Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Assessing the potential for bioethanol production from oxygen delignified corn stover and wheat straw Pope, Derek Stephen 2011

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

Item Metadata

Download

Media
24-ubc_2012_spring_pope_derek.pdf [ 2.75MB ]
Metadata
JSON: 24-1.0059082.json
JSON-LD: 24-1.0059082-ld.json
RDF/XML (Pretty): 24-1.0059082-rdf.xml
RDF/JSON: 24-1.0059082-rdf.json
Turtle: 24-1.0059082-turtle.txt
N-Triples: 24-1.0059082-rdf-ntriples.txt
Original Record: 24-1.0059082-source.json
Full Text
24-1.0059082-fulltext.txt
Citation
24-1.0059082.ris

Full Text

ASSESSING THE POTENTIAL FOR BIOETHANOL PRODUCTION FROM OXYGEN DELIGNIFIED CORN STOVER AND WHEAT STRAW  by  DEREK POPE B.Sc., The University of Victoria, 2007  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE  in  THE FACULTY OF GRADUATE STUDIES (Chemical and Biological Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  December 2011  © Derek Pope, 2011  Abstract Bioethanol derived from lignocellulosic biomass offers great potential as a low carbon energy source. Rather than using food resources as a feedstock, abundant non-food lignocellulosic alternatives can be utilized. Process economics represents the greatest barrier facing lignocellulosic bioethanol. An effective pretreatment that increases enzymatic hydrolysis sugar yield and reduces enzyme consumption is crucial for reducing ethanol production cost. Oxygen delignification was studied for its potential as a pretreatment of agricultural residues corn stover and wheat straw. This pretreatment was selected based on its ability to target, disrupt, and solubilise lignin; a component linked to poor enzymatic hydrolysis yields. Oxygen delignification reaction temperature (90-150°C), residence time (15-60 min), and caustic load (2-10%) were studied for their effect on enzymatic hydrolysis. Conditions that increase substrate hydrolysability while limiting sugar solubilisation allowed for a maximum total sugar yield of 81.7% (92.1% of total cellulose and 68.2% of total hemicellulose). Next, the substrate composition of pretreated substrates was analyzed. Results confirmed that reductions in lignin content improved substrate hydrolysability. An improvement to the Zhang et al. [1] model, which predicts sugar concentration during enzymatic hydrolysis, was made by enabling the model to account for changes in substrate lignin content. The addition was based on a hypothesis put forth that lignin reduces the availability of cellulase enzymes. The model (below) was successfully validated. P = Product (sugar) concentration So = Initial substrate (carbohydrate) concentration Eo = Initial enzyme loading concentration Lo = Initial lignin concentration LF = Lignin factor kd = Enzyme deactivation constant Ke = Equilibrium constant ii  k2 = Reaction rate constant Finally, an economic optimization of bioethanol production was conducted. To perform this analysis, an Aspen Plus simulation of the process was developed. A base case minimum ethanol cost of $0.55/L was observed under pretreatment conditions of 135°C, 8% caustic, and 60 minutes residence time. A sensitivity analysis demonstrated that ethanol cost was vulnerable to increases in the cost of biomass, enzyme, and NaOH. It was also found that changes in the cost of enzyme and biomass affected the optimal pretreatment reactor conditions.  iii  Table of Contents Abstract............................................................................................................................................ ii Table of Contents .......................................................................................................................... iv List of Tables ................................................................................................................................. viii List of Figures .................................................................................................................................. xi List of Abbreviations ..................................................................................................................... xiv Acknowledgements ....................................................................................................................... xv Dedication ..................................................................................................................................... xvi 1  Introduction..............................................................................................................................1 1.1  The need for a new transportation fuel ...........................................................................1  1.2  Biofuel as a clean transportation energy source ..............................................................1  1.2.1  Worldwide bioethanol production trends.................................................................3  1.2.2  Additional advantages of biofuel...............................................................................3  1.2.3  First generation bioethanol and the associated problems .......................................5  1.2.4  Second generation bioethanol and its advantages ...................................................6  1.2.5  Characteristics of ethanol as a fuel ...........................................................................7  1.3  Lignocellulosic biomass .....................................................................................................7  1.3.1  The main components of lignocellulosic biomass .....................................................7  1.3.2  Lignin as a co-product of bioethanol production ....................................................10  1.3.3  Lignocellulosic feedstocks .......................................................................................10  1.3.4  Agricultural residues as a lignocellulosic biomass feedstock ..................................12  1.4  Pretreatment of biomass for enzymatic hydrolysis ........................................................13  1.4.1  Current obstacles for lignocellulosic bioethanol and the need for substrate  pretreatment ..........................................................................................................................13 1.4.2  Establishing parameters for an effective pretreatment ..........................................14  1.4.3  Pretreatment methods ............................................................................................15  1.5  Oxygen delignification ....................................................................................................18  1.5.1  Oxygen delignification chemistry ............................................................................20  1.5.2  Oxygen delignification reaction variables ...............................................................22 iv  1.5.3  Process flow diagram for lignocellulosic bioethanol production using an oxygen  delignification pretreatment ..................................................................................................25 1.6  Hydrolysis of cellulose and hemicellulose ......................................................................26  1.6.1  Acid hydrolysis .........................................................................................................27  1.6.2  Enzymatic hydrolysis................................................................................................27  1.6.3  Kinetics of enzymatic hydrolysis of cellulose and hemicellulose ............................29  1.6.3.1  Cellulase enzyme inhibitors............................................................................. 30  1.6.3.2  Cellulose crystallinity ....................................................................................... 30  1.6.3.3  The effect of lignin on hydrolysis .................................................................... 31  1.6.4  Modelling cellulase hydrolysis kinetics and the effect of lignin content ................32  1.6.5  Empirically modelling of the effects of pretreatment on enzymatic hydrolysis .....33  1.6.6  Enzymatic hydrolysis conditions ..............................................................................34  1.7  Yeast fermentation .........................................................................................................35  1.7.1  Pentose fermentation..............................................................................................35  1.7.2  Fermentation process options.................................................................................36  1.8  Aspen Plus process modelling.........................................................................................37  2  Research Objectives ...............................................................................................................39  3  Materials and Methods ..........................................................................................................40 3.1  Lignocellulosic substrates ...............................................................................................40  3.1.1  Processing biomass for experimentation ................................................................40  3.1.2  Procedure for determining biomass moisture content ...........................................41  3.1.3  Procedure for compositional analysis of biomass ...................................................41  3.1.4  Procedure for measuring sugar concentration using HPLC .....................................43  3.2  Enzymatic hydrolysis .......................................................................................................44  3.2.1  Cellulase enzymes ....................................................................................................45  3.2.2  Hemicellulase enzymes............................................................................................45  3.2.3  Procedure for determining cellulase activity ..........................................................45  3.2.4  Procedure for determining cellulase concentration ...............................................47  3.2.5  Procedure for determining β-glucosidase activity ..................................................48 v  3.2.6  Procedure for determining endoxylanase activity ..................................................49  3.2.7  Procedure for lab-scale enzymatic hydrolysis .........................................................50  3.3  3.3.1  Defining oxygen delignification reactor conditions.................................................50  3.3.2  Apparatus used for oxygen delignification ..............................................................51  3.3.3  Procedure for oxygen delignification ......................................................................52  3.3.4  Procedure for determining solid recovery ..............................................................53  3.3.5  Procedure for determining substrate composition following pretreatment ..........53  3.3.6  Procedure for determining sugar content in pretreatment Liquor.........................54  3.4  4  Oxygen delignification pretreatment .............................................................................50  Method of simulating scale-up of processes for economic analysis ..............................54  3.4.1  Physical property database .....................................................................................55  3.4.2  Empirically-derived equations .................................................................................56  Experimental Results ..............................................................................................................57 4.1  Substrate compositional analysis of corn stover and wheat straw ................................57  4.2  Preliminary oxygen delignification experimentation .....................................................58  4.3  Oxygen delignification reaction conditions ....................................................................62  4.3.1  The effect of oxygen delignification reaction conditions on hydrolysis yield .........62  4.3.2  The effect of oxygen delignification reaction conditions on substrate (corn stover)  composition ............................................................................................................................69 5  Modelling the Effects of Substrate Lignin Content on Enzymatic Hydrolysis Kinetics ..........72  6  Empirical Model Development for Predicting Substrate Composition and Hydrolysability  Following Pretreatment .................................................................................................................80  7  6.1  Empirical model development ........................................................................................80  6.2  Model validation .............................................................................................................83  Aspen Plus Simulation of Bioethanol Process Using Empirical Models .................................85 7.1  Defining substrate composition ......................................................................................85  7.2  Simulating pretreatment ................................................................................................87  7.3  Simulating hydrolysis ......................................................................................................87  7.4  Simulating fermentation .................................................................................................87 vi  7.5  Simulating solid separation .............................................................................................88  7.6  Simulating distillation......................................................................................................88  8  Bioethanol Economic Optimization........................................................................................90 8.1  Capital cost analysis ........................................................................................................91  8.1.1  Unit mapping ...........................................................................................................91  8.1.2  Capital cost results ...................................................................................................94  8.2  Operating cost analysis ...................................................................................................99  8.2.1  Biomass ..................................................................................................................100  8.2.2  Enzyme...................................................................................................................100  8.2.3  By-Product .............................................................................................................100  8.2.4  Oxygen ...................................................................................................................101  8.3  Base case economic analysis.........................................................................................101  8.4  Sensitivity analysis ........................................................................................................104  9  Conclusions...........................................................................................................................111  10  Future Work ......................................................................................................................115  Works Cited .................................................................................................................................117 Appendix ......................................................................................................................................130 Sample calculations..................................................................................................................130 Additional information .............................................................................................................133  vii  List of Tables Table 1.1 Energy density of common transportation fuels sources [11] ....................................... 7 Table 1.2 Cellulose, hemicellulose, and lignin content in waste lignocellulosic residues [32] .... 11 Table 1.3 Cellulose, hemicellulose, and lignin content in dedicated lignocellulosic crops [32] [33] [34] ................................................................................................................................................ 11 Table 1.4 Classification and method of improvement for common lignocellulosic biomass pretreatments [44] [48] ................................................................................................................ 17 Table 1.5 Total sugar yield (wt %) for enzymatic hydrolysis of corn stover with 15 FPU/g glucan following each pretreatment method as observed by Wyman et al. [45] ................................... 18 Table 1.6 Comparison of key reaction parameters for dilute and concentrated acid hydrolysis [66] ................................................................................................................................................ 27 Table 1.7 Cellulose hydrolysis mechanistic models specific for lignocellulosic substrates.......... 33 Table 1.8 Literature review of empirically based models for lignocellulosic bioethanol hydrolysis ...................................................................................................................................................... 34 Table 3.1 HPLC operating conditions ............................................................................................ 44 Table 3.2 Enzyme dilution ............................................................................................................ 46 Table 3.3 Oxygen delignification experimental design ................................................................ 51 Table 3.4 Central composite design for compositional analysis of corn stover ........................... 54 Table 3.5 Component physical property assumptions ................................................................. 55 Table 4.1 Compositional analysis of corn stover and wheat straw .............................................. 57 Table 4.2 Enzyme consumed to produce 10 g of sugar for untreated and pretreated (30 minutes, 120°C, 6% caustic) corn stover ...................................................................................... 60 Table 4.3 Factorial design of pretreatment reaction conditions.................................................. 62 Table 4.4 Results from oxygen delignification factorial design for corn stover. Hydrolysability is based on an enzyme loading of 20 FPU/g substrate and 24 hour hydrolysis. ............................. 63 Table 4.5 Substrate composition based on pretreatment reaction conditions ........................... 69 Table 5.1 Comparison of lignin content and maximum sugar yield following different pretreatment conditions .............................................................................................................. 72 Table 5.2 Units .............................................................................................................................. 75 viii  Table 5.3 Fitted model constants ................................................................................................. 77 Table 5.4 Pretreatment conditions and lignin content of substrate used for model validation . 78 Table 6.1 Empirical models to predict substrate composition based on oxygen delignfication conditions ..................................................................................................................................... 81 Table 6.2 Empirical models to predict enzymatic hydrolysis based on oxygen delignification conditions ..................................................................................................................................... 83 Table 6.3 Model validation pretreatment conditions .................................................................. 84 Table 6.4 Percent error observed for empirical models .............................................................. 84 Table 7.1 Stoichiometry and conversion assumed for the fermentation of each sugar .............. 87 Table 7.2 Separator specifications................................................................................................ 88 Table 7.3 Distillation column specifications ................................................................................. 89 Table 8.1 Oxygen delignification sensitivity analysis pretreatment variables ............................. 91 Table 8.2 Contingency calculation assumptions .......................................................................... 91 Table 8.3 Reactor mapping ........................................................................................................... 92 Table 8.4 Pneumapress© pressure filter specifications............................................................... 94 Table 8.5 Summary of separators mapping.................................................................................. 94 Table 8.6 Equipment cost estimate for bioethanol production at a capacity of 2000 tonnes of biomass (dry weight) per day ....................................................................................................... 95 Table 8.7 Project capital cost summary (units: $ millions) ........................................................... 96 Table 8.8 Oxygen delignification reactor cost .............................................................................. 96 Table 8.9 Capital cost per litre of ethanol produced as observed in literature [108] .................. 97 Table 8.10 Expanding scope of capital cost assessment using results presented by Aden et al. [99] ................................................................................................................................................ 98 Table 8.11 Raw material, utility, and product prices ................................................................... 99 Table 8.12 By-product value assumptions ................................................................................. 101 Table 8.13 Oxygen cost assumptions ......................................................................................... 101 Table 8.14 Base case economic analysis assumptions ............................................................... 102 Table 8.15 Ethanol cost benchmarks .......................................................................................... 104 Table 8.16 Sensitivity analysis cost variables ............................................................................. 105 ix  Table 8.17 The effect of changes in NaOH, biomass, and enzyme cost on optimal pretreatment reactor conditions and minimum ethanol cost .......................................................................... 106 Table 8.18 The effect of enzyme cost on ethanol cost (NaOH cost = $430/tonne, biomass cost = $20/tonne) for fixed pretreatment conditions (60 minute, 135°C, and 6% caustic) ................. 108 Table 9.1 Fitted model constants ............................................................................................... 112  x  List of Figures Figure 1.1 Canadian greenhouse gas emissions (CO2 equivalents) by sector (2004) [2] ............... 1 Figure 1.2 Methods of producing biofuel as presented by Gomez et al. [3] ................................. 2 Figure 1.3 Bioethanol production by country in 2006 (billion litres) [7] ........................................ 3 Figure 1.4 The average worldwide price of oil from 1998-2011 [13]............................................. 4 Figure 1.5 Process flow diagram for the production of bioethanol from corn [14] ....................... 5 Figure 1.6 Cellulose repeating unit cellobiose [26] ........................................................................ 8 Figure 1.7 Hemicellulose represented as a chain of xylose sugars [27] ......................................... 9 Figure 1.8 Lignin [26] .................................................................................................................... 10 Figure 1.9 Composition of the lignocellulosic and non-lignocellulosic portions of corn and wheat [22] ................................................................................................................................................ 13 Figure 1.10 Illustration of the effects of pretreatment on lignocellulosic material [43] ............. 15 Figure 1.11 The guaiacyl representative unit for lignin [54] ........................................................ 20 Figure 1.12 Reaction mechanism for oxygen delignification proposed by Lucia et al. [54]......... 20 Figure 1.13 Stepwise reduction of oxygen to water that occurs during oxygen delignification, as proposed by Gierer et al. [59] ...................................................................................................... 21 Figure 1.14 Relative degree of crystallinity of cotton cellulose fibres as a function of oxidation time as observed by De Souza et al. [61] ..................................................................................... 22 Figure 1.15 Effect of temperature and oxygen partial pressure on kappa number observed by Charles et al. [57] .......................................................................................................................... 23 Figure 1.16 Effect of reaction time and caustic concentration on kappa number observed by Charles et al. [57] .......................................................................................................................... 23 Figure 1.17 Proposed process flow diagram for bioethanol production from corn stover using oxygen delignification pretreatment ............................................................................................ 25 Figure 1.18 Illustration of the actions of endoglucanase, exoglucanase, and β-glucosidase ...... 28 Figure 1.19 Separate hydrolysis and fermentation process option [97] ...................................... 37 Figure 1.20 Simultaneous saccharification and fermentation process option [97] ..................... 37 Figure 3.1 Cuisinart Mini-Prep Plus food processor ..................................................................... 40 Figure 3.2 Processed corn stover and wheat straw ..................................................................... 40 xi  Figure 3.3 Thermo Scientific Thermolyne bench-top furnace...................................................... 42 Figure 3.4 Determination of glucose concentration .................................................................... 46 Figure 3.5 Filter paper unit determination for cellulase activity.................................................. 47 Figure 3.6 Determination of p-Nitrophenol concentration .......................................................... 48 Figure 3.7 β–glucosidase activity determination ......................................................................... 49 Figure 3.8 PARR 4520 high pressure reactor ................................................................................ 51 Figure 3.9 Oxygen delignification reactor setup .......................................................................... 52 Figure 4.1 Experimental mass flow diagram ................................................................................ 58 Figure 4.2 Hydrolysability (20 FPU/g substrate) of untreated and pretreated corn stover and wheat straw samples. Pretreatment was conduced at 120°C with a 30 minute residence and 6% caustic. .......................................................................................................................................... 59 Figure 4.3 Total sugar yield during hydrolysis (20 FPU/g substrate) of corn stover and wheat straw. Pretreatment for 30 minutes at 120°C with 6% caustic ................................................... 61 Figure 4.4 The effect of pretreatment reaction variables on hemicellulose solubilisation ......... 65 Figure 4.5 The effect of caustic load on substrate recovery, hemicellulose solubilisation, substrate hydrolysability, and total sugar yield (24 h hydrolysis, 20 FPU/g substrate). Fixed pretreatment temperature (150°C) and residence time (60 min). .............................................. 67 Figure 4.6 The effect of reaction temperature and residence time on pretreatment recovery, substrate hydrolysability, and total sugar yield (24 h hydrolysis, 20 FPU/g substrate). Only the combined cellulosic and hemicellulosic sugar fraction is shown. Caustic load is fixed at 10%. .. 68 Figure 4.7 The effect of pretreatment reaction variables on lignin solubilisation ....................... 70 Figure 4.8 Substrate hydrolysability versus percentage of lignin solubilised during oxygen delignification ............................................................................................................................... 71 Figure 5.1 Sugar yield during enzymatic hydrolysis (20 FPU/g substrate) versus time for substrates with varied lignin content ........................................................................................... 72 Figure 5.2 Illustration of the proposed effect of lignin on enzymatic hydrolysis ......................... 75 Figure 5.3 Enzymatic hydrolysis (20 FPU/g substrate) experimental and model results for corn stover with varied lignin content.................................................................................................. 78 Figure 5.4 Hydrolysis model validation ........................................................................................ 79 xii  Figure 7.1 Aspen Plus simulation of bioethanol production using an oxygen delignification pretreatment ................................................................................................................................ 86 Figure 8.1 Rotary drum vacuum filter [106] ................................................................................. 93 Figure 8.2 The effect of plant capacity on capital cost as observed in literature [108]. .............. 99 Figure 8.3 Base case ethanol production cost analysis at optimal pretreatment reactor conditions (60 min, 135°C, and 8% caustic) ............................................................................... 102 Figure 8.4 Ethanol cost versus pretreatment reaction conditions for base case assumptions . 103 Figure 8.5 Ethanol cost versus feedstock costs for a single pretreatment reactor condition (60 minutes, 135°C, 8% Caustic) ....................................................................................................... 107 Figure 8.6 Cost of ethanol versus cost of enzyme for a range of caustic loads. Fixed pretreatment reaction time (60 min) and temperature (150°C). Fixed NaOH ($430/tonne) and biomass cost ($20/tonne). .......................................................................................................... 108 Figure 8.7 Cost of ethanol versus cost of NaOH for a range of caustic loads. Fixed pretreatment reaction time (60 min) and temperature (150°C). Fixed enzyme cost ($1/Million FPU) and biomass cost ($20/tonne). .......................................................................................................... 109 Figure 8.8 Cost of ethanol versus cost of biomass for a range of reaction temperature and time. Fixed pretreatment caustic load (10%). Fixed NaOH cost ($430/tonne) and enzyme cost ($1/Million FPU).......................................................................................................................... 110  xiii  List of Abbreviations $ ¢ © °C  DDG  US dollar US cent Copyright Degree Celsius Ammonia fibre explosion Acid soluble lignin Beta Cellobiose unit Centimetre Dried distillers grain  DNS  3,5-dinitrosalicylic  E10 E85  10% ethanol 90% gasoline blend 85% ethanol 15% gasoline blend Filter paper unit Foot Gram Hour High performance liquid chromatography Icarus Process Evaluator International unit Kilodalton Kilowatt hour Litre Molar Milligram Minute Millilitre Microlitre Millimeter Million Micromole Nanometer Pounds per square inch gauge Revolutions per minute Separate hydrolysis fermentation Simultaneous saccharification and fermentation Volume per volume Weight per weight  AFEX  ASL β CBU cm  FPU ft g h HPLC IPE  IU kDa  kWh L M mg  min mL  l mm MM mol nm psig rpm SHF  SSF v/v w/w  xiv  Acknowledgements First, a thank you for the funding received from the Agricultural Biorefinery Innovation Network (ABIN) and the National Science and Engineering Research Council (NSERC), without which I wouldn’t have had this opportunity. Next, I would like to give my most sincere thank you to my supervisors Dr. Sheldon Duff and Dr. Dusko Pasarac for their continued support, encouragement, guidance, and mentorship. You’ve both helped me to really grow as an engineer and an individual and for that I am truly grateful. I would also like to thank my lab mates for helping to create such an enjoyable environment to work in. Finally, I would like to thank my friends, family, Mom and Dad, and fiancée Jacqueline for all the support they provided throughout my journey.  xv  Dedication  For Jacqueline  xvi  1 Introduction 1.1 The need for a new transportation fuel With the growing acceptance that greenhouse gas emissions produced through human activities are directly involved in climate change, increased importance has been placed on discovering and implementing new methods of reducing greenhouse gas emissions. In 2004, Environment Canada reported that the transportation sector was responsible for 190,000 kilotonnes of carbon dioxide equivalents, or one quarter of Canada’s total greenhouse gas emissions [2]. Therefore, the need for clean transportation energy sources to help curb growing greenhouse gas emissions is urgent.  Energy Transportation Mining, Manufacturing & Industrial Processes Residential, Commercial & Institutional Sectors Land-Use, Forestry & Agricultrual Activities  Figure 1.1 Canadian greenhouse gas emissions (CO2 equivalents) by sector (2004) [2] 1.2 Biofuel as a clean transportation energy source Biofuel represents a green technology that has the potential to drastically reduce carbon dioxide emissions due to the carbon life cycle of plant-based fuels. Plants absorb carbon dioxide as a source of carbon through photosynthesis. Engineers can convert plant carbon into biofuels such as biodiesel and bioethanol. On combustion of these biofuels, the carbon dioxide which was previously absorbed by the plant source is emitted back into the atmosphere. It is 1  because of this carbon cycle that substantial carbon dioxide emission savings are possible. Figure 1.2 illustrates some of the different types of biofuel and methods of producing it.  Second Generation  First Generation Plant biomass  Non-food  Food  Lignocellulosic biomass  Starch  Sucrose  Oils  Pretreatment  Acid hydrolysis  Enzymatic hydrolysis  Hydrolysis  Extraction  Gasification  Condensation  Fermentation  Extraction  Fisher-tropsh diesel  Ethanol  Biodiesel  Figure 1.2 Methods of producing biofuel as presented by Gomez et al. [3] Bioethanol, which is ethanol derived from plant sources, is a particularly appealing biofuel because it can be blended with currently-used petroleum based fuels [4]. This would allow for the immediate yet gradual implementation of biofuel into current transportation systems, which will allow the biofuel industry to grow at its own pace. Blends up to 10% (v/v) ethanol (E10) are possible without any modification to current North American automobiles and is covered under current automobile warranties [5]. At higher ethanol blends, such as an 85% ethanol blend (E85), relatively minor engine modifications are required. Not only does blending ethanol with gasoline reduce the consumption of fossil fuels, but it also adds beneficial aspects to the combustion reaction. Ethanol has a higher octane rating, a measure of a fuel’s compression limit, than gasoline which can prevent preignition, improving engine  2  performance. Ethanol also introduces increased oxygen content, resulting in more complete combustion, reducing the formation of pollutants such as carbon monoxide and ozone [6]. 1.2.1 Worldwide bioethanol production trends A rapid increase in worldwide bioethanol production has been observed over the last decade. In 2009, 73.9 billion litres of bioethanol was produced, representing a 4.3 fold increase from bioethanol production in 2000 [7] [8]. The United States and Brazil are the largest bioethanol producers, accounting for over 90% of the total world production (Figure 1.3) [7]. Canada is currently a minor producer of bioethanol, however, recent government mandates (5% renewable fuel blend in gasoline as of 2010) and government funding (1.05 billion dollars for pilot and demonstration scale plants of cellulosic bioethanol) is expected to result in increased production [9]. India, 0.3  France, 0.25  Others, 1.65  China, 1  USA, 18.3 Brazil, 17.5  Figure 1.3 Bioethanol production by country in 2006 (billion litres) [7] 1.2.2 Additional advantages of biofuel The push to advance biofuel production in North America is not driven solely by environmental aspirations [10]. Desires to increase energy security and bolster the struggling agricultural sector are other important driving forces. Sims et al. [11] highlight the social and economic advantages to switching to biofuel, pointing at job creation, rural development, and the 3  potential for future carbon credit trading as three of the social-economic advantages to producing biofuel. At the root of energy security concerns in North America is the fact that oil is a finite resource that is rapidly being depleted. To add to this problem is the fact that North America has a disproportionately large appetite for oil, as indicated by the 23 million barrels consumed per day in 2009, accounting for more than 25% of annual worldwide oil consumption [12]. Finally, there are the fluctuating political relationships between North American countries and some oil rich countries such as those in the Middle East [9]. The result of this is unpredictable, yet generally rising oil prices. Oil prices peaked at around $136/barrel in 2008, and despite a drop to $35/barrel as a result of the global rescission, the cost has risen back to $120/barrel just three years later as shown in Figure 1.4 [13]. Part of this rise in cost is related to the depletion of conventional oil reserves. As a result, less conventional and more expensive oil, such as oil found in the Alberta oil sands is being exploited. As oil reserves continue to decline, it is predicted that the cost of oil will continue to rise. On the other hand, biofuels can be produced within North America from renewable feedstocks, reducing dependence on foreign oil and increasing energy security. 160 140  $/barrel  120 100 80 60 40 20 0 Apr 15, 1998  Jan 09, 2001  Oct 06, 2003  Jul 02, 2006  Mar 28, 2009  Dec 23, 2011  Figure 1.4 The average worldwide price of oil from 1998-2011 [13]  4  1.2.3 First generation bioethanol and the associated problems The term “first generation” is used to describe bioethanol produced from a plant source that would otherwise be used as food [3]. The concept behind it is simple, common food grains are composed almost entirely of starch, a glucose sugar polymer, making them ideal for bioethanol production. Starch is easily digested by the human body, and powerful enzymes have been isolated that are capable of reproducing this digestion of starch to glucose by facilitating a hydrolysis reaction. Yeast is then used to convert the glucose sugar to ethanol, which is distilled out as a concentrated product. The remaining undigested proteins can be used as animal feed called dried distillers grain (DDG). A typical process diagram for bioethanol production from corn is shown in Figure 1.5.  Enzyme  Yeast Ethanol (94%)  Corn  Milling  Condensate  Hydrolysis  Fermentation  Distillation  Evaporator  Centrifuge  Syrup  DDG Dryer  Figure 1.5 Process flow diagram for the production of bioethanol from corn [14] In 2001, 90% of the biofuel (and almost 100% of bioethanol) produced in the United States was derived from corn, and would therefore be considered first generation [15]. Several studies, however, have revealed that there are sustainability issues surrounding the production of first generation biofuel. The most obvious problem is that food is being consumed to produce fuel 5  rather than feed people. As a result, first generation biofuel production has been linked to rising food costs and has been deemed unethical by groups such as Oxfam [16] [17] [18]. Another problem stems from research suggesting that current first generation biofuel practices are doing little to reduce carbon emissions compared to conventional petroleum due to the large amount of energy, and therefore carbon dioxide emissions, required to grow and harvest the energy intensive crops. There are mixed reports regarding the energy output to input ratio of first generation bioethanol derived from common food sources such as corn. Some studies suggest a negative energy ratio, meaning more energy is consumed in the growth of plants and production of ethanol than can be produced through the combustion of the ethanol [3]. Pimentel et al. [19] report a 29% larger energy input than output for corn bioethanol [19]. This, however, contradicts the findings of Hill et al. [20] who report a 25% surplus in energy for corn bioethanol. The difference in findings likely stem from the different assumptions used when conducting their respective life cycle assessments. The bottom line is that even with the optimistic study results the energy gains are small, and could be improved with more suitable less energy-intensive feedstocks. 1.2.4 Second generation bioethanol and its advantages Second generation bioethanol refers to ethanol produced from lignocellulosic biomass. This process involves the conversion of cellulose and hemicellulose molecules found within lignocellulosic plant material, into fermentable sugars through hydrolysis. Yeast fermentation then converts the sugar into ethanol. The major advantage to lignocellulosic based bioethanol is threefold. First, there is an abundance of presently unused lignocellulosic material, as approximately 90% of the annual global production of plant mass is lignocellulosic material [21]. Secondly, lignocellulosic bioethanol does not use human food resources and can therefore end the fuel versus food debate. Finally, research indicates that bioethanol produced from lignocellulosic material could see a far greater return in carbon savings, with carbon dioxide reductions in the 80-90% range observed when compared to conventional gasoline [3] [22] [23]. This is largely due to the 6  reduced energy inputs for growing and harvesting lignocellulosic feedstocks such as switchgrass or waste residues, resulting in very favourable energy returns. For example, Schmer et al. [23] report that 540% more renewable energy is produced than non-renewable energy consumed for bioethanol produced from switchgrass. A life cycle analysis of this bioethanol demonstrated a 94% reduction in greenhouse gas emission compared to gasoline [23]. 1.2.5 Characteristics of ethanol as a fuel Opponents of bioethanol often point to its energy density as a problem. The energy densities of common automotive fuels are displayed in table 1.1. Ethanol has an energy density of 35.0 MJ/kg, roughly two thirds the energy density of gasoline. Opponents of bioethanol suggest that this makes it an inferior fuel, but in reality the only significant drawback resulting from this is the need for a larger fuel tank which could have a negative impact on fuel economy. Lynd et al. [5], however, point toward a Ford Motors study which concluded that due to the superior octane rating and combustion characteristics of ethanol, with an optimized engine, an ethanol fuelled car can actually travel about 80% of the distance that an equivalent gas car could on the same volume of fuel, reducing this drawback. Table 1.1 Energy density of common transportation fuels sources [11] Fuel Diesel Gasoline Natural gas Biodiesel Ethanol  Energy Density (MJ/kg) 48.6 51.6 55.7 43.7 35.0  1.3 Lignocellulosic biomass 1.3.1 The main components of lignocellulosic biomass Lignocellulosic biomass is comprised of three main components, cellulose, hemicellulose, and lignin. Cellulose and hemicellulose are carbohydrates and represent the portion of the biomass that can potentially be broken down into simple sugars for conversion into ethanol. Lignin is a structural polymer that gives rigidity to the plant structure. An increased understanding of 7  each of these components can aid in developing strategies that maximize the efficiency and yields of lignocellulosic bioethanol production. Cellulose (Figure 1.6), the most abundant polymer on earth, is composed solely of six carbon (hexose) glucose molecules linked by β-(1,4)-glycosidic bonds [24]. Cellulose can only form linear chains without branching, however, through hydrogen bonding tertiary structure between cellulose chains are formed, creating semi-crystalline cellulose microfibers [3]. This crystallinity is important in terms of biofuel production because a more crystalline structure is going to be more resistant to enzymatic hydrolysis [25]. The close interactions cellulose has with hemicellulose and lignin, through intramolecular interactions, can also reduce its accessibility to enzymes [24].  Figure 1.6 Cellulose repeating unit cellobiose [26] Hemicellulose is a heteropolymer, in that it can consist of different sugar subunits, both pentose and hexose. It is a highly branched polymer, and interacts strongly with cellulose through hydrogen bonding [3]. Hemicellulose in lignocellulosic plant material consists primarily of the pentose sugars xylose and arabinose, however, small amounts of the hexose sugars mannose, galactose, and glucose can also be present [24]. Figure 1.7 is a depiction of hemicellulose composed of xylose.  8  Figure 1.7 Hemicellulose represented as a chain of xylose sugars [27] Lignin is an aromatic polymer comprised primarily of phenolic rings (Figure 1.8). Lignin forms interactions with cellulose and hemicellulose, essentially surrounding these sugar chains, and adding rigidity to the cell wall structure. This has major implications to bioethanol production, as lignin is well known for its recalcitrance to enzymatic digestion, which can shield cellulose and hemicellulose from hydrolytic enzymes [3]. As a result much research is put into methods of detaching lignin from cellulose and hemicellulose to remove the resistant layer prior to attempting enzymatic degradation of the carbohydrates. Lignin is difficult to completely degrade, so research has also explored utilizing lignin as a combustible fuel to supply power needed for bioethanol production [5].  9  Figure 1.8 Lignin [26] 1.3.2 Lignin as a co-product of bioethanol production Lignin makes up a substantial portion of several lignocellulosic feedstocks, so utilizing it as a coproduct of bioethanol can dramatically improve the economics of the process [28] [29]. Due to the unique structure of lignin, which is high in aromatic rings, some research has gone into methods of converting lignin into valuable chemical co-products. Electrically conductive polymers, plasticizers, and soil stabilizing binders are among the potential uses touted for lignin [30] [31]. These technologies, however, are not yet ready for large scale applications, and instead lignin is commonly combusted for energy and steam production at the industrial level. Depending on the substrate and process used, good utilization of lignin can provide the potential for production of between 8,363–11,150 kJ per litre of ethanol, as estimated by Lynd et al. [5]. Therefore, it is important to consider not only ethanol production but also lignin utilization when exploring bioethanol production processes. 1.3.3 Lignocellulosic feedstocks Lignocellulosic biomass offers a diverse pool of substrates to work with, which includes both urban waste and waste materials from the agricultural and forestry sector (Table 1.2), as well as opportunities for the growth of dedicated energy crops such as switchgrass or fast growing trees like poplar hybrids (Table 1.3). High levels of cellulose and hemicellulose and low levels of lignin are sought after properties when identifying suitable substrates for bioethanol 10  production. For waste lignocellulosic biomass large quantities of waste in centralized locations are ideal. For energy crops, a high productivity and low energy input per unit land mass is optimal. Energy crops are typically suitable for growth on marginal land and waste residues are by-products from existing industry. Therefore, increasing production of lignocellulosic bioethanol can come without increased competition for farmland. Table 1.2 Cellulose, hemicellulose, and lignin content in waste lignocellulosic residues [32] Lignocellulosic Material Nut shells Corn cobs Grass cuttings Paper Wheat straw Sorted refuse Leaves Cotton seed hairs Newspaper Waste paper from chemical pulps Primary wastewater solids Swine waste Solid cattle manure Hardwoods stems Softwood stems  Cellulose (%)  Hemicellulose (%)  Lignin (%)  25-30 45 25-40 85-99 30 60 15-20 80-95 40-55 60-70  25-30 35 35-50 0 50 20 80-85 5-20 25-40 10-20  30-40 15 10-30 0-15 15 20 0 0 18-30 5-10  8-15  NA  24-29  6.0 1.6-4.7 40-55 45-50  28 1.4-3.3 24-40 25-35  NA 2.7-5.7 18-25 25-35  Table 1.3 Cellulose, hemicellulose, and lignin content in dedicated lignocellulosic crops [32] [33] [34] Lignocellulosic Material Switch grass Poplar hybrid Silver maple Weeping lovegrass  Cellulose (%)  Hemicellulose (%)  Lignin (%)  37-45 49 46 37  19-31 22 19 22  12 23 21 21  11  1.3.4 Agricultural residues as a lignocellulosic biomass feedstock The utilization of waste lignocellulosic residues offer a great opportunity for a low cost feedstock that does not compete with current farmland, nor does it require any deforestation for new land. Agricultural residues have gained interest because of the large quantities. For example, in Canada, an estimated 43.1 million tonnes of agricultural residues are produced annually [35]. In 2001 agricultural residues produced in the United States and the World were estimated to total 488 and 3758 million tonnes, respectively [36]. These numbers are substantial; however, not all of this biomass is likely to be readily available for biofuel use. Currently there is much research exploring the fraction of these residues that are surplus, as some of this biomass must be used as tillage to prevent soil erosion and maintain the soil nutrient and organic matter levels. Different reports have concluded between 20–70% of the biomass is available for use [36] [37]. This large discrepancy indicates that availability is region dependent, and that further research is required so that proper standards can be set. Even when tillage requirements are accounted for, a significant amount of agricultural residues are available for bioethanol production. Corn stover and wheat straw are the two largest sources of Canadian and North American agricultural residues [38]. Within North America, Kim et al. [37] reports that enough corn stover and wheat straw is available annually to produce 38 billion litres and 15 billion litres of bioethanol respectively. In this study, the amount of biomass required for tillage was set at enough to cover 60% of the farm soil surface, equalling roughly 40% of the available biomass. A combined 53 billion litres of bioethanol, more then half of the world’s current total production of bioethanol, from just North American corn stover and wheat straw indicates that agricultural residues can be utilized to produce a substantial amount of bioethanol. With an abundant source of agricultural lignocellulosic residue identified in corn stover and wheat straw, the next step was to determine whether the composition of these residues was suitable for bioethanol production. Figure 1.9 shows the availability of starch, cellulose, and hemicelluloses present in wheat and corn. Starch is found entirely in the edible portion of the crops (grains and ears), while the large majority of cellulose and hemicelluloses is present in the 12  non-edible lignocellulosic portion (straw and stover). Similar to how the starch of the food portion of the crop can be hydrolyzed and fermented into ethanol so can cellulose and hemicelluloses of the lignocellulosic portion. Therefore, the most striking observation from this figure is that the combined cellulose and hemicelluloses of the non-edible portion of the crops is greater then the starch present in the edible portion. This indicates that a greater theoretical ethanol yield per hectare is possible for lignocellulosic biomass than the food portion of the crop that is currently being used to produce first generation bioethanol.  Corn Ear Corn Stover Starch Cellulose Hemicellulose Lignin Other  Wheat Grain Wheat Straw  0  5  Tonne/ha  10  15  Figure 1.9 Composition of the lignocellulosic and non-lignocellulosic portions of corn and wheat [22] 1.4 Pretreatment of biomass for enzymatic hydrolysis 1.4.1 Current obstacles for lignocellulosic bioethanol and the need for substrate pretreatment Presently, lignocellulosic bioethanol is not cost competitive with petroleum-based fuels, and requires government subsidies, as there are still many process cost barriers left to overcome [3]. Due to the low yields and high costs associated with the hydrolysis portion of the process, it has been identified as a major obstacle for economic success [32]. Substrate pretreatment 13  has been shown to be capable of dramatically improving enzymatic hydrolysis yields, and a cost effective pretreatment method would go a long way in reducing costs by improving the utilization of enzymes. Another obstacle lies in the utilization of hemicellulose, which has the potential to improve bioethanol economics by increasing ethanol yields. Enzymes capable of hydrolyzing hemicellulose are available, however, only recently have genetically modified yeast strains capable of fermenting both hexose and pentose been developed [24]. A pretreatment that retains both cellulose and hemicellulose, while improving hydrolysis yields is essential for economic success [39]. 1.4.2 Establishing parameters for an effective pretreatment Enzymatic hydrolysis of untreated lignocellulosic biomass has been shown to display poor sugar yields [40] [41]. These poor yields have been proven to be the results of several major factors. The biggest contributor to low yields is the presence of lignin, which negatively affects hydrolysis in two ways. Firstly, because of the lignin molecule’s close interaction with cellulose and hemicellulose, it acts as a barrier, shielding the carbohydrates from the enzymes [3]. Secondly, the hydrolytic enzymes are known to display non-productive binding with lignin, reducing the concentration of active enzymes, which in turn reduces the rate of reaction [42]. Other factors that contribute to the enzymatic hydrolysis yields of untreated substrates include substrate porosity and cellulose crystallinity, both of which affect the accessibility of enzymes to their binding sites [25]. Figure 1.9 illustrates the theory behind how an effective pretreatment can aid in hydrolysis efficiency. Untreated biomass displays a lignin layer with known recalcitrance to enzymatic attack, and crystalline layers of cellulose chains tightly interacting with one another. An effective pretreatment must disrupt the lignin interaction with cellulose and hemicellulose, increase substrate porosity, and reduce cellulose crystallinity. By reducing the effect of these barriers, the result would be drastically increased efficiency of enzymatic hydrolysis.  14  Figure 1.10 Illustration of the effects of pretreatment on lignocellulosic material [43] With the main factors causing poor hydrolysis yields of untreated lignocellulosic biomass identified, parameters for an effective pretreatment have been established [32] [44]. An efficient pretreatment should: 1) Result in a high recovery of cellulose and hemicellulose 2) Result in a high degree of cellulose and hemicellulose digestibility (through lignin removal, reduction of cellulose crystallinity, or other methods) 3) Have a low operating and capital cost 4) Not produce any by-products that will interfere with downstream operations (i.e. fermentation) 1.4.3 Pretreatment methods Many approaches to pretreatment have been developed to achieve greater hydrolysis results for lignocellulosic biomass. The different pretreatment approaches are commonly separated into three main classes: physical, chemical, and physiochemical.  15  Within chemical pretreatments the two main approaches are to treat the substrate with either acidic or alkaline conditions, each with its own advantages. Exposing substrate to acidic conditions increases accessibility to cellulose by solubilising hemicellulose. Alkaline conditions, on the other hand, increase access to cellulose by causing substrate swelling and delignification [45]. An advantage of alkaline pretreatments is that they tend to display less sugar degradation, resulting in reduced production of fermentation inhibitors [46]. Physical applications such as high temperature or pressure swings can be combined with chemical pretreatments to enhance the overall pretreatment effects. Table 1.4 summarizes some of the more prominent pretreatments from each class (physical, chemical, or physiochemical). This form of classification, however, does not do much to describe the method of hydrolysis improvement. Within table 1.4 each pretreatment has also been classified based on its approach for hydrolysis improvement, which was broken down into the categories: delignification, hemicellulose solubilisation, substrate swelling, increased surface area, and reduced cellulose crystallinity. Physical pretreatments are energy intensive and are not considered cost effective on their own; therefore, they will not be considered further [28]. Dilute acid pretreatment exposes biomass to sulphuric acid with concentrations typically ranging from 0.2 to 2.5% (w/w) and heated to 121°C for 30-60 minutes. The presence of sulphuric acid at these temperatures initiates the hydrolytic reaction of cellulose and hemicellulose. It has been observed that dilute acid pretreatment preferentially targets the hydrolysis of hemicellulose over cellulose [45]. By removing the majority of hemicellulose, cellulose becomes more accessible for enzymatic hydrolysis. This method has been shown to greatly improve cellulose hydrolysis yields; however, the production of fermentation inhibitors through sugar degradation is a problem, as it leads to reduced ethanol yields [47].  16  Table 1.4 Classification and method of improvement for common lignocellulosic biomass pretreatments [44] [48] Pretreatment Milling Irradiation of cellulose Dilute acid Alkaline  Classification Physical Physical  Description of Method Size reduction Gamma rays  Method of Improvement Increased surface area Reduced cellulose crystallinity  Chemical Chemical  Sulphuric acid NaOH or slaked lime  Steam explosion  Physiochemical  Ammonia fibre explosion (AFEX)  Physiochemical  Oxygen Delignification  Physiochemical  High pressure steam with sudden pressure release, Temp ≈ 200°C, (Acid optional) Ammonia, Temp ≈ 100°C, High pressure with sudden pressure release NaOH, Oxygen, Temper ≈ 150°C  Hemicellulose solubilisation Swelling, Delignification Hemicellulose solubilisation, Reduced cellulose crystallinity, Lignin removal  Swelling, Reduced cellulose crystallinity  Lignin removal, Swelling, Some hemicellulose solubilisation, Possible reduction in cellulose crystallinity  AFEX is a pretreatment that exposes biomass to ammonia at moderate pressures (100-400 psi) and temperatures (70-200°C), followed by a rapid release of pressure. The result of this pretreatment is a decrystallization of cellulose, lignin removal, and an increase in cell wall micropore size, all resulting in enhanced accessibility of enzymes to carbohydrate [49]. This pretreatment looks promising, as not only are high sugar yields observed, but it also keeps all the sugar isolated as a solid substrate until hydrolysis (stage 2), reducing the complexity of the process. Finally, AFEX is shown to have little inhibitory effects on fermentation [47]. Lime pretreatment utilizes the alkalinity of slaked lime to remove lignin and create substrate swelling for enhanced carbohydrate accessibility [50]. This pretreatment can occur at low temperature (55°C) with a very long residence time (weeks), or at moderate temperature (100 17  150°C), with a reduced residence time (1-6 hours) [46] [51]. Advantages of this system include the low cost of slaked lime, and the possibility for lime recovery by precipitation of calcium carbonate. Three of the pretreatments outlined in table 1.4 were reviewed by Wyman et al. to observe the different hydrolysis yields obtained when each method was applied [45]. A summary of the results are given in table 1.5. Although hydrolysis yield is the primary indicator for evaluating a pretreatment; reagent costs, pretreatment time, and production of fermentation inhibitors should also be considered in the pretreatment evaluation. Table 1.5 Total sugar yield (wt %) for enzymatic hydrolysis of corn stover with 15 FPU/g glucan following each pretreatment method as observed by Wyman et al. [45] Pretreatment Dilute acid AFEX Lime  Stage 1 Yield (%) 35.1 0 10.2  Stage 2 Yield (%) 56.4 94.4 76.6  Total Yield (%) 92.4 94.4 86.8  Note: Stage 1 refers to pretreatment liquid fraction and stage 2 refers to liquid fraction following enzymatic hydrolysis of pretreated solid substrate  The sugar recoveries given for each pretreatment in Table 1.5 are high, with total sugar recoveries approaching 95%. The hydrolysis time used to achieve each recovery seemed to be inconsistent, with hydrolysis time varying from 72 hours for dilute acid pretreatment, 96 hours for lime pretreatment, and 144 hours for AFEX pretreatment [52] [51] [53]. Nevertheless, all these hydrolysis times represent ultimate yields using long hydrolysis. Reducing the hydrolysis time was shown to have significant effects, for example, total sugar recovery was around 60% for AFEX after 24 hour hydrolysis [52]. This exemplifies the difficulty of comparing the success of different pretreatments, as both improvements on rate of hydrolysis and sugar yield will be important in achieving efficient hydrolysis. Hydrolysis reaction times of 3-6 days may not be realistic for a large scale bioethanol plant, as the reactor sizes needed to accommodate this would be extremely large. 1.5 Oxygen delignification Oxygen delignification is a chemical pretreatment for the removal of lignin from lignocellulosic material. This method was made popular by the pulp and paper industry, which has 18  demonstrated its ability to effectively remove lignin to improve the pulp bleaching process [54]. Because the pulp and paper industry and the lignocellulosic biofuel industry share a common target, lignin removal, the potential for utilizing this process in the biofuel industry makes sense. Our laboratory has identified oxygen delignification as a pretreatment method with great potential for several reasons. First, lignin is a known barrier for enzymatic hydrolysis and oxygen delignification has been demonstrated to selectively remove up to 55% of lignin from pulp biomass [55]. Next, oxygen delignification has been thoroughly researched and become a common process used in the pulp and paper industry. Therefore, the technology is already proven to be cost effective at industrial levels, and the infrastructure already exists at several pulping plants [35]. Finally, oxygen delignification has been shown to produce little known fermentation inhibitors such as furfurals and hydroxymethyl-furfurals, a major problem observed with other pretreatments [56]. Past research has confirmed that oxygen delignification can be effective in enhancing the enzymatic hydrolysis sugar yield of lignocellulosic substrates [35] [41] [57] [56] [58]. For example, a 90% and 174% increase in sugar yields was observed when oxygen delignification was applied to primary clarifier sludge and softwood respectively [35]. For agricultural residues and woody yard waste, cellulose conversion in the 80% range has been shown to be achievable [41] [56] [58]. To achieve this high level of cellulose conversion, however, high temperatures (185-200°C) or high caustic loads (50%) were required [41] [56] [58], which subsequently resulted in a high level of hemicellulose (60-95%) being solubilised to the liquor. Sugar solubilised to the liquor is more difficult to recover because it is diluted to low concentrations and because liquor is not a suitable environment for yeast growth. Reducing sugar solubilisation during oxygen delignification will therefore represents one of the aspects that will be explored for improving the pretreatment, as high utilization of both cellulose and hemicellulose has been identified as being crucial to the economic success of lignocellulosic bioethanol [6].  19  1.5.1 Oxygen delignification chemistry Due to the complex and varied nature of lignin, simplified mechanisms describing the reactions occurring during oxygen delignification have been developed. Firstly, lignin can be simplified as being represented by a guaiacyl unit shown in Figure 1.11. The phenolic compounds present throughout lignin are thought to be the primary group involved in the reactions occurring during oxygen delignification [55]. One such mechanism for oxygen delignification is shown in Figure 1.12. An alkaline induced deprotonation of lignin initiates the reaction, allowing for the formation of radicals in the presence of oxygen. From there, oxidative cleavage of the phenolic ring is possible. This results in the formation of carboxylic acids, causing the destabilization and breakdown of the ring structure, increasing lignin solubility [54].  Figure 1.11 The guaiacyl representative unit for lignin [54]  Figure 1.12 Reaction mechanism for oxygen delignification proposed by Lucia et al. [54] 20  Gierer et al. [59] provides an illustration (Figure 1.13) of a proposed reaction mechanism for the formation of the superoxides necessary for the oxidative cleavage of the phenolic rings present in lignin. These same superoxides are also thought to be responsible for the degradation of carbohydrate [59].  Figure 1.13 Stepwise reduction of oxygen to water that occurs during oxygen delignification, as proposed by Gierer et al. [59] With kraft pulp it has been shown that between 35-55% of the remaining lignin can be removed before selectivity begins to decrease [55]. As the selectivity of oxygen delignification decreases, an increase in carbohydrate degradation is observed. The C-2 sites of cellulose are a likely target for oxidation, which leads to formation of carbonyl groups within the carbohydrates. The presence of these carbonyls, combined with the alkaline conditions of oxygen delignification, can results in the cleavage of glycosidic bonds, which is described as “peeling” [59]. Hemicellulose is known to be much more susceptible than cellulose to solubilisation during oxygen delignification and this has been used to separate cellulose and hemicellulose in two separate fractions [60]. It has also been shown that the extent of hemicellulose solubilisation is largely dependent on the oxygen delignification reaction conditions. As an example, Klinke et al. [60] found that between 27-50% of the hemicellulose in straw was solubilised over a range of reaction temperatures, caustic loadings, and reaction times. Aside from lignin removal, oxygen delignification has also been shown to play a role in altering another parameter key to hydrolysis efficiency, cellulose crystallinity. De Souza et al. [61] demonstrate that during oxygen delignification, there is an initial increase in cellulose crystallinity (0-15 min), followed by a decrease in cystallinity (15-30 min), and finally another gradual increase (30-60 min) (Figure 1.14). This trend may be important when optimizing 21  hydrolysis yields, as mid reaction times may present greater substrate accessibility to enzymes as a result of the reduced cellulose crystallinity.  Figure 1.14 has been removed because of copyright restrictions. The information removed is a visual representation of the effect of oxygen delignification residence time on cotton cellulose fibre crystallinity. The original source of the material is: De Souza, I. J., Bouchard, J., Methot, M., Berry, R., & Argyropoulos, D. S. (2002). Carbohydrates in Oxygen Delignification. Part I: Changes in Celllulose Crystallinity. Journal of Pulp and Paper Science , 167-170.  Figure 1.14 Relative degree of crystallinity of cotton cellulose fibres as a function of oxidation time as observed by De Souza et al. [61] 1.5.2 Oxygen delignification reaction variables The oxygen delignification reaction destabilizes lignin by exposing the substrate to pressurized oxygen in an alkaline environment, at temperatures in the range of 100-200oC. Reaction times typically range from five minutes to one hour. Five main reaction variables for oxygen delignification have been identified: reaction time, reaction temperature, caustic load, substrate load, and oxygen partial pressure. The general effect that reaction time, temperature, caustic load, and oxygen partial pressure have on kappa number, an indicator of lignin content, have been demonstrated within our laboratory by Charles et al. [57] the results of which are summarized in Figure 1.15-Figure 1.16. From these results the following observations can be made: 1. The effect of partial pressure on lignin removal is constant at pressures above 25 psig. 2. Increasing temperature from 125°C to 165°C significantly increases lignin removal. 3. Increasing NaOH concentration from 1% to 3% significantly increases lignin removal. 22  4. Delignification has two distinct stages. Stage one occurs in the first twenty minutes in which the rate of lignin removal is high. Stage two occurs from 20 to 60 minutes, in which the rate of lignin removal is dramatically reduced.  Figure 1.15 Effect of temperature and oxygen partial pressure on kappa number observed by Charles et al. [57]  Figure 1.16 Effect of reaction time and caustic concentration on kappa number observed by Charles et al. [57] Charles et al. [57] found that increasing both temperature and caustic load can result in increased delignification, a result consistent with the oxygen delignification reaction theory. Draude et al. [35] and Bjerre et al. [56] have both demonstrated that increasing reaction temperature and caustic also increase hydrolysis yield, linking lignin removal with enhanced hydrolysis. The roles that these reaction variables have on improving hydrolysis are as follows. Caustic load has a two fold ability to increase hydrolysis yields. First it plays a crucial role in the  23  reaction mechanism of delignification. Second, caustic has a swelling effect on the substrate, weakening intramolecular hydrogen bonding between cellulose and lignin, creating space between the molecules that enzymes can access [61]. Temperature simply enhances delignification by increased the reaction. Reaction temperature must, however, be optimized as it also has an increasing effect on the rate of carbohydrate solubilisation [59]. McGinnis et al. [62] demonstrates that rate of both lignin and glucose solubilisation increased with increasing temperature. The need for substrate specific reaction condition optimization is also demonstrated by McGinnis et al. [62] as a substantial difference in delignification is observed between two separate substrates pretreated under the same conditions. Reaction time ultimately controls the extent of delignification achieved. This requires optimization; however, as reduced specificity has been observed with longer reaction times, resulting in increased sugar loss as the reaction continues [54]. From a techno-economic perspective, a reduction in reaction time will be sought to reduce sugar loss, energy consumption and reactor size. This may mean stopping the reaction after the first 15 minutes, after which a substantial reduction in the rate of delignification is observed [57]. The two parameters oxygen partial pressure and substrate concentration were not explored in this research. With a very clear cut-off point observed for partial pressure around 25 psig (Figure 1.14) the partial pressure of oxygen will simply be set in excess of this point. In an industrial setting, substrate concentration has been shown to have an effect on delignification, likely a consequence of the dispersion of oxygen throughout the reactor [63]. The experiments to be conducted will take place at lab scale, however, in which an idealized oxygen concentration is desired throughout the reactor to eliminate sources of error. Therefore, a low and constant substrate concentration of 2% was employed which will promote ideal mixing conditions.  24  1.5.3 Process flow diagram for lignocellulosic bioethanol production using an oxygen delignification pretreatment The proposed process flow diagram for bioethanol production when oxygen delignification pretreatment is used is shown in Figure 1.17. Two streams leave the pretreatment, a solid and a liquid stream. The solid stream will contain a reduced fraction of lignin and the majority of the cellulose and hemicellulose. Following a wash, this can be fed into the hydrolysis reactor. The sugar produced through hydrolysis will then be fermented to produce ethanol. The liquid stream leaving the pretreatment, called the liquor, will contain solubilised lignin and low concentrations of solubilised sugar. The liquor will be alkaline due to the NaOH addition during the pretreatment.  Enzyme  O2 NaOH  Yeast  Oxygen Delignification Pretreatment  Ethanol (94%)  Solids  Corn Stover  Distillation  Milling CO2  Hydrolysis  Fermentation  Liquor Waste Water  Liquor  Centrifuge Centrifuge Lignin Lignin ppt Steam ppt Lignin Lignin  Dryer  Electricity Combustor  Figure 1.17 Proposed process flow diagram for bioethanol production from corn stover using oxygen delignification pretreatment Fermentation produces a dilute ethanol stream called beer (5-10% ethanol w/w). Through distillation, the concentration of ethanol can approach the azeotroph that exists at 95.6% 25  (w/w). To bring the ethanol to a fuel grade concentration (99.5% w/w), molecular sieve or pervaporation processes are used. The solid fraction remaining following hydrolysis, which is made up primarily of lignin, can be utilized as a by-product. Commonly the solids are combusted to produce steam and electricity to power the plant. There is also potential to recover the lignin that was solubilised to liquor during pretreatment. To recover the solubilised lignin, liquor pH can be adjusted to around pH 4, which has been shown to precipitate soluble lignin [42]. pH adjustment using carbon dioxide produced during fermentation is an appealing possibility. Past attempts at lignin recovery from oxygen delignification liquor have demonstrated only limited success, with 25% of lignin recovered [42]. Greater success (65-75% recovery) has been observed with kraft black liquor [64]. There is also the potential for recovery of sugar that was solubilised to the liquor during pretreatment. Solvent extraction of sugar from liquor is one potential recovery method, as Brennan et al. [65] have demonstrated that 90% sugar recovery is possible using naphthalene2-boronic acid. Both sugar and lignin recovery from the liquor was, however, outside the scope of this research. 1.6 Hydrolysis of cellulose and hemicellulose Hydrolysis is defined as the breaking of chemical bonds via the addition of a water molecule. The hydrolysis reaction is central to the production of bioethanol because it degrades the carbohydrates cellulose and hemicellulose into fermentable sugars that can be converted to ethanol. The hydrolysis of cellulose has the following formula: (C6H10O5)n + nH2O ↔ nC6H12O6  (1.1)  Acid hydrolysis and enzymatic hydrolysis are the two methods commonly employed for hydrolyzing cellulose and hemicellulose.  26  1.6.1 Acid hydrolysis Acid hydrolysis can be performed either as a dilute or concentrated sulphuric acid treatment, each with its advantages and disadvantages outlined in Table 1.6. Concentrated acid provides better yields but requires use of highly corrosive acid concentrations (30-70% w/w) and also suffers from high acid consumption and poor acid recovery. Dilute acid avoids some of these problems; however, it gives lower yields and requires a higher operating temperature [66]. Ethanol yields of 64% of theoretical are reported as the maximums observed for acid hydrolysis when the Scholler process, which incorporates a series of percolation reactors, is used [67]. The primary cause of ethanol yield loss when using acid hydrolysis is the degradation of sugar, producing fermentation inhibitors such as furfurals. This has a two fold effect both reducing the amount of sugar available to ferment, and reducing fermentation efficiencies [66]. Table 1.6 Comparison of key reaction parameters for dilute and concentrated acid hydrolysis [66] Parameter  Dilute Acid Hydrolysis  Acid Consumption Operating Temperature Reaction Time Equipment Corrosion Fermentation Inhibitors Sugar Yield  Low High High Medium High Medium  Concentrated Acid Hydrolysis High Low Low High Low High  1.6.2 Enzymatic hydrolysis Enzymatic hydrolysis, where hydrolytic enzymes specific to cellulose and hemicellulose are used, has been identified as the more promising method of hydrolyzing lignocellulosic carbohydrates [21] [32]. The cost of enzyme, however, remains as one of the roadblocks preventing bioethanol from being cost competitive with gasoline. Working on this problem, major enzyme producer Novozyme has been steadily reducing the cost of these enzymes at an average of 40% per year, and in 2010 released news of dramatic improvements, lowering the estimated cost of enzymes to $0.50/gallon ($0.13/L) of ethanol [68] [69]. With continued advances and the development of a sufficient market it is projected that the cost of these 27  enzymes could be reduced further to $0.10/gallon ($0.03/L) of ethanol [17]. These improvements will likely lead to a shift away from acid hydrolysis to enzymatic hydrolysis for the production of bioethanol. Mechanistically, enzymatic hydrolysis relies on the action of several different enzymes working simultaneously to achieve the hydrolysis of cellulose and hemicellulose. The main enzymes commonly accounted for in the hydrolysis of cellulose are endoglucanases, exoglucanases (cellobiohydrolases), and β-glucosidases (Figure 1.18) [70]. Endoglucanases attack β-1,4glycosidic bonds within a cellulose chain at random intervals, producing fragments of cellulose. It is the action of endoglucanases that is thought to initiate hydrolysis, as the internal cuts loosen the cellulose chains, reducing fractal restraints [71]. Exoglucanases attack at the ends of cellulose chains, and move along a chain in a progressive manner, producing glucose duplexes known as cellobiose units. Finally, β-glucosidases hydrolyse the cellobiose into monomer glucose molecules. Because β-glucosidase acts solely on cellobiose, it is not technically classified as a cellulase [70] [72].  Figure 1.18 Illustration of the actions of endoglucanase, exoglucanase, and β-glucosidase 28  The hydrolysis of hemicellulose requires a similar coordinated effort of enzymes as that seen for cellulose, however, an increased complexity exists because of the heterogeneity of hemicellulose, which is made up of xylose, galactose, mannose, glucose, and arabinose sugar molecules. Endonuclease and exonuclease enzymes specific towards the sugars that make up hemicellulose are required. Fortunately, organisms used to produce cellulase enzymes also naturally tend to produce enzymes specific for the hydrolysis of hemicellulose [70]. It is also common for cellulase enzymes to possess a degree of enzymatic activity towards hemicellulose [73]. To avoid being bogged down by the complexity of cellulose and hemicellulose enzymatic hydrolysis, the overall activity observed by the combined effort of the enzymes was quantified by observing the rate at which the endpoint sugar molecules are liberated per volume of enzyme solution. For measuring the cellulase activity, endo/exoglucanase activity was measured using the National Renewable Energy Laboratory (NREL) standardized filter paper assay. β-glucosidases activity was measured using a widely accepted method presented by Wood & Kellogg [74]. Xylose makes up the vast majority of sugar present in hemicellulose, so xylanase activity was considered a sufficient representation of the hemicellulase activity. Xylanase activity was measured using the method presented by Bailey et al. [75]. 1.6.3 Kinetics of enzymatic hydrolysis of cellulose and hemicellulose Reaction kinetics for the enzymatic hydrolysis of cellulose and hemicellulose are complex in nature due to the many variables affecting the rate at which hydrolysis occurs. Multiple enzymes working together simultaneously, product inhibition, substrate accessibility, enzyme deactivation, and enzyme jamming are the major variables at play. On top of that, the presence of lignin has been shown to play a major role in affecting these variables, and therefore is a significant factor when working with lignocellulosic biomass [42] [71]. When attempting to model the enzymatic hydrolysis of cellulose and hemicellulose, the first thing that can be simplified is the fact that several cellulase and hemicellulase enzymes are acting simultaneously. All the individual enzymes involved can be assumed to have a single combined effect [76] [1]. This allows for the determination of an overall rate constant, rather 29  then an individual rate constant for each enzyme involved. With this simplification, the generalized equation (Equation 1.2) can be used for deriving rate expressions.  (1.2) 1.6.3.1 Cellulase enzyme inhibitors  Now that a simplified equation for deriving enzyme rate constants has been established, it is important to consider all the variables that could be affecting the enzyme kinetics. Product inhibition is a well known effect to consider. Cellobiose is known to inhibit the activity of cellulase enzymes [77]. To avoid this product inhibition, a build-up of cellobiose can be prevented by providing an excess of the cellobiose-degrading enzyme β-glucosidase [74]. Previous studies have optimized the loading of β-glucosidase, showing that loading an activity five times more then the activity of cellulase is sufficient to avoid product inhibition [42]. Glucose, the product of β-glucosidase also displays some inhibitory effects towards both βglucosidase and cellulase enzymes. When concentrations of glucose reached 20 g/L, a 45% inhibition of β-glucosidase activity and 55% inhibition of cellulase activity was observed by Xiao et al. [78]. As glucose reaches concentrations high enough to inhibit β-glucosidase, this subsequently inhibits the cellulase enzymes as cellobiose concentration begins to rise, resulting in a two pronged inhibition of cellulase. Hemicellulose derived sugars such as xylose, galactose, and mannose were also shown to inhibit cellulase activity, but had no effect on β-glucosidase activity [78]. 1.6.3.2 Cellulose crystallinity  In experiments conducted on pure cellulose (without the presence of lignin), pretreatment has been shown to increase enzymatic hydrolysis yields [79]. A change in cellulose crystallinity is thought to be the reason for this. Cellulose chains, particularly in highly crystalline regions, form as parallel chains with little distance between them. This tight packing reduces enzyme accessibility. This can also cause problems when cellulase enzymes, which are in the 60 kDa size range, move along these cellulose chains in different directions and rates [77]. When two  30  enzymes moving along parallel chains in opposite directions impede one another, this is considered enzyme jamming. Experimentally, reduced cellulose crystallinity has been shown to improve enzymatic hydrolysis [25]. Pretreatment that can reduce cellulose crystallinity will therefore have a positive effect on hydrolysis yield [32]. It has been shown that when enzyme concentrations are too high, the rate of hydrolysis decreases, and this is attributed to increased jamming effects [77]. Although some amount of enzyme jamming is inevitable, maintaining an optimum level of enzyme can reduce this effect. 1.6.3.3 The effect of lignin on hydrolysis  The presence of lignin in lignocellulosic biomass has been shown to have significant effects on enzymatic hydrolysis, its presence often dramatically reducing hydrolysis yields. Two main factors accounting for this are lignin’s role in reducing enzyme accessibility to carbohydrate binding sites, and its role in causing non-productive enzyme binding, which reduces the rate of the enzyme-catalyzed reactions by lowering the concentration of active enzymes [71] [80]. As a result, the removal of lignin has become a primary target for many pretreatments aiming to enhance enzymatic hydrolysis. Enzyme accessibility is critical for high yields of hydrolysis to be achieved. Hemicellulose is commonly discussed as a barrier to cellulose; however, with the advancement of hemicellulase enzymes, the hydrolysis of hemicellulose simultaneously increases enzyme accessibility to cellulose by removing that barrier [25]. Lignin, however, remains a major barrier that cannot be removed enzymatically. More then just a physical barrier to enzymes, the chemical interaction between lignin and cellulose is also thought to limit the swelling of cellulose chains, further impeding enzyme accessibility [81]. Several studies have shown that the removal of lignin increases hydrolysis yield, an indicator of increase carbohydrate accessibility [35] [80]. Lignin’s role in nonspecific binding has been shown to both reduce the rate and extent of enzymatic hydrolysis of cellulose and hemicellulose [1] [71]. At the surface of cellulase enzymes are exposed to hydrophobic amino acids, which are though to interact with the 31  hydrophobic amino acids of lignin, creating hydrophobic interactions [82]. One study conducted by Kurabi [71] which looked at cellulase kinetics in the presence of pure cellulose and pure lignin demonstrated that lignin acted as an inhibitor to both cellulase and β-glucosidase [71]. 1.6.4 Modelling cellulase hydrolysis kinetics and the effect of lignin content When plotting the total sugar yield versus time for the enzymatic hydrolysis of lignocellulosic biomass, the shape of the resulting curve has some distinct features. Initial reaction rates are high, and most of the reaction takes place within the first 8-12 hours. At the 24 hour time point, the curve begins to plateau. These trends are observed consistently in literature [1] [83] [84]. Many approaches have been proposed to account for and model enzymatic hydrolysis reactions of cellulose and hemicellulose. Both empirical and fully mechanistic models have been developed, each approach with its own advantages [72]. Empirical models typically show a much higher level of fit with experimental data, however, these models cannot be used outside of the specific reaction conditions under which the model was developed. Mechanistic models offer the advantage of being theory based and do not have the same confining parameters as an empirical model. Creating a fully mechanistic model that accurately predicts results has, however, proven to be a difficult task. Bansal et al. [72] provides a review of the many mechanistic models that have been developed to predict the enzymatic hydrolysis of cellulose. The majority of the models reviewed modelled the hydrolysis of pure cellulose substrates such as Solka floc and Avicel, rather then lignocellulosic substrates. This presents problems when it comes to predicting the hydrolysis of a lignocellulosic substrate, as new factors that affect hydrolysis rate such as substrate accessibility and inhibition by lignin are introduced. Some of the mechanistic models reviewed by Bansal et al. [72] have been developed specifically for lignocellulosic substrates (Table 1.7). A common modelling approach is to include an enzyme deactivation rate term. This deactivation term acts as a catch-all for enzyme  32  deactivation, which is convenient for lignocellulosic substrates where multiple factors (i.e. product inhibition and lignin inhibition) contribute to the deactivation of the enzymes. Table 1.7 Cellulose hydrolysis mechanistic models specific for lignocellulosic substrates Author Luo et al. [85] Schell et al. [84]  Substrate  Pretreatment  Corn cob  Dilute acid (single condition)  Douglas Fir  Kadman et al. [86]  Corn  Ljunggren [83]  Softwood, sugarcane bagasse  Shen and Agblevor [76] Zhang et al. [1]  Dilute acid (single condition) Dilute acid (single condition) none  Steam explosion (single condition) Steam explosion Wheat straw (single condition) Cotton gin waste  Model Basis Adsorption of cellulase to substrate (Langmuir equation) Enzyme deactivation, adsorption  Variable Single condition  Varied enzyme loading  Cellulose adsorption (Langmuir equation)  Varied substrate loading and hydrolysis temp  Enzyme deactivation, enzyme adsorption  Single condition  Enzyme deactivation  Varied Enzyme loading  Enzyme deactivation  Varied enzyme loading  Of the mechanistic models developed for lignocellulosic substrates (Table 1.7), none attempted to account for the changes in hydrolysis resulting from varied lignin content. Instead, models were developed for substrates that had been pretreated with a single set of reaction conditions and, therefore, a fixed lignin composition. Successfully modeling the effect that delignification has on the hydrolysis reaction would not only be a novel finding, but would also be useful for optimizing pretreatment conditions and hydrolysis reaction time. Such a model would also provide information about the role lignin has on controlling hydrolysis reaction kinetics. 1.6.5 Empirically modelling of the effects of pretreatment on enzymatic hydrolysis Empirical models have been shown to be a useful tool for predicting response variables that are central to the evaluation of a bioethanol production process. This approach has been used 33  successfully to model the effect of various pretreatments on substrate composition and enzymatic hydrolysis yields. Since an empirical model cannot be used with confidence outside of the range of conditions used to develop the model, it is important to choose a suitable range of independent variables. A compilation of some of the empirical models developed for assessing bioethanol production is shown in Table 1.8. Table 1.8 Literature review of empirically based models for lignocellulosic bioethanol hydrolysis Substrate  Pretreatment  Rye straw Bermudagrass  Dilute sulphuric acid  Wheat straw  Alkaline  Poplar wood  Peracetic acid, KOH, Ball milling  Independent Variables Acid concentration, Reaction time Delignification, Crystallinity Delignification, Crystallinity, Acetyl content  Corn stover  Lime  Lignin  Sugar cane bagasse  Acid hydrolysis, Alkaline  pH, Enzyme loading, Temperature Hydrolysis pH, Hydrolysis temp, Enzyme concentration  Food waste  None  Response Variables  Reference  Monomeric sugar yield  Sun and Cheng [87]  Maximum degree of saccharification  Glucose and xylose yield  Koullas et al. [88] Chang and Holtzapple [89] Kim and Holtzapple [90]  Glucose concentration  Vasquez et al. [91]  Sugar concentration, Ethanol concentration  Kim et al. [92]  Total sugar yield  1.6.6 Enzymatic hydrolysis conditions Enzymatic hydrolysis at the laboratory scale is commonly performed in shaker flasks. Hydrolysis reaction conditions of 45-50oC and 150-200 rpm agitation speed are commonly used for the enzymatic hydrolysis of cellulose and hemicellulose [40] [42]. Other parameters that have been explored in literature as hydrolysis variables include substrate concentration and enzyme loading. An optimization of enzyme loading using Novozyme’s Celluclast has been previously conducted for recovered forestry fibres in our laboratory. Results showed little difference between 20 and 40 filter paper units (FPU) of cellulase per gram of substrate, and 34  therefore an enzyme loading of 20 FPU/g substrate was adopted as the regular enzyme loading concentration. β-glucosidase added at five times the concentration of cellulase has also been successfully used to limit cellulase inhibition due to cellobiose build up, and this practice was also adopted [42]. 1.7 Yeast fermentation Bioethanol is typically produced from the fermentation of hydrolysed sugars. Following the culturing of yeast at aerobic conditions, fermentation occurs at microaerophilic conditions to induce ethanol production, which converts one glucose molecule into two ethanol molecules as described in Equation 1.3. C6H12O6 → 2 C2H5OH + 2 CO2  (1.3)  With two molecules of carbon dioxide as a by-product, the theoretical yield attainable is 0.51 grams ethanol per gram of glucose [93]. 1.7.1 Pentose fermentation Traditionally, Saccharomyces. cerevisiae has been the yeast strain used to ferment ethanol from sugar, however, native S. cerevisiae cannot ferment pentose sugars. This has been acceptable for use with starch based biomass, but presents a major problem for lignocellulosic biomass in which a large fraction is made up of hemicellulose. The conversion of this hemicellulose to ethanol is considered as being essential for a profitable bioethanol from lignocellulosic biomass process [6]. As a result much work has been placed on developing a strain of yeast or bacteria capable of fermenting both pentose and hexose sugars. In nature, yeast and bacteria strains capable of fermenting pentose sugars exist, opening up the possibility for genetic engineering [94]. Zaldivar et al. [24] assessed the state of genetically modified organisms capable of pentose and hexose fermentation and concluded that modified forms of S. cerevisiae, Zymomonas mobilis, and Escherichia coli all have shown great potential. Each organism has a strain shown to be capable of producing ethanol yields from mixed sugar (pentose and hexose) sources in the 90% theoretical range [24]. Dien et al. [95] showed the potential for fermentation using an E.coli strain containing the plasmid pLOI297, which contains 35  genes from Zymomonas mobilis that are necessary for converting pyruvate to ethanol. For each sugar (arabinose, glucose, and xylose), or when a mixture of all three sugars was supplied, ethanol yields of 90-91% were demonstrated [95]. Finally, Yu et al. [96] demonstrated efficient simultaneous fermentation of pentose and hexose present in spent sulphite liquor using Candida shehatae. Combined, these findings suggest that the simultaneous fermentation of both pentose and hexose sugars is currently possible, further emphasising the need to develop a bioethanol process that effectively utilizes both the cellulose and hemicellulose fraction of biomass. 1.7.2 Fermentation process options There are two main process options used for the hydrolysis and fermentation of biomass: separate hydrolysis and fermentation (SHF), and simultaneous saccharification and fermentation (SSF). A simplified process diagram for each is shown in Figures 1.18 - 1.19. As the names suggest, for SHF the hydrolysis and fermentation reactions occur in separate reactors, while for SSF they occur in a single reactor. The advantage to SHF is simple, each reactor can use conditions optimized to the separate reactions taking place, unlike SSF where a compromise of reaction conditions that accommodates fermentation and hydrolysis is required. SHF is also useful in a lab setting, as each step can be evaluated easily. SSF, however, appears to be emerging as the more efficient process. By combining the reactions SSF has the following advantages [97]: 1. Sugar produced from hydrolysis that is inhibitory to cellulase enzymes is immediately consumed by yeast and converted to less inhibitory ethanol molecules 2. A single reactor reduces capital cost 3. The presence of low concentrations of ethanol within the hydrolysis/fermentation reactor reduces the risk of microorganism contamination The net result of these advantages is increased overall production efficiency, with higher observed ethanol yields [94]. SSF was not, however, incorporated into this research, as the focus is on improving pretreatment to enhance hydrolysis.  36  Enzyme Production  Lignocellulosic Biomass Pretreatment  Hydrolysis  Fermentation  Figure 1.19 Separate hydrolysis and fermentation process option [97]  Enzyme Production  Lignocellulosic Biomass Pretreatment  Hydrolysis  Fermentation  Figure 1.20 Simultaneous saccharification and fermentation process option [97] 1.8 Aspen Plus process modelling Aspen Plus is a commercial simulation program that allows for the modelling of each unit involved in a chemical process. Previous research has shown that with certain assumptions, a model of the entire bioethanol process can be developed and laboratory specific data can be incorporated into the model [40]. The National Renewable Energy Laboratory (NREL) has been 37  involved in the development of a biofuel database, which provides the physical properties of many components typically seen in bioethanol production that would not otherwise be available in the Aspen Plus database [98]. This database is available for public use. Following the development of the physical property database, NREL proceeded to model first generation bioethanol production from corn and second generation bioethanol production from corn stover using a dilute acid pretreatment [14] [99]. Aden et al. [99] demonstrated how Aspen Plus can be combined with Aspen IPE (Icarus Process Evaluator) to conduct a complete economic analysis of the bioethanol process. Through this, ethanol production cost, which included capital cost, was calculated to be $0.28/L. Aspen Plus was also utilised to conduct a sensitivity analysis to explore how changes in assumptions or reagent costs affected ethanol cost [99]. To date, no aspen simulation of bioethanol production from lignocellulosic biomass using an oxygen delignification pretreatment has been published. Aspen Plus can also be used to optimise process conditions to minimize process costs. An example of this is shown by Piccolo and Bezzo [100], in which energy consumption was minimised by optimising the use of heat exchangers to efficiently heat and cool the system. Utilization of an Aspen Plus to optimise oxygen delignification reactor conditions for minimal ethanol production costs will be explored.  38  2 Research Objectives The objective of this project was to evaluate the potential of oxygen delignification as a pretreatment for bioethanol production from corn stover and wheat straw using enzymatic hydrolysis. An economic analysis of bioethanol production was required to complete this assessment. To produce this economic analysis and evaluate whether oxygen delignification is an ideal pretreatment for bioethanol production, the following objectives were performed. First the effect of oxygen delignification reaction conditions on enzymatic hydrolysis had to be studied. The independent variables that were included in the study were reaction temperature, residence time, and caustic load. To gain an understanding of why oxygen delignification was affecting enzymatic hydrolysis, a compositional analysis of pretreated substrates was conducted. A mechanistic model that predicts the effect of lignin removal on enzymatic hydrolysis kinetics was then produced. This was followed by the development of empirical models to predict enzymatic hydrolysis sugar yield and substrate composition as a function of oxygen delignification pretreatment conditions. Next, to aid in the economic analysis of bioethanol production and economic optimisation of pretreatment reactor conditions, a simulation of the process was developed in Aspen Plus. The empirical models developed for predicting enzymatic hydrolysis sugar yield and substrate composition were incorporated into the simulation. Aspen Icarus Process Evaluator (IPE) was then used to calculate the capital cost of the bioethanol plant. Finally, utilising results from the Aspen Plus and Aspen IPE simulation, an economic analysis of bioethanol production was conducted in Excel. The effect of oxygen delignification pretreatment conditions on ethanol cost was examined and the conditions resulting in the lowest ethanol cost were determined. To complete the economic analysis, a sensitivity analysis was performed to explore how changes in the cost of enzyme, NaOH, and biomass would affect the cost of ethanol and the optimal pretreatment reactor conditions.  39  3 Materials and Methods 3.1 Lignocellulosic substrates Corn stover and wheat straw were selected as the representative agricultural lignocellulosic residues for experimentation. The corn stover was supplied by Dr. Shahab Sokhansanj from the University of British Columbia. The wheat straw was supplied by Viterra. 3.1.1 Processing biomass for experimentation Biomass was chopped using a Cuisinart Mini-Prep Plus food processor to a size of <20 mm. The biomass was stored at 4°C, and equilibrated to room temperature prior to experimentation.  Figure 3.1 Cuisinart Mini-Prep Plus food processor  Figure 3.2 Processed corn stover and wheat straw 40  3.1.2 Procedure for determining biomass moisture content To determine the percent moisture of each biomass substrate, a sample was taken and an initial wet weight measurement was recorded. The sample was then placed in an oven and dried overnight at 105°C. Following overnight drying, the sample was moved to a desiccator where it could cool without absorbing moisture. After one hour of cooling, a dry weight measurement was recorded. Sampling was performed in duplicate. (3.1) 3.1.3 Procedure for compositional analysis of biomass The composition of corn stover and wheat straw was analyzed to determine the cellulose, hemicellulose, lignin, and ash content present in the substrate. Samples were prepared for analysis by drying at 40°C until they reached a moisture content of <10%. This preparation was adopted from the NREL method of sample preparation for compositional analysis (LAP 510-42620). Follow substrate preparation, NREL’s procedure for the determination of structural carbohydrates and lignin in biomass (LAP 510-42618) was followed. In a glass test tube, 3.0 mL of 72% (w/w) sulphuric acid was added to 300 mg (dry weight) of substrate. The solution was stirred continuously with a glass rod for one minute. The test tube was then placed in a 30°C water bath for one hour, with vortexing every 10 minutes. The solution was then diluted with 84.0 mL of distilled H2O to achieve a 4% (w/w) sulphuric acid concentration, and was qualitatively transferred to a serum bottle. The serum bottle was sealed with butyl rubber septum and crimped aluminum seal and autoclaved in a Midmark M11 UltraClave at 121°C for one hour. Next, the serum bottles were cooled slowly to room temperature and the contents were suction filtered through pre-weighed medium coarseness glass crucibles, upon which the solid fraction was separated from the filtrate. Acid insoluble lignin and ash make up the solid portion captured by the crucible. The crucible was dried at 105°C overnight and weighed. From this the lignin plus ash weight was 41  determined quantitatively. The crucible was then placed in a Thermo Scientific Thermolyne bench-top furnace set at a temperature of 575°C for four hours (Figure 3.3). The crucible was then cooled in a desiccator and weighed. The remaining weight represents the weight of the ash. From this, the weight of acid insoluble lignin could be determined by subtraction.  Figure 3.3 Thermo Scientific Thermolyne bench-top furnace Acid soluble lignin and hydrolyzed sugar were present in the filtrate. HPLC was used to analyze the concentration of sugars present in the filtrate. Glucose was assumed to be hydrolyzed from cellulose, and xylose, arabinose, galactose, and mannose sugars were assumed to be hydrolyzed from hemicellulose. A conversion factor of 1.11 g sugar/g cellulose and 1.14 g sugar/g hemicellulose were applied to account for the addition of a water molecule during hydrolysis. The hemicellulose conversion factor was estimated based on the molecular weight of xylose. The procedure for HPLC analysis is described in section 3.1.4. Acid soluble lignin (ASL) was measured using spectroscopy at a wavelength of 320 nm. Samples were analyzed within six hours of generation, to ensure accurate results. Samples were diluted 42  with distilled H2O to ensure absorbance readings fell within the range of 0.7-1.0. Calculations for ASL are as follows: ASL (g/L) =  (3.2)  Where: A = Absorbance at 320 nm df = Dilution factor b = cell path length (1 cm) a = Absorptivity at lambda max, equals to 55 (L/g cm) for corn stover [101] 3.1.4 Procedure for measuring sugar concentration using HPLC High performance liquid chromatography (HPLC) was utilized for sugar analysis and quantification. HPLC works by passing a sample through a stationary column that has an affinity for the components being analysed. The strength of the affinity each component has for the column dictates the component retention time. Ideally, each component being analysed will have a unique retention time to allow for its identification. The Dionex DX600 HPLC system was used with the operating conditions outlined in Table 3.1. The Dionex CarboPac PA1 column was used as the stationary phase for sugar analysis. It separates sugars based on polarity, and displays a different affinity for each of the sugars (glucose, xylose, mannose, arabinose, and galactose) present in lignocellulosic biomass. A nanopure water mobile phase with a flow rate of 1.0 mL/min was used. The Dionex AS50 autosampler was used to sample 25 l from each vial for sugar analysis.  43  Table 3.1 HPLC operating conditions Parameter Temperature Pressure pH Injection Volume Analysis Duration Eluent (flow rate) Mobile Phase (flow rate)  Value 30°C 500-1000 psi 10-13 25 l 30 min NaOH (1 mL/min) Nanopure water (1 mL/min)  Samples were run for thirty minutes so that all sugars in a sample could be detected by the Dionex ED50 detector as they left the column. The ED50 detector is an electrochemical detector which required a stream of 0.25 M NaOH with a flow rate of 1.0 mL/min passing through it to enhance sugar detection. At the end of each sample run, the column was washed for 10 minutes with 0.25 M NaOH to strip any remaining compounds from the column. Finally, the column was washed with nanopure water for 10 minutes prior to running the next sample. Chromeleon software was used to identify and quantify each sugar peak relative to the retention times and peak areas observed for the sugar standard. To prepare samples for HPLC analysis, samples were vortexed and 100 l aliquots were diluted with 4.8 mL of nanopure water in 5 mL HPLC screw-top vials, such that the sugar concentration would fall within a range of 0.1-2.0 mg/mL. A 5 mg/mL fuctose internal standard was added to each sample to a final concentration of 0.1 mg/mL. A three point calibration curve (2, 0.5, 0.1 mg/mL) was created by diluting a sugar standard containing 5 mg/mL glucose, xylose, mannose, galactose, and arabinose. The Dionex AS50 autosampler was used to sample 25 l from each vial for sugar analysis. 3.2 Enzymatic hydrolysis Commercial enzyme preparations were used for the hydrolysis of cellulose and hemicellulose carbohydrates present in the lignocellulosic substrates. The hydrolysis yields of pretreated and untreated substrates were compared to observe the effect pretreatment had on hydrolysis.  44  3.2.1 Cellulase enzymes The commercial cellulase preparation used for the hydrolysis of cellulose was Novozyme Celluclast 1.5 L, which had activities of 71.2 FPU/mL and 27.4 CBU/mL. Novozyme-50010, a βglucosidase enzyme preparation with activity of 640.5 CBU/mL was used to elevate βglucosidase activity. 3.2.2 Hemicellulase enzymes Hemicellulase activity is approximated by measuring the endoxylanase activity. Endoxylanase activities of 5457 IU/mL and 359 IU/mL were measured for Novozyme Celluclast 1.5 L and Novozyme-50010 respectively. 3.2.3 Procedure for determining cellulase activity To measure cellulase activity the NREL developed method, know as the filter paper assay (LAP 510-42628), was used. This method measures cellulase activity in terms of filter paper units per mL of enzyme solution. A filter paper unit is defined as the amount of enzyme required to liberate 1 mol of glucose from cellulose substrate per minute. The measurement of cellulase activity has been standardized in this procedure by defining 2.0 mg of glucose liberated from 50 mg of filter paper in 60 minutes as the intercept for calculating FPU. To each assay mixture 1.0 mL of 0.05 M Na-citrate pH 4.8 was added to a test tube containing 50 mg Whatman No. 1 filter paper and the tubes were heated to 50°C. Enzyme samples were diluted and 0.5 mL of each enzyme dilution was added to an assay tube in triplicate. The enzyme dilutions had to be such that one sample results in slightly more, and one slightly less then 2.0 mg of total glucose liberated. Enzyme dilutions used are shown in Table 3.2. Samples were capped and incubated at 50°C for one hour. The enzymatic reaction was then stopped by adding 3.0 mL of 3,5-dinitrosalicylic (DNS) reagent and mixing the tubes.  45  Table 3.2 Enzyme dilution Enzyme Sample # 1 2 3 4 5  Concentration* 0.00875 0.00750 0.00500 0.00375 0.00250  * Concentration refers to the proportion of original enzyme solution in the enzyme dilution  All test tubes were then placed in a boiling water bath for five minutes to allow the colorimetric reaction between glucose and DNS take place, after which the samples were moved to an ice bath to terminate the reaction. The test tubes remained in the ice bath until all pulp had settled. Samples of 0.2 mL were diluted with 2.5 mL of distilled water in a cuvette for spectrophotometer absorbance measurement at 540 nm. Assay mixtures, blanks, and controls were compared to a glucose standard curve (Figure 3.4) to quantify glycosidic bond cleavage.  0.8000 0.7000  y = 0.1986x + 0.0035 R² = 0.9986  0.6000  Absorbance (nm)  0.5000 0.4000 0.3000 0.2000 0.1000 0.0000 -0.1000  0  0.5  1  1.5  2  2.5  3  3.5  4  Glucose Concentration (mg/0.5mL)  Figure 3.4 Determination of glucose concentration Glucose concentration for each enzyme dilution was plotted (Figure 3.5). The two enzyme dilutions spanning 2 mg of glucose released were used for interpolation of the enzyme dilution that would precisely liberate 2 mg of glucose. 46  Enzyme concentration versus glucose concentration 0.01  Enzyme conc. spanning 2.0mg glucose  Enzyme concentration*  0.009 0.008  y = 0.0027x - 0.0002  0.007 0.006 0.005 0.004 0.003 0.002 0.001 0 0  0.5  1  1.5  2  2.5  3  3.5  Glucose concentration (mg/0.5mL) * Concentration refers to the proportion of original enzyme solution in the enzyme sample  Figure 3.5 Filter paper unit determination for cellulase activity From this, the activity of the enzyme is calculated using Equation 3.3. The number 0.37 has units of mol/min-mL, which is obtained by converting the 2.0 mg of glucose liberated to mol and dividing by the 0.5 mL sample volume and the 60 minute reaction time. (3.3) [Enzyme] = The proportion of original enzyme solution present in the enzyme dilution 3.2.4 Procedure for determining cellulase concentration The Bio-Rad protein assay was used to measure cellulase concentration. A cellulase standard curve was developed within our lab by Oscar Rosales Calderon using Sigma Cellulase (from trichoderma reesei) powder. Five dilutions (0.2-1.0 mg/mL) of cellulase protein standard were prepared. To test tubes, 800 l of each standard and 200 l of dye reagent was added, and the tubes were vortexed. The test tubes were incubated for 10 minutes and the absorbance measured at 595 nm. From this the cellulase concentration present in Novozyme-50010 was determined to be 129 mg/mL. 47  3.2.5 Procedure for determining β-glucosidase activity β–glucosidase activity was quantified using the method published by Woods and Kellogg [102]. This method measures β–glucosidase activity using a p-Nitrophenyl-β-D-glucoside substrate. p-Nitrophenyl-β-D-glucoside is a cellobiose analog that contains a glucose molecule linked to pNitrophenol through a glycosidic bond. β–glucosidase cleaves this glycosidic bond, liberating the p-Nitrophenol molecule. Addition of glycine buffer causes a colormetric reaction with pNitrophenol, which can be utilised to determine its concentration from a standard curve. β–glucosidase activity in quantified as cellobiose units (CBU) per mL, which is a measure of the amount of enzyme solution required to release 1 mol of p-Nitrophenol per minute during the assay. A p-Nitrophenol standard curve was developed (Figure 3.6) to measure the concentration of pNitrophenol for each sample.  2.5  y = 1.9117x R² = 0.9955  abs @ 430nm  2  1.5  1  0.5  0 0  0.2  0.4  0.6  0.8  1  umol p-nitrophenol / 7mL  Figure 3.6 Determination of p-Nitrophenol concentration Assay test tubes were prepared with 1.8 mL of 50 mM sodium acetate buffer pH 4.8, and 1 mL of a 50 mM acetate buffer containing 5 mM p-Nitrophenyl-β-D-glucoside substrate. The test tubes were heated to 50°C in a water bath. Diluted enzyme samples of 200 l volume were 48  added to the test tubes and the samples were left in the water bath for 30 minutes. The reaction was then stopped by the addition of 4 mL of glycine buffer, which causes a colorimetric reaction. The absorbance of the samples was measured at 430 nm with a spectrophotometer. The concentration of p-Nitrophenol liberated during the reaction was measured from the standard curve (Figure 3.7).  Enzyme Activity (CBU/mL)  580 560 540 520 500 480 460 440 420 400 0  0.00005  0.0001  0.00015  0.0002  0.00025  Enzyme Concentration* * Concentration refers to the proportion of original enzyme solution in the enzyme dilution  Figure 3.7 β–glucosidase activity determination 3.2.6 Procedure for determining endoxylanase activity Endoxylanase activity was measured in the Jack Saddler laboratory at the University of British Columbia with the help of Valdeir Arantes, using the method published by Bailey et al [103]. Endoxylanase activity is expressed as international units per mL, which is a measure of the amount of enzyme solution required to release 1 mol of xylose per minute during the assay. Assay test tubes were prepared by adding 1.8 mL of substrate solution made up of 1.0% (w/w) xylan in 50 mM citric acid phosphate buffer (pH 5.3). The test tubes were heated to 50°C. Diluted enzyme samples of 200 l volume were added to the test tubes and the samples were incubated in the 50°C water bath for 5 minutes. The enzymatic reaction was then stopped by adding 3.0 mL of DNS and mixing the tubes. All test tubes were then placed in a boiling water bath for five minutes to allow the colorimetric reaction between xylose and DNS take place, 49  after which the samples were moved to an ice bath to terminate the reaction. Samples were added to a cuvette for spectrophotometer absorbance measurement at 540 nm. Assay mixtures, blanks, and controls were compared to a xylose standard curve to quantify xylose concentration. 3.2.7 Procedure for lab-scale enzymatic hydrolysis Erlenmeyer flasks (125 mL) were used to carry out enzymatic hydrolysis of pretreated and untreated substrates. Two grams dry weight of substrate was placed into duplicate flasks. The appropriate volume of 50 mM Sodium acetate pH 4.8 was calculated and added to the flask, such that the final volume would be 50 mL. Flasks were equilibrated to 50°C. Novozyme Celluclast (71.2 FPU/mL) was then added to the flask to obtain an activity of 20 FPU per gram of substrate (dry weight). To ensure there was limited enzymatic inhibition due to cellobiose end products, excess β-glucosidase enzyme was added. Novozyme-50010 (640.5 CBU/mL) was added at a CBU:FPU ratio of 5:1. Flasks were loaded into an incubator set at 50°C with 150 rpm shaking speed. 1 mL samples were taken at times: 0 h, 1 h, 8 h, 24 h, and 48 h. Samples were centrifuged for five minutes to pellet undigested substrate, and the supernatant was collected and stored at -20°C. 3.3 Oxygen delignification pretreatment 3.3.1 Defining oxygen delignification reactor conditions A full factorial experimental design was used to study the effects of oxygen delignification reaction time, temperature, and caustic load on enzymatic hydrolysis. The reaction conditions used in the experimental design are outlined in Table 3.3. A solids loading of 2% was used for all experiments to ensure adequate mixing by the light duty magnetic drive mixer. An oxygen partial pressure of 10 bar was selected for all experiments.  50  Table 3.3 Oxygen delignification experimental design Parameter Solids loading (%) Oxygen partial pressure (bar) Reaction time (min) Reaction temperature (°C) NaOH load (%)  Value 2 10 15, 30, 60 90, 120, 150 2, 6, 10  3.3.2 Apparatus used for oxygen delignification Oxygen delignification reactions were conducted in a PARR 4520, one litre bench top reactor (Figure 3.8). A PARR 4843 controller was used to control the temperature and mixing speed of the reactor (Figure 3.9). A gas inlet and outlet valve allowed for the addition of ultra pure grade oxygen and nitrogen, which was supplied by Praxair.  Figure 3.8 PARR 4520 high pressure reactor  51  Figure 3.9 Oxygen delignification reactor setup 3.3.3 Procedure for oxygen delignification Ten grams dry weight of substrate and approximately 200 mL of distilled water was added to a 500 mL tared Teflon cylinder liner. The desired amount of NaOH (2-10% of the substrate dry weight) was then added. Distilled water was then added to reach a final weight of 500 grams. The liner was then placed in the PARR 4520 high pressure reactor vessel. After preparing the reactants, the vessel was sealed by enclosing the split ring and tightening the cap screws and securing the safety drop ring around the split ring. The sealed vessel was placed onto the reactor mount and the heating jacket was raised to the reactor. The reactor was set to a mixing speed of 150 rpm. The reactor was sparged with nitrogen gas for five minutes to remove oxygen. The sparge valve was then closed allowing the reactor to become pressurized with the nitrogen feed to 8.0 bar. The reactor was then heated to the specified reaction temperate (90-150°C). Upon reaching reaction temperature, the nitrogen feed was eliminated and oxygen was introduced into the reactor. The gas release valve was opened slightly to allow a steady flow of oxygen through the reactor, which was kept at a partial 52  pressure of 10 bar. The reactor was maintained at these conditions for the specified reaction time (15-60 minutes). At the end of the run, the gas lines were closed and the reactor vessel was removed from the mount and placed in an ice bath to quickly cool the reactor, quenching the reaction. When the temperature decreased to 90°C the gas outlet was opened to depressurize the reactor. After cooling to room temperature, the reactor was opened and the reaction vessel was removed. The reaction mixture was passed through a Whatman No. 1 filter paper placed in a Buchner funnel under vacuum so that the pretreated substrate could be collected. The solid substrate was washed with four litres of distilled water, collected, and stored at 4°C. A 1 mL sample of the filtrate was frozen for subsequent HPLC analysis. 3.3.4 Procedure for determining solid recovery Following oxygen delignification, samples of the pretreated solid substrate were taken in duplicate and their moisture content was determined following the procedure given in section 3.1.2. This was used to calculate percentage of substrate recovered from the pretreatment. % Pretreatment Recovery =  *100%  (3.4)  3.3.5 Procedure for determining substrate composition following pretreatment To determine the composition of the substrate following oxygen delignification pretreatment, the compositional analysis procedure outlined in section 3.1.3 was performed. A central composite experimental design was used to determine the effects of oxygen delignification reactor conditions on the substrate composition of corn stover. Table 3.4 outlines the pretreatment reaction conditions of the samples included in the compositional analysis. Analysis of pretreated substrates was conducted in a randomized order.  53  Table 3.4 Central composite design for compositional analysis of corn stover Time (min) 15 15 15 15 60 60 60 60 15 60 30 30 30 30 30 30  Caustic Load (%) 2 2 10 10 2 2 10 10 6 6 2 10 6 6 6 6  Temperature (°C) 90 150 90 150 90 150 90 150 120 120 120 120 90 150 120 120  3.3.6 Procedure for determining sugar content in pretreatment Liquor Acid hydrolysis was performed on liquor samples to hydrolyse all solubilised carbohydrate into its monomeric sugar form, for detection with HPLC. Liquor samples were placed in 100 mL serum bottles and diluted with 8% (w/w) sulphuric acid such that the final concentration was 4% (w/w) sulphuric acid. The serum bottles were sealed with butyl rubber septum and crimped aluminum seal, and autoclaved in a Midmark M11 UltraClave at 121oC for one hour. The serum bottles were then cooled to room temperature and measured for sugar concentration using the HPLC procedure given in section 3.1.4. 3.4  Method of simulating scale-up of processes for economic analysis  Aspen plus was used to simulate the mass, energy, and utility balances of the bioethanol production process being studied. Empirically-derived substrate composition and hydrolysis yield equations, which express the response variable as a function of oxygen delignification pretreatment conditions, were incorporated into the simulation. This allowed for scale-up modelling and an economic sensitivity analysis to be performed. Results from the simulation were exported to excel where the economic analysis was conducted. 54  The activity coefficient NRTL model was selected as the physical property method. Henrys law was included to model the concentration of gas dissolved in liquid solutions. Physical properties for components not included in the Aspen Plus database were defined. 3.4.1 Physical property database The physical properties of many components present in the process were not available in the Aspen Plus physical property database, and therefore had to be defined. The physical properties of most unknown components were obtained from NREL’s physical property database for biofuels components [98]. For those unknown components not available in NREL’s biomass database, assumptions were made regarding their physical properties. Table 3.5 outlines the assumptions made for each component that required physical property specification. Table 3.5 Component physical property assumptions Component Biomass Cellulose Hemicellulose Lignin Other Soluble Cellulose Soluble Hemicellulose Soluble Lignin Soluble Ash Soluble Other Glucose Galactose Mannose Xylose Arabinose  State Solid Solid Solid Solid Solid Liquid (aqueous) Liquid (aqueous)  Physical Property Assumptions NREL Database NREL Database NREL Database – (Xylan) NREL Database NREL Database – biomass NREL Database – Solubilised solids NREL Database – Solubilised solids  Liquid (aqueous) Liquid (aqueous) Liquid (aqueous) Liquid (aqueous) Liquid (aqueous) Liquid (aqueous) Liquid (aqueous) Liquid (aqueous)  NREL Database – Solubilised solids Calcium carbonate NREL Database – Solubilised solids NREL Database – Glucose NREL Database – Glucose NREL Database – Glucose NREL Database - Xylose NREL Database - Xylose  Polymeric structures: Polymeric molecules such as biomass, cellulose, hemicellulose and lignin were assumed by NREL to be a single repeat unit when specifying the molecular formula [98].  55  Other: To maintain a mass balance, components that were not tracked in this simulation such as proteins, lipids, and extractives were classified as a single component called “other”. This group was assigned the physical properties of biomass. Solubilised solids: To account for the solubilisation of solid components into an aqueous phase, these soluble components were given the NREL database component properties of “soluble solids”. The vapour pressure of this group is kept low to ensure that the solids would not be flashed into a vapour stream. Hexose sugars: Galactose and mannose were assumed to have the same physical properties as NREL’s glucose component. Pentose sugars: Arabinose was assumed to have the same physical properties as NREL’s xylose component. 3.4.2 Empirically-derived equations To allow for the simulation of oxygen delignification and enzymatic hydrolysis, empirical equations derived from laboratory results were incorporated into the simulation in “calculator blocks” using FORTRAN. JMP (version 7) statistical software was used to perform linear regression on experimental data, producing equations that fit the output data curve. Interactive effects were included in the regression. The degree of fit produced by the equation was calculated within JMP and reported as an R2 value. One set of equations was derived from the central composite experimental design which examined the effects of pretreatment reactor conditions on substrate composition. These equations allowed for a mass balance to be simulated for oxygen delignification, as substrate composition following pretreatment could be accurately predicted. A second set of equations was derived from the full factorial design experiment which examined the effects of pretreatment reactor conditions on enzymatic hydrolysis yields. Because enzymatic hydrolysis sugar yields are affected by pretreatment conditions, these equations were necessary to allow this effect to be modelled in the simulation.  56  4 Experimental Results 4.1 Substrate compositional analysis of corn stover and wheat straw A compositional analysis of corn stover and wheat straw (Table 4.1) provided initial insight into the quality of these feedstocks for bioethanol production. Of primary importance was the carbohydrate content. Both substrates were shown to be relatively abundant in carbohydrate material (60.3%, 63.0%), a crucial requirement when selecting a suitable feedstock, as these carbohydrates can be converted into fermentable sugars for ethanol production. Table 4.1 Compositional analysis of corn stover and wheat straw Substrate Cellulose (%) Hemicellulose (%) Lignin (%) Ash (%) Other (%)  Corn Stover 34.1 (+/- 3.0) 26.2 (+/- 2.4) 19.6 (+/- 4.3) 5.3 (+/- 1.1) 14.8 (+/- 2.8)  Wheat Straw 36.7 (+/- 1.6) 26.3 (+/- 1.5) 19.2 (+/- 3.5) N/A N/A  With the emergence of pentose fermenting yeasts and hemicellulase enzymes, hemicellulose has become a component of interest for bioethanol production. Hemicellulose made up about 26% of corn stover and wheat straw, making it the second most abundant component (after cellulose). This highlights the importance of optimizing the process for maximum recovery and utilization of both hemicellulose and cellulose, rather then just focusing on cellulose recovery. Effective utilization of both carbohydrates is thought to be crucial for the economics of bioethanol production [6]. Finally, lignin, the third most abundant component was shown to make up about 20% of each substrate. This lignin content is lower than in most woody materials; however, it still represents a significant fraction of the biomass. As a consequence: 1. An effective pretreatment is necessary, as high lignin content is tied to low enzymatic hydrolysis efficiency. 2. The recovery of lignin, usually combusted as a carbon neutral energy source, should be explored for improving process economics. 57  The component labelled “other” is made up of extractives, acetate, lipids, and protein. Huang et al. provides a more in-depth analysis of these “other” components [104]. 4.2 Preliminary oxygen delignification experimentation To investigate the suitability of oxygen delignification pretreatment on corn stover and wheat straw, an initial probe into the effect of oxygen delignification on hydrolysis efficiency was conducted. A single pretreatment condition (150°C, 30 minutes residence time, 6% caustic) was tested for comparison of enzymatic hydrolysis against untreated substrate. Figure 4.1 shows the general mass flows observed during experimentation, and Equations 4.1-4.4 the typical responses measured. For yield calculations, to account for the addition of water during the hydrolysis reaction, a conversion factor of 1.11 g sugar/g carbohydrate was used.  Figure 4.1 Experimental mass flow diagram (4.1)  (4.2)  (4.3)  (4.4) In this work, hydrolysability is defined as the amount of sugar produced per mass of substrate hydrolysed (Equation 4.1). To compare the difference in hydrolysability between two 58  substrates, an equal mass (two grams) of pretreated substrate was exposed to the same hydrolysis conditions (20 FPU/g substrate enzyme loading, 48 hour reaction time). Note that hydrolysability does not take into account the composition of the substrate prior to hydrolysis, a point that will be discussed later. The difference in hydrolysability between pretreated and untreated samples (Figure 4.2) indicates that oxygen delignification has a strong positive effect on hydrolysability. Both an initial increase in the rate of hydrolysis and a final overall increase in sugar production are observed for the pretreated samples. Pretreated corn  0.7  Pretreated wheat  0.9  Untreated corn  Untreated wheat  0.8  0.6 Hydrolysability  Hydrolysability  0.7 0.5 0.4 0.3 0.2  0.6 0.5 0.4 0.3 0.2  0.1  0.1  0  0 0  20  40  Hydrolysis time (hours)  60  0  20  40  60  Hydrolysis time (hours)  Figure 4.2 Hydrolysability (20 FPU/g substrate) of untreated and pretreated corn stover and wheat straw samples. Pretreatment was conduced at 120°C with a 30 minute residence and 6% caustic. Although equal masses of the pretreated and untreated samples were loaded for hydrolysis, it is important to note that the mass of carbohydrate loaded was not equal because substrate composition was modified during pretreatment. So the increase in sugar production observed for the pretreated substrate could actually be the result of two factors: 1. An increase in initial mass of carbohydrate, resulting in greater theoretically achievable sugar production. 59  2. Enhanced hydrolysis efficiency as a result of the pretreatment. The extent to which each of these factors plays a role cannot be determined from this experiment alone. So why is hydrolysability an important measure? Hydrolysability is an indicator of the amount of enzyme that will be required for hydrolysis. By increasing hydrolysability of a substrate, the mass of substrate needed to undergo hydrolysis to produce a specific amount of sugar is reduced. Because enzyme is loaded based on substrate mass entering the reactor, increasing hydrolysability will reduce the amount of enzyme required. The cost of enzymes remains a barrier for the production of economically viable bioethanol, so reducing the demand of enzyme could significantly improve process economics [32]. To demonstrate the effect of hydrolysability on enzyme consumption, the amount of enzyme required to produce ten grams of sugar from the untreated and pretreated corn stover samples was calculated (Table 4.2). Hydrolysability was based on 24 hour enzymatic hydrolysis reaction time. From this it is shown that the pretreated sample with increased hydrolysability required less enzyme (364 FPU) then the untreated sample (834 FPU) to produce ten grams of sugar. This demonstrates that increasing substrate hydrolysability reduces enzyme demands. See appendix for sample calculations. Table 4.2 Enzyme consumed to produce 10 g of sugar for untreated and pretreated (30 minutes, 120°C, 6% caustic) corn stover  Untreated Corn stover Pretreated Corn stover  Enzyme loading (FPU/g substrate)  Hydrolysability (g sugar/gram substrate)  Mass of substrate required to produced 10 g sugar (g)  Enzyme required for hydrolysis (FPU)  20  0.24  41.7  834.0  20  0.55  18.2  364.0  60  Another critical parameter is the effect of pretreatment on total sugar yield (Equation 4.3). This will determine how much ethanol can be produced per mass of feedstock. During pretreatment some solubilisation of sugar into the pretreatment liquor is known to occur (Figure 4.1). Recovery of sugar solubilised to the liquor would be difficult, as the sugar concentration would be very dilute and the caustic liquor would not be a habitat environment for yeast. As a result, only sugar remaining in the solid fraction was considered for the total sugar yield calculation. Increasing hydrolysability while reducing sugar solubilisation will maximize sugar yield, a primary goal when pretreatment conditions are later optimized. The comparison of total sugar yield (Figure 4.3) for untreated and pretreated samples demonstrates that despite losing sugar to the liquor during pretreatment, the pretreatment still improves total sugar yield. This is a clear indication of improved hydrolysis efficiency due to pretreatment. With these encouraging initial results, an optimization of oxygen delignification reaction conditions was performed. Pretreated wheat  Pretreated corn Untreated corn  70  80 Total Sugar Yield (%)  60 Total Sugar Yield (%)  Untreated wheat  90  50 40 30 20 10  70 60 50 40 30 20 10  0  0 0  20 40 Hydrolysis time (hours)  60  0  20 40 Hydrolysis time (hours)  60  Figure 4.3 Total sugar yield during hydrolysis (20 FPU/g substrate) of corn stover and wheat straw. Pretreatment for 30 minutes at 120°C with 6% caustic  61  4.3 Oxygen delignification reaction conditions 4.3.1 The effect of oxygen delignification reaction conditions on hydrolysis yield Following the initial results, which indicated that oxygen delignification was a suitable pretreatment for improving the enzymatic hydrolysis of corn stover and wheat straw, an experiment designed to test a broad range of pretreatment reaction variables was conducted. The effect that reaction variables residence time, temperature, and caustic load had on response variables pretreatment recovery, substrate hydrolysability, and total sugar yield was analyzed. Optimal reaction conditions that maximize total sugar yield were then determined. A factorial experimental design was developed for this study. The range of each reaction variable was chosen based on results observed from a previous oxygen delignification experiment conducted on recovered fibre in our laboratory by Ruffell [80]. Increasing pretreatment reaction severity was previously demonstrated to produce opposite effects on the response variables substrate recovery and substrate hydrolysability. Conditions that maximize hydrolysability, while minimizing substrate loss were therefore sought after. For each reaction variable, a range from mild to strong conditions was selected so that the optimum balance between the opposing responses could be found. Table 4.3 outlines the reaction conditions studied. The results from this experiment on corn stover are presented in Table 4.4, and similar results for wheat straw are included in the appendix. Table 4.3 Factorial design of pretreatment reaction conditions Reaction Time (min) % Caustic (w/w) (NaOH/dry corn stover) Reaction Temperature (°C)  15, 30, 60 2, 6, 10 15, 30, 60  62  Table 4.4 Results from oxygen delignification factorial design for corn stover. Hydrolysability is based on an enzyme loading of 20 FPU/g substrate and 24 hour hydrolysis. Pretreatment Conditions Caustic Time Temperature Load (min) (°C) (% w/w) 15 2 90 15 2 120 15 2 150 15 6 90 15 6 120 15 6 150 15 10 90 15 10 120 15 10 150 30 2 90 30 2 120 30 2 150 30 6 90 30 6 120 30 6 150 30 10 90 30 10 120 30 10 150 60 2 90 60 2 120 60 2 150 60 6 90 60 6 120 60 6 150 60 10 90 60 10 120 60 10 150 30 14 120 30 18 120 120 6 120 60 6 180  Pretreatment Substrate Recovery (%) 99.3 103.1 95.8 91.6 68.9 72.4 78.0 76.9 51.4 91.4 98.9 86.7 88.8 69.6 66.9 83.1 71.4 58.6 101.2 83.1 77.2 82.2 59.5 70.2 74.8 63.3 58.0 62.9 53.5 57.6 58.8  Hydrolysability (g sugar/g substrate) Cellulosic Hemicellulosic Total Sugar Sugars Sugar (glucose) 0.14 0.07 0.21 0.23 0.14 0.37 0.10 0.08 0.19 0.16 0.12 0.28 0.30 0.19 0.49 0.35 0.19 0.54 0.31 0.21 0.53 0.45 0.26 0.71 0.52 0.25 0.77 0.12 0.06 0.18 0.16 0.08 0.24 0.20 0.11 0.32 0.18 0.12 0.30 0.34 0.21 0.55 0.39 0.20 0.59 0.28 0.19 0.46 0.37 0.21 0.58 0.42 0.20 0.62 0.15 0.07 0.22 0.17 0.11 0.27 0.24 0.15 0.40 0.25 0.16 0.42 0.31 0.23 0.54 0.44 0.20 0.64 0.41 0.22 0.63 0.42 0.24 0.66 0.58 0.24 0.83 0.45 0.23 0.69 0.49 0.25 0.74 0.37 0.19 0.56 0.48 0.16 0.65  63  A central composite experimental design for the analysis of sugar solubilised to the pretreatment liquor was also conducted. These results are important for explaining the observed effects that pretreatment reaction conditions had on hydrolysis sugar yields, and will therefore be presented first. The first observation made was that reaction conditions had very little effect on cellulose solubilisation. Only a minor amount of the total cellulose (1.0-2.5%) was found to be solubilised over the full range of reaction conditions. This high level of cellulose retention is crucial for an effective pretreatment. Hemicellulose was found to be much more vulnerable to solubilisation during pretreatment then cellulose, a finding that is consistent with literature [41] [58]. Depending on the reaction conditions, 3.5-27.0% of the total hemicellulose was solubilised during pretreatment (Figure 4.4). Understanding what causes hemicellulose solubilisation is important because maximizing total sugar yield will ultimately require a balance between increased substrate hydrolysability and reduced hemicellulose solubilisation. Increases in the reaction variables resulted in increased hemicellulose solubilisation. Caustic load and reaction temperature had the greatest effect on hemicellulose solubilisation. Increased residence time displayed only a minor increase in hemicellulose solubilisation, except for a dramatic drop observed for pretreatment reaction conditions of 150°C, 10% caustic, 15 minute residence time, which is suspected to be an anomaly attributed to experimental error.  64  Percent of Hemicellulose Solubilised  30  25  20  15  10  5  0  20  30  40  Residen  ce Time  50  (min)  60  90  100  120  110  elsiu re (C  140  150  s)  u  erat  p Tem  130  2% Caustic 6% Caustic 10% Caustic  Figure 4.4 The effect of pretreatment reaction variables on hemicellulose solubilisation The effect of the reaction variables on substrate recovery, substrate hydrolysability, and total sugar yield are shown in Figure 4.5 and Figure 4.6. Changes in caustic load (Figure 4.5) displayed the most significant effects on the response variables. An increase in caustic load caused a reduction in substrate recovery, with a subsequent increase in hemicellulose solubilisation. A large increase in hydrolysability was observed with increased caustic load, particularly for cellulosic sugars. When caustic load was increased from 6% to 10%, although hydrolysability increased, total sugar yield plateaued at 71.5% (89.5% total cellulosic sugars, 49.3% total hemicellulosic sugars). This indicates that the losses via hemicellulose solubilisation are cancelling out the gains in substrate hydrolysability. The effect of reaction temperature and residence time on the response variables (Figure 4.6) shows a decrease in substrate recovery with increased temperature. This was also observed for 65  residence time, though the trend was less apparent. Increased reaction temperature resulted in increased substrate hydrolysability, while residence time showed highest hydrolysability at 15 and 60 minutes. This suggests that increasing residence time from 15 to 30 minutes only contributes to hemicellulose solubilisation without improving hydrolysis, but that improved hydrolysis occurs with a 60 minutes residence time. A maximum total sugar yield of 81.7% (92.1% of the total cellulose and 68.2% of the total hemicellulose), occurred under reaction conditions of: 15 minutes, 120°C, and 10% caustic. These conditions did not produce substrate with the highest hydrolysability, but instead represent the optimum of increased hydrolysability and reduced sugar solubilisation.  66  Figure 4.5 The effect of caustic load on substrate recovery, hemicellulose solubilisation, substrate hydrolysability, and total sugar yield (24 h hydrolysis, 20 FPU/g substrate). Fixed pretreatment temperature (150°C) and residence time (60 min). 67  Figure 4.6 The effect of reaction temperature and residence time on pretreatment recovery, substrate hydrolysability, and total sugar yield (24 h hydrolysis, 20 FPU/g substrate). Only the combined cellulosic and hemicellulosic sugar fraction is shown. Caustic load is fixed at 10%. 68  4.3.2 The effect of oxygen delignification reaction conditions on substrate (corn stover) composition Substrate lignin content is hypothesized to have a direct effect on enzymatic hydrolysis efficiency. To confirm this hypothesis, a compositional analysis of the pretreated substrates was conducted. A central composite design was developed for the compositional analysis of pretreated substrates. This allowed for a reduced number of samples, while still producing statistically reliable results. Table 4.5 summarizes the compositional results obtained. Percentages for cellulose, hemicellulose, lignin, and ash were measured directly, while the fifth component “other” was calculated by completing the mass balance. Table 4.5 Substrate composition based on pretreatment reaction conditions t C T (min) (%) (°C) Untreated 15 2 90 15 2 150 15 10 90 15 10 150 60 2 90 60 2 150 60 10 90 60 10 150 15 6 120 60 6 120 30 2 120 30 10 120 30 6 90 30 6 150 30 6 120 30 6 120  % % % % Cellulose Hemicellulose Sugar Lignin 34.1 26.2 60.3 19.6 38.4 29.0 75.7 14.6 40.0 29.8 78.4 11.8 47.6 32.6 90.0 7.2 54.3 27.0 91.2 6.2 34.8 30.8 73.8 19.1 47.4 30.8 87.7 11.0 52.8 32.1 95.3 4.0 57.5 26.4 93.9 3.0 49.0 27.9 86.2 13.5 47.5 26.6 83.0 16.7 43.2 31.0 83.2 15.3 52.8 30.0 92.8 7.0 42.5 31.3 82.9 16.4 48.0 27.1 84.2 15.6 50.5 29.0 89.1 9.2 49.5 25.4 83.97 10.85  % Ash 5.3 2.1 1.8 0.9 2.6 1.5 1.3 0.7 3.1 0.2 0.3 1.5 0.3 0.5 0.3 1.7 0.03  % Other 14.8 15.9 16.6 11.7 9.9 13.8 9.5 10.4 10.0 9.4 8.9 9.1 9.9 9.3 9.0 9.6 14.2  Pretreatment conditions were shown to have a large effect on lignin solubilisation (Figure 4.7). Between 2.6-91.1% of lignin was solubilised over the full range of pretreatment reaction conditions. These results showed that caustic load has the greatest effect on lignin 69  solubilisation, with increased caustic resulting in increased solubilisation. Reaction temperature had the greatest effect when caustic load was 2%. As caustic load increases, the effect of temperature decreased suggesting that caustic load becomes the dominant factor for lignin solubilisation. Reaction residence time had little effect on lignin solubilisation, which suggests the majority of lignin is removed within the first 15 minutes of the reaction; a finding that is supported in literature [57]. Inexplicably, a small reduction in lignin solubilisation was observed with increased reaction residence time for samples with a 6% caustic load. This was considered to be a result of experimental error.  100  Percent of Lignin Solubilised  80  60  40  20  0  20  30  Reside  40  nce Tim  50  e (min  60  )  90  100  110  120  130  rature Tempe  140  (Celsiu  150  s)  2% Caustic 6% Caustic 10% Caustic  Figure 4.7 The effect of pretreatment reaction variables on lignin solubilisation When substrate composition results were combined with hydrolysis results, the relationship between lignin removal and hydrolysability could be observed (Figure 4.8). Increased substrate 70  hydrolysability was observed with increased lignin removal, however, data showed a poor fit with linear (R2=0.58) and polynomial models (R2=0.59). Data points with low lignin removal (25-45%) were the most scattered, suggesting that other factors affecting hydrolysability play a larger role when lignin removal is limited. Changes in cellulose crystallinity and substrate porosity, for example, are known to affect enzymatic hydrolysis [25] [32] [79]. A similar trend was observed by Nichola et al. [57]. 0.9  Substrate Hydrolysability  0.8 0.7 0.6 0.5 0.4 R² = 0.58  0.3 0.2 0.1 0 0  20  40 60 Percentage of lignin removed  80  100  Figure 4.8 Substrate hydrolysability versus percentage of lignin solubilised during oxygen delignification  71  5 Modelling the Effects of Substrate Lignin Content on Enzymatic Hydrolysis Kinetics In order to develop a model that takes into account the effect substrate lignin content has on enzymatic hydrolysis kinetics, three substrates with varied pretreatment conditions were used. The lignin content of each substrate is given (Table 5.1). The hydrolysis curve for each of the samples is shown in Figure 5.1. Table 5.1 Comparison of lignin content and maximum sugar yield following different pretreatment conditions Pretreatment conditions Time (min)  % Caustic  30 60  Untreated 6 10  Substrate Lignin Content (%)  Temp (°C)  19.6 11.4 1.7  150 150  Substrate Carbohydrate Content (%) 26.7 35.4 38.3  Maximum Observed Hydrolysis Yield (%) 44.5 63.7 70.9  90 80  Sugar Yield (%)  70 Substrate Lignin Composition  60 50  1.7% Lignin  40  11.4% Lignin  30  19.6% Lignin  20 10 0 0  10  20  30  40  50  60  Time (h)  Figure 5.1 Sugar yield during enzymatic hydrolysis (20 FPU/g substrate) versus time for substrates with varied lignin content 72  Based on the results in Figure 5.1 the following observations were made: 1. Reducing substrate lignin content increases the maximum hydrolysis sugar yield 2. Reducing substrate lignin content increase the initial rate of reaction 3. Reducing substrate lignin content causes the maximum sugar yield to be reached faster To try to predict these trends, a model which was first put forth by Shen and Agblevor [76] and later modified by Zhang et al. [1] was examined. This model incorporates an enzyme deactivation rate constant to account for the dramatic slowing in reaction rate with time. This concept of enzyme deactivation is backed by theory, as it has been shown that both product inhibition and non-productive binding of lignin can dramatically slow the rate of enzymatic hydrolysis [78] [71]. The initial assumptions required for this model include: 1. That all cellulase, β-glucosidase, and hemicellulase enzymes have a single combined effect of hydrolyzing carbohydrate 2. The substrate has a constant structure 3. That enzyme deactivation is responsible for the reduction in the rate of reaction during hydrolysis Shen and Agblevor [76] provide a detailed derivation of the model; however, it was the modified form of the model put forth by Zhang et al. [1] that was further built on in this research. A brief summary of the model derivation is as follows: The basis for this model stems from Equations 5.1 and 5.2, which describe the enzyme substrate complex kinetics, and the substrate mass balance necessary to obtain Equation 5.3. (5.1)  (5.2)  73  (5.3)  Where: E = Enzyme concentration, S = Substrate (carbohydrate) concentration, P = Product (sugar) concentration, k1, k-1, k2 = Reaction rate constants, So = Initial substrate (carbohydrate) concentration, SE = Substrate-Enzyme complex concentration. Assuming quasi steady state, Equation 5.4 is formed (5.4)  (5.5)  Where Ke is the equilibrium constant defined in Equation 5.5. By assuming the enzyme deactivation is a second order reaction (Equation 5.6), which becomes Equation 5.7 when integrated, the equation derived by Zhang et al. is formed (Equation 5.8) (5.6)  (5.7)  (5.8)  Where kd = Enzyme deactivation rate constant, Eo = Initial enzyme concentration. The units for the variables described are shown in Table 5.2.  74  Table 5.2 Units Variables E, Eo, S, So, P, SE k1, Kd k-1, k2 Ke t  Units g/L L/h g 1/h g/L h  By observing the results presented by Zhang et al. [1] and comparing them to the results observed in this study, an interesting observation was made. The effect that changes in enzyme load had on hydrolysis closely resembled the effect of varied lignin content presented in this study. Zhang et al. [1] showed that a reduction in enzyme loading both reduced hydrolysis reaction rate and the maximum sugar yield, effects also produced by an increase in lignin content. This observation led to the hypothesis that lignin was affecting hydrolysis by effectively eliminating the amount of available enzyme. Figure 5.2 illustrates this hypothesis.  Figure 5.2 Illustration of the proposed effect of lignin on enzymatic hydrolysis  75  To test this hypothesis, a modification that would be able to account for this proposed lignin effect was added to the Zhang et al. [1] model. In the revised model, initial enzyme loading (Eo) was subtracted by the lignin concentration Lo, multiplied by a lignin factor, LF (Equation 5.9). The lignin factor converts the lignin concentration term into units of g enzyme/L. The lignin factor therefore has units of g enzyme/g lignin, representing how much enzyme is made unavailable per gram of lignin. A conversion factor of 1.11 g sugar/g carbohydrate was also added to account for the addition of a water molecule during hydrolysis. (5.9)  Where Lo is the initial fraction of lignin present in the substrate (g/L), and LF is the lignin factor constant (g enzyme/g lignin). Carbohydrate and lignin concentrations were calculated for each sample based on the substrate composition and a 50 mL hydrolysis volume with two grams of substrate loaded. Enzyme loading was 20 FPU/g substrate, which was converted to an enzyme concentration (1.45 g/L) using a cellulase concentration standard curve generated experimentally. The data points produced from the enzymatic hydrolysis of the three substrates with varied pretreatment conditions were fit using Equation 5.9 to determine the models rate constants and lignin factor. This was accomplished using Solver in Excel by minimizing the sum of the standard error between experimental values and model predicted values. Table 5.3 shows the values obtained. These results are assumed to be valid only under the conditions used during experimentation. In solving for the models constants, the following parameters and assumptions were defined: 1. A fixed substrate (corn stover) 2. A fixed enzyme loading (20 FPU/g substrate) 3. An enzymatic hydrolysis reaction time ranging from 0 to 48 hours  76  4. Pretreatment does not affect the enzymatic hydrolysis rate constants (Ke, K2) or the enzyme deactivation rate constant (Kd) 5. (Eo - LoLF) assumed to be positive Changes to these fixed parameters would subsequently result in changes to the values of the equation constants. Table 5.3 Fitted model constants Enzyme Loading Eo Equation Constants Ke kd k2 LF  Defined Values 1.45 (20) Fitted Values 14.7 12.8 26.9 0.178 (98.2)  Units g/L (FPU/g substrate) Units g/L L/h g 1/h g enzyme/g lignin (FPU/g lignin)  Based on the fitted results, a lignin factor of 0.178 g enzyme/g lignin was determined. This demonstrates that under these reaction conditions, each gram of lignin had the capacity to make 0.178 grams of cellulase enzyme unavailable for reaction. Put into terms of enzyme activity, this is equivalent to 98.2 FPU/g of lignin. When the constants are incorporated into the model, Equation 5.9 becomes Equation 5.10. The high level of accuracy produced by this model can be observed in Figure 5.3. Equation 5.11 was used to convert the results of the model from concentration to hydrolysis sugar yield. (5.10)  (5.11)  77  80  Hydrolysis Sugar Yield (%)  70 60 50 40 1.7% Lignin Sample  30  Model Fit 11.4% Lignin Sample  20  Model Fit 10  19.6% Lignin Sample Model Fit  0 0  10  20  30  40  50  60  Time (h)  Figure 5.3 Enzymatic hydrolysis (20 FPU/g substrate) experimental and model results for corn stover with varied lignin content To validate this model, corn stover pretreated under a new set of conditions (30 minutes, 6% caustic, 120°C) was fitted using its unique lignin content (Table 5.4) and the previously determined lignin factor and k values (Table 5.3). The accurate prediction of this new set of data (Figure 5.4) acts to validate this model. Table 5.4 Pretreatment conditions and lignin content of substrate used for model validation Pretreatment Conditions Time % Temp (min) Caustic (°C) 30 6 120  Lignin Content (%) 12.9  78  70  Hydrolysis Sugar Yield (%)  60 50 40 12.9% Lignin  30  Model Fit  20 10 0 0  20  40  60  80  Time (h)  Figure 5.4 Hydrolysis model validation Combined, these findings act to confirm the hypothesis that lignin affects hydrolysis by eliminating availability of cellulase enzyme. This improved model can now accurately predict how changes in substrate lignin composition will affect enzymatic hydrolysis, greatly broadening the potential applications of the model.  79  6 Empirical Model Development for Predicting Substrate Composition and Hydrolysability Following Pretreatment Following the factorial and central composite design experiments exploring the effects of pretreatment reactor conditions on substrate composition and enzymatic hydrolysis, the development of empirical models to predict these results was conducted. 6.1 Empirical model development The data compiled from the factorial and central composite design experiments was used to develop empirical models for substrate hydrolysability, recovery, and composition following oxygen delignification. Models were developed using JMP statistical software to perform linear regression on the data sets. Exponential and interactive effects were included in the regression. The products of the regression were models that could predict compositional and hydrolysis outcomes as a function of the independent oxygen delignification reaction variables – residence time, temperature, and caustic load. The models developed to predict substrate composition following pretreatment are shown in Table 6.1. Due to the nature of experimental data, a perfect mass balance was not always obtained, and so the percent substrate recovery was adjusted such that the sum of cellulose and hemicellulose in the solids and liquor was consistent with the amount entering the system. The parameters that were viewed as statistically significant for inclusion in the empirical equations had a p-value of 0.15 or less. The R2 values represent the degree of fit produced by each model. Equation 6.1 – 6.3 show how the variables are scaled in the empirical equations. Reaction time is specified in minutes, temperature in degrees Celsius, and caustic in percentage (mass NaOH/mass of substrate). (6.1) (6.2) (6.3) 80  Table 6.1 Empirical models to predict substrate composition based on oxygen delignfication conditions  % Recovery  % Cellulose in solid substrate  % Hemicellulose in solid substrate  % Lignin in solid substrate  Significant Parameters (p-value) t=0.0060 C=0.0001 T=0.0001 (C)(T)=0.0231 T2=0.0021 C=0.0008 T=0.0133 T2=0.0716 C=0.1142 T=0.0001 (t)(C)=0.0723 (C)(T)=0.0002 t2=0.0095 C2=0.0019 T2=0.1490 C=0.0001 T=0.0633 (t)(C)=0.1124 (C)(T)=0.1458 C2=0.0024  R2  Empirical Model  0.99  69.74 - 1.35(t) - 8.45(C) - 4.36(T) + 1.17(C)(T) + 2.84(T2)  0.88  48.74 + 6.13(C) + 3.12(T) – 2.41(T2)  0.95  28.86 - 0.36(C) - 1.47(T) - 0.46(t)(C) 1.51(C)(T) - 1.54(t2) + 1.88(C2) + 0.60(T2)  0.89  14.61 + -4.52 (C) - 1.37(T) - 1.26(t)(C) + 1.11 (C)(T) - 4.69(C2)  Empirical models that predict the percentage of cellulose, hemicellulose, and lignin present in the solid fraction following pretreatment were developed. A fourth component termed “other” was the sum of the remaining components (proteins, extractives, and ash). “Other” was calculated by completing the mass balance on the solid fraction as shown by Equation 6-4. The cellulose, hemicellulose, lignin, and “other” solubilised to the liquor were calculated by completing the mass balance between components entering the system and components present in the solid fraction. An example for calculating lignin mass solubilised to the liquor is shown in Equations 6-5 to 6-6. %OS = 100% – (%CS+%HCS+%LS)  (6-4)  81  MLS = MBo*%R*%LS  (6-5)  MLL = (MBo*0.196) - MLS  (6-6)  Where: O = Other, M = Mass, %R = % substrate recovery, C = Cellulose, HC = Hemicellulose, L = Lignin, B = Biomass and the subscripts: S = Solid phase, L = Liquid phase, O = Initial Empirical models to predict the hydrolysability of each sugar type based on oxygen delignification pretreatment conditions were developed (Table 6.2) from the factorial design hydrolysis data. For these models, the hydrolysis time is set at 24 hours, and the cellulase and B-glucosidase concentrations are set at 20 FPU/g substrate and 100 CBU/g substrate respectively. A model to accurately predict mannose (R2=0.46) could not be produced through linear regression as indicated by the low R2 value. Mannose only makes up a small fraction of corn stover (1.6%), and is present in very low concentrations in the hydrolysate. Therefore, its omission will have little effect on sugar yield calculations.  82  Table 6.2 Empirical models to predict enzymatic hydrolysis based on oxygen delignification conditions  Arabinose  Galactose  Glucose  Xylose  Mannose  Significant Parameters (p-value) C=0.0001 (t)(T)=0.1433 (C)(T)=0.0245 T2=0.0001 t2=0.0855 C=0.0001 T=0.0044 (C)(T)=0.0005 T2=0.1467 C2=0.0158 t=0.0518 C=0.0001 T=0.0001 (C)(T)=0.0229 t2=0.1578 t=0.1386 C=0.0001 T=0.0001 C2=0.0124 T2=0.0080 t2=0.0456 C=0.0010 (C)(T)=0.1161 C2=0.1050  R2  Hydrolysability (g sugar / g pretreated substrate)  0.88  0.021 + 0.005(C) – 0.001(t)(T) – 0.001(C)(T) – 0.004(T2) + 0.002(t2)  0.74  0.002 - 0.0002(C) - 0.0002(T)– 0.0002(C)(T) - 0.0001(T2) + 0.0002(C2)  0.90  0.277 + 0.022(t) + 0.124(C) + 0.070(T) + 0.033(C)(T) + 0.032(t2)  0.92  0.46  0.154 + 0.007(t) + 0.057(C) + 0.023(T) + 0.018(t2) – 0.020(C2) – 0.022(T2) 0.007 + 0.002(C) - 0.001(C)(T) + 0.002(C2)  6.2 Model validation To validate the empirical models developed for substrate recovery, composition, and hydrolysability, two new samples were pretreated under oxygen delignification reaction conditions that had not been used in development of the empirical models (Table 6.3). Experimental measurements for recovery, composition, and hydrolysability following pretreatment were then compared to model predicted values and the level of error produced by the models was calculated.  83  Table 6.3 Model validation pretreatment conditions Sample 1 22.5 4 105  Reaction Time (min) Caustic load (%) Temperature (°C)  Sample 2 45 8 135  The results of the model validation (Table 6.4) show that some of the models predicted experimental values better then others. Composition models for percent recovery, cellulose composition, and hemicellulose composition all performed well, predicting to within 5% of the experimental values. The model for lignin, however, could only predict to within 20% of the experimental values. For the hydrolysis models, glucose was predicted to within 10% of the experimental values, and xylose to within 15%. The models for arabinose and galactose could only predict to within 22% of the experimental values. Fortunately, these two sugars combined make up <5% of the total substrate and so this larger error will only have minor effects on the overall predicted results. Table 6.4 Percent error observed for empirical models Sample Pretreatment Composition Analysis % Recovery Cellulose (%) Hemicellulose (%) Lignin (%) Hydrolysis Analysis Arabinose (g/g substrate) Galactose (g/g substrate) Glucose (g/g substrate) Xylose (g/g substrate)  22.5 min, 4% caustic, 105°C Experimental Predicted Value Value % Error 77.7 78.0 -0.4 43.8 43.5 0.7 28.1 29.2 -3.7 18.7 16.2 14.9  45 min, 8% caustic, 135°C Experimental Value 65.7 52.9 27.8 12.6  Predicted Value 63.9 52.8 27.9 10.6  % Error 2.8 0.2 -0.7 18.9  0.014  0.018  -18.5  0.020  0.022  -8.4  0.002  0.002  -22.4  0.001  0.002  -18.8  0.195  0.188  4.2  0.353  0.393  -10.3  0.091  0.107  -14.7  0.171  0.188  -9.0 84  7 Aspen Plus Simulation of Bioethanol Process Using Empirical Models A simulation of bioethanol production was developed using Aspen Plus, which combined the empirical models created using JMP software with the physical property models used in Aspen Plus (Figure 7.1). This was accomplished by incorporating the empirical models into the simulation using “Calculator blocks” written in FORTRAN. This approach allowed for the simulation of downstream operations central to bioethanol production, such as fermentation and distillation, which had not been empirically modelled. As a result, the effect that pretreatment conditions were having on downstream operations could also be observed. This approach enables an economic optimization of pretreatment conditions, as pretreatment and plant conditions can be readily changed within the simulation and the resulting mass and energy balances, and utility requirements efficiently calculated. 7.1 Defining substrate composition Substrate composition (lignin and carbohydrate content) can vary significantly between biomass types and even between strains of corn stover feedstocks. Prior to using this simulation, substrate composition must be measured and the values inputted into the calculator block BIO-IN. This block controls the first unit in the process BIO-COMP, which is an imaginary unit that converts the mass flow rate of biomass feedstock into mass flow rates of its main components (cellulose, hemicellulose, lignin, and other). A corn stover moisture content of 15% was assumed for all simulations, but can be changed by inputting a different water content in the BIOMASS stream. A corn stover flow rate of 2000 tonnes/day (dry weight) was used for all simulations.  85  Figure 7.1 Aspen Plus simulation of bioethanol production using an oxygen delignification pretreatment  86  7.2 Simulating pretreatment The oxygen delignification pretreatment was modelled with the yield reactor O2-DELIG, where the empirical models developed to predict substrate composition following pretreatment were incorporated using the calculator block O2DELIG. The reaction temperature must be specified within the unit, while reaction time must be specified in the calculator block O2DELIG, and caustic load must be specified in the calculator block O2D-NAOH. Solids loading in laboratory experiments was limited to 2% (w/w) due to the equipment available. Industrial oxygen delignification reactors operate with a solids loading of 8-28% (w/w) [42]. For this simulation, a solids loading of 10% (w/w) with equivalent mixing efficiency as the lab-scale reactor was assumed. 7.3 Simulating hydrolysis Hydrolysis was modelled with the yield reactor labelled HYDROLY, where the empirical models developed to predict hydrolysability of cellulose and hemicellulose based on pretreatment conditions were incorporated using a calculator block HYDROL. The pretreatment reaction time must be specified within this calculator block. 7.4 Simulating fermentation Fermentation was modelled using a stoichiometric reactor labelled FERMENT. Assumptions on fermentation efficiencies of pentose and hexose sugars consistent with the NREL Aden et al. [99] study were made (Table 7.1). At lab-scale, dual sugar (pentose and hexose) fermentation efficiencies at these levels have been achieved making this a reasonable assumption [24] [95]. Table 7.1 Stoichiometry and conversion assumed for the fermentation of each sugar Sugar  Stoichiometry  Glucose Galactose Mannose Xylose Arabinose  Glucose  2 Ethanol + 2 CO2 Galactose  2 Ethanol + 2 CO2 Mannose  2 Ethanol + 2 CO2 3 Xylose  5 Ethanol + 5 CO2 3 Arabinose  2 Ethanol + 2 CO2  Fractional conversion 0.95 0.85 0.85 0.85 0.85  87  7.5 Simulating solid separation Solids separation was accomplished using Aspen Plus Sep blocks. The cake moisture content and solid recovery were based on the specifications for the chosen process equipment, which is outlined in a latter section (Section 8.1.1). Based on the selection of a rotary drum vacuum filter for removing pretreatment liquor, and a Pneumapress© pressure filter for filtering distillation bottoms to produce a solid by-product the following specifications were used (Table 7.2). Table 7.2 Separator specifications Unit Sep-1, Sep-2 Sep-3  Process Equipment Rotary drum pressure filter Pneumapress© pressure filter  Solids Recovery (%)  Cake Moisture Content (%)  97  55  99  45  7.6 Simulating distillation The distillation column was designed and sized using Aspen Plus. The design goals are an ethanol purity of 94% and an ethanol recovery of 99.5%. Ethanol purity was set to approach the water-ethanol azeotroph that exists at 95.6% ethanol (w/w). To achieve fuel grade ethanol (99.5% w/w) a molecular sieve, simply modelled with a Sep block, would be used to take ethanol purity past the azeotroph. To achieve these design specifications for a 2000 tonne biomass per day flow rate, a column was designed as outlined in Table 7.3. Near atmospheric distillation pressure was assumed (1.2 atm), as this maximizes separation efficiency and reduces reboiler temperature. Based on this design a column diameter of 5.12 meter was calculated. This was comparable to the 4.37 meter distillation column diameter designed by Aden et al. [99] for the distillation of ethanol produced under the same biomass flow rate.  88  Table 7.3 Distillation column specifications Design Consideration Specification Design Assumptions Feed Temperature (°C) 36 Column Pressure (atm) 1.2 Feed Tray 30 Tray Spacing (m) 0.61 Tray Type Sieve Design Calculations Trays 60 Reflux Ratio 4.0 Maximum Diameter (m) 5.12  89  8 Bioethanol Economic Optimization Empirical models are commonly used to optimize process conditions for maximizing product yield. This approach, however, does not take into account process economics. This can be a problem, as the process conditions producing the greatest product yield do not necessarily represent the economic optimum. In this work, an economic optimization of pretreatment reactor conditions was performed. To conduct an economic optimization of oxygen delignification reactor conditions, the direct and indirect effects that each reaction variable has on process economics had to be accounted for. The direct effects accounted for were: 1. The reactor temperature effect on steam costs 2. The caustic load effect on NaOH costs 3. The effect of reactor residence time on the reactor capital cost 4. The effect of reaction conditions on ethanol yield The indirect effects accounted for were: 1. The effect hydrolysability has on enzyme consumption 2. The effect hydrolysability has on beer concentration, which affects distillation steam requirements 3. The effect pretreatment has on by-product yield To account for all these effects the Aspen Plus simulation, which incorporates the empirical models that predict substrate composition and hydrolysis yields based on pretreatment conditions, was utilized. The simulation calculates material and energy flows, utility requirements, and product yields. By performing a sensitivity analysis on the oxygen delignification reaction variables, results were calculated for a wide range of reactor conditions (Table 8.1). Through this, the effects that pretreatment reaction variables have on ethanol cost could be calculated, and an optimum reaction condition could be determined.  90  Table 8.1 Oxygen delignification sensitivity analysis pretreatment variables Reaction Variable Time Temperature Caustic  Range 15-60 90-150 2-10  Sensitivity Step 15 15 2  Units Min °C %  With the material and energy flows calculated for each set of reactor conditions, the capital and operating costs of the process could be calculated to determine which set of reactor conditions resulted in the cheapest ethanol. 8.1 Capital cost analysis Aspen IPE was utilized for conducting a capital cost analysis of the bioethanol production plant simulated in Aspen Plus. Plant size was based on a biomass flow rate of 2000 tonnes (dry weight) of corn stover per day. First, the process and project type had to be defined so that a contingency factor suitable for this type of project could be determined (Table 8.2). From these assumptions, a contingency factor of 26.4% was calculated with Aspen IPE. Table 8.2 Contingency calculation assumptions Contingency Variables Process Description Process Complexity Process Control Project Location Currency Project Type Calculated Contingency (%)  Specification Redesigned process Typical Digital North America USD Grass root 26.4  8.1.1 Unit mapping For an accurate capital cost estimation, each unit had to be correctly defined within Aspen IPE. In a procedure called “Mapping”, appropriate process equipment was selected from the Aspen IPE database for each of the units used in the Aspen Plus simulation. The material of  91  construction was then defined for each piece of equipment. A summary of the assumptions made for each unit is outlined in this section. Mapping reactors The reactors selected for oxygen delignification, hydrolysis, and fermentation are outlined in Table 8.3. Agitated enclosed tanks were selected for all three reactors; however, a lined vessel was selected for the oxygen delignification reactor due to the increased temperatures and pressures required. Stainless steel 316 was selected as the material of construction for the oxygen delignification reactor because increased corrosion resistance was required to handle the caustic added to the reactor. Carbon steel was the material used for the hydrolysis and fermentation reactors. For each reactor, a 75 kW agitator was selected. Table 8.3 Reactor mapping Aspen Unit  O2DELIG  HYDROLY  FERMENT  Residence Time (min)  Liquid Volume per Unit (L)  Vessel Diameter (m)  Vessel Height (m)  Material of Construction  3  15  1345  9.0  21.2  Stainless Steel 316  2  1440  1345  9.0  21.2  Carbon Steel  10  1440  1335  9.0  21.0  Carbon Steel  Number of Icarus Equipment Mapping Required Agitated Tank Reactor Agitated Tank Mixer Agitated Tank Mixer  Mapping separators A rotary drum vacuum filter was identified as being ideal for separation of liquor from the solids following pretreatment. This continuous unit can simultaneously separate and wash solids before drying (Figure 8.1), performing the tasks of the two separators, Sep-1 and Sep-2, used in the Aspen Simulation. Aqua Merik specifies their rotary drum filters as having a maximum capacity of 272 m3/h and up to 97% solids recovery [105]. With a flow rate of 778 92  m3/h leaving the oxygen delignifier, three rotary drum filters in parallel were required. To resist corrosion from the caustic liquor, stainless steel 316 was used. The moisture content of the cake leaving the rotary drum is dependent on the drums rotational speed, air-flow, and air temperature. Achieving very low moisture content was not required for this step, as wet hydrolysis was to follow, so an assumption of 55% cake moisture was made. A solids conveyor was included in the mapping, to move the solids sludge to the hydrolysis reactor.  Figure 8.1 Rotary drum vacuum filter [106] Following distillation, the solids fraction present in the bottoms stream needs to be dewatered so that the solid by-product can be combusted to produce energy. This requires either a high solids separation method, or separation followed by drying. Under the recommendation put forth by Aden et al. [99] a Pneumapress© pressure filter was selected for effective dewatering and high solids recovery (Table 8.4). Since these filters are capable of producing cakes with 55% solids, further drying is not required prior to by-product combustion. This eliminates the expense that would be associated with drying equipment, justifying the increased capital cost 93  compared to other separation methods. As a Pneumapress© pressure filter is not available in the Aspen IPE database, costing was based on the NREL findings [99]. A solids flow rate roughly half that observed in the NREL study (9704 kg/h compared to 22470 kg/h) meant that only two Pneumapress© pressure filters in parallel would be required. Table 8.4 Pneumapress© pressure filter specifications Cake Moisture Content Solids Recovery  45% 95-99.5%  Table 8.5 Summary of separators mapping Aspen Unit Sep-1, Sep-2 FP  Icarus # of Solids Material of Mapping Equipment Volume Construction Required (tonne/h) Rotary 3 52.9 Stainless drum Steel 316 Filter 2 9.7 Carbon Steel Press  Mapping distillation column and molecular sieve Mapping of the distillation column separates the unit into five parts: A condenser, condenser accumulator, reboiler, reboiler pump, and distillation tower. Carbon steel was the material of construction used for each part. A molecular sieve did not exist in the Aspen IPE database, so cost analysis was based on findings from NREL [99]. 8.1.2 Capital cost results Once Mapping was complete, Aspen IPE was used to conduct a capital cost analysis of the specified units. The equipment cost estimate for each piece of equipment is shown in Table 8.6. For the results shown, the oxygen delignification reactor is sized for a sixty minute residence time.  94  Table 8.6 Equipment cost estimate for bioethanol production at a capacity of 2000 tonnes of biomass (dry weight) per day Component Name Oxygen Delignification Reactor Oxygen Delignification Agitator Hydrolysis Reactor Hydrolysis Agitator Fermentation Reactor Fermentation Agitator Rotary Drum Filter Rotary Drum Conveyor Pneumapress Filter Distillation Condenser Distillation Condenser Accumulator Distillation Reboiler Distillation Reflux Pump Distillation Tower Molecular Sieve Total Cost  Number of Units Required 1 1 2 2 10 10 2 1 2 1 1 1 1 1 1  Total Equipment Cost (Million $) 3.92 0.09 2.20 0.18 10.96 0.86 0.96 0.04 2.40 0.07 0.03 1.58 0.01 4.03 2.70 30.02  Based on the equipment cost, total project capital cost was then calculated with Aspen IPE. Total project capital cost accounts for the costs associated with building a plant, which includes items such as piping, instrumentation, general and administrative overhead, and contingency, as outlined in Table 8.7. The category “other” covers the cost for design, engineering, procurement, material freight and tax charges; and construction indirect costs such as consumables/small tools, insurance, equipment rental, field services, field office construction, and plant start-up [107]. In total, an additional 40.7 millions dollars in capital was added to the equipment cost. This brings the total project capital costs to 70.7 million dollars.  95  Table 8.7 Project capital cost summary (units: $ millions)  Purchased Equipment Equipment Setting Piping Civil Steel Instrumentation Electrical Insulation Paint Other G and A Overheads Contract Fee Contingencies Total Project Cost  Construction Material  Design, Eng, Procurement  Construction Manpower  Construction Indirects  Total Cost  30.02  -  -  -  30.02  -  -  0.20  -  0.20  2.94 1.69 0.75 1.65 1.37 1.20 0.06 4.15  3.89  0.98 1.14 0.15 0.20 0.21 1.24 0.12 -  4.87  3.92 2.83 0.90 1.86 1.58 2.45 0.17 12.91  0.13  0.15  1.61  1.34 0.92 7.94  0.21 0.74  0.30 0.84  0.35 0.97  1.78 10.49  54.10  4.83  5.51  6.33  70.71  The oxygen delignification reactor size and therefore cost changes with the residence time. The effect this has on capital cost was taken into account when selecting optimal reactor conditions. The total plant cost for varied oxygen delignification reaction times is shown in Table 8.8. This value was then annualized using the assumptions of a 10 year plant life and 8% interest rate, where annualized capital cost was observed to range from 10.0-10.6 million dollars/year. Table 8.8 Oxygen delignification reactor cost Residence Reactor Reactor Reactor Reactor Time Capacity Diameter Height Cost 3 (min) (m ) (m) (m) ($ Million) 15 267 5.2 12.6 1.63 30 524 6.6 15.5 2.67 45 774 7.5 17.7 3.58 60 1063 8.2 19.3 4.45 * Based on a 10 year plant life and 8% interest rate  Total Plant Capital Cost ($ Million) 66.8 68.3 69.7 70.8  Annualized Capital Cost ($ Million/year)* 10.0 10.2 10.4 10.6  96  With an annualized total capital cost calculated, the effect that capital cost was having on ethanol cost could be determined by dividing it by annual ethanol production. Ethanol production varies based on pretreatment reactor conditions. Over the range of reactor conditions, capital cost was contributing between 0.05-0.16 $/L to the cost of ethanol. This demonstrates the need for high yielding reactor conditions, to reduce the effect of capital cost. This capital cost was found to be at the mid to lower end when compared to results found in literature (Table 8.9). The wide discrepancy of capital costs observed throughout literature is largely a result of the varied scope of the capital cost analysis conducted in each study. The capital assessment conducted in this study looked only at the major reactors and separators used for bioethanol production. Table 8.9 Capital cost per litre of ethanol produced as observed in literature [108] Source Aden et al. [99] Hinman et al. [109] Douglas [110] US Department of Energy [111] Isaacs [112] Wright et al. [113] von Sivers & Zacchi [108] Nystrom el al. [114]  Plant Capacity (Tonnes Biomass per Day) 2000 1635 939  Capital Cost ($/L) 0.111 0.135 0.154  1635  0.157  1971 1249 282 381  0.202 0.309 0.342 0.710  A more in-depth capital assessment conducted by Aden et al. [99] incorporated the capital cost of a water treatment system, by-product boiler and turbogenerator, feedstock handling facility, and biomass storage facility [99]. To observe how an increased scope of capital cost would affect the selling price of ethanol, capital costs generated by Aden et al. [99] were applied to this study. A boiler/turbogenerator system that is half the size of that designed by Aden et al. [99] was required for the current plant design due to reduced by-product production. A simple 42% added cost factor, as used by Aden et al. [99], approximated the additional capital costs  97  (piping, instrumentation, electrical, etc.) for these systems. Combined, these expenditures were calculated to add an additional 51.1 million dollars of capital investment (Table 8.10). When these additional capital expenses were included into this studies capital analysis, a total capital investment of 117.9-121.9 million dollars was calculated depending on oxygen delignification reactor size. This amounted to a 0.03-0.10 $/L increase to the cost of ethanol depending on the reactor conditions. With this increased scope, capital costs would be 0.080.26 $/L depending on reactor conditions. Table 8.10 Expanding scope of capital cost assessment using results presented by Aden et al. [99] System  Capital Cost ($ Million) 7.5 3.3  Feed Handling Waste Water Treatment Storage 2.0 Boiler/Turbogenerator 38.3 Total 51.1 * Assume an economy of scale factor of 0.6  Scale Factor 1 1 1 0.5  Adjusted Capital Cost ($)* 7.5 3.3  Additional Costs ($) 3.2 1.4  2.0 25.3 38.1  0.8 10.6 16.0  Plant scale is often explored for capital cost savings based on the principle of economy of scale. With bioethanol plants, however, increasing plant size does not necessarily result in cost savings. Due to the close relationship that biomass cost has with transportation distance, as plant size increases, so does the cost of biomass [104]. Data collected from industrial-sized bioethanol plants has demonstrated this point, as findings have shown that ethanol plants with a production capacity smaller then 225 million litres per year (~2500 tonnes biomass per day) are more profitable then larger plants [115]. Furthermore, based on the results found in literature presented by von Sivers and Zacchi [108], a clear relationship between plant capacity and the capital cost per litre of ethanol was not observed (Figure 8.2). Due to the complex relationship between plant size and feedstock cost, combined with the small benefit of increasing plant size observed in literature and in industry, a sensitivity analysis on plant scale was concluded as being outside the scope of this research.  98  Capital Cost ($/L ethanol)  0.4  Literature Values  0.35  Calculated Values  0.3 0.25 0.2 0.15 0.1 0.05 0 0  500  1000  1500  2000  2500  Plant Capacity (tonnes of dry biomass/day)  Figure 8.2 The effect of plant capacity on capital cost as observed in literature [108]. 8.2 Operating cost analysis To calculate the operating cost of ethanol production, costs and values had to be assigned to the materials, utilities, and by-products that are consumed and produced during ethanol production (Table 8.11). An overview of how these costs were determined is outlined in this section. Table 8.11 Raw material, utility, and product prices Price  Units  Source  Corn Stover  33-55  $/tonne  NaOH (Caustic Soda) Cellulase + βglucosidase Oxygen Utilities Steam, 1135 KPA Cooling water Process water By-Products  430-870  $/tonne  0.1-0.5  $/gallon ethanol  157-403  $/h  Huang et al. [104] Aden et al. [99] Chang [116] Aden et al. [99] Geiver et al. [117] Wilcox [118]  0.0105 0.020 0.20  $/kg $/m3 $/m3  Seider et al. [119] Seider et al. [119] Seider et al. [119]  Lignin Carbohydrate  82.0 55.0  $/tonne  Lau et al. [47] McLaughlin et al. [120] Rolls et al. [121]  Raw Materials  99  8.2.1 Biomass The cost of agricultural biomass residues such as corn stover is a function of transportation cost, bale and storage costs, farmer premiums, and fertilize costs (required to make up for the nutrients lost by removing the material). Based on a US biomass availability study conducted by Oak Ridge National Laboratory [122], biomass cost was estimated to be 38-55 dollar per tonne. Aden et al. [99] adopted a biomass cost of $33/tonne under the assumption that advances in collection efficiency would reduce the price of biomass. In this work, the effect of changes in biomass cost on ethanol cost will be assessed. 8.2.2 Enzyme The cost of enzyme assumed by Aden et al. [99] for their bioethanol economic analysis was $0.1 per gallon of ethanol. With an enzyme loading of 12 FPU/g glucan and the enzyme and ethanol flow rates observed in the report, the enzyme cost was back-calculated to be approximately $2.4/million FPU. This enzyme cost estimate seems optimistic, as even with the release of a new line of cellulase enzyme cocktail in 2010, Novozyme could only put forth an enzyme cost estimate of $0.5 per gallon ethanol [117]. With some uncertainly regarding enzyme cost, a sensitivity analysis exploring the effect of varied enzyme cost is necessary. 8.2.3 By-Product The value of lignin and unhydrolysed carbohydrates remaining as a by-product in solid form was calculated based on the amount of electricity that could be generated through their combustion. Heat of combustion values for lignin and carbohydrate (Table 8.12) are an average of the values observed in literature. The value of each component was calculated by assuming an electricity selling price of $0.041/kWh and 28% efficiency for electricity generation. This electricity generation efficiency is consistent with the efficiencies achieved when municipal solid waste is used as the fuel [123]. The calculated value of lignin was consistent with the value ($80/tonne) used in a similar bioethanol by-product assessment [124].  100  Table 8.12 By-product value assumptions By-Product  Heat of Combustion (KJ/g)  Value ($/tonne)  Lignin  25.8  82.0  Carbohydrate  17.1  55.0  Reference Lau et al. [47], McLaughlin et al. [120] McLaughlin et al. [120], Rolls et al. [121]  8.2.4 Oxygen The cost of oxygen needed for oxygen delignification was calculated based on the approach put forward by Wilcox [118]. This approach assumes a commercial arrangement with Praxair to own and operate an oxygen plant on site. Annual oxygen cost is based on annual facility fees (a percentage of the total capital cost), and annual electricity costs which are a function of the required oxygen flow rate. Oxygen flow rate was calculated based on the reactor size (varies with residence time) and pressure (10 atm), assuming ideal gas law. Table 8.13 provides the assumptions required for the remaining calculations. Table 8.13 Oxygen cost assumptions Assumption Electricity Cost Capital cost for a 1000 ton/day oxygen plant Scale factor Power consumptions for a 1000 ft3/day oxygen plant Calculated oxygen cost  Value 0.041  Unit $/kWh  27.0E6  $  0.6 15  kWh  1.3-3.5  (Million$/yr)  8.3 Base case economic analysis With capital and operating costs in place, a base case economic analysis of bioethanol production cost was conducted. The base case scenario was established by making assumptions that were consistent with the bioethanol economic assessment conducted by Aden et al [99]. A summary of these assumptions is listed in Table 8.14. The cost of NaOH selected for this base case was $430/tonne, representing the lower end of the cost range listed by the ICIS Chemical index [116]. 101  Table 8.14 Base case economic analysis assumptions Biomass flow rate Enzyme cost Biomass cost NaOH cost  Assumption 2000 2.4 33 430  Units Tonnes/day $/MM FPU $/tonne $/tonne  Based on these assumptions, the minimal ethanol cost was found to be $0.55/L which occurred under the pretreatment reaction conditions: 60 minute residence time, 135°C, and 8% caustic. The cost or revenue of each component, and the percentage it makes up of the entire cost of ethanol is shown in Figure 8.3.  Distillation Cooling 0.00 ($/L) 0.3%  By-Product Revenue Annualized 0.05 ($/L) Total 8% Capital Cost 0.05 ($/L) Distillation 8% Steam 0.07 ($/L) 12%  Biomass 0.12 ($/L) 18%  Reactor Steam 0.08 ($/L) 13%  Enzyme 0.12 ($/L) 18% Oxygen 0.02 ($/L) 3%  NaOH 0.12 ($/L) 19%  Process Water 0.01 ($/L), 1%  Figure 8.3 Base case ethanol production cost analysis at optimal pretreatment reactor conditions (60 min, 135°C, and 8% caustic) Figure 8.4 demonstrates how each of the pretreatment reaction conditions affected the cost of ethanol. Increased caustic load up to between 6-8% was shown to improve process economics, 102  after which further increase in caustic became detrimental. High reaction temperatures produced lower ethanol costs, indicating that the increase in ethanol yield was worth the increased reactor steam costs. Finally, reaction time demonstrated a trend where mid-point (30 minute) reaction times resulted in the highest ethanol costs. This was consistent with reduced ethanol yields observed for 30 minute reaction time, thought to be a result of increased hemicellulose solubilisation without increased hydrolysability. As reaction time increases further, to 60 minutes, gains in ethanol yield overcome the increases in capital cost (from increasing reactor size), reducing ethanol cost.  1.2  Ethanol Cost ($/L)  1.1  1.0  0.9  0.8  0.7  0.6  0.5 140  Tem  130  per  atu  60  120  50  110  re ( 100 Ce lsiu s)  40 90  30 20  im tion T Reac  )  e (min  2% Caustic 6% Caustic 10% Caustic  Figure 8.4 Ethanol cost versus pretreatment reaction conditions for base case assumptions The optimized ethanol cost ($0.55/L) is well above the $0.28/L reported by Aden et al. [99], however, it does fall into the lower-mid range of ethanol costs reported in the von Sivers and Zacchi [108] literature review where ethanol prices ranging from 0.27-1.51 $/L were observed.  103  To determine if this ethanol price is cost competitive, several ethanol benchmarks can be used for comparison (Table 8.15). Based on information provided by Petro-Canada it was shown that in 2007 on average tax accounted for 32% of the cost of gasoline in Canada [125]. With the price of gasoline averaging roughly $1.20/L currently in BC (2011), the pre-tax cost of gasoline would be $0.82/L. Based on this, after adjusting for the loss in energy content inherent with ethanol, a benchmark of $0.54/L ethanol is observed. Under this benchmark, ethanol cost was shown to be nearly economically viable. With the recent adoption of a carbon tax in BC, which ethanol would be except form, a savings of $0.056/L of gasoline (equivalent to $0.037/L of ethanol) would be had [126]. With the carbon tax included the ethanol benchmark increases to $0.58/L, putting bioethanol below this price benchmark. Another method of developing an ethanol price benchmark is by observing the market price of ethanol. ICIS chemical index provides chemical market prices based on 2007 pricing [116]. From this, ethanol market prices ranging from 0.59-0.79 $/L were observed. Based on this benchmark, bioethanol production was shown to be economically viable. Finally, there is the US Department of Energy ethanol benchmark of $0.28/L [99]. The US Department of Energy suggests that to be competitive, ethanol production costs must be well below market price. Under this benchmark, bioethanol production was shown to not be economically viable. Table 8.15 Ethanol cost benchmarks Benchmark Method Petro Canada Gasoline Conversion [125] Ethanol Market Price [116] US Department of Energy [99] * With carbon taxation  Ethanol Cost Benchmark ($/L) 0.54-0.58* 0.59-0.79 0.28  8.4 Sensitivity analysis A sensitivity analysis on the cost of process materials used for bioethanol production was conducted to explore the effects on: 1. The cost of ethanol 104  2. The oxygen delignification reaction conditions that resulted in the cheapest bioethanol The parameters selected for this sensitivity analysis were biomass cost, NaOH cost, and enzyme cost. The range of prices include in the study are outlined in Table 8.16. Justification for the parameters and ranges in cost selected is discussed. Table 8.16 Sensitivity analysis cost variables Raw material Biomass NaOH Enzyme  Base case cost 33 430 2.4  Sensitivity Analysis Cost 20, 40, 60 430, 650, 870 1, 5, 25  Units $/tonne $/tonne $/Million FPU  The cost of NaOH was selected for this study because of its demonstrated high degree of impact on the cost of ethanol, accounting for 19% of the base case cost. With the market price of NaOH ranging from 430-870 $/tonne [116], the cost of ethanol could see significant increases. Caustic load is also a reaction variable used for optimizing the oxygen delignification pretreatment, therefore, the cost of NaOH may also dictate the optimal reactor conditions. Biomass and enzyme costs were selected based on their impact on the base-case ethanol cost (18% each) and the cost uncertainty that currently exists for these two feed stocks. Agricultural waste residues are not commonly sold in North America and so a market does not currently exist, meaning prices are based on estimates. Biomass cost is also related to plant size and plant location [104]. With these factors producing a large degree of cost uncertainty, a biomass cost of 20-60 $/tonne was selected for the sensitivity analysis. The cost of cellulase enzymes also remains very speculative, as no large-scale lignocellulosic biomass plants utilizing cellulase enzymes exist world-wide. Aden et al. [99] suggest a cost of $2.4/Million FPU is possible; however, a recent enzyme cost estimate by Novozyme [117] of $0.5/gallon of ethanol would put enzyme cost closer to $12/Million FPU. Even this price could be optimistic, so the scenario of higher enzyme costs was included. On the other hand, Novozyme has demonstrated consistent reductions in enzyme cost, and so a potential futuristic enzyme cost of $1/Million FPU was also included in the analysis. 105  A compilation of the minimal ethanol costs and the reactor conditions used to achieve it is shown for each set of enzyme, NaOH, and biomass costs (Table 8.17). The effect that each sensitivity variable had on the cost of ethanol and the ideal reactor conditions is assessed. Table 8.17 The effect of changes in NaOH, biomass, and enzyme cost on optimal pretreatment reactor conditions and minimum ethanol cost Enzyme Cost ($/Million FPU) 1 1 1 1 1 1 1 1 1 5 5 5 5 5 5 5 5 5 25 25 25 25 25 25 25 25 25  NaOH Cost ($/tonne) 430 430 430 650 650 650 870 870 870 430 430 430 650 650 650 870 870 870 430 430 430 650 650 650 870 870 870  Biomass Cost ($/tonne) 20 40 60 20 40 60 20 40 60 20 40 60 20 40 60 20 40 60 20 40 60 20 40 60 20 40 60  Minimal Cost of Ethanol ($/L) 0.44 0.51 0.58 0.49 0.57 0.64 0.55 0.62 0.70 0.63 0.70 0.76 0.69 0.76 0.83 0.75 0.82 0.89 1.46 1.53 1.60 1.53 1.60 1.67 1.61 1.68 1.74  Optimal Reactor Conditions Time (min) 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60  Temp (°C) 135 150 150 135 150 150 150 135 135 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150  Caustic (%) 8 8 8 6 8 8 6 6 6 8 10 10 8 8 8 8 8 8 10 10 10 10 10 10 10 10 10  106  To observe the role that each cost variable was having, the ethanol cost for a single pretreatment condition was tracked for each combination of sensitivity cost variables (Figure 8.5). From this, enzyme cost was observed to have the greatest effect on ethanol cost. The changes in cost of biomass and NaOH were found to have similar effects on ethanol cost, increasing cost by roughly ten cents over their full ranges.  1.0  0.9  Ethanol Cost ($/L)  0.8  0.7  0.6  0.5  0.4 500 550 600 650 700 750 800 850  NaOH Cost  ($/Tonne)  55 60 45 50 40 nne) 30 35 ($/To t s 20 25 o C  B io m  ass  Enzyme Cost = $1/Million FPU Enzyme Cost = $5/Million FPU  Figure 8.5 Ethanol cost versus feedstock costs for a single pretreatment reactor condition (60 minutes, 135°C, 8% Caustic) A low enzyme cost has been shown to be crucial for maintaining economically viable ethanol costs. Table 8.18 demonstrates how, even at enzyme costs of $5/Million FPU, enzyme cost begins to dominate the cost of ethanol accounting for 35% of the total cost. For each dollar per million FPU raise in enzyme cost, the cost of ethanol rose by approximately $0.04/L. This 107  indicates that based on the current Novozyme cost estimate of $0.5/gallon of ethanol (~$12/million FPU) [117], bioethanol production would not be economically viable. It is crucial for enzyme cost to approach the cost estimate of $0.1/gallon of ethanol ($2.4/Million FPU) put forth by Aden et al. [99]. Table 8.18 The effect of enzyme cost on ethanol cost (NaOH cost = $430/tonne, biomass cost = $20/tonne) for fixed pretreatment conditions (60 minute, 135°C, and 6% caustic) Enzyme Cost ($/Million FPU) 1 5 25  Optimal Ethanol Cost ($/L) 0.44 0.63 1.57  Enzyme Percentage of Total Ethanol Cost (%) 10 35 73  With respect to optimal pretreatment reactor conditions, an increase in the cost of enzyme was shown to favour high caustic load (Figure 8.6). The same was observed for reaction temperature. This is because at these conditions, low levels of lignin are observed and substrates have a high hydrolysability. This reduces the amount of enzyme required, as the presence of lignin has been shown to eliminate enzyme availability. 3.000  Cost of Ethanol ($/L)  2.500 2.000  2% 4%  1.500  6%  1.000  8% 10%  0.500 0.000 0  5  10 15 20 Enzyme Cost ($/Million FPU)  25  30  Figure 8.6 Cost of ethanol versus cost of enzyme for a range of caustic loads. Fixed pretreatment reaction time (60 min) and temperature (150°C). Fixed NaOH ($430/tonne) and biomass cost ($20/tonne). 108  An increase in NaOH cost by $220/tonne was shown to increase ethanol cost by between 0.050.07 $/L. As expected, increases in NaOH cost beings to favour a reduction in pretreatment caustic load (Figure 8.7). This demonstrates that a reduction in ethanol yield can be justified by reduced pretreatment costs. 0.650  Cost of Ethanol ($/L)  0.600 2%  0.550  4% 6%  0.500  8% 10%  0.450  0.400 300  400  500  600  700  800  900  NaOH($/tonne)  Figure 8.7 Cost of ethanol versus cost of NaOH for a range of caustic loads. Fixed pretreatment reaction time (60 min) and temperature (150°C). Fixed enzyme cost ($1/Million FPU) and biomass cost ($20/tonne). Increasing biomass cost by $20/tonne was shown to increase the cost of ethanol by $0.07/L. The increased biomass cost had little effect on dictating optimal reactor conditions (Figure 8.8).  109  0.800  0.650  0.700  Temp (°C)  0.650  90 105  0.600  120  0.550  135  0.500  150  0.600 Cost of Ethanol ($/L)  Cost of Ethanol ($/L)  0.750  0.550  15 min 30 min  0.500  45 min 60 min  0.450  0.450 0.400  0.400 0  20 40 60 Biomass Cost ($/tonne)  80  0  20 40 60 Biomass Cost ($/tonne)  80  Figure 8.8 Cost of ethanol versus cost of biomass for a range of reaction temperature and time. Fixed pretreatment caustic load (10%). Fixed NaOH cost ($430/tonne) and enzyme cost ($1/Million FPU). This sensitivity analysis demonstrates the current vulnerability that exists for bioethanol economic viability. Without ideal costs of enzyme, NaOH, and biomass, ethanol production quickly became unviable. Enzyme cost was found to be particularly important in this respect, as achieving an enzyme cost approaching $0.1/gallon (2.4 FPU/million FPU) was crucial for process economics. An increased cost certainty for all the reagents is necessary for reducing this economic vulnerability. Increased cost certainty would also allow for optimization of pretreatment reactor conditions, as changes in the cost of enzyme and NaOH were shown to dictate the economic optimal reactor conditions.  110  9 Conclusions An increase in substrate hydrolysability was observed when oxygen delignification reaction variables time (15-60 min), temperature (90-150°C), and caustic load (2-10%) were increased. Caustic load was found to be the most significant variable for affecting hydrolysability. The highest substrate hydrolysability (0.83 g sugar / g substrate) occurred with the most severe pretreatment reaction conditions (60 minutes, 150°C, 10% caustic). The highest total sugar yield of 81.7% (92.1% of the total cellulose and 68.2% of the total hemicellulose), however, occurred under a different set of pretreatment conditions (15 minutes, 120°C, 10% Caustic). This indicates that under conditions that are too severe, sugar loss (to liquor) begins to have a larger detrimental effect on sugar yield then the gains resulting from increased hydrolysability. Analysis of sugar content present in pretreatment liquor, and a compositional analysis on pretreated substrates revealed that only small amounts of cellulose (1.0-2.5%) was being solubilised to liquor over the full range of pretreatment conditions. Pretreatment conditions were shown to have a greater effect on hemicellulose and lignin solubilisation, with increased reaction severity resulting in increased solubilisation. Between 3.5-27.0% of hemicellulose and 2.6-91.1% of lignin was solubilised over the full range of pretreatment reaction conditions. Both caustic and temperature had a large effect on hemicellulose solubilisation. Caustic load had the greatest effect on lignin solubilisation. Increased substrate hydrolysability was observed for reduced substrate lignin content, however, data showed a poor fit with linear (R2=0.58) and polynomial models (R2=0.59). This indicates that other factors are also affecting substrate hydrolysability. Data points with low lignin removal (25-45%) were the most scattered, suggesting that other factors affecting hydrolysability play a larger role when lignin removal is limited. A fully mechanistic model to predict enzymatic hydrolysis sugar concentration versus time, which takes into account changes in substrate lignin content, was developed. A modification to the Zhang et al. [1] model was made based on a proposed hypothesis that lignin was effectively reducing the availability of enzyme. To account for this effect the initial concentration of lignin,  111  multiplied by a lignin factor, was subtracted from the initial enzyme concentration. The final model is given by equation 5.9.  (5.9)  Where: P = Product (sugar) concentration So = Initial substrate (carbohydrate) concentration Lo = Initial lignin concentration LF = Lignin factor Eo = Initial enzyme concentration kd = Enzyme deactivation constant Ke = Equilibrium constant k2 = Reaction rate constant For corn stover, the rate constants and the lignin factor were determined based on the model fit with experimental data (Table 9.1). From this, a lignin factor of 98.2 FPU/g lignin (0.178 g enzyme/g lignin) was determined. This demonstrates that under the experimental conditions, each gram of lignin had the capacity to make 98.2 FPU (0.178 g) of cellulase enzyme unavailable for reaction. The model was validated by demonstrating a high level of fit with a pretreated sample that was not used in the model development. Based on these findings, the proposed hypothesis was accepted. Table 9.1 Fitted model constants Equation Constants Ke kd k2 LF  Fitted Values 14.7 12.8 26.9 0.178 (98.2)  Units g/L L/h g 1/h g enzyme/g lignin (FPU/g lignin)  112  Empirical models to predict substrate composition and hydrolysability following pretreatment were developed based on experimental data. R2 values ranged from 0.74-0.99 for the models produced. The model for predicting mannose production was rejected due to its low R2 value (0.46). These models were then validated using two pretreatment samples that were not used for model development. The error observed for each model was generally found to be <15%, however, the models predicting lignin, arabinose, and galactose generated somewhat larger errors (18.9%, 18.5%, and 22.4% respectively). These empirical models were incorporated into an Aspen Plus simulation of the complete bioethanol production process. Based on the base case simulation in Aspen Plus, the capital cost of the bioethanol plant was calculated to be 70.7 million dollars. This amounts to 0.051-0.159 $/L of ethanol produced depending on the pretreatment reactor conditions. Increasing the scope of the capital cost analysis to include a water treatment facility, boiler, and feedstock processing/storage facility resulted in a total capital cost of 120.9 million dollars. This increased ethanol cost by 0.0310.098 $/L. A base case economic optimisation of bioethanol production determined an optimum ethanol cost of $0.55/L, which occurred under pretreatment conditions of 60 minutes residence time, 135°C, and 8% caustic. This ethanol cost can be considered as economically viable when compared to the price benchmarks of untaxed gasoline with adjustment for reduced energy content ($0.58/L), or for the market price of ethanol (0.59-0.79 $/L). When compared to the ethanol price benchmark suggested by the US Department of Energy ($0.28/L), which is well under ethanol market price, production would be considered unviable. A sensitivity analysis of biomass, NaOH, and enzyme cost demonstrated that the cost of ethanol was vulnerable to cost variability in these components. Ethanol costs ranged from 0.44-1.74 $/L, most of which were above the economic viability benchmarks. This sensitivity analysis also demonstrated that changes in the sensitivity costs affected optimal pretreatment reactor conditions.  113  The potential for utilising Aspen Plus and Aspen IPE to aid in the economic analysis and economic optimisation of bioethanol production has been successfully established. The incorporation of experimental models into the process simulation was also shown to be a useful tool for enhancing simulation results. Ultimately, this work has demonstrated that there is a future for bioethanol produced from lignocellulosic residues rather than food. Oxygen delignification has also been verified as an effective pretreatment for agricultural residues. The potential for economic viability has been demonstrated under ideal conditions. With future oil costs speculated to increase, the cost of cellulase enzymes decreasing, and further research being conducted to improve process efficiency, cost competitive lignocellulosic bioethanol could be a reality in the near future.  114  10 Future Work Bioethanol production using an oxygen delignification pretreatment was shown to be approaching economic viability; however, several areas have been identified which have the potential to improve the process economics. Oxygen delignification improves hydrolysis by solubilising a substantial portion (up to 90%) of the lignin fraction present in the substrate. This subsequently reduces the amount of solid lignin by-product produced, which has value as a fuel source. Additional research into a cost effective way of recovering this lignin needs to be explored. One appealing option is to use carbon dioxide produced during fermentation to lower the pH of the liquor and precipitate the lignin. This approach has been met with poor results thus far, but increased research is required as the economics could be greatly improved through increased by-product yields [42]. Another approach would be to find alternative uses for the lignin by-product. This approach has worked well for the sulfite pulping process, which produces lignosulfonates that are now being used for acid rock drainage water treatment by NORAM Engineering [127]. Oxygen delignification is also responsible for the solubilisation of some of the carbohydrate present in the substrate. Up to 3% of the total cellulose and 27% of the hemicellulose was shown to be solubilised depending on pretreatment conditions. Recovery of this sugar fraction would improve ethanol yields and should be explored. One recent study has demonstrated the potential for sugar recovery from liquor by solvent extraction [65]. Modifying the liquor to improve its fermentability is another potential approach. Enzyme cost was shown to account for between 10-73% of the total cost of ethanol, depending on the price of enzyme. Approaches for reducing enzyme consumption should be explored, as this could dramatically reduce ethanol cost. Exploring the effects of enzyme load on ethanol cost is necessary, as enzyme load was fixed at 20 FPU/g substrate in this study. Enzyme recycling is another area requiring consideration for improving process economics; as enzymes, much like catalysts, have the potential to be used repeatedly. Reduced substrate lignin content has been shown to improve enzyme recycling, suggesting oxygen delignification would be an ideal pretreatment for employing enzyme recycling [128]. 115  Finally, the possibility of using Aspen Plus to conduct the entire economic optimization of pretreatment reactor conditions should be explored. For this project, Aspen Plus was used to generate mass, energy, and utility balances over a range of pretreatment conditions. This data was then exported to Excel where an economic analysis was conducted. Economic optimization could be performed within Aspen Plus utilizing the optimization tool, and as a result a complete set of optimum process conditions could be determined. Improved optimisation of reactor conditions would reduce ethanol costs.  116  Works Cited [1] Zhang, Y., Hu, J.-L., Xu, H.-J., Yuan, Z.-H., & Guo, Y. (2010). Cellulase Deactivation Based Kinetic Modeling of Enzymatic Hydrolysis of Steam-Exploded Wheat Straw. Bioresource Technology, 8261-8266. [2] Environment Canada. (2004). National Inventory Report. Retrieved February 5, 2009, from Greenhouse Gas Sources and Sinks in Canada: http://www.ec.gc.ca/pdb/ghg/inventory_report/2004_report/ta8_2_e.cfm [3] Gomez, L. D., Steele-King, C. G., & McQueen-Mason, S. J. (2008). Sustainable Liquid Biofuels from Biomass: the Writing's on the Walls. The New Phytologist, 473-485. [4] Champagne, P. (2008). Bioethanol from Agricultural Waste Residues. Environmental Progress, 51-57. [5] Lynd, L. R., Cushman, J. H., Nichols, R. J., & Wyman, C. E. (1991). Fuel Ethanol from Cellulosic Biomass. Science, 1318-1323. [6] Wyman, C. E. (1994). Ethanol from Lignocellulosic Biomass: Technology, Economics, and Opportunities. Bioresource Technology, 3-16. [7] Balat, M., & Balat, H. (2009). Recent Trends in Global Production and Utilization of BioEthanol Fuel. Applied Energy, 2273-2282. [8] Commodities Now. (2010). Global Ethanol Production to Reach 85.9 bn Litres in 2010. Commodities Now. [9] Kerr, R. A. (1998). The Next Oil Crisis Looms Large - and Perhaps Close. Science, 1128-1131. [10] Johnson, T., Johnson, B., Scott-Kerr, C., & Kiviaho, J. (2010). Bioethanol - Status Report on Bioethanol Production from Wood and Other Lignocellulosic Feedstocks. Beca. [11] Sims, R. E., Hastings, A., Schlamadinger, B., Taylor, G., & Smith, P. (2006). Energy Crops: Current Status and Future Prospects. Global Change Biology, 2054-2076. 117  [12] North American Crude Oil Consumption by Year. (2009). Retrieved April 26, 2011, from Index Mundi: http://www.indexmundi.com/energy.aspx?region=na&product=oil&graph=consumption [13] Petroleum and Other Liquids. (2011). Retrieved March 10, 2011, from U.S. Energy Information Administration: http://tonto.eia.doe.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=WTOTWORLD&f=W [14] Wallace, R., Ibsen, K., McAloon, A., & Yee, W. (2005). Feasibility Study for Co-Locating and Integrating Ethanol Production Plants from Corn Starch and Lignocellulosic Feedstocks. National Renewable Energy Laboratory. [15] Dien, B. S., Bothast, R. J., Nichols, N. N., & Cotta, M. A. (2002). The U.S. Corn Ethanol Industry: An Overview of Current Technology and Future Prospects. International Sugar Journal, 204-211. [16] Odling-Smee, L. (2007). Biofuels Bandwagon Hits a Rut. Nature, 483. [17] DiPardo, J. Outlook for Biomass Ethanol Production and Demand. Energy Information Administration. [18] The Biofuel Controvery. (2008). De Economist, 507-527. [19] Pimentel, D. (2003). Ethanol Fuels: Energy Balance, Economics, and Environmental Impacts are Negative. Natural Resources Research, 127-134. [20] Hill, J., Nelson, E., Tilman, D., Polasky, S., & Tiffany, D. (2006). Environmental, Economic, and Energetic Costs and Benefits of Biodiesel and Ethanol Biofuels. PNAS , 206-210. [21] Lin, Y., & Tanaka, S. (2006). Ethanol Fermentation from Biomass Resources: Current State and Prospects. Applied Microbiology and Biotechnology, 627-642.  118  [22] Brehmer, B., & Sanders, J. (2009). Implementing an Energetic Life Cycle Analysis to Prove the Benefits of Lignocellulosic Feedstocks with Protein Separation for the Chemical Industry from the Existing Bioethanol Industry. Biotechnology and Bioengineering, 767-777. [23] Schmer, M. R., Vogel, K. P., Mitchell, R. B., & Perrin, R. K. (2008). Net Energy of Cellulosic Ethanol from switchgrass. PNAS, 464-469. [24] Zaldivar, J., Nielsen, J., & Olsson, L. (2001). Fuel Ethanol Production from Lignocellulose: a Challenge for Metabolic Engineering and Process Integration. Applied Microbiology and Biotechnology, 17-34. [25] Yoshida, M., Liu, Y., Uchida, S., Kawarada, K., Ukagami, Y., Ichinose, H., et al. (2008). Effects of Cellulose Crystallinity, Hemicellulose, and Lignin on the Enzymatic Hydrolysis of Miscanthus sinensis to Monosaccharides. Bioscience Biotechnology Biochemistry, 805-810. [26] Biomass Pyrolysis Chemistry. (n.d.). Retrieved March 21, 2011, from Princeton University Blog Service: http://blogs.princeton.edu/chm333/f2006/biomass/2007/01/pyrolysis_chemistry.html [27] Leung, C. (2010). Potential for Ethanol from Urban Cellulosic Wastes. University of British Columbia. [28] Alvira, P., Tomás-Pejó, E., Ballesteros, M., & Negro, M. J. (2010). Pretreatment Technologies for an Efficient Bioethanol Production Process Based on Enzymatic Hydrolysis: A Review. Bioresource Technology, 4851-4861. [29] Arato, C., Pye, E. K., & Gjennestad, G. (2005). The Lignol Approach to Biorefining of Woody Biomass to Produce Ethanol and Chemicals. Applied Biochemistry and Biotechnology, 871-882. [30] Halil, C., Kasthurirangan, G., & Sunghwan, K. (2010). Soil Stabilization with Bioenergy Coproduct. Transportation Research Board.  119  [31] Ghosh, P., & Ghose, T. K. (2003). Bioethanol in India: Recent Past and Emerging Future. Advance Biochemical Engineering / Biotechnology, 1-27. [32] Sun, Y., & Cheng, J. (2002). Hydrolysis of Lignocellulosic Materials for Ethanol Production: a Review. Bioresource Technology, 1-11. [33] Pan, X., Gilkes, N., Kadla, J., Pye, K., Saka, S., Gregg, D., et al. (2006). Bioconversion of Hybrid Poplar to Ethanol and Co-Products Using an Organosolv Fractionation Process: Optimization of Process Yields. InterScience. [34] Himmel, M. E., Adney, W. S., Baker, J. O., Elander, R., McMillan, J. D., Nieves, R. A., et al. (1997). Advanced Bioethanol Production Technologies: A Perspective. American Chemical Society. [35] Draude, K. M., Kurniawan, C. B., & Duff, S. J. (2001). Effect of Oxygen Delignification on the Rate and Extent of Enzymatic Hydrolysis of Lignocellulosic Material. Bioresource Technology, 113-120. [36] Lai, R. (2005). World Crop Residues Production and Implications of its Use as a Biofuel. Environment International, 575-584. [37] Kim, S., & Dale, B. (2004). Global potential bioethanol production from wasted crops and crop residues. Biomass and Bioenergy, 361-375. [38] Mabee, W. E., Fraser, E. D., Mcfarlane, P. N., & Saddler, J. N. (2006). Canadian Biomass Reserves for Biorefining. Applied Biochemistry and Biotechnology, 129-132. [39] Saha, B. C., Iten, L. B., Cotta, M. A., & Wu, Y. V. (2005). Dilute Acid Pretreatment, Enzymatic Saccharification and Fermentation of Wheat Straw to Ethanol. Process Biochemisty, 3693-3700. [40] Sassner, P., Gable, M., & Zacchi, G. (2008). Techno-economic Evaluation of Bioethanol Production from Three Different Lignocellulosic Materials. Biomass and Bioenergy, 422-430.  120  [41] Lissens, G., Klinke, H., Verstraete, W., Ahring, B., & Thomsen, A. B. (2004). Wet Oxidation Pre-treatment of Woody Yard Waste: Parameter Optimization and Enzymatic Digestibility for Ethanol Production. Journal of Chemical Technology and Biotechnology, 889-895. [42] Ruffell, J. (2006). Pretreatment and Hydrolysis of Recovered Fibre for Ethanol Production. UBC Masters Thesis. [43] Hector, H., Hughes, S., & Liang-Li, X. (2008). Developing yeast strains for biomass-toethanol production. Biomass Magazine. [44] Galbe, M., & Zacchi, G. (2007). Pretreatment of Lignocellulosic Materials. Advance Biochemical Engineering / Biotechnology, 41-65. [45] Wyman, C. E., Dale, B. E., Elander, R. T., Holtzapple, M., Ladisch, M. R., & Lee, Y. Y. (2005). Comparative sugar recovery data from laboratory scale application of leading pretreatment technologies to corn stover. Bioresource Technology, 2026-2032. [46] Kaar, W. E., & Holtzapple, M. T. (2000). Using Lime Pretreatment to Facilitate the Enzymic Hydrolysis of Corn Stover. Biomass and Bioenergy, 189-199. [47] Lau, M. W., Gunawan, C., & Dale, B. E. (2009). The Impacts of Pretreatment on the Fermentability of Pretreated Lignocellulosic Biomass: a Comparative Evaluation Between Ammonia Fiber Expansion and Dilute Acid Pretreatment. Biotechnology for Biofuels. [48] Silversteina, R. A., Chena, Y., Sharma-Shivappa, R. R., Boyettea, M. D., & Osborne, J. (2007). A comparison of chemical pretreatment methods for improving saccharification of cotton stalks. Bioresource Technology, 3000-3011. [49] Bals, B., Roger, C., Jin, M., Balan, V., & Dale, B. (2010). Evaluation of Ammonia Fibre Expansion (AFEX) Pretreatment for Enzymatic Hydrolysis of Switchgrass Harvested in Different Seasons and Locations. Biotechnology for biofuels.  121  [50] Wang, Z., & Cheng, J. J. (2011). Lime Pretreatment of Coastal Bermudagrass for Bioethanol Production. Energy and Fuels, 1830-1836. [51] Kim, S., & Holtzapple, M. T. (2005). Lime Pretreatment and Enzymatic Hydrolysis of Corn Stover. Bioresource Technology, 1994-2006. [52] Teymouri, F., Laureano-Perez, L., Alizadeh, H., & Dale, B. E. (2004). Ammonia Fiber Explosion Treatment of Corn Stover. Applied Biochemistry and Biotechnology, 113-116. [53] Lloyd, T. A., & Wyman, C. E. (2005). Combined Sugar Yields for Dilute Sulfuric Acid Pretreatment of Corn Stover Followed by Enzymatic Hydrolysis of the Remaining Solids. Bioresource Technology, 1967-1977. [54] Lucia, L. A., & Smereck, R. S. (2001). Effect of Lignin Content and Magnesium-toManganese Ratio on the Selectivity of Oxygen Delignification in Softwood Kraft Pulp. Pure Applied Chemistry, 2059-2065. [55] Yang, R., Lucia, L., Ragauskas, A. J., & Jameel, H. (2003). Oxygen Delignification Chemistry and Its Impact on Pulp Fibers. Journal of Wood Chemistry and Technology, 13-29. [56] Bjerre, A. B., Olesen, A. B., Fernqvist, T., Ploger, A., & Schmidt, A. S. (2000). Pretreatment of Wheat Straw Using Combined Wet Oxidation and Alkaline Hydrolysis Resulting in Convertible Cellulose and Hemicellulose . Biotechnology and Bioengineering, 568-577. [57] Charles, N., Mansfield, S. D., Mirochnik, O., & Duff, S. J. (2003). Effect of Oxygen Delignification Operating Parameters on Downstream Enzymatic Hydrolysis of Softwood Substrates. Biotechnology Progress, 1606-1611. [58] Varga, E., Schmidt, A. S., Reczey, K., & Thomsen, A. B. (2003). Pretreatment of Corn Stover Using Wet Oxidation to Enhance Enzymatic Digestibility. Applied Biochemistry and Biotechnology, 37-50.  122  [59] Gierer, J., Reitberger, T., Yang, E., & Yoon, B.-H. (2001). Formation and Involvement of Radicals in Oxygen Delignification Studied by the Autoxidation of Lignin and Carbohydrate Model Compounds. Journal of Wood Chemistry and Technology, 313-341. [60] Klinke, H. B., Ahring, B. K., Schmidt, A. S., & Thomsen, A. B. (2002). Straw, Characterization of Degradation Products from Alkaline Wet Oxidation of Wheat. Bioresource Technology, 15-26. [61] De Souza, I. J., Bouchard, J., Methot, M., Berry, R., & Argyropoulos, D. S. (2002). Carbohydrates in Oxygen Delignification. Part I: Changes in Celllulose Crystallinity. Journal of Pulp and Paper Science, 167-170. [62] McGinnis, G. D., Wilson, W. W., & Mullen, C. E. (1983). Biomass Pretreatment with Water and High-Pressure Oxygen. The Wet-Oxidation Process. Industrial and Engineering Chemistry Product Research and Development, 352-357. [63] Tench, L., & Harper, S. (1987). Oxygen Bleaching Practives and Benefits. Tappi Journal. [64] Loutfi, H. B. (1991). Lignin recovery from kraft black liquor: preliminary process design. Tappi Journal. [65] Brennan, T. C., Datta, S., Blanch, H. W., Simmons, B. A., & Holmes, B. M. (2010). Recovery of Sugars from Ionic Liquid Biomass Liquor by Solvent Extraction. Bioenergy Resources, 123-133. [66] Taherzadeh, M. J., & Karimi, K. (2007). Acid-Based Hydrolysis Processes for Ethanol from Lignocellulosic Materials: a Review. BioResources, 472-499. [67] Larsson, S., Palmqvist, E., Hahn-Hagerdal, B., Tengborg, C., Stenberg, K., Zacchi, G., et al. (1999). The Generation of Fermentation Inhibitors During Dilute Acid Hydrolysis of Softwood. Enzyme and Microbial Technology. [68] Novozymes. (2009). Cellulosic Cost Savings with Novozymes Expertise. [69] Smith, R. (2010). Novozymes Reports Enzyme Breakthrough for Biofuel Production. Franklington: WRALtechwire. 123  [70] Lynd, L. R., Weimer, P. J., van Zyl, W. H., & Pretorius, I. S. (2002). Microbial Cellulose Utilization: Fundamentals and Biotechnology. Microbiology and Molecular Biology Review , 506-577. [71] Kurabi, A. (2004). The Inhibitory Role of Lignin in the Enzymatic Hydrolysis of Softwoods. UBC Masters Thesis. [72] Bansal, P., Hall, M., Realff, M. J., Lee, J. H., & Bommarius, A. S. (2009). Modelling Cellulase Kinetics on Lignocellulosic Substrates. Biotechnology Advances, 833-848. [73] Rabinovich, M. L., Melnick, M. S., & Bolobova, A. V. (2002). The Structure and Mechanism of Action of Cellulolytic Enzymes. Biochemistry (Moscow), 67 (8), 850-871. [74] Wood, W. A., & Kellogg, S. T. (1988). Methods for measuring cellulase activities. Methods in enzymology, 87-92. [75] Bailey, M. J., Biely, P., & Poutanen, K. (1992). Interlaboratory Testing of Methods for Assay of Xylanase Activity. Journal of Biotechnology, 257-270. [76] Shen, J., & Agblevor, F. A. (2008). Kinetics of Enzymatic Hydrolysis of Steam-Exploded Cotton Gin Waste. Chemical Engineering Communications, 1107-1121. [77] Bommariusa, A. S., Katonaa, A., Chebena, S. E., Patela, A. S., Ragauskasc, A. J., Knudsonc, K., et al. (2008). Cellulase kinetics as a function of cellulose pretreatment. Metabolic Engineering, 370-381. [78] Xiao, Z., Zhang, X., Gregg, D. J., & Saddler, J. N. (2004). Effects of sugar inhibition on cellulases and β-glucosidase during enzymatic hydrolysis of softwood substrates. Applied Biochemistry and Biotechnology, 113-116. [79] Sinitsyn, A. P., Gusakov, A. V., & Vlasenko, E. Y. (1991). Effect of Structural and PhysicoChemical Features of Cellulosic Substrates on the Efficiency of Enzymatic Hydrolysis. Applied Biochemistry and Biotechnology, 43-59. 124  [80] Ruffell, J., Levie, B., Helle, S., & Duff, S. (2010). Pretreatment and enzymatic hydrolysis of recovered fibre for ethanol production. Bioresource Technology. [81] Mooney, C. A., Mansfield, S. D., Touhy, M. G., & Saddler, J. N. (1998). The Effect of Initial Pore Volume and Lignin Content on the Enzymatic Hydrolysis of Softwoods. Bioresource Technology, 113-119. [82] Palonen, H., Tjerneld, F., Zacchi, G., & Tenkanen, M. (2004). Adsorption of Trichoderma reesei CBH I and EG II and their catalytic domains on steam pretreated softwood and isolated lignin. Journal of Biotechnology, 65-72. [83] Ljunggren, M. (2005). Kinetic Analysis and Modeling of Enzymatic Hydrolysis and SSF. Lund Institute of Technology, Department of Chemical Engineering. [84] Schell, D. J., Ruth, M. F., & Tucker, M. P. (1999). Modeling the Enzymatic Hydrolysis of Dilute-Acid Pretreated Douglas Fir. Applied Biochemistry and Biotechnology. [85] Luo, J., Xia, L., Lin, J., & Cen, P. (1997). Kinetics of Simultaneous Saccharification and Lactic Acid Fermentation Processes. Biotechnology Progress. [86] Kadam, K. L., Rydholm, E. C., & McMillan, J. D. (2004). Development and Validation of a Kinetic Model for Enzymatic Saccharification of Lignocellulosic Biomass. Biotechnology Progress. [87] Sun, Y., & Cheng, J. J. (2005). Dilute Acid Pretreatment of Rye Straw and Bermudagrass for Ethanol Production. Bioresource Technology, 1599-1606. [88] Koullas, D. P., Christakopoulos, P., Kekos, B., Macris, B. J., & Koukios, E. G. (1992). Correlating the Effect of Pretreatment on the Enzymatic Hydrolysis of Straw. Biotechnology and Bioengineering, 113-116. [89] Chang, V. S., & Holtzapple, M. T. (2000). Fundamental Factors Affecting Biomass Enzymatic Reactivity. Applied Biochemistry and Biotechnology, 84-86. 125  [90] Kim, S., & Holtzapple, M. T. (2006). Effect of structural features on enzyme digestibility of corn stover. Bioresource Technology, 583-591. [91] Vasquez, M. P., Da Silva, N. J., De Souza, M. B., & Pereira, N. J. (2007). Enzymatic Hydrolysis Optimization to Ethanol Production by Simultaneous Saccharification and Fermentation. Applied Biochemistry and Biotechnology, 136-140. [92] Kim, J. K., Ohb, B. R., Shinc, H.-J., Eomd, C.-Y., & Kim, S. W. (2008). Statistical optimization of enzymatic saccharification and ethanol fermentation using food waste. Process Biochemistry, 1308-1312. [93] Demirbas, A. (2005). Bioethanol from Cellulosic Materials: A Renewable Motor Fuel from Biomass. Energy Sources, 327-337. [94] Olsson, L., Soerensen, H. R., Dam, B. P., Christensen, H., Krogh, K. M., & Meyer, A. S. (2006). Separate and Simultaneous Enzymatic Hydrolysis and Fermentation of Wheat Hemicellulose with Recombinant Xylose Utilizing Saccharomyces Cerevisiae. Applied Biochemistry and Biotechnology, 129-132. [95] Dien, B. S., Hespell, R. B., Wyckoff, H. A., & Bothast, R. J. (1998). Fermentation of Hexose and Pentose Sugars Using a Novel Ethanologenic Escherichia Coli Strain. Enzyme and Microbial Technology, 366-371. [96] Yu, S., Wayman, M., & Parekh, S. K. (2004). Fermentation to Ethanol of Pentose-Containing Spent Sulphite Liquor. Biotechnology and Bioengineering, 1144-1150. [97] Wyman, C. E., Spindler, D. D., & Grohmann, K. (1992). Simultaneous Saccharification and Fermentation of Several Lignocellulosic Feedstocks to Fuel Ethanol. Biomass and Bioenergy, 301-307. [98] Wooley, R. J., & Putsche, V. (1996). Development of an Aspen Plus Physical Property Database for Biofuels Components. National Renewable Energy Laboratory.  126  [99] Aden, A., Ruth, M., Ibsen, K., Jechura, J., Neeves, K., Sheehan, J., et al. (2002). Lignocellulosic Biomass to Ethanol Process Design and Economics Utilizing Co-Current Dilute Acid Prehydrolysis and Enzymatic Hydrolysis for Corn Stover. National Renewable Energy Laboratory. [100] Piccolo, C., & Bezzo, F. (2009). A Techno-Economic Comparison Between Two Technologies for Bioethanol Production from Lignocellulose. Biomass and Bioenergy, 478-491. [101] Sluiter, A., Hames, B., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D., et al. (2011). Determination of Structural Carbohydrates and Lignin in Biomass. [102] Wood, W., & Kellogg, S. (1988). Methods of enzymology: Methods for measuring cellulase activities. [103] Bailey, M. J., Biely, P., & Poutanen, K. (1992). International Testing of Methods for Assay of Xylanase Activity. Journal of Biotechnology, 257-270. [104] Huang, H.-J., Ramaswamya, S., Al-Dajania, W., Tschirnera, U., & Cairncross, R. A. (2009). Effect of Biomass Species and Plant Size on Cellulosic Ethanol: A Comparative Process and Economic Analysis. Biomass and Bioenergy. [105] Aqua Merik. (2010, Dec 29). Rotary Drum Filters. Retrieved June 15, 2011, from Aqua Merik: http://www.aquamerik.com/catalogue/produits.cgi?category=filtration_filtreatambourrotatif& lg=eng [106] Aushulz. (2009, October). Rotary Vacuum-Drum Filter. Retrieved July 2, 2011, from Wikipedia: http://en.wikipedia.org/wiki/Rotary_vacuum-drum_filter [107] Aspen Technology Inc. (2009). Aspen Process Economic Analyzer V7.1 User Guide. Burlington.  127  [108] von Sivers, M., & Zacchi, G. (1996). Ethanol from Lignocellulose: A Review of the Economy. Bioresource Technology. [109] Hinman, N. D., Schell, D. J., Riley, C. J., & Bergeron, P. W. (1992). Preliminary estimate of the cost of ethanol production for SSF technology. Applied Biochemistry and Biotechnology. [110] Douglas, L. J. (1989). A Technical and Economic Evaluation of Wood Conversion Processes. ERM, Rewnewable Energy Branch. [111] U.S. Department of Energy. (1993). Assessment of Costs and Benefits of Flexible and Alternative Fuel Use in the U.S. Transportation Sector. [112] Isaacs, S. H. (1984). Ethanol production by enzymatic hydrolysis. Parametric analysis of a base-case process. SERI. [113] Wright, J. D., Power, A. J., & Douglas, L. J. (1986). Design and Parametric Evaluation of an Enzymatic Hydrolysis Process (Separate Hydrolysis and Fermentation). Biotechnology Bioengineering Symposium. [114] Nystrom, J., Greenwald, C. G., Hagler, R. W., & Stahr, J. (1985). Technical and Economic Feasibility of Enzyme Hydrolysis for Ethanol Production from Wood. NYERDA. [115] Jessen, H. (2010). Report: Plant Size not a Factor in Profitability. Ethanol Producer Magazine. [116] Chang, J. (2006, August 28). Indicative Chemical Prices A-Z. (ICIS) Retrieved June 5, 2011, from ICIS: http://www.icis.com/StaticPages/a-e.htm [117] Geiver, L. (2010). Novozymes: Enzyme Cost Drops to 50 Cents. Biomass Power and Thermal. [118] Wilcox, W. R. (2004). Oxygen Cost. Retrieved 06 07, 2011, from Clarkson University: http://people.clarkson.edu/~wwilcox/Design/oxycost.htm#_ftn2  128  [119] Seider, W. D., Seader, J. D., Lewin, D. R., & Widagdo, S. (2009). Product and Process Design Principles: Synthesis, Analysis and Design 3rd Edition. New York: McGraw-Hill. [120] McLaughlin, S. B., Samson, R., Bransby, D., & Wiselogel, A. (1996). Evaluating Physical, Chemical, and Energetic Properties of Perennial Grasses as Biofuels. National Bioenergy Conderence: Partnership to Develop and Apply Biomass Technologies. [121] Rolls, B. J., Bell, E. A., Castellanos, V. H., Chow, M., Pelkman, C. L., & Thorwart, M. L. (1999). Energy Density but Not Fat Content of Foods Affected Energy Intake in Lean and Obese Women. The American Journal of Clinical Nutrition. [122] Walsh, M. E., Perlack, R. L., Turnhollow, A., Ugarte, D. d., Becker, D. A., Graham, R. L., et al. (2000). Biomass Feedstock Availability in the United States: 1999 State Level. [123] International Energy Association. (2007). IEA Energy Technology Essentials: Biomass for Power Generation and CHP. [124] Nguyen, Q. A., & Saddler, J. N. (1991). An Integrated Model for the Technical and Economic Evaluation of an Enzymatic Biomass Conversion Process. Bioresource Technology. [125] Petro Canada. (2007). Gasoline Taxes Across Canada. Retrieved June 11, 2011, from Petro Canada: http://retail.petro-canada.ca/en/fuelsavings/2139.aspx [126] BC Ministry of Finance. (n.d.). How the Carbon Tax Works. Retrieved May 28, 2011, from Ministry of Finance Climate Action: http://www.fin.gov.bc.ca/tbs/tp/climate/A4.htm [127] Zhuang, J. M., & Walsh, T. (2004). Lignor Process for Acid Rock Drainage Treatment. Environmental Technology. [128] Qi, B., Chen, X., Su, Y., & Wan, Y. (2011). Enzyme Adsorption and Recycling During Hydrolysis of Wheat Straw Lignocellulose. Bioresource Technology, 2881-2889.  129  Appendix Sample calculations 1) Enzyme demand required to produce ten grams of sugar based on substrate hydrolysability  Substrate hydrolysability = 0.24  2) Empirical model %error For percentage of hemicellulose present in solid substrate after pretreatment of 22.5 min, 4% caustic, 105°C.  3) Annualized capital cost Assume interest rate (i) = 0.08 and plant life (n) = 10 Capital cost = 66.8 Million dollars  130  4) Enzyme cost from Aden et al. [99] study Table A1 Process description Known Information Enzyme Cost Enzyme loading Ethanol production Cellulose flow rate into saccharification reactor  0.1 12 8244.1 28432  Units ($/gallon of ethanol) (FPU/g cellulose) (gallons/h) (kg/h)  5) Biomass by-product value Table A2 Heat of combustion Component Lignin Carbohydrate  Heat of Combustion (kJ/g) 25.8 17.1  Assumptions: Electricity = $0.041/kWh Electrical efficiency = 28% Lignin 25.8 kJ/g * $0.041/kwh * 0.28 * (1h/3600 s)*1000 g/kg = $0.082/kg Carbohydrate 17.1 kJ/g * $0.041/kwh * 0.28 * (1h/3600 s)*1000 g/kg = $0.055/kg 131  6) Oxygen cost Table A3 Reaction conditions Residence time (min) Vessel size (m3) Reaction temperature (°C) O2 partial pressure (atm)  60 994.9 90 10  = 141.2 ton O2/day To calculate capital cost fee, the capital cost of 27 Million dollars for a 1000 ton/day oxygen plant was provided for reference and an economy of scale of 0.6 was assumed.  Assume a facility fee of 2.75% of the capital cost per month Annual capital fee = $8.4 Million * 0.0275 * 12 = $2.77 Million To calculate the electricity costs, a reference of 15 kWh/1000 ft3 of oxygen was provided. = 15 kWh/28.32m3  Total annual cost = $2.77E6 + $7.1E5 = $3.48 Million  132  Additional information Table A4 Wheat straw hydrolysis results Pretreatment conditions Time Caustic Temperature (min) load (°C) (% w/w) 15 2 90 15 2 120 15 2 150 15 6 90 15 6 120 15 6 150 15 10 90 15 10 120 15 10 150 30 2 90 30 2 120 30 2 150 30 6 90 30 6 120 30 6 150 30 10 90 30 10 120 30 10 150 60 2 90 60 2 120 60 2 150 60 6 90 60 6 120 60 6 150 60 10 90 60 10 120 60 10 150  Pretreatment substrate recovery (%) 101.7 101.5 87.0 86.1 81.8 61.5 81.8 64.7 62.4 89.0 84.8 78.5 76.9 75.4 59.8 74.8 77.6 65.8 86.7 87.2 84.0 71.3 74.6 62.2 77.2 65.8 59.9  Hydrolysability (g sugar/g substrate) Cellulosic Hemicellulosic Total sugar sugars sugar (glucose) 0.07 0.01 0.07 0.07 0.01 0.08 0.10 0.13 0.24 0.15 0.25 0.40 0.24 0.34 0.58 0.24 0.33 0.57 0.21 0.33 0.54 0.21 0.41 0.62 0.24 0.50 0.74 0.06 0.13 0.19 0.08 0.17 0.25 0.10 0.17 0.27 0.17 0.29 0.45 0.29 0.50 0.79 0.27 0.56 0.83 0.27 0.37 0.63 0.28 0.39 0.67 0.27 0.51 0.78 0.06 0.01 0.07 0.09 0.08 0.17 0.11 0.13 0.24 0.23 0.35 0.58 0.25 0.38 0.63 0.25 0.54 0.79 0.24 0.35 0.60 0.27 0.41 0.68 0.27 0.50 0.77  133  Table A6 Required input for Aspen Plus simulation Specification Oxygen delignification reaction time Oxygen delignification reaction temperature Oxygen delignification reaction caustic load (g NaOH/ g biomass) Oxygen delignification percent solids Biomass input (solids and water content)  Where to make specification Fortran: O2DELIG, Fortran: HYDROL Unit: O2-DELIG  Example Values  Fortran: O2D-NAOH  6%  Fortran: O2D-H20  10%  Stream: BIOMASS  2000 dry tonnes biomass per day at 15% water content Biomass = 83333kg/h Water = 14705.9kg/h Cellulose = 34.1% Hemicellulose = 26.2% Lignin = 19.6% Other = 20.1% 10%  Biomass composition  Fortran: BIO-IN  Hydrolysis percent solids  Fortran: HYDR-ENZ  30 min 120°C  134  2.500  Cost of Ethanol ($/L)  2.000 Temperature (oC) 1.500  90 105 120  1.000  135 150  0.500  0.000 0  5  10  15  20  25  30  Enzyme Cost ($/Million FPU)  Figure A1 Cost of ethanol versus cost of enzyme for a range of reaction temperatures. Fixed pretreatment reaction time (60 min) and caustic load (10%). Fixed NaOH ($430/tonne) and biomass cost ($20/tonne)  135  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0059082/manifest

Comment

Related Items