were, however, other circumstances, such as equipment malfunctions and maintenance events,

which might reasonably have been expected to affect Daisy’s performance and stability.

The six most significant such events are summarized in Table 8.5. Events 1 and 5 were solved without

requiring shutdowns, Events 2 and 4 required partial shutdowns to make modifications to Daisy’s

heaters, Event 6 required emptying, removal, servicing and reinstallation of the UASB, and Event 3 was

a 10-day shutdown of all pumps and heaters. In the case of Event 1, the persistent heater controller

problem, there was no observable change in Daisy’s performance. In the remaining five examples,

Daisy’s performance returned to the status quo ante within 12 hours, once normal operation had been

restored. Event 4, the installation of leach bed heaters, resulted in an apparent permanent 20%

increase in biogas production that persisted from week to week. Daisy also yielded more biogas for the

next two weeks, but this was believed, at least in part, to result from conversion of COD accumulated

during the shutdown. It proved impossible to separate the effect of increased leach bed temperature

from that of delayed COD conversion.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0 10 20 30 40 50 60 70 80 90

COD

Con

cent

ratio

n in

Lea

chat

e g/

L

Elapsed Time - weeks

COD conc. before feeding COD conc. after feeding

P1 P2 P3 P4 P5 P6

Figure 8.5 Concentration of COD in Daisy. Blue line is the weekly low and the red line the weekly high. The low remains remarkably steady; the high rises and falls with FW addition

110

Throughout 83 weeks of continuous operation Daisy remained completely stable, easily

accommodating changes in feedstock and temporary process disruptions. Undoubtedly, the high C:N

ratio contributed to this, but the buffering capacity of six leach beds, each at a different stage of the

digestion process, appears to also have contributed; this will be discussed further in Section 8.4. Tank

removal and UASB removal (Events 2 and 6 respectively) were complex undertakings requiring

Table 8.5 Major Malfunctions and Maintenance Events

Dates Description Response

Impacts on stability or performance

Comments

Event Start End Potential Actual

1 2015/06/04 Week 11

2015/08/05 Week 20

TI heater repeatedly cut out; the cause was never

determined with certainty

Raised the RTD about 5cm, out of what was believed to be a cool mixing zone

lower destruction efficiency, decline in biogas production

None observed

Believed to be caused by an undisclosed failsafe circuit in the controller

2

2015/07/22 Week 18

2015/08/06 Week 20

one day event

one day

event

Tank heaters underpowered, wrong

type of heat tape

T1 heat tape replacement T2 heat tape replacement

Ingress of air, loss of biogas, impairment of microbial community

Short term loss in biogas production; made up in subsequent days

Required removal and replacement of each tank

3 2015/09/04

Week 24 2015/09/14

Week 25 Temporary shutdown of

heaters and pumps

Daisy restarted - carefully planned; within 14h all normal; gas production picked up quickly.

VFA inhibition; plugging of valves or lines; drop in destruction efficiency

observed; biogas production caught up over next two weeks

Deliberate action, partly to test the effect of shutdown and restart (partly to have a short holiday)

4 2015/09/14

Week 25 20/15/2015

Week 26 Installation of heaters on each leach bed

Removed leach beds one at a time; rest of Daisy operating

Hydraulic upset - pressure loss from removal and replacement of leach beds

None; system ran fine

Biogas production rose 20% after heaters turned on; temp rose from 35°C to 40°C

5 2016/03/09

Week 51 2016/03/10

Week 51

Hydraulic imbalance; clarifier ran dry; starved UASB; gas pumped through UASB; some granules lost to T2

Pumps to 100% for five minutes, flushed lines; palpated balance tube between T1 - T2

Loss of UASB function; hydraulic imbalance

Temporary transfer of biogas to GM1; rapid return to normal

Probably a blockage of the tank balance tube; never fully explained; no recurrence

6 2016/05/12

Week 55 2016/05/12

Week 60 UASB performance

declined;

Shut down UASB, emptied; re and re; flow distributor plugged; removed geotextile;

Permanent loss of UASB function; malfunction of balance of system

None; UASB out of service 09.00h returned to service 14.30h; back to normal 23.00h

System unperturbed by temporary absence of UASB

111

emptying and dismantling, while the rest of Daisy remained operational. To exclude oxygen to the

greatest extent possible, the leachate and sludge were removed, stored (at 4°C), and returned under an

argon blanket. The rapidity with which Daisy returned to normal biogas production following major

maintenance events is attributable, in part, to the careful preservation of the microbial community and

the exclusion of oxygen.

8.3 Operation by Period - Synergy and the Bulking Agent Effect

The performance of Daisy, in response to changes in feedstock and bulking agent, was measured in

terms of COD added, COD destroyed, and the quantity of biogas and methane generated versus time. A

summary of the data, and the calculations used to

generate it, are presented below. The methods used

for measuring COD are described in Chapter 6 (Section

6.1). The COD measurements used to calculate the

total input into Daisy are presented in Table 8.6. The

COD content of the digestate from each leach bed

(together with TS and VS) was measured individually

in sextuplicate; all the results can be found in

Appendix C. Table 8.3 shows substrate destruction efficiencies averaged by Period, with and without

BA. Biogas production was measured as L.h-1, L.d-1 and L.wk-1, under laboratory conditions of

temperature and pressure, for the entire operating period of 88 weeks (Weeks 1 to 5 were omitted

from data analysis because of anomalies related start-up). These measurements were converted to

biogas STP, and to methane

at STP, and are presented in

Table 8.7. The percent

methane content in both gas

meters (58.5% in GM1 and

51.7% in GM2), and the

weighted average methane

content of 52.4% are also included. Overall, 90% of the methane came from the leach beds and the

tanks and was measured in GM2 but, immediately following the addition of fresh waste, gas production

in the UASB rose to as much as 25% of the total for several hours (depending on FW addition). The

utility of the UASB is discussed further in Chapter 10.

Table 8.6 Comparison of COD Analysis Methods

COD g/gTS Sample Value

Dry STDEV

Cardboard 1.14 0.05

Boxboard 1.13 0.02

Newsprint 1.33 0.02

Fine paper 1.00 0.03

Food waste 1.30 0.11

Bulking agent 1.29 0.05

Digestate By Serial #

Table 8.7 Summary of Biogas and Methane Production (83 weeks) Biogas Methane

As meas. At STP At STP

Total GM1 GM2 Total GM1 GM2 Total GM1 GM2

Litres 39500 3830 35700 36300 3520 32800 19000 2060 16950

% of Total 1000 9.7 90.3 100 9.7 90.4 100 10.8 89.2

% Methane 52.4 58.5 51.7 52.4 58.49 51.7

STDEV n=40 3.7 3.7 3.6

STDEV% 6.3 6.3 6.9

112

The analysis of the elemental composition and ash content of all the substrates (Table 8.8) was used for

two main purposes, to calculate the C:N ratio (see Table 8.9), and to calculate the stoichiometry of the

digestion process for the average substrate, based

on the entire 83 week digestion period (Figure 8.8).

From the stoichiometric equation, the average methane content of the biogas was calculated at 52.5%

and corresponded satisfactorily with the measured value (weighted average of GM1 and GM2) of

52.4±3.7%. Table 8.10 is a complete summary of the mass balance for each operating Period.

Table 8.8 Elemental Analysis - Average of three samples

Substrate

Analest1 TKN Diosady2

TS/VS

Weight percent C H O N Ash

CB 41.3 5.96 44.1 0.17 8.55

BB 39.4 5.67 42.2 0.11 12.7

NP 46.8 6.39 45.8 0.06 0.94

FP 37.2 5.57 41.0 0.03 16.3

FW 47.7 6.92 32.4 3.04 9.97

BA 46.6 6.19 45.1 0.25 1.84 1 Department of Chemistry, 2Department of Chemical Engineering

Table 8.9 Carbon Nitrogen Ratio

CODFW% Substrate Samples Avg. STDEV

Set 1 Set 2 Set 3 n = 3

0 356 356 343 352 7.6

7.9 133 133 123 130 5.4

12.9 95.3 95.3 88.2 92.9 4.1

17.2 76.4 76.4 70.6 74.4 3.3

21.7 63.3 63.3 58.5 61.7 2.8

29.3 49.0 49.0 45.3 47.8 2.1

Substrate: from elemental analysis of FW+FB+BA, average fed to Daisy (weeks 5 – 88): C90

H155

O67

N (n=90 a=155

b=67 c=1), nitrogen source: ammonia.

d = (4n + a – 2b – 3c) = 378

fe = 1 – f

s = 1 – 0.08 = 0.92 (f

s from Rittmann and McCarty for anaerobic reactions – low biomass yield)

ra: 0.125CO

2 + H

+ + e

- = 0.125CH

4 + 0.25H

2O

rc: 0.2CO

2 + 0.05NH

4

+ + 0.05HCO

3

- + H

+ + e

- = 0.05C

5H

7O

2N + 0.45H

2O

rd: (c(n-c)/d)CO

2 + (c/d)NH

4

+ + (c/d)HCO

3

- + H

+ + e

- = (1/d)C

nH

aO

bN

c + ((2n-b+c)/d)H

2O

- rd: 0.235CO

2 + 0.00264NH

4

+ + 0.00264HCO

3

- + H

+ + e

- = 0.00264C

90H

155O

67N + 0.301H

2O

fer

a: 0.115CO

2 + 0.92H

+ + 0.92e

- = 0.115CH

4 + 0.23H

2O

fsr

c: 0.016CO

2 + 0.004NH

4

+ + 0.004HCO

3

- + 0.08H

+ + 0.08e

- = 0.004C

5H

7O

2N + 0.036H

2O

0.00264C90

H155

O67

N + 0.035H2O = 0.004C

5H

7O

2N + 0.115CH

4 + 0.104CO

2 + 0.00136NH

4

+ + 0.00136HCO

3

-

1.94C90

H155

O67

N + 25.7H2O = 2.98C

5H

7O

2N + 84.6CH

4 + 76.4CO

2 + NH

4

+ + HCO

3

-

COD content: substrate = 1.25 gCOD.gVS-1

cells = 1.42 gCOD.gVS-1

Calculated methane content: (84.6/84.6+76.4)*100 = 52.5%

Average operating methane content (weeks 5 to 88) = 52.4%

Figure 8.8 Stoichiometry of digestion of the 83 week average substrate; calculation of percent methane in biogas from first principles based on elemental analysis.

113

The mass balance calculation is a measure of the percentage of the COD destroyed that can be

accounted for by the COD in the methane generated (Equation 8.1). The numerator is the volume of

methane produced at STP converted to grams of COD at 350 mL per gram, plus the COD estimated to

have been consumed to generate new cells (collectively products). The denominator is the amount of

COD destroyed (reactants);

𝑀𝑎𝑠𝑠 𝑏𝑎𝑙𝑎𝑛𝑐𝑒 =

.

. . ( )× .

( )× 100% [8.1]

The percent mass balance between one operating period and the next varies for two reasons; firstly, at

low biogas production the calculation of substrate destruction becomes the difference between two

large numbers with its inherent uncertainty; secondly, it is difficult to match substrate destruction with

biogas production particularly during periods when digester conditions are changing. To illustrate this

second point, Table 8.11 shows a segment of the chart (Weeks 43 to 56 inclusive) of all the leach beds’

serial numbers through the duration of the research programme (the entire chart is in Appendix B). The

period between the two double red lines, Weeks 44 to 49 inclusive, is the period of the lowest biogas

production, Period 4c. Leach bed serial numbers in deep yellow, S.043 to S.048, are the leach beds to

which no food waste was added. It can readily be seen that, while those leach beds are the

Table 8.10 Mass Balance by Period Operating Data COD g

Methane Mass

balance %

Period Weeks Serial Nos. (a) Subs. in

(b) DG out

(c) Destr.

(d) Conv. Cells

(e) Conv. to CH4

Total out (g) + (d)

(f) Vol. out L.

(g) COD out g.

1 6 - 15 S.008 - S.017 20970 12710 8260 660 7600 7680 2460 7020 93.0

2 16 - 24 S.018 - S.024 14680 8920 5760 461 5300 5460 1750 5000 94.8

3 25 - 31 S.025 - S.030 14680 6240 5000 340 4600 5170 1610 4600 98.9

4a 32 - 37 S.030 - S.035 12220 8480 3740 299 3440 3770 1210 3470 100.7

4b 38 - 44 S.036 - S.042 13620 10560 3060 245 2820 3030 975 2790 98.9

4c 44 - 49 S.043 - S.048 11080 9560 1520 1228 1400 1340 428 1220 88.3

5a 50 - 57 S.049 - S.056 16770 11090 5690 4558 5230 5220 1670 4760 91.8

5b 58 - 63 S.057 - S.062 12580 8400 4180 3356 3850 4510 1460 4170 107.8

5c 64 - 70 S.063 - S.068 12580 8590 3990 319 3670 4160 1350 3840 104.3

5d 71 - 74 S.069 - S.073 10480 6520 3960 317 3640 3820 1230 3500 96.5

6a 74 - 80 S.074 - S.079 13080 7640 5440 435 5010 5410 1740 4970 99.4

6b 81 - 86 S.080 - S.087 14080 7120 6960 5578 6400 6960 2240 6400 100.0

Totals 6 - 88 S.008 - S.087 167300 108900 58390 4670 53720 58830 18960 54160 100.8

mass balance = (g+d)/c*100% c = a - b d = c*0.08 e = c - d g = f/0.35l.gCOD-1

114

predominant contributors to biogas during the six-week period, five other leach beds, S.038 to S.042,

also contribute. This is discussed further in Section 8.4.

Table 8.11 Leach Bed Serial Number Sequence

Week LL01 LL02 LL03 LL04 LL05 LL06

43 S.037 S.038 S.039 S.040 S.041 S.042

44 S.043 S.038 S.039 S.040 S.041 S.042

45 S.043 S.044 S.039 S.040 S.041 S.042

46 S.043 S.044 S.045 S.040 S.041 S.042

47 S.043 S.044 S.045 S.046 S.041 S.042

48 S.043 S.044 S.045 S.046 S.047 S.042

49 S.043 S.044 S.045 S.046 S.047 S.048

50 S.049 S.044 S.045 S.046 S.047 S.048

51 S.049 S.050 S.045 S.046 S.047 S.048

52 S.049 S.050 S.051 S.046 S.047 S.048

53 S.049 S.050 S.051 S.052 S.047 S.048

54 S.049 S.050 S.051 S.052 S.053 S.048

55 S.049 S.050 S.051 S.052 S.053 S.054

56 S.055 S.050 S.051 S.052 S.053 S.054

Period 4b Period 4c Period 5a

Some of the discussions which follow rely upon the establishment of stable biogas production at

different levels of food waste addition, as a basis for the calculation of the synergistic effect of food

waste addition on the digestion of fibres. Weeks 44 to 49 represent biogas production from fibre alone,

and are thus of particular importance to these calculations.

8.3.1 Periods 2 and 3 - Seven Week SRT

Table 8.12 Composition of Digester Feedstock per Leach Bed - Periods 2, 3 and 5

Category Component Code Quantity (g)

grams

Percent of TS Percent of VS Percent of COD

Fibre Cardboard CB 440 33.0 33.5 32.5 Boxboard BB 350 28.1 27.3 27.5 Newsprint NP 110 8.3 9.2 9.6 Fine paper FP 200 15.3 14.5 13.2

Sub-total fibre FB 1,100

Food waste Green bin waste FW 675 15.3 15.4 17.2

Bulking agent Ash wood chips BA 500

Water 3,800

Total 6,075

115

The original plan was to complete two seven-week cycles in Period 2. The SRT was lengthened to seven

weeks from six by introducing a fresh leach bed each week for six weeks in a row, then skipping the

seventh. The feedstock composition during Periods 2, 3 and 5, Table 8.12, is unchanged from Period 1.

Figure 8.9 shows that biogas production in Period 2 fell to a nine-week average of 367 L.wk-1 (from 509

L.wk-1 in Period 1) a drop of 28%, but COD destruction efficiency (Table 8.3) was unchanged at 53.5%

compared to 53.7% in Period 1. Appendix D contains complete tables showing substrate destruction

efficiency, and mass balance for all 6 Periods. Essentially, the yield was the same but the digestion rate

had fallen by almost 30%; nothing could be gained from a longer SRT. The elimination of a fresh leach

bed every seven weeks made biogas production uneven, hence the dips at Week 16 and Week 23. The

experiment was terminated after 9 weeks at Week 25, and Period 3 was begun by returning Daisy to a

six week SRT; no other changes were made.

The more pronounced drop in biogas production at Week 25 was the result of a 10 - day shutdown of

pumps and heaters. This was to accommodate a short vacation but also to test robustness and stability

on restart; the restart went smoothly with a rapid return to normal operation. The purpose of Period 3

was simply to re-establish stable biogas production with a six week SRT as a starting point from which

to begin the next experiment. As Figure 8.10 shows, biogas production rose to more than 600 L.wk-1,

about 100 L.wk-1 greater than Period 1, despite the identical operating conditions.

0

100

200

300

400

500

600

15 17 19 21 23 25

Biog

as p

rodu

ctio

n L/

wk

at 2

97K

Elapsed Time - weeks

Total GM1 GM2

Period 2 - 7 week SRT

Figure 8.9 Weekly biogas production – Period 2; 7 wk. SRT; LB6 replacement postponed one week at wk. 16 and at wk. 23 to create 7wk. cycle; beginning of short shutdown at wk. 25

116

Higher concentrations of accumulated COD were also measured within Daisy during this period (Figure

8.5); presumably a rise in unconsumed hydrolysis intermediates created a temporary peak in biogas

production upon restart. The mass balance closed at 94.8% in Period 2 and 98.9% in Period 3 (Table

8.10).

8.3.2 Period 4 - Reduction in Food Waste Addition – Discovery of Synergy

The objective in Period 4 was to reduce the amount of food waste in the feed, in small decrements,

from the starting level of 17.2% CODFW to 12.9%, then 7.9% and finally 0% CODFW, measure Daisy’s

performance and stability under these changing conditions, and determine if the food waste itself had

any effect on the digestion of the fibres. Table 8.13 shows the composition of the feedstock at each

stage; the total quantity of fibre remained unchanged throughout. The drop in biogas production, in

response to the reduction in food waste, proceeded stepwise (Figure 8.11). Gas production stabilized at

each level very quickly; the weekly gas production figure shown on the graph is the average of the last

four weeks of data under each condition. Within four weeks of making the first change, to 12.9%

CODFW, it was apparent that something strange was happening. The reduction in food waste caused

biogas production to fall as expected, but the extent of the drop was far greater than could be

accounted for by the amount of CODFW withheld from the feedstock - even assuming 100% conversion

of CODFW.

0

100

200

300

400

500

600

700

25 26 27 28 29 30 31 32

Biog

as p

rodu

ctio

n L/

wk

at 2

97K

Elapsed Time - weeks

Total GM1 GM2

Period 3 - return to 6 week SRT

Figure 8.10 Weekly biogas production – Period 3; restart after 10d shutdown; return to 6 wk. SRT; weekly biogas production rises above prior level of 509 L.wk-1

117

For example, at the first plateau (CODFW12.9%), biogas production should have been no less than 462

L.wk-1 if the reduction in food waste alone was responsible, instead it was 414 L.wk-1. By the time food

waste addition had been reduced to zero, the discrepancy had grown to almost 200 L.wk-1 of biogas (at

297K); in other words, almost 40% of the biogas produced at CODFW17.2% where the experiment

began, could not be properly accounted for. Substrate destruction efficiency, on a COD basis, fell with

each reduction in food waste, to 38.9% then 31.8% and finally to 18.6% at CODFW0% (Table 8.3). A

leachate analysis confirmed that unconverted COD was not accumulating within Daisy, so what was

happening?

Table 8.13 Composition of Digester Feedstock per Leach Bed– Periods 4a, 4b and 4c

Category Component Code Quantity (g) Percent of TS Percent of VS Percent of COD Fibre Cardboard CB 440 Boxboard BB 350 Newsprint NP 110 Fine paper FP 200 Sub-total fibre FB 1,100 Bulking agent Ash wood chips BA 500 Food waste Green bin waste FW 480 11.4 11.4 12.9 Water 3,800 Total 4a 5,880

Food waste Green bin waste FW 280 7.0 7.0 7.9 Water 3,800 Total 4b 5,680 Food waste Green bin waste FW 0 0.0 0.0 0.0

Water 3,800 Total 4c 5,400

0

100

200

300

400

500

600

30 35 40 45 50

Biog

as p

rodu

ctio

n L/

wk

at 2

97K

Elapsed Time - weeks

Total GM1 GM2

Period 4 - FW reduction to zero in 3 steps

414L/wk.

287L/wk.

132L/wk.

Figure 8.11 Weekly Biogas Production – Period 4: FW to zero in 3 steps; stable biogas production established at each level. Biogas drop greater than explicable by reduction in FW.

118

The initial, tentative, conclusion was that an unexpectedly large synergistic effect, created by adding

food waste, was significantly enhancing the digestibility of the fibres and was related in some way to

the amount of food waste added. The implications for process design (SRT in particular) prompted a

mid-course correction. Four steps were decided upon.

a) The literature was revisited to see if this behaviour had been reported. I was unable to find

any references to a synergistic effect on the digestion of wood-derived lignocellulosic fibres

with the help of food waste.

b) A simple calculation to measure synergy was devised, based upon two assumptions; firstly,

that fibre alone produced 122 L.wk-1 of biogas at STP (132 L.wk-1 as-measured); secondly,

that the total amount of biogas produced without synergy could not exceed the sum of 122

L.wk-1 plus the food waste contribution calculated at 100% COD conversion. Thus the

synergy could be calculated as follows:

𝑉 . 𝑉 − 𝑉 − 𝑉 [8.2]

I subsequently found a reference to a synergistic effect of food waste addition on the

digestion of waste activated sludge, in which this same calculation was used to measure

synergy (Yun et al., 2015).

c) The plan for BMP 2 was modified and expanded, specifically to study synergy; the results

are presented in Chapter 9. BMP 3, digesting larger quantities of substrate under conditions

similar to Daisy’s is also presented in Chapter 9.

d) The experimental programme was revised to concentrate on quantifying the apparent

synergistic effect of food waste within Daisy. In Period 5 Daisy was returned to CODFW17.2%

in a single step, firstly to test stability under more extreme changes, and secondly to re-

establish a stable baseline of biogas production from which to explore synergy at higher

levels of food waste addition.

The mass balance closed at 100.7% in Period 4a, 98.9% in Period 4b and 88.3% in Period 4c.

8.3.3 Period 5 – Raising Food Waste Addition; the Bulking Agent Effect

The plan for Period 5 was to raise CODFW to 17.2% in one step and return Daisy to stable biogas

production, presumably at just over 500 L.wk-1 based on Period 1, and to test Daisy’s stability under this

more abrupt change in food waste addition.

119

The assumption was that this would take 8 to 10 weeks. From Week 59 to Week 56, everything went as

planned, but at Week 56 gas production took a sharp upturn and rose to just under 600 L.wk-1. By this

point, supplies of the fourth bulking agent shipment, BA4, were running low so at Week 58 Daisy was

switched to BA5 (see Figure 8.12).

Bulking agent is used for the specific purpose of maintaining the permeability and structure of the

waste mass in the leach beds for the six weeks of SRT. Of the materials fed to Daisy, BA is the only one

that varied substantially in its properties during the course of the research programme. It always took

the form of ground up ash wood and was supplied by Miller Waste Systems as part of their in-kind

contribution to the project. In all, six batches were delivered over the entire duration of the

experiment; Batches BA1, BA2, BA4 and BA6 were ground in a Roto-chopper and screened, Batch 3 was

ground in a Peterson Grinder, and BA5 was ground by an independent tree company who used a

chipper. Batches BA1, BA2, and BA4 worked well in the sense that switching from one to another

caused no noticeable changes in Daisy’s performance; Batch 3 was rejected because the particle size

was far too coarse.

BA5 had a different physical appearance; it was coarser, shorter and fatter. It was first used at the

beginning of Week 58, one week after gas production had returned to prior levels, and used for six

0

100

200

300

400

500

600

700

49 54 59 64 69 74

Biog

as p

rodu

ctio

n L/

wk

at 2

97K

Elapsed Time - weeks

Total GM1 GM2

BA4 BA5 BA4 BA6

513L/wk.

Figure 8.12 Weekly biogas production – Period 5; return to CODFW17.2%; properties of BA5 apparently lowering biogas yield and substrate destruction efficiency

120

weeks. Biogas production declined throughout the next 6 weeks, slowly at first, but then more rapidly.

Although there was no specific reason to believe that the bulking agent was the cause, nothing else had

changed and there was no other plausible explanation. A supply of BA4 had been retained in case BA5

did not perform well, so at the beginning of Week 64 Daisy was switched back to BA4. The decline in

biogas production first levelled out and then, at Week 67, started to climb. It finally reached stability at

513 L.wk-1, between Weeks 68 and 74. In the meantime, a shipment of fresh bulking agent, BA6,

prepared to the same specifications as BA4, had been requested from Miller; it arrived before the

remaining BA4 was all consumed. What I thought would take 8 to 10 weeks had taken 25. The

differences between BA5 and BA4 are readily apparent in Plate 14, as are the similarities between BA4

and BA6.

To try and shed light on this behaviour, some simple experiments were performed to compare the

physical properties of the three bulking agent samples. Some extended leach bed permeability tests

were also performed. The particle size distribution, water retention capacity and bulk density of BA4

and BA5 were compared. The methods can be found in Appendix C, and the results are presented in

Table 8.14. The particle size distributions (screened at +3.4 mm, +0.5 mm and ±0.2 mm) were similar,

a

1cm 1cm

b

1cm

c

Plate 14. Bulking agent a) BA4; b) BA5; c) BA6; BA5 coarser and of higher bulk density

121

though BA5 had a higher proportion of coarse material, however their bulk densities were very

different, 0.12 kg.l-1 for BA4 vs. 0.22 kg.l-1 for BA5, and BA4 had a greater water retention capacity. The

morphology of the two batches was also quite different. BA4 consisted of longer, finer, pieces of wood

which may serve to enhance lateral movement of leachate in the leach beds.

The substrate destruction efficiency

reflected the trends in biogas production.

During Period 5a destruction efficiency rose

sharply to 51.7%, from 18.6%, then fell to

43.8% and 43.9% in Periods 5b and 5c

respectively. Finally, in period 5d, it rose to

53.0%, reestablishing the performance of

Periods 1, 2 and 3. An extended series of

leach bed permeability tests was conducted,

using a procedure similar to that described in Chapter 7 (Section 7.1). Leachate (565mL) was delivered

automatically by P3 at 510 mL/min. and the drainage rate of leachate measured until 520 mL had been

recovered. These tests were planned and conducted in the hope that they would make it possible to

distinguish between the behaviour of leach beds with BA4 from those with BA5 (and ultimately BA6) in

a way that might explain the drop in biogas production and digestion efficiency during this period.

Unfortunately, no clear pattern emerged, and no correlation between leach bed drainage properties

and substrate destruction or biogas production was found. A subset of the data is provided in Appendix

D, and the entire data set in EDF6. This finding was contrary to the earlier permeability tests, which had

confirmed that BA2, which was very coarse, did indeed show different drainage behaviour; perhaps, in

the case of BA5 vs. BA4, the difference was too subtle. The mass balance closed at 91.8% in Period 5a,

107.8% in Period 5b, 104.3% in Period 5c and 96.5% in Period 5d. Despite the unexpected behaviour

biogas production and substrate destruction ultimately returned to previous levels, and no process

instability was observed during this period. It still seems most probable that the changes in BA both

caused, and rectified, this anomalous behaviour. The answer may lie in the moisture retention

properties of the BA and this is discussed further in Chapter 10.

8.3.4 Period 6 - High Levels of Food Waste Addition

The final step was to increase food waste addition further, to test the effect on synergy and on Daisy’s

stability. In Period 6a, CODFW was raised to 21.7% and in Period 6b to 29.3%. Biogas production levelled

off at 620 L.wk-1 at CODFW21.7% and 797 L.wk-1 at CODFW29.3% (at 297K) as shown in Figure 8.13. Of

this amount, 244 L.wk-1 and 300 L.wk-1 respectively (at 297K) were attributable to synergistic digestion

Table 8.14 Physical Properties of Bulking Agents BA4 and BA5

Particle Size Distribution %

>3.36mm <3.36mm>0.50mm <0.50mm>0.21mm <0.21mm

BA4 44.7 37.9 12.5 3.4

BA5 52.8 31.2 11.7 2.9

Water Retention gH2O/gBA

BA4 2.95

BA5 2.33

Bulk Density kg/L

BA4 0.119

BA5 0.220

122

of fibres; At the end of Week 88 Daisy was shut down and all six leach beds removed at once for

analysis so that digestion efficiency versus digestion time could be measured. This was the only

occasion on which the contents of all six leach beds could be analyzed at one time; the results, in Figure

8.14, show that digestion is about 85% complete after only two weeks and 93% after three weeks

(assuming destruction efficiency is

essentially at the asymptote after six

weeks). These results were obtained of

course at the highest level of food waste

addition tested, and this must influence the

shape of this curve, and yet food waste still

only represented 29% of total COD. Figure

8.14 illustrates very well the futility of

lengthening the SRT as described in 8.3.1.

Substrate COD destruction efficiency rose

again to 56.0% in Period 6a and then to

65.4% in Period 6b (from 18.6% at CODFW0%). The mass balance closed at 99.4%, and at 100.0% in

Periods 6a and 6b respectively. The mass balance deviates further from 100% during periods of low

food waste addition and periods of changing operating conditions.

0

100

200

300

400

500

600

700

800

900

1000

74 76 78 80 82 84 86 88 90

Biog

as p

rodu

ctio

n L/

wk

at 2

97K

Elapsed Time - weeks

Total GM1 GM2

Period 6 FW increase in 2 steps

620L/wk.

797L/wk.

Figure 8.13 Weekly biogas production – Period 6; CODFW to 21.7% then 29.3% in 2 steps; stable biogas production at each step

01020304050607080

0 2 4 6

Perc

ent C

OD

des

truc

tion

(cum

ulat

ive)

Digestion Time (weeks)

COD Destruction Efficiency at CODFW 29%

Figure 8.14 Destruction efficiency vs. digestion time – S.082 to S.088; substrate destruction high in Wks 1 and 2, essentially complete by Wk. 4

123

8.3.5 Synergy

Using the simple calculation described earlier, synergy was quantified, converted to STP, and the results

summarized in Table 8.15. The input to Daisy is expressed in grams of COD added per week as food

waste, and as fibre. The latter remains constant throughout. Food waste COD is then expressed as a

percentage of the total. The biogas produced is presented in litres per week at STP as four numbers, the

total, the portion attributable to food waste at 100% COD conversion, the portion attributable to

fibre alone (the difference between total and food waste), and the portion attributable to synergy (the

difference between fibre total and fibre at zero food waste addition). Specific biogas is expressed as

L.kg-1 CODadded. The results are

displayed graphically in Figures

8.15, 8.16 and 8.17.The growth in

synergistic biogas with increasing

food waste addition is clearly

evident. At CODFW29.3%, the biogas

generated from fibre was 325% of

that generated from the same

amount of fibre at CODFW0%. The

foundation for these calculations is

the measurement of 122 L.wk-1 of

biogas (at STP) from fibre alone;

how reliable is this number? Is it

trustworthy? While it would have been preferable to operate at CODFW0% for a more prolonged period,

it is unlikely that gas production would have risen, but it may have gone lower with time. The value for

Table 8.15 Calculation of Synergy at 100% FW conversion

CODIN (per week) Biogas Production (L.wk-1 at STP) Specific Production L.kg-1 CODadded

FW ST

DEV FB Total FW

ST

DEV

Total meas

ST DEV

n FW FB FB

w/o syn.

Syn. FW FB Total Syn. STDEV

g g g % n = 27

0 0.0 1,200 1,200 0.0 0.0 122 7.7 4 0 122 122 0 0 101 101 0 6.4

104 8.7 1,200 1,310 7.9 0.1 264 6.4 4 69 195 122 73 53 162 202 61 4.9

178 15.0 1,200 1,380 12.9 0.1 379 11.4 4 119 261 122 139 86 217 275 115 8.2

250 21.0 1,200 1,450 17.2 0.2 468 23.1 16 167 302 122 180 115 250 322 149 15.9

333 28.1 1,200 1,540 21.7 0.4 569 21.1 6 223 346 122 224 145 288 370 186 13.7

500 42.1 1,200 1,700 29.3 0.7 731 13.7 6 334 397 122 275 196 330 429 229 8.1

0

100

200

300

400

500

600

700

800

0.0 7.9 12.9 17.2 21.7 29.3

Biog

as -

Litr

es/w

k (a

t STP

)

Percent CODFW

Synergisticbiogasfrom fibre

Foodwaste at100% CODconversion

Fibre alone

n = 4

n = 4

n = 4

n = 16

n = 6

n = 6FW at 100% conversion

Figure 8.15 Synergistic biogas production from FB vs. FW addition; 𝑉 . 𝑉 − 𝑉 − 𝑉

124

the plateau in biogas production is the

average of four measurements over four

weeks with a standard deviation of only

6.2% (Table 8.4); this supports the

conclusion that the figure for biogas

production at CODFW0% is dependable.

Figure 8.16 shows a strong linear

relationship between synergistic biogas

production and CODFW addition. Figure

8.17 presents the same results differently,

on the basis of specific biogas production

per kgCODFWadded. Despite their linearity, this relationship must eventually change and the synergistic

effect must decline as the digestible portion of the fibres becomes consumed. The results make a

strong and persuasive case for a powerful synergistic effect in the anaerobic digestion of lignocellulosic

fibres, resulting from the addition of food waste to wood-based lignocellulosic fibres. Furthermore,

they show that the relationship between

the amount of FW added and synergistic

biogas produced is almost linear. But they

also raise a number of questions; what is

the mechanism at work; is the combination

of substrates the sole cause, or does the

reactor design contribute; are all the fibres

affected in the same way, or is there a

difference among them; is it a rate effect as

well as a yield effect and at what level of

FW addition does synergy reach a plateau?

Daisy’s coupon data provide some of the

answers, and the BMP experiments and the enzyme assays described in Chapter 9 provide additional

insight. The coupon results represent substrate digestibility, at 42 days, at different levels of FW

addition. They are subject to inaccuracies; the samples are small, ranging from 1.3 to 6.5 g (depending

on the substrate density) and exposure to the same digestion conditions in Daisy cannot be

guaranteed. Nevertheless, the large number of samples, a total of 277 fibre coupons and 27 bulking

agent coupons, have made it possible to calculate decent average values. The coupon data led to three

y = 8.04x + 4.033R² = 0.99

0

50

100

150

200

250

300

0 10 20 30 40

Biog

as -

Litr

es/k

gCO

Dad

ded

(at S

TP)

Percent CODtotal as FW

FW at 100% conversion.

For all x axis values n = 27

n = 4

n = 4

n = 16

n = 6

n = 6

Figure 8.17 Specific Synergistic biogas production in L.kg-1VSadded calculated at 100% FW conv.

y = 9.68x + 4.85R² = 0.99

0

50

100

150

200

250

300

350

0.0 10.0 20.0 30.0 40.0Syne

rgist

ic B

ioga

s -Li

tres

/wk

(at S

TP)

Percent CODFW

FW at 100% conversion

n = 6

n = 6

n = 4

n = 16

n = 4

Figure 8.16 Synergistic biogas production vs FW addition; L.wk-1

calculated at 100% FW conversion.

125

principal findings. The digestibility of the individual fibres is enhanced by the addition of food waste,

the extent of the enhancement is

related to the amount of FW

added, and the individual fibres

behave quite differently from one

another (Figure 8.18). The number

of replicates for each fibre at each

%CODFW are in a table in Appendix

D; note that COD measurement of

the coupon samples was not

attempted because of their

physical form after digestion and

small sample size. Their digestibility

can be ranked FP> BB> CB> NP>

BA, confirming the results of Pommier (2010), Eleazer (1997) and Buffiere (2008a), all of whom ranked

digestibility FP> CB> NP. Of greater interest is the relationship between digestibility and food waste

addition, which confirms the observations from Daisy’s results and broadens the understanding of what

is occurring. The most impressive examples are BB and CB whose digestibility increased 3-fold between

CODFW0% and CODFW29%, and FP an amazing 6.8-fold. It is noteworthy that the fibres that were the

product of chemical pulping, FP, CB and BB, are much more digestible than NP, which is the product of

mechanical pulping, and the BA which is not pulped at all; the BA coupons were larger (and thus

singular, not triplicates) because of the physical properties of the material (Chapter 6, Section 6.1). It is

perhaps possible to view the pulping process as a form of serendipitous pretreatment which aids

digestion, especially in the presence of food waste. The decline in the digestibility of NP at higher levels

of food waste addition looks to be genuine, the error bars are quite small: though it is hard to visualize

a mechanism.

While the existence of synergy had been demonstrated and quantified, the mechanism remained

obscure and needed to be investigated; the first basic question being, is synergy primarily the result of

an enhancement of methanogenesis or substrate hydrolysis? Two pieces of information suggest that it

is the latter. Firstly, by definition, the synergistic biogas comes from the lignocellulosic fibres; these are

known to be recalcitrant to anaerobic digestion as seen in Fig. 8.18. Secondly, if a depleted population

of methanogens were responsible for a drop in biogas production, a build-up of unconverted

intermediates (e.g. VFAs) would be seen. Figure 8.19 shows the concentrations of individual VFA’s,

Figure 8.18 Effect of FW addition on fibre digestibility – coupon data; substrate destruction at 42d SRT

0

10

20

30

40

50

60

70

80

90

0 10 20 30Pe

rcen

t VS

dest

ruct

ion

Percent COD as Food Waste

Fine paper

Cardboard

Boxboard

Newsprint

Bulking agent

126

expressed as gCOD.L-1 from Day 240 to Day 616. Both short-term variations and long-term trends are

evident; no long-term build-up of VFAs was ever observed and VFA production virtually ceased at

CODFW0%. Each VFA peak represents the addition of fresh waste. This close look at what is going on

inside Daisy suggests that fresh waste, especially food waste, gives rise to a sharp peak in VFA

concentration which is rapidly consumed (within 48 hours), the concentration remaining low until the

next feeding event.

From this point three possibilities were considered:

Changes in the diversity and abundance of the microbial community.

Changes in C:N ratio.

The presence of enzymes or other EPS in the FW which directly assist the hydrolysis of

the FB, and which may be the by-products of FW digestion.

The complementary research being carried out by Peter HyunWoo Lee should be able to address the

questions of microbial abundance and diversity and their potential role in synergy. Although changes in

the C:N ratio inevitably accompany changes in food waste addition (because of the great disparity

between the nitrogen content of food waste and fibres), the effect of such changes is known to be slow

in manifesting itself which would make it an inadequate explanation of the rapidity of response when

food waste is withheld (Kayhanian, 1999). However, there are clues from the results that may help the

search for mechanisms and causes. Firstly, when the FW content in Daisy’s feed was reduced, the effect

on biogas production was almost instantaneous and within 4 to 6 weeks biogas production had

0

50

100

150

200

250

300

350

400

240 290 340 390 440 490 540 590 640

VFA

Conc

entr

atio

n as

CO

D m

g/L

Elapsed Time - days

Acetate Propionate Butyrate

COD

FW 1

2.9% CODFW

7.9%CODFW Zero%

CODFW 17.2%BA#5

CODFW 17.2%BA#4

CODFW17.2%BA#6

CODFW 17.2%BA#4

CODFW 29.3%BA6

CODFW 21.7% BA#6

Figure 8.19 Concentration of individual VFAs vs. time. Effect of 6 levels of CODFW addition; Effect of changes in BA properties. Propionate usually dominant contributor except at Day 44 after FW added back, and at Day 568 at CODFW29%; butyrate concentration also starts rising at Day 519.

127

stabilized at a lower level (Figure 8.11). The next reduction, and the next, produced exactly the same

effect. When the CODFW was all added back in Week 50, gas production jumped in the first week, but

grew more modestly over the next five weeks. At Weeks 56 and 57, biogas production underwent two

big jumps, perhaps suggesting that some critical condition had been reached that promoted synergy; in

all it took 8 weeks to restore gas production to its former level (Figure 8.12). These observations would

suggest that, though the effect diminishes quickly as the CODFW addition rate is reduced, when CODFW is

added back it takes longer for the synergistic effect to recover and for gas production to fully return.

Yun et al. (2015) showed that the addition of small amounts of FW enhanced the digestion of waste

activated sludge (WAS), and that extracellular hydrolytic protease was primarily responsible. Yuan et al.

(2012) showed that pre-treatment of waste paper and cardboard with a microbial consortium

enhanced methane production, and used the same method to quantify synergy as the one used in this

thesis. The consortium was specifically developed to degrade cellulose under thermophilic conditions

but has only been partly characterized.

To examine synergy further, two BMPs, and an analysis of enzymatic activity in Daisy’s leachate, were

carried out; these are described in Chapter 9.

8.4 Influence of Reactor Design

Daisy has been shown to operate for extended periods of time, under a variety of conditions, without

becoming unstable or showing any signs of process upset. The ability of the digester to degrade solid

organic waste to high levels of destruction efficiency has been demonstrated, and much learned about

the synergistic effects of food waste on the digestion of lignocellulosic fibres.

But what can be said about the relationship between the reported performance and the design

characteristics of Daisy? Are there features of the design which contribute to the performance

described, particularly digestion efficiency and stability? For example, is it possible, from the data

already to hand, to draw any conclusions about the way in which sequentially-fed leach beds contribute

to stability and performance? A closer look at short-term behaviour, in relation to biogas production

and reactor conditions, provides some insight. Figure 8.20 shows that at CODFW0%, biogas production

(Fig. 8.20a) is very low, though the slight rise following addition of fresh waste is discernible. The

corresponding temperature graph (Fig. 8.20b) shows the temperature of one leach bed diving below

the X axis and then returning as it is removed and replaced, but the temperatures of the other five

remain virtually constant throughout the 24 hours.

128

Conversely, at CODFW29%, temperatures in all the leach beds (Fig. 8.20d) rose by about 3°C over a

period of five hours, and biogas production soared (Fig. 8.20c), and then both temperature and biogas

production began a slow decline. This contrasting behaviour was observed week after week during

these two separate periods of operation. What is notable is that in graph Fig. 8.20d, all five leach beds

containing partially digested waste behaved in exactly the same manner, irrespective of the age of the

waste they contained. This would seem to imply that all five were acting as methanogenic reactors,

consuming the readily available soluble COD in the fresh food waste in an exothermic reaction, raising

the temperature and producing the peak in biogas production. As was described Section 7.4, the

measurement of gas production from a single leach bed is difficult to do. The recirculation of the

leachate between the tanks and the leach beds, and the need to maintain the interconnection between

their respective head spaces, restricts the ability to isolate a single leach bed.

3536373839404142434445

0 2 5 7 9 11 14 16 18 20 23

Tem

pera

ture

Deg

. C

Elapsed Time Hrs

Leach Bed Temperatures2016-10-04

LB1 (degC)

LB2 (degC)

LB3 (degC)

LB4 (degC)

LB5 (degC)

LB6 (degC)

d

0

2

4

6

8

10

12

14

1 3 5 7 9 11 13 15 17 19 21 23

Biog

as l

itres

/hr

Elapsed Time Hrs

Biogas Production2016-02-02

GM1

GM2

Total

a

0

2

4

6

8

10

12

14

1 3 5 7 9 11 13 15 17 19 21 23

Biog

as l

itres

/hr

Elapsed Time Hrs

Biogas Production2016-10-04

GM1

GM2

Total

c

3536373839404142434445

0 2 4 6 8 10 12 14 16 18 20 22

Tem

pera

ture

Deg

. C

Elapsed Time Hrs

Leach Bed Temperatures2016-02-02

LB1 (degC)

LB2 (degC)

LB3 (degC)

LB4 (degC)

LB5 (degC)

LB6 (degC)

b

Figure 8.20 Effect of fresh waste on LB temperatures and biogas production a) Biogas production at 0%FW b) Initial LB temp. rise at 0%FW c) Biogas production at 29%FW b) Initial LB temp. rise at 29%FW.

129

A method was devised that made it possible to make this measurement while still maintaining leachate

recirculation. Unfortunately, at lower rates of gas production, the measurement was very inaccurate. A

second method, in which the leach bed in question was completely isolated from the rest of Daisy

(other than its connection to GM1) for a period of four or five hours, was developed. Gas production

from a single leach bed was measured seven times over a three-week period. The results are shown in

Figure 8.21. The first measurement was made

when the waste was three days old. In the next two

measurements gas production continued to rise

and then began a decline, ending at 180 mL per

hour at Day 42. Two things are notable; firstly, the

biogas production peak, at Day 10, did not

correspond to the immediate peak shown in Figure

8.16c, suggesting that peak biogas production in

the system (first day following leach bed

replacement) does not coincide with that of an

individual fresh leach bed (Day 10 following

installation). Secondly, even a six-week old leach bed is still producing 4 or 5 L per day of biogas which

suggests that, while the substrate it contains may be spent (Figure 8.14), it may still be serving as a

reservoir for methanogens and a continuing source of biogas.

Combining all of this information, I concluded that there is a persuasive case to be made that leach

beds containing older waste serve as methanogenic reactors to consume the high concentrations of

COD, particularly VFAs, which occur immediately after fresh food waste is added and thus the design

contributes to process stability. The other related aspect of a design consisting of sequentially-fed leach

beds, is that the system is very hard to shock because not more than 16% of the contents are changed

at any one time. The other 84% therefore, by definition, serves as a buffer. This also raises the question

of the utility of the UASB and whether or not it can be eliminated. It does most of its work immediately

after Daisy is fed, during which time the UASB produces up to 25% of the biogas generated, as

measured by GM1, compared to its long-term average of 10%. Only a comparative experiment, with

and without the UASB, can answer this question definitively. Finally, Figure 8.22 consists of six graphs,

each representing a six-week period of biogas production, one for each level of food waste addition

from CODFW0% to CODFW29.3%. These graphs provide a summary of biogas production at all levels of

FW addition tested. The stability and consistency of Daisy’s operation, and response to changing food

waste addition rates, are apparent.

0

100

200

300

400

500

600

700

800

900

0 10 20 30 40 50Biog

as g

ener

atio

n ra

te m

l/h (2

97K)

Digestion Time - Days

Figure 8.21 Biogas production from a single leach bed vs. digestion time; peak at 10d does not correspond with fresh waste addition

130

02468

101214

Biog

as l

itres

/hr

02468

101214

Biog

as l

itres

/hr GM1 GM2 Total

02468

101214

Biog

as l

itres

/hr

02468

101214

Biog

as l

itres

/hr

02468

101214

Biog

as l

itres

/hr

02468

101214

Biog

as l

itres

/hr

Elapsed Time - weeks

Period 6b 29.3%CODFW%

Period 6a 21.7%CODFW%

Period 1 17.2%CODFW%

Period 4a 12.9%CODFW%

Period 4b 7.9%CODFW%

Figure 8.22 Six-week biogas cycle for FW addition at six levels from CODFW0% to CODFW29.3% showing consistency of operation.

Period 4c 0%CODFW%

131

Chapter 9. Supporting Experiments

Three supporting experiments were conducted; biochemical methane potential (BMP) tests (three

separate experiments); enzyme activity assays of leachate, and an aerobic curing test performed on

Daisy’s digestate in a single large experiment. Each is described in its own subsection below. The

enzyme activity assays were conducted by Greg Brown in Prof. Sasha Yakunin’s lab; the leach bed

permeability tests were conducted by Scott Mitchell, a summer student in 2016, the second of the two

BMPs was the joint work of myself and Prof. Savia Gavazza (but mostly Savia), and the aerobic curing

experiment was conducted by Kyle Schumacher of Miller Waste Systems Inc.

9.1 BMP experiments

The first of the three biochemical methane potential tests (BMPs), testing the digestibility of synthetic

feed and comparing two different inocula, was described in Chapter 7, Section 7.2. The second,

comprising 20 experiments performed in triplicate, measured the digestibility of the individual

substrates fed to Daisy, and in particular the synergistic effect described in Chapter 8. The third also

measured the synergistic effect of food waste addition on the digestion of fibres, but was conducted on

a larger scale (1 L bottles) with coarsely shredded substrates in the same physical form as they were fed

to Daisy.

9.1.1 Biochemical Methane Potential Test 2 – Substrate Digestibility and Synergy

BMP Set-up BMP 2 was designed with three objectives:

Evaluate the BMP of each substrate separately.

Evaluate the BMP of a combination of food waste (FW) and each of the other substrates.

Evaluate the effect of the CODFW addition rate on the BMP of fibres (FB).

The bulking agent (BA) and all the substrates except FW were prepared as solids:

The BA was dried at 105°C and ground directly in a Wiley mill (-20 mesh, 0.84mm).

Six grams of each fibre, and of a mix of fibres replicating Daisy’s feed (40%CB, 32%BB, 10%NP

and 18%FP on a dry weight basis) were blended with 1 litre of water, dried at 105°C, and

ground in a Wiley mill (-20 mesh, 0.84 mm).

The FW was prepared as both a solid (dried) and a liquid, to reduce the risk that the enzymes or other

extra-cellular polymeric substances (EPS), that may play a role in synergy, would be lost during

substrate drying. It was further determined that 80% of CODFW should be added as solid and 20% as

liquid.

132

FWsolid (representing 80% of CODFWadded)

o 225 g of FW was blended 1:1 with water, 2x 50 g of the resulting paste was dried at

37°C for 2 days, then ground in the Wiley mill (-20 mesh, 0.84 mm).

FWliquid (representing 20% of CODFW added)

o 25 g of FW was blended 8:1 with water, centrifuged at 3,000 g for 10 min., the

supernatant recovered and stored at 4°C.

Table 9.1 Total Quantities of Substrates, Inoculum and Medium for BMPs

BMP No. FB or BA (mg)

FWsupernt (µL)

FW Dry (mg)

Inoculum (mL)

Med. (mL)

Water (mL)

1 a,b,c CB 80.6 1.34 60 18.7

2 a,b,c BB 82.1 1.34 60 18.7

3 a,b,c NP 72.3 1.34 60 18.7

4 a,b,c FP 90.6 1.34 60 18.7

5 a,b,c BA 85.3 1.34 60 18.7

6 a,b,c FW 0.0 1571 70.4 1.34 60 17.1

7 a,b,c FBFW0 81.8 1.34 60 18.7

8 a,b,c FBFW6 77.7 78 3.5 1.34 60 18.6

9 a,b,c FBFW8 75.1 129 5.8 1.34 60 18.5

10 a,b,c FBFW13 72.7 175 7.8 1.34 60 18.5

11 a,b,c FBFW17 70.1 225 10.1 1.34 60 18.4

12 a,b,c FBFW20 67.9 267 11.9 1.34 60 18.4

13 a,b,c FBFW27 62.6 369 16.5 1.34 60 18.3

14 a,b,c CBFW 71.7 173 7.7 1.34 60 18.5

15 a,b,c BBFW 72.9 176 7.9 1.34 60 18.5

16 a,b,c NPFW 65.1 157 7.0 1.34 60 18.5

17 a,b,c FPFW 79.5 192 8.6 1.34 60 18.5

18 a,b,c BAFW 75.4 97 8.1 1.34 60 18.6

19 a,b,c NCon 1.34 60 18.7

20 a,b,c FWFBnoIN 72.7 175 7.8 60 19.8

CODsubstrate:VSinoculum ratio set at 3:1

CODsubstrate = 92 to 110 mg per bottle 37 mg VS per bottle

VSinoculum = 27.7g.l-1 (at 1:1 dilution) which gives 1.34 ml IN per bottle

CODinoculum = 1.07g/gTS CODinoculum = 42mg COD per bottle

The set-up of the BMP is summarized in Table 9.1. BMPs 1 to 6 contained individual substrates: CB, BB,

NP, FP, BA FW, FB; BMPs 8 to 13 contained the FB mix, plus FW representing 6%, 8%, 13%, 17%, 20%,

and 27% of substrate COD; BMPs 14 to 18 contained individual fibres, plus FW added at 13% of

substrate COD. The TS, VS and COD content of the inoculum, each substrate and the BA were measured

using standard methods.

133

The inoculum used was granular sludge from the Tembec pulp and paper mill digester, from the

same shipment as that used to charge Daisy at start-up.

The bottles were prepared as follows:

Each bottle contained 110 mg COD as substrate based on substrate COD measurements made

using the “wet” method. When it was subsequently determined that the dry method gave more

accurate results, the input and output data were adjusted accordingly. In practical terms, this

means that the total COD per bottle varied between 93 mg.L-1 and 110 mg.L-1. All the dry

substrates were weighed into small plastic vials (one for the fibres or fibre mix, and one for the

solid portion of the FW). These were then inserted into the bottles. This step of the process was

performed aerobically.

The bottles, FWliquid , inoculum, medium and anaerobic water, together with all the equipment,

were placed in the glove bag, which was purged twice with nitrogen, and left overnight to

equilibrate. The following morning the glove bag was purged twice with purge mix (20% CO2

and 80% N2).

The FWliquid , inoculum, medium and anaerobic water were added to each bottle which was then

stoppered and capped, and placed in an incubator at 37°C.

The bottles were sampled and biogas measured for the next 74 days; samples were taken at intervals of

3 days for the first two weeks, 7 days for the next 6 weeks, and 14 days for the last 4 weeks. During the

middle of the experiment, on three consecutive sampling days one week apart, one of each triplicate

set was sampled and analyzed for methane concentration. A spreadsheet was used to calculate dilution

in the headspace and withdrawal during sampling, to determine actual methane generation rates. The

pH of several bottles was checked periodically. Further details of the setup of this BMP can be found in

Appendix E.

Results The results of the BMP tests are presented in four groups; controls, individual fibres, individual fibres

plus CODFW at 13% of CODTotal, and finally a mix of fibres replicating Daisy’s feed (40%CB, 32%BB,

10%NP and 18%FP on a dry weight basis) plus CODFW varying from 6% to 27% of CODTotal. The detailed

results, and all the associated calculations, can be found in EDF3. For the first three sets of data, error

bars representing ±1 standard deviation are included. The graphs of mixed FB plus varying CODFW are so

crowded together that error bars would be indecipherable and were thus omitted. All the graphs

contain the theoretical maximum biogas yield line, based on total COD added, converted to biogas at

310K assuming 56% CH4, the average measured content of the BMP bottles.

134

Controls The results are presented in Figure 9.1. Two controls were run; the negative control contained

inoculum only and no substrate; the substrate control contained FB, plus FW at 13% of CODtotal, and no

inoculum. The negative control, which essentially

stopped producing gas at 25d, was subtracted

from all the data in the subsequent figures.

Although the substrate control showed a slow but

steady production of gas, repeated analysis

showed that it contained virtually no methane.

There is no obvious explanation for this result.

Through an oversight, a positive control was not

performed.

Individual Substrates The results are presented in Figure 9.2. Seven

individual substrates were tested: CB, BB, NP, FP, BA, FW and FB (a mix of 40%CB, 32%BB, 10%NP and

18%FP on a dry weight basis). FW digested very rapidly with no lag time; FP followed almost the same

pattern, but with a short lag time of 3d. Both reached about 78% COD destruction efficiency after 60d.

CB and BB were almost

identical, reaching about

46% COD destruction

efficiency at 60d with a lag

time of about 4d. NP

digested very poorly, with a

lag time of 4d and ultimate

destruction of only 11%

COD. The BA behaved

differently from the other

substrates; it began with a

7d lag followed by a steady,

slow, climb to 28% COD

destruction after 74d.

0

10

20

30

40

50

60

70

80

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Max Substr. Con. Neg. Con.

Figure 9.1 BMP2 – Controls; substrate alone and inoculum alone

0

10

20

30

40

50

60

70

80

90

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Max.

FW

FP

FB

CB

BB

BA

NP

Neg.Con.Substr. Con.

Figure 9.2 BMP2 - CB, BB, NP and FP without FW, FW alone, BA alone; neg. control subtracted

135

Individual Substrates plus FW The results are presented in Figure 9.3.

The same individual substrates were co-

digested with CODFW added in an

amount equal to 13% of CODTotal. In all

cases the biogas yield increased; as will

be shown in subsequent graphs, some

of this increase was attributable to the

CODFW directly, but the balance was the

result of synergy. The comparison of

results for each of these fibres, digested

with and without FW added, are

presented side-by-side in Figure 9.4. Figure 9.4a shows food waste alone and illustrates the

methodology for all the graphs. Each displays a

lag time in days; the graph itself is plotted as

three straight lines (in the case of mixed fibres,

graphs j) and k) and BA, graphs l) and m), just

two straight lines). The implication of this

presentation is that digestion is proceeding

stepwise in a series of pseudo-zero order

reactions (independent of substrate

concentration). This is consistent with

enzymatic processes such as hydrolysis.

0

10

20

30

40

50

60

70

80

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Max.

FPFW

CBFW

FBFW13

BBFW

BAFW

NPFW

Figure 9.3 BMP2 Individual FB plus FW at 13% of CODTotal neg. con. subtracted

y = 2.7284x - 10.94R² = 0.98

y = 0.35x + 21.46R² = 0.98

y = 0.0687x + 31.05R² = 0.91

0

10

20

30

40

50

60

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Cardboard

Lag = 4d

b

y = 2.8737x - 0.73R² = 0.97

y = 0.0982x + 40.75R² = 0.93y = 0.3724x + 34.20

R² = 1.00

0

10

20

30

40

50

60

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Food Waste + Cardboard

Lag = 0

c

y = 5.82x + 1.07R² = 0.99

y = 0.34x + 45.97R² = 0.98

y = 0.047x + 55.31R² = 0.89

0

10

20

30

40

50

60

70

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Food Waste

Lag = 0

a

136

y = 2.4662x - 9.12R² = 0.99

y = 0.38x + 19.04R² = 0.98

y = 0.051x + 30.25R² = 0.91

0

10

20

30

40

50

60

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Boxboard

Lag = 4d

d

y = 2.63x - 0.50R² = 0.97

y = 0.079x + 38.68R² = 0.93y = 0.46x + 29.29

R² = 1

0

10

20

30

40

50

60

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Food Waste + Boxboard

Lag = 0

e

y = 0.62x - 2.99R² = 0.99

y = 0.15x + 2.47R² = 0.96

y = 0.044x + 6.03R² = 0.91

0

2

4

6

8

10

12

14

16

18

20

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Newsprint

Lag = 7d

f

y = 1.15x - 0.19R² = 0.99

y = 0.21x + 8.13R² = 0.99

y = 0.084x + 11.18R² = 0.92

0

2

4

6

8

10

12

14

16

18

20

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Food Waste + Newsprint

Lag = 0

g

y = 9.64x - 31.58R² = 1

y = 1.03x + 29.87R² = 0.98

y = 0.068x + 55.27R² = 0.95

0

10

20

30

40

50

60

70

80

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Fine Paper

Lag = 4d

h

y = 6.93x - 5.93R² = 1.00

y = 0.60x + 42.32R² = 0.99

y = 0.13x + 53.76R² = 0.90

0

10

20

30

40

50

60

70

80

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Food Waste + Fine Paper

Lag = 1d

i

137

Yoshida et al. (2008) studied the effects of cellulose crystallinity, hemicellulose and lignin of the

enzymatic hydrolysis of miscanthus, using commercially available enzymes. They found that hydrolysis

took place in stages, amorphous cellulose, then crystalline cellulose, then hemicellulose; they also

found that when lignin was chemically removed (cf FP), the yield of monosaccharides increased. The

hydrolysis curves look like those in Figure 9.4. The implication is that substrate hydrolysis is the rate-

determining step in Daisy.

The rates and lag-times are summarized in Table 9.2. In all cases, the addition of food waste eliminates,

or greatly reduces, the lag at the beginning of the digestion process. In all cases except fine

paper and mixed fibre, the initial digestion rate is the same or higher with food waste addition, though

in the case of mixed fibre they are very similar. The final digestion rate of fibres is always higher with

food waste addition, though BA is unchanged. Generally, the addition of food waste improves both the

digestion rate and the ultimate yield, but again fine paper is the exception. For practical reasons, the

y = 3.69x - 8.69R² = 0.96

y = 0.27x + 30.70R² = 0.91

y = 0.058x + 38.35R² = 0.87

0

10

20

30

40

50

60

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Fibres

Lag = 4d

j

y = 3.40x - 0.63R² = 0.98

y = 0.087x + 39.15R² = 0.93

0

10

20

30

40

50

60

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Food Waste + Fibres

Lag = 0

k

y = 0.82x + 0.52R² = 0.99

y = 0.24x + 16.70R² = 0.96

0

5

10

15

20

25

30

35

40

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Food Waste + Bulking Agent

Lag = 0

m

y = 0.59x - 1.86R² = 0.98

y = 0.23x + 11.46R² = 0.98

0

5

10

15

20

25

30

35

40

0 20 40 60 80

Biog

as P

rodu

ced

(mL

at 3

10K)

Digestion Time (days)

Bulking Agentl

Figure 9.4 Comparison of digestion of each fibre, with and without FW addition; a) FW alone; b) and c) CB w and w/out FW; d) and e) BB w and w/out FW; f) and g) NP w and w/out FW; h) and i) FP w and w/out FW; j) and k) FB w and w/out FW; l) and m) BA. w and w/out FW. Graphs presented as a series of pseudo zero order reactions.

Lag = 7d =

138

total COD added to each bottle for the BMP tests had to remain constant and not increase as a result of

food waste addition (as it does in Daisy). Thus all the data presented above represent substrate

digestibility and synergy under conditions of constant COD, but without being able to measure the

synergistic effect separately. This was addressed by normalizing the BMP data to separate the

contributions of food waste and fibre substrates and obtain a true measure of the synergistic effect as

an increase in biogas production from each fibre; the calculations and results are presented next.

Calculating true synergy A method was developed to

compare the volume of biogas

produced from the FB plus FW

co-digested (FBFW) with the sum

of the volumes of biogas

produced from the same

quantities of the same individual

substrates digested separately

(FB+FW). By subtracting the latter from the former it was possible to determine true synergy as follows:

𝑉 = 𝑉 (0.13𝐶𝑂𝐷 + 0.87𝐶𝑂𝐷 ) − (𝑉 [0.13𝐶𝑂𝐷 ] +

𝑉 [0.87𝐶𝑂𝐷 ]) Eq 9.1

The calculations were based on the data in Figures 9.1 and 9.2 and Tables 9.1 and 9.3. The method is

described below using the combination of CB and FW as an example. The volume of biogas produced

from CB and FW as individual

substrates (Figure 9.2), was

used to calculate a biogas yield

per g of COD added for each.

a) In the co-digestion of

CB with FW (Table 9.1 BMP 14),

87% of the COD (71 mg) came

from CB and 13% (10 mg) from

FW (Table 9.3).

b) Using this information, the biogas curves for CB and FW digested individually (Table 9.1 BMPs 1

and 6 respectively), were re-plotted based on the prorated quantities of COD – 87% for CB and

13% for FW; the results are presented in Fig. 9.5a - individually in graphs 3 (CB) and 2 (FW)

Table 9.3 Biochemical Methane Potential Test 2 – Substrate Calculation

Substrate COD (g/gdry feed) COD (g) COD fraction COD (mg)

CB 1.14 0.71 0.87 80.9 BB 1.13 0.00 0.00 0.0

NP 1.33 0.00 0.00 0.0

FP 1.00 0.00 0.00 0.0 FW 1.25 0.10 0.13 12.5

BA

0 0.0 Total 1.000

0.81 1.00 93.4

Table 9.2 BMP graphs - biogas production rate in mL.d-1.mg-1CODadded at 310K Fibres alone Fibres + Food Waste

Substrate Lag - days 1 2 3 Lag - days 1 2 3

CB 4 0.030 0.004 0.0008 0 0.031 0.004 0.0011

BB 4 0.027 0.004 0.0005 0 0.028 0.005 0.0008

NP 7 0.006 0.002 0.0004 0 0.012 0.002 0.0008

FP 4 0.11 0.011 0.0008 1 0.074 0.006 0.0014

FW 0 0.053 0.004 0.0005 NA NA NA NA

FB 4 0.034 0.003 0.0007 0 0.031 NA 0.0082

BA 7 0.006 NA 0.0025 0 0.009 NA 0.0026

139

respectively. These graphs were then summed as a combined total in graph 4 (CB+FW). This

calculated prorated total was then compared to the result of co-digestion of the same

substrates, in the same proportions (Table 9.1 BMP14), shown in graph 1 of Fig. 9.5a (CBFW).

c) The difference between graph 1 and graph 4 (the yellow and the grey lines) therefore

represents synergy, the additional biogas produced from two substrates co-digested compared

to the sum of the same substrates, in the same quantities, digested separately.

d) This process was repeated for BB, NP, FP and BA, and the results presented in Figures 9.5c, 9.5d

and 9.5e respectively, using the same labelling convention. These calculations make it possible

to calculate the initial rate of biogas production from fibre, with and without FW added. This

shows the true effect of FW addition on rate and yield from fibre.

Rates and lag times for the normalized BMP data are

compared in Table 9.4. In all cases the addition of food

waste resulted in an increase in the digestion rate; the

effect is more pronounced with these normalized data

than with those presented in Table 9.2. Additionally, all

of these results confirm the ranking of digestibility of

the fibres: FP>CB>BB>NP as determined from the

coupon data presented in Chapter 8. It is important to

remember that the digestion of the substrates in the BMP bottles takes place under idealized

conditions of particle size distribution, with surplus inoculum, in a medium containing enough essential

nutrients to meet the requirements of the microbial community for the duration of the test. Figure 9.6

shows the effect of synergy on the destruction of each substrate (expressed as cumulative millilitres of

biogas) at 39 days and at 74 days. The synergistic effect of food waste, measured both in absolute

terms and as a percentage, is ranked as follows: CB>BB>FP>NP; the benefit of food waste addition to

the digestion of fine paper is principally the elimination of the initial lag time; beyond that, the

digestion rate and yield for FP are not much enhanced by the presence of the food waste, but note that

the coupon data tell a different story (Section 8.3.5, Fig. 8.18). The cardboard and boxboard always

appear ranked in that order, the difference between them is small. This is not so surprising in the BMP,

given the trip through the Wiley mill and a 60 mesh screen, but it still seems true from Daisy’s coupon

data for which the particle size is anywhere from 5 mm to 20 mm or larger. The BA did give higher yield

in the BMP than in Daisy’s coupons, probably as a result of size reduction in the Wiley mill.

Table 9.4 BMP graphs – rates Biogas ml/d/mgCOD at 310K

Substrate Lag - days Lag - days

CB 4 0.025 0 0.036

BB 4 0.024 0 0.033

NP 7 0.009 0 0.010

FP 4 0.066 1 0.074

FW 0 0.053 NA NA

BA 7 0.009 0 0.009

140

y = 0.97x + 0.30R² = 0.97

y = 0.82x - 0.67R² = 0.98

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60 70

Cum

ulat

ive

Biog

as (m

l at 3

10K)

Digestion Time (days)

Newsprint

Max.

FWNP (1)

NP+FW (4)

NP (3)

FW (2)

c

y = 3.15x - 2.93R² = 0.97 y = 2.19x - 3.27

R² = 0.97

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60 70

Cum

ualti

ve B

ioga

s (m

l at 3

10K)

Digestion Time (days)

Boxboard

Max.

FWBB (1)

FW+BB (4)

BB (3)

FW (2)

y = 3.40x - 3.26R² = 0.98

y = 2.29x - 3.85R² = 0.96

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60 70

Cum

ulat

ive

Biog

as (m

l at 3

10K)

Digestion Time (days)

Cardboard

Max

FWCB (1)

FW+CB (4)

CB (3)

FW (2)

a

b

141

In an experiment designed to compare the digestibility of the fibre mixture at different levels of food

waste addition (Figure 9.7), it was found that the increase in digestion rate is very high at the beginning

of the experiment (>110%), then rapidly falls as digestion proceeds, finally levelling off at about 30% at

around 30 days. The extent of this effect seems to be completely independent of the amount of food

waste added, contradicting the coupon data obtained from Daisy, which showed that the greater the

amount of food waste added, the greater the digestibility of the substrates (with the exception of

newsprint and bulking agent). It seems highly likely that the optimum conditions of a BMP, namely

ample inoculum, macro and micro nutrients, and the fine particle size distribution of the substrate play

a role in this discrepancy.

y = 6.93x - 5.93R² = 1.00

y = 6.02x - 8.65R² = 0.91

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60 70

Cum

ulat

ive

Biog

as (m

l at 3

10K)

Digestion Time (days)

Fine Paper

Max.

FPFW (1)

FP+FW (4)

FP (3)

FW (2)

y = 1.01x - 0.36R² = 1.00

y = 0.88x - 0.10R² = 0.98

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60 70

Cum

ulat

ive

Biog

as (m

l at 3

10K)

Digestion Time (days)

Bulking Agent

Max.

FWBA (1)

FW+BA (4)

BA (3)

FW (2)

Figure 9.5 Calculation of synergy – normalized for COD content a) CB; b) BB; c) NP; d) FP; e) BA. The difference in initial gradients represents the effect of FW on the rate of biogas production from each fibre

e

d

142

BMP 2 contributed a great deal to the understanding of the digestibility of the fibres and the

synergistic effect brought about by food waste addition, but it did not answer questions about the

underlying mechanism. There seem to

be four possibilities;

Changes in the diversity and

abundance of the microbial

community

Changes in C:N ratio

Loss of nutrients present in

FW

The presence of enzymes in

the FW (or as products of

digestion of the FW) which

directly assist the hydrolysis of

the FB or

Some combination of these.

The complementary research being carried out by Peter HyunWoo Lee should be able to address the

first. The second and third are possible, but the effect on biogas production of changes in FW addition -

58.1

36.3 35.124.7

12.9

60.0

38.4 36.731.6

14.3

1.3

7.25.6

0.8

0.5

2.6

9.0 7.5

2.0

2.7

0

10

20

30

40

50

60

70

Cum

ulat

ive

Biog

as m

L. (a

t 310

K) 39 days 74 days

Figure 9.6 Synergy by fibre at 39d and 74d; normalized to constant COD; synergistic contribution in dark shading

0

20

40

60

80

100

120

140

0 20 40 60 80

Perc

ent I

ncre

ase

in B

ioga

s Pro

duct

ion

from

Fib

re

Digestion Time (days)

Synergy vs. Digestion Time - Effect of FW addition

FW 6%CODFW 8%CODFW 11%CODFW 17%CODFW 20%CODFW 27%COD

Figure 9.7 Percent increase in biogas yield, as a result of FW addition vs time, at six levels of food waste addition

143

rapid when FW is reduced, slower when FW is increased - would suggest something more complex. A

single leach bed with less food cannot alter the nutrient concentration by more than 3 or 4%; likewise

the C:N ratio. When the FW content in Daisy’s feed was reduced, the effect on biogas production was

almost instantaneous, and within 2 to 4 weeks biogas production had stabilized at a lower level (Figure

8.11). The next reduction, and the next, produced exactly the same effect. When the CODFW was all

added back in Week 50, gas production jumped in the first week, grew more modestly over the next

five weeks, then produced two big jumps in Weeks 56 and 57 (Section 8.3.3, Fig. 8.12). This perhaps

suggests that some critical condition had been reached that promoted synergy; in all it took 8 weeks to

restore gas production to its former level (Figure 8.11). These observations would suggest that the

synergy is in some way caused by the FW, and that the magnitude of the effect is related to the CODFW

addition rate, but that once CODFW is added back, it takes time for gas production to fully return. In

contrast, when Daisy was restarted at Week 26 after a 10d shutdown, gas production immediately rose

to exceed the prior weekly rate by about 100 L.wk-1 (20%). This lasted for three weeks suggesting that

Daisy was converting intermediate products that had accumulated during the shutdown.

Yun (2015) showed that the addition of small amounts of FW enhanced the digestion of waste activated

sludge, and that extracellular hydrolytic protease was primarily responsible. Yuan (2012) showed that

pre-treatment of waste paper and cardboard with a microbial consortium enhanced methane

production, and used the same method to quantify synergy as the one used in this work. The

consortium was specifically developed to degrade cellulose under thermophilic conditions, but has not

been fully characterized. This work offers some insight into possible causes, and more study will be

required, but there is evidence in the literature that enzymatic activity is the mechanism by which

biogas yields from paper and cardboard are enhanced. This is explored further in Section 9.2.

9.1.2 Biochemical Methane Potential Test 3 – One Litre Bottles and Coarse Substrate

Although BMP 2 provided great insight into the digestibility of individual substrates, and the effect of

food waste addition on biogas yield and rate, it was conducted under idealized conditions of finally-

divided substrates, excess inoculum and an anaerobic medium complete with all necessary macro and

micro nutrients, and the entire system was appropriately buffered. This approach falls short of

mimicking the behaviour in Daisy in four specific ways; substrate preparation, quantity of inoculum,

lack of leachate recirculation, inability to add food waste progressively to sealed bottles.

BMP 3 was designed to obviate three of these four differences (introducing leachate recirculation was obviously not possible);

144

Add substrates in exactly the same form as they were fed to Daisy (necessitating the use of 1 L bottles).

Add 120 g of substrate COD per bottle, 1/20 of the quantity of COD fed to one of Daisy’s leach beds.

Use leachate and sludge recovered at the conclusion of Period 6 from Tank 2 of Daisy as inoculum.

Equip three of the bottles with a gas-tight food waste feeding port. On paper, this concept was fine but it did create a

practical problem; measuring the volume of biogas

produced on this scale would be a challenge. A 160

mL serum bottle could be expected to produce 60

mL of biogas over a 10 week period. Under the

proposed plan each bottle would generate

approximately 60,000 mL of biogas over the

duration of the experiment. Professor Savia

Gavazza offered her knowledge and expertise to

solve this and other related challenges.

Experimental Setup Figure 9.8 is a schematic of Savia’s experimental

setup. As biogas is produced in Bottle a) it leaves

under slight pressure through plastic tubing and a

hypodermic needle into Bottle b) where it passes

through a solution of sodium hydroxide that scrubs

out the CO2; it also contains bromothymol blue

which turns green at pH 7.6. As gas accumulates in

b CH4

Biogas

Substrate

Water bath 37°C

NaOH in

Displacement gas meter and CO2 scrubber

NaOH soln.

Vent

Figure 9.8 Large BMP experimental set-up

c a

Plate 15. BMP 3; Substrate bottles a) in water bath (rear); gas meter bottles b) above and c) bottom foreground

Plate 16 BMP 3 Entire set-up with 11 bottles

145

Bottle b) it displaces the scrubbing medium to Bottle c). Periodically, as required, Bottle c) is removed

and weighed and the contents returned to Bottle b). This last step requires clamping the gas line and

physically removing and replacing Bottle b). Occasionally, scrubbing solution accumulates in the line

between Bottle b) and Bottle c); the clamped side arm is used to correct this condition. Plate 15 shows

a close-up of how the bottles are assembled; Plate 16 shows the complete set-up, with the water bath

in a spill containment tray, and the entire experiment housed within a fume hood. Though Savia and I

planned the experiment, built the system, and started it up together, the design concept and all the

subsequent gas measurement and analytical work were performed by Savia.

The experiment was set up as follows;

Experiments 1, 2 and 3 all contained cardboard (CB), boxboard (BB), newsprint (NP) and fine

paper (FP) in the same proportions as were fed to Daisy and in the same physical form; the

total mass added was 55 g (cf Daisy at 1,100 g). To ensure uniformity, each fibre was weighed

separately for each bottle.

Experiments 1 and 2 had food waste (FW) added at CODFW28%; Experiment 3 contained fibre

alone.

Experiment 1 was equipped with a gas-tight port to allow periodic injection of additional FW.

Two controls were run; one bottle containing glucose plus inoculum and one bottle containing

inoculum alone.

Table 9.5 Total Quantities of Substrates, Inoculum and Medium for BMP3, Large BMP (g unless otherwise stated)

Expt. No. CB BB NP FP FB FW (pulp)

FW/wk FWTotal Inoc LT (mL) Glucose

1 a,b,c FB+FW+FWadded 22.0 17.5 5.5 10.0 55.0 156.0 4.5 192.0 600

2 a,b,c FB+FW 22.0 17.5 5.5 10.0 55.0 156.0 156.0 600

3 a,b,c FB 22.0 17.5 5.5 10.0 55.0 600

4a NCON 600

4b PCON 46 554 2.0

CODsubstrate = 83.62 g/bot. (1 and 2) 62.17 g/bot. (3) VSinoculum = 15.5 g.L-1 9.3 g VS per bottle CODsubstrate:VSinoculum ratio = 9.0:1 lower than previous BMPs, lower than lit. recommends, but replicates Daisy CODglucose:VSinoculumratio= 3.0:1 0.71 gTS inoculum per bottle 45.8 mL LT+ sludge

Expt No. Description COD Content long-term averages (dry method)

1 a,b,c FB blend + FW at 28% by COD + 4.5g FWadded/wk Subs. CB BB NP FP FW Glucose

2 a,b,c FB blend + FW at 55 wt% (28% by COD) CODg/g 1.14 1.13 1.33 1.00 1.30 1.07

3 a,b,c FB blend 4a Negative control, Inoculum alone

146

All bottles received 600 mL of inoculum (leachate and sludge from Daisy recovered at

shutdown) except the positive control which received 46ml; the details are provided in Table

9.5.

Results The first thing to report is the failure of Experiment 1. On Day 25 all three bottles popped their septa,

and deposited most of their contents in the fume hood. The proximate cause was a blockage of the

hypodermic needles in the gas lines; the underlying cause is unknown, but was probably related to the

quantity of material in the bottles and

the smallness of the hypodermic needle.

In retrospect a sample of half the size

would probably have given equally good

results and been less difficult to handle.

The results for the controls and the

remaining two experiments are

displayed in three graphs. Figure 9.9

shows the two control experiments.

Figure 9.10 is a close-up of the first 25

days, and Figure 9.11 shows the entire

Figure 9.10 Large scale BMP 3; first 25 days showing initial acid inhibition in Experiment 2 FW + FB

0

2,000

4,000

6,000

8,000

10,000

0 5 10 15 20 25 30

Cum

ulat

ive

Met

hane

Pro

duct

ion

(mL)

Digestion Time (days)Food Waste + Fibers + Leachate Fibers + Leachate Leachate

0

200

400

600

800

1000

1200

1400

1600

0 2 4 6 8 10

Cum

ulat

ive

met

hane

(mL)

at 2

97K

Digestion Time (days)

Glucose

Leachate

Figure 9.9 BMP 3 Negative (leachate) and positive (glucose) controls

147

experiment as cumulative methane production versus time. The positive control converted 96% of the

glucose added. The negative control generated 737 mL of methane (at 297K) by day eight. The negative

control is also shown in Figures 9.10 and 9.11; it has not been subtracted from the totals. The

differences between Experiment 2 (fibre + food waste) and Experiment 3 (fibre alone) are striking.

During the first three days, Experiment 2 produced a short-lived burst of methane then slowed to a

crawl for the next 10 days. It appears likely that the bottles became acidic as a result of the rapid

hydrolysis of food waste, and until that acid was consumed digestion was partially inhibited. This is

supported by the finding that at Day 17, shortly before it failed, Experiment 1 had dropped to pH 5.2,

an inhibitory condition presumably brought about by the FW it contained. During the same period,

methane production in Experiment 3 (fibre alone) began more slowly, caught up with Experiment 2 by

Day 10, then overtook it. The curves crossed again at day 17, at which point methane production in

Experiment 2 rose sharply for about 50 days before beginning to slow, after which the two cumulative

gas curves ran more or less parallel again. Experiment 3 produced methane at a very steady rate

throughout, finally ending at 11,378 mL of methane and 49% COD conversion, while Experiment 2

ended at 24,315 mL of methane and 81% COD conversion. Synergy was calculated as the percent

increase in methane production from fibre as a result of food waste addition, using the method

previously described. It was found to be 73% at 70 days, and 45% at 112 days. This decline in synergistic

contribution with time was also observed in the previous BMPs. Given enough time, fibre without food

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

18,000

20,000

22,000

24,000

26,000

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115

Cum

ulat

ive

Met

hane

Pro

duct

ion

(mL)

Experimental Time (days)

49% COD conversion

737 mL of CH4 from leachate, 8167 mL CH4 from FW at 297K

81% COD conversion

Leachate

Experiment 3 FB

Experiment 2 FB + FWSynergy at 71d = 73%

Synergy at 112d = 45%

Figure 9.11 Large-scale BMP to 112 days; acid inhibition reverses itself in Exp. 2, methane production rises sharply, synergy evident

148

waste will partially close the gap. The detailed results are provided in EDF7. Once again, the synergistic

effect of food waste on the digestion of fibres was demonstrated, this time using substrates in exactly

the same form and under the same conditions as those prevailing in Daisy, except for leachate

recirculation. Two important observations; firstly, without recirculation of leachate and the buffering

effect of leach beds of stable waste, the addition of food waste at CODFW28% created inhibitory

conditions of low pH which lasted for about three weeks; secondly, the synergistic contribution, while

still large at 73%, was much lower than the 225% achieved in Daisy under similar conditions (275%

based on the less conservative more realistic assumption of 78%FW conversion). A plausible

explanation for this large difference is that leachate recirculation in Daisy prevents inhibition by VFA

formation and low pH, and does this by distributing soluble COD to the older leach beds, thereby

enhancing substrate conversion.

9.2 Enzyme Activity Assays

In Section 8.3 a case was made for studying enzyme activity in the search for a mechanism for synergy.

After some thought, and discussions with Sasha Yakunin, we decided that initially five samples would

be analyzed. Sasha kindly offered the services of Greg Brown to perform the analytical work.

Sample preparation was carried out as follows. Leachate was extracted (physically squeezed with a

garlic press) from three digestate samples obtained from leach beds S.043, S.050 and S.055 (high,

medium, and low COD destruction efficiencies respectively), a FW sample was blended with water and

a leachate sample taken from Daisy. Twenty mL of each sample were centrifuged at 20,000 g for 30

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Digestate19.5%

Digestate43.0%

Digestate59.4%

Food waste Leachate

nM o

f sub

stra

te c

onv.

/min

./m

L le

acha

te

Source of samples (with substrate destruction efficiencies where applicable)

Esterase Protease Cellulase

Figure 9.12 Enzyme activity assays. Activity higher at high substrate destruction and in food waste and leachate

149

minutes; two of the samples did not pellet well, and were centrifuged again at 24,000 g; the

supernatant was recovered with a pipette. Each sample was spin-filtered at 3,600 g to concentrate the

proteins. Greg performed the following analyzes on each sample; total protein (Bradford Assay), and

enzyme activity for esterase, protease and cellulase; the procedures, including standard substrates, can

be found in Appendix C. Protein separation by denaturing gradient gel electrophoresis was also carried

out, and samples of the bands excised, and analyzed by Andy Quaile using a gas chromatograph/mass

spectrometer (GCMS) in an unsuccessful attempt to identify specific enzymes. Ideally, a metagenome

for the microbial population was required to do this successfully. The results for all the samples are

presented in Table 9.6 and Figure 9.12. The measured esterase, protease, and cellulase activity,

expressed as µM of subst. conv./min./ml extract, correlates well with COD destruction efficiency; both

the leachate and FW showed enzymatic activity too. The activity, in all cases, appears related to the

amount of food waste added, and consequently the substrate digestion efficiency. The food waste itself

also exhibits a level of enzyme activity that at least implies that FW is a source of enzymes similar to

those present in the digestate. The inference from this analysis is that synergy, at least in part, is an

enzymatic phenomenon and dependent upon the amount of food waste added.

Table 9.6 Enzyme Activity Assays

Sample Total Protein in Gel Enzyme Activity Activity

Description COD Destr. Effic.

%

OD - corrected for spin filter conc. µM of subst.

conv./min/mg of protein

nM of subst. conv./min/ml extract

Vol. µl

Conc. µg/µl

Mass µg

Ester Prot Cell X Conc

Ester. Prot. Cell. Ester. Prot. Cell.

S.043 19.5 12.5 0.16 2.00 0.089 0.015 0.020 42 8.27 0.23 1.90 7.87 0.22 0.30

S.050 43.0 12.5 0.4 5.00 0.125 0.019 0.039 36 4.00 0.10 1.24 11.1 0.28 0.57

S.055 59.4 12.5 0.15 1.88 0.489 0.086 0.116 7.4 8.58 0.25 2.03 43.5 1.3 1.7

FW (S.073)

12.5 0.62 7.75 0.389 0.194 0.098 12.7 2.83 0.24 0.71 34.6 2.9 1.5

Leach. 19/08/16 12.5 0.03 0.38 0.241 0.059 0.034 11.6 33.19 1.36 4.69 21.5 0.9 0.5

9.3 Composting of Digestate

The digestate produced by Daisy is still biologically active, but it is not stable, it is not sterilized and

must be aerobically cured (composted). The initial plan was to cure the digestate by aerating it in the

leach beds at the conclusion of the digestion period, generally 42 days. As reported previously it was

not possible to do this on such a small scale. All the digestate was saved, stored at 4°C until the end of

the experiment, then transported to Miller Waste Systems for aerobic curing in a single batch

experiment of about 300 kg. The work was conducted by Kyle Schumacher of Miller, and it is his

summary, with my edits, which follows.

150

9.3.1 Aerobic Static Pile Design

The purpose of this trial was to assess the compostability of the digestate recovered from Daisy,

accumulated over a period of 88 weeks, using a specially constructed “bunker style” aerated static pile

(ASP) system; a schematic plan view and side elevation of the system are shown in Figures 9.13 a) and

b) respectively. The ASP unit was constructed using a 1.2 m3 metal container, and a ¼ hp aeration

pump. The aeration pump feeds a manifold connected to a series of lateral perforated pipes. These

pipes were covered by a coarse layer of compost “overs” (the coarse material screened out in the

manufacture of finished compost) to ensure proper air distribution to the composting mass.

Airflow is controlled by an on-off timer and adjusted according to the internal temperature of the

material. The entire system was housed inside Miller’s composting building in Pickering, Ontario.

9.3.2 Description of Experiment

Approximately 320 kg of digestate was mixed with 10% by weight of actively composting leaf and yard

waste (LYW). The resulting 352 kg of “mix”, occupying a volume of approximately 471 L, was placed

between two insulating layers of yard waste, each separated from the mix by layers of nylon mesh.

Aeration was cycled continuously, 15 minutes on and 15 minutes off, for 10 weeks. The temperature

was monitored throughout the trial using a Reotemp compost temperature probe, inserted into the

approximate center of the composting digestate. Once per week the top insulating layer was removed

to gain access to the composting mix. Once uncovered, the digestate was turned using a pitchfork and

observations were noted. Following the turning process the top insulating layer of yard waste was

returned to the vessel. The digestate showed a tendency to form clumps, which composted more

slowly. Because of this, the trial was extended beyond the 6 weeks originally planned, to allow

Figure 9.13 Aerobic Static Pile Composter; a) plan view, b) side elevation

a b

151

additional time and additional turning to improve the quality of the product. Samples of the leaf and

yard waste, digestate, starting mix, and resulting composted digestate were sent to A&L labs for

analysis and the results are presented below. Plate 17 shows the ASP in operation; what it lacks in

beauty it makes up for in performance.

9.3.3 Results

Upon receipt, the raw digestate had a mild “rotting/putrid” odour, which only lasted for about three

weeks. By week four the digestate had taken on a pleasant earthy odour. The digestate consisted

primarily of remnants of cardboard/paper materials from the anaerobic digestion process, and had a

moisture content of 75%. To help induce proper composting, 10% by weight of actively composting leaf

and yardwaste was added, reducing density and moisture content slightly. The most challenging

characteristic of the digestate for composting purposes was the clumping of the material. Some of

these clumps were already formed, some formed during the trial. Efforts were made to break them up

using a pitchfork during the turning process, to increase surface area for microbial activity;

nevertheless, some clumping remained. On a commercial scale these clumps would be broken up by

mechanical equipment (Scarab turners, loaders, etc.) and would not be an issue. The extension of the

compost trial to 10 weeks allowed most of the clumps to be broken up in a way that simulated a

commercial operation.The light beige/grey appearance of the mix changed quickly once the materials

Plate 17. Aerobic Static Pile Composter; housed inside compost building; note blower on the left and piping

152

began to compost. From visual inspection of the composting mix at each turning event, it was apparent

that materials were becoming darker, more decomposed, and of finer particle size. By the end of the

trial, most of the large clumps were gone and a dark fine compost material remained. There was still

visible evidence of woody materials (sticks, woodchips) from the digestate and the LYW, which is typical

in composting.

The digestate took about 9 days to rise above 35oC; typically this is accomplished in 3-5 days (Czekala et

al., 2017). A maximum temperature of 45oC was reached and maintained for approximately two weeks

before beginning to cool down. A temperature of 55OC is typical (Czekala et al., 2017; Drennan &

Distefano, 2010; Pandey et al., 2016). During the trial steam was visible, and even more so during the

turning operation (Plate 19). Various conditions probably contributed to the inability to reach 55OC; the

small size of the vessel, moisture content, density and porosity, and ambient temperature of the

building may all have been factors. Reaching high temperatures would probably not be a concern at an

industrial scale, because of pile size alone. After this active phase, the material cooled down to about

25-28OC for the remainder of the trial. This is typical behaviour during curing (Czekala et al., 2017;

Drennan & Distefano, 2010; Pandey et al., 2016).

At the conclusion of the trial, lots of white mold was speckled throughout the composted digestate.

This is considered a good indicator that the materials have become mixed, composted, and stable

enough to support smaller colonies of mold amongst the materials, rather than the larger masses seen

Plate 18. Compost feed and product over time. Progressively turned black and less coarse

Digestate Mix 29/04/2017 17/05/2017 29/05/2017

153

earlier in the process. Plate 18 shows a photographic progression of the material from the compost

vessel.

Samples were sent for laboratory analysis for pH, moisture/TS, C:N ratio, bulk density, particle size

distribution by sieve analysis and Solvita (maturity) tests. The results are presented in Table 9.7, and are

discussed below.

pH Normally during composting there is a characteristic rise in pH as

the materials decompose, however this was not observed. Drennan

& Distefano (2010) reported pH 6.4, Czekala et al. (2017) pH 7.9 and

Pognani et al. (2010) pH 8.6. The digestate, leaf and yardwaste, and

mix had an average pH of 7.54, 6.77, and 7.42 respectively while the

finished composted digestate had a pH of 5.19. The pH of the final

compost was abnormally low. The reason for this is unknown; these

data were not used as part of the determination of compostability.

Moisture/TS The average starting moisture content of the materials was

digestate: 75%, LYW: 60%, and the mix: 68%. The higher starting

moisture content of digestate (Yazdani et al. (2012)reported 61%) is

probably why the composting mass took longer to reach

thermophilic temperatures and why the desired temperature of

50°C to 60°C (Czekala et al., 2017; Yazdani, 2010) was never

reached. Comparing the moisture content of the starting mixture to that of the finished compost

showed a decline of approximately 20%, ending at 49.7%, close to the figure of 47.7% reported by

Yazdani et al. (2012).

Carbon:Nitrogen Ratio The C:N ratio plays an important part in aerobic composting. In theory, the digestate, and leaf and

yardwaste should have a fairly high carbon content since they consist of sticks, leaves, paper products,

Table 9.7 Properties of digestate, leaf and yard waste and mixture

pH moisture % TS % C:N Ratio Bulk density kg.m-3 Solvita Index

Digestatea 7.11 68.1 31.9 66 216 5

Leaf and yard wastea 7.07 67.2 32.8 60 214 5

Digestate Mixa 7.43 70.8 29.2 31 209 5

Finished compost 5.19 49.7 50.4 23 236 8 aaverage of two samples

Plate 19 Steam rising from active compost

154

and ash wood chips. However, the reported C:N ratios for the starting materials on the first set of

samples analyzed showed a large discrepancy and were obviously incorrect. Based on the more realistic

values from the second set of samples, the C:N ratio of the mix was 31:1, falling to 23:1 at the end of

the composting period; this is within the normal range as reported by Czekala et al. (2017).

Bulk density The bulk density on a dry weight basis increased from 201 kg.m-3 to 236 kg.m-3 for the finished compost;

these are typical values measured at full scale by Miller Waste Systems Inc.

Solvita (maturity) The most important method for determining compostability is the measurement of microbial activity

using the Solvita test, a commercially available test which combines measurements of the decline in

production of NH3and CO2 as composting progresses (Drennan & Distefano, 2010). The result is

compared to a Solvita index scale of 1 to 8; the higher the number the lower the CO2/NH3 emissions

and the better the product. A Solvita index of 7 or 8 is required in Ontario to demonstrate mature

finished compost.

The starting materials, including the digestate mix, all showed a Solvita index of 5. Though raw organic

waste typically has an index closer to 3, thisvalue was expected since the digestate had already been

anaerobically treated and the volatile carbon content reduced. During the trial, several samples were

collected and had their Solvita index measured to track progress. The index remained at 5 until about

Week 7 when the index fell to 6. At 10 weeks however the finished compost showed a fully mature

index of 8. Having gone from an index of 5 to 8 confirms the compostability of digestate, and confirms

its potential use as a finished compost material.

Sieve analysis Sieve analyses of the beginning mix and the finished compost were performed using screen sizes typical

for compost analysis. The two samples of digestate mix from the start of the experiment had quite

different particle size distributions, but the finished compost was certainly finer than the starting

materials. An increase from 26% to 43% of material passing the ¼” sieve was observed, giving further

confirmation of the compostability of

the digestate. The difference was easily

seen in a visual comparison of the two

materials.

Table 9.8 Sieve analysis of digestate mix and compost - percent minus

Sieve size 2" 1" 1/2" 3/8" 1/4"

Digestate Mixa 100.0 97.8 60.2 44.6 26.0

Finished compost 100.0 82.9 63.7 57.2 43.0 aaverage of two samples

155

9.3.4 Aerobic Curing Conclusions

The final compost had a desirable “earthy compost” odour. The physical evidence of composting was

very apparent; the colour changed from beige/gray to a very dark earthy colour with a considerable

proportion fine particles. There was still evidence of clumps that had not been broken up properly

during the trial. Smaller particle size was also measured by the final sieve analysis. The Solvita test, the

best indicator of compostability, resulted in an index of 8 (fully mature compost).

Unfortunately the C:N ratio and pH data could not be used with confidence to prove or disprove

compostability. However, with the remaining data, physical observations, and temperature behavior

observed, it can be concluded that the digestate is compostable.

The composting process could have been improved by increasing the active yard waste proportion to

20-25% to decrease composting time, and encourage higher temperatures. On a commercial scale,

results would also benefit from larger pile size, higher ratios of active compost to digestate, and the use

of mechanical turning to break up clumping.

156

Chapter 10. Discussion, Commercial Implications, Conclusions and Recommendations

In this chapter, the results obtained from the operation of Daisy and presented in Chapter 8, are

combined with the findings from the supporting experiments presented in Chapter 9 to provide a

complete picture of Daisy’s performance. That performance is then compared to the most relevant

published information available from the literature, recognizing that this is a particular challenge

because of the limited published information related to the digestion of solid waste with a high content

of wood-derived lignocellulosic fibre.

This is followed by a discussion of how well the research and the results met the objectives set out in

Chapter 2, and whether the hypothesis, in whole or in part, was supported by the results obtained.

The new information obtained through this research and the understanding gained about the

performance of this technology are used to reevaluate the commercial prospects of the technology in a

comparison with the previous analysis, presented in my M.Eng. thesis.

Lastly, final conclusions are drawn and recommendations made for future research.

10.1 Discussion of Research Results

10.1.1 Synergy

The results presented in Chapter 8 describe behaviour that offers persuasive evidence that food waste

synergistically enhances the digestibility of lignocellulosic fibres and that the enhancement manifests

itself as increased substrate destruction efficiency, and an increase in the rate and yield of biogas

production. The coupon data provided strong supporting evidence that food waste addition was the

motive force behind synergy and revealed that the synergisitic effect was different for different fibres.

In Chapter 9 the results of BMP 2 confirmed that synergy was real, and confirmed that it affected the

digestibility of different fibres differently i.e. FP>CB>BB>NP, in exactly the way suggested by the coupon

data. The results of BMP 3, designed to more closely mimic conditions inside Daisy, also confirmed

synergistic methane production, but displayed early signs of incipient inhibition caused by low pH. In

both these BMPs the size of the synergistic effect was much smaller than in Daisy, where the effect was

found to be very large. For example at CODFW29%, the highest food waste addition rate tested, the

synergistic effect in Daisy, measured as the increase in biogas from fibre, was 225% at 42 days. Under

nominally identical conditions, the third BMP showed the aforementioned inhibitory behaviour, and

achieved a maximum synergistic increase of “only” 73% in 112 days. The results presented in Section

157

8.4, describing biogas production from an individual leach bed, and rising leach bed temperatures after

fresh waste was added, suggest that the buffering effect of partly digested substrates, combined with

leachate recirculation, enhanced the synergistic effect in Daisy, compared to the static conditions of the

BMP. If this is true, the sequentially-fed leach beds, central to the design of Daisy, play an important

role in synergy as well as enhancing stability, and overall performance.

A plausible mechanism for synergy was needed, together with some additional confirmatory evidence.

A mechanism involving the presence or absence of specific methanogens was discussed in Section 8.3.5

and shown to be inconsistent with VFA creation and consumption. Another possibility, that synergy is

an enzymatic process whereby hydrolytic enzymes already present in the FW or created as a product of

FW digestion, seems more plausible.

The enzyme activity assay provided evidence that enzymatic activity is at least partly responsible for

synergy. Gan et al. (2003), in a review paper on the enzymatic hydrolysis of cellulose, determined that

this is a surface phenomenon, and that only a small fraction of the cellulosic substrate is undergoing

active hydrolysis at any given moment (less than 2% for particles smaller than 100 µm). Zhang et al.

(2007) conducted experiments on the digestion of vegetable and flower wastes in a two-stage digester

comprising a single leach bed and a methanogenic reactor. Leachate was recirculated through the leach

bed but was diluted with the effluent from the methanogenic reactor. They were able to show that this

experimental arrangement accelerated the degradation of the solid wastes, and that hydrolysis was

enhanced by the dilution effect of the methanogenic reactor effluent. They also found that the

hydrolysis of these organic solid wastes was mainly attributable to cell-free enzymes, followed by

biofilm-associated enzymes; the relative contribution of each remaining a matter of debate. Pommier

et al. (2010) in an experiment digesting samples of paper and cardboard, under conditions in which

enzyme concentration was not a limiting factor, concluded that substrate size reduction did not

improve accessibility to enzyme binding sites. Judging from the pseudo-zero order graphs of biogas

production for individual fibres in BMP 2 (Section 9.1.1), hydrolysis was being governed by substrate

availability in the form of the surface phenomenon described by Gan et al. (2003). If this is correct, and

the synergistic effect is attributable to enzymes present in food waste, or created by its digestion,

hydrolysis in Daisy must still be enzyme-limited, because the synergistic effect had not levelled off even

at the highest CODFW addition (Figure 8.16). In summary there is persuasive evidence, in both literature

and in the results of this study, that synergy is the result of enzymatic activity and that this process is

enhanced by the leachate recirculation in a two-stage anaerobic digester and particularly the return of

UASB effluent to the leach beds.

158

One last word on synergy. All of the previous calculations of the magnitude of the synergistic effect

have been based on the assumption that 100% of CODFWadded is converted; but we know from BMP 2,

carried out under presumably ideal

conditions of inoculum/substrate ratio,

macro and micro nutrients, and very fine

particle size, that CODFW is only 78%

converted. If this conversion were translated

to Daisy’s results, what would this mean to

the magnitude of the synergistic effect?

Figure 10.1 reproduces Figure 8.16 (FW at

100% conversion) with FW at 78%

conversion added. The amount of synergistic

biogas produced is of course larger (by

26.6% at CODFW29.3%) and the relationship

between food waste added and synergistic biogas produced per kilogram of CODFB is linear, with a very

good regression coefficient. At some level of food waste addition this relationship must end. Perhaps

this will occur when the only

undigested portion of the fibres

remaining consists of lignin.

Plate 20 illustrates why 22% of

CODFW remains undigested. A

piece of corn cob from Daisy’s

digestate has been picked clean

of more easily digested

material, leaving a

lignocellulosic ‘skeleton’

behind, apparently resistant to

the effects of synergy. It is also possible to calculate the destruction efficiency of the fibres, on a COD

basis, from the synergy data. This is shown, calculated two ways, assuming 100%FW conversion and the

more realistic 78%FW conversion in Figure 10.2. Destruction efficiency is 63.6% at 29%CODFW vs. 16.5%

at 0%CODFW, a 285% increase (assuming 78% FW conversion).

y = 10.17xR² = 0.99

y = 8.04x + 4.03R² = 0.99

0

50

100

150

200

250

300

350

0 5 10 15 20 25 30 35

Syne

rgist

ic B

ioga

s L/k

gCO

Dfib

re a

dded

(at S

TP)

Percent CODtotal as FW

FW at 78% conv. FW at 100% conv.

n = 4

n = 4

n = 4

n = 16

n = 6

n = 6

For all x axis values n = 27

Figure 10.1 Synergistic biogas - FW at 78% conv. vs FW at 100% conv.

3 cm

Plate 20. Partially digested corn cob; unconvertible COD

159

Like the biogas yield curve, this suggests that still higher FB destruction efficiencies might be possible at

higher FW addition.

10.1.2 Aerobic Curing

The digestate from the digestion of solid waste must be composted to ensure that it is no longer

biologically active, and to turn it into a potentially usable product. This was the last unit operation

which needed to be demonstrated and was always considered to be essential to overall success; lab

scale tests on a small scale during Daisy’s operation failed, but a single experiment with all of Daisy’s

digestate was successful. The digestate is not difficult to compost and can be combined with other

wastes, such as leaf and yard waste, and co-composted without difficulty.

10.1.3 Comparison of Daisy’s Performance to the Literature Daisy’s performance is compared to the somewhat limited published information on the digestion of

similar substrates. The comparisons, shown in Table 10.1, are based on methane yield expressed as

mLCH4.g-1VSadded and destruction efficiency expressed as a % VSdestroyed. At CODFW29.3% Daisy’s yield was

296 mLCH4.g-1VSadded at a substrate destruction efficiency of 69.4% (as VS). Daisy’s results are shown at

six different levels of food waste addition and compared to those of four published studies; Pommier et

al. (2010), Yuan et al. (2012), Eleazer et al. (1997), and Zhang et al. (2012a). The first three specifically

studied wood-derived lignocellulosic fibres such as CB, BB, NP, and FP. Zhang et al.’s substrate was mr-

OFMSW but also included are their results for source separated food waste (ss- FW). For each, the type

Figure 10.2 Fibre destruction efficiency. Shows synergistic effect of FW addition

0

10

20

30

40

50

60

70

0 5 10 15 20 25 30

Perc

ent C

OD

fibre

Des

truc

tion

Percent CODtotal as Food Waste

100% FW conv. 78% FW Conv.

160

and size of the experiment (for example 35 L CSTR, or 6 g BMP etc.), the substrate(s) digested, and the

SRT in days, are provided. In a study of the effect of particle size on the digestion of a mixture of fine

paper, cardboard, newsprint, magazines and corrugated cardboard, Pommier et al. (2010) reported a

Table 10.1 Comparison of Daisy's performance with published data

1 Nigel Guilford SRT 42d Seq.-fed LB reactor with UASB

Substrate: FB plus FW

CODFW% Methane yield Destr. Eff. mLCH4.g-1VSadded % as VS

0 66.8 20.0

7.9 134.8 33.2

12.9 184.4 43.2

17.2 218.7 56.3

21.7 246.0 63.5

29.3 296.3 69.4

2 Pommier et al. SRT 90d 6g BMP tests

Substrate: CB, BB, NP, FP + mag.

Methane yield Destr. Eff.

bmLCH4.g-1VSadded % as VS

149.6 43.6

3 Yuan et al. SRT 60d 1L BMP tests

Substrate: CB + FP + NP w and w/out microbial pretreatment

Methane yield Destr. Eff.

mLCH4.g-1VSadded % as VS

W/out pretreatment 92.9 N/A

With pretreatment 209.0 N/A

4 Eleazer et al. SRT 600d 2L BMPs w leachate recirc.

Methane yield aDestr. Eff.

Substrate mLCH4.g-1VSadded % as VS

FW 320.6 77.4

NP 75.4 31.1

CB 155.0 54.4

FP 288.0 54.6

MSW 122.3 58.4

5 Zhang et al. SRT 30d 35L CSTRs fed semi-cont.

Substrate Methane yield Destr. Eff.

mLCH4.g-1VSadded % as VS

mr-OFMSW 304.0 62.0

ss - FW 425.0 83.9

ameasured as cellulose+hemicellulose destruction. bconverted from mLCH4.g-1DMadded

methane yield of 149.6 mLCH4.g-1VSadded and a VS destruction efficiency of 43.6%. They also found that

the effect of particle size on both yield and destruction efficiency was negligible between 100 mm x 100

mm and <1 mm. This unexpected result may be a consequence of the pulping process used to create

161

these products all of which, to varying degrees, are permeable. They also reported an inverse linear

relationship between lignin content and biodegradability – FP>CB>NP. The work of Yuan et al.(2012)

compared the digestion of mixed fibres (CB, FP and NP) with and without microbial pretreatment. The

purpose of the pretreatment was to enhance digestibility; a 125% increase in methane yield from 92.9

to 209.0 mLCH4.g-1VSadded was reported. Eleazer et al. (1997), in a simulation of anaerobic digestion in a

landfill of individual fibres (hence the 600 d SRT) obtained results similar to those of BMP 2 presented

in this thesis, except in the case of FP which appears to show a destruction efficiency which is too low

for the amount of methane produced. Daisy’s performance matched or surpassed that of all the

comparable work cited, except for Eleazer et al.’s (1997) result for food waste alone. In broad terms,

Daisy’s performance, processing coarser material, matched that of Zhang et al.’s (2012a) CSTR using

only 25% of the volumetric reactor capacity. Although Daisy’s SRT was 42 days, Figure 8.14 shows that

97.6% of ultimate substrate destruction had been achieved within 28 days.

In contrast, Zhang et al’s (2012a) results for ss-FW (source separated food waste) measured shortly

before reactor failure, show a methane yield of 425 mLCH4.g-1VSadded and destruction efficiency of

83.9% (compared to 304 mLCH4.g-1VSadded and 62% destruction efficiency for mr-OFMSW). This is partly

because FW is more digestible, but partly because it has a higher COD content per gVS than does

lignocellulosic fibres. Does this mean that Daisy could be operated with an SRT of 30 days instead of 42

days as suggested by Figure 8.14, the graph of substrate destruction versus time, and if so how could

this be accomplished? The data would suggest that SRT could be reduced but the supporting evidence

is limited (a single measurement obtained at CODFW29%). It also seems likely, based on Daisy’s

performance, that the curve shown in Figure 8.14 would look quite different at lower food waste

addition rates. Section 8.4 makes the argument that leach beds nearing the end of their digestion cycle

are still operating as methanogenic reactors and thereby contributing to process stability, thus

shortening the SRT could diminish or eliminate this benefit. Lastly, the possibility of eliminating the

UASB, also discussed in Section 8.4, must be factored in. A systematic study to address these

interconnected questions is proposed as part of the recommendations for future work in Section

10.3.2.

10.1.4 Properties of Bulking Agent

The addition of bulking agent to the waste provides structure and ensures continued drainage as the

waste digests. The physical characteristics of the bulking agent were not given much serious thought

until a decline in gas production and waste digestibility were observed during Period 5. At the time, it

appeared as though the manner in which the wood had been ground was at least partially responsible.

162

The experiments conducted to investigate this behaviour suggested that bulking agent properties were

the cause, but did not conclusively prove it; in the absence of any other explanation, I still believe it to

be true. Also, the amount of bulking agent used was based on the work of others (Murto et al., 2013)

and was not optimized for Daisy.

10.1.5 Meeting the Objectives

Five objectives were set out in Section 2.2:

1. Design, build, and commission a laboratory scale version of the Biopower process.

2. Test and troubleshoot the initial operation to confirm operability and create a baseline

for subsequent experiments.

3. Develop analytical methodology and a sampling plan compatible with the operation of

the digester that will enable mass and energy balance calculations and performance

monitoring and benchmarking to other systems.

4. Operate the reactor under a variety of test conditions to measure efficiency in terms of

biogas yield, COD destruction and solids stabilization, as well as robustness and

stability, starting with examining the effects of changes in substrate composition.

5. Use the performance data obtained to refine economic predictions for a full scale

system.

All of these objectives were met.

10.1.6 Testing the Hypothesis

The hypothesis comprises three sentences and is tested against the findings of this research

programme sentence by sentence. I have split the hypothesis accordingly and restated it below for

convenience:

1. An anaerobic digester, comprising multiple sequentially-fed leach beds feeding a single UASB is

capable of successfully processing mixed organic wastes, consisting of varying proportions of

cellulosic wastes (paper, cardboard, boxboard) and food waste, to produce commercial

quantities of biogas and a stabilized digestate.

Daisy was designed and built to conform to the concept of the patented BioPower process, with six

sequentially-fed leach beds and a single UASB. The capability of this design for processing a mixture of

cellulosic wastes and food waste was clearly demonstrated. The quantity of food waste added,

measured as a percentage of total COD, varied from 0 to 29%. At a solids retention time of 42 days

(which was maintained throughout the experiment with one minor exception), substrate digestibility

163

and biogas production varied considerably as a result of changes in food waste addition. It is doubtful

that commercial quantities of biogas could be generated under these conditions without the synergistic

effect contributed by the food waste. The upper limit of food waste addition, evidenced by the onset

of digester instability, was never reached. Given that food waste addition is shown to be so beneficial

to performance, and thus to the commercial prospects of the technology, the upper boundary of this

component of the hypothesis should be determined. Daisy’s performance matched that of a CSTR

digesting a similar substrate, and did so in a similar solids retention time. Section 10.3 compares

economic performance in 2009 with 2017. Despite a number of significant changes since 2009 (e.g.

cancellation of Ontario’s feed-in-tariff programme for electricity which was worth 14.7ȼ/kWh), the

financial returns (assisted by carbon credits at $5/GJ and robust tipping fees of $100/t) remain very

strong – a ten-year IRR of 41% and a year-10 EBIT of 54%. As was the case in 2009, an annual capacity

of 50 – 60,000 tonnes/y of organic waste is needed to achieve these results, and financial performance

is quite sensitive to capacity (Guilford, 2009).

2. This can be accomplished by managing the leach beds independently of one another, balancing

leachate flow between recirculation to the leach beds and delivery to the UASB, allowing

adequate solids retention time to achieve the maximum practical conversion of biomass to

biogas, and by aerobically curing the digestate.

The leach beds were managed independently of one another, the flow was balanced between

recirculation and delivery to the UASB and this worked successfully, but it would not be accurate to say

that this relationship was optimized. Varying flow to the leach beds requires that their temperature be

thermostatically controlled and, although leach bed heaters were installed in September 2015 with a

resultant increased biogas production, their control was manual. The necessity for a UASB was also not

demonstrated. On the one hand the UASB converts a lot of soluble COD immediately after fresh waste

(especially FW) is added, but on the other hand, when it was partially plugged and malfunctioning from

Week 55 to Week 60, there were no obvious ill-effects on overall performance. On the basis of the

single test near the end of the experimental programme (Figure 8.14) the solids retention time was

shown to be more than adequate at high levels of food waste addition. The aerobic curing step worked

well.

3. Digestion rate and yield can be enhanced by managing the proportion of food waste to fibres;

the production of waste water (except for water-of-saturation in the digestate) can be

minimized or even eliminated; process stability can be maintained, despite feedstock variability,

by controlling two principal variables; C:N ratio and alkalinity.

164

The discovery of synergy, and determination of its magnitude, demonstrated beyond all reasonable

doubt that adding food waste enhanced the digestion of fibres, and did so without upsetting Daisy’s

stability. The complete absence of wastewater (other than as water of saturation in the digestate) was

a pleasant surprise; managing effluent from anaerobic digesters is often a significant expense. All the

water of saturation in the digestate can be expected to evaporate in the aerobic curing step. Near the

end of the experiment, water addition to the feedstock was reduced in order to prevent leachate levels

rising too high in the tanks. From start to finish Daisy remained stable, the C:N ratio was always above

40 and the pH and alkalinity well within acceptable ranges, though it cannot be said that either one had

to be actively controlled in order to maintain stability. In fact, Daisy’s six leach beds meant that no more

than 16% of the waste could be changed at any one time, so it is hard to create conditions of instability

in the entire system: the design is inherently well-buffered and stable. There were signs towards the

end of the experiment that the concentration of inorganic salts within Daisy, particularly sulphate, was

starting to rise. This was attributable to increasing food waste addition and could ultimately require

that food waste addition be reduced. In summary, with some caveats, the hypothesis was found to be

well-supported.

10.2 Commercial Implications

In 2009 my M.Eng. thesis included a detailed financial analysis of the commercial prospects of the

BioPower technology. Much has changed since then; some of the changes have been favourable but

others not. In 2009 the model was based upon a generic design for a 60,000 tonne per year plant. In

2017, the model is based upon a specific project to be built in Ottawa with construction starting in

about two years, and is thus fundamentally different in certain respects. Table 10.2 summarizes the

major assumptions then and now. Plant capacity has been reduced by 10,000 tonnes per year but the

tipping fee has increased to $100 per tonne; this is the currently-prevailing market rate in the Ottawa

area. The amount of biogas recovered per tonne of

incoming waste has increased, and the SRTanaerobic has

been shortened to one year from three years and the

SRTaerobic from 18 months to nine months, all as a

direct consequence of Daisy’s results. These SRTs

differ markedly from those used in Daisy and this

requires some explanation. Compared to Daisy, a

commercial plant will be disadvantaged in three ways; firstly, the feedstock, as placed in the digester,

will be more heterogeneous and the distribution of leachate, and its penetration into the waste mass

Table 10.2 Financial Model - Major Assumptions

2009 2017

Plant capacity 60,000 50,000 tonnes/year

Tipping fee 70 100 $/tonne

Biogas recovered 150 250 m3/tonne

Power purchase 14.7 0 ¢/kWh

Natural gas sales 0 8 a$/tonne

Residue (10% incoming t) 70 60 $/tonne

Carbon credits 0 5 $/GJ

165

more uneven and slower as a result. Secondly, the digesting waste mass will partially collapse under its

own weight potentially affecting digestion efficiency. Thirdly, the regulators will insist that the waste be

stabilized and, as far as possible odour-

free, before it is exhumed. With time and experience,

shorter SRTs are probably achievable, so the design

must be conservative and yet accommodate this

possibility. The SRTs chosen come directly from the work

of Yazdani (2012) on a landfill-based anaerobic/aerobic

digester processing yard waste only and are thus very

conservative. Revenue from the sale of power has been eliminated because of the cancellation of

Ontario’s feed-in tariff (FIT) programme for renewable electricity from biogas (14.7ȼ/kWh). Carbon

credits have been added at $5/GJ.

The capital cost estimate in Table 10.3 shows an increase of about 11% but this is less of an effect of

inflation and more a reflection of changes in scope. The land purchase is a pro rata allocation of the

total cost of the lands acquired for the entire Ottawa project. The environmental approvals cost is

shown as zero because the total is not known, and its allocation to the various components of the

entire project has not been made. Costs have been reallocated between the treatment unit and site

preparation; this is largely an accounting exercise. The electrical plant has of course been eliminated.

Table 10.4 compares projected financial performance in 2009 and

2017. The underlying assumptions and a summary income

statement can be found in my M.Eng. thesis for 2009 (Guilford,

2009), and in Appendix G for 2017. The major differences are as

follows. The tipping fee revenues have increased, the net effect of

lower tonnages at a higher price per tonne. Energy sales are

much lower because of the cancellation of Ontario’s feed-in-tariff

programme which was worth 14.7ȼ/kWh . An allowance for

carbon credits has been added at $5/GJ. The compost revenue

has also been eliminated; the value of the compost is very

dependent upon the cleanliness of the incoming waste; this is less

certain for commercial waste. The EBITDA and EBIT numbers

given are for year 10, and the NPV and IRR are both calculated

over 10 years. Despite all the changes, the financial performance is still very strong. Apart from the

inevitable uncertainties associated with scale up, the financial performance depends most heavily on

Table 10.3 Capital Cost Summary

2009 2017

Land purchase $430,000 $1,670,000

Environmental Approvals $250,000 $0

Site preparation $3,243,000 $121,000

Treatment Unit $1,881,000 $7,741,000

Equipment $4,075,000 $5,386,000

Electrical plant $3,500,000 $0

Total $13,379,000 $14,915,000

Table 10.4 Year 10 Financial projections

2009 2017

Tipping fee 5,019,000 5,975,000

Energy sales 2,371,000 572,000

Carbon credits 0 1,430,000

Compost sale 752,000 0

Total Revenue 8,142,000 7,977,000

O&M costs 3,041,000 2,985,000

EBITDA 5,101,000 4,923,000

Deprec. 1,298,000 1,773,000

EBIT 3,803,000 3,220,000

NPV 6,053,000 6,534,000

IRR 43% 41%

166

tipping fees. However, if these were to drop to $50 per tonne the project would still be breakeven.

Under the terms of the environmental assessment approval granted for this project, the anaerobic

digestion facility will first be built at a demonstration scale of up to 20,000 tonnes per year.

10.3 Conclusions and Recommendations

10.3.1 Conclusions

Six principal conclusions can be drawn from the research. In order of significance, they relate to

synergy, digester performance, digester stability, reactor design, commercial prospects and greenhouse

gas emissions.

1. Synergy

The addition of food waste enhances the digestion of a mixture of lignocellulosic fibres to improve both

digestion rate and methane yield. This synergistic effect is proportional to the amount of food waste

added, expressed as percent of total COD, in a relationship that is almost linear to CODFW29%. The

biogas yield from the mixture of lignocellulosic fibres increased from 101 L.kg-1CODFBadded at CODFW0% to

330 L.kg-1CODFBadded at CODFW29% (at STP), an increase of 225%. This figure might well increase further

at higher food waste addition. The available evidence suggests that the mechanism is in part enzymatic,

but this is not proven. Substrate conversion and the recovery of renewable energy from lignocellulosic

fibre wastes can be greatly enhanced by the addition of food waste.

Synergy, as measured in Daisy, the confirmatory BMP experiments, the enzyme assays experiments and

Daisy’s operating data, will form the basis of a paper. This thesis will be the principle supporting

document.

2. Digester Performance

A solid state, two-stage anaerobic digester, comprising six sequentially fed leach beds and a UASB, is

able to achieve substrate destruction efficiencies and methane yields equivalent to those obtainable in

a CSTR at a similar solids retention time, digesting similar waste, and do so without extensive feedstock

pretreatment.

3. Digester Stability

Digester stability, measured in terms of alkalinity, COD concentration and VFA concentration, was

essentially unaffected by six changes in food waste addition, some large and some small, between

167

CODFW0% and CODFW29%, throughout an experiment that lasted 88 weeks. Periodic equipment

malfunctions and partial process shutdowns had no lasting effect on digester stability.

4. Reactor Design

Throughout the experiment, a near-perfect hydraulic balance was maintained; water added with the

fresh waste was equal to water removed with digestate, and no free wastewater was generated.

Aerobic curing of the digestate will either consume, or cause the evaporation of, the water it contains.

Frequent leachate recirculation minimized the build-up of COD and VFAs and prevented inhibition. The

UASB effectively converted the high concentrations of COD that resulted from the addition of fresh

waste, especially food waste. Spent leach beds, no longer contributing much fresh COD to the system,

continued to serve as methanogenic reactors producing biogas.

5. Commercial Prospects

A financial analysis of a 50,000 tonne per year AD plant employing this technology demonstrated that it

is an attractive commercial proposition capable of competing with landfill disposal. Under current

market conditions tipping fees remain the principle source of revenue, and carbon credits, even at

$5/GJ, are likely to contribute more than energy sales.

6. Greenhouse Gas Emissions

Based on these results, and in particular the effect of food waste on the digestion of lignocellulosic

fibres, it is possible that municipal programmes designed to divert organic wastes away from landfills

may be lengthening the contaminating lifespans of the landfills. The organic waste which still goes to

landfill after diversion (and there is a lot of it) may actually take much longer to decompose in the

relative absence of food waste. This could have significant implications for the costs of long-term care

and the efficient recovery of landfill gas. It is also possible that a greater percentage of the organic

carbon in the landfill might be sequestered than was previously the case.

10.3.2 Recommendations for Further Research

Future research should be conducted in five distinctly separate directions; additional work with Daisy,

research associated with a forthcoming commercial deployment of the technology, generating products

of higher value, exploring the applicability of the technology in different socio-economic settings and

investigating the environmental implications of organic waste diversion from landfill.

Daisy a) Solids retention time, UASB and digester stability

168

The objective would be to determine if Daisy can operate with a shorter SRT, with and without a UASB,

and maintain performance while remaining stable. A three-step approach is proposed.

Measure substrate destruction efficiency versus time over a range of food waste addition

rates without shutting Daisy down. This would require development of a modified coupon

test.

Shut down the UASB and measure performance of Daisy, particularly the conversion of high

concentrations of COD immediately after feeding, in the absence of the UASB.

Shorten the SRT in stages, either by operating with only five leach beds then four leach

beds, or by continuing to operate with six leach beds but replacing them at five week and

then four week intervals.

b) Optimize leachate recirculation.

There are several approaches. For example, it might be advantageous to maintain a higher recirculation

rate with a leach bed of fresh waste and reduce the rate to more mature leach beds, thus saving

operating costs and perhaps enhancing performance. This would require a refinement of the

techniques used to measure biogas production from a single leach bed, and the installation of

thermostatically controlled heaters on the leach beds, to allow variations in leachate flow rate without

causing unwanted temperature fluctuations. The design is simple, the cost of materials is about $2,000,

and the work could be completed in about two weeks without the need for outside help.

c) Mechanism of Synergy

There is considerable scope and need to further examine the mechanisms behind synergy, for example

the role of enzymes in the digestion fibres; what are they, where do they come from, and how do they

function? What does the separate investigation of the microbial population have to say about synergy?

In conjunction with this, an investigation of the effect of interim storage of FW (from kitchen to curb to

digester) might also prove interesting.

d) Additional analyses

In retrospect, there are several additional analyses that I wish I had done, or done more of, and which I

recommend future work include:

Substrate analysis for cellulose, hemicellulose and lignin

Ammonia in the leachate

Biogas from a single leach bed – more extensively

Refine the coupon technique to improve accuracy and perform CODout measurements

169

Determine the reliability of sulphate measurements

Commercialization A demonstration scale version of Daisy is to be built in Ottawa once all the remaining approvals are

secured. This will entail a considerable amount of design followed by applied research in the field to

measure performance under very different conditions from those in the lab, and solve the technical

problems that will undoubtedly arise from scale up. I believe that the Department of Chemical

Engineering and Applied Chemistry has a significant role to play in this work.

Generating products of higher value The economics of digesting organic solid waste are supported by the generation of renewable energy in

the form of methane, but the energy revenues are modest. There is the possibility of generating

products of greater value; isoprene is an interesting example. Isoprene is an industrial chemical used in

the production of synthetic rubber and made predominantly from petroleum. Because it has a boiling

point of 34.7°C, simplified product recovery in the gaseous phase is possible.

In 2014, Professors Sasha Yakunin, Elizabeth Edwards and Grant Allen submitted an unsolicited

proposal to the Environmental Research and Education Foundation (EREF) for a 3-year study to

investigate the possibility of generating isoprene from solid waste. The proposal, which was not

approved for funding, had three components; the introduction of cellulose-degrading enzymes into E.

coli; the engineering of the mevalonate pathway for isoprene synthesis and its introduction into the

cellulose-degrading E. coli strains, and the optimization of isoprene production from MSW by the

engineered E. coli strains in flask experiments and anaerobic bioreactors. In view of the results obtained

from Daisy this proposal should be resurrected to see if it is worth pursuing further.

Applicability in different socio-economic settings As described in detail in Chapter 1, the basic objective of this undertaking was to devise a simple robust

technology to anaerobically digest large quantities of organic waste at a price that was affordable

within the economic framework of waste management in North America. To do this it had to be

relatively simple with as few process steps as practicable. With solid waste processing, the primary

objective is always to treat the waste to reduce its volume, and its environmental impact.

The recovery of value in the form of usable materials, or in this case energy, is usually treated as a

potential offset for operating costs rather than as the principal goal. At a commercial scale in the

current North American market, a minimum capacity of 50,000 tonnes a year of organic waste

(corresponding to a population of about 400,000) would be needed to ensure a financial return. This

same plant could generate about 1.7 MW of electricity net of parasitic load, enough to supply the

170

needs of about 1,500 homes, say about 6,000 people. Thus the waste of about 65 people can supply

the electrical needs of only one.

The question thus arises, is this technology, either in its current form or modified, applicable in other

socioeconomic settings? For example, could it be deployed to generate energy for remote communities

who otherwise rely on diesel powered generators? Or could it be deployed in developing countries to

dispose of organic waste and recover some usable energy? The answer to the first question would

appear to be a resounding ‘no’, unless there were a large, and continuing, demand for the anaerobic

digestion of an industrially-generated organic waste stream. The answer to the second question could

be ‘yes’, in principle. In densely populated areas it might be possible to find a large enough quantity of

waste that doesn’t have to be transported very far. Three things would be essential; confirmation of

the availability of suitable waste in sufficient quantities (for example, food waste might be plentiful but

lignocellulosic wastes in short supply thus altering the technical requirements) financial subsidies, and

education. The possibility of miniaturization for local use on a smaller scale is also worth investigating,

but the requirement for back-up power may be its Achilles’ heel.

Environmental Impact of Food waste Diversion The long-term implications of the diversion of large quantities of food waste from landfills should be

investigated.

Does it slow down the decomposition of the (considerable) amount of organic waste still being

landfilled?

Does it lengthen the contaminating lifespan of the site?

What are the impacts on the scope and costs of long-term care?

Are larger quantities of carbon are being sequestered as a result of food waste diversion?