UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The viability of lignocellulosic ethanol production as a business endeavour in Canada Stephen, James Duncan 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2013_spring_stephen_james.pdf [ 3.29MB ]
Metadata
JSON: 24-1.0073518.json
JSON-LD: 24-1.0073518-ld.json
RDF/XML (Pretty): 24-1.0073518-rdf.xml
RDF/JSON: 24-1.0073518-rdf.json
Turtle: 24-1.0073518-turtle.txt
N-Triples: 24-1.0073518-rdf-ntriples.txt
Original Record: 24-1.0073518-source.json
Full Text
24-1.0073518-fulltext.txt
Citation
24-1.0073518.ris

Full Text

THE VIABILITY OF LIGNOCELLULOSIC ETHANOL PRODUCTION AS A BUSINESS ENDEAVOUR IN CANADA by JAMES DUNCAN STEPHEN B.Sc., Queen’s University, 2003 M.Sc., The University of British Columbia, 2008  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Forestry) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) January 2013  © James Duncan Stephen, 2013  ABSTRACT The competitiveness and operational viability of a wood-based lignocellulosic ethanol biorefinery located in Canada were assessed using techno-economic and logistical models, combined with fuel and feedstock market data analyses. Scenario analyses included the ability to compete for forest feedstocks with alternative bioenergy options; for market share with ethanol producers using alternative feedstocks (corn, sugarcane, Brazilian eucalyptus); and for financing, given the market risks and alternative investor options. The model variables used to identify the most competitive lignocellulosic ethanol production conditions included feedstock type and properties, feedstock logistics system design, facility site, facility scale, capital cost, pretreatment technology, operating tactics, co-products, ethanol and co-product revenues, enzymes and other process inputs, ethanol and co-product yields, and taxes and renewable energy support policies. It was determined that a wood-based lignocellulosic ethanol facility in Canada is technically viable, but will find sustainable profitability difficult at current energy prices and without a change in the historical volatility of those prices. Using scenario analyses, it was determined that the minimum ethanol selling price (MESP) of lignocellulosic ethanol produced from Canadian forest biomass is $0.80-1.10 L-1, with most scenarios in the $0.90-0.95 L-1 range. This compares with recent corn and sugarcane ethanol MESP of $0.30-0.40 L-1, highlighting the difficulty of competing with these conventional biofuels. In addition, other bioenergy products, such as wood pellets and combined heat and power, will compete with lignocellulosic ethanol facilities for feedstock but offer more stable markets. Largely due to the lack of correlation between transportation fuel markets and forest feedstock costs, the gross processing margin of lignocellulosic ethanol production was shown to be decidedly volatile. This volatility results in an anticipated cost of capital (>11%) that exceeds other fuel production facilities. Although supplying large (e.g., 800 ML yr-1) lignocellulosic ethanol facilities that maximize economiesof-scale (and minimize cost per unit) with feedstock is logistically possible, high feedstock costs, exceeding $100 bdt-1, are projected to put Canadian producers at a disadvantage relative to tropical country producers. Significant production cost reductions must occur before lignocellulosic ethanol can compete for market share with gasoline and conventional ethanol.  ii  PREFACE Two papers, based upon Chapters 3 and 4 of this work, have been published to date. The primary author for the two manuscripts was James Stephen and he was the primary designer of the methodology and research plans as described in the manuscripts and in this thesis. Guidance and oversight were provided by Dr. Jack N. Saddler and Dr. Warren E. Mabee. All modelling and analysis research work was carried out solely by James Stephen, with recommendations and explanation on some model inputs provided by Dr. Saddler and Dr. Mabee. The publications arising from this thesis are: Stephen JD, Mabee WE, Saddler JN, 2010. Biomass logistics as a determinant of 2nd generation biofuel facility scale, location and technology selection. Biofuels, Bioproducts and Biorefining 4: 503-518. (Thesis Chapter 4) Stephen JD, Mabee WE, Saddler JN, 2012. Will second generation ethanol be able to compete with first generation ethanol? Opportunities for cost reduction. Biofuels, Bioproducts and Biorefining 6: 159-176. (Thesis Chapter 3)  iii  TABLE OF CONTENTS ABSTRACT.................................................................................................................................................. ii PREFACE .................................................................................................................................................... iii TABLE OF CONTENTS ............................................................................................................................. iv LIST OF TABLES ..................................................................................................................................... viii LIST OF FIGURES ..................................................................................................................................... ix LIST OF FORMULAS ................................................................................................................................. x LIST OF ABBREVIATIONS ...................................................................................................................... xi ACKNOWLEDGEMENTS ....................................................................................................................... xiii DEDICATION ........................................................................................................................................... xiv 1  INTRODUCTION ................................................................................................................................ 1 1.1  Oil-based Transportation Fuels ..................................................................................................... 1  1.1.1  Overview ............................................................................................................................... 1  1.1.2  The Economic Challenge of Oil ............................................................................................ 3  1.1.3  The Social Challenge of Oil .................................................................................................. 5  1.1.4  The Environmental Challenge of Oil .................................................................................... 8  1.1.5  Looking for Alternatives...................................................................................................... 10  1.2  Bio-based Transportation Fuels .................................................................................................. 10  1.2.1  Overview of Conventional Bio-based Transportation Fuels............................................... 12  1.2.1.1  Corn Ethanol Overview .................................................................................................. 12  1.2.1.2  Corn Ethanol Technology: Dry Mill & Wet Mill ........................................................... 15  1.2.1.3  Sugarcane Ethanol Overview .......................................................................................... 16  1.2.1.4  Sugarcane Ethanol Technology ...................................................................................... 17  1.2.2  Overview of Advanced Bio-based Transportation Fuels .................................................... 18  1.2.2.1  Lignocellulosic Biomass within the Energy System ....................................................... 19  1.2.2.2  Lignocellulosic Feedstocks ............................................................................................. 20  1.2.2.3  Advanced Lignocellulosic Technologies: Biochemical Conversion .............................. 23  1.2.2.4  Advanced Lignocellulosic Technologies: Thermochemical Conversion........................ 25  1.2.3  The Economic Performance of Bio-based Transportation Fuels........................................ 25  1.2.3.1  Feedstock Economics Overview ..................................................................................... 26  1.2.3.2  Feedstock Supply Risk .................................................................................................... 27  1.2.3.3  Integration of Biofuels within the Forest Sector ............................................................. 30  1.2.3.4  Forest Feedstock Logistics .............................................................................................. 32 iv  1.2.3.5  Financial Markets and Processing Margin Risk Management ........................................ 35  1.2.3.6  Techno-Economic Performance of Conventional Biofuels ............................................ 38  1.2.3.7  Techno-Economic Performance of Advanced Lignocellulosic Biofuels ........................ 40  1.2.4  The Social Performance and Policy Options for Bio-based Transportation Fuels ............ 44  1.2.4.1  Energy Security ............................................................................................................... 45  1.2.4.2  Competition with Food ................................................................................................... 48  1.2.4.3  Justification for Government Support ............................................................................. 50  1.2.4.4  Policy Support: Tax Incentives ....................................................................................... 52  1.2.4.5  Policy Support: Blending Mandates................................................................................ 54  1.2.4.6  Policy Support: Import Tariffs ........................................................................................ 55  1.2.5  The Environmental Performance of Bio-based Transportation Fuels ................................ 57  1.3  Thesis Objective and Themes ..................................................................................................... 58  1.4  Research Approach ..................................................................................................................... 64  2 ECONOMIC PERFORMANCE AND RISK OF LIGNOCELLULOSIC ETHANOL RELATIVE TO OTHER WOOD-BASED BIOENERGY OPTIONS ........................................................................... 72 2.1  Introduction ................................................................................................................................. 72  2.2  Biomass Conversion and Competition........................................................................................ 72  2.2.1  Biopower and CHP ............................................................................................................. 72  2.2.2  Wood Pellets ....................................................................................................................... 74  2.2.3  Transportation Biofuels ...................................................................................................... 74  2.3  Relative Value ............................................................................................................................. 76  2.4  Methodology ............................................................................................................................... 77  2.4.1  Facility Review.................................................................................................................... 81  2.4.2  Base Case ............................................................................................................................ 81  2.5  Results ......................................................................................................................................... 82  2.5.1  Scale .................................................................................................................................... 83  2.5.2  Feedstock ............................................................................................................................ 86  2.5.3  Revenue ............................................................................................................................... 87  2.6  Discussion ................................................................................................................................... 89  2.7  Conclusion .................................................................................................................................. 92  3 COMPETITION WITH CONVENTIONAL ETHANOL AND OPPORTUNITIES FOR COST REDUCTION ............................................................................................................................................. 94 3.1  Introduction ................................................................................................................................. 94  3.2  Methods....................................................................................................................................... 95 v  3.3  Results ......................................................................................................................................... 97  3.3.1  Progress Ratio .................................................................................................................... 98  3.3.2  Cost Reduction: Feedstock Selection .................................................................................. 99  3.3.3  Cost Reduction: Enzymes .................................................................................................. 102  3.3.4  Cost Reduction: Capital costs ........................................................................................... 105  3.3.5  Cost Reduction: Co-product Credits................................................................................. 112  3.3.6  Cost Reduction: Energy and Administration .................................................................... 114  3.4  Discussion ................................................................................................................................. 114  3.5  Conclusion ................................................................................................................................ 118  4 THE INFLUENCE OF FEEDSTOCK LOGISTICS ON MAXIMUM FACILITY SCALE, LOCATION AND TECHNOLOGY SELECTION.................................................................................. 119 4.1  Introduction ............................................................................................................................... 119  4.1.1 4.2  Study Design and Assumptions ................................................................................................ 125  4.3  Results ....................................................................................................................................... 126  4.3.1  Bulk Density and Transport Modes................................................................................... 128  4.3.2  Estimating Maximum Deliveries ....................................................................................... 131  4.4  Discussion ................................................................................................................................. 132  4.4.1  Siting and Feedstock Supply ............................................................................................. 132  4.4.2  Transportation Mode Limitations ..................................................................................... 134  4.5 5  Logistics and Scaling ........................................................................................................ 122  Conclusion ................................................................................................................................ 136  COMPETITIVE ADVANTAGES IN SITING FOR CANADIAN FACILITIES ........................... 139 5.1  Introduction ............................................................................................................................... 139  5.2  Study Design and Assumptions ................................................................................................ 141  5.3  Results ....................................................................................................................................... 146  5.3.1  Feedstock Characteristics and Delivered Cost ................................................................. 146  5.3.2  Cost of Ethanol Delivery ................................................................................................... 150  5.3.3  Labour ............................................................................................................................... 151  5.3.4  Cost of Capital .................................................................................................................. 151  5.3.5  Cost of Construction ......................................................................................................... 153  5.3.6  Electricity Production and Revenues ................................................................................ 153  5.3.7  Taxes, Insurance, and Permits .......................................................................................... 154  5.3.8  Sensitivity to Energy Rates ................................................................................................ 156  vi  6  5.4  Discussion ................................................................................................................................. 157  5.5  Conclusion ................................................................................................................................ 160  FINANCING EXPECTATIONS AND THE IMPACT ON FEEDSTOCK COST ......................... 162 6.1  Introduction ............................................................................................................................... 162  6.2  Study Design and Assumptions ................................................................................................ 166  6.3  Results ....................................................................................................................................... 168  6.3.1  MESP Base Case and Sensitivity ...................................................................................... 168  6.3.2  Diesel Consumption .......................................................................................................... 170  6.3.3  Fuel Price and Margin...................................................................................................... 170  6.3.4  Expected Return ................................................................................................................ 173  6.3.5  Impact on Maximum Feedstock Cost ................................................................................ 173  6.4  Discussion ................................................................................................................................. 174  6.5  Conclusion ................................................................................................................................ 177  7  GENERAL DISCUSSION ............................................................................................................... 179  8  GENERAL CONCLUSIONS ........................................................................................................... 192  REFERENCES ......................................................................................................................................... 195  vii  LIST OF TABLES Table 1.1 Estimates on global bioenergy potential ..................................................................................... 20 Table 1.2 Literature examples of estimated delivered cost of forest biomass ............................................ 34 Table 1.3 Estimates on net cost of production of conventional ethanol ..................................................... 40 Table 1.4 Estimates on net cost of production of lignocellulosic biofuels ................................................. 40 Table 2.1 Assumptions for base case 200,000 bdt yr-1 biomass processing facility models....................... 80 Table 2.2 Financial results for base case 200,000 bdt yr-1 biomass facilities ............................................. 83 Table 2.3 Feedstock and select conversion assumptions for feedstock sensitivity ..................................... 86 Table 2.4 Volatility of real prices for select fuels (2007 US$) ................................................................... 88 Table 3.1 Assumptions for base case advanced lignocellulosic ethanol production cost ........................... 96 Table 3.2 Holocellulose content and theoretical ethanol yield of various lignocellulosic feedstocks ...... 100 Table 3.3 Properties and ethanol yield of feedstocks from the forest system ........................................... 101 Table 3.4 Impact of hydrolysis residency time on biomass feedstock-product margin ............................ 110 Table 3.5 Comparison of selected lignocellulose pretreatments for ethanol production .......................... 111 Table 3.6 Revenue comparison of possible biomass-to-ethanol co-products ........................................... 113 Table 4.1 Optimal and near-optimal facility scales for advanced lignocellulosic biofuel production...... 124 Table 4.2 Largest-of-type processing facility comparison on feedstock energy input ............................. 127 Table 4.3 Estimated conversion efficiencies: intermediates-to-ethanol, intermediates-to-FT liquids ...... 128 Table 4.4 Truck weight limitations for selected jurisdictions ................................................................... 129 Table 4.5 Potential maximum biofuel facility capacity ............................................................................ 132 Table 5.1 Overview of study scenarios ..................................................................................................... 142 Table 5.2 Base case techno-economic facility and delivery model assumptions...................................... 144 Table 5.3 Feedstock content and theoretical and assumed ethanol yields ................................................ 146 Table 5.4 Feedstock harvest assumptions ................................................................................................. 148 Table 5.5 Feedstock delivery parameters.................................................................................................. 149 Table 5.6 Non-feedstock site-specific variables impacting MESP ........................................................... 155 Table 6.1 MESP and hydrolysis spread sensitivity analysis ..................................................................... 169 Table 6.2 Diesel fuel consumption and contribution to MESP................................................................. 170 Table 7.1 Research opportunities to reduce minimum ethanol selling price ............................................ 191  viii  LIST OF FIGURES Figure 1.1 World oil consumption, 1980-2010 ............................................................................................. 2 Figure 1.2 U.S. total refiner acquisition cost of crude oil, 1968-2011 (2011 US$) ...................................... 5 Figure 1.3 World oil production, 1980-2010 ................................................................................................ 6 Figure 1.4 OPEC crude oil exports, 1986-2009 ............................................................................................ 7 Figure 1.5 Distribution of 1.35 trillion barrels total proved reserves of oil, 2010 ........................................ 8 Figure 1.6 Sources of 29 Gt CO 2 eq of total global GHG emissions from fuel combustion, 2009 ............ 10 Figure 1.7 Cellulose within the plant cell wall ........................................................................................... 21 Figure 1.8 Historical pulp production for selected jurisdictions ................................................................ 29 Figure 2.1 Nominal U.S. pricing of primary fossil fuels ............................................................................ 76 Figure 2.2 Impact of facility scale on capital investment and internal rate of return.................................. 85 Figure 2.3 Impact of feedstock type on IRR from base case scenario ........................................................ 87 Figure 2.4 Impact of market volatility on IRR from base case scenario..................................................... 88 Figure 3.1 2007 and future ethanol production costs from U.S. corn, Brazilian sugarcane, and U.S. lignocellulose feedstocks ............................................................................................................................ 98 Figure 3.2 EISA cellulosic ethanol production mandate from 2010 to 2020.............................................. 99 Figure 3.3 Maximum biomass cost for lignocellulosic ethanol to be competitive with corn ethanol on a net feedstock cost basis ................................................................................................................................... 102 Figure 3.4 Impact of plant capacity and scaling factor on lignocellulosic ethanol production cost ......... 106 Figure 3.5 Hydrolysis profile of steam-pretreated pine under different severities ................................... 108 Figure 3.6 Tonnes of hydrolyzed cellulose for the same 100 t cellulose digester operated for 48 hours with varying pretreatment severities and hydrolysis residency times ............................................................... 109 Figure 4.1 Bulk energy density for various fuels ...................................................................................... 127 Figure 4.2 Maximum energy load capacity for U.S. Interstate trucks ...................................................... 130 Figure 5.1 MESP for all siting scenarios .................................................................................................. 155 Figure 5.2 MESP sensitivity to electricity rates and oil cost .................................................................... 156 Figure 6.1 Path to technology commercialization .................................................................................... 163 Figure 6.2 Cost components for ethanol production and delivery, with an MESP of $0.85 L-1 ............... 169 Figure 6.3 Real price comparison of gasoline and ethanol with cost of primary feedstocks .................... 171 Figure 6.4 Comparison of real crack and hydrolysis spreads over time, with base prices adjusted for U.S.Canada exchange rate and inflation .......................................................................................................... 171 Figure 6.5 Maximum feedstock cost payable by a lignocellulosic ethanol facility to meet the risk-free cost of capital, with gross processing margins (GPM) of $0.46 L-1 (321 L bdt-1) and $0.34 L-1 (385 L bdt-1) 172 Figure 6.6 Maximum feedstock cost payable by a lignocellulosic ethanol facility with a risk-adjusted cost of capital, with gross processing margins (GPM) of $0.57 L-1 (321 L bdt-1) and $0.42 L-1 (385 L bdt-1) 174  ix  LIST OF FORMULAS 1.1  Oil refining crack spread  2.1  Biomass energy content  2.2, 3.3  Facility scaling and capital cost  2.3  Pellet pricing  3.1  Progress ratio  3.2  Cellulase enzyme cost  5.1  Average transportation distance from every point of a circle to the centre  5.2  Average transportation distance from every point of a circle to the centre, accounting for tortuosity  5.3  Average transportation distance using Cartesian coordinate system, accounting for tortuosity  5.4  Maritime shipping cost  5.5, 6.1  Capital asset pricing model  6.2  Total beta  2 1 𝛼𝑐𝑠 = 𝑃𝑔 42 + 𝑃ℎ 42 − 𝑃𝑜 3 3 𝐵𝑤𝑒𝑡 = 𝐵𝑑𝑟𝑦 (1 − 1.1405𝑥) 𝐶1 𝑀1 𝑠 =� � 𝐶0 𝑀0  𝑓 𝑃 = 𝑃𝑁 � � 18.8 Q 1 = Q 0 •Nu  𝐸=  𝑃𝑝 ∙ 𝐿 ∙ 𝐶𝑐𝑒𝑙 × 1000 𝐴∙𝑌  d=  2 r 3  d =τ  2 r 3  d = τ [(x − x ) + ( y − y ) ] 2  i  j  i  j  1 2 2  𝐶𝑠 = (𝑃𝑐ℎ × 𝑡) + �𝑃𝑏𝑓 × 𝑛 × 𝑑� 𝑅𝑖 = 𝑅𝑓 + 𝛽�𝑅𝑚 − 𝑅𝑓 � + 𝑅𝑐 𝛽𝑇 =  𝜎ℎ𝑠 𝜎𝑐𝑠  x  LIST OF ABBREVIATIONS A  Enzyme activity  LCA  Life cycle assessment  AAC  Annual allowable cut  M  Scale of facility  𝛼𝑐𝑠  3  Crack spread  m  Cubic meter  ADM  Archer Daniels Midland  MESP  Minimum ethanol selling price  β  Volatility  Mha  Million hectares  bbl  Barrel  ML  Megalitre  𝐵𝑑𝑟𝑦  Energy content of bone dry biomass  Mt  Megatonnes (metric)  bdt  Bone dry tonne  MTBE  Methyl-tert-butyl ether  bpd  Barrels per day  MWe  MegaWatt (electrical)  𝐵𝑤𝑒𝑡  Energy content of wet biomass  n  Bunker fuel consumption  C  Capital cost of facility  N  Progress ratio  C5  Five carbon molecule  NEV  Net energy value  C6  Six carbon molecule  NO x  Nitrogen oxides  C cel  Cellulose content  O3  Ozone  CH 4  Methane  OECD  Organization for Economic CoOperation and Development  CO  Carbon monoxide  OPEC  Organization of Petroleum Exporting Countries  CO 2  Carbon dioxide  OSB  Oriented strand board  CO 2 eq  Carbon dioxide equivalent  P bf  Price of bunker fuel  Cs  Cost of shipping  P ch  Price of daily charter (shipping)  CTL  Cut-to-length  Price of gasoline  d  Distance  𝑃𝑔  DDGS  Dried distillers grains with solubles  PM  Particulate matter  DIN  Dissolved inorganic nitrogen  Price of normal (whitewood) pellet  E  Price of enzyme  𝑃𝑁  EIA  Energy Information Administration (United States Department of Energy)  EISA  𝑃ℎ  Price of heating oil  𝑃𝑜  Price of crude oil  Pp  Price of protein (enzyme)  Energy Independence and Security Act  PR  Progress ratio  EJ  Exajoules  Q  Cost of production  EPA  Environmental Protection Agency (United States)  r  Radius of a circle  EtOH  Ethanol  Rc  Return of country risk premium xi  EU  European Union  Rf  Return of risk free investment  f  Energy content of pellet  Ri  Return of investment  FAO  Food and Agriculture Organization (United Nations)  Rm  Return of market  FPU  Filter paper units  s  Scaling factor  FT  Fischer-Tropsch  σ cs  Standard deviation of crack spread  Gal  Gallons  σ hs  Standard deviation of hydrolysis spread  Gge  Gallons gasoline equivalent  t  Time  GHG  Greenhouse Gas  τ  Tortuosity -1  GJ  Gigajoule  t ha  Tonnes per hectare  GL  Gigalitre  t ha-yr-1  Tonnes per hectare per year  GREET  Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation  u  Number of doublings (progress ratio)  Gt  Gigatonne  U.S.  United States  H2  Hydrogen  USDA  United States Department of Agriculture  HFCS  High fructose corn syrup  VOC  Volatile organic compound  IEA  International Energy Agency  WDG  Wet distillers grains  ILUC  Indirect land-use change  WTO  World Trade Organization  Kg  Kilogram  x  Moisture content  L  Enzyme loading rate  Y  Enzyme activity  xii  ACKNOWLEDGEMENTS There are a number of people that have truly helped make this thesis possible. I want to thank Dr. Jack Saddler and Dr. Warren Mabee, my PhD supervisors, for their guidance, insight, and patience throughout the PhD program. I look forward to working with both of you in the future on a number of exciting endeavours. My committee members Paul McFarlane (Wood Science) and Shahab Sokhansanj (Chemical and Biological Engineering) provided critical assessment of my work and I strongly believe their discerning feedback has resulted in a much higher quality final product. I must thank the Forest Products Biotechnology and Bioenergy Group in the Faculty of Forestry at UBC for their intellectual and moral support, but more importantly, for their high-quality research and analyses which have been incorporated into this thesis. Gladys Tecson of UBC was also incredibly helpful in preparation of documents and coordination of all aspects to do with the examination and defense. Finally, I want to thank my family, those in Canada and in Europe, for their constant and unwavering support through this process. In particular, my mother, whose dedication to research and the pursuit of knowledge was critical to my decision to start this journey, and my partner in all things, Stéphanie. Cookie, you kept me motivated through the ups and downs and I cannot wait to embark on the next chapter of our life together.  xiii  DEDICATION To the two important ladies of my life -J  xiv  1  INTRODUCTION Liquid transportation fuels are essential to the world’s economy. They enable the movement of  goods and people, and support a global marketplace where consumers can access lowest-cost goods and where companies can maximize profits and customer base. They dramatically increase the diversity and availability of goods to consumers around the world, providing them with products that, in many cases, are not available or cannot be produced locally. Technological advances in the production and use of liquid fuels have also played a key role in the dramatic increases in productivity of both humans and agriculture, and have enabled exponential improvements in the quality of life of the developed world [1]. This thesis summarizes the challenges of oil-based fuels and provides an overview of the existing biofuels sector. The use of woody biomass for the production of ethanol for the transportation sector in Canada is discussed, and the ability of this biofuel to compete with other wood-based bioenergy options explored.  The relationship between conventional ethanol made from sugarcane and corn and the  emerging wood-based ethanol industry is considered, with particular emphasis given to the influence of feedstock handling on design of new wood-to-ethanol capacity. Finally, the competitive advantages of a Canadian wood-to-ethanol facility are examined, as are financing expectations and the potential impact on feedstock cost.  1.1 Oil-based Transportation Fuels 1.1.1  Overview  As the primary source of transportation fuels including gasoline, diesel, jet fuel, and bunker fuel (shipping), oil has become the lifeblood of the world’s economy. With the current global economic structure, it would be very difficult, if not impossible, to support the world’s seven billion people without oil. As the global population increases and people in developing nations seek to attain a standard living equal to, or greater than, those in the developed regions, oil demand has continued to steadily climb. Oil consumption increased from 63 million barrels per day (M bpd) in 1980 to 87 M bpd in 2010 [2]. Oil demand is expected to increase at a steady 1.5% per year, even accounting for improvements in efficiency  1  [3]. This rate of increase would result in a demand of 117 M bpd by 2030 (calculated from [2,3]). BP P.l.c. (formerly British Petroleum), who is more conservative with its demand forecast, projects that oil demand will reach 102 M bpd by 2030 [4]. Historical oil demand is presented in Figure 1.1 [2]. Figure 1.1 World oil consumption, 1980-2010  Source: [2] Notes: OECD = Organization for Economic Co-Operation and Development (34 countries, loosely defined as ‘developed’); OPEC = Organization of Petroleum Exporting Countries (12 major oil exporting countries including Algeria, Angola, Ecuador, Iran, Iraq, Kuwait, Libya, Nigeria, Qatar, Saudi Arabia, the United Arab Emirates, and Venezuela).  While oil and its derivatives have brought prosperity to much of the world, production and consumption are not without significant challenges. The challenges of relying upon oil as the dominant source of transportation fuels include economic, social, and environmental factors.  The economic  challenge of utilizing oil for virtually all transportation is its pricing impact on the global economy and the ability of oil price shocks to dramatically curtail demand and affect prices of other goods. The social challenge of oil is that oil is unevenly distributed across the globe and a large proportion of production comes from nations or regions that may have opposing geopolitical priorities to consumers. Military intervention can be required to secure such supplies, resulting in a decreased social acceptance of oil as the primary source of transportation fuel. Finally, the environmental cost of using oil is borne in the form of greenhouse gas (GHG) emissions, and more locally, oil spills and air pollution. Each of these challenges is discussed in more detail below. 2  1.1.2  The Economic Challenge of Oil  As the largest internationally-traded commodity by volume and value, oil is a key factor in the health of the world’s economies. As price and production margins rise, oil producers benefit from increased revenues and the opportunity to invest in additional production capacity. However, increases in oil price can negatively impact oil importing nations, as well as companies that rely upon oil and its derivatives as key inputs for their operations. This is the situation faced by many OECD countries. In the United States (U.S.), oil production peaked in 1970, and net imports doubled from 6 M bpd in 1973 to over 12 M bpd from 2004-2007 [5]. The same situation is true across much of Europe. Within the OECD, only Canada, Mexico, Norway, and Denmark are long-term net exporters of crude oil, with the U.S. becoming a net exporter for the first time in 20 years in 2011 primarily due to high prices and improved technologies for unconventional oil extraction [6,7]. Overall, the OECD imported 22-24 M bpd from 2005-2009. This is a significant drain on the trade balance of the OECD and presents a large risk to the economic wellbeing of these nations. The Hubbert model accurately predicted that U.S. oil production would peak in 1970 and recent research, utilizing an updated Hubbert model, suggests that the world’s production of conventional crude will peak in 2014 and that of the Organization of Petroleum Exporting Countries (OPEC) will peak in 2026 [8]. While the world will not ‘run out’ of oil in the physical sense, an increase in oil demand that is not met with an increase in supply will result in price increases. Proved oil reserves, defined as those that can be recovered commercially with a high degree of confidence, are estimated at 175.2 billion barrels (B bbl) in Canada and 20.7 B bbl in the U.S. [9]. Technological innovation and increased production of nonconventional oil sources, such as the oil sands and shale oil, will extend supply significantly, but these non-conventional sources are much more costly to develop and recent studies suggest the break-even oil price for a new oil sands development in Alberta, Canada is between $50 and $80 bbl-1 [10]. This compares with the estimated all-in extraction cost (US$2008) for Saudi Arabian oil of $4-6 bbl-1 [11,12]. It is worth noting that extraction cost differs significantly than the ‘breakeven’ price in countries such as Saudi Arabia, where the government relies heavily on oil revenues to fund government expenditures and 3  ensure geopolitical stability for oil production. From 2010 to 2011, Saudi Arabia’s breakeven oil price rose from $68 bbl-1 to $88 bbl-1 [13,14]. Given that oil is an internationally traded commodity and all inhabited continents have a mixture of conventional and non-conventional resources, security of energy supply would be better termed security of energy price. Energy security is one of the overall chief policy drivers in not only the U.S., but also other large energy importers such as the European Union (EU), China, and India. According to Brown and Huntington (2008), “...an improvement in energy security is achieved by reducing the vulnerability of economic activity in a country to potential disruptions in energy supply” [15]. The primary focus of energy security and reduction in vulnerability to supply disruptions and price volatility, particularly for the U.S., is oil. It is well known that oil shocks, or a rapid increase in oil price, hurt oil importing economies [16]. Data published by Delucchi and Murphy (2008) indicate that ten of the eleven recessions in the U.S. between 1947 and 2001 were preceded by an oil shock and ten of the eleven oil shocks were followed by a recession [17]. Jones et al. (2004) concluded these recessions were directly caused by a rapid increase in oil price (shock), to which the economy could not rapidly adapt, and could not have been prevented by alternative monetary policy [16]. Rubin (2009) argued that the 2008/09 financial crisis and recession was caused by an oil price spike, not by a housing bubble; the latter was only a symptom made possible by the availability of cheap credit, which in turn was made possible by the globalization (and hence low cost products) afforded by oil [1]. Shell and McKinsey and Co. both anticipate that with oil production failing to keep up with continuous increases in demand, oil price shocks will be difficult to avoid in the coming decades [3,18]. Therefore, given the economic challenges and risks of a total reliance upon oil, there has been a strong push by oil importing nations to diversify their transportation fuel supplies and insulate against spikes in oil prices. As evidenced by Figure 1.2, which describes historical real U.S. oil prices (2011 US$), oil price volatility is not a recent phenomenon.  4  Figure 1.2 U.S. total refiner acquisition cost of crude oil, 1968-2011 (2011 US$)  Source: [19]  1.1.3  The Social Challenge of Oil  It became clear after the Arab Oil Embargo of 1973-74 (first oil crisis) and the Iranian Revolution 1979-1980 (second oil crisis) that the oil-importing nations of the west were heavily dependent upon a small number of volatile countries for their transportation energy supply [20]. The Persian Gulf War from 1990-1991 resulted in a complete cut-off of oil exports from OPEC members Kuwait and Iraq, with the latter not resuming exports until 1997 and, as of 2011, not yet back to pre-war export volumes [2]. Additional geopolitical events in OPEC nations that have affected oil price include the Nigerian oil workers’ strike in 1994, uprisings and a General Strike in Venezuela in 2002-2003, invasion of Iraq in 2003, Nigerian militant attacks in 2006, and the Arab Spring of 2011 – including the overthrow of Muammar Gaddafi in Libya [21,22,23,24,25]. The social challenge for the U.S. and other OECD nations derives from the fact that oil exports are highly concentrated (most exports come from only a few nations) and that oil production is becoming increasingly concentrated. This concentration of resources can lead to conflict over resource access. The availability of affordable energy has a strong influence on economic and social stability.  In order to secure oil supplies for domestic economic and social stability, OECD  5  nations have made geopolitical decisions that are socially contentious. These decisions call into question issues around human rights, military support or aggression, and trade in other products [17]. As contributors to world oil production, the OECD nations dropped from a peak of 35.6% in 1985 to lows of 24.5% in 2008 and 24.6% in 2010. This highlights an increasing reliance upon non-OECD nations, and OPEC in particular, for transportation fuel supply. Oil production by country group is presented in Figure 1.3 [2]. Figure 1.3 World oil production, 1980-2010  Source: [2]  OPEC consists of twelve oil exporting nations, many of which have tumultuous relations, both in the past and presently, with the U.S. and other OECD members. The twelve nations and their historical crude oil exports are presented in Figure 1.4 [2].  6  Figure 1.4 OPEC crude oil exports, 1986-2009  Source: [2]  The world’s current reliance upon the twelve OPEC nations for oil is substantial, and that reliance is likely to grow. The International Energy Agency (IEA) estimates that more than 90% of the additional production (from current levels) required to meet demand out to 2035 will be met by production from OPEC members, based on proved reserves and rates of extraction [26,27]. While official reserve numbers should be taken with a grain of salt due to their strategic importance and the ability to increase numbers through technological innovation and exploration, they do provide a picture of the relative weighting of various countries and country groups. Based upon U.S. Energy Information Administration (EIA) data, OPEC currently holds 70% of the world’s proven reserves [2]. The OECD reserves are dominated by Canada, while those of non-OECD and non-OPEC countries are dominated by Russia – which has historically had a tumultuous relationship with the OECD and western nations, but has been the number two exporter globally behind Saudi Arabia from 1992 onwards [2]. As of 2010, global proved reserves totalled 1.35 trillion barrels (Figure 1.5).  7  Figure 1.5 Distribution of 1.35 trillion barrels total proved reserves of oil, 2010  Source: [2]  1.1.4  The Environmental Challenge of Oil  The negative environmental impacts of oil include oil spills, air pollution, and the production of greenhouse gases (GHGs). Oil spills on water kill birds, fish, mammals and plants; spills on fresh water contaminate drinking water supplies, rendering them inconsumable for humans and animals [28,29,30]. Large oil spills in oceans can be hard to contain, covering large geographic areas – examples of incidents with wide-ranging impacts include the 1989 Exxon Valdez spill in Alaska (42 million litres (ML)[31]) and the 2010 BP Deepwater Horizon spill in the Gulf of Mexico (776 ML [32]). Oil spills on land can result in long-term impacts on soil and groundwater quality, with aquifer contamination having been cited as a major reason for delaying or prohibiting the expansion of oil pipelines [33,34]. Human health effects of oil spills can be long-lasting and range from physical (e.g., migraine, dermatitis, respiratory illness, ocular irritation) to psychological (e.g., anxiety and post-traumatic stress disorder) [35]. Examples of air pollutants produced during the combustion of oil-sourced products such as gasoline and diesel include nitrogen oxides (NO x ), volatile organic compounds (VOCs), carbon monoxide (CO), and particulate matter (PM) [36]. In addition, air pollutant ground-level ozone (O 3 ) forms when sunlight reacts with VOCs and NO x molecules. These pollutants have both acute and chronic effects on 8  humans, including respiratory distress and impaired lung function, increased cancer risk, and premature death [37,38]. The U.S. Environmental Protection Agency (EPA) estimated that approximately half of all cancers attributed to outdoor air pollution are the result of mobile source (car, bus, truck) emissions [39]. As with oil spills, the human health impacts of air pollutants also have negative social and economic impacts on health systems and budgets. Air pollutant environmental and health impacts of oil tend to be concentrated near the site of combustion. This contrasts with the primary non-localized environmental challenge of oil, which is the production of GHGs and resulting climate change impacts. The combustion of oil, which results in the production of carbon dioxide (CO 2 ), is the second largest source of anthropogenic GHG emissions after the combustion of coal. In 2009, 37% of CO 2 emissions of the total 29.0 Gigatonnes (Gt) CO 2 from fuel combustion were from oil, with 43% and 20% from coal and natural gas respectively [40]. Due to the limitations on supply and price constraints, oil-sourced CO 2 emissions are predicted to decrease as a percentage of the total – to 35.5% by 2035. However, actual emissions are still anticipated to increase to 12.6 Gt CO 2 in 2035 from 10.6 Gt CO 2 in 2009 [40]. In addition, transportation emissions as a whole currently account for 23% of total global CO 2 production from fuel combustion and demand for transportation is expected to increase by 40% by 2035 [41]. Figure 1.6 shows total CO 2 emissions from fuel combustion by sector in 2009. It is important to note that although climate change has been identified as a potential significant economic influence that could cause severe strain on the global economy [42], a worldwide binding agreement on GHG emission reduction has not been established. According to traditional economic theory, should no economic value be allocated to the reduction of GHG emissions, it is unlikely emissions reductions will be a primary driver for reducing the consumption of fossil fuels [43,44].  9  Figure 1.6 Sources of 29 Gt CO 2 eq of total global GHG emissions from fuel combustion, 2009  Source: [40]  1.1.5  Looking for Alternatives  Although oil is a high energy density fuel with properties that allow it to be used for a multitude of applications, the challenges discussed previously have led governments and industry to investigate and develop both renewable and non-renewable alternatives for transportation.  Examples of currently  commercial individual mobility options include compressed natural gas fuel, electric vehicles, and the focus of this thesis, biofuels. Although the importance of these alternatives is expected to increase in the coming decades, most projections of transportation fuel supply predict that oil derivatives will continue to dominate the mix for the foreseeable future [26,45].  Therefore, when examining alternatives, any  analysis must take into account integration and/or co-existence with the primary oil-based transportation fuel system.  1.2  Bio-based Transportation Fuels Liquid biofuels are transportation fuels produced from biological materials, with examples  including ethanol, a two carbon alcohol produced from corn and sugarcane, and biodiesel, a methyl ester produced from vegetable or animal oils and fats [46,47]. Biofuels have typically been blended with 10  dominant oil-derived transportation fuels; ethanol and other alcohols with gasoline, and biodiesel with diesel. They have also been blended with jet fuels in aviation applications. Biofuels can be categorized into two major groups: 1) conventional biofuels (sometimes referred to as “first generation” biofuels), which includes sugar- and starch-based ethanol and lipid-based biodiesel; and 2) advanced biofuels (sometimes referred to as “second generation” biofuels), which includes fuels produced from lignocellulosic feedstocks, such as wood or agricultural crop residues, but also algae and other unconventional feedstocks and technologies [46,48]. While the environmental, social, and economic performance of conventional biofuels has been questioned at national and international levels [49,50,51,52], advanced lignocellulosic biofuels are often proposed as a much better alternative [53,54,55]. Lignocellulose is the most abundant biological material on Earth and has been cited as the most realistic feedstock for sustainable, large-scale (>10% market share) production of renewable fuels [56,57,58]. Biomass is the only renewable, carbon-based source of energy, enabling it to be used for a multitude of applications. Conventional biofuels have been produced commercially in large volumes for more than 25 years and have become a large (>5%) component of the transportation fuel supply for several nations (e.g., Brazil, U.S.). Conversely, advanced lignocellulosic biofuels are still under development, ranging from bench-scale lab experiments to near completion of large, first-of-kind commercial facilities. Both pilotand demonstration-scale facilities are currently producing advanced lignocellulosic biofuels in North America, but these are typically heavily government subsidized.  Several existing wood-fibre pulp  ‘biorefineries’, such as the Temiscaming facility of Tembec Inc. [59] and the Norwegian company Borregaard, produce ethanol as a co-product in their sulphite mills [60]. However, no facilities with advanced lignocellulosic biofuel as the primary product have become commercial as of 2011. Biofuels must not only be viable from a technical perspective, in a capitalist market economy they should also be a viable business that remains sustainable over the long term without ongoing government subsidies [61,62]. Commercial production of biofuels will ideally be competitive with, or out compete, oil-derived transportation fuels in three major areas: economically, socially, and environmentally. There 11  are significant lessons to be learned from conventional biofuels, from both positive and negative perspectives, that can be applied to the implementation of advanced lignocellulosic biofuels. At the same time, new policy options and recent research, development, and deployment (RD&D) results present many different, and contrasting, options for the sustainable implementation of biofuels [46,55]. Numerous questions still exist about the role of biomass in future energy systems, including the defining characteristics of future transportation systems, and the role that liquid biofuels will play within those systems. Consideration must be given to alternative transportation options such as electric, natural gas, and hydrogen fuel cell vehicles, in addition to increased use of public transportation [36]. The research detailed in this thesis focuses upon the microeconomic performance of lignocellulosic ethanol biofuel production. However, the technical, microeconomic, and macroeconomic aspects of conventional biofuels are highly relevant to the research due to technology transfer from conventional to advanced biofuels, the lessons learned from inclusion of conventional biofuels in the transportation fuel system (technically, economically, socially, and environmentally), and the potential for competition for market share between conventional and advanced biofuels. As a starting point in the discussion of biofuel performance, it is important to establish an understanding of the current biofuel industry and the production processes utilized. To undertake research, including projections on future scenarios, one must consider both the past and the present situations. 1.2.1 1.2.1.1  Overview of Conventional Bio-based Transportation Fuels  Corn Ethanol Overview The U.S. produces 51.9% of the world’s ethanol, having surpassed Brazil as the largest producer  in 2006 [63]. Total production in 2008 was 34 gigalitres (GL), with a total installed capacity of 47.2 GL in January 2009 [63]. Approximately 97% of the ethanol produced in the U.S. is made from corn (maize), with minor contributions from sorghum and wheat. Ethanol production from corn is a mature technology, with both wet milling and dry milling processes well established (see Section 1.2.1.2). Corn starch is hydrolyzed into individual glucose molecules for fermentation into ethanol. All gasoline vehicles sold in developed nations can use ethanol blends up to 10% without modifications to the engine [64]. Recently, 12  certain regions (such as the corn belt) in the U.S. reached the ‘blend wall’ of 10% and the blend limit was increased in 2010 and 2011 to allow 15% (E15) blend to be used light-duty vehicles of model year 2001 or newer [65]. Model year 2000 and older vehicles are not approved to operate on E15. North American Flex-fuel engines can operate on fuels from E0 (0% ethanol; 100% gasoline) to E85 (85% ethanol; 15% gasoline) (in contrast, in Brazil, hydrous ethanol, which contains approximately 98% ethanol and 2% water, is also used in flex fuel vehicles). Although historically a major feed crop, corn did not dominate the sugar market in the U.S. until a method of converting dextrose (corn syrup) into fructose was developed. High fructose corn syrup (HFCS), which is sweeter than sugar, can be mixed with corn syrup to replace table sugar. Archer Daniels Midland (ADM), the leading producer of HFCS, along with other producers and farmers, sought alternative markets for corn sugars and proposed ethanol as a gasoline alternative [66].  Ethanol  production from corn dramatically increased in the U.S. after methyl-tert-butyl ether (MTBE), the leading gasoline oxygenate, was found to contaminate groundwater supplies.  From 1999-2003, states  representing the majority of transportation fuel consumption in the U.S. banned MTBE [67,68] and a 13.2 GL oxygenate market for ethanol was created [69]. A large number of state bans were effective as of January 1, 2004 [67]. U.S. corn ethanol production tripled between 2002 and 2007 – a feat that had previously taken 16 years (1986-2002) to accomplish [63]. Other policy drivers that have had a large influence on ethanol industry growth in the U.S. include energy security and rural development, which are discussed in Section 1.2.4.1. Environmental benefits have also been a factor influencing the support of increased ethanol production. While a variety of feedstocks can be used for ethanol production, corn currently dominates the market in North America and therefore any alternative feedstocks proposed for ethanol production (e.g., lignocellulose) will need to compete with corn for feedstock market share. Ethanol currently represents 44% (128.8 megatonnes (Mt)) of all U.S. domestic use of corn, surpassing animal feed as the largest end use in 2011 [70]. Since 88% of U.S. corn is now consumed for ethanol and animal feed, animal product markets will inherently impact the availability and price of corn used for ethanol production [71,72]. As demand for animal products, including meat and milk, increases, 13  prices for corn will rise unless supply is increased in line with demand. Ethanol producers in the U.S. largely do not have long-term forward purchase contracts with farmers, but purchase feedstock at the market price. This operating structure and susceptibility to market prices contributed to the bankruptcy of several ethanol producers – particularly following the spike in corn prices in the summer/fall of 2007 [73,74,75]. Corn is an internationally traded commodity, with the U.S. the world’s largest exporter by far with over 60%, or 50 Mt, of the export market [76]. Real corn prices, like most agricultural commodities, have historically followed a downward trend. However, like many agricultural commodities, corn prices can be volatile and subject to seasonal disruptions and low productivity years [77]. In the U.S., corn production is concentrated in the mid-west ‘corn belt’, with the five states of Illinois, Indiana, Iowa, Nebraska, and Minnesota accounting for over 63% of the country’s corn production [78]. While wheat and barley dominate western production in Canada, corn is the most important grain crop in Ontario and Québec. Combined, those provinces produce 96% of Canada’s corn crop (63% and 33%, respectively) of 8.8 Mt (2004/05) [79]. This figure pales in comparison to the yearly 327 Mt of corn produced in the U.S. (2007 production) [78]. Corn is planted on roughly one quarter (~36 million hectare (M ha)) of the U.S.’s 138 M ha of working cropland in any given year [80]. U.S. corn is traditionally a relatively high-input crop, with variable costs of production averaging US$77.50 t-1 in 2008/09 [81]. Farm level costs are highly sensitive to input costs – predominantly fuel and fertilizer. Fertilizer pricing has a direct correlation with natural gas prices due to the use of natural gas to produce ammonia [80], while fuel costs are dependent on oil pricing. Productivity has steadily increased over the past half century, at an average annual rate of 0.12 tonnes per hectare (t ha-1), to an average U.S. yield of 10.36 t ha-1 in 2009 [80,81,82]. Therefore, variable costs of corn production equated to US$802.90 ha-1 in 2008/09, and at an average farm price of US$174.87 t-1 [70], gross operating earnings of US$930.86 ha-1. A publicized 2030 goal of ‘300 bushel corn’, taken as 300 bushels per acre (18.6 t ha-1), would require unprecedented yearly yield increases of 0.37 t ha-1 [83].  Kim et al. (2009)  14  estimated that on average, a 1 ha cornfield in the Midwest, with yields rising to 18 t ha-1, would produce 0.65 ML of ethanol over 100 years, which could transport an E85 vehicle 8.1 M km [84]. 1.2.1.2  Corn Ethanol Technology: Dry Mill & Wet Mill For ethanol production from corn and other grains, it is the starch component of the kernel that is  converted into ethanol. Starch is found in two forms: amylose, which contains only α-1-4 glycosidic bonds between individual glucose (dextrose) monomers, and amylopectin, which contains both α-1-4 and α-1-6 bonds between the glucose units. Amylose, like cellulose in lignocellulose, is an unbranched linear polymer, whereas amylopection consists of an α-1-4 chain with branches of α-1-6 linked glucose [85]. Enzymes are utilized to catalyze hydrolysis of the bonds between glucose units, releasing the individual monomers. The enzyme α-amylase acts upon the internal α-1-4 bonds, releasing the dimer maltose, which is cleaved into individual glucose monomers by amyloglucosidase [86]. This latter enzyme can act upon α-1-6 bonds and non-reducing end α-1-4 bonds. The ratio of α-amylase to amyloglucosidase is dependent upon the feedstock being hydrolyzed. Hydrolysis is followed by fermentation of the sugars into ethanol using brewer’s yeast (Saccharomyces cerevisiae) [87]. There are two primary corn ethanol production pathways: dry and wet milling. In dry milling, the corn kernels (or other grain) are ground into flour, to which water is added, thereby forming a mash. Amylase enzymes are added to the mash, breaking down the starch into gluco-oligosaccharides. Prior to fermentation, the mash is cooked to reduce bacteria levels and ammonia is added to control pH and provide nutrients for the fermenting yeast. Following fermentation, the ethanol is generally purified via distillation (to 96% purity) and further concentrated to 100% purity using a molecular sieve [87,88]. Dried distillers grains with solubles (DDGS) are the primary co-product of dry grind mills and are sold as livestock feed. In wet mills, the grain is soaked in dilute sulphuric acid for 24-48 hours to assist with separation of starch from other components; namely corn oil, fibre, and gluten. This separation of the seeped slurry is accomplished using grinders, centrifugation, screens, and hydroclonic separators [87]. The remaining steps of conversion to ethanol are very similar to dry grinding mills and include the addition of amylase 15  enzymes for hydrolysis and yeast for fermentation. Wet milling produces co-products including corn oil, corn gluten feed, and corn gluten meal, along with small quantities of vitamins and food/feed additives [88]. Wet milling yields approximately 7.5% less ethanol than dry milling [89]. Given the priority placed upon high ethanol yield and the large size of the ethanol market relative to that of wet mill coproducts, a new wet mill has not been constructed in the U.S. since 1990 and wet mills currently constitute only 20% of U.S. ethanol mills [90]. 1.2.1.3  Sugarcane Ethanol Overview Sugarcane is the second-most commonly used crop to produce ethanol and therefore must be  considered a competitor for any emerging feedstocks, such as lignocellulose. The vast majority of sugarcane ethanol is produced in Brazil, with the country accounting for 30.2%, or 26.2 GL, of total world ethanol production in 2010 [63]. As discussed below, given the flexibility in sugar and ethanol production facilities, this could be increased if ethanol and sugar prices justified increased ethanol production. Combined, Brazil and the U.S. produced 87.8% of the world’s fuel ethanol in 2010, with the EU, China, and Canada accounting for an additional 9.0% [63]. Ethanol is produced from sugarcane by fermenting the sucrose sugars pressed from the stalk. Over 300 processing plants in Brazil currently produce ethanol from sugarcane [91]. Development of the sugarcane ethanol industry in Brazil was originally driven by Brazil’s reliance in the early 1970’s on imported oil, which created large supply problems during the 1973 Arab Oil Crisis that were compounded by hyperinflation from 1980-1994 [92,93,94]. In addition, low market prices for sugar, a major Brazilian export, enabled the production of an economically competitive domestic alternative to oil-based products [95]. This development extended the existing sugar industry in a way that is still evident today, with most ethanol produced in plants that enable processors to produce sugar and/or ethanol depending upon market conditions. Sugarcane cultivation in Brazil is centered in the state of São Paulo, which accounted for 61% of the country’s 2008/09 569 Mt production. Over the past 5 years, production has increased at an average annual rate of 9.6% [96], while area planted to sugarcane in São Paulo has increased at a rate of 3.8% [97]. Sugarcane is grown in a ratoon system, in which the 16  root and stubble remain following harvest and serve as a base for new cane to grow, enabling the same stock to be harvested 5-7 times. Productivity of the sugarcane decreases approximately 15% after the first harvest, and 6-8% after each of the following harvests [98]. Only the stem (stalk) is currently processed and the leaves are either burned prior to manual harvest or removed during mechanical harvest. Burning is being replaced by mechanical harvest due to the reduction in stem sucrose yield associated with burning and the additional benefit of improved environmental performance. Delays in processing also cause a reduction in sucrose yield [91]. Sugarcane is a highly productive crop with a cane yield of 80-90 t ha-1 per year (t ha-yr-1). A more accurate indicator of useful productivity is total reducible sugars, the amount of sugar that can be recovered for conversion to ethanol. This is estimated to average about 140 kg t-1 of cane in Brazil [91]. Approximately 53% of the aboveground biomass is the harvested stem, with sugar making up 47% of the stem on a dry matter basis [99]. Therefore, 25% of the dry, aboveground biomass produced by sugarcane is in the form of sugars (largely sucrose).  Approximately 33% of the energy contained in the  aboveground biomass remains in produced ethanol from the sucrose component [99]. Van den Wall Bake et al. (2009) reported a 60% reduction in real sugarcane costs from US$35 t-1 in 1975 to US$13 t-1 in 2004 [91]. This can be primarily attributed to increases in yield, although increases in harvest efficiency and automation have also assisted in bringing down the cost. Feedstock cost as a proportion of overall ethanol production cost has remained fairly steady at 60% [91]. 1.2.1.4  Sugarcane Ethanol Technology The first processing step of converting sugarcane into ethanol is washing and size reduction of the  stalk into pieces 20-25 centimeters (cm) in length. The pieces are then pressed several times to extract sucrose from the cane, resulting in separate sugar juice and solid (bagasse) streams. Sugarcane sugar is in the form of sucrose, which is a dimer of glucose and fructose [100]. The moisture content of the bagasse is reduced to enable its use as a fuel for boilers producing process steam and electricity. The sugar juice is concentrated via evaporation, and then crystallized using heat from the boilers. Clear sugar crystals and molasses are produced, with the molasses further pasteurized and supplemented with lime to remove 17  impurities. Fermentation of the molasses to ethanol is accomplished using yeast, which can be later recovered using centrifugation. Yeast invertase catalyzes the hydrolysis of sucrose. Ethanol is extracted from the mixture using distillation, resulting in hydrous ethanol with a concentration of 96%. Cyclohexane or other means of dehydration (e.g., molecular sieves) is required to achieve ethanol purity of greater than 99.7% necessary for blending with gasoline [91,101].  Most facilities that process  sugarcane are flexible manufacturing facilities, meaning they can produce greater quantities of either ethanol or sugar depending upon market prices [91,102]. 1.2.2  Overview of Advanced Bio-based Transportation Fuels  While production of conventional biofuels is fully commercialized, with yields reaching near theoretical maximum and plant scale consistently increasing over time, advanced lignocellulosic biofuels are still being tested and demonstrated. Sims et al. (2008) provided a comprehensive assessment of the numerous conversion pathways and technologies being developed to produce advanced lignocellulosic biofuels [46]. These can be classified into two very distinct pathways, or platforms: biochemical and thermochemical. While a small number of cross-platform options, such as that proposed by Coskata, Inc., have been developed [103], the majority of advanced lignocellulosic biofuel production pathways adhere to one of these options.  Feedstock properties (e.g., woody, herbaceous), presence of existing  infrastructure, and scale all play an important role in determining which platform is better suited for a given situation. In this thesis, feedstocks are generally grouped as agricultural and forestry biomass, although there is some overlap for feedstocks such as short-rotation woody crops (SRWC). In addition, ‘waste’ feedstocks, such as municipal solid waste, can also be converted into advanced biofuels. Although not dealt with in this thesis, it is worth noting that these feedstocks are often negative value, or in other words, the facility processing facility obtains revenue for disposal rather than paying for feedstock. The business model of these facilities typically relies upon this revenue (‘tipping fees’) for economic viability and therefore differs from agricultural and forestry feedstock facilities. Facility design must also take the heterogeneous nature of the material into consideration.  18  It is also important to note that ethanol is not the only biofuel product to consider. Ethanol has several drawbacks, including being hydrophilic, corrosive, volatile, and has a low energy density relative to gasoline and longer chain alcohols [104,105]. Therefore, alternative end-products such as N-butanol, isobutanol, and biologically-based hydrocarbons are also being targeted [106,107,108]. These can be based upon the same sugar source as ethanol and can therefore be potentially feedstock neutral; the feedstock choice will be based upon sugar cost and, in some cases, environmental or social performance. 1.2.2.1  Lignocellulosic Biomass within the Energy System The potential scale of biofuels deployment, and the ability of biofuels to contribute in relation to  the existing energy system, is a primary determinant of the investment and policy support (and hence rate of development) that biofuels receive. It is an indication of the ‘size of the opportunity’ and the potential revenue from a new technology development. While single facility performance and operation is the emphasis of this thesis, an individual facility could be considered a proxy for the viability of the industry as a whole. Policy decisions that affect the industry as a whole can also be based upon the modelled performance of an individual facility, and vice versa. The ability of lignocellulosic biomass to make a significant contribution to the energy mix has been a popular research topic (e.g. [56,109]). Arable land availability is often cited as the key limiting factor in terrestrial biomass, and hence lignocellulosic biofuel, production. This is directly tied to water and nutrient availability. Several global bioenergy assessments have been conducted, with a large range of estimates on the future potential contribution of bioenergy. The estimates from several studies are presented in Table 1.1. To put these estimates in perspective, current global bioenergy consumption is estimated at 50 exajoules (EJ) yr-1 [110,111], which is approximately 10% of total world energy use [111]. U.S. finished gasoline sales in 2011 were 500 GL (770 GL ethanol equivalent) [112], while Canadian consumption in 2009 was 41 GL (62 GL ethanol equivalent) [113]. The theoretical potential of biomass energy at the terrestrial surface is approximately 3500 EJ yr-1, but realizing this potential is impossible as it would require all plant matter to be used for energy applications [114]. Berndes et al.  19  (2003), summarizing 17 previous works, reported that estimates of bioenergy potential ranged from 47450 EJ yr-1 in the year 2050 [56]. Table 1.1 Estimates on global bioenergy potential Reference Hoogwijk et al. (2003) [114]  Bioenergy (EJ yr-1) 33-1135  Lignocellulosic Ethanol (GL yr-1)a 550-18,916  Hakala et al. (2009) [115] Hakala et al. (2009) [115] Demirbas (2009) [116] Smeets et al. (2007) [117]  47-133 44-110 137 215-1272  783-2,217 733-1,833 2,283 3,582-21,199  Hoogwijk et al. (2005) [114]  130-410  2,166-6,833  Hoogwijk et al. (2005) [114]  240-850  4,000-14,166  Campbell et al. (2008) [118] Kim and Dale (2004) [119]  32-41  533-683 491  a  Notes A diversity of assumptions and scenarios at 2050 Current estimate Estimate for 2050 By 2040 Potential from surplus agricultural land only by 2050 Abandoned agricultural land only by 2050 Abandoned agricultural land only by 2100 Abandoned agricultural land Crop residues and wasted crops  Assumes biomass energy content of 18 GJ bdt-1 and ethanol yield of 300 L bdt-1 [except Kim and Date (2004)]  What can be concluded from these studies is that estimates of global bioenergy potential range significantly and that results are dependent upon assumptions on land use, yield, population, and climate. While some studies (e.g., Hoogwijk et al. (2005) [114], Smeets et al. (2007) [117]) find biomass is being significantly underutilized, others (e.g., Hakala et al. (2009) [115], Hoogwijk et al. (2003) [114]) determine that world bioenergy consumption may be near or at its peak potential for some scenarios. However, all scenarios identify that lignocellulosic biomass is produced in large quantities and can make a significant contribution to the global energy mix. Therefore, the potential opportunity is also large. 1.2.2.2  Lignocellulosic Feedstocks Lignocellulose is the most abundant biological material on Earth, and is comprised of three  polymers: cellulose, hemicellulose, and lignin [57,58]. Cellulose is similar to starch because the molecule is a polysaccharide composed of a series of glucose monomers in linear chain form. However, while the glucose monomers in starch are held together by 1,4-α-glycosidic bonds, glucose monomers in cellulose are held together by 1,4-β-glycosidic bonds [46]. This bond difference results in a much more resilient, crystalline structure for cellulose. Hemicellulose is a branched polysaccharide with a more random structure. Whereas cellulose consists exclusively of 6-carbon (C6) glucose sugars, hemicellulose can  20  contain both C6 (glucose, mannose, galactose) and 5-carbon (C5) sugars (xylose, arabinose). Hemicellulose is much more easily hydrolyzed than cellulose [46,120]. Finally, lignocellulose also contains lignin, a biopolymer composed of aromatic compounds with varying side chains that functions as a binder for holding together the lignocellulose polymer. A simplified description of the lignocellulose wood cell structure is a cellulose skeleton, composed of microfibril chains, wrapped in a hemicellulose matrix, with both components bound and encrusted by lignin [121]. The cell wall is composed of several layers, which are themselves composed of combined microfibrils and varying proportions of lignin and hemicellulose [121]. In addition to the three primary components, lignocellulosic materials also typically contain a certain quantity of ash (inorganics) and extractives [121,46]. A diagram of cellulose as it appears in plant cell walls is presented in Figure 1.7. Figure 1.7 Cellulose within the plant cell wall  Image courtesy of Office of Biological and Environmental Research of the U.S. Department of Energy Office of Science  21  Agricultural residues, which have been identified as the largest source of raw material in the U.S. [122], include corn stover, wheat straw, and trimmings from crops such as olives and fruit trees. The amount of residue that can be removed from a given field for bioenergy or other purposes is site and species specific, and it is clear that excessive residue removal can result in soil erosion and reduced yields [123]. Since agricultural residue availability is dependent upon the production of the primary agricultural crop (e.g., corn stover is not available unless corn is produced), and since the primary agricultural crop is typically valued much higher than the residues, crop residues will only be available for bioenergy applications when the crop is produced and only if removal does not negatively affect the long-term profitability of that crop [123]. Biomass crops are typically perennial crops grown exclusively for their lignocellulosic material and include both herbaceous (i.e., grasses) and woody biomass varieties. Examples of herbaceous biomass crops that are being considered as feedstocks for advanced lignocellulosic biofuels include switchgrass, miscanthus, and reed canary grass [124,125]. Options for woody biomass crops are willow, hybrid poplar, and acacia [126,127]. While the lignocellulose material yield of biomass crops is usually higher for biomass crops than agricultural residues on a per hectare basis, all the cost of establishment, land rent, management, and harvest must be borne by the lignocellulose product [128]. Historically, the establishment of biomass crops has often been supported by government policies [128,129]. While use of biomass crops usually entails harvest of the whole plant for energy applications, the multi-stage processing and interconnected nature of the forest industrial sector means a variety of feedstocks can be utilized for biofuel production. Just as agriculture-based biofuel producers can utilize grain, straw, stover, chaff, and cobs as feedstocks, forestry-based biofuel producers can utilize whole logs, harvest residues, and processing residues including pulp chips, sawdust, shavings, and hog fuel. Harvest residues are typically the branches and tops of trees left on-site at a cut-block after the whole logs have been harvested. Pulp chips can either be sourced from whole-tree chipping or lower cost residual chips from lumber production.  This latter option has historically been prevalent in British Columbia.  [130,131]. During lumber sawing, a substantial amount of sawdust is produced, while lumber planing 22  results in significant quantities of shavings. These residues are typically used to make either pellets or particle board. Finally, hog fuel is the catch-all phrase for lower quality biomass that is primarily used for heat, and sometimes power, generation on-site at processing facilities. Bark typically constitutes a significant percentage of hog fuel [132,133]. Canada is home to approximately 10% of the world’s forests, covering 42% or 4.2 M km2 of Canada’s total 10 M km2 landmass [134]. Approximately 71% of Canada’s forests are provincially owned, 22% are federally owned, and only 7% are privately held [134].  Despite a large and strong  agricultural crop sector in Canada (particularly in the Prairies), the largest annual accumulation of biomass in Canada by far occurs in Canada’s forests [135]. When determining the amount of biomass available for lignocellulosic biofuel production, forestry represents an even greater proportion of the total. This is particularly true after discounting the agricultural residue availability for sustainability requirements [123].  Therefore, when examining the potential for Canada’s lignocellulosic biomass  resources to contribute to transportation fuel supply, the emphasis should be placed upon forest biomass. Since this thesis is produced at a Canadian university (University of British Columbia) and seeks to provide knowledge and value for Canadian industry, government, academia, and other stakeholders, it is fitting for the thesis to focus on forest resources. However, in a market economy, firms will seek to utilize the lowest cost feedstock available and therefore feedstock comparisons of Canada’s forest resources to alternatives, such as agricultural residues, is critical to determining overall competitiveness. 1.2.2.3  Advanced Lignocellulosic Technologies: Biochemical Conversion Most biochemical conversion routes of lignocellulosic biomass to liquid biofuels involve the  separation of lignin, cellulose, and hemicellulose components (pretreatment); hydrolysis of hemicellulose bonds and β (1-4) glycosidic bonds between glucose monomers in cellulose (hydrolysis); and fermentation of individual sugars into alcohols (fermentation) [46,47,102]. The fermentation step may involve either five and/or six carbon sugars, depending upon the technology employed [46]. Ethanol is the traditional fuel product, but once the sugars are in the form of monomers, they can be converted to alternative fuel options, such as isobutanol. Biochemical conversion biofuels, including ethanol, are 23  typically positioned as gasoline replacements. Since ethanol is by far the largest volume biofuel in production at present, lignocellulosic-based ethanol has been emphasized as an improved alternative to corn ethanol. Previous research by the Forest Products Biotechnology Group at the University of British Columbia has also concentrated on biochemical ethanol production from lignocellulosic materials and therefore the research detailed in this thesis focuses on ethanol production from forest resources using the biochemical pathway. Biochemical conversion begins with pretreatment of the biomass to separate sugar-containing components hemicellulose and cellulose from the glue-like lignin that holds lignocellulosic biomass together. Lignin is composed of phenylpropanoids p-hydroxyphenyl (H), guaiacyl (G), and syringal (S), with their monolignol monomers being p-coumaryl alcohol, coniferyl alcohol, and sinapyl alcohol respectively [121]. Examples of pretreatments being considered for commercial application include acidcatalyzed steam pretreatment [136,137], organosolv pulping [138,139], ammonia fibre expansion [140,141], and dilute acid [142]. Hydrolysis of cellulose bonds is carried out using acids or cellulases, a class of enzymes that function in concert to release glucose monomers from the cellulose fibre. Four primary types of enzymes function to hydrolyze cellulose into glucose monomers: 1) endocellulases (endoglucanases), which catalyze cleavage of the cellulose chain at random intervals; 2) exocellulases (exoglucanses or cellobiohydrolases), which cleave glucose dimer (two glucose) cellobiose units from the ends of exposed chains – cellobiohydrolase I works from the reducing (formyl group) end of the chain, while cellobiohydrolase II works from the non-reducing end; 3) β-glucosidase (cellobiase), which cleaves the cellobiose units into glucose monomers; and 4) accessory enzymes (e.g., xylanase), which assist in the hydrolysis process [143,144].  A primary source of cellulase enzyme is the filamentous fungus  Trichoderma reesei [145]. Acid hydrolysis tends to have lower yields (50-70% of cellulose sugars converted) of ethanol production from cellulose compared to cellulase hydrolysis (75-85% of cellulose sugars converted), with the latter having the potential of reaching >95% yield [146]. Fermentation of individual glucose monomers is carried out using yeast (e.g., S. cerevisiae) or bacteria (e.g., Zymomonas mobilis) [147]. 24  1.2.2.4  Advanced Lignocellulosic Technologies: Thermochemical Conversion Thermochemical conversion to biofuels is based upon gasification of biomass feedstock and  Fischer-Tropsch synthesis of liquid fuels from the resulting synthesis gas. Biomass typically requires size reduction, to increase surface area, and drying, to minimize tar formation, prior to gasification [148,149]. Pyrolysis for the production of bio-oil, a multi-chemical viscous liquid derived from the heating of biomass in the absence of oxygen, can also be utilized as a pretreatment [149]. The resulting synthesis gas (or syngas) from gasification is a mixture of hydrogen (H 2 ), CO, methane (CH 4 ), and CO 2 . Syngas is purified to remove tars, and then converted into liquid fuels using a catalytic process known as FischerTropsch (FT) conversion [148]. Most of the catalysts used for FT synthesis are Group VIII transition metal oxides [46]. Many different chemicals can be produced from FT synthesis or from other syngas catalysis systems, but the typical primary product from the thermochemical process (as explored by Neste and other industrial concerns) is FT biodiesel, or FT liquids. Thermochemical conversion can yield 200 litres of FT liquids bdt-1, with an energy content of 36 MJ L-1 (LHV) [150]. Overall system energy efficiency, defined as energy content of the product transportation fuel relative to energy content of the biomass feedstock, is estimated to be 33-40% for atmospheric gasification and 42-50% for pressurized gasification systems [148]. Thermochemical pathway FT liquids production is considered a feedstock and investment competitor for biochemical lignocellulosic ethanol production within this thesis. Pyrolysis as a stand-alone thermochemical biofuel technology is not included here due to the requirement for additional upgrading and processing at either existing oil refineries or secondary fuel production facilities. It is considered a pretreatment throughout this thesis. 1.2.3  The Economic Performance of Bio-based Transportation Fuels  Biofuels must be economically competitive and present attractive opportunities for private industry [61,62]. This is the primary determinant for whether or not biofuels will be produced in a capitalist society. As such, it is the target of the research detailed within this thesis. The research focuses upon an individual production facility and the cost of biofuel production relative to alternatives. A large number of production variables exist, particularly for advanced biofuels, and identifying the most 25  economically competitive options will be of value to companies pursuing, and governments encouraging, commercial production. Primary among these is feedstock selection and the impact this selection has on the viability of lignocellulosic biofuel production.  Given the thesis emphasis on Canadian forest  feedstocks, these will be discussed in greater detail in this section. This thesis is focused on the techno-economic performance and risk of lignocellulosic ethanol production. In order to identify knowledge gaps, which form the basis for unique research questions included in this thesis, a review of previous techno-economic biofuels assessments was necessary. The methodology and findings from previous work also serve to inform research design and model creation while providing an important comparison for research results validation. 1.2.3.1  Feedstock Economics Overview  Biomass feedstock in the lignocellulosic biofuels industry is analogous to oil in the oil-based transportation fuels industry. Whoever has easiest access to the lowest cost supply of the highest quality raw material has a competitive advantage. Oil refiners seek to obtain the lowest cost crude oil that can meet the quality requirements of their facilities, while biofuel producers do the same for biomass feedstock. Technological advances developed by one company can be readily transferred to facilities around the globe. Therefore, the long-term sustainable competitive advantage for both an oil refinery and biofuels production facility is feedstock. Although lignocellulosic biomass is naturally more broadly distributed around the world than oil, delivered feedstock costs and the risks associated with feedstock supply vary by jurisdiction. Just as oil is available from a variety of different sources, including conventional, oil sands, oil shale, and deep-ocean, with a large range in extraction and upgrading costs and supply risks, lignocellulosic biomass is also available in a variety of forms with varying levels of quality, supply risk, and harvesting, extraction, and upgrading costs. Therefore, although conversion technologies receive a large amount of the attention when it comes to lignocellulosic biofuels, feedstock supply and risk management is a critical component of a successful enterprise, and over the long run, will be the primary source of sustainable competitive advantage.  26  A predictable, manageable, and cost-competitive supply of feedstock is essential for the operation of a biomass-consuming production facility [151,152].  Although advanced lignocellulosic biofuel  feedstocks are anticipated to constitute a smaller share of overall operating costs than conventional, foodbased feedstocks [153,154,155], effective biomass supply management will remain a prerequisite for a successful biofuels operation. In determining the long-term viability of a biofuels industry, profitability of both feedstock producers and biofuel producers (feedstock consumers) is required [80]. Since the impact of feedstock prices on profitability is inversely correlated between the two groups (i.e. high feedstock prices increase the profitability of feedstock growers but decrease the profitability of feedstock consumers), a comfortable level of profitability for both will require market balance. Alternatively, firms can be vertically integrated and control both their own feedstock supply and conversion facility. The availability of feedstocks is also dictated by competition for those feedstocks with other users. In the case of conventional biofuels, this is largely food and feed markets [54,72]. For advanced lignocellulosic biofuels, fibre (including structural materials, pulp for paper, animal bedding, etc.) and energy products (e.g., wood pellets, biomass-to-electricity) are more important competitors [72]. In addition, competition for land to grow biofuel feedstocks will have a large impact on the scale of the biofuels industry. Suitable land availability is the underlying limiting factor in the contribution terrestrial biofuels, and bioenergy as a whole, can make to the overall energy mix [109,156,157]. 1.2.3.2  Feedstock Supply Risk Feedstock supply risk is a significant factor in decision making both macro- and micro-  economically. At a micro-scale, this means sufficient, reliably available feedstock of the required quality delivered to a biofuel facility. At a macro-scale, this primarily relates to energy security and domestic energy supply (and the potential contribution from biomass to the regional or national energy mix), with the added complexity of trade.  Unlike conventional corn ethanol, which utilizes a market-traded  commodity feedstock that can be transported by truck, rail, or ship, for ethanol production, advanced lignocellulosic biofuel facilities are largely expected to utilize local biomass as the primary feedstock source [155]. While Babcock et al. (2011) identified this reliance on local supplies as a competitive 27  advantage for advanced lignocellulosic ethanol facilities over conventional corn ethanol facilities, local reliance can also create feedstock insecurity [155]. At an individual facility level, feedstock insecurity means investors may not be willing to invest in infrastructure and facilities that are reliant upon a single, local biomass source. Facilities that rely upon annual feedstocks, such as crop residues, run the risk of having very little or even no biomass available in a given year, as productivity will be impacted by shifting climates [158]. Stephen et al. (2010) found that variability in crop residue availability for biofuels is even larger than that of grain, due the necessity to leave a consistent amount of residue on the field to maintain soil organic carbon levels [158]. Diversification of feedstock production areas and feedstock types, which could be possible for advanced lignocellulosic biofuel facilities, could decrease the sensitivity of biofuel prices to extreme weather (e.g., droughts), pest infestations, and other region- and species-specific yield-reducing impacts [80]. An example of this feedstock supply risk is the British Columbia Mountain Pine Beetle infestation that began in 1999 and is expected to have killed approximately 1 billion m3 (500 Mt) of lodgepole pine by 2014 [159]. These losses represent 77% of the lodgepole pine inventory of British Columbia and cover an area of 15 M ha, the combined area of Portugal and Denmark [159]. This type of biomass feedstock disruption may become more common as the combined risks of fires, insect and other pest outbreaks, and extreme weather events reshape biological systems [160]. Forest product markets, namely the structural lumber and pulp markets, will also impact the availability of forest feedstocks for lignocellulosic ethanol production. The Canadian forest sector has already experienced the impact of uncertain feedstock supplies, particularly when facilities such as pulp mills are reliant upon chips or residues from upstream sawmills. This is the case for many pulp mills in British Columbia; those reliant upon pulp chips from sawmills were negatively affected when those sawmills were shuttered or production curtailed due to a downturn in lumber markets [130,131]. Local feedstock risk can be reduced by drawing only a small fraction of the total feedstock in an area in a single year. In addition, feedstock risks and draw area for lignocellulosic biomass can be reduced by sourcing feedstock from areas with a high growth rate. Should a disturbance (e.g., fire, 28  drought) occur, only a single year of production is lost for annual crops and the long-term impacts are felt less for harvest rotations of 5-7 years than 80 years. A single year of lost production due to feedstock disturbance would have less impact on the lifetime profitability of a facility than 10 or 20 years of lost production due to lack of feedstock. An example of the benefits of high growth rate is Brazil, where eucalyptus plantations are grown in 7-9 year rotations for pulp and require a much smaller harvest area than temperate climates [161]. Brazil has benefited from a trend of increasing pulp production in tropical regions and stagnant or declining production (as is the case in Canada) in temperate countries [162]. Historical total pulp production from several leading producers is presented in Figure 1.8. The relative performance of pulp plants in various countries can provide important insights on the preferred siting selection, and feedstock supply, of lignocellulosic biofuel facilities. Higher growth rates can mean lower feedstock, and hence production, costs. High growth rates dramatically reduce the average trucking distance and draw area required to supply an advanced lignocellulosic biofuel facility. Figure 1.8 Historical pulp production for selected jurisdictions  Source: [162] As stated by Faaij (2008), “...investments in infrastructure and conversion capacity rely on minimisation of risks of supply disruptions (both in terms of volume, quality and price)” [163]. It is clear that a stable market and reliable supply of biomass are critical to justify the investment required for the development of advanced lignocellulosic biofuel conversion capacity. 29  1.2.3.3  Integration of Biofuels within the Forest Sector Since this thesis focuses on the production of lignocellulosic ethanol from Canada’s forest  biomass resources, it is important to consider how this production could potentially integrate with the existing industry. While bioenergy has been promoted for its potential economic benefits to the forest industrial sector, there is also concern from existing companies about competition for feedstock [164]. This is particularly true when energy products, be it electricity, pellets, or biofuels, are provided with government support while structural and pulp products are not. At all levels of the wood processing chain, competition could potentially exist between traditional wood products, biofuels, and other forms of bioenergy. This latter option includes wood pellets (a solid fuel suitable for export or domestic use) and combined heat and power (CHP) production (i.e., a more localized use of the resource). Harvest residues, forest thinnings, downed/unmerchantable trees, and wood processing residues such as sawdust and shavings are the typical targeted bioenergy feedstocks and development of new facilities and facility expansion has created competition for these feedstocks with particle board and oriented strand board (OSB) manufacturers. In addition, pulp producers have expressed concern about competition for pulp chips [164,165]. For advanced lignocellulosic biofuel project developers, this means considering the role of an existing, established industry. The relationship between advanced lignocellulosic biofuel producers and the traditional forest sector will be influenced by the relative cost of production and market value of competing uses of woody biomass. Forest-based biofuels and bioenergy have been extensively studied both academically and by the forestry industry as a means to increase sector profitability, diversify their product mix, and reduce wastes [132,166,167,168]. Integration of solid biofuels (pellets) and CHP within the forest industrial system has been shown to be beneficial, with an increase in profitability and increased overall stability [133]. The developing ‘industrial ecology’ – utilizing by-products of other processes as downstream feedstocks – of the forest sector can take many forms; the most common includes one or multiple sawmills, a pulp mill, often a pellet mill, and a heat plant (with or without power production). Heat can be used for industrial processes or community heating systems [132]. Hog fuel, the lowest value residue from sawmilling 30  operations, is often used by either sawmills or pulp mills for heat and electricity production [169]. This reduces or eliminates natural gas or oil consumption, and in doing so, also eliminates a by-product that previously involved a disposal cost. Wolf et al. (2006) concluded that integration of wood pellet mills with pulp mills or sawmills increases their profitability and maximizes energy efficiency relative to stand-alone pellet mills [170]. In the past 15 years, global wood pellet production from sawdust and shavings has dramatically increased, from 1.5 Mt in 2000 to 14.3 Mt in 2010 [171,172]. As of June 1, 2011, there were 37 facilities in Canada and domestic wood pellet production capacity had reached 2.9 Mt [173]. Although some of this fine residue was used in particle board production, a large proportion was previously disposed in beehive burners or CHP facilities. By moving up the value chain, the forest sector has been able to diversify their product mix and buffer against decreases in lumber and pulp prices. Given transportation fuels command a price premium over stationary energy, the forest sector could seek to increase profitability through the production of transportation biofuels. The term ‘biorefinery’ is often used to represent a facility analogous to an oil refinery and many designs have been proposed (e.g., [174,175,176,177]). However, given the existence of an established, integrated forest sector with material flows between different facilities, it may be that a forest-based biorefinery could also be analogous to an industrial cluster. The critical component is that the entire potential of the raw material is maximized and that fibre streams are used for products that increase the profitability of the biorefinery as a whole [178]. Beyond transportation biofuels and bioenergy, new biorefinery products could eventually include plastics, adhesives, nutraceuticals, paints, fabrics, and lubricating oils [179]. Sathre and Gustavsson (2009) used linear programming to determine the ‘value add’ of 14 traditional and emerging forest products, including ethanol and FT liquids, from saw logs, pulp logs, and harvest residues [180]. Within a European market context, they found structural wood products, such as lumber and glue-laminated beams, to add the greatest value, and biofuels and bioenergy to add the least. They also found biomass type to have a strong influence on the potential value-add, with saw logs significantly more valuable than pulp logs because they can be used for structural product production. 31  However, co-production of multiple products from a single raw material was found to increase overall value. In a related study, Van Heiningen (2006) estimated than an integrated pulp-based biorefinery producing pulp, structural, and biofuel products garnered 3.5 times the revenue of a biofuel-only biorefinery [181]. Sipilä et al. (2009) found there was significant opportunity for incorporation of wastebased ethanol production within the European pulp and paper industry [182]. The Forest Products Association of Canada compared 13 different ‘biopathways’ using traditional and emerging technologies for their return on capital employed (ROCE) [168,183]. Results varied significantly by region, but as a guide, large lumber and laminated veneer lumber operations were more attractive than most energyrelated cases, with a ROCE of over 20%. Those energy options that were more attractive than structural options were largely government subsidized and varied from region to region. Although biofuels were not considered directly in the study, full fractionation, essential to biochemical biofuels production, was considered middling for ROCE at approximately 17%. Wood pellets had the lowest ROCE at 11% (equivalent to the assumed cost of capital) of the options considered [183].  Clearly, a variety of  alternative uses for forest feedstocks beyond lignocellulosic ethanol exist, making feedstock competition and integration with other users an important consideration in this thesis research on enterprise viability. 1.2.3.4  Forest Feedstock Logistics The logistics and cost of harvesting and transporting biomass has been extensively studied in the  literature [e.g.,184,185,186,187 ]. The three primary methods of forest harvest are full tree, tree-length, and cut-to-length (CTL) [188].  In full tree harvest, whole trees, including branches and tops, are  forwarded (transported) to the roadside, where they are delimbed and the tops removed. This leaves a large quantity of harvest residues in a centralized, near-road location. With tree-length harvest, stems are delimbed and tops removed within the cut block, often right at the stump location. CTL is similar to treelength in that processing is done at the stump, with the primary difference between the two methods being CTL logs are cut to shorter lengths within the forest for easier transport to a landing. CTL is often used on larger diameter trees because transporting an entire stem may not be possible or cost-effective [188]. The Forest Engineering Research Institute of Canada (now part of FP Innovations) prepared a 32  comprehensive report of harvest systems and equipment used in the Province of British Columbia, but many of these methods are applicable worldwide [189]. Included in the report were descriptions and use explanations of primary transport equipment (e.g, skidders, yarders, forwarders), falling equipment (e.g., hand, feller-buncher), processing equipment (e.g., hand bucking, stroke delimber, dangle-head processor), and loading equipment (e.g., front-end, line, self-loading trucks). Harvest method is a major determinant of biomass removal impact on forest soils and future forest productivity. Of these harvest methods, CTL and tree-length typically have less negative impacts on soil quality than full tree due to the avoidance of disruptions incurred when forwarding full trees to the roadside [190]. In addition, CTL and tree-length harvest methods also result in greater quantities of harvest residues, which have a higher concentration of nutrients than boles (trunks), remaining at the felling site [191]. The majority of studies on the impact of residue removal on soil quality and stand productivity conclude that residue retention, particularly of biomass fines (needles, leaves, twigs) enhances soil organic carbon and helps maintain nitrogen and nutrient levels [192,193,194,195]. However, results are largely site-specific and some studies find that residue can be removed with little to no negative impact on soil nitrogen and carbon (e.g., [196,197]). Soil disturbance is a critical component of site productivity, particularly in select-cut applications where damage to tree roots can hinder the quality of the forest. Given the tree bole is the main product of harvest and the source of most revenue, harvesting practices that hinder forest productivity would be counterproductive to value extraction over the long term [190]. Harvest methods assumed in the Canadian forest biomass scenarios detailed in the thesis are consistent with existing practices. No change in harvest method for forest biomass is assumed. In most cases, full tree harvest is the primary harvest method – particularly in cases where harvest residues (i.e., tops and branches) are considered a source of biomass. These are already located at or near roadside. Previous work that reviewed logistics research for both forestry and agricultural biomass, including harvest, handling, and transportation, identified a large number of logistics models, methods, and decision support systems that have been used by academic researchers to determine the delivered cost 33  of biomass within a specific region [123]. Delivered biomass cost differed significantly between regions – the result of differences in feedstock type, delivery distance (a function of facility scale and hence biomass demand), biomass pre-processing, transportation infrastructure, and mode of transportation [123]. Examples of previous studies on forest biomass logistics are presented in Table 1.2. Table 1.2 Literature examples of estimated delivered cost of forest biomass Reference Ranta and Rinne (2006) [198]  Leinonen (2004) [199]  Bradley (2007) [200] Kumar et al. (2005) [201]  British Columbia Ministry of Forests and Range [202] British Columbia Ministry of Forests and Range [202] Johansson (1996) [203]  Feedstocks Harvest residues in loose residues, chips, and bundles form Whole tree thinning and harvest residue chips Roadside harvest residues chipped or bundles Whole-tree chips from Mountain Pine Beetle-kill areas Unchipped harvest residues  Delivered Cost ($ bdt-1) 24.50-38.50 (bundles lowest cost, loose residues highest cost) 39 for residue chips in Finland; 31-43 for whole tree thinning chips in U.S. 46 for chips; 50 for bundles 50-66  Notes Transportation 100 km  distance  of  Transportation distance of 80 km Transportation distance of 100 km in Canada British Columbia, includes road building and silviculture  22-44  Whole tree  67-105  Harvest residue chips  77 in 1996; projected to be 62 in 2015  Provides indication of cost of whole logs for sawmill operations Transportation distance of 75 km  Numerous techno-economic analyses (e.g., [204,205]) have weighed capital cost economies-ofscale against biomass delivery cost diseconomies-of-scale. As facility scale increases, installed capital cost per unit decreases. However, as facility scale increases, greater quantities of biomass are required which means transporting biomass further distances. Frombo et al. (2009) found feedstock cost increased from €32 bdt-1 ($44 bdt-1) to €174 bdt-1 ($237 bdt-1) when the size of an Italian CHP facility was increased from 1 MegaWatt electrical (MWe) to 6.7 MWe [206]. Jack (2009) created a simple model contrasting facility capital cost economies-of-scale with feedstock cost diseconomies-of-scale to identify overall optimal scale [207]. Several studies, including Graham et al. (2000) and Elmore et al. (2008), have considered the impact of feedstock availability and transportation costs on facility scale in a given region, often at the State or Province level [208,209]. Kumar et al. (2003) identified the optimal scale, defined as the lowest cost of electrical generation, of an Alberta biopower plant based upon feedstock delivery cost and scaling factors [210]. Searcy and Flynn (2008) established functions relating biomass yield and  34  processing cost for power (combustion and integrated gasification combined cycle), FT liquids, and cellulosic ethanol production [211]. Wright et al. (2008) examined a distributed processing system involving delivery of pyrolysis biooil to a central biofuel production facility [150]. Overall, the optimal scale of a facility is a trade-off between facility cost and delivered feedstock cost, with technology efficiency, feedstock type and properties, and transportation infrastructure all impacting the result. Almost all bioenergy feedstock management studies (eg., [212,213]) have focused upon local supply for bioenergy facilities, principally determining the draw radius and average/incremental transportation costs by truck. However, biomass is increasingly seen as a commodity that can be traded, both inter-regionally and internationally, thereby requiring consideration of other modes of transportation. The feasibility of importing biomass using rail and ship to large, coastal facilities has been proven positive, both in literature [146,214] and as evidenced by the 1.5 Mt+ of wood pellets exported from British Columbia, Canada to Europe each year [171]. Uslu et al. (2008) also found alternative forms of densified biomass, namely pyrolysis biooil and torrefied biomass, could also be competitive with pellets in intercontinental transportation [215]. Therefore, long distance transportation of biomass is worth exploring as an operating model for a lignocellulosic ethanol facility. Feedstock logistics, including long distance transportation, and their impact on facility operation and viability are a recurring theme throughout the thesis. 1.2.3.5  Financial Markets and Processing Margin Risk Management In addition to feedstock supply risk, biofuel producers face notable market risk.  This is  manifested as changes in gross processing margin, the difference between feedstock costs and product revenues. Historically, derivatives have been used as a means to manage risk and to transfer risk between parties. Although financial derivatives, such as currency and interest rate swaps and options on stocks currently dominate the volume of transactions, derivatives were originally used by farmers to manage their risk to market price changes. For example, a corn farmer could ‘lock-in’ the price that he would receive for his crop several months in advance of harvest in order to avoid potential downturns in the market price. Forwards, including exchange-traded futures and options, are derivatives that allow buyers 35  and sellers to increase or decrease their risk. A forward is an agreement between two parties on a trade that will be carried out in the future, while an option is also an agreement that gives the purchaser the right, but not the obligation, to buy or sell an asset at a certain price and a specified point in time. Futures and options are now applied across most commodity markets, including energy, agriculture, and forestry, and are used by producers, speculators, and traders to manage their risk exposure [216]. Ethanol and biodiesel producers currently use futures and options to mitigate gross processing margin risk. For ethanol producers, their profit increases when corn prices are decreasing and ethanol prices are increasing. Vice versa, their profit decreases when corn prices are increasing and ethanol prices decreasing. This gross processing margin, the primary metric of profitability, is known as the ‘crush spread’ in the ethanol industry [217]. Ethanol producers can lock in their margins for future production by buying and selling futures and options. Biodiesel producers who utilize virgin oils, such as soybean or rapeseed, also utilize a crush spread to manage gross processing margins [218]. A related spread, location spread, is used to manage price differentials between the same or a very similar product located at two different sites. Because futures and options must contain a provision for delivery of the underlying commodity, the basis of their worth, a site of delivery must be specified. The different costs of delivering the commodity to that site must then be included in the price. The inclusion of biofuel and biofuel feedstock products in commodity exchanges provide several benefits to producers, the most obvious being price discovery and risk management (futures and options). Commodity exchanges enable increased liquidity of a product and coming together of buyers and sellers in a marketplace governed by rules and standards [216]. Conventional biofuels and their feedstocks are traded extensively on spot markets and futures exchanges, and an extension of this trade could be possible with advanced lignocellulosic biofuels.  The major challenges associated with developing strong  bioenergy exchanges for advanced lignocellulosic biofuels are feedstock and product standardization and location pricing. Given biomass feedstocks are much more costly to transport than both fossil fuels and conventional biofuel feedstocks such as corn, and delivery of the physical commodity must always be an option [216], there are uncertainties as to whether futures or options, which are exchange-traded, could 36  accurately be used to price biomass in a given region far from a trading centre. For example, corn contracts, which are used by corn ethanol producers to hedge feedstock risk, are sold at the Chicago Board of Trade (CBOT) and specify Chicago as the delivery point (although actual physical delivery is very rare) [217]. All pricing must take delivery to that point into account. Oil refineries also use a spread – the crack spread – to determine their gross processing margin. This can be used as a general proxy of refinery profitability [219]. The crack spread, which is analogous to the crush spread of biofuel producers, is the difference between the price of oil and the price of its commodity fuel products, namely gasoline and heating oil/diesel [220]. The most commonly used spread ratio is 3:2:1 – three barrels oil, two barrels of gasoline, and one barrel of heating oil. Given gasoline and heating oil are priced on a per gallon basis and there are 42 gallons per barrel, the crack spread (α cs ) can be given by: 2 3  1 3  𝛼𝑐𝑠 = 𝑃𝑔 42 + 𝑃ℎ 42 − 𝑃𝑜  (1.1)  Where P g is the price of gasoline, P h is the price of heating oil, and P o is the price of crude oil [220]. As with agricultural commodities, refineries can use financial derivatives to manage their price risk – in this case, oil and gasoline/fuel oil markets. They can ‘lock-in’ the crack spread, thereby ensuring their margins are of an acceptable level. Forwards and exchange-traded futures can be used to guarantee a price for their feedstock (oil) or product (gasoline/heating oil) [220]. Swaps, which are essentially consecutive forward contracts, allow refiners to lock in their prices over extended periods of time. Options can be used to put floors or ceilings on prices, but carry a premium for purchase. In addition to buying derivatives, refiners can also write contracts and receive the premiums afforded writers [216]. Previous research has shown that since ethanol functions as a gasoline substitute, thereby effectively increasing the supply of gasoline in the market, gasoline prices drop relative to oil prices as ethanol production increases.  This relative drop decreases the crack spread, and hence profits, of  independent refiners [219]. However, ethanol also reduces imports of gasoline, which negatively impacts refinery profit margins as well.  37  Margin volatility for lignocellulosic ethanol producers has not been studied in great detail in the past and therefore this thesis seeks to quantify this volatility and its impact on operating and investment risk. The same margin risk calculations that are applied to oil refining and corn ethanol production are used, with modifications to compensate for alternative operating models, for the processing spread research. 1.2.3.6  Techno-Economic Performance of Conventional Biofuels Advanced biofuels must compete with not only transportation fossil fuels such as gasoline and  diesel, but conventional biofuels. They will compete with the latter not only for market share, but to some extent, for inputs such as land when energy crops are considered as feedstock. Surveys of the current and projected cost of production of conventional biofuels provide a moving target for production of advanced lignocellulosic biofuels. Examples of previous studies on production costs of conventional ethanol are presented in Table 1.3. Shapouri and Gallagher (2005) found the average 2002 cost of ethanol production in the corn dry milling process to be $0.25 L-1, including net feedstock costs of $0.14 L-1 (gross feedstock cost of $89 t-1 or $0.21 L-1 ethanol) [89]. This is similar to estimates of Hettinga et al. (2009), who found net corn costs equivalent to approximately 50% of production costs [88]. Shapouri and Gallagher (2005) calculated coproducts DDGS and CO 2 provided feedstock cost credits equivalent to $0.07 L-1, with the former responsible for 98% of co-product revenues. Capital costs for greenfield conventional plant construction ranged from $0.277 L-1 to $0.79 L-1 yearly capacity, with an average of $0.42 L-1 [89]. Expansion costs were approximately $0.13 L-1 yearly capacity, indicating a much higher return on investment for expansions than greenfield construction. Ethanol yield has increased to more than 424 L t-1 of corn in 2005 from less than 378 L t-1 in 1980 [89]. In an earlier study, McAloon et al. (2000) found ethanol production costs to be $0.232 L-1 (1999 dollars) [221]. This included the corn feedstock component at $0.18 L-1 and a DDGS credit of $0.077 L-1. The authors assumed a yield of 410 L t-1 and a yeast cost of $0.003 L-1. Approximately 0.77 kg of DDGS is produced per litre of ethanol, with a sale price of $0.17 kg-1. Capital costs were estimated at $0.39 L-1 38  capacity. F.O. Lichts (2007) found that production costs had increased from $0.28 L-1 in 2004/2005 to $0.40 L-1 in 2006/2007, partially due to an increase in cost of capital [222]. Hettinga et al. (2009) used experience curves to determine the progress ratios (PRs) for U.S. corn ethanol production costs from 1980-2005 [88]. These PRs represent the reduction in cost with each doubling of cumulative production. Hettinga et al. (2009) found a PR of 0.55 for U.S. corn production, meaning costs have been reduced by 45% with each doubling of cumulative production. This compares with a PR of 0.87 for industrial processing costs, which are now estimated at less than $0.13 L-1. The total ethanol production cost, including net corn cost, was calculated to be $0.30 L-1 [88]. This is higher than that identified by Shapouri and Gallagher (2005) and McAloon et al. (2000), but this is largely due to the increase in corn price over the period 2000-2005 [89,221]. The primary drivers for decreases in production cost are increased ethanol yield, reduced energy consumption, and a focus on fuel-ethanol rather than beverage-alcohol optimized technology. Hettinga et al. (2009) also notes that the average size of dry grind ethanol plants increased by 235% between 1990 and 2007 [88]. Van den Wall Bake et al. (2009) used an experience curve approach to project future cost reductions in ethanol production from Brazilian sugarcane based upon historical PRs [91].  They found  feedstock costs to have a PR of 0.68 and industrial costs, which exclude feedstocks, to have a PR of 0.81. Total Brazilian ethanol production cost had a PR of 0.80, indicating a 20% reduction in cost for each doubling of cumulative production. Van den Wall Bake et al. (2009) found that current production cost is approximately $0.34 L-1 [91]. This finding is similar to earlier findings (adjusted for inflation) by Moreira (2000) and Larson et al. (2001) of production costs of $0.28 - $0.34 L-1 [223,224]. F.O. Lichts (2007) determined 2006/07 Brazilian sugarcane ethanol production cost to be $0.30 L-1, $0.10 L-1 lower than their estimate of U.S. corn ethanol production costs for that same year [222]. Sugarcane feedstock constituted approximately 80% of total ethanol cost. Given existing progress ratios, Van den Wall Bake et al. (2009) estimated that ethanol production cost in 2020 will be between $0.20 and $0.26 L-1 [91].  39  Table 1.3 Estimates on net cost of production of conventional ethanol Reference Shapouri and Gallagher (2005) [89]  U.S. corn  Production Cost ($ L-1) 0.25  McAloon et al. (2000) [221] F.O. Lichts (2007) [222]  U.S. corn U.S. corn  0.23 0.28-0.40  Hettinga et al. (2009) [88] Hettinga et al. (2009) [88]  U.S. corn U.S. corn  0.30 0.22  Van den Wall Bake et al. (2009) [91] Moreira (2000) [223] F.O. Lichts (2007) [222]  Brazilian sugarcane Brazilian sugarcane Brazilian sugarcane  0.34 0.20 0.30  Van den Wall Bake et al. (2009) [91]  Brazilian sugarcane  0.20-0.26  1.2.3.7  Feedstock  Notes Survey year of 2002 Model year 1999 Year range 20042007 Study year 2008 Projected cost for 2020 Study year 2008 Study year 1999 Study period 2006/07 Projected cost for 2020  Techno-Economic Performance of Advanced Lignocellulosic Biofuels Despite the lack of an operational commercial facility, numerous techno-economic studies have  attempted to model advanced lignocellulosic biofuel production cost. Examples of results are presented in Table 1.4. This previous work is highly useful to the present research as it provides an indication of reasonable modelling assumptions, estimates on cost, and operational heuristics for lignocellulosic biofuel production. Within the research chapters, comparisons of the current research to previous studies are presented. A review of previous techno-economic research also informed the design of the current research methodology by identifying knowledge gaps and opportunities to investigate novel questions on the viability of lignocellulosic ethanol production. Table 1.4 Estimates on net cost of production of lignocellulosic biofuels Reference  Feedstock/Technology  Aden (2008) [225] Aden (2008) [225] McAloon (2000) [221] Hamelinck et al. (2005) [146] Hamelinck et al. (2005) [146] Piccolo and Bezzo (2009) [226] Tijmensen et al. (2002) [148]  Corn stover/biochemical Corn stover/biochemical Corn stover/biochemical Crop residues/biochemical Crop residues/biochemical Woody biomass/biochemical  Tijmensen et al. (2002) [148] Wright and Brown (2007) [227] Wright et al. (2008) [150]  Woody biomass/thermochemical Woody biomass/thermochemical Crop residues/thermochemical Biooil/thermochemical  Production Cost ($ L-1 ethanol equivalent) 0.64 0.35 0.40 0.69 0.35 0.75  Study year 2007 U.S. DOE 2012 target Study year 1999 Study year 2005 Projection for 2025 Study year 2008  0.26  Study year 2002  0.15  Projection for 2020  0.26-0.38  Range due to facility scale  0.25-0.27  Range due to facility scale  Notes  40  Aden (2008) prepared a state of technology report for ethanol production from corn stover, including a techno-economic analysis on a 210 ML yr-1 facility [225]. He found the minimum ethanol selling price to be $0.642 L-1 on a yield of 302 L t-1. This was a reduction from $0.666 L-1 in 2005 and $1.57 L-1 in 2001. Total installed equipment cost was $133.5 M, with a total project investment of $231.7 M. This works out to $1.10 L-1 of annual capacity. Operating costs were dominated by feedstock at $0.222 L-1 ($67 dry t-1), followed by enzymes at $0.085 L-1. After tax internal rate-of-return was 10%. The stated 2012 target for the United States Department of Energy, which uses this report and previous versions for establishing goals, is a minimum ethanol selling price of $0.35 L-1, including ethanol yield of 379 L t-1, feedstock at $51 t-1, and enzyme cost of $0.026 L-1 [225]. McAloon (2000) calculated an ethanol production cost of $0.40 L-1 from corn stover (1999 dollars) [221]. Of this cost, approximately 1/3 was feedstock and 1/3 was capital cost. An electricity production credit of $0.03 L-1 was included. Total capital investment for a 95 ML yr-1 facility was estimated at $136 M ($1.43 L-1 yearly capacity), approximately 3.7 times that of a comparable capacity starch ethanol facility. This meant capital cost contribution was $0.143 L-1 compared to $0.03 L-1 for corn ethanol [221]. Hamelinck et al. (2005) found an investment cost of €1.71 L-1 ($2.33 L-1) annual production capacity in the short term, with a reduction in the medium term (10-15 years) due to technological learning and plant scale-up of €0.97-1.30 L-1 ($1.32-1.77 L-1) annual production capacity [146]. This assumes a 91% capacity factor. Capital constituted 40% of the cost of production, which Hamelinck et al. (2005) projected will decrease from €0.51 L-1 ($0.69 L-1) in 2005 to €0.26 L-1 ($0.35 L-1) in the long term (+20 years) [146]. Given the large emphasis placed by the U.S. government on agricultural residues and energy crops such as switchgrass, most studies have focused on those feedstocks. However, several studies have focused on woody biomass as the lignocellulosic biofuel feedstock. Piccolo and Bezzo (2009) compared the biochemical and thermochemical platforms for ethanol production using the Aspen Plus process simulator [226]. They found a total ethanol production cost, including capital, of €0.556 L-1 ($0.75 L-1), 41  when produced using the biochemical platform. Of this, €0.173 L-1 ($0.234 L-1) or €54 t-1 ($73 t-1) was for woody biomass feedstock. The plant capacity was assumed to be 700,000 bdt yr-1 or 218 ML yr-1, with a yield of 312 L t-1 [226]. Thermochemical platform capital costs are estimated to be higher than those of biochemical conversion. Tijmensen et al. (2002) prepared an extensive assessment of the technology and estimated a production cost (2000) of $16 GJ-1 ($0.44 L-1 or $0.26 L-1 ethanol equivalent). This assumed a feedstock cost of $37 bdt-1, yearly feedstock input of 640,500 bdt yr-1, and total investment cost ranging from $312 $449 M depending upon equipment selection [148]. In the long-term (10-20 years from 2002), Tijmensen et al. (2002) anticipated production costs to drop to $9 GJ ($0.25 L-1 or $0.15 L-1 ethanol equivalent), with reductions driven by decreased capital costs [148]. Wright and Brown (2007) determined the production costs for biomass gasification to FT liquids to be $2.10 per gallon gasoline equivalent (gge-1; $0.38 L-1 ethanol equivalent) for a 40 million gge (151 ML gasoline equivalent or 227 ML ethanol equivalent) capacity facility [227]. These costs were reduced to $1.53 gge-1 ($0.26 L-1 ethanol equivalent) when the facility scale was optimized at 486 M gge. Using a biooil-as-feedstock biorefinery production model, Wright et al. (2008) found FT-liquids could be produced, using a 550 million gge yearly capacity (3.15 GL ethanol equivalent capacity), at a cost of $1.58 gge-1 ($0.27 L-1 ethanol equivalent) [150]. This cost was reduced to $1.43 gge-1 ($0.25 L-1 ethanol equivalent) for facilities with a capacity greater than 2,500 million gge (14.2 GL ethanol equivalent capacity). The existence of forest-based pulp biorefineries producing ethanol in a flexible manufacturing structure, along with logistical and infrastructure benefits, has led both researchers and pulp companies to investigate the feasibility of either co-locating an advanced lignocellulosic biofuel facility at a pulp mill or retro-fitting a pulp mill to produce advanced lignocellulosic biofuels.  A combined Kraft  mill/biochemical platform was proposed by Red Shield Environmental LLC., who purchased the former Georgia Pacific mill in Old Town, Maine and based their plans upon the research of the Forest Bioproducts Research Initiative at the University of Maine (e.g. [228,229]). In the literature, Huang et al. (2010) found a 2000 bdt day-1 input forest biorefinery based upon the Kraft pulping process could 42  produce 38 ML ethanol yr-1 from the hemicellulose sugars, at a cost of $0.49 L-1 [175]. Total incremental (beyond the existing pulp mill) investment was $30-40 M, or $0.79-1.05 L-1 yearly capacity.  For the  thermochemical platform paired with a Kraft pulp mill, Consonni et al. (2009) examined production of dimethyl ether (DME), FT liquids, and ethanol-rich mixed alcohols. They estimated capital costs of $0.60-1.50 L-1 yearly ethanol equivalent capacity [176]. This result shows that add-on facilities to pulp and paper facilities, while notably lower in capital cost than a greenfield lignocellulosic biofuel facility, are nevertheless still projected to be higher than a greenfield corn ethanol plant. Although there are no commercial-scale lignocellulosic ethanol facilities currently in operation, announced projects and government support programs may be able to provide a reasonable indication of expected installed costs.  The United States Department of Energy (U.S. DOE) has had several  solicitations for major investments in lignocellulosic biofuel technology development and deployment, with the first of these announcements coming in 2007 [230]. Six biorefineries were granted a total of $385 M for a yearly capacity of 492 ML [231]. Planned facility capacity ranged from 43-151 ML yr-1, with most facilities in the 50-75 ML yr-1 capacity range [231]. Of these projects, two were withdrawn, three are in various stages of development, and one was constructed [230,232]. However, the company has since gone bankrupt and never managed to reach full commercial operation [232]. The U.S. DOEonly portion of funding for these facilities represents an average installed cost portion of $0.78 L-1 capacity.  Total investment was expected to be $1.2 B, or $2.44 L-1 of yearly capacity, although  investment also included 355 ML of corn ethanol capacity [231]. After deducting the total investment by standard installed costs ($0.40 L-1 of yearly capacity) for 355 ML of corn ethanol [89], the installed cost for the lignocellulosic ethanol portion of the program was $2.15 L-1 yearly capacity. Also funded with an additional $200 M were seven small projects, which could be considered demonstration scale and included both thermochemical and biochemical conversion routes, ranging in capacity from 5-10 ML yr-1 at a capital cost of $73-136 M [233]. This is an installed cost of $7-18 L-1 annual capacity. In December 2009, the DOE announced a major investment of $564 M in 19 pilot and demonstration biorefineries [234]. Overall, capital costs for advanced lignocellulosic biofuels are expected to be upwards of 300% 43  greater than those of comparable conventional biofuels [221]. To make up for this cost discrepancy, feedstock costs must be significantly less than that of conventional feedstocks to compete on a production cost basis. 1.2.4  The Social Performance and Policy Options for Bio-based Transportation Fuels  Biofuels would ideally provide equal or improved societal impacts relative to oil-based fuels, with indicators including public acceptance, job creation, energy security and defense spending, wealth distribution, and tax structure. These macroeconomic factors, while not the primary focus of the thesis, are important considerations that affect the operating environment of a biofuel production company and the market for its product. As such, a discussion on the macroeconomic and policy impacts of both conventional and advanced biofuels, as they relate to operation of a biofuels firm, is included within the thesis. In addition, the macroeconomic and biofuels/transportation policy impacts of the research results detailed in Chapters 2-6 are extensively discussed in the thesis general discussion. All energy sources require trade-offs in terms of economic, environmental, and social performance. Biofuels are no exception [15]. Government support, within a capitalist country, for one fuel source over another should only be considered if there is a market failure [235]. This could be macroeconomic (e.g., foreign government subsidize production to distort the global market), environmental (e.g., no recognition of negative environmental impacts or beneficial ecosystem services), or social (e.g., taxpayer-funded military involvement required to secure supplies). Unless biofuels provide a measurable benefit over oilderived alternatives in one of these areas, it is difficult to justify subsidies for production or consumption [235]. Firms do not operate in isolation and must therefore, when making decisions, consider the macroeconomic and policy environment in which they operate. This environment has direct implications for the enterprise viability and economic competitiveness of lignocellulosic biofuels commercialization and production and therefore requires consideration in this thesis. In all aspects, lignocellulosic ethanol must be weighed against the alternatives and their performance and the influence policy can have on that relative performance.  44  1.2.4.1  Energy Security As noted in Section 1.1, reliance upon oil as the primary source of transportation fuels presents  significant energy security challenges. Energy security is improved by replacing types and sources of energy that are susceptible to disruptions with types and sources that are not [236]. In addition, if a given source of energy represents only part of the market or can easily be replaced by another source, a supply disruption will have a smaller impact on the overall economy. Diversification of energy supply is therefore a key component of energy security [237]. It is in this vein that biofuels are promoted as a means to increase energy security, particularly in the U.S. Since Canada is a net energy exporter, energy security is not as often cited as a driver for supporting biofuel development.  However, this is a  simplification of the transportation fuel supply situation in Canada. Large geographic distances and a lack of oil transportation infrastructure from resources in the west to markets in central Canada results in the country importing large quantities of crude oil [113]. In 2010, approximately 6.3% of U.S. gasoline consumption was displaced by corn ethanol on an energy equivalent basis [238,239]. However, the net energy gain, which is often used as a proxy of performance and contribution to energy security, was relatively poor. Assuming a net energy balance of 1.25, a net energy gain equivalent to 8 GL of gasoline was realized, which represents a 1.6% energy gain for the overall system. Without the co-products, namely DDGS, the net energy gain would be close to zero [240]. These results are similar to Hill et al. (2006), who found converting the whole U.S. corn crop (2005) to ethanol production would provide a net energy gain equivalent to 2.4% of U.S. gasoline consumption [241]. While this may be interpreted as poor performance by ethanol to improve energy security, Lavigne and Powers (2007) argued that net energy gain is a poor metric by which to judge energy options, particularly from an energy security perspective [242]. It does not include information on imported oil (and hence domestically-sourced percentage of the fuel supply mix), use of biomass as a fuel for processing, how much energy in the balance can be utilized for transportation, land use changes, and geographic distribution of emissions. The metric by which energy security should be judged, they argued, is increased domestic energy supply and decreased dependence upon imports. In addition, energy that 45  can be utilized for transportation has traditionally been priced at a premium above energy sources not acceptable for transportation (i.e., stationary uses). Lavigne and Powers concluded, “Most of the energy used to create ethanol is generally inaccessible for the purposes of operating an automobile” [242]. If a country is a net importer of transportation fuel and wishes to decrease reliance on imports, the emphasis needs to be the net transportation fuel output. In addition, the ability to utilize alternative transportation fuel options in the existing infrastructure is a critical prerequisite for a gasoline substitute. Ethanol, while not easily transportable due to its hydrophilic nature, can nevertheless be distributed using existing fuel stations [105]. Lavigne and Powers (2007) found lignocellulosic (corn stover) ethanol was much more compatible with energy security goals than corn grain ethanol. They estimated that foreign energy inputs to net energy output were 7.6% (1.5 MJ litre-1) for lignocellulose vs. 17% (3.1 MJ litre-1) for corn grain [242]. The significant advantage afforded lignocellulosic ethanol was the use of the lignin fraction of biomass to produce electricity and heat. In either case, foreign energy was only a small component of the energy input required to produce ethanol. Therefore, lignocellulosic ethanol produced on a large scale could make a notable contribution to improving domestic energy security and diversifying the liquid transportation fuel supply. When considering the energy security performance of biofuels relative to oil-based fuels, it is important to determine whether geopolitical risk in other countries (oil exporters) is being exchanged for biofuel feedstock yield risk. Due to the current centralization of ethanol production in the Corn Belt states of the U.S. and the São Paulo state of Brazil, ethanol prices are subject to large price swings due to seasonal growing conditions in specific geographic areas [243]. This has a direct impact on the viability of single firm, since they must manage this revenue volatility and protect gross processing margins. The U.S. is highly reliant upon corn as the primary feedstock for ethanol. Eaves and Eaves (2007) found that corn production in the U.S. was more variable than oil imports (standard deviation of 11.9% vs. 6.8%) over the period 1960-2005 [71]. However, their analysis accounted for neither price nor elasticity of demand. 46  Due to structural imbalances in gasoline vs. diesel production and demand in Europe (there is an oversupply of gasoline production relative to a strong diesel demand), gasoline can be imported from Europe to North America should domestic refining capacity not meet demand [219]. However, ethanol production serves as pseudo refining capacity and a buffer to peaks in gasoline prices caused by very high marginal costs when gasoline demand reaches the limit of North American gasoline production capacity and European supply is limited. According to Du and Hayes (2009), “in the absence of this additional crude oil refining capacity, the [fuel price] impact of eliminating ethanol would be extreme.” [219] However, Eaves and Eaves (2007) reported U.S. corn ethanol has a poor ability to assist in absorbing transportation fuel demand shocks, given most shocks occur during the Northern Hemisphere summer when corn stocks are at their lowest [71]. Since corn ethanol currently dominates the North American biofuel market, the season variations in ethanol production and prices must be taken into account by lignocellulosic ethanol producers in operational and financial planning. In addition, developments in domestic oil refining capacity could also impact the market price of ethanol, and hence revenue, for lignocellulosic ethanol producers. Rubin (2009) challenged the premise of energy security, a justification for biofuels support, and the ability of a free-market nation to achieve energy security [1]. He argued that energy security, as represented by insulation from energy price volatility, is not possible in nations that practice free trade and open markets. This contrasts with many OPEC nations that subsidize domestic oil consumption and set local prices, thereby insulating their citizens from global market prices [1]. This argument has significant impacts for biofuels and bioenergy, since neither would provide energy security to a producer nation unless a protectionist, insular energy policy was adopted whereby the prices of domesticallyproduced fuels were not subject to international free-market prices. This would require a model similar to that employed in OPEC nations and would be a dramatic reversal of policy for trade-liberalized western nations. However, since biofuels do not constitute a large proportion of the fuel market, using biofuel policy to control greater transportation fuel prices could be challenging.  47  Until recently, the U.S. provided substantial support policies for domestic corn ethanol in the form of a $0.14 L-1 ($0.54 gal-1) import tariff on ethanol from countries such as Brazil and a $0.12 L-1 ($0.45 gal-1; previously $0.51 gal-1) blenders credit (Volumetric Ethanol Excise Tax Credit (VEETC)) for inclusion of ethanol in the gasoline pool [244,245]. Both these support policies altered fuel market dynamics to favour domestically produced fuel and discourage importation from other countries. Now that the import tariff and blenders credit for domestic fuel have been removed, blenders will be free to choose the lowest priced ethanol available, and in turn, provide the lowest cost ethanol-gasoline blends to consumers.  This development may lead to greater imports from other countries and increased  competition in U.S. biofuel markets, with the anticipated impact of reducing prices. Elimination of these market-distorting policies could in fact increase energy security due to the diversification of fuel supply [237]. 1.2.4.2  Competition with Food A perceived competition between food/animal feed and biofuel producers for agricultural  commodities such as corn, soybeans, and rape has led to the ‘food vs. fuel’ debate. When food prices spiked in 2007-2008, biofuels were identified as a potential contributor to the increase [246]. According to one World Bank analyst, 75% of the increase in food prices between 2002 and 2008 could be attributed to biofuels [247]. Although this number was largely discredited [248,249] and alternative estimates of the influence of biofuels demand on food price increases ranged from 10-15% [250,251], it is nevertheless a reality that conventional biofuels impact food prices and are expected to impact food security in some regions [252]. However, the vice-versa is also true, and food prices will invariably impact feedstock prices for conventional (and potentially advanced lignocellulosic) biofuels. That is why discussion of food markets is justified when examining the viability of biofuel enterprises. The role of biofuels in agriculture has become so significant that in 2008 the United Nations Food and Agriculture Organization (FAO) devoted their annual ‘State of Food and Agriculture’ report to biofuels [72]. In depth analyses determined that biofuels are only one of many drivers, including weather-driven reductions in production, low global cereal stocks, income growth leading to a greater 48  demand for meat (and hence feed) consumption, population growth, urbanization, and high energy prices (which strongly impact production costs) that drove prices higher [72,253,254]. The interdependencies of these drivers, combined with the interconnected nature of energy and food markets, mean a single variable cannot be attributed a direct causal impact for the majority of the increase [253,254,255]. Biofuels create a convergence between agriculture and energy markets, reversing a historical trend that began with the development of fossil fuels to replace horses and working animals. While agricultural inputs to the energy markets were minimal until the rise in biofuels, fossil fuel inputs, in the form of fertilizer, pesticides, herbicides, and diesel fuel, to agriculture have steadily been increasing [72]. Without fossil fuels, and in particular oil, our current industrial food supply structure would not be possible [1]. Utilization of forest-based feedstocks, while attracting its own host of controversial issues (e.g., deforestation, biodiversity), will nevertheless avoid the food vs. fuel and associated moral debate. This could be used as a justification by producers to receive preferential treatment in the form of policy support. The FAO (2008) concluded that policies that support the consumption of biofuels in developed countries could provide net food exporting developing countries with increased revenues, assuming an absence of trade barriers [72]. However, conventional biofuel support policies in developed nations will, in the short and medium term, have negative impacts on net food importing developing nations. This is due to the general upward pressure biofuels exert on food prices and therefore the increased food importation cost [250,251]. Net food exporting nations, in contrast, benefit from the higher prices. Biofuel consumption support programs in developed nations will continue to be a net detriment to net food importing developing nations if higher prices do not stimulate agricultural production to the point a country becomes a net exporter [72,235]. In regards to advanced lignocellulosic biofuels, biomass crops will not directly compete on grain markets with food products, but “...acreage availability is a binding natural limitation that could lead to conflict with food production.” [235]. While the FAO (2008) predicted increasing demand for biofuels will have negative impacts on net food importers, both at a national and individual level, they also recognized that there are substantial opportunities in developing 49  countries to export biofuels and “...hasty decisions to restrict biofuels limit opportunities for sustainable agriculture that could benefit the poor.” [72] Based upon results from their global agricultural model, Searchinger et al. (2008) projected that meeting the U.S. corn ethanol mandate of 56 GL by 2022 would result in a decline in U.S. exports of corn by 62%, of wheat by 31%, and of soybeans by 28% [52]. However, recent corn export data show that U.S. corn exports in 2007/08 reached their highest level since 1979/80 and world trade in corn has been climbing since 1985/86 [256]. This is during the same time as a large expansion in U.S. corn ethanol capacity [63]. This has direct implications for the lignocellulosic ethanol industry because it means that a substantial quantity of corn is still available for ethanol production, should prices justify conversion. In the absence of subsidies, lignocellulosic ethanol will need to compete on price point with corn ethanol. Global food demand is expected to double over the next 50 years as both population and per capita caloric intake increase [257]. Over 70% of all corn grown worldwide is fed to animals [84], a highly inefficient means of supplying food energy to a growing population. Meat consumption has been directly linked to a higher standard of living, and as the per capita of income of developing countries increases, so does the meat consumption [258,259]. Federoff and Cohen (1999) predicted that this will exponentially increase the demand for animal feed and the land to grow the feed [257]. Therefore, according to the authors, even without production of biofuels, food prices are expected to increase [257, 259]. This prediction, in which food demand will outpace supply, is a Malthusian argument [260] and contrasts with the historical trend of a decrease in food prices as a percentage of income [261]. If food prices, and hence corn prices rise, thereby increasing the cost of conventional ethanol production, lignocellulosic biofuels could become increasingly competitive. In the case of forest resource-based biofuels, forested land can be unfit for agriculture, making land competition minimal in such instances. 1.2.4.3  Justification for Government Support Despite the seemingly win-win-win convergence of the three main policy drivers (GHG  reduction, energy security improvement, and job creation) for support of biofuels, Brown and Huntington (2008) found that policy makers face a trade-off between these drivers [15].  This is because of 50  opportunity cost: due to a limited quantity of resources, supporting one technology, that is the preferred option for one policy driver, comes at the expense of another technology, which could be the preferred option for another policy driver. Since none of the alternative energy options currently available are the preferred option for all policy drivers, there is inherently a trade-off. By supporting one technology, the opportunity to support an alternative, competing technology is reduced.  As stated by Brown and  Huntington, due to limited resources, “At any given cost, the increased use of any one technology must result in the reduced use of another technology” [15]. Compared to all other options, lignocellulosic ethanol was identified by Brown and Huntington as having the most ‘balanced’ trade-off between energy security and climate change mitigation. However, options such as nuclear power are identified as a better option for mitigating climate change and the implementation of lignocellulosic ethanol reduces the ability to implement nuclear power due to resource (i.e., financial) limitations. Tilman et al. (2009) argued biofuels should only receive policy support when they have a positive impact on energy security, GHG balance, biodiversity, and security of the food supply [53], while Hill et al. (2006) claimed that for a biofuel to be viable, it must have a positive net energy balance (i.e. energy gain), be environmentally improved over fossil fuels, be economically competitive, and be producible in large quantities without reducing food supplies [241]. Peters and Thielman (2008) asserted that biofuel support policies should not be indefinite and should only be used to support an emerging industry [235]. This has been the case in Brazil, with the elimination of ProAlcool [95,262], in the United States, with the elimination of the ethanol import tariff and ethanol blenders credit [244,245], and in Germany, with the elimination of the biodiesel tax credit [263]. Sanderson (2006) argued that U.S. corn ethanol support is motivated by the strong corn lobby and the fact that the primaries for party nominations for U.S. president start in Iowa, the country’s largest corn producer [104]. The current policy climate, with financial austerity a key component of many countries’ plans, does not suggest large, ongoing subsidies for biofuels. While lignocellulosic ethanol is an emerging industry and can therefore be supported under the premise of technological innovation, long-term subsidies for production of lignocellulosic ethanol may be  51  difficult to justify by policymakers at the present time. This must be taken into account by lignocellulosic ethanol producers when predicting fuel prices and revenue over the long-term. Du and Hayes (2009) studied ethanol production elasticity relative to gasoline prices and oil refinery profit margins [219]. They found that, based on wholesale prices from January 1995 to March 2008, ethanol production kept wholesale gasoline prices $0.14 gal-1 ($0.04 L-1) lower than would otherwise have been the case. The impact varied significantly by region, with the Midwest having the greatest reduction of $0.28 gal-1 ($0.074 L-1) and the Rocky mountain region the least at $0.07 gal-1 ($0.02 L-1) [219]. The authors calculated that ethanol support policies precipitated a net welfare loss of $0.5 B in the U.S. in 2007, largely driven by reduced profits for refiners from a reduced gasoline price ($1.33 bbl-1) and losses by taxpayers. This is approximately half the net yearly (2007) biofuel support policies economic cost of $1B identified by Hahn and Cecot (2008) [264]. In both studies, welfare losses were found to be at least partially offset by gains in the income of crop growers and land owners, and consumers’ benefit of reduced gasoline prices. In effect, ethanol support policies, while having a net cost, were found to transfer funds from large oil refiners/blenders and taxpayers to farmers, landowners, and fuel consumers. All else being equal, since ethanol reduces price paid by consumers, economic theory suggests the amount of transportation fuel consumed (on an energy basis) will increase [219]. Based upon these previous analyses, it is reasonable to assume that policies supporting the production (and consumption) of lignocellulosic ethanol from Canada’s forest resources are likely to result in a net fiscal transfer from fuel blenders and tax payers to foresters and fuel consumers. However, since most forested lands in Canada are public, the cost to the tax payers would be partially received by the Government of Canada (and hence tax payers) as landowners. 1.2.4.4  Policy Support: Tax Incentives The primary means of ongoing (i.e., not one-off) support for biofuels has been an exemption from  the fuel taxes applied to fossil gasoline and diesel. An example was the U.S. VTEEC (blenders credit) of $0.12 L-1 ($0.45 gal-1) provided to gasoline blenders before it was repealed in 2011 [244,245,265]. Exemption from fuel taxes is a significant benefit for biofuel producers, as it can dramatically improve 52  competitiveness since fuel taxes can constitute a significant portion of the cost of transportation fuel. According to Newbery (2005), fuel taxation is justified by 1) being the second-best means for charging for road infrastructure (the best and most direct being road tolls); 2) as a means to internalize in the fuel price the external costs caused by fuel use; 3) as part of a second-best tax system to improve the efficiency of other taxes; and 4) as an optimal import tariff to shift the scarcity rent from exporting to importing countries [266]. Given the rationalizations for taxing, the omission of taxes from biofuels must be justified by environmental, social, or economic benefits. As production grows, omitting biofuels from fuel taxes could be a major burden on the national budget. Therefore, assuming a long-term fuel tax exception at the individual firm level could be a risky proposition for lignocellulosic ethanol producers. As highlighted by Peters and Thielman (2008), fuel taxes are a major source of government revenues and charging for road use is vital to road maintenance and extension financing [235]. Given vehicles utilizing biofuel will require the same road infrastructure as fossil fuel powered vehicles, a reduction in taxes may not be a sustainable option or politically justifiable. In fact, biofuels will likely increase road traffic in some regions due to increased transportation of feedstocks and fuels [235]. Tax reduction based upon fewer externalities (i.e. lower GHG production) to be internalized in the pricing of fuels may be justified for advanced lignocellulosic biofuels or sugarcane-based ethanol. However, other conventional biofuels have limited GHG benefits and increase other externalities such as water consumption and water contamination, thereby reducing the justification for a complete elimination of fuel taxes [266]. Under the premise of fuel tax as an import tariff, shifting scarcity rent from exporting to importing countries, elimination of the fuel tax on domestically produced biofuels is better justified. De Gorter and Just (2009) determined that tax incentives for biofuels, in the absence of consumption or blending mandates, will reduce the cost of transportation fuels and therefore increase their consumption [267]. This is due to a shift in the cost burden from fuel consumers to tax payers. Given the increase in overall fuel consumption due to decreased fuel costs, benefits such as reduced fuel GHG intensity could be overwhelmed at a macro scale by increases in fuel consumption [267]. Tax incentives may also be a less effective way of creating new capacity; Mabee (2007) found that direct 53  start-up funding support is a greater determinant of ethanol production capacity increases than excise tax exemption [268]. 1.2.4.5  Policy Support: Blending Mandates Blending mandates, which require fuel sellers to include a percentage of biofuel in the  transportation fuel they sell, are a regulatory approach to increasing biofuel use. Instead of reducing the cost of blenders to include biofuel within the mix, blenders are required to purchase biofuel at the market price for inclusion [267]. A mandate essentially creates a guaranteed market for biofuel that may otherwise not be included in the fuel supply. Instead of placing the burden upon taxpayers, as is the case with fuel tax reductions for biofuels, the burden of increased fuel cost is borne by blenders, which in turn, pass it on to the fuel consumers – both households and firms [267]. Peters and Thielman (2008) pointed to the budget effect, whereby the increased cost of fuel reduces consumers’ budgets for the purchase of other goods [235]. This feed-on effect could mean a negative impact on the economy as a whole. Many States of the U.S., most Canadian Provinces, and the Canadian Federal Government have mandated biofuel percentages within the transportation fuel mix. In Canada, the national renewable content mandate is 5% in gasoline and 2% in diesel [269]. While the U.S. Energy Independence and Security Act (EISA) of 2007 includes mandates for lignocellulosic biofuel blending, production volumes have fallen far short of the requirements [270]. Therefore, it is clear that blending mandates, in the absence of an economically competitive or technically viable production process, are difficult to enforce in a capitalist economy. Mandates alone are insufficient to entice commercial production when large private capital investment is required and the business proposition is unproven. Alternatively, mandates must be accompanied by strict penalties for non-compliance. According to Gorter and Just (2009), the implementation of blending mandates alongside tax incentives may actually subsidize fossil fuel consumption instead of biofuels [267].  This policy  combination places the burden of fuel consumption on tax payers and causes an increase in oil dependence and GHG emissions, the exact opposite of the stated goals of most biofuel support programs. This may be the case until biofuels make up a much larger proportion of the fuel supply mix. Thompson 54  et al. (2009) used a multi-year, multi-commodity equilibrium model to determine the relationship between ethanol, oil, and corn in an ethanol-mandated fuel policy regime [271]. They found that mandates weaken the link between ethanol use and ethanol price, and serve to make ethanol demand relatively unresponsive to corn yield or oil price. This corn and gasoline price inelasticity due to blending mandates was confirmed by McPhail and Babcock (2012) [272]. In contrast, ethanol consumption exhibits a high responsiveness to corn yield and oil price in the absence of a mandate [271,272]. At the same time, the presence of a mandate strengthens the influence of corn yield on ethanol price. Therefore, the gross processing margin of ethanol producers becomes more sensitive to corn yields and less sensitive to oil prices. A significant corn yield decrease will cause a shock to ethanol market prices under a binding mandate, as consumption will not change to offset changes in the corn price due to reduced supply. Biofuel mandates can therefore result in higher price variability in both feedstock (e.g., corn) and ethanol markets [272]. As long as mandate volume exceeds industry production, pricing control will be in the hands of producers. However, if industry production exceeds mandated volume, producers will be subject to traditional supply and demand pricing.  Therefore, under a volumetric mandate policy,  individual firms must consider current biofuel production capacity relative to mandate volume, as exceeding mandate volume can result in a dramatic drop in prices that were previously artificially supported by the mandate. 1.2.4.6  Policy Support: Import Tariffs In order to reap the job creation, rural economic development, and energy security benefits  associated with biofuel production [273] domestically, many countries have implemented import tariffs on biofuels from other countries. For example, until 2011, the U.S. placed a US$0.54 gal-1 (US$0.14 L-1) duty on imported ethanol [265]. The largest exporter of biofuels in 2007 was Brazil, with 3.5 GL of ethanol exported [274]. The largest importers were the U.S., the Netherlands, and Japan. The Caribbean region was also a major importer, but also a major exporter, highlighting the flow-through nature of ethanol shipments destined for the U.S. This trade through the Caribbean was due to the duty-free status, constituting up to 7% of the U.S. ethanol market, of imports from select Central American and Caribbean 55  countries [265]. However, now that the U.S. import tariff has been dropped, this is likely to change. China has also become a major exporter in recent years [275]. Although lignocellulosic biofuels have not reached full commercial status, many government support policies nevertheless target domestic production to serve domestic markets and the disputes observed in conventional biofuels trade could potentially also occur for lignocellulosic biofuels. Protectionist trade policies in agriculture and forestry, the two primary feedstock suppliers for lignocellulosic biofuels, have a long history and are a major point of contention in the drive towards free trade.  Biofuels could either exacerbate disagreements on  agriculture trade or could be used to mitigate these disagreements by providing an additional domestic market for feedstock supplies [276]. While import tariffs may protect domestic jobs, they can also result in sub-optimal outcomes for other drivers of biofuels, including GHG reductions and energy security (as dictated by transportation fuel prices). Brazilian sugarcane-based ethanol has a significantly smaller GHG footprint than corn-based ethanol, even after accounting for ocean shipping from Brazil to North American or European markets [277,278]. Therefore, placing a tariff on imported Brazilian sugarcane ethanol has the effect of reducing the potential of biofuels to contribute to GHG reductions. Environmental benefits are sacrificed at the expense of protecting domestic producers. Import tariffs also distort markets and increase consumer prices, thereby reducing national energy security. Limiting the opportunity of imported fuels to enter domestic markets could reduce the suppliers in the market and decrease competition. This situation also makes the market overly responsive to developments in domestic production. In major conventional biofuel producing nations, reliance upon only domestic crops and feedstocks for biofuel production increases the risk of price shocks [158]. This is because crop yields can be highly variable and low crop yields will inherently lead to higher feedstock (and hence biofuel) prices – particularly when combined with a volumetric mandate policy. Volatility in market biofuel prices could be reduced by diversifying sources of feedstocks and biofuels. Biofuel support policies aimed at increasing energy security and domestic transportation fuel supplies can lead to support of not only domestic producers, but foreign ones as well – particularly in the 56  absence of import tariffs. Ironically, Brazil was the world’s largest ethanol importer from 1989 to 1996 [279], despite being the world’s largest producer. It must be recognized that there are potential trade-offs when implementing policies to support biofuels. Should imported biofuels have a lower delivered cost than domestically-produced fuels, policies that support biofuel use could be subsidizing foreign producers. However, import tariffs, or subsidies that only apply to domestic producers, could result in higher consumer prices and sub-optimal biofuel benefits (e.g., GHG reductions). This situation may apply to not only conventional biofuels, but also advanced lignocellulosic biofuels and further investigation is required. In the absence of ethanol import tariffs and domestic production incentives, Canadian lignocellulosic ethanol will need to compete with imported conventional and advanced lignocellulosic ethanol in the Canadian marketplace. A techno-economic comparison of Canadian lignocellulosic ethanol production to that in Brazil is a key component of the thesis. This comparison is important not only from an individual facility/firm competitiveness standpoint, but also from a macroeconomic policy standpoint. It is essential that policymakers understand relative competitiveness of firms, as this will dictate trade flows and determine which market actors will be negatively impacted, and which will be positively impacted, as a result of government policies. 1.2.5  The Environmental Performance of Bio-based Transportation Fuels  The environmental performance of biofuels includes the impacts on air, water, and soil and the health impacts of biofuels on humans and other species. The environmental performance of advanced lignocellulosic biofuels is generally expected to be superior to that of conventional ethanol and gasoline. This includes environmental indicators such as GHG production [241,280,281,282,283,284,285], air pollutants [36,286], water consumption [287,288,289,290], water quality [55,291,292], soil quality and erosion [123,293,294], and biodiversity [55,293,295,296,297,298]. Concerns of inferior environmental performance relative to fossil fuels include direct and indirect land use change [52,299,300,301,302] and increased acetaldehyde emissions for ethanol [303]. Although environmental performance is not a focus of this thesis, it will be an important operating consideration for facilities. Environmental performance 57  would have a greater influence on the microeconomics and design of a lignocellulosic ethanol facility (and hence relevancy for this thesis) if impacts were allocated a financial cost or benefit.  These  ‘ecosystem services’ would then be included in the financial performance of facilities and appear as revenue or expenses on the financial statements of operating companies. As the value of recognizing ecosystem services, including carbon emissions, clean water, and air quality, increases, the importance and impact of environmental performance of energy options on the balance sheet of companies rises. Therefore, environmental sustainability becomes intertwined with economic performance. If ecosystem services were fully recognized financially, a techno-economic assessment would be incomplete if they were not included. However, since allocating financial value to ecosystem services is currently limited in North America, inclusion would be presumptuous and could distort the results of the research.  1.3 Thesis Objective and Themes Commercial operation in a capitalist society implies a profitable enterprise and therefore, this thesis will focus on the primary determinant of whether lignocellulosic ethanol is a viable product: its economic competitiveness.  Many developed countries have invested significant public funds in  developing advanced lignocellulosic biofuels in order to meet energy and policy goals. The most commonly cited drivers for public investment in this area are climate change mitigation, energy security, and rural development/job creation, with varying levels of emphasis given the jurisdiction [53, 235,241,242,304]. To date, little work has been done to address the fundamental question of whether biofuels can actually be an economically viable co-product within a temperate forest industrial system over a processing facility lifetime, given the variability in feedstock costs, reliance upon revenue from coproduct markets, changes in gross processing margins, competition from southern hemisphere producers, and the presence of established conventional biofuel producers. Despite optimistic projections that biomass could meet more than 100% of the world’s energy demands (e.g., Hoogwijk et al., 2003 [114]), the reality is that biomass supplies are not infinite. This is particularly true for forest-based biomass, where existing industries already utilize much of the fibre for  58  established markets. Lignocellulosic ethanol is but one of the potential products that can be produced from forest-sourced lignocellulosic biomass. There are a number of competing uses for this feedstock, including traditional forest products lumber and pulp, wood pellets, electricity, and heat.  From a  technical perspective, biomass is the only renewable material that can directly replace fossil fuels in the existing energy infrastructure, whether transportation systems (e.g., biofuels) or electricity production facilities (e.g., biomass chips or wood pellets). Woody biomass will therefore be in increasing demand from companies and sectors wishing to retain existing infrastructure but reduce fossil fuel consumption. The importance of the incumbent fossil energy fuel characteristics and industrial structure, and that of the existing forest industrial system structure, should not be underestimated. Lignocellulosic ethanol must fit within both systems; complementing existing forest products in production and fossil fuels in utilization. Demand for biomass from the traditional markets of food, fibre, and shelter will remain, which means competition for material will inevitably lead to making choices about whether lignocellulosic ethanol is a viable and long-term competitive product. Whether or not biofuels are in fact a profitable product for forest industries remains to be proven. At approximately 5.4% efficiency, (compared to 80% for heat) [305], transportation biofuels are a relatively inefficient use of biomass in terms of total energy applied to useful work – that is energy utilized for a primary activity or purpose (in the case of biofuels, moving a vehicle). However, they remain a primary focus of bioenergy technology research, development, and deployment programs in both North America and Europe. While it is important to take into account processing efficiency, market demand will inevitably dictate use of the raw material. Given that transportation energy (gasoline) has historically commanded a +700% premium over stationary energy (coal) on a fuel energy basis, conversion of biomass to biofuels clearly adds value compared to heat and power [306,307]. Whether or not biofuels create sufficient value relative to other forest products for various feedstocks needs to be determined. Not only will lignocellulosic ethanol producers need to compete with the traditional forest sector for feedstock, they will need to compete with incumbent conventional biofuel producers – namely corn 59  and sugarcane ethanol companies. With several decades of experience of industrial fuel production, the conventional ethanol industry has a significant first-mover competitive advantage. However, as presented earlier in this thesis, there are many lessons that can be learned from conventional biofuels and applied to the production of lignocellulosic ethanol. Opportunities to decrease the cost and risk of production must be found in order for this emerging product to make a significant contribution to the transportation fuel mix in Canada and other nations. The production of biofuels is an inherently volatile business, with risk categorized into three primary areas: feedstock, technology, and market.  Conventional ethanol (e.g., corn-based ethanol)  producers have suffered from a divergence in feedstock and product pricing, caused by competition for feedstock from food and animal feed producers on one hand, and exposure to volatile oil prices on the other. The inflexibility of corn ethanol producers to respond to these variables has led to a number of bankruptcies and a tarnished reputation for the industry as a sector for investment. Lignocellulosic ethanol producers will be faced with similar gross processing margin risk, with competition for feedstock from traditional biomass users, such as lumber and pulp producers, as well as other bioenergy (e.g., electricity) producers. Similarly to the existing domestic pulp sector, biorefineries operating in Canada must be able to compete with producers operating in tropical and sub-tropical climates. Given the extensive research on the technical aspects of lignocellulosic ethanol production but limited commercialization success, the primary objective of this thesis is to identify the structural, operating, and financial conditions that would permit economical production of the fuel at a commercial scale. Since Canada is home to a large forestry industry and extensive forest resources, forest-based woody biomass is a natural choice for feedstock when investigating domestic lignocellulosic biofuel options. The existing forestry and forest products industry is seeking to diversify its product mix beyond lumber and pulp and paper, and bioenergy has been highlighted as a potential opportunity. Lignocellulosic ethanol is but one of a number of bioenergy options that could be considered. Therefore, the First Theme of the thesis focuses on the competitiveness of lignocellulosic ethanol production from 60  woody biomass relative to other bioenergy options, including other lignocellulosic biofuels (namely the thermochemical platform) and existing commercial products such wood pellets, stand-alone biopower, and combined heat and power. This theme is about the ability to compete for woody feedstock from the forest products production chain, including harvest residues, sawmill residues (sawdust/shavings, pulp chips, and hog fuel), and whole logs. The ability to compete is impacted by product revenue, operational scale, and feedstock selection, which are primary components of the theme. Securing feedstock at a manageable cost will be critical to the long-term economic production of lignocellulosic ethanol, since without feedstock, a plant cannot operate and ethanol cannot be produced. Lignocellulosic ethanol will need to be competitive in existing markets with established transportation fuels – namely gasoline and conventional (corn and sugarcane) ethanol.  Although  produced from different feedstocks with differing life cycle impacts than conventional ethanol, the end product is chemically identical to the existing fuel. Thus, in the absence of government policies that manipulate the fuel market to favour a specific fuel produced from a specific feedstock, lignocellulosic ethanol must be able to compete with conventional ethanol for market share. Although the presence of corn ethanol in North American fuel markets benefits lignocellulosic ethanol from a fuel handling, distribution, and public/OEM acceptability perspective, the long history of corn ethanol production and the presence of an established industry could make market penetration difficult for a new technology. Therefore, the Second Theme of the thesis focuses on the ability of lignocellulosic ethanol produced from woody biomass to compete with conventional ethanol – namely corn ethanol in the North American marketplace. Production costs do not necessarily have to be the same for all ethanol producers, but capital will inherently flow to those companies and technologies that provide the greatest margins and return on investment. Therefore, this theme requires analysis of each of the primary contributors to production cost, from feedstock to capital to enzymes, for lignocellulosic ethanol relative to corn ethanol. This assessment is critical in order to identify targets for lignocellulosic ethanol production costs and where lignocellulosic ethanol has competitive advantages and disadvantages relative to the established competitor product in the marketplace. 61  Industrial facilities typically benefit from notable economies-of-scale, wherein per unit production costs decrease as the scale of the production facility increases. When it comes to bioenergy, these production facility economies-of-scale are often contrasted to feedstock diseconomies-of-scale: as a facility increases in scale, locally-sourced biomass (i.e., transported by truck) costs increase as transportation distance increases. However, bioenergy feedstocks can also be transported longer distances by ship and rail, reducing the influence of feedstock diseconomies-of-scale. Given the often large production cost benefits obtained from production facility economies-of-scale, the Third Theme of the thesis is a biomass feedstock logistics assessment that seeks to determine how these economies-of-scale for lignocellulosic biofuel production can be maximized through various feedstock supply system designs and operations. The feedstock form, type, and delivery mode, designed to maximize economies-of-scale, has implications for preferred siting location and technology selection (namely biochemical vs. thermochemical platforms).  These siting and technology preferences, based upon feedstock  characteristics, are also included within the theme, along with an assessment of the existing feedstock logistics systems for large commodity-based production industries as potential models for lignocellulosic biofuel operations. The optimization of logistics systems that enable production facility economies-ofscale to be realized will be essential for the economically competitive production of lignocellulosic ethanol. Lignocellulosic ethanol producers, typically technology developers at the current stage in commercial development, will site their facilities in locations that result in the most profitable enterprise with the lowest risk. In an open global market exempt of import and export tariffs and government policy supporting preferred production locations (e.g., subsidization of domestically produced biofuels only), the most economically competitive lignocellulosic ethanol producers will be the ones that supply the market demand at the lowest possible MESP. Since there is no differentiation in fuel properties or characteristics from one ethanol shipment to the next, price at point of sale is the sole determining factor in selecting a supplier. Investment will flow to those producers that have a sustainable competitive advantage in production costs. In this regard, optimal siting that minimizes the combination of feedstock cost, capital 62  cost, operating cost, and product delivery cost will be a fundamental competitive advantage for lignocellulosic ethanol producers. Building upon the production cost assessment of Theme Two and the economies-of-scale maximization in Theme Three, the Fourth Theme assesses the importance of siting on the ability of Canadian producers to supply the lowest cost fuel possible and compete with foreign producers. With a continued focus on woody biomass, potential facility sites in British Columbia and Ontario (i.e. domestic sites) are contrasted with a site in Brazil for production of lignocellulosic ethanol to supply Canadian fuel markets. Model facilities use identical technology platforms and have a capacity that maximizes economies-of-scale, as defined in Theme Three.  This Fourth Theme has policy  implications regarding the support of biofuels by governments – domestic production vs. domestic consumption. The relative cost of production at multiple sites from different woody feedstocks will be valuable in determining the viability of Canadian-based enterprises and their ability to compete globally. Lignocellulosic ethanol facilities are anticipated to be very large capital projects and as such, would typically be financed through a large amount of debt. The cost of capital, or the return expected by financiers, will be proportional to the risk profile of the operation. However, as an emerging industry, the market and operating risks of lignocellulosic ethanol production are not fully understood. Previous studies have typically assumed a standard average cost of capital (ACC) based upon existing, technologically and operationally proven industries. While it may be advantageous for lignocellulosic ethanol producers to assume a standard ACC, the risk profile of this emerging industry is likely to be different than that of either conventional ethanol or other transportation fuel processing facilities. Therefore, the Fifth Theme of the thesis focuses on the risk profile of lignocellulosic ethanol production from woody biomass in Canada, as dictated by theoretical facility gross processing margin volatility, and utilizes this profile to determine a minimum average rate of return (MARR) demanded by facility financiers. For a lignocellulosic ethanol production facility that utilizes woody biomass as feedstock, the raw material price is not directly linked with transportation fuel markets, whereas the revenue from ethanol sales is dictated by liquid fuel market prices. This is in stark contrast to independent oil refineries, in which the prices of both the raw material (oil) and product (gasoline, diesel, etc.) are 63  dictated by transportation fuel markets. Clearly, this difference will affect gross processing margin and hence, MARR.  Identifying the operation risk profile and the MARR provides justification for  assumptions in financing scenarios and can also be used to identify the maximum cost payable by lignocellulosic ethanol producers for process inputs such as feedstocks. This brings the thesis full circle by addressing the ability to compete for feedstock, given gross processing margin and potential revenue. These five themes are all based around the common elements of enterprise competitiveness and risk management. With the objective of identifying realistic structural, operating, and financial conditions for economically viable lignocellulosic ethanol production from Canada’s forest sector, this thesis sheds light on the commercialization hurdles that must be overcome and where resources must be allocated in order to maximize enterprise profitability. The three primary operating risks for lignocellulosic ethanol producers – feedstock, technology, and markets – are dealt with throughout the research, with recommendations on managing and/or reducing risks provided. Identification of optimal, but realistic, commercial operating conditions can assist in the successful commercialization of lignocellulosic ethanol.  1.4 Research Approach This thesis assesses the economic viability and competitiveness of forest-based lignocellulosic ethanol production in Canada and provides strategies to address the risks involved. This assessment of competitiveness has five major themes as identified above. Competitiveness is defined as the ability to compete with other options from a financial perspective – the ability to compete with other bioenergy options for feedstock purchase, the ability to gain market share in the fuel ethanol market currently dominated by conventional ethanol, the ability to reduce production costs by optimal scaling and siting of facilities, and the ability to attract investment by providing financial returns commensurate with the market and operating risks. Competitiveness is a relative concept and therefore, in order to determine the competitiveness for the purposes of this thesis, comparisons between lignocellulosic ethanol production and other potential investment, fuel, and product options must be undertaken. While the absolute financial performance of lignocellulosic ethanol production, and the identification of opportunities to  64  improve that performance, is an important component of the thesis, it is the performance relative to other feedstock consumers, technologies, and end products that is particularly unique to the thesis. Investment does not occur in isolation and it must be recognized from the start that investment will flow to those opportunities that provide the highest financial return, adjusted for risk. This thesis identifies how lignocellulosic ethanol production from Canada’s forest resources measures up relative to other options and the conditions of operation that will enable it to be competitive. The comparison of lignocellulosic ethanol production relative to other options was conducted using techno-economic spreadsheet models. The reasons for using spreadsheets, rather than a dynamic, complex techno-economic model developed using a program such as Aspen®, were three-fold. First, lignocellulosic ethanol is currently not being produced at commercial scale and the industry is in its infancy. Production therefore lacks a proven business model and no facility design standards or heuristics are in place. Therefore, analyses that utilize a complex dynamic process model that requires selection of individual pieces of equipment with a range of assumptions on performance could be rendered obsolete in a short period of time and assumptions on equipment pricing and performance would need constant updating. In addition, the results would be specific to that individual facility design and could therefore lack applicability to alternative facility designs. By utilizing a spreadsheet and focusing on key metrics such as installed cost per unit of yearly capacity rather than individual pieces of equipment, the results will remain relevant for longer. Secondly, complexity does not necessarily equate to greater accuracy when there is great uncertainty in commercial production performance and design, as is the situation with lignocellulosic ethanol. Spreadsheets enable rapid modification of inputs and general production models can be utilized as a basis for more complex models as commercialization occurs. Improvements in technology, which are likely for a rapidly developing field such as lignocellulosic ethanol, can easily be integrated without redesigning the entire model. Finally, spreadsheets are highly accessible and all assumptions can be clearly explained to not only reviewers, but potentially other users. They can be readily replicated and the inputs challenged by peers. This is notably different than a complex, black box model that requires thousands of lines of code. 65  The research described in this thesis is not basic, lab research. It is techno-economic modelling and is therefore dependent upon data from other sources as inputs. Most data were drawn from published peer reviewed journals, white papers and reports from government and NGOs, and government and international organization databases. Since the focus of the thesis is commercialization of lignocellulosic ethanol, industry press releases and presentations were also considered when relevant. In many cases, the research benefited from association with the Forest Products Biotechnology and Bioenergy Group (FPB) in the Faculty of Forestry at the University of British Columbia. FPB has a 25-year history of research on lignocellulosic ethanol production from woody biomass and large amounts of data and results are available (published and unpublished). Models were verified for accuracy and relevancy by comparing results to previous studies, but also adjusting the models to compare outputs against real-world performance. An example of this verification is modification of the lignocellulosic ethanol model to a corn ethanol production model and comparison of model outputs with data from corn ethanol mill surveys from the USDA. Since the availability and affordability of biomass feedstock is a critical prerequisite for the planning and establishment of any bioenergy facility, the ability of lignocellulosic ethanol facilities to compete for feedstock with other wood-based bioenergy options is a logical starting point for the assessment. This First Theme required the creation of spreadsheet techno-economic models representing not only a lignocellulosic ethanol facility, but also FT liquids (thermochemical biofuels platform), wood pellets, combined heat and power (CHP), and stand-alone biopower facilities. The same basic model design was used for all technology/product options, with modifications to reflect the unique processing and operation requirements of each facility. The primary metric for comparison was the 20-year internal rate of return (IRR) of the facilities, with the premise of the analysis being that facilities with a higher IRR can potentially pay more for feedstock and therefore outbid facilities with a lower IRR, leaving them without feedstock (or investment). Income statements projected out 20 years were created for calculation of the IRR. Extensive sensitivity analyses on feedstock type and cost, facility scale, and product revenue were used to determine preferable operating conditions for lignocellulosic ethanol relative to other 66  bioenergy products. Risk in this First Theme was assessed by reviewing historical price volatility of primary products for each facility. Overall, this First Theme provided an indication of expected financial returns for bioenergy projects at existing market prices. However, it must be noted that due to the lack of commercial lignocellulosic ethanol and FT liquids production facilities, there is significant uncertainty in model inputs on capital costs, operating costs, and long-run feedstock costs. This contrasts with pellet and CHP facility model outputs, where commercial data are available and model results could be considered to better reflect real-world performance. Therefore, the key focus, as is the case throughout the thesis, must be identification of conditions that will make lignocellulosic ethanol competitive. While the First Theme focuses on the feedstock aspect of competition, the Second Theme addresses the opposite end of the of the lignocellulosic ethanol production business – competition in the ethanol fuel market. A similar spreadsheet techno-economic model was utilized to determine current lignocellulosic ethanol production cost. However, this was also used as a starting point to identify the cost reductions that are necessary for lignocellulosic ethanol to compete with corn ethanol on a cost-ofproduction basis. Instead of focusing on where production costs are currently, the emphasis shifts to where they need to be. Each primary contributor to production cost was compared for lignocellulosic ethanol relative to corn ethanol, with the data for the latter sourced from surveys and reports on existing facilities. The premise for this theme is that as long as fertile land is available for additional corn production, investment will flow to corn ethanol plants rather than lignocellulosic ethanol plants if the returns are better and the risk is lower. This assumes a lack of specific government mandates. In addition, as this chapter has already discussed, substantial volumes of corn are still used for purposes other than ethanol and could potentially be redirected to fuel production. Essentially, while the economic attractiveness of lignocellulosic ethanol production is strongly influenced by ethanol market prices and high market prices could potentially support a commercial industry, it may be difficult to attract financing or investment if conventional ethanol production provides much greater returns and lower risk. As a new industry, lignocellulosic ethanol production will inherently have a higher technology and operating risk profile and therefore production costs may need to be equal to or lower than corn ethanol in order for the 67  newer fuel to gain market share. Included in the Second Theme analysis were sensitivity analyses on facility design and operation, including feedstock selection, scale, process residency times, enzyme performance, pretreatment technology, and co-product production. The emphasis for these sensitivity analyses was the potential to reduce production costs and improve competitiveness relative to corn ethanol for all production cost components. A progress ratio analysis was also utilized to project production costs in 2020, although it should be noted that production cost projections out 8-10 years are accompanied with a notable uncertainty. The approach for the Third Theme differs from the previous two themes in that techno-economic facility models were not the primary tools used. Instead, the focus of the theme on logistics and scaling required the design and use of a spreadsheet logistics model for feedstock delivery. Having established the importance of facility scale on lignocellulosic ethanol production cost (as is the case with all commodity processing facilities), the purpose of this theme was to determine how economies-of-scale can be maximized and thus inform the subsequent themes of the thesis. In essence, the Third Theme identifies the maximum scale of facility that can reasonably be pursued for commercial development. While previous logistics analyses have examined the concept of optimal scale from a capital cost perspective, this assessment assumed that logistics are the limiting factor in scale maximization and used existing industrial logistics structures (and delivery patterns) to determine this maximum facility scale for various biofuel production scenarios. The primary determinant of maximum scale was the number of deliveries possible, whether by trucks, rail, or ship. This is a unique research approach to logistics and scaling of lignocellulosic biofuel production facilities that relies upon real-world experience and proof of viability. The use of a spreadsheet-based model instead of a dynamic model is appropriate, since number of deliveries can be calculated using the simpler approach. This is a strategic level assessment, rather than an operational level assessment; the latter would require the use of a dynamic model to track and run analysis on actual truck/train/ship movements. Costs were not included in the analysis due to the scale of facilities and the inherent large variability in cost of sourcing large volumes of biomass (as necessitated by the large facility scale) from diverse regions around the world. 68  The Fourth Theme is largely focused on the importance of siting on facility economic performance. Under a free-trade regime, ethanol blenders and fuel distributors will seek to fulfill market demand by purchasing ethanol at the lowest price possible, regardless of site or country of production. The lignocellulosic ethanol production spreadsheet techno-economic model developed for the First and Second Themes was utilized and modified for the Fourth Theme. However, instead of using country averages for the model, this analysis was highly site-specific and inputs were dictated by local conditions. Two Canadian sites (one in Ontario, one in British Columbia) were compared to a site in Brazil, with the analysis emphasizing the identification of the primary competitive advantages and disadvantages for locating a facility at a given site. This analysis continued the comparative approach considered in the first three themes, but focuses on geographic location rather than technology, product, or facility design. It is largely a theoretical study asking the question, “If one were to build a lignocellulosic ethanol facility utilizing woody biomass that achieved economies-of-scale to make it competitive with conventional ethanol and gasoline, would that facility be built in Canada?” The production cost comparison is to Brazilian eucalyptus feedstock.  Whether a Brazilian company would commercially pursue ethanol  production from eucalyptus, given the large sugarcane ethanol opportunity in Brazil, is highly questionable from a real world perspective.  However, the question of location selection, while  theoretical, is very relevant within the broader thesis question of commercial competitiveness. The product (lignocellulosic ethanol from woody feedstocks) and facility scale (as determined by the analysis in Theme 3) remained the same for all scenarios in the siting assessment, but a host of other model inputs were determined by site-specific conditions.  These included feedstock type and characteristics,  construction cost (and hence total capital cost), cost of capital, labour cost, other operating costs, and tax rate.  Other than feedstock properties, most of these site-specific variables are policy and macro-  economically related, since theoretically, the exact same facility could be built at any site. In addition, any technical advances made at one site could be transferred to other facilities and therefore technical improvements may not always provide a sustainable competitive advantage for one site. The trucking logistics model developed in Theme 3 was modified to be used in this analysis and was accompanied by a 69  complete trucking cost techno-economic model to determine delivered cost of feedstock for each scenario. The primary point of comparison for this siting analysis was the delivered MESP for two Canadian markets, since delivery costs for the ethanol product needed to be included to provide a true comparison. All four previously discussed themes required the use of engineering economics/project management heuristics and assumptions from previous studies to select an average cost of capital (ACC) (with modifications to reflect local investment risk for the Fourth Theme). However, since there is very limited commercial experience with large-scale production of lignocellulosic ethanol and operating risks are not fully understood, ACC heuristics that are used for established commodity processing industries may not be applicable to the lignocellulosic ethanol industry. The Fifth Theme seeks to identify an appropriate ACC that is commensurate with the non-technology operating risk of lignocellulosic ethanol production, as dictated by gross processing margin volatility. The techno-economic model from Themes 1, 2 and 4 was utilized to determine gross processing margin, with the inputs and facility design based upon the results from Theme 4. The combined logistics and techno-economic trucking model utilized in Theme 4 was also used in this analysis. While feedstock cost is one side of the gross processing margin calculation, product revenue is also required and historical ethanol rack prices were used for this calculation. Gross processing margin volatility was determined over a 35 year period based upon historical ethanol price and feedstock price. The latter was adjusted to account for inflation over time and by the price of oil inputs (namely diesel) used in feedstock harvest and delivery. This gross processing margin volatility (beta) is a primary indicator of investment risk and served as an input to the capital asset pricing model (CAPM), a standard financial equation used to determine the minimum acceptable rate of return (MARR) demanded by financiers based upon the risk profile of an investment. Historical returns for risk free investments and the stock market, combined with data on the beta of independent oil refineries (as a comparative industry to lignocellulosic ethanol production), also served as inputs for CAPM.  All data for this analysis were sourced from government and stock exchange databases.  Technology risk, which is much more difficult to quantify and is significantly more subjective, was 70  excluded from the analysis because financing exceeding $1 B is unlikely to be granted if a large amount of technology risk still exists. By combining information on gross processing margin with historical rack ethanol prices, it was also possible to determine how much could be paid for feedstock historically. This brings the thesis and analysis full circle by showing the ability of lignocellulosic ethanol producers to compete for feedstock historically.  71  2  2.1  ECONOMIC PERFORMANCE AND RISK OF LIGNOCELLULOSIC ETHANOL RELATIVE TO OTHER WOOD-BASED BIOENERGY OPTIONS Introduction A significant advantage that biomass has over other renewable energy sources is that it can be  used to produce a variety of forms of energy, including electricity, high-efficiency heat, and liquid transportation fuels, all of which can be integrated within the existing fossil-fuel dominated energy infrastructure. This flexibility presents a challenge for those wishing to enter this sector, as decisions must be made today about the potential markets and values of different energy options over the life of the facility. Biomass combined heat and power (CHP) and wood pellet production stand out as successful bioenergy industries due to both the increasingly large number of commercial facilities that exist and the overall rapid rate of growth in total output [308,309]. However, substantial investment in research & development and commercialization of lignocellulosic biofuels could lead to the successful and profitable production of liquid biofuels from lignocellulosic biomass, including lignocellulosic ethanol and FT liquids. Multiple bioenergy options for forest feedstocks would inherently lead to competition for these raw materials. This chapter focuses on the question of whether lignocellulosic ethanol will be able to compete with other bioenergy options for feedstock and which feedstock-technology-product chains will be the most profitable relative to risk.  2.2  Biomass Conversion and Competition 2.2.1  Biopower and CHP  In North America, biomass is one of the largest sources of renewable power, providing 8.2 million MWh each year in Canada [310] and 55 million MWh in the U.S. [311]. Proportionally, however, these figures are small, representing approximately 1.3% of total production in both countries. The vast majority of biopower generation in the U.S. and Canada is by the forest sector, with 80% of the wood-to-energy sourced from indirect sources such as the burning of black liquor (a lignin-dominated byproduct of wood pulp production) and sawmill residues [312]. 72  Boilers are the most widely utilized technology for electricity generation, where the heat of combustion is used to produce high pressure steam. The two primary technologies for combustion of biomass are grates and fluidized beds, with both options providing significant flexibility in biomass feedstock properties and the ability to co-fire with coal [308]. There are four main types of grates, with progressively higher carbon burnout efficiency and combustion control, including stationary sloping, travelling, reciprocating and vibrating [313]. Fluidized beds are of three main types including, bubbling, circulating, and entrained-flow, with the latter requiring very fine particles [308,314]. Grates were historically used for all solid fuel combustion, but the introduction of other technologies, including powder injection suspension-fired burners for coal, has meant grates are now largely reserved for biomass and municipal solid waste (MSW) fuel [313]. While production of biopower at a scale smaller than 2 MWe is technically feasible using steam generation, large economies-of-scale and regulations regarding operating personnel (e.g., power engineer on-site at all times) typically means that it is not economically viable to operate in many jurisdictions. Options for biopower generation at scales less than 2 MWe include Organic Rankine Cycle (ORC) and gasification combined with an internal combustion engine (ICE) [315]. Electricity is produced in ORC systems by evaporation and superheating of an organic fluid which runs a turbine connected to a generator [316]. Gasification involves the conversion of solid or liquid biomass into a synthetic gas (syngas) composed primarily of hydrogen (H 2 ) and carbon monoxide (CO), with notable amounts of methane (CH 4 ) and CO 2 [317,318]. Specially-designed or modified gas engines can operate on this syngas [319]. Many projects also utilize ‘off-the-shelf’ diesel engines, with syngas co-fired with diesel fuel in ratios of up to 4:1 (syngas:diesel) [320]. Electricity produced from biomass must be competitive with fossil fuel-based generation (e.g., coal-fired, natural gas), nuclear power, and renewables (e.g., wind, solar PV, solar thermal, geothermal, hydroelectric). Modern biomass power generation facilities typically range in net efficiency from 2035%, with larger units at the higher end of this spectrum [210,321]. To increase overall plant efficiency, combined heat and power (CHP) systems capable of capturing heat as well as electricity can be 73  employed, provided that a use for the heat can be found [308]. Biomass competes with coal, peat, and natural gas in fuel-flexible power and CHP applications, while also competing in the heating market with natural gas in gas-grid connected areas, and oil/propane and electrical heat in off-grid areas. 2.2.2  Wood Pellets  Wood pellets are one of the most common biomass-based fuels – an internationally-traded commodity utilized for residential and district heating, in combined heat and power systems, and potentially as feedstock for advanced lignocellulosic liquid biofuels (further details are provided in Chapter 4). Wood pellets are typically produced from sawdust and shavings generated during lumber production. Particles are dried to <10% moisture content (MC) and reduced in size using a hammer mill. Pellets are typically 6-8 mm in diameter and <38 mm long, and are formed using heat and extrusion through a circular pellet die [322]. Pellets produced from woody biomass require no additional binders to retain their structural integrity, due to the melting and re-hardening of lignin (which functions as the pellet binder) during production [309]. All pellet production equipment, including hammer and pellet mills, dryers, and coolers are off-the-shelf, with several manufacturers competing in the marketplace [309]. Wood pellets can be fired in coal power plants at rates of 10-100%, depending upon the boiler technology employed [323]. Potential for growth is huge; a worldwide 10% co-firing rate, based on about 8,200 TWh of generation, would require 579 Mt of wood pellets (assuming 30% conversion efficiency and 17 GJ t-1 pellets), compared to 10 Mt of wood pellet demand in 2007 [324]. 2.2.3  Transportation Biofuels  As previously discussed in Chapter 1, advanced liquid transportation biofuels produced from lignocellulosic biomass have received significant investment, particularly in the U.S. where an emphasis on transportation fuel energy security has driven government support.  The U.S. DOE announced  investments of up to US$365 million in six commercial biorefineries in 2006, US$200 million in seven small-scale facilities in 2008, and US$564 million in 14 pilot-scale and four demonstration-scale projects in 2009 [234,325].  74  The biochemical and thermochemical approaches are the two primary platforms for conversion of solid biomass to liquid biofuels. Biochemical conversion, the primary focus of this thesis, involves pretreatment (e.g., steam, organosolv, and ammonia fibre explosion), which enhances fractionation and recovery of the hemicellulose and lignin streams, and increases accessibility of the cellulase enzymes to the cellulosic component of the biomass. This is followed by hydrolysis, carried out by either acid or by cellulase enzymes sourced from microorganisms such as Trichoderma reesei, which cleaves the β-1-4 glycosidic bonds of cellulose to release individual glucose monomers. The glucose, and potentially the other 5 and 6-carbon sugars of hemicellulose, including xylose, galactose, mannose and arabinose, can be fermented to ethanol or to other alcohols such as butanol using organisms such as  yeasts (e.g.  Saccharomyces cerevisiae) or bacteria (e.g. Zymomonas mobilis) [147]. Biochemical conversion is projected to produce 300-340 L ethanol per bone dry tonne (bdt) of feedstock, equivalent to approximately 7.2 GJ bdt-1 [326]. Several competing technologies and process pathways exist for thermochemical conversion of biomass to liquid biofuels, including gasification of biomass (as described above) [317]. Syngas of high purity, often achieved through filtration and scrubbing processes following gasification, can be subjected to catalytic reformation to liquid fuels in the Fischer-Tropsch (FT) synthesis reaction. This reaction is catalyzed by Group VIII transition metal oxides – with cobalt and iron dominating previous experience – supported by large surface area bases such as zeolites. The reaction temperature (150-300°C), pressure, and catalyst used strongly affect the chain length and type of chemical synthesized, with lower temperature favouring desirable longer-chain alkanes but with a lower yield and reaction rate, while high pressure promotes longer-chain alkanes with a higher yield and reaction rate [327,328]. In addition to alkanes, FT synthesis can be used to produce ethanol and mixed alcohols [329]. Thermochemical conversion could deliver liquid fuels at a yield of 180-200 L bdt-1 of feedstock, with an energy output value as high as 7.2 GJ bdt-1 [46,150].  75  2.3  Relative Value Given the limited quantities of biomass available in any given region, choices must be made  about which product and conversion technology will yield the greatest economic benefit for a project developer. Historical pricing on fossil fuel equivalents can be used to determine the attractiveness of various markets and assess the competitiveness of biomass relative to other fuels and energy sources. A comparison of fossil fuel pricing in the U.S. on an energy unit (GJ) basis from 1976-2010 highlights the differences in both relative pricing and volatility of coal, natural gas, and oil (Figure 2.1) [330,331,332]. Coal is priced free-on-board (FOB), natural gas is priced at wellhead, and oil is a composite refiner acquisition cost (including domestic and imports). Figure 2.1 Nominal U.S. pricing of primary fossil fuels  Over the past 35 years, oil has commanded an average price premium of 260% over coal and a 76% price premium over natural gas on an energy content basis (as shown in Figure 2.1). This implies a premium paid for the energy content of transportation fuel compared to feedstocks principally used for stationary energy applications. Recent natural gas prices, extending below $2.50 GJ-1, have created an even greater divergence between oil and gas prices [333]. It terms of final products and assuming a gasoline to motive power efficiency of 17.5% [334], natural gas to heat efficiency of 85% [335,336] and a coal to useful power efficiency of 30% [321], transportation has been afforded a 581-1730% price premium over electricity and a 756-2252% price premium over heat from 1978-2010 [337]. Clearly,  76  transportation is attractive as a target market if product value is the sole metric; however, profitability and processing margin volatility are the true tests of attractiveness. This chapter provides a high level comparison of the competitiveness of advanced lignocellulosic biofuels - ethanol and FT liquids - relative to other forest-based bioenergy products, based on investment return and volatility of that return.  2.4  Methodology Facility scale, feedstock selection, and product revenue (as dictated by target market) were  chosen as primary variables for investigation.  A hypothetical lignocellulosic ethanol facility was  compared with facilities that could produce FT liquids, wood pellets, electricity, or both electricity and heat (CHP). Financial models were created for all forest biomass conversion facilities using spreadsheet software. The time period for the comparison study was set at 20 years. An installed capital cost per unit output – $ L-1 (yearly capacity) for ethanol and FT liquids, $ MW-1 for power and CHP, and $ t-1 (yearly capacity) for wood pellets – was estimated based upon previously installed and announced projects. This enabled comparisons across technology types and ensured relevancy regardless of specific equipment selection. All of the base case assumptions are listed in Table 2.1. As with all the models described in this thesis, every input could potentially be modified from the research assumptions for a particular real-world case and will inherently result in challenges to input assumptions. It was the goal of this research to use assumptions that were reasonable expectations for real-world performance and an extensive review of previous academic studies, industry surveys, and industry/government reports was used to identify reasonable assumptions and model inputs. The inputs are ‘defendable’ and the models were designed in such a way that they could be readily updated to reflect advances in technology or changes in market dynamics and costs. Financing for all of the projects was assumed to be 100% debt, with cost of capital assumed to be 8% for calculation of interest and taxes. The base case scenario was assumed to be 200,000 bdt yr-1 feedstock input, which was chosen due to the availability of data on existing or planned facilities of this scale for all conversion technologies. It also enabled a middle ground comparison between technologies that have traditionally operated at a scale less than 200,000 bdt yr-1 (e.g., wood 77  pellets), and those that are projected to operate at a much larger scale (e.g., FT liquids) (see Chapter 4 for facility scale analyses). Variables such as plant operating factor (uptime) and plant capacity factor were assumed to be the same for all technologies to permit a focus on capital costs, conversion efficiency, and relative product value. Capital cost allowance rate (depreciation) was assumed to be 4% for income tax purposes. Operating costs, excluding feedstock, were divided into labour, maintenance and repairs, stores/supplies (other than feedstock), insurance/permits, property tax, and supervision.  Other than  labour, these operating costs were assumed to be a percentage of installed cost. Base case product revenues were based upon proxy product reference prices (e.g., diesel fuel for FT liquids). All operating expenses and revenues were increased year over year by 2% to account for inflation. Feedstock, for the base case scenario, was assumed to be 50% moisture content chipped whole logs. Although not the most likely feedstock for wood pellets, the use of standing timber allowed for realistic sensitivity analyses. Douglas fir (Pseudotsuga menziesii) was used as representative species. Harvest and whole-tree chipping cost was assumed to be $85 bdt-1 [338,339,340,341,342] and transportation cost was determined using a trucking model discussed in detail in Chapter 5. For the base case, delivered feedstock cost was determined to be $95.87 bdt-1. An additional on-site grinding cost of $5 bdt-1 was assumed for FT liquids (entrained flow reactor) and pellets and whole logs were assumed to be debarked for ethanol production. Lower heating value (LHV) was assumed to be 18.5 GJ bdt-1 for white wood and 21 GJ bdt-1 for bark. The pretreatment assumed for ethanol production was steam and the facility was considered to be revenue neutral (self-sufficient) in electricity and heat. Heat was considered a co-product of FT liquids. All prices were adjusted for inflation and currency exchange to 2007 U.S. dollars. The sensitivity analysis component of the assessment focused on facility scale, feedstock type, and product pricing. Scaling factors derived from the literature were used to adjust capital costs for various scales, ranging from 100,000 bdt yr-1 input to 2 M bdt yr-1. Calculation of scaling factors is detailed in section 2.5.1 and the scaling factors themselves are listed in Table 2.1. Feedstock costs for scaling were estimated based upon a case study project site of Williams Lake, British Columbia. However, the results could be applied to similar sites in other jurisdictions. Cost of harvest and whole78  tree chipping was assumed to be the same cost as the base case for all scales. At the base case 200,000 bdt yr-1 scale, several types of forest-based biomass were considered for a feedstock comparison sensitivity including whole logs, pulp chips, sawdust/shavings, harvest residues (‘slash’), and hog fuel (largely bark and low-grade mill residues). Feedstock costs for residues were based upon previous studies and government and industry data [198,199,201,202,203]. Product yield from each type of feedstock was calculated  using  conversion  rate  results  from  the  literature  and  industrial  standards  [46,137,138,309,321,343,344]. Finally, historical volatility (standard deviation) of fossil fuel prices from 1976 to 2007 was used to determine potential revenue volatility of the modelled facilities at the base case scale [330,331,332,345,346]. The internal rate of return (IRR), the interest rate at which the net present value of cash flows is equal to zero, was used as the primary metric of comparison.  79  Table 2.1 Assumptions for base case 200,000 bdt yr-1 biomass processing facility models Facility Overview Plant Capacity  Power Only  CHP  Pelletsa  Ethanol  FT Liquids  41.8 MW e  41.8 MWe; 68 MW th Electrical: 30% (1541 kWh bdt-1) Thermal: 43.7% (7.67 MMBTU bdt-1) [321,343] 90% 95% $3000 kW e -1 [343]; 0.7 [210] $125,400,000  207,000 t yr-1  75.1 ML yr-1  42.1 ML yr-1  1.02 tonnes bdt-1  Ethanol: 321 L EtOH bdt-1; [137,138]  FT liquids: 180 L bdt-1; Heat: 25% (4.38 MMBTU bdt-1) [46]  90% 95% $175 t-1 (yearly capacity) 0.84 [352] $36,225,000  90% 95% $2.15 L-1 (yearly capacity) 0.65 [363] $161,500,000  90% 95% $4.15 L-1 (yearly capacity) [344]; 0.5 [344] $174,715,000  8% 20 years 4%  8% 20 years 4%  8% 20 years 4%  8% 20 years 4%  200,000  200,000  200,000  200,000  18.8  18.8  18.8  18.8  50  50  50  50  95.87  100.87  95.87  100.87  6  6  8  8  35  35  35  35  3% 1% 1% 0.75% 0.75% -  3% 1.5% 1% 0.75% 0.75% 112 kWh t-1 [354]; $0.064 kWh-1 [355] -  3% 2.0% 1% [353] 0.75% 0.75% Heat and electricity selfsufficient $2.20 kg-1 proteinc; 600 FPU g-1 protein [356,357,358]; 20 FPU g-1 cellulose  3% 4% 1% 0.75% 0.75% Heat and electricity selfsufficient -  $120 t-1  $0.59 L-1 [360];  $0.59 L-1 [361]; $3.00 MMBTU-1  Conversion Efficiency  30% (1541 kWh bdt-1) [321,343]  Plant Operating Factor Plant Capacity Factor Unit Installed Cost§  90% 95% $3000 kW e -1 [343] 0.7 [210] $125,400,000  Scaling Factorb Total Capital Cost (US$) Financing Interest Rate 8% Amortization Period 20 years Capital Cost 4% Allowance/Depreciation Feedstock Biomass consumption 200,000 (bdt yr-1) Biomass LHV (GJ bdt18.8 1 ; 10% bark) Moisture Content (wet 50 basis %) Feedstock Cost ($ bdt-1) 95.87 Non-Feedstock Operating Costs Labour (workers per 6 shift) Labour (incl. benefits) 35 ($ hr-1) Maintenance & repairs 3% Stores/supplies 1% Insurance/permits 1% Property Taxes 0.75% Supervision & overhead 0.75% Electricity -  Enzymes  -  -  Revenues Specific revenue (year 1)  $0.0572 kWh-1 [359]  $0.0572 kWh-1 [359]; $3.00MMBTU-1  a  Assumes feedstock is dried by combustion of additional feedstock and not natural gas Capital cost and scaling factors based upon facility review and literature (as indicated in the table)  b  c  Calculated using the equation 3.2 in Section 3.3.3 and assuming 20 FPU g-1 cellulose, 600 FPU g-1 protein, and contribution to MESP of $0.132 L-1 [362]. This is a figure consistent with protein for amylase, but high compared to estimate of Humbird et al. (2011) [363]. Should loading be 10 FPU g-1 cellulose, activity 600 FPU g-1 protein, and contribution to MESP be $0.132 L-1, protein would be $4.40 kg-1. This is consistent with Humbird et al. (2011) [363].  80  2.4.1  Facility Review  A review of installed and announced/planned projects of each technology enabled the calculation of scaling factors for each technology type, with 200,000 bdt yr-1 feedstock input used as the base for per unit capacity installed capital cost. The results are included in the ‘Facility Overview’ section of Table 2.1. The relative capital cost of these facilities to ‘nth’ facilities is addressed in the discussion. The advanced lignocellulosic biofuel facility capital cost data for both technology types (biochemical and thermochemical) is based upon planned projects only, with commercial facilities expected by the end of 2013 [347,348,349,350,351]. The majority of biopower plants reviewed were found to lie in the 35-55 MWe capacity range, significantly smaller than most coal-fired power plants (200-450 MWe per unit, with multiple units often operating at a single plant) [201,210]. While many of the newer pellet plants are in the 100,000-150,000 t yr-1 range, the largest pellet plants approach 1 Mt yr-1, with the largest concentration of these facilities in the U.S. southeast. These capacities are significantly larger than many of the older pellet plants in North America and Europe, which range from 2,000 -150,000 t yr-1 [309]. 2.4.2  Base Case  The assumptions for each base financial model are listed in Table 1. Biomass energy content, B wet , was calculated using the equation: 𝐵𝑤𝑒𝑡 = 𝐵𝑑𝑟𝑦 (1 − 1.1405𝑥)  (2.1)  Where x is the moisture content of the biomass (wet basis) and B dry is the energy content of bone dry biomass. This was based upon the combined specific and latent heat requirement to evaporate water in wet biomass of 2.6 MJ kg-1 H 2 O. Biomass was priced on a bone dry tonne basis, with wet feedstock priced according to dry content percentage. Total installed costs were calculated using the equation: 𝐶1 𝐶0  𝑀 𝑠 𝑀0  = � 1�  (2.2)  81  Where M 0 is the capacity of the base facility, M 1 is the capacity of the study facility, C 0 is the total installed cost of the base facility, C 1 is the total installed cost of the study facility, and s is the scaling factor. The base case specific revenue rates were based upon average U.S. wholesale prices for the same (e.g., electricity) or comparable [e.g. ultra low sulphur diesel (ULSD) for FT liquids] products in the year 2007. In the case of heat, industrial prices for delivered natural gas, was used as a general gauge for pricing, but heat pricing is going to be highly location-specific and the $3 MMBTU-1 is considered a general indicator. It is important to note that electricity, heat, pellet, and FT liquid revenues do not have bioenergy-specific subsidies included in the assumed sale price. However, the assumed sale price for ethanol is the average rack price for ethanol (largely corn-based) in 2007, which includes support from the U.S. federal government in the form of a $0.51 gal-1 ($0.135 L-1) blender’s credit and a $0.54 gal-1 ($0.143 L-1) ethanol import tariff [364]. The market price for pellets is strongly influenced by carbon taxes (which could be considered a subsidy) in Sweden, Denmark, and other EU nations [309].  2.5  Results When the base case financial results were compared, all facilities had incalculable IRR  (substantially negative) and significant 20-year net income losses (Table 2.2). Therefore, assumed specific revenue was adjusted to identify a price (specific revenue) that would permit achievement of an acceptable IRR of 15%. The resulting values are listed in Table 2.2. This artificially-set specific revenue was used for additional analyses on scale, feedstock, and revenue volatility and therefore results of these additional analyses represent relative impact on return (IRR) for the bioenergy options rather than absolute or real-world performance.  82  Table 2.2 Financial results for base case 200,000 bdt yr-1 biomass facilities Power Only Cost of Production (8% cost of capital) Net Income ($M)  $0.173 kWh-1  CHP -1  $0.173 kWh (electricity only)  Pellets  Ethanol  FT Liquids  $185 t-1  $0.94 L-1  $1.70 L-1  20-Year Results -$503 -$463 -$231 -$407 Modified Specific Revenue Requirements (First Year Rates, Increased 2% yearly)  -$754  20-year Breakeven (0% IRR)  $0.120 kWh-1  $0.115 kWh-1 (heat same)  $169 t-1  $0.79 L-1  $1.34 L-1  15% IRR  $0.178 kWh-1  $0.172 kWh-1 (heat same)  $197 t-1  $1.12 L-1  $1.96 L-1  311%  301%  164%  190%  332%  Price difference of 15% IRR case relative to base case (market)  2.5.1  Scale  Each facility was scaled up and down, from 100,000 bdt yr-1 to 2,000,000 bdt yr-1 biomass input, with the scaling factors presented in Table 2.1 used for capital costs. A scaling factor of 0.836 was used previously by Gallagher et al. [365] for corn ethanol production, but this very high compared to other industrial facilities. Nyguen and Prince [366] used a scaling factor of 0.7 for modelling sugarcane and sweet sorghum ethanol plants and Humbird et al. (2011) [363] provided scaling factors for a number of pieces of equipment within a cellulosic ethanol facility. Based upon the Humbird et al. (2011) [363] figures, 0.65 was assumed for lignocellulosic ethanol production. Kumar et al. (2003) used a scaling factor of 0.75 for power plants in Western Canada, but also conducted a sensitivity analysis of 0.6. Traditional scaling factors for larger capital energy projects are in this range so 0.7 was applied to the power and CHP facilities [210]. This was consistent with the biopower/CHP installed facility review. Mani et al. [367] used the scale factor of 0.6 for estimating the wood pellet plant capital cost from a base case of 45,000 t yr-1, although the 0.84 used here and by Sultana et al. [352] for pellet mills was more consistent with the pellet plant review survey. This higher scaling factor intuitively makes sense because pelleting mills are typically limited to a capacity of 5 t hr-1 capacity and plants above that output require multiple pelleting mills [309]. This results in relatively poor economies-of-scale. The relatively low  83  scale factor of 0.5 used for the FT liquids facility is based upon the engineering assessment of biomass-toliquid plants by Boerrigter for capacities from 1,000-5,000 bbl/d [344]. The capital investment and IRR relative to facility capacity are described in Figure 2.2 and a labour scaling factor of 0.5 was used to determine the number of workers required. Scaling factor clearly has a very large impact on the capital cost and IRR of the facilities and due to the variability of scaling factors for specific projects and uncertainty of scaling factors for non-commercial technologies, the results should only be considered a rough estimate. A lower scaling factor will favour facilities as scale is increased, while facilities with a high scaling factor (e.g., wood pellets) will perform poorly at a large scale.  84  Figure 2.2 Impact of facility scale on capital investment and internal rate of return  *Bubble size represents facility scale  85  2.5.2  Feedstock  The second sensitivity examined the impact of feedstock type on IRR, with the base case capacity of 200,000 bdt used for analysis. The primary feedstocks considered for bioenergy operations were whole logs (base case), sawdust/shavings, harvest residues, pulp chips, and hog fuel (a bark-dominated mill residue).  The facility capacities and conversion efficiency/yield were modified to reflect the  differing characteristics of the feedstock, which are listed in Table 3. Pellet prices were modified to reflect the relative value of bark inclusion in the pellet and differing energy content. The equation used to calculate pricing was: 𝑓 � 18.8  𝑃 = 𝑃𝑁 �  (2.3)  Where P is the market price of the study pellet, P N is the price of the base case pellet with 18.8 GJ t-1 energy content, and f is the energy content of the pellet in GJ t-1. A 5% price adder was applied to pellets with 1% bark or less. A 10% price discount was applied to harvest residues and a 20% price discount was applied to hog fuel to recognize the increased cost to power and CHP facilities for ash handling. Table 2.3 Feedstock and select conversion assumptions for feedstock sensitivity Property Moisture Content (w.b.) Bark Content Energy Content (Dry LHV) [370] Energy Content (Wet) Delivered Biomass Cost Pellet Pricing Ethanol Yieldb  Whole Logsa 50%  Sawdust/Shavings 25%  Harvest Residuesa 40%  Pulp Chipsa 45%  Hog Fuela 55%  10%c 18.8 GJ bdt-1  1% 18.5 GJ bdt-1  25% 19.1 GJ bdt-1  1% 18.5 GJ bdt-1  80% 20.5 GJ bdt-1  8.1 GJ t-1  13.2 GJ t-1  10.4 GJ t-1  9.0 GJ t-1  7.6 GJ t-1  $95 bdt-1  $35.00 bdt-1  $55 bdt-1  $70 bdt-1  $15 bdt-1  Base case $197 t-1 Ethanol: 321 L EtOH bdt-1 (70% of a theoretical 458 L bdt-1);  Premium $203t-1 Ethanol: 321 L EtOH bdt-1 (70% of a theoretical 458 L bdt-1);  10% deduction $180 t-1 Ethanol: 269 L EtOH bdt-1 (70% of a theoretical 385 L bdt-1);  Premium $203 t-1 Ethanol: 321 L EtOH bdt-1 (70% of a theoretical 458 L bdt-1);  20% deduction $172 t-1 Ethanol: 154 L EtOH bdt-1 (70% of a theoretical 220 L bdt-1);  a Whole logs, harvest residues, pulp chips, and hog fuel are assumed to be chipped to <2”. Additional preprocessing to sawdust particles (as applicable for pellets and FT liquids) assumed to cost $5 bdt-1-1. b Ethanol yield based upon [369] c Whole logs are assumed to be debarked for ethanol production  86  Ethanol yield was modified to reflect the bark and moisture content; yield decreases with a decrease in moisture content and an increase in bark content [137,368]. Bark has significantly higher lignin content (~50%) and hence, lower polysaccharide content, than white wood, which limits the amount of sugar available for fermentation [369,370]. The results of the feedstock analysis are presented in Figure 2.3. It is clear that the base case whole log feedstock is the highest cost and switching to harvest or mill residues results in an increase in IRR for all cases. Pellets have the greatest increase in IRR from feedstock switching, particularly when using sawdust/shavings for production. Figure 2.3 Impact of feedstock type on IRR from base case scenario  2.5.3  Revenue  The third sensitivity analysis examined the impact of product price and overall revenue on the IRR, with historical proxy prices serving as a basis for modelling range and volatility. Prices for U.S. free on board (FOB) coal, wellhead gas, refiner acquisition, retail gasoline and retail electricity were all used to compare volatility [330,331,332,345,346] as shown in Table 4. Retail gasoline prices and electricity rates were employed due to their long term availability; as long as profit margins were consistent throughout the period (reflecting the difference in wholesale and retail prices) the volatility results should be representative of relative shifts.  87  Table 2.4 Volatility of real prices for select fuels (2007 US$) Coal ($ GJ-1) Mean Standard Deviation %SD a  1.95 0.67 34.3%  Gas ($ GJ-1)  Oil ($ GJ-1)  3.95 1.52 38.6%  6.51 2.83 43.5%  Gasoline ($ GJ-1) 15.46 3.87 25.0%  Electricity ($ GJ-1)a 28.68 4.48 15.6%  Average (residential, commercial, industrial, transportation, other) retail prices used  The revenue obtained in the base case scenario of sawdust/shavings feedstock and a plant capacity of 200,000 bdt yr-1 was modified for each conversion technology to reflect the product price volatility. Gasoline was used a proxy for both ethanol and FT liquids volatility, while natural gas was used for wood pellets. A normal distribution was estimated, with prices within two standard deviations assumed to be representative of 95% of possible scenarios. When the impact on the IRR of the bioenergy facilities was compared (Figure 2.4), pellets were found to have the greatest range in IRR, followed by transportation fuels ethanol and FT liquids. Power was the most stable product due to low volatility in electricity markets. Again, the more relevant values are not the absolute IRR but rather how the relative changes in IRR’s compare. Figure 2.4 Impact of market volatility on IRR from base case scenario  *-25% indicates incalculable  88  2.6  Discussion The economic viability of biomass to energy projects is very sensitive to changes in scale,  feedstock type, and product revenue. It is apparent that lignocellulosic ethanol and the other bioenergy products assessed here will have a difficult time competing with conventional fossil fuels and alternatives at 2007 market prices for the base case scale of 200,000 bdt yr-1 when whole logs from dedicated bioenergy forest harvest are the primary feedstock. However, these results are highly dependent upon assumed product revenue and in many instances new renewable energy projects may receive revenue at above-market prices. In addition, the choice of whole logs as the base case feedstock resulted in higher operating costs than for all other feedstocks. Although dedicated forest harvest for bioenergy products permitted modelling of large-scale facilities and has been analyzed extensively in the literature in the past (e.g., [201]), it is not widely practiced commercially at present. The vast majority of bioenergy projects in operation at present rely upon residues from the primary wood products industry (e.g., lumber, pulp). Therefore, it is not surprising that all base case scenarios showed significantly negative IRR and 20-year net income at 2007 market prices. To reach an IRR of 15%, ethanol required a doubling of specific revenue to $1.12 L-1, while stand-alone power required a tripling to $0.178 kWh-1. This indicates that a lignocellulosic ethanol plant, at the scale of 75 ML yr-1, is closer to competitiveness with fossil fuels than is a 42 MWe biomass power plant. This is a dramatic finding, as it shows lignocellulosic ethanol significantly benefits from competing in a higher value market (transportation fuels) than the lower value electricity market. It was also clear that FT liquids are uncompetitive at smaller scales, such as the base case of 42.1 ML yr-1, which is consistent with the conclusions of previous studies [227,344]. The inclusion of heat as a co-product with power or with FT liquids helped improved the economic competitiveness of both processes, but the historically low heat revenue of $3 MMBTU-1 limited the revenue provided by heat. When considering options to reduce production cost and improve IRR for whole log feedstock, it was apparent that for bioenergy products other than pellets, increasing scale up to 2,000,000 bdt yr-1 will likely improve project economics. Pellets are the exception because they are highly sensitive to feedstock 89  costs but have relatively poor economies-of-scale. As scale is increased, higher feedstock costs, driven by greater transportation distance, overwhelm the capital cost benefits of a larger facility. Scaling and feedstock logistics are discussed in greater detail in Chapters 4 and 5, but from this analysis, it was clear that the economic performance of technologies with a low scaling factor (e.g., 0.5 for FT liquids) benefit significantly from increasing facility scale. In this case, the reduction in per unit capital cost is greater than the marginal increase in feedstock cost. The difference in scaling factors between technologies also significantly affects relative installed cost at different scales. While there was a notable lower installed capital cost for lignocellulosic ethanol compared to FT liquids at the base scale 200,000 bdt yr-1, increasing the yearly feedstock demand 10-fold resulted in a lower capital cost for FT liquids compared to lignocellulosic ethanol. Clearly, scaling factors have a very large impact on the expected capital cost and hence, economic viability of the facilities described in this work. The lack of any commercial advanced lignocellulosic biofuel facilities and a standardized technology approach means that the scaling factors employed here are estimates, derived from experience gained in operating conventional corn ethanol plants and coal/natural gas FT liquid plants. Verifying scaling factors is further hampered by the absence of large scale bioenergy plants; the world’s largest planned pellet mill will have a capacity of 1 M tons yr1  (907,000 t yr-1) [371] and the largest dedicated biomass power plant has a capacity of 240 MW [372].  Therefore, the results presented here should only be used as a general indicator of relative performance. Improved accuracy in scaling factors would be a worthwhile research endeavour in the future. The data also indicated that the impact of labour on IRR declined substantially as the scale increased.  It is  recognised that economies-of-scale not only benefit capital investment, but also operating costs such as labour [210]. While a goal of bioenergy might be job creation [373], small-scale bioenergy facilities will be hampered by high labour costs. Feedstock selection was shown to impact the IRR of the five considered technologies differently. Switching to hog fuel resulted in the largest increase in IRR for biopower, CHP, and FT liquids, highlighting the low cost of hog fuel per unit of energy output for these thermal processes. In 90  comparison, switching to sawdust/shavings resulted in the largest increase in IRR for wood pellets and ethanol. For wood pellets, this is consistent with existing industrial production, which is based largely on sawdust from sawmills and planer shavings. The large IRR gap between sawdust/shavings feedstock and other feedstocks indicates it may be difficult for Canadian pellet facilities to switch to other feedstocks and maintain profitability. Sawdust and shavings are the lowest cost whitewood feedstock considered here, which means they have a high ethanol yield potential and the lowest feedstock cost contribution to ethanol production cost. Surprisingly, despite the low yield, utilization of hog fuel resulted in the second greatest increase in IRR for ethanol production, driven by the low feedstock price. Although some investigation has already been conducted on the effect of bark inclusion on ethanol yield [369,374], little work has been done on very high percentage bark feedstocks and this needs further exploration before the assumed yield of 154 L bdt-1 can be confirmed. In addition, research on pretreatment and hydrolysis (for ethanol production) of dry feedstocks, such as shavings, could confirm the assumed yield in this work and determine how moisture content impacts yield. The revenue volatility sensitivity analysis indicated that pellets, followed by ethanol and FT liquids, are most susceptible to changes in market price. This sensitivity of pellets to revenue change is due, in part, to the high volatility of natural gas prices, the proxy competitor used for modelling pellet revenue, but also the relatively lower base revenue per tonne of feedstock compared to other products. It should be noted that the use of natural gas to gauge heat markets also has a major pitfall, in that demand for natural gas is also driven by the power and chemicals sectors, thereby reducing the correlation of natural gas and heat markets. In addition, heat revenue from a bioenergy facility will be highly dependent upon local conditions due to the inability to transport heat (in the form of hot water or thermal oil) long distances. The power and CHP options benefited from the low volatility of electricity prices, which additionally may be guaranteed through Feed-in tariff contracts or similar mechanisms, as employed in various jurisdictions [375,376].  All types of bioenergy facilities could benefit from long-term  agreements, which would reduce or eliminate their exposure to market volatility [377], with either  91  electrical utilities or large oil refiners/blenders.  Market volatility from month-to-month may also be  important due to seasonal demand shifts [378,379]. The purpose of this chapter was not to predict IRR for future facilities (many of which are negative at current market prices), but to compare bioenergy facilities for their ability to compete for feedstock and investment. In the future, the capital costs ‘nth’ facilities for liquid biofuels are expected to be notably lower than those facilities modelled here [146,148]. When considering unsubsidized current North American market rates for various energy products, lignocellulosic ethanol is not at a disadvantage relative to other bioenergy options. However, subsidies and above-market rates for other bioenergy options, including wood pellets and electricity, mean that lignocellulosic ethanol will have difficulty competing for feedstock (and investment) without a consistent subsidy regime for all bioenergy products relative to market prices.  2.7  Conclusion It is clear that advanced lignocellulosic biofuels ethanol and FT liquids will be challenged to  compete with fossil fuels gasoline and diesel at current market prices. They will also face significant competition from other bioenergy products in their ability to access forest-based biomass feedstocks. Dedicated biopower and CHP facilities have the advantage of being highly flexible in terms of the biomass feedstocks that can be used and do not require the same degree of feedstock consistency and quality as liquid biofuel production facilities. However, it is apparent that the investment return of both biochemical and thermochemical-based biofuel processing facilities can accommodate changes in feedstock to a greater degree than can wood pellet production facilities. This provides, at least from a financial perspective, enhanced feedstock risk management options and a reduced reliance on a single feedstock supply. While the accuracy of scaling factors for advanced lignocellulosic biofuel facilities is lower than that of facilities in the established biopower and wood pellet industries, FT liquid facilities clearly benefit more from an increase in scale than do biochemical/enzyme based ethanol production facilities. This means that for smaller investments (e.g., <$100 M), lignocellulosic ethanol is likely to be the preferred biofuel conversion platform. 92  While many biopower and CHP facilities utilizing forest-sourced biomass already exist in North America and Europe, it appears that projects in North America are only profitable when they receive above-market prices for the power they produce. The revenue per unit output required to reach 15% IRR is closer to existing market prices for advanced lignocellulosic ethanol facilities than it is for biopower plants. Therefore, it appears that market structure, selective policy support, and technology risk are enabling biopower to outperform advanced lignocellulosic biofuels in commercial implementation. A consistent pricing regime for bioenergy products, which takes into consideration relative market prices, would enable lignocellulosic ethanol to better compete for feedstock with other bioenergy options. Although the financial return of advanced lignocellulosic ethanol facilities will likely increase as scale increases, it is proving difficult to attract significant financing for larger energy projects without the experience of operating smaller, commercially-proven facilities. It is also likely that a sustained increase in transportation fuel prices will be required before small- to medium-scale advanced lignocellulosic ethanol facilities can operate in an economical manner and pave the way for investment in larger, more profitable biorefineries. Conventional ethanol is already a proven fuel, produced in large scale facilities, that competes with gasoline for market share and can provide a potential model for commercialization of lignocellulosic ethanol. The conventional ethanol industry may enable market access for lignocellulosic ethanol, present opportunities for technology transfer, and ease investors’ concerns about production and use of biofuels. However, conventional ethanol will also compete in the marketplace with lignocellulosic ethanol and this is the focus of the next chapter.  93  3 3.1  COMPETITION WITH CONVENTIONAL ETHANOL AND OPPORTUNITIES FOR COST REDUCTION Introduction Conventional ethanol has been rapidly adopted, initially as an oxygenate and then as a primary  transportation fuel, in the U.S., Brazil, Canada and several European nations. Total world ethanol production grew from 17.1 GL in 2000 to 73.9 GL in 2009, which is a 430% increase or 17.7% compound annual growth rate [380]. In addition to rising gasoline prices, this success has been largely driven by existing and historical government policy support, including the ProAlcool program of Brazil [95,262], and the Energy Policy (2005) and Energy Independence and Security (2007) Acts in the United States [270,304]. The justification for ongoing government support of ethanol has been based upon the three primary drivers of increasing domestic energy security [242,304], reducing greenhouse gas (GHG) emissions/criteria air (CA) pollutants [53,241], and supporting rural industries and communities [235]. Conventional ethanol, and the success it has achieved, serves as a basis for the introduction of lignocellulosic ethanol to the transportation fuel supply, but, as this chapter emphasizes and forms the basis for Theme 2, is also a competitor in the marketplace for advanced biofuels. In Chapter 2, the economic performance of lignocellulosic ethanol relative to other bioenergy options was modeled and analyzed. However, real world commercialization of lignocellulosic ethanol has been slow and the current stage of development is large demonstration, with several facilities operating in North America and Europe.  A first-of-kind commercial facility is expected to be in  operation by the end of 2012 [381]. This slow progress has been due to both large technological risk, large capital cost (driven by economies-of-scale), and the predicted poor economic performance of biorefineries in the short term [382,383].  Most of the current demonstration facilities have been  financially supported by significant direct investments from national or regional governments. Despite its likely better performance than conventional ethanol on environmental and energy balance criteria, advanced lignocellulosic ethanol must be able to compete with conventional biofuels economically.  With carbon legislation uncertain and the low likelihood of a high price on GHG 94  emissions, lignocellulosic biofuels must meet or exceed the economic performance of their conventional counterparts without the need for significant, preferential policy support. In this chapter, a comparison of the competitiveness of ethanol derived from lignocellulose with ethanol derived from corn is presented. The comparison includes an estimation of the opportunities for reductions in production costs, including feedstock, capital, and operational processing costs. Building upon the results of Chapter 2 (Theme 1), opportunities for production cost reduction will be examined to quantify the potential for increased competitiveness of lignocellulosic ethanol.  3.2  Methods The techno-economic spreadsheet model created for Theme 1 and presented in Chapter 2 was  used to determine the current production cost of lignocellulosic ethanol and the impact on total production cost of various cost-reducing strategies. The model included inputs on feedstock type and properties, facility scale, operating and capacity factors, installed cost, financing structure and operational considerations such as labour, enzymes, and supplies. The assumptions for the base case in this Theme are listed in Table 3.1. An emphasis was placed upon ethanol production from woody feedstocks, although these were also compared to corn stover. Modelled lignocellulosic ethanol production costs were compared to those of conventional ethanol, both current and projected for the year 2020. As previously discussed, use of models inherently leads to debate on inputs and assumptions. For this analysis, information on conventional ethanol production costs was derived from commercial ethanol plant surveys, research reports, and academic literature. The lignocellulosic ethanol model was verified for accuracy by adjusting variables to match those of the corn ethanol studies and production costs compared. The lignocellulosic ethanol model base case assumptions were sourced from a variety of publications – both academic and industry. Wherever possible, inputs and assumptions were crosschecked with other publications for accuracy. The progress ratio, which is the rate at which production costs decrease for each doubling of aggregate production [384], required for lignocellulosic ethanol to be competitive with corn ethanol by 2020, was calculated using projected production volumes. The U.S. ethanol mandates, as detailed in the 95  Energy Independence and Security Act (EISA) of 2007 [304], were used to define the expected production volumes for the period of 2010 to 2020. This progress ratio was compared to those defined in previous studies of the conventional ethanol industries [88,91]. Table 3.1 Assumptions for base case advanced lignocellulosic ethanol production cost Facility Overview Plant Capacity Ethanol Yield Plant Operating Factor Plant Capacity Factor Unit Installed Cost Total Capital Cost Financing Interest Rate Amortization Period Capital Cost Allowance/Depreciation Feedstock Biomass consumption Biomass LHV Cellulose/Hemicellulose/Lignin/Extractives & Ash Moisture Content Feedstock Cost Non-feedstock Operating Costs Labour (workers per shift) Labour (incl. benefits) Maintenance Stores/supplies Insurance/permits Property Taxes Management & Administration Electricity Enzymes  50 ML yr-1 250 L EtOH bdt-1 [137,138] 0.95 0.95 $2.50 L-1 yearly capacitya US$125 M 8% [363] 20 years 4% [386] 180,500 bdt yr-1 18.5 GJ bdt-1 42%/23%/29%/6% [121] 45% (wet basis) $ 50 bdt-1 [199,200,201] 6 [363] $35 hr-1 [210,363] b 3% [363] 2% [353,363] c 1% [353] 0.75% [387] d 0.55% [363] b 0.9 kWh L-1 [353]; $0.064 kWh-1 [355] $2.20 kg-1 proteine; 600 FPU g-1 protein [356,357,358]; 20 FPU g-1 cellulose [136,138,388]  a  Installed yearly capacity cost based upon a survey of announced cellulosic ethanol facilities, summarized in Chapter 2. A listing of projects was compiled by Bacovsky et al. (2010)[385]. b Total labour and management costs based upon Humbird et al. (2011) [363], with hourly wage based upon Kumar et al. (2003) [210] and adjusted for inflation c Stores and supplies includes process chemicals such as sulphuric acid and/or sulphur dioxide (pretreatment catalyst); sodium hydroxide (pH balance and lignin-cellulose separation); diammonium phosphate (yeast nutrient); sorbitol (yeast culturing). Adapted and based upon processes described in Humbird et al. (2011) [363] and Sassner et al. (2008) [353]. d Property taxes range significantly by jurisdiction. In British Columbia, heavy industry property taxes range from zero to 8.7% of assessed value [387]. It is assumed that the facility would be located at a lower property tax jurisdiction. e Calculated using the equation 3.2 in Section 3.3.3 and assuming 20 FPU g-1 cellulose, 600 FPU g-1 protein, and contribution to MESP of $0.132 L-1 [362]. This is a figure consistent with protein for amylase, but high compared to estimate of Humbird et al. (2011) [363]. Should loading be 10 FPU g-1 cellulose, activity 600 FPU g-1 protein, and contribution to MESP be $0.132 L-1, protein would be $4.40 kg-1. This is consistent with Humbird et al. (2011) [363].  Opportunities for production cost reductions were then identified and quantified. These included feedstock and pretreatment technology selection, enzymes, and capital costs, with components of the  96  latter including scaling and optimization of hydrolysis rate (and hence ethanol output). An emphasis was placed on those processing components that differed from conventional ethanol production, since cost reductions in equipment/operations shared by both would inherently be applied to both conventional and advanced lignocellulosic production schemes.  Co-products, including energy, were the final  consideration of ‘cost reduction’ in the form of production credits. All financial figures are listed in 2007 U.S. dollars, unless otherwise stated.  3.3  Results The results of the base case techno-economic model from Theme 2 are provided in Figure 3.1,  along with costs reported in the literature. Both U.S. corn and Brazilian sugarcane ethanol production costs are in the range of $0.30-0.40 L-1, while lignocellulosic ethanol is 250-300% higher at approximately $0.90 L-1.  Feedstock costs are the largest source of variation in reported costs for  conventional ethanol production. It should be noted that the facility scale, which strongly influences the capital and some processing costs, is not consistent for the cost comparison. The U.S. corn and Brazilian sugarcane results are based upon the average size of a facility in these countries of 150 ML yr-1 [88] and 365 ML yr-1 [91] respectively, while the lignocellulosic base case is much smaller at 50 ML yr-1. While this is clearly a small plant by industrial standards, this scale was chosen based both on the projects funded under the U.S. DOE’s commercial biorefinery program and those operating in other jurisdictions such as Denmark. Thus, this scale was intended to represent facilities currently being developed. The Brazilian Real – U.S. dollar exchange rate, which was assumed to be 2.3 for this study (2007), has a large impact on the results and the overall competitiveness of sugarcane ethanol in U.S. markets. This exchange rate has been highly volatile over time and therefore future comparisons to Brazilian cases must take this into consideration. Production costs for conventional (sugar- or starch-based) ethanol are expected to continue to decline, down to an overall production cost of about $0.22-0.25 L-1 by 2020. Thus advanced lignocellulosic ethanol will be, for some time, trying to play “catch-up” with this moving target of conventional ethanol costs [88,91].  97  Figure 3.1 2007 and future ethanol production costs from U.S. corn, Brazilian sugarcane, and U.S. lignocellulose feedstocks  Sources: [88,89,91,389]  3.3.1  Progress Ratio  According to EISA, cellulosic ethanol production schedule/blend mandate (Figure 3.2), 40 GL of ethanol must be produced in 2020 with a cumulative production increase over ten years (2010 to 2020) of 160 GL [304]. From a 2010 base figure of 0.38 GL, this is equivalent to 8.65 doublings. Based upon current and future production costs, the progress ratio which defines the cost reduction for each doubling of production volume can be calculated using the formula:  Q 1 = Q 0 • Nu  (3.1)  Where Q 1 is the future (or ending) cost of production, Q 0 is the current (or starting) cost of production, N is the progress ratio, and u is the number of doublings of production from the point 0 to point 1. A progress ratio of 0.5 would indicate a reduction of 50% in production cost for each doubling of production, whether due to technological learning, increases in scale, increases in efficiency, or other  98  factors [384]. The base case modelled for C 0 was the current cost of production of $0.904 L-1. The C 1 future target cost of production used was $0.249 L-1, the projected cost of corn ethanol production in the year 2020, as modelled by Hettinga et al. (2009) [88] using progress ratio methods. This figure was chosen to represent a stage at which lignocellulosic ethanol would be competitive with corn- and sugarcane-derived ethanol [88,91]. Figure 3.2 EISA cellulosic ethanol production mandate from 2010 to 2020  Based upon these cost assumptions and 8.65 doublings over 10 years, a progress ratio of 0.86 will be required to reduce the production cost to $0.249 L-1 by 2020, assuming the production schedule outlined in EISA [2]. This ratio was similar to the ratios previously determined for corn ethanol (0.82) [88] and sugarcane ethanol (0.8) [91]. If a higher production cost target (e.g., $0.40-0.50 L-1) was assumed, the required progress ratio would be higher and therefore easier to achieve. 3.3.2  Cost Reduction: Feedstock Selection  Since significant cost reductions must occur, albeit with a reasonable progress ratio, the question becomes where and how will the cost reductions occur. Biomass is highly heterogeneous, from the overall proportions of cellulose, hemicellulose, lignin, extractives, and ash through to its microstructure at the cellular level and the methods used to harvest, store, and commute the feedstock. However, the structure and cellulose/hemicellulose content will generally determine the potential ethanol yield from a given feedstock. In general, herbaceous biomass is higher in hemicellulose, in particular 5-carbon (C5)  99  hemicellulosic xylose/arabinose sugars, and lower in lignin content than woody lignocellulosic materials. Hardwoods, such as poplar and eucalyptus, are lower in lignin and higher in hemicellulose than softwoods, such as pine and fir. The type of lignin also differs between phylum, with gymnosperm lignin composed almost entirely of guaiacyl with low levels of p-hydroxyphenyl, while angiosperms have varying levels of syringal, guaiacyl, and p-hydroxyphenyl depending upon the species [121]. When the composition of a range of potential biomass-to-ethanol feedstocks was compared (Table 3.2) it was apparent there could be substantial difference in their theoretical ethanol yield, with corn stalk and stover having the largest potential yield. However, if only six-carbon (C6) sugars are considered for fermentation, woody biomass, and Douglas fir in particular, has a much greater theoretical yield than do most agricultural residues. Due to differences in difficulty of hydrolysis based upon structural differences between biomass types, theoretical yield alone should not be used as the basis for choosing the most attractive feedstock. In addition, the contents of materials presented here are from a small selection of samples and real-world materials will inherently have a range in material composition. Table 3.2 Holocellulose content and theoretical ethanol yield of various lignocellulosic feedstocks Feedstock  Glucan (kg bdt-1)  Mannan (kg bdt-1)  Galactan (kg bdt-1)  Xylan (kg bdt-1)  Arabinan (kg bdt-1)  Theoretical Yield (L bdt-1; all sugars)  Corn leaf [390] Corn stalk [390] Poplar stem [391] Douglas fir heartwood [392] Douglas fir sapwood [392] Conifers normal (27 species avg) [393] Conifers compression wood (27 species avg) [393]  342 365 403 445  18 17 31 126  25 24 7 38  221 216 176 26  35 32 6 10  467 476 452 466  Theoretical Yield (L bdt-1; C6 sugars only) 278 293 318 439  435  119  32  20  10  445  423  387  88  19  63  14  413  356  293  48  93  58  8  362  313  While it is possible to isolate differing forms of wood (e.g., compression wood vs. noncompression wood) for experimental purposes, feedstocks available in an integrated forest industrial system are often mixed and can include multiple wood species. An example is the spruce-pine-fir (SPF) lumber mix which is marketed in British Columbia, which indicates that the lumber could be from trees  100  of any one of the three genera. Trees from different species and genera are often cut together at the same mill, resulting in mixed mill residues, such as pulp chips, sawdust/shavings, and hog fuel. This latter feedstock is composed largely of residues from debarking operations – primarily bark. Table 3.3 provides a carbohydrate content and ethanol yield comparison of feedstocks from the forest industrial system in British Columbia. Table 3.3 Properties and ethanol yield of feedstocks from the forest system Feedstock  Moisture Content (% w/w wet basis)  Bark Content (% w/w)  Energy (GJ bdt-1)  Energy (GJ wet tonne-1)  Whole 50 12.5 18.8 8.1 logs Harvest 40 25 19.1 10.4 residue Sawdust/ 25 1 18.5 13.2 shavings Pulp chips 45 1 18.5 9.0 Hog fuel 55 80 20.5 7.6 Based upon results from Robinson et al. (2002) [369]  Glu (kg bdt-1)  Man (kg bdt-1)  Gal (kg bdt-1)  Xyl (kg bdt-1)  Ara (kg bdt-1)  Theoretical Yield (L bdt-1; all sugars)  418  99  17  50  8  383  Theoretic al Yield (L bdt-1; C6 sugars only) 345  397  96  16  48  8  365  329  431  107  18  54  9  405  359  431 292  107 68  18 12  54 35  9 6  405 267  359 240  While feedstocks from the forest system differ in their theoretical ethanol yield, their market value also varies significantly based upon their typical end use. For example, whole logs can be used to produce dimensional lumber, while hog fuel has traditionally been used only for low value heat and power production. Thus, the maximum feedstock cost that could be paid for forest biomass feedstocks was compared to the net feedstock cost (after co-product credits) of corn ethanol, taking into consideration ethanol yield (Figure 3.3). Scenarios 1 and 2 are feedstock cost equivalent to a 2007 figure of $0.148 L-1 [88], while scenario 3 is for the 2020 projection of $0.11 L-1 [88]. Yields are 50%, 70%, and 80% respectively, with ‘A’ scenarios including all sugars and ‘B’ scenarios including C6 sugars only. The maximum feedstock cost ranges from less than $15 bdt-1 for hog fuel to over $50 bdt-1 for pulp chips and sawdust. It should be noted that this figure does not consider whether ethanol yields are realistic for a given feedstocks, since feedstock characteristics such as low moisture content and small particle size have been shown to have a considerable impact on yield [137].  101  Figure 3.3 Maximum biomass cost for lignocellulosic ethanol to be competitive with corn ethanol on a net feedstock cost basis  Scenarios: 1A: 2008 feedstock cost of $0.148 L-1 and 50% conversion of all sugars; 1B: 2008 feedstock cost of $0.148 L-1 and 50% conversion of C6 sugars only; 2A: 2008 net feedstock cost of $0.148 L-1 and 70% conversion of all sugars; 2B: 2008 feedstock cost of $0.148 L-1 and 70% conversion of C6 sugars only; 3A: 2020 projected feedstock cost of $0.11 L-1 and 80% conversion of all sugars; and 3B: 2020 projected feedstock cost of $0.11 L-1 and 80% conversion of C6 sugars only. Net feedstock costs based upon 2008 and projected 2020 net corn ethanol (including DDGS credits) feedstock cost from Hettinga et al (2009) [88]  3.3.3  Cost Reduction: Enzymes  After capital and feedstock costs, cellulase enzymes are the largest cost associated with lignocellulosic ethanol production as modelled. Commercial cellulases always contain endoglucanases, cellobiohydrolases, β-glucosidase (cellobioase) as well as other ancillary enzymes. They are also often supplemented with additional β-glucosidase to limit end-product inhibition caused by accumulated cellobiose [46]. Over the last few years, significant progress has been made in reducing enzyme cost, with Novozymes claiming a 30-fold reduction in enzyme cost between 2001 and 2005 from $1.32 L-1 to between $0.026 and $0.048 L-1 at the bench scale [394], although this figure was later revised to $0.132 L-1 in 2010 to represent industrial costs [362]. Cellulase enzyme cost contribution to the total production cost of lignocellulosic ethanol can be approximated by the equation:  𝐸=  𝑃𝑝 ∙𝐿∙𝐶𝑐𝑒𝑙 𝐴∙𝑌  × 1000  (3.2)  102  Where E is the price of enzyme in $ L-1 ethanol, P p is the price of protein in $ kg-1, L is the enzyme loading in Filter Paper Units (FPU) g-1 cellulose, C cel is the cellulose content of the biomass (%), A is the enzyme activity in FPU g-1 protein, and Y is the ethanol yield in L bdt-1. Apart from enzyme recycling/reuse, the options to reduce enzyme cost are a) reduce the enzyme loading; b) increase enzyme activity; c) reduce the cost of protein; and d) increase the overall hydrolysis yield. Each of these options is described below and the potential for reductions assessed. a) Reduce the enzyme loading. A review of the literature and prior work shows that enzyme loading generally varies from low (5 FPU g-1 cellulose) to high (>50 FPU g-1 cellulose), but typically ranges from 10 to 30 FPU g-1 cellulose for most lignocellulosic substrates [136,368,388,395]. Enzyme loading tends to be higher for softwoods and lower for herbaceous biomass (e.g. corn stover, switchgrass), likely due to the higher lignin content of the former and the lower lignin content of the latter. The base case modelled here uses 20 FPU g-1 cellulose [136,388]. Although some have argued that 5 FPU g-1 cellulose will be sufficient loading for hydrolysis of the majority of cellulose, previous work has shown that 5 FPU g-1 cellulose is insufficient for softwoods and 10 FPU g-1 cellulose is a more likely target for 2020 [395]. Unproductive binding of cellulase to lignin highlights the importance of effective pretreatment to remove lignin and open the lignocellulosic matrix to enable enzyme access to cellulose and complete hydrolysis of the substrate in a timely manner [396]. b) Increase enzyme activity. Enzyme activity varies dramatically, from below 400 FPU g-1 protein to over 6000 FPU g-1 protein in enzymes derived from Trichoderma reesei mutants [397,398,399]. However, these high enzyme activities are typically accompanied by very low organism enzyme production and are therefore unsuitable for industrial ethanol applications. Current commercial enzyme products, such as Celluclast® and Cellic® CTec2 from Novozymes, and Accellerase® DUET and Spezyme® CP from Genecor, have activities in the range of 600 FPU g-1 protein [356,357,358]. While bioprospecting continues to uncover new enzymes and various enzyme  103  “cocktails” have shown improvements, substantial increases in enzyme activity have, disappointingly, not been a significant source of cost reductions. c) Reduce the cost of protein. Corn ethanol, at a loading of 0.192% amylase by mass (combined alpha and gluco) with dry corn and an ethanol yield of 418 L bdt-1, uses approximately 4.6 g protein l-1 ethanol [400]. Comparatively, at 600 FPU g-1 protein, 20 FPU g-1 cellulose, a 42% cellulose content, and a yield of 250 L bdt-1 biomass, 56 g protein l-1 ethanol are required for lignocellulosic ethanol. This is a protein requirement which is 12 times greater than that required for corn ethanol and might be considered conservative given that previous protein loading estimates were in the range of 40 – 100 times more protein [401]. The enzyme cost component of corn ethanol is approximately $0.011 L-1 [89], which results in a protein value of $2,417 t-1. The current price of cellulose enzymes, as identified by industry producers, is $0.132 L-1 [362], valuing cellulase protein at $2,359 t-1 based upon a calculation from Equation 3.2, or approximately the same as amylases used for starch saccharification. This is consistent with previous findings [401]. Alternatively, a loading of 10 FPU g-1 cellulose and assuming the same contribution to MESP would double the cost of protein to $4,800 – a cost consistent with results from Humbird et al. (2011) [363]. However, the U.S. DOE Biomass Program places the ‘state of art’ figure at $0.092 L-1 and targets $0.032 L-1 by 2012 [389]. This puts a value on protein of $1,643 t-1 for 2008 and $571 t-1 for 2012. This latter number is approximately 5 times lower than of amylase protein value and is in fact 22.6% lower than the market price for soymeal protein of $700 t-1* d) Increase the overall hydrolysis yield. As previously discussed, ethanol yield is an interplay between biomass properties, including cellulose/hemicellulose/lignin ratios, lignin type, fibre structure, and non-lignocellulosic components and processing conditions, including pretreatment technology and severity, enzyme loading/activity and hydrolysis time and completeness of hydrolysis. Sugar content determines the maximum theoretical yield, but a lower theoretical yield does not inherently lead to greater enzyme costs as a lower cellulose percentage may require a lower enzyme loading per tonne of biomass. While overall yields of 85% or greater have been 104  achieved in bench-scale settings [138], the 72% yield attained by corn ethanol in 1980 [88] is a more realistic target in the short- to medium-term. Fermentation of C5 sugars is particularly important to obtain a high yield for herbaceous biomass due to its high xylan content, while pretreatment selection strongly influences the cellulose-to-glucose hydrolysis yield of softwoods [136,138,388]. *Based upon market price of soymeal of $335 t-1, with a protein content of 48% [402] Under a ‘best-case’ scenario with enzyme loading of 10 FPU g-1 cellulose, enzyme activity of 800 FPU g-1 protein, a cellulase protein cost 75% that of amylase ($1,800 t-1), 45% cellulose in biomass, and an ethanol yield of 80% of a theoretical 475 L bdt-1, enzyme cost would be $0.027 L-1 ethanol. This shows that while the DOE target for enzyme cost of $0.032 L-1 ethanol is theoretically possible, breakthroughs are required that meet or exceed those achieved in the production of amylases over the past 30 years in the corn ethanol industry. 3.3.4  Cost Reduction: Capital costs  The $0.278 L-1 capital cost component of production costs modelled here for the base case advanced lignocellulosic ethanol facility is between 530% ($0.0529 L-1) and 730% ($0.038 L-1) that of recently built corn ethanol plants, as reported by Shapouri and Gallagher (2005) [89] and Hettinga et al. (2009) [88] respectively.  The primary drivers for this discrepancy are scale of facility, increased  complexity of technology, inclusion of a boiler and steam turbine, increase in number of unit operations (especially pretreatment), lower sugar output per tonne of feedstock (and hence larger equipment), higher grade construction materials to handle harsh pretreatment conditions, and first-mover costs associated with being first-of-kind facilities [401,403]. a) Scaling Although the scale of 50,000 L yr-1 capacity for the first, biomass-to-ethanol commercial facilities is small relative to newly constructed corn and sugarcane ethanol plants, which averaged 150 ML yr-1 for corn [88] and 365 ML yr-1 [91] for sugarcane ethanol in 2005, it was chosen based upon the scale of facilities that have received support from the U.S. DOE’s biorefinery program. 105  As shown in Figure 3.4, increasing the capacity (scale) of the facility can make a significant contribution to reducing the capital cost component of a biomass-to-ethanol process. Since no commercial facilities currently exist and multiple facilities will be required to determine relative metrics, several scenarios have been modelled to showcase the impact of scaling factor and capacity on capital costs. Scenario A represents a 50 ML yr-1 installed cost of $3.50 L-1 yearly capacity with scaling factors ranging from 0.6 to 0.8. Scenario B, which represents a 50 ML yr-1 installed cost of $2.50 L-1 yearly capacity, is the most consistent with previously announced projects discussed in Chapter 2. The installed cost was calculated for capacities of 150 ML yr-1, 250 ML yr-1, 500 ML yr-1, and 1 GL yr-1, based upon the capacity cost scaling formula: 𝐶1 𝐶0  𝑀  = �𝑀1 � 0  𝑠  (3.3)  Where M 0 is the capacity of the base facility, M 1 is the capacity of the study facility, C 0 is the total installed cost of the base facility, C 1 is the total installed cost of the study facility, and s is the scaling factor. The capital cost contribution was calculated using an 8% cost of capital amortized over 20 years with monthly repayment. Figure 3.4 Impact of plant capacity and scaling factor on lignocellulosic ethanol production cost  ‘A’ scenarios are based upon an installed cost of $3.50 L-1 yearly capacity at 50 ML yr-1, ‘B’ scenarios on an installed cost of $2.50 L-1 yearly capacity at 50 ML yr-1, and ‘C’ scenario on an installed cost of $1.50 L-1 yearly capacity at 50 ML yr-1.  106  Despite the contribution of increasing scale to reductions in capital contribution per litre production cost, the capital cost for a lignocellulosic ethanol facility with a capacity of 1 GL yr-1 or smaller is higher than for an average capacity corn ethanol plant. This result is true for all scaling factors and base scenario costs ($1.50-3.50 L-1 yearly capacity) that were modelled. b) Equipment The inherent recalcitrance of lignocellulosic biomass results in much greater pretreatment costs and lower sugar yields per unit feedstock for a lignocellulose-to-ethanol process than a corn-to-ethanol process.  This is evidenced by the large equipment differences for this part of the process. Thus  pretreatment and the capacity/throughput of the hydrolysis digester should be the focus of equipment cost reductions in a lignocellulose-to-ethanol process. Unit operations downstream of hydrolysis, including fermentation, CO 2 scrubbing, beer vapour flashing, distillation, and ethanol recovery (using rectifier, stripper, and molecular sieves), are similar for both conventional and advanced lignocellulosic processes [356,400,389]. However, notable differences include the inclusion of feedstock degradation products such as furfural, hydroxymethylfurfural (HMF), acetic acid, formic acid, extractives, and low molecular weight phenolics in the lignocellulosic hydrolysate, which negatively affect yeast fermentation, and the considerably more dilute sugar streams obtained from lignocellulose hydrolysis as compared to starch, necessitating larger equipment or concentrators downstream of hydrolysis. c) Digester/hydrolysis reactor Compared to starch hydrolysis, the capital requirement of cellulose hydrolysis differs primarily in the required reactor capacity per unit sugar output. This is largely due to the longer hydrolysis times and lower yield per tonne of feedstock typically achieved using a lignocellulosic feedstock [401,404]. Various groups have tried to increase the solids loading (ratio of solids to non-solids) of hydrolysis, which should result in smaller reactors and ideally increased yield [405]. A typical hydrolysis profile [based on earlier work by Ewanick et al. (2007)] [136] of steam-pretreated lignocellulose is shown in Figure 3.5. While near complete (94%) hydrolysis occurs after 48 hours treatment at the higher severity (where, unfortunately, significant hemicellulose degradation occurs) the maximum reaction rate, V max , occurs 107  between 0 and 3 hours. After this point, the reaction rate decreases significantly. Although this is just one set of results at low solids loading (2% w/v), they are consistent with previous findings from the Forest Products Biotechnology Group at UBC. Figure 3.5 Hydrolysis profile of steam-pretreated pine under different severities  * Based upon Ewanick et al. (2007) [136]  It was apparent that a substantial difference in tonnes of hydrolyzed cellulose produced using the same digester operated over 48 hours, but with differing hydrolysis residency times, could be achieved (Figure 3.6). Under scenarios of less than the full 48 hours residency, it was assumed that the residual, recalcitrant substrate was removed and new substrate added. Although the sugar yield is significantly lower on a per tonne basis using short residency times, the rate of hydrolysis, and hence tonnes of hydrolyzed cellulose, was maximized. For example, the average rate of hydrolysis for the 3 hour residency (high pretreatment severity) scenario is 11.67 t hr-1, while the average rate of hydrolysis for the 48 hour residency (high pretreatment severity) scenario is 1.96 t hr-1 and averages 0.13 t hr-1 between 24 and 48 hours.  A significantly greater sugar yield could be achieved from the same equipment if the in-  situ feedstock was removed and new feedstock added.  108  Figure 3.6 Tonnes of hydrolyzed cellulose for the same 100 t cellulose digester operated for 48 hours with varying pretreatment severities and hydrolysis residency times  While increased sugar yields (hydrolyzed cellulose) resulted in increased ethanol revenues, the high throughput required when adding biomass at more frequent intervals also leads to an increase in feedstock costs. This suggested the need to examine reduced biomass hydrolysis residency time and the resulting impact it had on revenues and the feedstock-product margin (gross margin) when using the unhydrolyzed cellulose and associated lignin for combined heat and power or other products at equal (case 1) and half (case 2) the value of the original biomass feedstock (Table 3.4). A sensitivity analysis of enzyme cost showed the significant impact that enzyme consumption has on the gross margins for the overall process. It was apparent that the higher throughput scenarios of 3 hour and 6 hour residency significantly increased the feedstock margins relative to a 24 or 48 hour residency, despite the considerably higher raw biomass costs. The biomass cost was 1600% greater for 3 hour residence compared to 48 hour, but resulted in a 590% greater feedstock margin. However, it should be noted that while both 24 and 48 hours residency have positive margins on sugar revenue alone, this is not the case for 3 and 6 hour residencies. Due to the lower yield, revenue from residues was critical to achieving positive margins. 109  Table 3.4 Impact of hydrolysis residency time on biomass feedstock-product margin 3 h Residence 6 h Residence Cellulose 1600 t 800 t Throughput Cellulose 440 t 234 t Hydrolyzed Biomass 3556 t 1778 t Consumption Financial Sugar Revenue $143,000 $76,050 ($325 bdt-1) Residue Revenue $155,778 $77,189 ($50 bdt-1) Biomass Cost $177,778 $88,889 Feedstock Margin $121,000 $64,350 Sensitivity: Residue revenue reduced to $25 bdt-1 Feedstock Margin $43,111 $25,756 -1 Sensitivity: Enzyme cost of $0.0317 L included Feedstock Margin $111,949 $59,537  24 h Residence 200 t  48 h Residence 100 t  122 t  75 t  444 t  222 t  $39,650  $24,375  $16,122  $7,361  $22,222 $33,550  $11,111 $20,625  $25,489  $16,944  $31,040  $19,082  d) Pretreatment Along with the inclusion of a biomass boiler and steam turbine in most designs, the largest capital cost difference between conventional and advanced lignocellulosic ethanol is the pretreatment component. However, without pretreatment, the efficiency of the overall process decreases significantly [406]. The U.S. DOE Biomass Program calculates it as the second largest production cost, behind feedstock, for the existing state of technology but also for future commercial production [389]. The high pretreatment capital cost has been shown to be due to the two major factors of needing reactors able to withstand high pressures and/or temperatures, often in conjunction with corrosive catalysts, and the complex or difficult catalyst recovery systems needed to reduce operating costs [407].  Given the  importance of pretreatment and the lack of commercial facilities, a large number of pilot- and demonstration-scale pretreatment technologies and processing conditions exist. This differs significantly from corn ethanol, where two primary technology pathways (dry grinding and wet milling) exist, but where a single pathway and plant design (dry grinding) has come to dominate the industry. This standardization allows economies-of-scale in plant component production and a clear definition of best practices.  With multiple, competing, commercially-unproven technologies, economies-of-scale will 110  likely not be achieved for some time. In addition, many of these technologies have not achieved extended commercial operation and therefore actual real-world performance at a commercial scale is uncertain. Table 3.5 Comparison of selected lignocellulose pretreatments for ethanol production Hot water  Steam  Organosolv (ethanol)  AFEX  Dilute acid  Feedstocks that can achieve >80% yield within 48h hydrolysis and loading 20 FPU g-1 Unit operations  Hardwood; herbaceous  Softwood; hardwood; herbaceous  Softwood; hardwood; herbaceous  Herbaceous  Hardwood; herbaceous  2 (hot water treatment; solid/liquid fractionation)  3 (organic solvent cooking; solid/liquid fractionation; ethanol recovery)  3 (prewetting; ammonia cooking; ammonia recovery)  3 (pre-soaking; dilute acid treatment; fractionation and washing)  Residence time (min) Operating temperature (° C )/ pressure (bar) Solids loading (% w/w) Catalyst (%) Solvent  15  4 (pre-soaking; steam impregnation and release; solid/liquid fractionation; alkaline post treatment) 2-5  60  5  5  190/12.5  190/12.5  170/13.3  90/26.0  160/6.2  16  30  10  25  19  None Water (superheated)  SO 2 (3) Water (steam)  H 2 SO 4 (1.1) Ethanol (65%); water (35%)  H 2 SO 4 (0.98) Water (superheated)  Recovery  None  Catalyst  Solvent cost (including energy inputs) Capital cost (US$ M 2007) @ 2000 t day-1 Operating cost rank based upon solvent, catalyst, recovery, and batch/ continuous (1= highest)  Low  Medium  Catalyst and Solvent High  None Liquid anhydrous ammonia (50%); water (50%) Solvent High  Low  4.8  30.9  27.3  26.5  5  3  1  4  2  Catalyst  Based upon previous results [139,290,407,409,410,411,412,413] It should be noted that not all pretreatments can be used effectively with all types of biomass feedstocks and that the inherent physiochemical differences in feedstock type will likely lead to diverging processing pathways. However, a comparison of leading pretreatment technologies was undertaken and 111  included a ranking on number of unit operations, cost of capital and operating costs (Table 3.5) based on the work reported by the Biomass Refining Consortium for Applied Fundamentals and Innovation (CAFI) [408]. Due to the early stage of commercialization of these technologies, trying to rank the preferred pretreatment and accurately gauge performance is difficult. Table 3.5 is intended to provide a general understanding of the options available. Due to its high lignin content, softwood is typically the most difficult feedstock to hydrolyze [326]. Pretreatments such as AFEX and hot water are insufficiently harsh to disrupt the lignocellulose matrix and enable enzyme access to cellulose fibres [414,415]. Water and steam-based pretreatments are typically lower in operating cost per tonne of feedstock than those using strong acid or chemical solvents such as AFEX and organosolv [326,416]. This is particularly true when there are fewer unit operations. In the case of softwoods, steam pretreatment is often viewed as the lowest cost pretreatment, but the use of organosolv enables more rapid and complete hydrolysis combined with the ability to isolate lignin as a potential co-product [326]. This operating cost vs. co-product credit conundrum has received substantial attention in the literature, but companies continue to pursue both processing routes with differing business models [413,417]. 3.3.5  Cost Reduction: Co-product Credits  Co-product credits for corn ethanol, namely in the form of dried distillers grains (DDGS)/wet distillers grains (WDGs) and carbon dioxide, are a major contributor to reducing the net feedstock costs for corn ethanol production. These co-products equate to approximately $0.078 L-1, thereby reducing feedstock costs by 35% from a gross of $0.226 L-1 to a net of $0.148 L-1 [88]. As mentioned earlier, the pretreatment employed for a particular lignocellulose-to-ethanol process has a large impact on the co-products that can be produced. For example, while organosolv can be used to isolate a high purity and active lignin [413], steam pretreatment generally results in a highly condensed lignin that will likely only have value as an energy product. Additionally, the type of feedstock used will also dictate the quantities of co-products that can be produced. For example, furfural, an industrial chemical used in the production of solvents and solid resins, and xylitol, a lower-calorie 112  glucose/sucrose substitute, can both be produced from xylose. Since herbaceous plants and hardwoods have significantly higher xylan content than softwoods, the ability to derive co-product credits based on this C5 sugar alcohol is also higher. Table 3.6 Revenue comparison of possible biomass-to-ethanol co-products Feedstocka  Lignin content (%)  C5 content (%)  Heat onlyb  Heat & electricityc  Heat, electricity & furfurald  High purity lignin & furfurale  High purity lignin + xylitol + furfuralf  $ bdt-1 $ L-1 $ bdt-1 $ L-1 $ bdt-1 $ L-1 $ bdt-1 $ L-1 $ bdt-1 $ L-1 Corn 18.5 25.3 33.75 0.158 153.75 0.718 294.43 1.38 178.73 0.835 381.31 1.78 stover Poplar 23.5 18.2 37.98 0.159 173.02 0.724 274.80 1.15 137.13 0.574 299.93 1.26 stem SPF 32.0 5.0 38.81 0.157 176.82 0.716 204.31 0.827 59.13 0.239 102.92 0.417 harvest residue SPF pulp 30.1 5.0 35.31 0.131 160.86 0.598 187.62 0.697 57.53 0.214 115.87 0.431 chips aAssumes a 75% ethanol yield from C6 sugars and product value assumptions: heat = $4 GJ-1; electricity = $0.14 kWh-1; furfural = $1.50 kg-1; high purity lignin = $0.12 kg-1; xylitol = $3.50 kg-1 b 75% efficiency based upon bone dry basis LHV and calculated by subtracting cellulose energy content of 16.1 GJ t-1 and C6 conversion to ethanol of 75% [418] and hemicellulose LHV of 13.6 GJ t-1 [181] c 50% heat efficiency and 30% electrical efficiency based upon LHV above d 50% heat efficiency and 30% electrical efficiency based upon LHV above. 43% furfural yield from C5 sugars based upon 75% conversion of xylan to xylose of theoretical maximum of 1.136 g g-1 [389,419] and 60% conversion of xylose to furfural of a theoretical maximum of 0.72727 g g-1 [420], plus 11% xylan conversion to furfural from pretreatment [389] e 70% high purity lignin yield plus 43% furfural yield from C5 sugars as above f 70% high purity lignin yield. 41% w/w xylitol from xylan yield based upon 75% conversion of xylan to xylose of theoretical maximum of 1.136 g g-1 [389,419] and 60% conversion of xylose to xylitol of a theoretical maximum of 91% (w/w) [421]. 9% w/w furfural from xylan yield based upon 11% xylan conversion to furfural from pretreatment [389] with 43% w/w (see above) conversion of arbinan to furfural  Some of the co-products that could be produced in a lignocellulose-to-ethanol process were assessed for their ability to contribute to the economic performance of the lignocellulosic ethanol facility by estimating the expected revenues that could be generated from one tonne of various biomass feedstocks assuming a 75% ethanol yield from C6 sugars and no fermentation of C5 sugars. The results are presented in Table 3.6. It should be noted that these are gross revenue figures enabling a comparison across products and that all material not converted to ethanol is included. In reality, should this material be used for co-products, it could result in increased operational costs for a facility due to the absence of material for process heat if an external heat source were not available at no cost (for example, the Inbicon A/S facility in Kalundborg, Denmark sources heat from a neighbouring coal power plant) [422]. Inherently, operating and capital expenditure decisions will be based upon the relative return of various options. A complete processing pathway and co-product return financial analysis was beyond the project  113  scope but it was useful for the purposes of this thesis to derive a rough, high level comparison of possible co-product revenues. These possible co-product credits should not be compared directly to the $0.078 L-1 DDGS credit of corn ethanol since this figure is net income from DDGS and not simply revenue. 3.3.6  Cost Reduction: Energy and Administration  As identified above, both heat and electricity can be obtained as co-products of a lignocelluloseethanol process. With a yield of 300 L bdt-1, approximately 1/3 of the inherent energy in biomass (assuming 18.5 GJ bdt-1), or 6.3 GJ bdt-1, is produced in the form of ethanol. The remaining 2/3 of inherent energy, or 12.2 GJ bdt-1, can be used for process heat and electricity generation. Assuming 65% heat efficiency and 25% electrical efficiency, heat production would equal 26.4 MJ L-1 and electrical production of 2.8 kWh L-1. This compares with estimated internal process heat requirement of 16-19 MJ L-1 and electrical requirement of 0.9 kWh L-1 (varies by technology and feedstock) [353], highlighting process energy self-sufficiency and the potential for sale of heat and electricity. This is a significant advantage over conventional corn ethanol, where energy (heat and electricity) is the largest operating cost of production at an average of $0.0525 L-1 [89]. It is recognised that new technologies that are located in jurisdictions that are unfamiliar with biofuel production will inherently suffer from first-mover regulatory hurdles. However, these problems are typically overcome as more projects are built and are shown to operate safely with minimal environmental impacts.  Time and experience, along with cooperation of regulatory agencies and  governments, lignocellulosic ethanol production cost components such as insurance/permits, property taxes, and supervision should become cost competitive with corn ethanol.  3.4  Discussion This high level estimation of anticipated lignocellulosic ethanol production costs was carried out  to determine if the significant cost reductions in production components, such as the cellulase enzymes, and the predicted rapid commercialization of lignocellulosic ethanol could translate into competitive production relative to corn ethanol processes.  It was apparent that for this to happen, substantial  reductions are required for all production cost components of a lignocellulose-to-ethanol process. 114  Currently, lignocellulosic ethanol is economically uncompetitive with conventional ethanol production, with the cost proving to be 250-300% greater. The volumes of cellulosic ethanol mandated by EISA, combined with a reasonable cost reduction progress ratio of 0.86, could enable production cost reductions that make lignocellulosic ethanol competitive with corn ethanol by 2020. However, the original 2010 mandate of 100 M gal (378.5 ML), which was used to calculate the progress ratio used in the analysis presented here, has been scaled back to 6.5 M gal (24.6 ML) [423], while the 2011 mandate has been set at 6.8 M gal (25.7 ML), down from the originally targeted 250 M gal (945 ML) for 2011 [424]. Therefore, while the progress ratio for lignocellulosic ethanol is reasonable (albeit using an accelerated cost reduction on a time scale relative to conventional ethanol), the volumes so far have been significantly lower than initially proposed. Without a large commercial industrial production volume and multiple doublings of cumulative production, and the typically associated cost reductions, it is unlikely lignocellulosic ethanol will be competitive with corn ethanol by 2020. Much of the lignocellulosic ethanol efforts in the U.S. have focused on switchgrass and corn stover as feedstocks, since agriculture is projected to have a significantly greater potential for large ethanol volumes than forestry. This is evident in the U.S. DOE and USDA Billion Ton annual biomass supply study, in which it is suggested that agricultural lands could supply 998 million tons (905 Mt), while forested lands could only supply 368 million tons (334 Mt) [425]. However, while agricultural residues such as corn stover have an advantage over woody biomass in potential ethanol yield from total sugars (largely due to their lower recalcitrance, lower lignin content and hence greater sugar content of agriculture-derived biomass), if only C6 sugars are considered, softwood biomass has the highest potential ethanol yield. As C5 fermenting microorganisms have yet to be fully proven in an industrial setting to the same extent as C6 fermenting yeasts [46], this needs to be a consideration when selecting feedstocks and siting facilities. While it is evident that the theoretical ethanol yield itself is not an indication of cost of production and the inherent heterogeneity of biomass feedstocks, such as the quantity and type of lignin, will significantly influence actual production costs, it does highlight how feedstock selection will impact overall production cost and yield. 115  When examining fibre flows in the forest industrial system, the maximum feedstock cost to be competitive with corn ethanol on a feedstock basis ranges from $15-50 bdt-1, with hog fuel the lowest due to the high bark content and bark-free pulp chips being the highest. The maximum feedstock costs for the competitive lignocellulose-derived ethanol production costs calculated here were in the general range of biomass delivery costs modelled in previous studies. Agricultural residues typically range in cost from $20-75 bdt-1 [426,427] while forest harvest residues range from $25-50 bdt-1 [199,200].  This is  substantially less than whole tree harvest and chipping for dedicated bioenergy operations, where feedstock can reach over $100 bdt-1 [202]. Therefore, the lignocellulosic biofuel and bioenergy industries will be more competitive when utilizing residues as a secondary processer than dedicated harvest operations. When these figures are compared to the maximum competitive feedstock cost calculations, it is clear that advanced lignocellulosic ethanol does not appear to have a significant feedstock cost advantage over corn ethanol. Some of the modelled lignocellulosic biomass feedstocks, such as harvest residues and whole logs, may in fact have a higher delivered cost than corn on a per litre ethanol basis [199,202]. A large distinction must be made between cost of delivered feedstock and the price of delivered feedstock for both biomass and corn. Since corn is traded as a commodity and many ethanol facilities purchase grain at contracted price tied to the market price, this could be significantly higher than actual production costs (i.e. difference in price vs. cost). However, ethanol production companies that own or control land and grow corn through cooperatives or contract farming (vertical integration) should have a feedstock price that is at least equal to the cost of production. The reduction in cellulase enzymes costs has been a focal point for the lignocellulose-to-ethanol process and it has proven to be a considerable success [362,389]. However, based upon the analysis presented here, it will be extremely difficult for enzyme suppliers and ethanol producers to meet the U.S. Biomass Program 2012 target of $0.032 L-1 ethanol. The $0.132 enzyme cost contribution to ethanol production announced by Novozymes in 2010 [362] could represent a protein cost of $2400 t-1 (assuming 600 FPU g-1protein and 20 FPU g-1 cellulose) - equivalent to the cost of amylase protein used in  116  conventional ethanol. However, assuming typical methods of enzyme production, it will be a challenge for cellulases to be produced at a substantially lower cost than amylases. Due to the recalcitrant nature of most lignocellulosic feedstocks, a lignocellulose-to-ethanol process will inherently be more complex and require more unit operations than starch-based ethanol processes. Although this initially results in higher capital costs, a combination of strategies could make these capital costs more competitive with corn ethanol production. Chief among these is increasing the scale of facilities, as previous studies have shown that the optimal scale of biochemically-based lignocellulosic biomass-to-ethanol facility is greater than 500 ML yr-1 [204,290], with several estimates exceeding 1 GL yr-1 [146,227]. Chapter 4 examines the potential of supplying these facilities with biomass feedstocks using truck, rail, and marine transportation. However, scaling the plant capacity up to 1 GL yr-1 was not sufficient in itself to reduce the capital costs to a point where it would be competitive with corn ethanol at the average facility scale of 150 ML yr-1 capacity.  Therefore, technological  improvements and real equipment cost reductions must be combined with increases in scale to reduce per unit capital costs. Increases in solids loading could also significantly increase facility throughput without increasing the actual size of equipment. Past experience from the sugarcane ethanol industry has shown that the majority of capital cost reduction over the 30 years from 1975 to 2004 was primarily due to increases in scale rather than any dramatic improvement in technology [91]. This indicates that capital cost reduction, other than increasing the scale of the facility, may be challenging. The chief equipment difference between conventional and advanced lignocellulosic ethanol is the necessity for pretreatment equipment. Catalyst and solvent recovery systems, which might be used to reduce operating costs, also increase the capital costs. Another difference is the generally long hydrolysis times and relatively low sugar concentrations that are achieved in a typical lignocellulose-to-ethanol process. Since hydrolysis of cellulose to sugar monomers occurs rapidly in the first few hours but slows quickly, it might be advantageous to reduce the residence times for hydrolysis (1-12 hours) in order to maximize hydrolysis rate and to use the recalcitrant, unhydrolyzed biomass for energy products (heat and electricity), or other co-products (such as furfural, high purity lignin, and/or xylose). This is where some 117  pretreatments such as organosolv might have an advantage over other pretreatments due to the production of higher value chemical products (such as Lignol’s High Purity Lignin (HPL)) and the associated reduction in net feedstock costs.  3.5  Conclusion It is apparent that, in order for advanced lignocellulosic ethanol to compete with conventional  ethanol, large cost reductions must occur in a number of areas, including the capital and operational costs of most of the units within an enzyme-based lignocellulose-to-ethanol process. Improvements in one area alone will be insufficient to provide the cost reductions necessary for an economically viable enterprise. As conventional ethanol production costs are a moving (and generally decreasing) target, lignocellulosic ethanol producers must not look to the current ethanol production cost, but the future, reduced cost. The projected superior environmental impacts from the use of lignocellulosic ethanol relative to corn ethanol may not be realized unless fuel production is not economically viable and commercialized. It is likely that, at least initially, lignocellulose-derived ethanol will require some type of subsidy or policy driver beyond those currently afforded corn ethanol in order for it to be competitive beyond 2020. As identified in this chapter, facility scale and feedstock play a pivotal role in the ability of lignocellulosic ethanol producers to minimize production costs. Economies-of-scale mean larger facilities will have a lower capital cost contribution to the MESP. Although most previous assessments of lignocellulosic ethanol production assume a local feedstock supply and use that to determine maximum facility scale, it is possible that producers should look to long distance transportation of feedstocks in order to maximize economies-of-scale. The following chapter is focused on the ability to achieve those economies-of-scale, identified as so important to reducing unit capital costs, from a logistical feedstock perspective.  118  4 4.1  THE INFLUENCE OF FEEDSTOCK LOGISTICS ON MAXIMUM FACILITY SCALE, LOCATION AND TECHNOLOGY SELECTION Introduction As shown in the previous two chapters, capital costs make a significant contribution to the MESP  and therefore, reduction of these costs by maximizing economies-of-scale is a key opportunity for reducing total production costs and improving the competitiveness of lignocellulosic ethanol relative to corn ethanol and gasoline. In addition, identification of the maximum facility scale can be used at a macro level by government policymakers to determine the number of facilities required to supply bioenergy mandates and targets. This provides a more tangible measure of policy reasonableness. As previously discussed, there are two primary competing process platforms, biochemical and thermochemical, that have been proposed for liquid biofuel production from lignocellulosic forest feedstocks. Both platforms are able to deliver about 1/3 of the initial energy content of the feedstock in the final biofuel product. Both platforms are also able to process a variety of lignocellulosic feedstocks, including both hardwood and softwood forest species. However, desirable feedstock characteristics differ in critical ways. Typically, biochemical based conversion works best with high moisture content (green) feedstocks, as these are less likely to have undergone hornification, an effect of drying which reduces the ability of moisture to travel through cellulose fibres, the accessibility of the feedstock to chemicals and steam during pretreament, and the subsequent ease of access for enzymes in the hydrolysis step [137,368]. Although limited research has been done on the use of pellets for biochemical production of biofuels, experience from the pulp and paper industry shows that dry biomass restricts effective biomass pretreatment and fractionation, resulting in lower quality fibre [428].  In contrast, thermochemical  conversion typically works best with dry feedstocks [46]. Small particle size, such as is exemplified with sawdust or “wood flour”, is also a favourable feedstock characteristic for processes such as entrained flow gasification, which can be considered the “generic” route for thermochemical conversion to liquid fuels [429]. Thus, wood pellets that have been macerated will likely be an excellent feedstock for most thermochemical routes for biofuel production. This major feedstock preference difference between the 119  thermochemical and biochemical conversion routes indicates that the scale and site location of plants will likely vary dramatically, simply due to feedstock characteristics. It is also likely that fundamental technology differences will result in different cost drivers for the two conversion routes [226]. In order to reduce biomass feedstock transportation costs while enhancing conversion efficiency through economies-of-scale, several groups have looked at increasing the bulk density of lignocellulosic feedstock [367,430].  For the purposes of this study, densified feedstocks are considered to be  intermediates to the liquid biofuel production process. Examples of intermediates include residues such as wood chips, solid products such as wood pellets and briquettes [367], and liquid products such as biooil and methanol [431,432,433]. In addition to logs from the forest, three of these intermediate products – wood chips, wood pellets, and biooil – were also assessed as potential feedstocks for liquid biofuel production. Wood chips are typically generated as the co-product of modern sawmilling operations. In British Columbia, approximately 40% of the harvested lumber volume is converted to chips during the production of solid lumber products [434]. In some jurisdictions, such as Eastern Canada or in some regions of Scandinavia, full logs (pulpwood) may be converted to chips at pulp mills. In other locations in Central and Eastern Canada and the interior of British Columbia, chipping can be carried out at, or very near, the point of harvest, using portable machines that debark, delimb and chip entire trees in a single, continuous manner.  For example, the Daishowa-Marubeni International Ltd. plant in Peace River,  Alberta, Canada, no longer processes pulpwood at the mill but relies on nine in-woods chippers instead [434,435]. As chipping is carried out on wet (green) logs, wood chips could be a good intermediate for use with biochemical conversion processes, or if dried and further comminuted, could act as a feedstock for the thermochemical route. Wood pellets, discussed extensively in Chapter 2 as a potential feedstock competitor to lignocellulosic ethanol, are the largest volume internationally-traded solid fuel bioenergy commodity. In this chapter, they are considered a potential feedstock rather a feedstock competitor. Global pellet production was close to 10 Mt in 2008, with 25% of production exported outside of the country of origin 120  [436]. Europe was the largest market, with consumption estimated at 8.5 Mt in 2008 [437]. Industrial pellet mills range in capacity from less than 5,000 t yr-1 to greater than 600,000 t yr-1. Wood pellets are internationally traded, have established quality standards (eg., CEN/TS 14961:2005) [322], and wood pellet futures are traded on commodity exchanges (eg., ENDEX). The highest volume trade routes are currently from British Columbia, Canada to Stockholm, Sweden, Rotterdam, Netherlands, and the United Kingdom [438]. Pellets are traditionally used for heat applications, ranging from household stoves and furnaces to city-scale district heating systems. However, they are also used for electricity production in both stand-alone and co-firing (with coal) applications. Pellets are considered an intermediate as they could be used as a feedstock for liquid biofuel production, although this is not currently practiced commercially.  Pellets can also be produced from agricultural residues and herbaceous materials.  However, due to their low percentage of lignin, which functions as the pellet binder and provides higher calorific value, pellets made from these feedstocks generally have lower durability and value [439]. Biooil, or pyrolysis oil, is the dominant product (approximately 70-75% by weight) derived from fast pyrolysis. This typically involves heating the biomass feedstock at 400-600° C in an anaerobic environment [440,441], resulting in three main streams of liquid biooil, syngas, and biochar. At a 70% yield and a biooil density of 1.2 kg L-1, about 1.7 t of wood at 20% moisture content is usually required to produce 1000 L biooil. Typical constituents of biooil are water (20-30%), lignin fragments (15-30%), aldehydes (10-20%), carboxylic acids (10-15%), carbohydrates (5-10%), and small fractions of phenols, furans, alcohols, and ketones [440]. However, biooil composition can change dramatically based upon the pyrolysis conditions used and the nature of the feedstock [442,443]. The heterogeneous nature of the biofuel can make it challenging for transportation, storage, and utilization by downstream processors [443,444]. Biooil has a relatively high viscosity when compared to fossil fuels at room temperature. The viscosity decreases when the temperature increases with a typical viscosity of 40 to 100 cp achieved at 40°C and 25% water [440,443]. These viscosity challenges, combined with biooil’s high water content and low pH, make it more difficult to pipeline than most oil and significantly more difficult to pipeline than finished oil products [441]. World biooil production is currently very small and biooil can be 121  considered as a specialty product rather than a typical tradeable commodity [432]. However, it should be noted that biooil can be used as an intermediate feedstock for both ethanol and FT liquids.  The  conversion of the levoglucosan fraction of biooil into ethanol has been investigated [445], as has the distributed production of biooil for subsequent conversion to FT liquids [150]. 4.1.1  Logistics and Scaling  Investment in large-scale lignocellulosic biofuel facilities will only occur if feedstock supply can be secured and risk minimized. Numerous techno-economic analyses have estimated the scale and cost of a biorefinery based upon conversion technology economies-of-scale weighed against transportation costs and feedstock supply [205,204]. An assessment of previous work on the optimal scales proposed for biorefineries based on biochemical and thermochemical conversion processes (Table 4.1) shows the optimal scale for bioconversion-based facilities ranges between 0.24 and 2.5 GL ethanol yr-1, while thermochemical-based facilities ranges between 1.4 and 8.4 GL FT liquids yr-1. The range in these figures is explained, in part, by the small number of commercially viable operations available for comparison and the large variety of technologies currently being pursued. It should also be noted that, given the relative novelty of these technologies and the lack of demonstration and commercial data to confirm these values, these estimates should be considered with caution.  Other factors such as  government policies on maximum scale, ease of enviornmental permit obtainment, and pollutant release permit limitations will also play a role in the scale at which facilities will eventually be built. In order to identify optimal scale, Jack (2009) created a simple model contrasting facility capital cost economies-of-scale with feedstock cost diseconomies-of-scale [207]. Several studies, including Graham et al. (2000) and Elmore et al. (2008), have considered the impact of feedstock availability and transportation costs on facility scale in a given region, often at the State or Province level [208,209]. Kumar et al. (2003) identified the optimal scale of a biopower plant based upon feedstock delivery cost and scaling factors [210]. Searcy and Flynn (2008) established functions relating biomass yield and processing cost for electricity (combustion and integrated gasification combined cycle), FT liquids, and cellulosic ethanol production [211]. Wright et al. (2008) examined a distributed processing system 122  involving delivery of pyrolysis biooil to a central biofuel production facility [150]. Almost all bioenergy feedstock management studies have focused upon local supply for bioenergy facilities, principally determining the draw radius and average/incremental transportation costs by truck [212,213]. However, the possibility of importing biomass using rail and ship to large, coastal facilities has been proven to be feasible, both theoretically [214,433], and practically, as evidenced by the 1.5+ Mt of wood pellets exported from Canada to Europe each year [171]. Uslu et al. (2008) also examined converting biomass into intermediates biooil, pellets, and torrefied biomass for international supply chains using a technoeconomic assessment [215]. It is apparent that a “not-in-my-backyard” (NIMBY) mentality has challenged the drive towards renewable energy and industrial projects and biomass processing facilities are no exception [447]. This is particularly true for large-scale facilities. For example, the proposed AU$1.7 B Gunns Limited Bell Bay Pulp Mill in Tasmania has received significant opposition from community and environmental groups [448]. NIMBYism must be taken into account when planning biofuel facilities to reach ambitious market penetration targets, such as the 60.8 GL cellulosic biofuels by 2022 mandated by the U.S. under EISA. Whether it is easier (politically or technically) to build 122 x 500 ML yr-1 facilities or 24 x 2.5 GL yr-1 facilities [449] remains to be determined and will require further investigation and experience. By taking a distributed facility model, some processing plants will need to be located in regions where energy processing facilities have not previously been located and for which signficant local resistance may be encountered. This change in energy system structure and impact on communities needs to be taken into consideration during biofuel production system planning.  123  Table 4.1 Optimal and near-optimal facility scales for advanced lignocellulosic biofuel production Optimal plant capacity ML MLGE per year a per year b  Conversion pathway  Feedstock  Product  Source reference  238 c 156 Biochemical Yellow Poplar Ethanol 356 511 d 335 Biochemical Aspen, Poplar Ethanol 204 767 e 503 Biochemical Corn Stover Ethanol 290 1,104 f 725 Biochemical Biomass Crops Ethanol 146 1,384 908 Biochemical Biomass Crops Ethanol 227 1,402 g 1,246 Thermochemical Unspecified FT Liquids 148 1,636 1,840 Thermochemical Biomass Crops FT Liquids 227 1,752 h 1,557 Thermochemical Unspecified FT Liquids 446 1,851 2,082 Thermochemical Biomass Crops FT Liquids 150 2,555i 1,677 Biochemical Unspecified Ethanol 449 8,412+ 9,464+ Thermochemical Biooil FT Liquids 150 a Assumes 32 MJ L-1 for gasoline, 21 MJ L-1 for ethanol, and 36 MJ L-1 for FT liquids b Million Litres Gasoline Equivalent c Modelled capacity and may not represent optimal scale d Based upon 4,000 bdt day-1 feedstock input and assumed conversion of 350 L bdt-1 e Based upon 6,000 bdt day-1 feedstock input and assumed conversion of 350 L bdt-1 f Based upon 2000 MW HHV input (8,640 bdt day-1 feedstock input) and assumed conversion of 350 L bdt-1 g Based upon capacity of 1600 MW th FT liquids h Based upon capacity of 2000 MW th FT liquids i Based upon 20,000 t day-1 feedstock input and assumed conversion of 350 L t-1  While biomass is a distributed resource, over time, economies-of-scale will drive the push towards larger, higher efficiency and lower installed cost-per-unit capacity plants.  This has been  observed in related industrial sectors. For example, between 1990 and 2005, the average size of a dry mill ethanol plants in the United States increased by 235%. At the same time, industrial processing costs of corn ethanol declined by 45% between 1983 and 2005 [88]. Pulp and paper facilities have also continued to grow, with a typical eucalyptus-based facility in Brazil producing up to 2.3 M air-dried tonnes pulp yr-1 [450,451]. In comparison, North American mills tend to be older and smaller; an average Canadian pulp mill has a capacity of approximately 200,000 t yr-1, which is 11.5 times smaller than a Brazilian mill [452,453]. It is apparent that substantial differences exist between fossil fuel facilities and biomass facilities with regard to their fuel management, sourcing, handling, pre-processing and transportation.  Fuel  differences between biomass and fossil feedstocks, with bulk energy density in particular, may significantly affect the scale at which biomass facilities can operate relative to oil refineries and coal-fired 124  power plants. In addition, the properties of the biomass fuel, such as moisture content and particle size, will have a large impact on the preferred logistics and choice of conversion technologies. This is analogous to coal properties dictating whether it is used for heat, electricity, steel, or coal-to-liquid applications. This theme focuses on maximizing the economies-of-scale for biorefineries, based on the process under consideration, feedstock characteristics (using logs and three intermediate feedstocks), and the known ‘best case’ logistics for feedstock handling and delivery.  4.2  Study Design and Assumptions The purpose of this chapter was to determine the maximum scale, from a logistical perspective, at  which a biomass-to-liquid biofuel facility might be constructed. The world’s largest ethanol plant, pulp mill, coal-based power plant, and oil refinery were used as models and compared for their capacity, with the primary metric for comparison being the energy content of the fuel entering the facility. Their logistical structures were analyzed and quantified, including identification of fuel delivery modes (road, rail, ship, pipeline, multi-modal), fuel source, and number of deliveries. Energy input to oil refineries was calculated using 6.1 GJ per barrel of oil equivalent (boe). For coal-fired power plants, the fuel energy content was estimated based on electrical output at 30% conversion efficiency and the assumption of bituminous coal with a LHV of 27.1 GJ t-1. For pulp and paper mills, pulp outputs were converted to energy inputs using a pulp yield of 50% (typical of a largescale Kraft paper mill), wood density of 440 kg m-3 [454] and a biomass LHV of 18.5 GJ t-1. For corn-toethanol, the plants’ output (in millions of litres) was divided by the 2005 industrial average yield of 410 l t-1 of corn. A LHV of 19.2 GJ t-1 was used for corn. A logistical analysis using a spreadsheet model was performed to determine the number of deliveries required to supply each facility with its yearly feedstock requirement. The delivery modes considered in the analysis were Panamax-size ships, North American rail, Swedish/Finnish logging trucks (gross vehicle weight limit of 60 t), and American trucks (gross vehicle weight limit of 36.3 t). A reasonable estimate, based on the required deliveries for the model facilities to operate at full capacity using their respective feedstocks, of maximum delivery and handling capacity for a single site was 125  established for each mode of transportation. Four feedstocks types of whole logs, wood chips, pellets, and biooil were considered and the amount of biomass in each form that could be delivered in a single year was calculated. Using data from the literature on advanced lignocellulosic biofuel yield, this biomass delivery amount was translated into a maximum capacity biofuel output.  4.3  Results The relative scale of the world’s largest ethanol, pulp, power (coal), and oil facilities were ranked  by the relative size of feedstock inputs, in terms of energy content (Table 4.2). It should be noted that these facilities were selected for size alone and as such, may be atypical in design or operation. For example, the Paraguana Refining Complex in Venezuela is actually two linked oil refineries (Amuay and Cardon) with shared feedstock supply systems and product shipping. The Aracruz Celulose S.A. Barra do Riacho pulp and paper mill, now owned by successor company Fibria, which consists of three fibre lines, is a multi-modal facility, with road, rail, and ship deliveries, and imports 46% of its eucalyptus plantation feedstock by ship [455]. This structure differs significantly from older pulp mills in the Northern Hemisphere, which tend to be land-locked and rely upon feedstock delivered exclusively by truck. Also, facilities may not operate at full capacity. For example, the Jilin Tianhe ethanol plant operates at approximately ¼ capacity and China has put a halt on development of new conventional generation biofuel facilities [456,457]. However, since the facilities were built under the assumption that they would operate at full capacity and the feedstock supply infrastructure was designed to handle feedstocks for full capacity operation, it was assumed that full capacity would be viable. The focus of the research Theme was on logistical feasibility, not plant construction and operation heuristics. Although the average plant size for all four industries studied is significantly smaller than these largest-of-type facilities, industrial trends towards larger plants that provide economies-of-scale make it worthwhile considering these capacities for logistical comparison. Lignocellulosic biomass from the forest, whether in its native form (logs), reduced to chips, or compressed into pellets, has a much lower bulk energy density than most fossil fuels. On a volumetric basis, bituminous coal (21.4 GJ m-3) contains twice as much energy as a similar volume of wood pellets 126  (10.0 GJ m-3), and four times as much energy as green (wet) logs (5.4 GJ m-3). Liquid fuels such as oil (36.7 GJ m-3) and gasoline (32.1 GJ m-3) are significantly higher in energy content on a bulk volumetric basis. Processed liquid biofuels (ethanol and biooil) are similar to coal in bulk energy density (Figure 4.1) [458]. Table 4.2 Largest-of-type processing facility comparison on feedstock energy input Facility type  Feedstock Largest by inputs  Feedstock input Mt Mm3 PJ  Estimated # deliveries by mode (%) Trucks Rail cars Panamax Pipeline  Ethanol  Corn, 15% MC Whole logs, 46% MC  3.4  4.7  56  8.6  15.8  86  75,695 (50%) 169,907 (45%)  18,966 (50%) 9,460 (10%)  21.0  25.2  541  --  49.7  56.9  2092  --  Jilin Tianhe [456] Pulp and Aracruz paper Celulose SA/Fibria Barra do Riacho [450] Electrical Bituminous Taichung generation coal, 5% generating MC plant [459] Oil refinery Crude oil, Paraguana 0% MC Refining Complex [460]  --  --  107 (45%)  --  --  393 (100%)  --  --  --  100%  Figure 4.1 Bulk energy density for various fuels  127  The efficiency of producing both ethanol and FT liquids from each of the four intermediates was estimated using assumptions and information from the literature (Table 4.3). It is important to note that these yield figures are on a wet basis, as required for this logistical analysis, and not a bone dry basis as has been discussed in previous thesis research chapters. It can be seen that the bioconversion process works best with wood chips at 34% moisture content, providing over 200 L per tonne of feedstock. More heavily processed intermediates, such as pellets or biooil, give lower ethanol yields than wood chips when used as feedstocks in the bioconversion process. Conversely, the production of FT-liquids rises with the density of the intermediates, as thermochemical processes tend to be better able to utilize the carbon within these feedstocks. Thus, it is estimated that biooil (at 25% moisture content) is the optimal feedstock for the thermochemical process, which can deliver over 250 L FT-liquids t-1 of biooil. Table 4.3 Estimated conversion efficiencies: intermediates-to-ethanol, intermediates-to-FT liquids Feedstock (Moisture Content)  Ethanol Yield (L t-1)  Logs (46%)  153  Chips (34%)* Pellets (10%)  Biooil (25%)  Notes  FT-Liquids Yield (L t-1)  Notes  Assumes 42% feedstock glucose content and steam pretreatment; Based upon [137,461]  97  Assumes FT biodiesel density of 0.78 kg L-1; Based upon [462,463]  208 151  Assumes 50% feedstock glucose content and steam pretreatment; Based upon [137,461]  125 160  57  Based upon [445]  254  Based upon [150,440,464]  *A moisture content of 34%, lower than for green chips, was chosen to reflect storage for a year in a temperate climate  4.3.1  Bulk Density and Transport Modes  The density and moisture content of both raw logs and the three intermediates is of particular interest given current legislation in North America and Europe, where road transport is limited by weight rather than volume [444]. Countries where forestry constitutes a significant part of the domestic economy, namely Sweden, Finland, and Canada, have higher gross vehicle weight (GVW) limits. Under European Commission Directive 96/53, member states are free to implement legislation allowing 60 t  128  internal transit, which has been tested in Netherlands, Denmark, and Germany [465,466]. Several U.S. states also allow increased loads within state (i.e. not interstate), including Michigan (74 t) [467] and Washington (47.8 t) [468]. Using typical payload weighting, and assuming a maximum volume of 160 m3, the maximum load bulk densities for trucks in several jurisdictions are calculated and shown in Table 4.4. Table 4.4 Truck weight limitations for selected jurisdictions Country  Gross vehicle weight limit (t)  Maximum net payload* (t)  Maximum bulk density** (kg/m3)  Source  United States 36.3 22.8 142.5 469 United Kingdom/European Union 44 30.5 190.6 470 Sweden/Finland 60 40.0 250.0 471 Canada 62.5 42.5 265.6 472 * Net payload is based on 62-69% of gross vehicle weight ** Maximum bulk density based on a maximum truck volume of 160 m3 [this is equivalent to an 8-axle b-train (one tractor + two trailers)]  The minimum bulk density that utilizes the entire volume of an 8-axle b-train is approximately 265.6 kg m-3 in Canada, and lower in other jurisdictions. The average bulk density of green (non-dried) woodchips can range from 300-400 kg m-3 depending upon moisture content. It is apparent, therefore, that truck shipping is limited by weight for high moisture content woodchips and not volume and further densification is unwarranted. It should be noted that not all jurisdictions allow b-train trucks due to their length, and so in some countries and urban areas, volume may become a limiting factor. Smaller-volume b-train trucks with higher clearance may also be required for many forestry roads, meaning volume could be limiting. However, over longer distances on highways in many jurisdictions, weight will dominate. Assuming weight is the limiting factor in transportation and the same vehicle curb weight across all products, it can be shown that the maximum energy load for U.S. Interstate trucks ranges from 228 GJ (green logs) to 1008 GJ (gasoline) (Figure 4.2). Wood pellets, while lower in bulk energy density compared to biooil, can result in a higher maximum energy per truckload due to the removal of water. Water removal is essential to maximize the energy payload in weight-limiting transportation situations.  129  Rail and ship transportation is primarily employed for longer distances than road. A typical rail car has a net payload capacity of 91 t. Specialized chip cars can reach a volumetric capacity of 210 m3, indicating a maximum bulk density of 433 kg m-3 and thereby making volume the limiting factor for chip transportation by rail [473]. However, weight is still the limiting factor for green logs, wood pellets, and biooil. Figure 4.2 Maximum energy load capacity for U.S. Interstate trucks  Using a Panamax (maximum size that can traverse the Panama Canal) bulk carrier with a net tonnage of 23,665 tons, or 66,500 m3 (net tonnage is volumetric measurement, where 1 ton = 2.83 m3) [474], with a 53,500 t cargo weight capacity [72,000 deadweight tonne capacity (dwt)] [475], volume becomes the limiting factor for shipping wood chips, logs, and wood pellets. This is because their bulk density ranges from 350 kg m-3 to 610 kg m-3. However, biooil, with a high bulk density of 1200 kg m-3, transportation is still limited by weight. These results are consistent with bulk shipping heuristics of rates based upon weight when the bulk density of cargo is greater than 800 kg m-3 and on measure (volume) when the bulk density of cargo is less than 800 kg m-3 [476]. 130  4.3.2  Estimating Maximum Deliveries  The maximum number of deliveries using each mode of transportation was determined by the maximum observed at each facility. By this method, the 170,000 U.S. trucks received by the Fibria (formerly Aracruz Celulose S.A.) Barra do Riacho pulp and paper mill is considered a maximum for truck deliveries. This equates to a truck delivery every three minutes. Kumar et al. (2003) stated that no fossil fuel processing facilties of industrial scale receive their feedstock by truck and therefore, coal power plants and oil refineries need not be considered as potential models for truck delivery [210]. The greatest number of railcars for the four model facilities is 19,000 for the Jilin Tianhe ethanol plant. However, this is not considered a maximum, given this facility is by far the smallest model facility here and larger coal-fired power plants exist that receive their fuel exclusively by rail. The largest U.S. coal consumer is the Scherer Power Plant in Juliette, Georgia. This facility uses approximately two thirds of the amount of coal consumed by Taichung [459]. Coal is delivered by train from Wyoming’s Powder River Basin, 2900 km away [477]. Approximately 154,000 railcars deliveries are required per year to supply the facility with coal. This is equivalent to 3 trains, of 141 railcars each, per day and has been considered a likely maximum for this study The coal-fired power plant is the largest facility of the four cases studied to receive all of its feedstock by ship. Therefore, 393 Panamax deliveries per year will be used as an estimate of maximum number of ship deliveries. It is worth noting that should all the oil be transferred onto ships at the Paraguana Oil Refining Complex for export, 930 Panamax ships per year would be required. However, much larger tanker ships would likely be used, such as Very Large Crude Carriers (VLCC), with 150,000320,000 dwt, or Ultra Large Crude Carriers, with 300,000-550,000 dwt [478]. This compares with the 60,000-75,000 dwt of Panamax. If the oil were to be exported from the Paraguana Oil Refining Complex using only 393 ships (as per the coal power plant), the average ship capacity would need to be approximately 170,000 dwt; a figure at the lower end of VLCC capacity. Therefore, 393 ship deliveries could be considered a reasonable maximum.  131  The maximum deliveries of each mode of transportation were used to determine the maximum scale, from a logistics perspective, of thermochemical (FT liquids) and biochemical (ethanol) advanced lignocellulosic biofuel facilities. The 4 different feedstocks of logs, chips, pellets, and biooil were considered, with capacity calculated using yields sourced from the literature (Table 4.5). Table 4.5 Potential maximum biofuel facility capacity Feedstock (Moisture content)  Inputs MT / year PJ / year  Logs (46%) Chips (34%) Pellets (10%) Biooil (25%)  3.9 3.9 3.9 3.9  39 48 65 61  Logs (46%) Chips (34%) Pellets (10%) Biooil (25%)  14.0 11.3 14.0 14.0  140 138 233 220  Logs (46%) Chips (34%) Pellets (10%) Biooil (25%)  14.2 9.1 15.8 21.0  142 112 264 331  4.4  Ethanol potential Ethanol Gasoline production equivalent (ML / year) (ML / year)  A. US Trucks 597 811 589 222 B. Rail cars 2,142 2,350 2,114 798 C. Panamax ships 2,173 1,893 2,386 1,197  FT potential FT Gasoline production equivalent (ML / (ML / year) year)  392 532 387 146  378 488 624 991  425 549 702 1,115  1,405 1,542 1,387 523  1,358 1,413 2,240 3,556  1,528 1,590 2,520 4,001  1,426 1,242 1,567 786  1,377 1,138 2,528 5,334  1,549 1,280 2,844 6,001  Discussion 4.4.1  Siting and Feedstock Supply  Previous studies have shown that the optimal size for a advanced lignocellulosic biofuel facility, particularly for thermochemical conversion, is typically greater than 1 GL yr-1 capacity [227]. The results of this current work show that this scale is feasible from a logistics perspective. For the bioconversion process, wood chips are the most attractive feedstock when truck or rail delivery modes are considered. Truck deliveries could supply enough wood chips to support a plant with a capacity of up to 800 ML ethanol yr-1, while the use of rail as the primary delivery mode could increase the maximum plant size to 2.3 GL yr-1. The most attractive feedstock for the bioconversion process using marine transit is wood pellets, due to the tradeoff between mass and density. A facility based on wood pellets at the maximum  132  number of ship deliveries could produce almost 2.4 GL ethanol yr-1, and thus would be very similar in scale to a maximized rail-only facility. As stated earlier, the thermochemical platform is capable of processing biooil in a more efficient manner than would a biochemical platform, and thus, it can take advantage of the density of this type of feedstock. The maximum size of a FT liquids production facility using biooil delivered by truck could be as high as 990 ML yr-1. It was apparent that rail delivery could increase the maximum size to 3.5 GL yr-1, while marine delivery using Panamax ships could support a 5.3 GL yr-1 facility. It is clear that feedstock characteristics and the ability of different platforms to process these feedstocks, combined with the transportation mode that could be used, will dictate the maximum scale at which a facility can operate. The final delivered cost of feedstock will be a function of the transportation distance and mode and will be combined with capital cost economies-of-scale to optimize the facility scale for a specific site. The use of Panamax ships allows dense feedstocks such as biooil to be transported most effectively. Thus, thermochemical platforms that can process biooil have a potential scale advantage over land-locked facilities that require delivery by truck or rail. Ultimately, multi-modal delivery is likely at the largest facilities. Only the smallest of the four largest-of-type model facilties considered, that of the Jilin Fuel Alcohol corn-to-ethanol plant, does not have ocean access. The Aracruz pulp mill, which may be the best representation of logistics for a large-scale, advanced lignocellulosic biofuel facility, receives biomass by road, rail, and ship [455]. This multi-modal logistics strategy enables the biomass managers to mitigate feedstock supply risk that would be associated with a single source area. It is apparent that a multi-modal delivery siting enhances a facility’s potential capacity, which is not possible with a typical, land-locked pulp plant that is dependent upon road transportation and a regional chip supply. High capital cost projects, such as biorefineries, will want to have a guarantee of a long term, stable, and low cost feedstock. British Columbia, where the Mountain Pine Beetle infestation will have killed enough lodgepole pine to provide approximately 1 billion cubic meters (500 Mt) of wood by 2014, provides a perfect example of feedstock availability but also of potential risk. These dead and dying trees 133  represent 77% of the lodgepole pine inventory of British Columbia and cover an area of 15 million hectares (ha), the combined area of Portugal and Denmark [159]. This type of local biomass feedstock disruption may become more common as the combined risks of fires, insect and other pest outbreaks, and extreme weather events, reshape biological systems [160]. In addition, facilities that rely upon annual feedstocks, such as crop residues, run the risk of having very little biomass available in a given year, as productivity will be impacted by shifting climates [158]. Although most previous techno-economic assessments have assumed a local feedstock supply due to the inherent low energy density, for the large facility capacities described here, this is a high-risk operational plan. As discussed in Chapter 1, recent market trends have shown an increase in the proportion of pulp production sourced from tropical and sub-tropical countries such as Brazil. This trend could also apply to lignocellulosic ethanol and FT liquids facilities. Higher growth rates mean lower feedstock, and hence production, costs. High growth rates also dramatically reduce the average trucking distance and draw area required to supply an advanced lignocellulosic biofuel facility. Lignocellulose-based liquid biofuel facilities supported by only truck deliveries are unlikely to exceed feedstock consumption of 4 Mt yr-1 due to delivery logistics limitations. The trucking distance involved to support the maximum number of deliveries has not been discussed in this chapter, but transportation distance is a primary model output described in Chapter 5. Countries such as Brazil, which can grow significantly more biomass within a smaller radius around a facility than in temperate regions, may be more likely to support advanced lignocellulosic biofuel production facilities at the larger scale and this is the primary research focus of Chapter 5. If temperate countries such as the U.S. give domestic fuel production priority regardless of feedstock source, importation of biomass feedstocks from high productivity regions, such as Brazil, may be considered a possibility. 4.4.2  Transportation Mode Limitations  Whether transportation of the feedstock is weight or volume limited has significant implications for the preferred mode/feedstock combination, potential maximum scale of the facility, and technology selection. Ship transport is volume limited for all feedstocks except biooil, which is weight limited. 134  Three times the energy can be delivered in a biooil form as compared to wood chips using the same number of ships. However, when transporting by trucks (assuming a large volume b-train), which are weight limited for all four feedstocks, the advantages of densified intermediate feedstocks biooil and pellets are substantially reduced. Only 1.3 times the amount of energy in chips can be transported in a U.S. Interstate truck when it is in the form of biooil. Given the processing energy and additional costs associated with pellet and biooil production, the moderate truck transportation benefits provided by these intermediate feedstocks may not justify their production. While biooil may seem to be the ideal intermediate for thermochemical conversion to liquid biofuels, particularly for large, port-based facilities, there are signficant challenges associated with its production and logistics. Biooil is typically highly acidic with a high water content and can therefore not be transported using traditional oil fuel trucks, railcars, and tankers [440]. Its transportation would need specially designed, dedicated equipment. In addition, companies do not currently produce biooil in supplies sufficient to justify even one dedicated tanker [432]. Therefore, although pellets are lower in bulk energy density than biooil, they are currently the most likely feedstock for advanced lignocellulosic biofuel facilities with outputs greater than 1 GL yr-1. When comparing oil refineries and coal power plants, both of which employ well-proven technology and have had extensive time to scale-up, it is noteworthy that the largest oil refinery is approximately four times the size of the largest power plant. A primary reason for this is that the feedstock (oil) for the refinery arrives by pipeline. However, the refinery products (namely gasoline and diesel/heating oil) are exported by ship. The Paraguana Refining Complex, located on the coast and representing greater than 70% of Venezuela’s refining capacity [479], is serviced by over 10 berths for large tankers that enable the refinery to operate at this scale. The liquid form of both the fuel and product is a defining characteristic that allows these economies-of-scale to be realized. A thermochemical biofuel facility relying upon biooil as the feedstock was the largest capacity facility (5.3 GL) described in Table 4.4. An 8.4+ GL FT liquids production facility utilizing biooil as a feedstock in a distributed processing system was also the largest facility contemplated by previous studies 135  [150]. However, this scale could be increased by the utilization of pipelines supplying biooil to a central processing facility [441]. Although the significant capital cost would likely make this industrial structure unlikely over a large geographic area, biooil pipelines could potentially be used to connect two or more processing facilities, as is the case with the Paraguana Refining Complex. Thermochemical conversion, which is more similar to traditional fossil fuel processing than is the biochemical conversion option, benefits more from economies-of-scale due to technology complexity (and hence large capital costs) and the ability to accept a dry, compacted pellet feedstock [46,226]. Biochemical facilities, which have relatively poor economies-of-scale, are likely to be much smaller and will require less feedstock [365,480]. Unlike thermochemical facilities, they do not benefit significantly from importing biomass by ship due to the preference of low bulk density (and hence volume limited) wood chips and whole logs as feedstocks. The conversion process is much more akin to a pulp and paper mill or conventional ethanol plant, which provide a means of scale comparison. The results of the feedstock comparison indicate the logistical benefits of utilizing pellets are overwhelmed by poor hydrolysis yields in biochemical facilities. However, further work on biochemical conversion of pellets to ethanol is required to quantify, and potentially overcome, the full impacts of drying on biomass hydrolysis and subsequent ethanol production.  Research at the University of British Columbia is  currently being conducted to address this knowledge gap. The risk of depending on local feedstocks may negatively affect the development of inland biochemical facilities, although the ability to operate at a smaller scale, combined with an effective feedstock management plan, will partially offset this risk. If facilities must be located inland, such as is the case for most crop residue-based facilities, biochemical conversion has an advantage due to the ability to operate economically at a smaller scale [227,356]. In addition, the ability to import feedstocks by rail will reduce the feedstock supply risk and make the construction of substantially larger scale facilities more attractive.  4.5  Conclusion From a logistics perspective, the analysis in this chapter shows that it is possible to deliver forest-  based lignocellulosic biomass in sufficient quantity to supply advanced lignocellulosic biofuel production 136  facilities such that they can be substantially larger than existing conventional biofuel facilities. Wood chips seem to be the optimal woody feedstock for bioconversion facilities, supporting plants up to a maximum scale of 800 ML ethanol yr-1 when supplied by U.S.-sized trucks, or as much as 2.3 GL ethanol yr-1 with chips delivered by railcar. Other than feedstock supply diversity, there may be no major advantage in moving to marine delivery for biochemical facilities because this type of platform may be relatively poor at processing densified intermediates biooil and wood pellets. However, thermochemical facilities can make use of densified intermediate feedstocks, such as biooil and pellets, which, because of their higher bulk energy density, are able to support larger facilities than when using undensified feedstocks.  It is possible that a biooil feedstock could support a maximum potential scale for  thermochemical facilities ranging from 990 ML yr-1 (for truck deliveries) to 3.5 GL yr-1 (supported by rail) to as large as 5.3 GL yr-1 (supplied by Panamax ship). However, wood pellets are likely to remain as an intermediate feedstock for thermochemical facilities in the near to intermediate term, given the constraints of limited biooil production and transportation infrastructure. While commercial advanced lignocellulosic biofuel plants in the first wave of development are expected to be much smaller, in the 50-70 ML yr-1 capacity range [231,381], economies-of-scale will likely drive a rapid increase in largest-of-kind capacity. This will make a notable contribution to reducing the MESP of lignocelluloisc ethanol, as shown in Chapters 2 and 3. The combination of technology platform, feedstock characteristics, and transportation mode will dictate the scale of biorefinery operations. Thermochemical conversion may be concentrated in coastal regions, due to the ability of this platform to process dense, low moisture content intermediates. These facilities will be able to utilize multiple modes of delivery, including shipping, in order to reduce supply risk. Bioconversion facilities will likely be significantly smaller, and sited primarily to maximise access to “local” biomass (agriculture and forestry residue, logs, wood chips) sources using road and rail corridors. It is also apparent that increasing the scale of biofuel production facilities beyond that of conventional ethanol and current pulp plants is possible logistically, but it will require importing biomass from other regions by rail or ship. This is particularly true for temperate forest-dependent facilities, 137  where relatively low growth rates mean longer transportation distances and a much larger impact of disturbance on the lifetime economics of a facility. Similarly, dependence on an annual feedstock, where a poor year can leave little to no crop residue available, can also be problematic. Thus, it is likely that production of biomass feedstock intermediates, namely pellets and biooil, will have to be increased substantially in order to allow scale-up of advanced lignocellulosic biofuel facilities. Even with densification and increased ease of handling of intermediates relative to raw biomass such as logs, bioenergy is at a fundamental disadvantage compared to fossil fuels. Biooil, the most energy dense of intermediates considered here, is still half as energy dense as crude oil. The moisture content of biomass must be reduced to be transported without becoming weight-limited. Biomass is also more geographically dispersed and intermittently available (due to seasonality) than are fossil fuels, reducing supply reliability.  Efficient feedstock preprocessing, transportation, and logistics will be  essential to the economic viability of future lignocellulosic ethanol facilities. Although logistical lessons can be learned from the fossil fuel-based energy system, it cannot be replicated due to differences in material energy density, flowability, and distributed availability.  Therefore, lignocellulosic ethanol  facilities will not resemble oil refineries in terms of feedstock delivery, management, or handling and the industrial production model will likely be more akin to what is currently used in the forestry and agricutural sectors. This thesis theme has shown that economies-of-scale for lignocellulosic ethanol facilities utilizing forest resources can reach 800 ML yr-1 when supplied using a trucking logistical system. Now that this has been identified, the question becomes one of minimizing those delivered feedstock costs, along with other contributers to MESP identified in Chapters 2 and 3, by selecting the most appropriate and economically competitive site for facility establishement and operation. A comparative site analysis is required to identify the key factors influencing the attractiveness of a site and this is presented in the following chapter.  138  5 5.1  COMPETITIVE ADVANTAGES IN SITING FOR CANADIAN FACILITIES Introduction As presented in Chapters 2 and 3, lignocellulosic ethanol producers will require every advantage  possible when trying to compete with gasoline and conventional ethanol produced from corn and sugarcane. Site selection may be a key component in maximizing chances of success for lignocellulosic producers. Poor site choice may be an ongoing strategic disadvantage for those firms that do not consider all variables in selection since once a choice is made and the capital invested, it cannot be readily changed. In addition, from a Canadian perspective, it is important to determine the relative competitiveness of domestic production compared to international producers. This will have significant influence on the impact of policy choices, such as volumetric mandates, excise tax exemptions, and import tariffs (see Sections 1.2.4.4-1.2.4.6). For businesses comparing jurisdictions for manufacturing or chemical processing operations, labour is typically the most significant contributor to location-specific costs [481]. However, biomass, due to its low energy density, is not traditionally transported long distances without preprocessing and therefore feedstock (raw material) is a location-specific cost component critical to the biofuel production industry [215]. Since feedstock is the largest variable and operating cost for a lignocellulosic ethanol facility, as described in Chapters 2 and 3 and other sources [46,344,363], regardless of the technology platform used, it is also the largest contributor to location-specific costs [204]. Other major locationspecific biofuel production cost components include raw material and labour transportation cost, cost of capital, cost of construction, income and other taxes, utilities (energy and water), and government and insurance fees [481]. When determining the ability to compete in a target market, physical proximity to that market is an essential consideration for large-volume commodity producers, and hence transportation cost of product(s) is an additional factor [482,215]. Finally, the exchange rate must be considered if intercountry trade is involved, which affects all cost components and can greatly impact competitiveness of production [483]. 139  In Chapter 4, it was shown that advanced lignocellulosic biofuel facilities are likely to resemble pulp mills in terms of scale, feedstock supply, and logistical structures. Therefore, it is worthwhile considering trends in the pulp industry for implications and lessons for advanced lignocellulosic biofuels, particularly those utilizing woody feedstocks. Of particular importance is the general trend of a greater percentage of pulp supplied by tropical and subtropical producers at the expense of more temperate producers. Information on pulp production trends is presented in Chapter 1. Although labour savings contribute to the lower cost production in countries such as Brazil and Chile, the greatest factor by far is cost of feedstock [484]. Eucalyptus can be grown in 7-year rotations in Brazil [161], while hardwood and softwood rotations in Canada range from 50-90 years. Brazil is already the world’s largest short-fibre pulp producer, has risen to the third largest producer overall, and is expanding rapidly [485,486]. As described in Chapter 4 and other sources, the ability to utilize high productivity eucalyptus has enabled Brazilian pulp and paper mills to achieve much greater economies-of-scale than their northern hemisphere counterparts [487]. While the largest pulping complex in Brazil exceeds 2.3 million tonnes (Mt) pulp per year and the largest individual mill is 1.5 Mt, the average mill capacity in Canada is 200,000 tonnes (t) pulp [451,452,453,487]. Pulp companies in Brazil and other tropical nations benefit from being vertically integrated; owning their own land, growing and harvesting their own fibre, and then processing at their mills. Comparatively, companies with operations in temperate and subtropical climates have little or a negative financial benefit for landownership and vertical integration due to low productivity [488]. Even in the relatively high productivity U.S. south-east, land expectation values do not support investment in land purchase [488]. In other words, it is lower cost to purchase fibre at market value than could be produced by the pulp companies themselves. Not controlling feedstock directly poses a substantial operating risk for a large biorefinery requiring potentially millions of tonnes of biomass per year. Nevertheless, this multi-million tonne per year capacity, which enables large economies-of-scale and lower production costs, is the competition faced by Canadian pulp mills and biorefineries.  140  The difficulty of temperate pulp companies to compete has forced Canadian pulp and forestry companies to investigate alternative or co-products such as lignocellulosic ethanol [489]. However, the question of forest sector competitiveness must apply to both pulp and new products such as ethanol. Given the push towards advanced lignocellulosic biofuels in countries such as Canada and the U.S., it must be determined whether domestic production is economically competitive in a world with global trade. As a renewable resource, it is possible to dramatically increase production of biofuels worldwide if justified by market prices and demand (see Section 1.2.2.1). Therefore, the purpose of the research described in this chapter was to determine the ability of domestically-produced lignocellulosic ethanol to compete economically with the same fuel produced in tropical regions for markets in Canada. This is largely a theoretical question, since it is deemed unlikely that tree plantations in tropical regions will be developed in order to support lignocellulosic ethanol production – largely due to the ease and low risk of production from sugarcane. However, relative costs of lignocellulosic ethanol production from woody biomass between Canada and tropical regions will be a strong indicator of enterprise competitiveness and the potential effects of policies supporting or subsidizing ethanol production from forest resources.  5.2  Study Design and Assumptions The primary theme in this chapter is identification of competitive advantages in production and  delivery of lignocellulosic ethanol. The metric for comparison in this analysis was the cost including freight (CIF) minimum ethanol selling price (MESP) of ethanol delivered to two Canadian markets – Montréal and Vancouver – the second and third largest cities in Canada respectively. Both cities have year-round port access for accepting Panamax-sized ships, a key criterion for the study. To undertake the comparative analysis, the creation of six lignocellulosic ethanol production scenarios – two at each of three potential production sites – was required.  These are presented in Table 5.1 and utilize the  lignocellulosic ethanol production techno-economic spreadsheet model designed for the analyses in Chapters 2 and 3.  The assumed processing chain was steam explosion pretreatment, followed by  enzymatic hydrolysis and fermentation by S. cerevisiae (yeast). Steam explosion was chosen to be the assumed pretreatment due to low operating and capital costs relative to other pretreatment technologies, 141  as presented in Chapter 3 [326]. The scale chosen for modelling was 800 ML ethanol per year, based upon the results on maximum scale using feedstock truck delivery presented in Chapter 4. This permits maximization of economies-of-scale on capital expenditures and is mid-range compared to the optimal scale of a lignocellulosic ethanol facility identified in Chapter 4 and previous studies [146,204,227,290,356,449]. Based upon the results presented in Chapter 3, a CHP plant operating on lignin and the unhydrolyzed solid cellulose fraction from the ethanol production process is co-located with the ethanol facility. The facility meets all heat and electricity requirements for ethanol production and excess electricity is sold to the grid. This is used to reduce net electricity cost and provide zero-cost process heat to the facility. Base case assumptions for the techno-economic model, which require adjustment based upon local conditions analyzed in this chapter, are presented in Table 5.2. Table 5.1 Overview of study scenarios Scenario BC-Forest BC-Poplar ON-Forest ON-Willow BC-Imports QC-Imports  Plant Site Williams Lake, British Columbia Williams Lake, British Columbia Prescott, Ontario Prescott, Ontario Barra do Riacho, Brazil Barra do Riacho, Brazil  Delivered Fuel Site Vancouver  Feedstock Mixed Softwoods  Vancouver  Poplar  Montréal Montréal Vancouver Montreal  Mixed Hardwoods Willow Eucalyptus Eucalyptus  The sites chosen for comparison were Prescott, Ontario; Williams Lake, British Columbia; and the Barra do Riacho port, Brazil. Prescott, Ontario is a town of 4,200 people located on the St. Lawrence Seaway 350 km northeast of Toronto and 190 km southwest of Montréal. It is home to the only deepwater port between these two cities and is on both the major trucking and rail routes between Canada’s two largest cities. The newly established Ontario East Wood Centre, which is intended to become a bioenergy/bioproduct industrial park, is located at the Port of Prescott. Feedstock is assumed to be sourced from both Canada and the U.S. (upper New York State). Although the Prescott site is located at a port, it is in close proximity to the Prescott-Ogdensburg International Bridge and therefore no additional cost for water transportation is included. However, in some cases, it may be economically competitive to import feedstocks by barge. Williams Lake is located in the central interior of British 142  Columbia (Cariboo region), 550 km northeast of Vancouver, and has a population of 10,750. The region includes both interior-Douglas fir and Sub-Boreal Pine-Spruce biogeoclimatic zones [490]. Its primary industry is forestry and is home to North America’s second largest biopower plant, at 66 MWe. The University of British Columbia’s Alex Fraser Research Forest, a 9,802 hectare (ha) parcel, is located not far from Williams Lake. Barra do Riacho, owned by Fibria, is the world’s largest pulp production complex, with a total capacity of 2.3 million tonnes of air-dried pulp per year [451]. Co-located with the Portocel shipping facility, Barra do Riacho is on the Brazilian coast 90 km northeast of Vitória and 600 km northeast of Rio de Janeiro. The port-based location (and hence reduction in surrounding land area by approximately 50%) was taken into account in the analysis.  Feedstock trucking assumptions are  presented in Table 5.2.  143  Table 5.2 Base case techno-economic facility and delivery model assumptions Facility Overview Plant Capacity Plant Operating Factor (uptime) Plant Capacity Factor Unit Installed Cost Total Capital Cost Financing Risk Free Cost of Capital Amortization Period Capital Cost Allowance/Depreciation Non-feedstock Operating Costs Labour (workers per shift) Maintenance Stores/Supplies Electricity/Process heat Enzymes  800 ML yr-1 0.90 0.95 $1.48 L-1 yearly capacitya US$1,185 M 2.8% 20 years 4% [386] 24b 3% [363] 2% [353,363]c 0.9 kWh L-1 [353] / 18 MJ L-1 $2.40 kg-1 protein (Chapter 3)d; 600 FPU g-1 protein [356,357,358]; 20 FPU g-1 cellulose [136,138,388]  Feedstock Delivery Fuel Cost $1.30 L-1 Fuel Consumption 2.0 km L-1 [492]e Net Truck Load 42.5 tonnesf Average Trucking Speed 60 km hr-1 Days Active; Time in Day Active 90%; 80% Truck Capital Cost $120,000 [491] Trailer Capital Cost $50,000 Equipment Useful Life 1,500,000 km Cost of capital 8% Amortization Period 4 years (assuming 380,000 km yr-1) Driver Wages Same as facility worker (see Labour below) Other Operating Costs (repairs, maintenance, $0.593 km-1 [491] insurance, licensing, permits, overhead) and maintenance Co-located Co-generation Facility Gross Electrical Capacity Feedstock dependent (base case 112 MWe)g Electrical efficiency 20%h Heat Efficiency 75% Net : Gross Electrical Capacity 0.90 Plant Operating Factor 0.90 Plant Capacity Factor 0.95 Unit Installed Cost $2450/kW i (@ 112 MWe) Total Capital Cost $274 M (@ 112 MWe) Operating Costs 4% of capital cost a Installed yearly capacity cost based upon a survey of announced cellulosic ethanol facilities, summarized in Chapter 2. A listing of projects was compiled by Bacovsky et al. (2010) [385]. Calculated using $2.50 L-1 yearly installed capacity for 58.5 ML facility and scaling factor of 0.8 b Based upon 6 workers for a 50 ML yr-1 facility [363] and a labour scaling factor of 0.5 c Stores and supplies includes process chemicals such as sulphuric acid and/or sulphur dioxide (pretreatment catalyst); sodium hydroxide (pH balance and lignin-cellulose separation); diammonium phosphate (yeast nutrient); sorbitol (yeast culturing) [353,363] d Calculated using the equation 3.2 in Section 3.3.3 and assuming 20 FPU g-1 cellulose, 600 FPU g-1 protein, and contribution to MESP of $0.132 L-1 [362]. This is a figure consistent with protein for amylase, but high compared to estimate of Humbird et al. (2011) [363]. Should loading be 10 FPU g-1 cellulose, activity 600 FPU g-1 protein, and contribution to MESP be $0.132 L-1, protein would be $4.40 kg-1. This is consistent with Humbird et al. (2011) [363]. e Based upon Canadian heavy truck average of 2.9 km L-1 and reduced by 30% to account of added weight of b-train [492] f Based upon gross vehicle weight (GVW) of 62.5t and curb (unloaded) weight of 20 t for b-train chip truck g Assumes 25% lignin, 42% cellulose feedstock, and 2.05 M bdt feedstock  144  h  Relatively conservative electrical efficiency is due to prioritization of process heat. Base case facility would require 12,312,000 GJ of process heat. With 75% heat conversion efficiency and process residue (lignin & unhydrolyzed cellulose) average energy content of 23.4 GJ t-1, 700,800 t of residue are required. i Based upon a survey of announced biopower facilities, summarized in Chapter 2  Two of the six scenarios, BC-Forest and ON-Forest, utilize existing domestic biomass resources as feedstock for ethanol and electricity production, servicing local markets – Williams Lake for Vancouver and Prescott for Montréal. The other two domestic scenarios, BC-Poplar and ON-Willow, focus on purpose-grown short-rotation woody crops. In Prescott, these would largely be grown on marginal agricultural land, while in Williams Lake, they could be grown in areas where the Mountain Pine Beetle epidemic has wiped out Lodgepole pine stocks. The two imported ethanol scenarios, BCImports and QC-Imports, utilize eucalyptus as a feedstock, grown in 7-year rotations intended to maximize productivity. With a common facility design and operation across all three sites, the focus of the comparative study was the site-specific variables impacting the CIF MESP. These included delivered feedstock type and cost; net electricity cost (as dictated by feedstock and local pricing); ethanol delivery to market; labour; cost of capital; construction; utilities (energy, water); permits and insurance; income, property, and other taxes; and exchange rate. Spreadsheet trucking models were created to determine the delivered cost of feedstock for each scenario, taking into account harvest area and local conditions. Assumptions are presented in Tables 5.4 (harvest) and 5.5 (delivery) and are based upon data from the literature. All financial inputs were adjusted to 2010 Canadian dollars using the United States Consumer Price Index (CPI) and an exchange rate of 1:1 for US dollars to Canadian dollars, 1.6 for Brazilian Reals to Canadian dollars, and of 1.4:1 Canadian dollars to Euros unless otherwise specified. Ethanol yield is assumed to be 70% of theoretical yield from all sugars. Theoretical and assumed yield for all feedstocks is presented in Table 5.3. Douglas fir is assumed to be a representative species for scenario BC-Forest and Red maple is assumed to be a representative species for scenario ON-Forest. Lignin and unhydrolyzed cellulose are assumed to be the co-generation facility fuel, with lignin recovery at 95% and unhydrolyzed cellulose at 25% (following steam explosion of softwood, the fraction of 145  feedstock components recovered in the solid fraction is typically 70%, including 95-100% of initial lignin and 70-90% of initial glucan) [388]. Table 5.3 Feedstock content and theoretical and assumed ethanol yields Feedstock  Glucan (kg/bdt)  Mannan (kg/bdt)  Galactan (kg/bdt)  Xylan (kg/bdt)  Arabinan (kg/bdt)  Lignin (kg/bdt)  Theoretical Yield (L/bdt; all sugars)  Theoretical Yield (L/bdt; C6 only)  Facility biomass consumption (bdt yr-1)  421  Assumed Yield (L/bdt; 70% of all sugars) 321  Douglas fir [369]a Poplar [493] Red Maple [494]b Willow [495] Eucalyptus [496]b  430  128  26  39  13  285  458  403  31  7  176  6  259  452  318  316  2,164,557  466  35  6  173  5  240  497  365  348  1,965,517  430  32  20  149  12  266  466  348  326  2,098,159  461  4  15  171  8  245  478  346  335  2,041,791  2,130,841  a  10% Bark Debarked  b  5.3  Results The site-specific variables are presented by subsection, with a summary in Table 5.6. 5.3.1  Feedstock Characteristics and Delivered Cost  Delivered feedstock costs were compared on a bone dry tonne (bdt) basis, with the assumption that wood is delivered in green form. It was assumed that all firms are vertically integrated and feedstock is supplied to the biorefinery at cost, regardless of harvest scheme. Therefore, costs presented here do not necessarily reflect market prices. Land rental, establishment, silviculture, stumpage, harvest, chipping, and associated costs (e.g., road building) were sourced from the literature.  The harvesting and  transportation regime selected for each scenario was based upon heuristics and the most likely real-world approach given existing methods. It was assumed harvest residues would be insufficient in volume to supply the facility for scenarios BC-Forest and ON-Forest and that whole-tree logging would be required. It was assumed that all feedstocks are chipped on-site and transported to the biorefinery in chip form. Scenario ON-Willow was based upon a direct-chip harvesting regime, in which a specially-designed willow header was used to harvest and chip the willow in a single pass, as reported by Tharakan et al. (2005) [497]. Both eucalyptus scenarios replicated the existing harvesting standard of a cut-to-length  146  harvester with debarking and bucking head and a forwarder for transportation to roadside [498]. A $15 bdt-1 chipping cost was assumed for all Canada scenarios with the exception of ON-Willow, and a $10 bdt-1 chipping cost is assumed for Brazil scenarios [201,210]. Harvest and preprocessing costs can vary significantly from site-to-site within the same region. Reasonable assumptions, based upon published literature, industry organization trials, and interviews with experts in the field, used for this analysis are presented in Table 5.4. Average transportation distance was calculated using the following equations. The average transportation distance from every point of a circle to the centre is given by:  d=  2 r 3  (5.1)  Where, d is the average transportation distance for a circle and r is the radius of that circle. However, real-world transportation distance is not straight line and is dictated by the tortuosity (τ), or bendiness, of a road network [426]. Therefore, average transportation distance can be given by:  d =τ  2 r 3  (5.2)  Or when using a Cartesian coordinate system and two (x,y) points:  [  d = τ (xi − x j ) + ( yi − y j ) 2  ]  1 2 2  (5.3)  For this study, an assumed tortuosity factor of 1.4 was used to calculate the average transportation distance for Ontario and Brazil scenarios. A tortuosity factor of 1.5 was used for the British Columbia scenarios to account for the rugged terrain and undulating road network. These figures are consistent with previous studies [213,426,513]. The harvest area and delivery requirement, including average transportation distance, are presented in Table 5.5. The cost of delivery is based upon the transportation model inputs presented in Table 5.2 and the average transportation distance. The truck power unit is assumed to be 9.5 t, b-train trailers 10.5 t, and the Gross Vehicle Weight (GVW) 62.5 t [472,491,514]. As detailed in Chapter 4, the maximum truck payload for both chips and logs was therefore 42.5 tonnes and with 50% moisture content wood chips, trucks legally max out on weight before volume in all cases. 147  Table 5.4 Feedstock harvest assumptions Scenario  Site  Feedstock  Land ownership  Rotation  Public  Average yield (bdt/ha-yr) 3  Harvest regime  Equipment  70  Yield at harvest (bdt/ha) 210  BCForest  Williams Lake, British Columbia  BCPoplar  Williams Lake, British Columbia  Mixed softwoods dominated by lodgepole pine Poplar  Public  ONForest  Prescott, Ontario  Mixed Hardwoods  ONWillow  Prescott, Ontario  Willow  BCImports  Barra do Riacho, Brazil  Eucalyptus  QCImports  Barra do Riacho, Brazil  Eucalyptus  Harvest & chipping cost to roadside ($/bdt) 85  Source  Whole tree harvest and chipping  Feller-buncher; grapple skidder; whole-tree chipper  6.3  20  126  Whole tree harvest and chipping  Harvester; cutto-length and debarking head; forwarder; whole-tree chipper Feller-buncher; grapple skidder; whole-tree chipper  75  [499,500]  Public and private; separate land owner and operator Public and private; separate land owner and operator Private; vertically integrated  3.2c  60  192  Whole tree harvest and chipping  65  [500,501, 502, 503, 504]  11.25  3  34  Direct chip harvesting  Direct-chip harvester with willow header  55  [497,500,505]  15.6  7  109  Cut-to-length and debarking; whole tree chipping  35  [488,498]  109  Cut-to-length and debarking; whole tree chipping  Harvester; cutto-length and debarking head; forwarder; chipper Harvester; cutto-length and debarking head; forwarder; chipper  Private; vertically integrated  15.6  7  35  [488,498]  [338,339, 340,341,342]  c  Based on stem MAI of 5.3 m3 ha-yr-1, with a 20% allowance for branches and tops [501]  148  Table 5.5 Feedstock delivery parameters Scenario  Required yearly input (bdt)  Harvest Yield (bdt ha-1)  Yearly Rotation Land in Percentage Total Radius Tortuosity Average Average harvest (yrs) production of cut draw area (km) transportation delivery area (%) allocated (ha) distance (km, costi ($ (ha) one way) bdt-1) a f BC1 2,130,841 210 10,147 70 30.0% 20% 11,838,006 194 1.5 194 40.38 BC2 2,164,557 126 17,179 20 6.0%b 100% 5,726,341 135 1.5 135 29.62 ON1 1,965,517 192 10,237 60 17.0%c 60%g 6,021,805 138 1.4 129 28.52 ON2 2,098,159 34 61,711 3 4.5%d 100% 4,114,037 114 1.4 107 24.51 BR-BC 2,041,791 109 18,732 7 12.0%e 100% 1,092,702 83h 1.4 78 16.36 BR-ON 2,041,791 109 18,732 7 12.0%e 100% 1,092,702 83h 1.4 78 16.36 a Based upon 60% of land in productive forest [506] and 50% of forest providing feedstock b Based upon allocation of 10% of forested land to poplar plantations c Based upon 34% forested area [507] and 50% of forest providing feedstock d Based upon 10% of agriculture land [508] e Based upon 35% of land non agricultural nor pasture use [509] and a ratio of 1.7:1 eucalyptus plantation to native forest management plan [510]. Additional 10% allowance for urban and water areas. f Based upon 50% of fuel wood component of saw log/fuelwood/non-harvestable rate of decay over 20 years for mountain pine beetle infected stands [511] g Based upon pulp log mix of 50% [512] and 80:20 saw log: residue ratio for remaining 50% h Takes into consideration port location (50% of surrounding area considered water) i Includes $5 bdt-1 for loading/unloading  149  5.3.2  Cost of Ethanol Delivery  For the Williams Lake and Prescott scenarios, ethanol is transported by rail to the ports of Vancouver and Montréal respectively. It was also assumed that the ethanol plants have a dedicated siding for ethanol export. The maximum cargo payload for a typical rail car is 91 t [473]. With an ethanol density of 0.789 t m-3, the maximum load is 115 m3 or 115,000 litres. For the Brazilian (Barra do Riacho) scenarios, a liquid bulk Panamax (maximum size that can traverse the Panama Canal) vessel was assumed for transportation. This vessel has a net tonnage of 23,665 tons, or 66,500 m3 (net tonnage is volumetric measurement, where 1 ton = 2.83 m3) [474], with a 53,500-55,000 t cargo weight capacity [72,000 deadweight tonne capacity (dwt)] [475]. In the case of ethanol, the vessel will max out on volume before maxing out on weight, which is consistent with bulk shipping heuristics of rates based upon weight when the bulk density of cargo is greater than 800 kg m-3 and on measure (volume) when the bulk density of cargo is less than 800 kg m-3 [476]. Delivery is assumed to be cost, insurance, and freight (CIF), which includes delivery to a port location but not handling beyond the transportation vessel/car boundaries. Ethanol blending and storage costs are not included as they are considered to be consistent for all scenarios. The rail transportation cost for a carload of ethanol from Williams Lake to Vancouver is assumed to be $2226 per car or $0.019 L-1, while from Prescott to Montréal is assumed to be $1715 per car or $0.015 L-1. Both cost estimates were calculated using CN Rail’s Carload Price eBusiness tool [515]. To determine shipping rates from Brazil, it was assumed that shipping cost is composed of two components: vessel charter cost and bunker fuel cost. Shipping cost can be given by the equation: 𝐶𝑠 = (𝐶𝑐ℎ × 𝑡) + �𝐶𝑓 × 𝑛 × 𝑑�  (5.4)  Where C s is the total cost of shipping, C ch is the charter cost per day, t is the charter time in days, C f is the cost of bunker fuel per tonne, n is the tonnes per nautical mile, and d is the distance in nautical miles. Charter cost, a good indicator of overall economic activity, is highly volatile and driven by an inelastic supply but elastic demand. From 1990 to 2005, the mean rate for Panamax bulk ships was 150  US$11,552 with a standard deviation of US$7,485 [516]. In the past five years, this volatility has been increased, peaking in 2008, dropping by over 60% in 2009, and rebounding in 2010 [517]. However, a longer-term sustainable rate may be partially determined by futures on the Baltic shipping indexes. Based upon these futures, it was assumed that the daily charter rate for a Panamax ship is CA$20,000 per day. The Panamax ship averages 13.6 knots (25.2 km hr-1; 15.65 mi hr-1) and uses 30 tonnes per day of bunker fuel, or 20 km t-1 (12.5 mi t-1) [516]. The distance from Vitória, BR to Montréal, CA is 5,224 nautical miles and from Vitória, BR to Vancouver is 8,047 nautical miles, making trip times 13.9 days and 21.4 days respectively. Assuming 1.5 days for loading and 1.5 days for unloading, the total charter times are 16.9 days for Vitória-Montréal and 24.4 days for Vitória-Vancouver. Cost of fuel is based upon the Bloomberg 380 Bunker Index and was assumed to be CA$675 per tonne [518]. Based upon these figures, the cost for transportation of a Panamax load of ethanol from Vitória to Montréal is CA$620,096 or $0.009 L-1. The cost from Vitória to Vancouver, with a $93,110 toll for the Panama Canal [519], is CA$975,648 or $0.015 L-1. 5.3.3  Labour  The cost of labour was calculated on an hourly basis based upon data from the Division of International Labor Comparisons of the U.S. Department of Labor. Hourly manufacturing labour costs were determined for direct pay, directly paid benefits, and social insurance in the local currency for 2009, with conversion from Brazilian reals to Canadian dollars at an exchange rate of 1.6 [520]. Total cost of labour in the manufacturing sector in Canada for 2009 was estimated at $33.78 per hour, with direct pay accounting for 80% of the cost. The Brazilian cost of labour was R$16.64, or CA$10.40, with direct pay accounting for 68% of the cost [521]. 5.3.4  Cost of Capital  Systematic risk, defined as risk that is not diversifiable, can vary significantly by country [522]. This can be termed ‘country risk premium’ and the expected returns of investors are commensurate with the amount of risk posed by investment in any particular country. A higher level of risk requires greater premiums for investment, and hence, cost of capital. Typically, investments in developing nations pose a 151  higher country risk than developed nations due to differences in financial, social, and political stability and other factors which may adversely affect the operating profits or asset value but which are largely out of the control of the business operators [522]. Euromoney publishes a country risk index, with a rating out of 100. The highest ranked is Norway at 93.44, while Canada ranks in the top 10 at 86.35. Brazil has a rating of 63.53 [523]. Debt ratings seek to define the likelihood of default, and hence risk. Canada has a Moody’s Aaa rating, the highest possible, while Brazil has a rating of Baa3 [524]. The capital asset pricing model, which is used to determine the appropriate rate of return of an asset, takes into account both systematic (undiversifiable risk) and unsystematic risk (diversifiable risk), the expected rate of return of the market, and the theoretical return on a risk free asset. The CAPM systematic risk-adjusted return is given by: 𝑅𝑖 = 𝑅𝑓 + 𝛽�𝑅𝑚 − 𝑅𝑓 � + 𝑅𝑐  (5.5)  Where R i is the expected (required) return of the investment, R f is the rate of return of a theoretically risk-free investment, β is the risk of the investment compared to the overall market (the market = 1), R m is the expected return of the market, and R c is the country risk [522]. For the purposes of this analysis, it is assumed that R f is equivalent to the average return of 1-year Government of Canada Treasury Bills from 2000-2010 of 2.8%. During the same period, the annual return of the S&P/TSX index was 4.52%, resulting in an equity risk premium (R m – R f ) of 1.72%. However, historically this is low and therefore the long-term Canadian equity risk premium of 5.7% was used for the analysis [525]. This is slightly higher than the average global equity risk premium of 5% [525]. It was assumed that β is equal to 1 (i.e. the company developing the project has the same risk as the market as a whole). When comparing two identical facilities built by the same (Canadian) company but in two different countries, any difference in risk premium will largely be systematic, undiversifiable, country risk. Based upon analyses by Damodaran (2011), it is assumed Brazil has a country risk premium of 3% [526]. Therefore, the cost of capital for this analysis for the facilities in Williams Lake and Prescott is 8.5% and for Barra do Riacho is 11.5%.  152  5.3.5  Cost of Construction  Several companies provide international cost of construction indexes that compare major cities around the world. The U.S. dollar and a U.S. city are typically used as the basis for the index, making the indexes susceptible to changes in exchange rates and the health of the U.S. economy. The closest indexed cities (Vancouver, Toronto, and São Paulo) were used as a comparative metric to determine relative cost of construction at the plant sites. While construction costs at the actual site may vary substantially compared to the indexed city, since all sites are outside of major cities, the relative comparison was consistent. Based upon the Faithful + Gould International Construction Index, which uses a Chicago, IL baseline of 100, Toronto’s construction cost index was estimated at 85.6 and São Paulo’s cost index at 67.6 [527]. However, a Vancouver cost index was unavailable. The exchange rate, which greatly affects the index results, was assumed to be 1.25 for $CA to $US and 2.32 for R$ to $US. Current exchange rates place the Canadian dollar on par or higher with US dollar, and the assumed exchange rate for the Real in this analysis, based upon recent trading, is 1.6. When these changes in exchange rate are taken into account, Toronto’s index becomes 102.7 and São Paulo’s index becomes 88.56. Based upon the KPMG Guide to International Business Location, Vancouver is on par with Toronto for the chemicals sector cost index and therefore 102.7 was assumed for Vancouver as well [481]. 5.3.6  Electricity Production and Revenues  Electricity rates are significantly lower in British Columbia than both Ontario and Brazil. The high cost of electricity in Brazil, which is almost 300% that of British Columbia, has been attributed to taxes on consumption [528]. While this is a major challenge for facilities that must purchase electricity from the grid operator, it presents an opportunity for a facility that is a net producer of electricity. All jurisdictions considered in this study have implemented targeted policies for the purchase of renewable biomass-based electricity. In Canada, this electricity has been dominated by residues from the forestry industry, while in Brazil, residues from the sugarcane industry have been the leading feedstock. Feed-inTariff rate (Ontario), Standing Offer Program rate (British Columbia) and renewable energy auctions 153  (Brazil) have been used to project electricity revenues, which have been used to reduce the net electricity cost. There is a notable difference in the capacity of the co-located biopower plants, as this capacity is dictated by the quantity of lignin produced by the ethanol facility. Since all ethanol facilities are producing the same amount of ethanol, facilities using a feedstock with a high lignin content (and associated lower sugar content) will produce more lignin and therefore have a higher capacity co-located co-generation facility. Based upon the assumptions presented in Table 5.2, an 800 ML yr-1 ethanol facility would consume 615,600 MWh yr-1, with a capacity demand of 82.2 MWe. Net electrical capacity of the co-generation facility ranged from 102.0 MWe to 124.4 MWe for the six scenarios, indicating a grid supplying capacity of 19.8 MWe to 42.2 MWe. The softwood-based facility (scenario BC1) was at the top end of this range. 5.3.7  Taxes, Insurance, and Permits  For projects as large and unique as a lignocellulosic ethanol facility, insurance, permits, and property taxes are difficult to predict.  They would likely be negotiated on a case-by-case basis.  However, general heavy industry rates have been used here to represent a likely situation for each biorefinery site. Income taxes are notably higher in Brazil than Canada, although property taxes are relatively similar. It is worthwhile to note that within a single province such as British Columbia, yearly property taxes for large industry range by municipality from zero to 8.4% of assessed value [387]. The minimum ethanol selling price (MESP), the price at which the producer breaks even, is presented in Figure 5.1 for all six scenarios. The Brazilian scenarios BR1 and BR2 were the lowest cost with an MESP below $0.75 L-1, while the British Columbia softwood scenario (BC1), at $1.02 L-1, is higher than all hardwood scenarios. Delivered feedstock cost, including harvest and delivery, is the primary contributor to MESP differences between scenarios. Property taxes are also a major contributor to MESP.  154  Table 5.6 Non-feedstock site-specific variables impacting MESP Cost of ethanol delivery ($ m-3) Labour ($ hr-1) Cost of capital (discount rate)a Cost of construction (from 100 baseline) Industry Electricity Cost ($ MWh-1)b Biopower Revenues ($ MWh-1) Insuranced Permits Corporate income taxe Property taxf Management & Administration Exchange rate  Williams Lake, BC $19.36  Prescott, ON $14.91  $33.78 8.5% 102.7  $33.78 8.5% 102.7  Barra do Riacho, BR $14.67 (Vancouver); $9.32 (Montréal) $10.40 11.5% 88.56  53.20 [529] 99 [530] 1% 1% 25.5% [533] 8.4% [387] 0.55% 1  96.50 [529] 130 [531] 1% 1% 27% [533] 6.8% [535] 0.55% 1  155.00 [528] 103c 1% 1% 34% [534] 6.5% [536]g 0.55% 1.6  a  CA$1 = R$1.6 and financing is assumed to be 100% debt Based upon representative cities of Vancouver, Toronto, and Rio de Janeiro respectively c Based upon 2008 auction rate of US$80 MWh-1 [532] for sugarcane bagasse and adjusted for Brazilian inflation d Rate based upon discussions with anonymous industrial insurance broker e As of January 1, 2012 f On assessed value g Based upon 5% ISS (Municipal Service Tax) and 1.5% IPTU (Municipal Property Tax) [536] b  Figure 5.1 MESP for all siting scenarios  155  5.3.8  Sensitivity to Energy Rates  Since all the process heat required for the production of ethanol is assumed to be produced by onsite residues and that there are no heat export revenues, the two forms of energy considered for the sensitivity analysis are the electricity rate, which impacts the net cost of electricity to the facility, and oil cost – namely diesel and bunker fuel – which impacts the delivered cost of feedstock and cost of product export. The impact of a 50% increase in the revenue per kWh sold to the electrical grid and a 50% increase in oil-input cost was calculated for the sensitivity analysis. The base rate for electricity revenue is site specific (Table 5.6) and the base case cost for diesel is $1.30 L-1 and for bunker fuel is $675 t-1, as stated above. Consumption is assumed to be completely inelastic in regards to cost and consistent with base case assumptions. Rail diesel consumption is assumed to be 168 t-km L-1 [537]. The impact of an increase in energy rates on the MESP is presented in Figure 5.2. Figure 5.2 MESP sensitivity to electricity rates and oil cost  156  5.4  Discussion Despite the long distance to market, Brazilian-produced eucalyptus ethanol has a notably lower  MESP than Canadian-produced lignocellulosic ethanol. Large-volume seaborne deliveries from Brazil to Vancouver and Montréal markets have a low cost on a per litre-km basis and result in product shipping contributing a small percentage (<2%) to the total MESP. As in the pulp sector, the largest difference in cost of production of lignocellulosic ethanol between Brazil (tropical) and Canada (temperate) is the cost of feedstock [484]. The delivered cost of feedstock, which includes harvest and transportation, is approximately 2.5 times greater for BC softwood than Brazilian eucalyptus. Scenario ON-Willow is the only non-Brazilian scenario with a feedstock contribution to MESP under $0.25 L-1. This presents a challenging situation for Canadian lignocellulosic producers, because it means that even if they are able to compete with other bioenergy product producers for feedstocks domestically, those feedstocks may have a higher cost than those in other jurisdictions. While labour cost is substantially lower in Brazil than Canada, impacting both delivery (driver) and facility operating costs, it has a small contribution to MESP and therefore provides only a limited advantage to Brazilian facilities. This is consistent with the pulp industry [484]. The sensitivity analysis showed diesel and bunker fuel prices impact BC and Ontario-based facilities to a greater extent than Brazilian-based facilities. This is due to the higher proportion of fuel used in feedstock transportation compared to ethanol transportation and the relatively longer feedstock transportation distances for Canadian biomass. Despite the lower construction cost of a facility in Brazil, driven by construction labour cost, the contribution of capital to the MESP of a Brazilian facility is actually greater than one located in Canada. This is due to the higher investment risk and therefore higher cost of capital. This dichotomy of total capital cost could be modified in the future as Brazilian wages rise, driving up the cost of construction index, or Brazil’s debt rating could be upgraded, reducing the financing interest rate. Capital is the largest single cost contributor to the MESP for Brazilian scenarios and the second largest, following delivered feedstock cost, for all Canadian scenarios. 157  Enzymes cost contributed approximately $0.10 L-1 for all scenarios, with 20 FPU g-1 cellulose assumed for all scenarios. In practice, enzyme loading is likely to be lower for hardwood than softwood feedstocks on a tonne of raw material basis, regardless of absolute cellulose content. Although a 70% theoretical yield was assumed for all feedstocks, softwood is known to be significantly more recalcitrant and difficult to hydrolyze than hardwood, which could reduce the actual hydrolyzed sugar yield and hence ethanol yield for softwoods relative to hardwoods [326,415]. This would increase the MESP for scenario BC-Forest beyond the already high $1.02 L-1. This MESP is higher than that identified in Chapter 3 for softwood due to the higher feedstock costs – a result of utilizing whole logs from dedicated harvest instead of harvest and mill residues – in addition to higher property taxes. The co-location of a CHP facility is an important means of attaining a competitive MESP. Should the facilities purchase electricity at local industrial rates, MESP would be $0.014-0.098 L-1 greater, with the greatest difference being for Brazilian scenarios. As noted by the mining industry, high electricity prices in Brazil could limit development of high electricity-consuming projects [528]. The relatively high feed-in tariff rates in Ontario also help reduce the net electricity cost for those projects. A high rate for bio-based electricity supplied to the grid, which results in lower operating costs, could be used by governments to attract lignocellulosic ethanol facilities. The sensitivity analysis showed that increasing electricity rates paid by the grid by 50%, thus reducing the net electricity cost for ethanol, can reduce the cost of ethanol production by greater than $0.02 L-1. The electricity efficiency, as modelled, was low compared to most large-scale (>50 MWe) biopower plants operating on raw biomass, which typically operate at 25-30% efficiency. However, the large process heat demand for ethanol production, at 18 MJ L-1, meant that 75% heat efficiency for conversion of lignin (95% recovery) and cellulose (25% recovery) was required. This also showed that cellulose hydrolysis rates beyond 75% would result in heat demand exceeding available heat from process lignin and unhydrolyzed cellulose. Additional heat inputs, in the form of supplementary biomass, natural gas, or coal, may be required.  158  The scenarios presented here do not include the sale of co-products, other than electricity, which could potentially generate co-product credits – as detailed in Chapter 3. Lignin-derived compounds have been promoted as a means to provide substantial co-product credits [413], although a large market for lignin-derived products has not yet been proven. Production of lignin co-products would reduce the amount of lignin available for process heat, and would result in a requirement for supplementary fuel and added operating cost. In addition, the sale of lignin co-products is dependent upon purification of lignin and utilization of pretreatment technology such as organosolv. Steam explosion, as assumed here and preferred by industrial concerns such as Mascoma [417], while lower in capital and operating costs than organosolv, does not produce lignin of sufficient purity for sale as a co-product [326,415]. Both business models, built around competing pretreatments, are being pursued but neither has thus far resulted in commercial production. Given the substantial capital cost of the modelled facility, and that property taxes are typically calculated on an ‘assessed value’ basis, the yearly tax liability is substantial for an 800 ML facility – ranging from $68-102 M per year (not including assessed value depreciation). The assumed rates are based upon stated large-scale industrial rates for the local municipalities. However, it is likely that in order to attract the investment and jobs associated with a large scale biorefinery, some communities would be willing to offer lower, ‘custom’ property tax rates. In addition, many communities within the same region as the chosen sites already offer lower property tax rates. For example, property tax rates in British Columbia range from 0.0-8.4% of assessed value, giving Williams Lake the second highest property tax rate (8.4%) of all jurisdictions in British Columbia [387]. Previous chapters have assumed a rate of 0.75%, which would reduce the property tax burden and contribution to MESP to $0.0133 L-1 for Canadian scenarios and $0.0115 L-1 for Brazilian scenarios. Clearly, given the large capital investment, a low property tax jurisdiction is critical to long-term economic competitiveness and a low MESP. If property taxes were eliminated for the Brazilian scenarios, ethanol could be delivered to Montréal for less than $0.64 L-1. This is less than 200% of the MESP of both sugarcane and corn ethanol [88,91], indicating a more competitive position than was determined for Canadian feedstocks in Chapters 159  2 and 3 and other analyses [146]. However, this does not address the fundamental question of why a Brazilian company would pursue technically risky and challenging advanced lignocellulosic biofuel production from eucalyptus when conventional production from sugarcane is technically and economically preferred, despite the fact eucalyptus ethanol has been shown here to be the lowest cost woody biomass ethanol option.  5.5  Conclusion Advanced lignocellulosic ethanol facilities in Canada and elsewhere will face the same challenge  posed by the pulp and paper industry – competing with lower cost imports from southern hemisphere countries. In addition, there is less product differentiation for ethanol production than pulp (e.g., Northern Bleached Softwood Kraft Pulp v. Eucalyptus Hardwood Pulp). Since shipping constitutes only a small fraction of the MESP of ethanol for major Canadian markets of Vancouver and Montréal, operating costs – in particular feedstock – and taxes will be the driving factors behind decisions on siting biorefineries for advanced lignocellulosic ethanol production. Just as North American lignocellulosic ethanol producers will have a challenge competing for market share with conventional ethanol, as presented in Chapter 3, they will also be challenged to compete for ethanol fuel market share with lignocellulosic ethanol producers from tropical countries. This is particularly true if there are no import tariffs, as discussed in Section 1.2.4.6, or domestic fuel only excise tax exemptions, as discussed in Section 1.2.4.4. Since domestically-produced product may not be the lowest cost option for including lignocellulosic ethanol in the Canadian fuel supply, federal and provincial governments in Canada wishing to incentivize biofuels are faced with the dilemma of supporting lignocellulosic ethanol production or use. Supporting domestic production may result in the consumer receiving a higher-priced fuel than could be achieved through imports, thus hindering fuel acceptance and competitiveness relative to conventional ethanol and gasoline, while supporting domestically-produced lignocellulosic ethanol consumption could result in ethanol imports and limited domestic production and job creation. This chapter, focused on thesis Theme 4 on facility siting, combined with Chapter 3, which contrasted lignocellulosic ethanol production cost to that of conventional corn and sugarcane ethanol, has 160  shown that lignocellulosic ethanol produced from Canada’s forest resources is not likely the lowest cost ethanol fuel available to Canadian markets. It has competitive disadvantages in terms of feedstock costs and processing requirements (relative to conventional ethanol) that may be difficult to overcome. However, these findings by themselves do not necessarily mean that lignocellulosic ethanol production in Canada will not be a profitable enterprise in the future. The transportation fuel supply is currently dominated by oil-based products and therefore, all alternatives are judged relative to the price of oil. A high oil price, and hence high gasoline price, could make lignocellulosic ethanol production in Canada profitable – albeit perhaps not as profitable as domestic conventional ethanol production or lignocellulosic ethanol production in Brazil. This profitability, discounting the influence of preferential government policy, must be viewed from a facility lifetime perspective and the risk associated with continual operation of a facility and the gross processing margins of that facility (as dictated by revenue and feedstock cost). Even if feedstock costs are static, revenue volatility, and hence gross processing margin volatility, will largely be driven by oil and conventional ethanol prices. Returns for financiers must be commensurate with gross processing margin volatility and the risk it implies, even under profitable conditions. The next chapter examines the impact of this volatility on the anticipated average cost of capital for financing a lignocellulosic ethanol facility in Canada and how gross processing margin and revenue volatility impact the ability of producers to compete for feedstock.  161  6 6.1  FINANCING EXPECTATIONS AND THE IMPACT ON FEEDSTOCK COST Introduction Despite numerous industry support policies, large-scale commercial production of lignocellulosic  ethanol has not occurred in the time frames anticipated. An example of this is the 2010 and 2011 lignocellulosic ethanol mandates under the Renewable Fuel Standard (RFS) of EISA, which were initially set at 100 million gallons (M gal; 378.5 ML) and 250 M gal (945 ML) for those years respectively. These mandates have been since revised to 6.5 M gal (24.6 ML) for 2010 and 6.6 M gal (25.7 ML) for 2011 [538]. This is in stark contrast to the U.S. corn ethanol industry, which grew to 50 GL in 2010 [539], required an increase in the U.S. blend wall (the allowable content of ethanol in gasoline) from 10% to 15% for vehicles model years 2001 and newer to enable demand to meet supply [540,541], and as of 2010 production, was only 1.8 G gal (6.8 GL) short of the 2022 15 G gal (56.8 GL) RFS volumetric mandate under EISA. Additionally, it is estimated that 2010/2011 was the first year that ethanol surpassed animal feed as the number one use of corn in the U.S. [542]. With a total world ethanol production of 85.8 GL in 2010 [543] and ethanol constituting over 2% of the world transportation fuel supply [46], lignocellulosic ethanol has a relatively strong base on which to build production capacity and develop the industry. However, building this production capacity will require not only competitive firms, as discussed in the previous four chapters, but also the financing to development of projects. It is this financing, and the risk associated with financing lignocellulosic ethanol facilities, that is the focus of this chapter. Commercialization of new energy technologies typically follows a four-stage path, with sources of financing and key milestones changing along the path. The traditional route is presented in Figure 6.1. However, in the field of advanced lignocellulosic biofuels, governments, and in particular the U.S. and Canadian governments, have expanded beyond their traditional role of investment in technology research and have provided support at every stage in the commercialization process, including Roll Out & Market Volume (highlighted in orange in Figure 6.1). This expansion means that governments and the public sector are not only taking on technology risk, which tends to involve relatively small sums of capital in 162  the research and development stage of technologies, but a large amount of operational risk.  This  operational risk can dwarf technology risk in absolute financial terms due to the size of capital investments and the quantity of product being produced. Since lignocellulosic biofuels must compete in transportation fuel markets, as discussed in Chapters 3 and 5, Roll-Out & Market Volume risk also includes exposure to volatile oil prices. Figure 6.1 Path to technology commercialization  *Adapted from [544,545,546] Governments have prioritized biofuels for research and development relative to other renewables (e.g., wind, solar, geothermal, small hydro), making it the single largest category for investment, with US$2 B invested by the public sector in 2010 compared with US$0.3 B by the private sector. This differs significantly from solar technology, in which the private sector invested US$2.1 B compared to US$1.5 B from the public sector [544].  While venture capital and private equity (VC/PE) investment in  163  conventional biofuels (largely corn/sugarcane ethanol and biodiesel) dropped by 90% to $71 M in 2010, indicating a mature technology sector and lack of VC/PE opportunities, investment in second and third generation (i.e., drop-in) biofuels jumped 57% to $630 M [544]. Although this investment is promising for scale-up and roll-out of technologies in the future, the investment requirements and scale (i.e., value) for VC/PE are significantly different than those for credit (debt) financing. The risk profile typically accepted by VC/PE financing is dominated by technology risk, while plant financing and roll-out is dominated by operational risk [545]. Despite the interest in lignocellulosic ethanol by VC/PE and public sector (government) investment, traditional commercialization patterns dictate that roll-out and growth in market volume will require private-sector financing of large capital projects. This has traditionally been attained through public equity markets (e.g., Initial Public Offerings), debt, and/or mergers and acquisitions by other operations. In the lignocellulosic biofuels sector, large integrated energy companies, such as Shell, BP, and Exxon-Mobil, could be potential acquirers. However, investors in, and financiers of, lignocellulosic ethanol facilities will demand compensation appropriate for the risks encountered and thus far this has been a major hurdle to commercialization of lignocellulosic ethanol. Investment in the transportation biofuels sector as a whole peaked in 2006 at US$20.4 B, but was only US$5.5 B, or 26.7% of that peak, in 2010.  Biofuels, as a percentage of total clean energy  investment, dropped from 26.8% in 2006 to 3.8% in 2008 [544] (UNEP and BNEF, 2011). Part of this drop can be attributed to the general poor economic conditions resulting from the global financial crisis and freezing of credit markets, but the associated drop in oil prices from US$147 to US$35 per barrel was also a primary reason. No other clean energy sector category dropped so significantly. As discussed in Chapter 2, the primary differences between biofuels and other renewables (apart from other biomass technologies such as electricity production) is the dependence upon a variable cost input (feedstock), which makes up a large proportion of the operating costs of a facility, and the target market of transportation fuels, which is a significantly more volatile market than that of the electricity sector for producers. This is particularly true when electricity producers receive revenues through governmentregulated renewables programs such as Feed-In Tariffs that guarantee above-market electricity rates for 164  extended operating periods.  Therefore, given these differences, biofuels face significantly greater  operational, financial, and market risks than the other clean energy sectors of wind, solar, geothermal, and small hydro. However, the market risk premium for lignocellulosic ethanol and other advanced biofuel investment has not been a subject of extensive evaluation in the literature. Unlike lignocellulosic ethanol, the investment risk and return profile of conventional ethanol has been the subject of much analysis, particularly the corn-ethanol processing margin and management of this primary operational risk. Similar to the conventional ethanol technical analysis in Chapter 3, work in the conventional ethanol financing and financial research areas can be used to inform research on lignocellulosic ethanol commercialization and this thesis. Schmit et al. (2009) used a real options (financial options applied to ‘real world’ investments such as processing facilities) and net present value (NPV) analysis to determine entry-exit decisions for dry-grind corn ethanol plants [547], while Dal-Mas et al. (2011) used a mixed integer linear program to assess economic performance and investment risk along the entire processing chain for dry-grind corn ethanol [548]. Gallagher et al. (2007) also used an NPV analysis to determine the impact of scale, siting, and organization on the ethanol processing margin and profitability [549]. Leach et al. (2011) focused on the commodity risk faced by investors in a smallscale wheat-based ethanol facility and policies and strategies that could be used to mitigate this commodity risk [550]. These studies highlight the importance of feedstock cost and gross processing margin, or ‘crush spread’, in facility economic viability. The economic performance of lignocellulosic ethanol production is also highly dependent upon the ethanol gross processing margin. However, unlike corn or sugar, lignocellulosic biomass, apart from wood pellets, is not an internationally-traded feedstock commodity. Therefore, margin volatility is likely to follow a different pattern than conventional ethanol. Typically, lignocellulosic ethanol facilities are modelled as relying exclusively on local biomass supplies, although international feedstock supply chains have been shown to be economically competitive through analysis [215] and evidenced by the >1.5 million tonnes of wood pellets exported from British Columbia to Europe each year [438]. When relying upon local feedstocks only, and particularly when that feedstock is available via long-term supply 165  agreements such as forest tenure, lignocellulosic ethanol will have a gross processing margin risk profile similar to that of vertically integrated operations such as corn ethanol production by large land owners and operators such as Archer Daniels Midland. In this case, feedstock price is equivalent to cost of harvest/production, rather than market price dictated by supply and demand with multiple sellers and buyers [488]. Babcock et al. (2011) identified this lack of an integrated market for lignocellulosic feedstocks as an advantage for lignocellulosic biofuel over corn ethanol produced at independently operated mills [155], although vertically-integrated operations of any type would have the same advantage of insulation from feedstocks markets – not considering the opportunity cost when market prices rise. Independent mills, similar to independent oil refineries, must buy feedstock on the open market and typically use financial futures to ‘lock-in’ their gross processing margin between commodity inputs and outputs. This margin is termed the ‘crack spread’ for oil refineries and ‘crush spread’ for ethanol producers. Determining the gross processing margin for the production of lignocellulosic ethanol and the volatility of that margin is a primary research task detailed in this chapter. This information can be used to identify the conditions of production that provide an acceptable investment return, taking into consideration the market risks faced by a forest-based lignocellulosic ethanol facility. Identification of the minimum financing rate (i.e., average cost of capital) that accounts of these risks was also a primary objective of the research.  6.2  Study Design and Assumptions The spreadsheet techno-economic model described in Chapters 2, 3, and 5 was used to determine  the cost of production of lignocellulosic ethanol from woody feedstocks at the risk-free cost of capital. This rate was based upon the average nominal return of 1-year Government of Canada Treasury Bills over the past 10 years and was assumed to be 2.8% [551]. The facility, with an annual ethanol capacity of 800 ML yr-1, as informed by the logistics scaling analysis in Chapter 4, was assumed to be located at Williams Lake, British Columbia, Canada. This replicates the BC1 scenario conditions described in Chapter 5, except property tax is assumed to be 2% rather than the very high default 8.4% for Williams Lake. The 166  base case minimum ethanol selling price (MESP) at a major market (Vancouver, British Columbia, Canada) and minimum gross processing margin – termed the ‘hydrolysis spread’ – were calculated for the base case risk-free nominal cost of capital. Primary variables modelled for sensitivity at this base case included feedstock cost, ethanol yield, enzyme cost, and capital cost. Data from the literature on forest harvest cost for the interior of British Columbia was used to create a baseline cost of feedstock. Harvest site whole tree and residue chipping was assumed. Diesel fuel consumption for both harvest and trucking was calculated based upon the literature and used to determine the direct impact of diesel cost on delivered cost of feedstock. Delivered feedstock cost, adjusted for historical diesel cost and consumer price index (CPI, as an inflation proxy), was contrasted with historical ethanol prices to determine the theoretical historical hydrolysis margin and volatility.  Historical Omaha rack ethanol prices were  obtained from the Nebraska Energy Office, State of Nebraska [360] and were also used to calculate historical lignocellulosic biomass feedstock cost required to attain a rate of return equivalent to the riskfree cost of capital. For this analysis, it was assumed that there is no price premium for lignocellulosic ethanol over conventional ethanol. Prices were adjusted for historical exchange rates and Canadian inflation (complete CPI), but rates and cost of capital are considered to be nominal. All financial figures are provided in 2010 Canadian dollars, at par with U.S. dollars, unless otherwise stated. The Capital Asset Pricing Model (CAPM), previously described in Chapter 5, was used to determine the expected rate of return, given the risks involved, for an investment in a lignocellulosic ethanol plant.  The CAPM takes into account systematic (undiversifiable risk) and unsystematic  (diversifiable risk) risk, the expected rate of return of the market, and the theoretical return on a risk free asset. The CAPM systematic risk-adjusted return is given by: 𝑅𝑖 = 𝑅𝑓 + 𝛽�𝑅𝑚 − 𝑅𝑓 �  (6.1)  Where R i is the expected (required) return of the investment, R f is the rate of return of a theoretically risk-free investment, β represents volatility (systematic risk) of the investment compared to the overall market (the market = 1), and R m is the expected return of the market. Inflation could be included in the CAPM, but for the purposes of this thesis, the calculation will be nominal rates of return. 167  Typically, R m is the return of the stock market, which in Canada is the S&P/TSX index that has a longrun equity risk premium (R m – R f ) of 5.7% [525]. In this analysis, R i was calculated for the integrated and independent oil refining industries, using a beta previously reported by Damodaran (2011) [552]. This is the return that an investor would expect to justify an investment in an independent or integrated oil company. The independent oil refining R i was then used as the expected market return, R m , for a second CAPM calculation to determine the expected return of a lignocellulosic ethanol biorefinery. In this case, Total Beta (β T ) was calculated based upon an assessment of the processing margin volatility for lignocellulosic ethanol (hydrolysis spread) relative to independent oil refining (crack spread) and is given by the equation: 𝛽𝑇 =  𝜎ℎ𝑠 𝜎𝑐𝑠  (6.2)  Where σ hs is the standard deviation of the hydrolysis spread and σ cs is the standard deviation of the crack spread. Total Beta can be used since correlation between the independent oil refinery returns and lignocellulosic ethanol biorefinery returns is not a focus of this analysis. NYMEX historical futures prices, provided by the United States Energy Information Administration [553] for gasoline [Reformulated Regular Gasoline (1985-2005) and Reformulated Blendstock for Oxygenate Blending (2006-2010), New York Harbour] , No 2 heating oil (proxy for diesel, New York Harbour), and crude oil (Light Sweet, Cushing, Oklahoma) were used to calculate historical crack spread on a 3:2:1 basis (three units oil for two units gasoline and one unit diesel. The resulting R i , the expected return for an investment in a lignocellulosic ethanol biorefinery to account for market risks (but assuming the same technology and non-commodity risks as an independent oil refinery), was used in the techno-economic model to determine the MESP that would meet the return requirements of a financier.  6.3  Results 6.3.1  MESP Base Case and Sensitivity  The minimum ethanol selling price (MESP) for production from softwood at an 800 ML yr-1 capacity facility located in British Columbia’s interior and delivered to Vancouver, BC is $0.85 L-1,  168  assuming a risk-free cost of capital. Cost components are presented in Figure 6.2 and are based upon the model presented in Chapter 5. Figure 6.2 Cost components for ethanol production and delivery, with an MESP of $0.85 L-1  The combined feedstock harvest and delivery cost, assuming enterprise vertical integration (i.e., price is equal to cost), is $0.39 L-1, resulting in a minimum hydrolysis spread (gross processing margin), from the MESP, of $0.46 L-1. Results of the sensitivity analysis on variables delivered feedstock cost, yield, enzyme protein cost and loading, capital cost, and property taxes are presented in Table 6.1. Table 6.1 MESP and hydrolysis spread sensitivity analysis Variable  Base case value  Change from base case  Revised value  MESP ($ L-1)  Delivered feedstock cost Ethanol yield  $125.38 bdt-1  -20%  $100.30 bdt-1  0.78  Minimum Hydrolysis spread ($ L-1) 0.46  321 L bdt-1  +20%  385 L bdt-1  0.83  0.44  Enzyme protein cost Enzyme loading  $2400 t-1  -20%  $1920 t-1  0.83  0.44  20 FPU  -20%  16 FPU  0.83  0.44  Capital cost  $1,216 M  -20%  $972.8 M  0.80  0.41  Property taxes All combined  2.0% -  -50% -  1.0% -  0.84 0.65  0.45 0.34  169  6.3.2  Diesel Consumption  Feedstock harvest, processing including whole tree chipping, and transportation diesel consumption were calculated to determine diesel cost impact on ethanol MESP. Data sourced from previous studies is presented in Table 6.2. Oil-based motor oil and lubrication were estimated to be equivalent to 15% of diesel fuel by volume [339], resulting in a total oil product consumption of 21.4 L bdt-1. Pricing was assumed to follow that of diesel. Based upon the sensitivity analysis of Chapter 5, diesel use for ethanol delivery from Williams Lake to Vancouver (550 km) on a per litre basis was assumed to be negligible. Table 6.2 Diesel fuel consumption and contribution to MESP Task  Source  Machine  Diesel consumption (L bdt-1)  Diesel contribution to MESP ($ L-1)a  0.8 3.5 3.0b 0.4  Diesel contribution to feedstock ($ bdt-1) 1.04 4.55 3.90 0.52  Felling Forwarding Whole-tree chipping Loader  [554] [554] [339] [339]  Off-highway trucking (30 km) Unloader/Loader On-highway trucking (170 km) Total  [492]  Large feller-buncher Grapple-skidder Trelan 23 chipper Komatsu PC200 loader Semi-trailer  2.3c  2.99  0.0093  [339] [492]  Komatsu PC200 B-train chip truck  0.6 8.0d  0.78 10.4  0.0024 0.032  -  -  18.6  24.18  0.075  0.0032 0.014 0.012 0.0016  a  Assumes ethanol yield of 321 L bdt-1 and a diesel cost of $1.30 L-1 b Assumes 75 L PMH-1 and 25 bdt PMH-1. This is mid-range from biomass grinding estimates, including 2.4 L bdt-1 [555] and 4.8 L bdt-1 [556] c Assumes 2 km L-1 and a load of 26 t with a 50% moisture content. Adjusted from Canadian heavy trucking average of 2.9 L km-1 [492] by 30% to account for terrain and consistency with previous estimates (e.g. 1.7 km L-1 by Jones