UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Steam explosion of biomass to produce durable wood pellets Lam, Pak Sui 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_2011_fall_lam_paksui.pdf [ 3.68MB ]
Metadata
JSON: 24-1.0059133.json
JSON-LD: 24-1.0059133-ld.json
RDF/XML (Pretty): 24-1.0059133-rdf.xml
RDF/JSON: 24-1.0059133-rdf.json
Turtle: 24-1.0059133-turtle.txt
N-Triples: 24-1.0059133-rdf-ntriples.txt
Original Record: 24-1.0059133-source.json
Full Text
24-1.0059133-fulltext.txt
Citation
24-1.0059133.ris

Full Text

STEAM EXPLOSION OF BIOMASS TO PRODUCE DURABLE WOOD PELLETS  by  PAK SUI LAM  M. Sc., Hong Kong University of Science and Technology, Hong Kong, 2005 B. Eng., Hong Kong University of Science and Technology, Hong Kong, 2004  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (CHEMICAL AND BIOLOGICAL ENGINEERING)  THE UNIVERSITY OF BRITISH COLUMBIA (VANCOUVER) May 2011 © Pak Sui Lam, 2011  ABSTRACT Wood pellet is regarded as a clean fuel for combustion with low ash content (less than 1% by weight) and a high heating value around 21500 MJ/m3 compared to a heating value of 5400 MJ/m3 for dry wood chips. However, pellet is easily disintegrated into fines due to impact or moisture sorption during handling and storage. Fines may promote dust explosion during handling or self-heating of pellets in storage. The present study investigates the use of steam explosion pretreatment to improve the pellet durability in terms of mechanical strength and moisture sorption resistance. In this research, a batch steam explosion unit consisting of a steam generator, a steam treatment reactor, and control devices was developed. Steam explosion experiments were carried out on Douglas Fir at 2 temperatures (200oC and 220oC), 2 treatment durations (5 min, 10 min), and 2 particle sizes (0.4 mm and 0.9 mm). It was found that the bulk density and tapped density of steam treated wood increased with the treatment severity. The pellets made with biomass treated at different combinations of temperature-time were 1.4 to 3.3 times stronger than untreated pellets. The steam treated biomass required 12% to 81% more energy to form durable pellets than the untreated biomass. Energy input to produce 45000 metric ton regular pellets and steam exploded pellets was estimated. The input energy ranged from 2.80 to 3.52 MJ/kg. Producing pellets from untreated biomass consumed the least energy while pellets made from biomass treated with saturated steam at 220oC for 10 minutes consumed the highest. A kinetic model for pseudolignin formation during steam explosion was developed. Based on the experimental data in this research and published literature, it was postulated that the creation of pseudolignin is responsible for improved durability of steam exploded pellets. A reaction model was developed to predict the formation of pseudolignin and evaluate the optimized treatment condition for making durable and water repellent wood pellets.  ii  PREFACE The literature review and the design of the research plan in chapter 1 were prepared by the major author (Pak Sui Lam) for the PhD comprehensive exam. The design and construction of steam explosion unit (Chapter 2) and pellet die system (Chapter 3) and the control system installation (Appendix I) were done by the major author. The major supervisors, Dr. Sokhansanj, Dr. Bi and Dr. Lim provided guidelines and advice on the work from Chapter 1 to Chapter 6. Parts of the chapter 2 of the lab work (TGA, moisture content, sieving and moisture sorption measurement) were carried out by Dr. Sylvia Larsson, a postdoctoral researcher at UBC in 2009. The design of experiment, sample preparation of steam explosion at different conditions, other physical and chemical characterizations, statistical analysis, and regression analysis were done by the major author. Parts of the chapter 2 were presented in the ASABE Annual International Meeting held in Pittsburgh from June 20th to June 23rd, 2010 under the title “Effect of Temperature, Time, Particle size and Moisture content on Physical and Chemical Properties of Steam Exploded Woody Biomass”. Parts of Chapter 2 were also submitted for publication to the journal of “Fuel” on April 25th, 2010 and under revision. Other than Sylvia Larsson, other co-authors include Shahab Sokhansanj, Xiaotao Bi, Jim Lim and Staffan Melin. Parts of Chapter 3 were presented in the AICHE annual meeting held in Salt Lake City from Nov 7th to Nov 12th, 2010 under the title of “Effect of steam explosion on wood pellet quality”. A version of Chapter 3 was prepared and submitted for publication with the help of co-authors Shahab Sokhansanj, Xiaotao Bi, Jim Lim and Staffan Melin. This publication was accepted for publication in the journal of “Energy and Fuel” on Feb 15th, 2011. Parts of Chapter 4 were presented in the ASABE Annual International Meeting held in Pittsburgh from June 21th to June 24rd, 2009 under the title “Steam Treated Sawdust Pelleting and Systems Energetics”. A version of Chapter 4 was prepared for  iii  publication. Co-authors include Shahab Sokhansanj, Xiaotao Bi, Jim Lim and Staffan Melin. Parts of Chapter 5 were presented in the 8th World Congress of Chemical Engineering held in Montreal from August 23rd to 27th, 2009 under the title “Kinetic Modeling of Pseudolignin Formation in Steam Exploded Woody Biomass”. A version of Chapter 5 was prepared for publication. Co-authors include Shahab Sokhansanj, Xiaotao Bi, Jim Lim and Staffan Melin.  iv  TABLE OF CONTENTS ABSTRACT ........................................................................................................................ ii  PREFACE .......................................................................................................................... iii  TABLE OF CONTENTS.................................................................................................... v  LIST OF TABLES ........................................................................................................... viii  LIST OF FIGURES ............................................................................................................ x  ACKNOWLEDGEMENTS ............................................................................................. xiii  Chapter 1  Introduction and Objectives .............................................................................. 1  1.1 Background and Structure of Thesis ......................................................................... 1  1.2 Literature Review on Steam Explosion of Biomass Granules .................................. 4  1.2.1 Model of Operation ............................................................................................ 5  1.2.1.1 Batch System .............................................................................................. 6  1.2.1.2 Continuous System ..................................................................................... 6  1.2.2 Steam Treatment Severity .................................................................................. 7  1.2.3 Physical, Chemical and Morphological Changes of Biomass ........................... 9  1.2.3.1 Hemicellulose ........................................................................................... 10  1.2.3.2 Cellulose ................................................................................................... 11  1.2.3.3 Lignin ........................................................................................................ 12  1.2.4 Enhancing the Binding and Hydrophobicity by Pseudolignin Lignin ............. 13  1.3 Biomass Densification ............................................................................................ 14  1.4 Quality Properties of Biomass Pellets..................................................................... 16  1.4.1 Water Sorption Ability ..................................................................................... 17  1.4.2 Durability and Hardness.................................................................................. 18  1.4.3 Moisture Content of Feedstock ........................................................................ 18  1.4.4 Particle Size Distribution of Biomass Grinds .................................................. 20  1.4.5 Klason Lignin and Extractives ......................................................................... 21  1.4.6 Compaction Pressure, Holding Time and Preheat Temperature .................... 21  1.4.7 Pellet Dimensions, Bulk Density and Single Pellet Density ............................ 23  1.5 Rationale and Objectives ........................................................................................ 24  Chapter 2  Steam Explosion Experiments ........................................................................ 33  2.1  Introduction ........................................................................................................... 33  2.2 Materials and Methods ............................................................................................ 34  2.2.1 Equipment ........................................................................................................ 34  2.2.1.1 Operation Procedures ................................................................................ 34  2.2.1.2 Preliminary Run Data Analysis ................................................................ 35  2.2.2 Materials .......................................................................................................... 36  2.2.3 Sample Preparation ......................................................................................... 36  2.2.4 Experimental Design of Steam Explosion Pre-treatment ................................ 37  2.2.5 Solid Yield, Moisture Content and Volatile Contents ...................................... 37  2.2.6 Particle Size Distribution ................................................................................. 38  2.2.7 Bulk, Tapped and Particle Density and Porosity............................................. 38  2.2.8 Image Analysis ................................................................................................. 39  2.2.9 Scanning Electron Microscope (SEM)............................................................. 40  2.2.10 Drying Kinetics .............................................................................................. 40  2.2.11 Moisture Sorption Isotherm ........................................................................... 41   v  2.2.12 Statistical Analysis ......................................................................................... 42  2.3 Results and Discussion ........................................................................................... 43  2.3.1 Solid Yield, Moisture Content and Volatile Content ........................................ 43  2.3.2 Pressure and Particle Size Distribution by Sieving ......................................... 45  2.3.4 Particle Size Distribution Determined by Scanner Imaging............................ 46  2.3.5 Bulk, Tapped and Specific Density .................................................................. 47  2.3.6 Particle Morphology Analysis ......................................................................... 48  2.3.7 Drying Kinetics of Untreated and Steam Treated Particles ............................ 49  2.3.8 Equilibrium Moisture Content of Steam Treated Particles ............................. 49  2.4 Conclusions ............................................................................................................. 50  Chapter 3  Pelletization Experiments ............................................................................... 69  3.1  Introduction ........................................................................................................... 69  3.2 Materials and Methods ............................................................................................ 71  3.2.1 Materials .......................................................................................................... 71  3.2.2 Pelletization ..................................................................................................... 71  3.2.3 Pellet Density and Specific Density ................................................................. 72  3.2.4 Breakage Test................................................................................................... 73  3.2.5 Chemical Composition ..................................................................................... 73  3.2.6 X-ray Diffraction .............................................................................................. 74  3.2.7 Moisture Sorption Rate Test ............................................................................ 74  3.2.8 Scanning Electron Microscope of Cross Section of Pellet .............................. 75  3.3 Results and Discussion ........................................................................................... 75  3.3.1 Physical Dimensions and Density of Pellets .................................................... 75  3.3.2 Force and Energy to Make Pellets................................................................... 76  3.3.3 Chemical Composition ..................................................................................... 78  3.3.4 Hardness Test................................................................................................... 79  3.3.5 Moisture Adsorption ........................................................................................ 80  3.4 Conclusions ............................................................................................................. 80  Chapter 4  Energetic Analysis of Steam Explosion and Pelletization Using Laboratory Data ......................................................................................................................... 91  4.1 Introduction ............................................................................................................. 91  4.2 Development of Analysis Method .......................................................................... 92  4.2.1 Receiving .......................................................................................................... 93  4.2.1.1 Conventional ............................................................................................. 93  4.2.1.2 Steam Explosion ....................................................................................... 94  4.2.2 Drying .............................................................................................................. 96  4.2.3 Size Reduction .................................................................................................. 99  4.2.4 Pelletization ................................................................................................... 100  4.2.5 Cooling ........................................................................................................... 101  4.2.6 Screening........................................................................................................ 101  4.3 Results and Discussion ......................................................................................... 101  4.3.1 Energy Input of Steam Explosion ................................................................... 101  4.3.2 Energy Input of Pelletization ......................................................................... 102  4.3.3 Overall Process of Wood Pellets Made from Untreated and Steam Treated Sawdust ................................................................................................................... 104  4.4 Conclusions ........................................................................................................... 107   vi  4.5 Notation................................................................................................................. 108  Chapter 5  Kinetic Modeling of Polysaccharides Depolymerization and Pseudolignin Formation of Softwood during Steam Explosion ........................................................... 119  5.1 Introduction ........................................................................................................... 119  5.2 Materials and Methods .......................................................................................... 121  5.2.1 Sample Preparation ....................................................................................... 121  5.2.2 Chemical Composition Characterization ...................................................... 121  5.3 Model Development.............................................................................................. 122  5.4 Results and Discussion ......................................................................................... 123  5.4.1 Chemical Composition of Untreated and Steam Treated Douglas Fir.......... 123  5.4.2 Estimation of Arrhenius Parameters of Polysaccharides Depolymerization 125  5.4.3 Kinetic Model of Galactan Depolymerization ............................................... 126  5.4.4 Net Galactose in Condensed Steam or in Liquid Prehydrolysate.................. 127  5.4.5 Hydroxylmethylfurfural Production from Galactose Degradation ............... 128  5.4.6 Kinetic Model of Lignin Solubilization and Condensation ............................ 129  5.4.7 New Mathematical Kinetic Model to Predict the Pseudolignin Formation .. 130  5.5 Conclusions ........................................................................................................... 132  5.6 Notation................................................................................................................. 133  Chapter 6  Conclusions and Future Research ................................................................. 140  6.1 Overall Conclusions .............................................................................................. 140  6.2 Recommendations for Future Work...................................................................... 143  References ....................................................................................................................... 146  Appendix I Data Acquisition of Steam Explosion Unit.................................................. 164   vii  LIST OF TABLES Table 1.1 Summary of Operational Conditions Used in Hydrothermal Treatments of Lignocellulosic Materials (Garrote et al., 1999) ............................................................... 25  Table 1.2 Quality Requirement of the Wood Pellets (Alakangas et al., 2006; Lehtikangas et al., 2001) ....................................................................................................................... 26  Table 1.3 Quality Requirement of the Wood Pellets (Alakangas et al., 2006; Lehtikangas et al., 2001) ....................................................................................................................... 27  Table 2.1 List of Valves in the Experiment Unit .............................................................. 53  Table 2.2 List of Thermocouples and Pressure Transducers in the Experiment Unit ...... 54  Table 2.3 Factorial Design of Experiment consisting of Four Factors with Two Levels . 54  Table 2.4 Solid Yield, Moisture Content and TGA Analysis of Untreated and Pretreated Feedstock .......................................................................................................................... 55  Table 2.5 Size Reduction Effect of Steam Explosion at Different Processing Conditions and Feedstock Variables (n = 3) ....................................................................................... 56  Table 2.6 Summary of Imaging Analysis of Particle Length and Width with and without Steam Explosion Pretreatment at Different Treatment Severities .................................... 56  Table 2.7 Bulk, Tapped and Specific Density of Untreated and Steam Treated Samples at Different Severities. (n=5) ................................................................................................ 57  Table 2.8 ANOVA Results of Bulk, Tapped Density and Geometric Mean Diameter with respect to Steam Explosion Conditions and Materials Parameters................................... 58  Table 2.9 Multiple Linear Regression Results of Bulk Density, Tapped Density and Geometric Mean Diameter with respect to Steam Explosion Conditions and Materials Parameters ......................................................................................................................... 59  Table 2.10 Estimated Parameters for Drying Kinetics of Untreated and Steam Treated Douglas Fir Particles ......................................................................................................... 59  Table 2.11 Estimated Parameters for the GAB Equation fitted to Untreated and Steam Treated Equilibrium Moisture Data for Douglas Fir Particles.......................................... 60  Table 3.1 Moisture Content and Mean Size of Particles prior to Pelletization and Particle Moisture Content, and Pellet Mass and Dimensions after Pelletization ........................... 82  Table 3.2 Forces and Input Energy to Make Pellets and Pellet Density ........................... 83  Table 3.3 Chemical Composition Analysis of Pellets Made From Untreated and Treated Feedstock .......................................................................................................................... 84  Table 3.4 Data on Hardness Tests, Crystallinity and Moisture Adsorption ..................... 85  Table 4.1 Direct Energy Input for Steam Explosion at Different Severity with 0.22 Steam to Biomass Ratio ............................................................................................................. 112  Table 4.2 Energy Saving in Size Reduction of Steam Treated Samples ........................ 112  Table 4.3 Summary of Physical and Chemical Properties of Steam Treated Wood at Different Treatment Severity .......................................................................................... 113  Table 4.4 Direct Energy Input of Pelletization of Untreated and Steam Treated Wood at Different Severity............................................................................................................ 114  Table 4.5 Direct Energy Input Per Unit Kilogram of Produced Pellets at 8 % m.c. (w.b.) to the Biomass Pelletization Process with the Laboratory Data of Steam Explosion at Different Treatment Severity and Pelletization with 0.22 Steam to Biomass Ratio. ...... 115   viii  Table 4.6 Direct Energy Input Per Unit Kilogram of Produced Pellets at 8 % m.c. (w.b.) to the Biomass Pelletization Process with Optimized Steam Explosion at Different Treatment Severity and Pelletization with 0.22 Steam to Biomass Ratio. ..................... 116  Table 5.1 Chemical Composition of the Solid Phase of the Untreated and Steam Treated Douglas Fir based on the Steam Treated Material (n=2) ................................................ 134  Table 5.2 Chemical Composition of the Solid Residue Part of the Untreated and Steam Treated Douglas Fir based on the Feedstock prior to Steam Explosion (n=2) ............... 134  Table 5.3 Chemical Composition of the Water Soluble Fraction of the Steam Treated Douglas Fir in g/100 g Feed (n=2).................................................................................. 135  Table 5.4 Estimated Arrhenius Parameters of Carbohydrates Hydrolysis Rate Constant (n=2)................................................................................................................................ 135  Table 5.5 Fitted Kinetic Parameters of Rate Constant for HMF Formation, Lignin Depolymerization and Lignin Condensation at Different Temperature .......................... 135   ix  LIST OF FIGURES Figure 1.1 Process Flow Diagram of a Plant with the Densification of Biomass Process (Mani et al., 2006a) ........................................................................................................... 28  Figure 1.2 Steam Explosion Equipment: (a.) Batch (Turn et al., 1998) and (b.) Continuous System STAKE II Pilot Plant Facility Having a Maximum Capacity of 4 t/h Located in Sherbooke, Quebec, Canada (Heitz et al., 1990). ........................................... 29  Figure 1.3 Schematic of the Ultrastructure of the Wood Cell Wall, showing the Middle Lamella (ML), the Main Cell Wall Layers and the Associated Microfibrillar Orientation (P: Primary Cell Wall, S1, S2, S3: Secondary Cell Wall with Different Orientation of Cellulose Nanofibers) (Kettunen, 2006). .......................................................................... 29  Figure 1.4 Microfibril Cross-Section showing Strands of Cellulose Molecules (Horizontal Line) Embedded in a Matrix of Hemicellulose (Curved Fibers) and Lignin (Outer Dark Layer) (Ramos et al., 2003). ............................................................................................. 30  Figure 1.5 Schematic of Goals of Deconstruction of Lignocellulosic Material (Hsu et al., 1980) ................................................................................................................................. 30  Figure 1.6 Hydrolysis of Xylan (Pentosan) into Xylose (Pentose) and Subsequent Degradation of Pentose to Produce Furfural (Ramos et al., 2003). .................................. 31  Figure 1.7 Lignin Coalescence caused by Hydrothermal Pretreatment (Donohoe et al., 2008). (A): Three Pairs of Droplets that Appear to be Fusing (Arrowheads) to Form Larger Droplets. (B): Spherical Droplets with a Rough, Coated Surface (Arrowheads). The Droplets in A and B are Naturally Redeposited on the Cell Wall Surface following Dilute Acid Pretreatment. Scale Bars = 0.5 μm. ............................................................... 31  Figure 1.8 Model of a Proposed Mechanism of Lignin Coalescence caused by Hydrothermal Pretreatment. Tm is Lignin Melting Temperature (Donohoe et al., 2008) 32  Figure 1.9 The Proposed Compaction Process (Compaction Force vs. Time) to Form a Dense Form of Loose Bulk Granular Solids (Rumpf, 1962) ............................................ 32  Figure 2.1 Closed System Steam Explosion Unit at the UBC Clean Energy Research Center ................................................................................................................................ 61  Figure 2.2 Process Flowsheet of the Closed System Steam Explosion Unit at the UBC Clean Energy Research Center (B: Ball Valve, PS: Pressure Relief Valve, T: Thermocouple, P: Digital Pressure Transducer) ............................................................... 62  Figure 2.3 A Typical Plot of Temperature for the Steam Explosion Experiment at 200oC for 5 minute. T1 is the Temperature of the Steam in Steam Boiler, T2 is the Surface Temperature of the Steamline before the Ball Valve (B-2), T3 is the Surface Temperature of the Steam Line after the Ball Valve 2 (B-2), T4 is the Temperature of the Reaction in the Reactor. ....................................................................................................................... 63  Figure 2.4 Treatment Temperature Profile (T4) of Biomass Steam Explosion Experiment at 200oC for 5 minute (n=3) .............................................................................................. 64  Figure 2.5 A Typical Plot of Pressure vs. Time during Steam Treatment at 200oC for 5 minute and at the Moment of Pressure Release (n=3). ..................................................... 64  Figure 2.6 Physical Appearance of Steam Exploded Particles Treated at Different Steam Explosion Conditions. From Left to Right: (a.) Untreated Douglas Fir, (b.) 200oC, 5 min, (c.) 200oC, 10 min, (d.) 220oC, 5 min, (e.) 220oC, 10 min. .............................................. 65   x  Figure 2.7 Solid Yield of Douglas Fir Ground Particles at Different Severities of Steam Explosion Pretreatment ..................................................................................................... 65  Figure 2.8 Particle Size Distribution of Untreated and Pretreated Douglas Fir Sawdust at Different Severities ........................................................................................................... 66  Figure 2.9 Relative Tapped Density vs. Severity of Steam Explosion for Samples Treated at Different Severity of Steam Explosion Pretreatment .................................................... 66  Figure 2.10 SEM Pictures of (a.) Untreated Ground Particle by Hammermill with 1.7 mm Screen Opening and (b.) Steam Treated Particles at 4.53 Severity with x35 Magnification ........................................................................................................................................... 67  Figure 2.11 SEM Pictures of (a.) Untreated Ground Particle by Hammermill with 1.7 mm Screen Opening and (b.) Steam Exploded Particles at 4.53 Severity with x500 Magnification .................................................................................................................... 67  Figure 2.12 Moisture Content vs. Time for the Untreated Douglas Fir Particles and Steam Treated Particles at Oven Temperature 103oC.................................................................. 68  Figure 2.13 Moisture Sorption Isotherms for Untreated Douglas Fir Particles and Steam Treated Particles at Different Temperature and Treatment Time (Solid Lines show GAB Model). .............................................................................................................................. 68  Figure 3.1 The MTI Machine with Computer Control System and Temperature Controller. ........................................................................................................................................... 86  Figure 3.2 The Piston – Cylinder Pellet Making Assembly installed in a MTI Universal Testing Equipment. Picture shows the 6.2 mm Diameter Piston, Piston Bolted to the Top Cross Head and the Die with the Heating Tape Bolted to the Bottom Plate. ................... 87  Figure 3.3 Physical Appearance of Wood Pellets Made from Untreated and Treated Douglas Fir at Different Steam Explosion Conditions. From Left to Right: (a.) Untreated, (b.) 200oC-5 min, (c.) 200oC-10 min, (d.) 220oC-5 min, (e.) 220oC-10 min. Pellets were 6.62 mm in Diameter and about 18 mm in Length. .......................................................... 88  Figure 3.4 SEM Photos of the Cross Section of Pellets Made from (a.) Untreated and from (b.) Steam Treated (220oC-10 min) Douglas Fir at Low Magnification (x30). Untreated Pellets show the Stack of Fibrous Particles. The Fibrous Structure is not Visible in the Treated Sample. .......................................................................................... 88  Figure 3.5 A Typical Force v.s. Displacement Graph for Producing Pellets. (1: Untreated, 2: 200oC-5 min, 3: 200oC-10 min, 4: 220oC-5 min, 5: 220oC-10 min.)............................ 89  Figure 3.6 Force Displacement Graph of Hardness Test of Compressing Different Pellets Made from Untreated and Steam Treated Douglas Fir Particles. (1: Untreated, 2: 200oC-5 min, 3: 200oC-10 min, 4: 220oC-5 min, 5: 220oC-10 min.) .............................................. 89  Figure 3.7 Moisture Adsorption of Pellets Made from Untreated and Steam Treated Douglas Fir Particles. Both Types of Pellets were Dried to 0% Moisture Content before Placed in the Environment Chamber Set at 30oC, 90% RH ............................................. 90  Figure 4.1 Biomass Preprocessing with and without Steam Explosion Pretreatment (Dash Line shows Steam Explosion Pretreatment, Numbers Indicate the Streamline to Indicate the Mass Balance) ........................................................................................................... 117  Figure 4.2 Relative Compression Energy Against Steam Explosion Pretreatment Severity ......................................................................................................................................... 117  Figure 4.3 Relative Total Pelletization Energy (Compression + Extrusion) per Mass and per Density against Steam Explosion Pretreatment Severity .......................................... 118   xi  Figure 4.4 The Energetic ratio, Relative Meyer Hardness and Relative Maximum Breaking Force of Pellets Treated at Different Severity of Steam Explosion Pretreatment with Steam to Biomass Ratio of 0.22.............................................................................. 118  Figure 5.1 5-Hydroxylmethylfurfural Formation at Different Temperature .................. 136  Figure 5.2 Arrhenius Plot of Reaction Rate of Arabinan, Galactan and Mannan in Natural Logarithm at Different Temperature ............................................................................... 136  Figure 5.3 Galactan Depolymerization at Different Temperature .................................. 137  Figure 5.4 Galactose Formation and Degradation at Different Temperature ................. 137  Figure 5.5 Hydroxylmethylfurfural (HMF) Formation at Different Temperature ......... 138  Figure 5.6 Acid Insoluble Lignin Depolymerization and Condensation at Different Temperature .................................................................................................................... 138  Figure 5.7 Acid Soluble Lignin Formation in Liquid Phase at Different Temperature .. 139  Figure 5.8 Pseudolignin Formation at Different Temperature........................................ 139  Figure AI.1 Data Acquisition Front Panel Labview Program ........................................ 164  Figure AI.2 Block Diagram of Temperature and Pressure Transducer Data Acquisition Labview Program ............................................................................................................ 165  Figure AI.3 Ball Valve (B-1) Opening (Left) and Closing (Right) Controlled by Labview Program ........................................................................................................................... 165  Figure AI.4 Block Diagram of Ball Valve (B-1) opening Labview Program ................ 166   xii  ACKNOWLEDGEMENTS I would like to gratefully acknowledge my academic advisor, Dr. Shahab Sokhansanj, Distinguished Research Scientist, Oak Ridge National Laboratory. This thesis would not have been completed without his patient guidance, support and inspiration. I would also like to express my sincere gratitude to my co-supervisors, Dr. Xiaotao Bi and Dr. Jim Lim, Professors of the Department of Chemical and Biological Engineering (CHBE), University of British Columbia. I appreciate their insights in process design, reactor design and instrumentation of my experiments. I would also like to acknowledge Mr. Staffan Melin, Research Director of the Wood Pellet Association of Canada (WPAC) for providing his valuable insight in the current critical challenges of the biomass handling and bioenergy industry. His knowledge of the industry is impetus for my research topic. Special thanks are also expressed to my Doctoral defense committee: External examiner: Dr. Franco Berruti, Director of Institute for Chemicals and Fuels from Alternative Resources and Professor of Department of Chemical and Biological Engineering, University of Western Ontario; University examiners: Dr. Jack Saddler, Professor of Faculty of Forestry, and Dr. Paul McFarlane, Professor of Department of Wood Science, University of British Columbia, Dr. Anthony Lau, Associate Professor of Department of Chemical and Biological Engineering, University of British Columbia and Dr. Tim Durance, Professor of Department of Food Nutrition and Health, University of British Columbia for their critical comments and valuable suggestions to improve my thesis work. Sincere thank is also given to the visiting Professor Katia Tannous, Associate Professor of Department of Thermo-fluid Dynamics of School of Chemical Engineering of the University of Campinas for her valuable comments on particle dynamics. I would also like to acknowledge the former postdoctoral fellows in UBC Biomass and Bioenergy Research Group (BBRG) for their valuable comments and guidance on my research work, Dr. Sudhagar Mani, Assistant Professor of the Faculty of Biological and Agricultural Engineering of University of Georgia, Dr. Jaya Shankar Tumuluru, Research Scientist in Bioenergy Group of Idaho National Laboratory, Dr. Igathinathane Cannayen, Assistant Professor of Department of Agricultural and Biosystems Engineering of North Dakota xiii  State University and Dr. Sylvia H. Larsson, Postdoctoral Fellow at the Unit of Biomass Technology and Chemistry, Faculty of Forest Science of Swedish University of Agricultural Sciences. I would like to thank the BBRG colleagues, CHBE workshop and administrative colleagues for their technical support to my experimental and administrative work. My special thanks are expressed to the Dean Professor Jack Saddler’s research group of the Faculty of Forestry of University of British Columbia for the technical support in carbohydrate analysis of biomass. I extend my sincere thanks for providing financial support to this research from the University of British Columbia (UBC), Wood Pellet Association of Canada (WPAC) and Natural Sciences and Engineering Research Council of Canada (NSERC). I would also like to acknowledge Dr. Ping Gao, Dr. Guohua Chen and Dr. Po Lock Yue, Professors of Department of Chemical and Biomolecular Engineering (CBME) of Hong Kong University of Science and Technology (HKUST) for their inspiration, fully support and encouragement for me to pursue the Ph.D. study. My sincere gratitude would be given to Dr. Ping Gao, for motivating me into the exciting materials science field including physical, morphological, mechanical and rheological characterization of biomaterials. Finally, I would also like to acknowledge my family for their endless love and continuous support throughout my graduate study.  xiv  Chapter 1 Introduction and Objectives 1.1 Background and Structure of Thesis The global warming issues caused by the emission of greenhouse gases from electric power plants (Bernstein et al., 1999) and vehicles (Greene et al., 2003) have called for a renewable, environmental friendly and sustainable energy development. Among various types of renewable energy, bioenergy is attractive because it can be produced from renewable biomass. Wood residues and agricultural crops can replace the fossil fuels and reduce greenhouse gas emissions because these plants absorb CO2 from the atmosphere and turn it into usable biomass (DOE, 2005). Biofuel pellets are a type of solid fuels made from biomass with uniform shape and dimensions. Pellets are made by densifying the grinds from various biomass including wood. Biomass usually comes in a bale form after harvesting from the field. Their bulk densities are around 40 – 60 kg/m3 dry basis depending on species and moisture content. The long-stem plants have to be chopped before fine grinding. Drying is required to ensure that the size-reduced feedstock is good for pelletization (densification) to produce durable pellets. The bulk densities of the biomass pellets are around 550 -700 kg/m3 depending on the pellet particle size. The volume reduction helps to increase the loading of the materials into the vehicles for transport. For storage, biomass densification helps to reduce the space required to store the materials. For example, the agricultural crop residues including corn stover and switchgrass can only be collected during a limited duration and their low bulk density is not suitable for storage to feed the processing facilities year round. Biomass pellets improve the heating efficiency and have lower emissions during combustion than using the low bulk density and fluffy biomass. The typical example is the co-firing plant using wood pellets and coal as feedstock where the difference between these two materials densities cause difficulties in feeding due to the uneven, fluffy and low bulk density of the biomass feedstock. Canada has become a major exporting country to supply logs and wood products to U.S., Japan and European countries. One of the major recent wood products supplied is  1  the export of wood pellets to Europe from the province of British Columbia. The typical production process of biofuel pellets is collecting the residues from saw mill, and following by drying using a rotary drum dryer, further size reduction to the granular form by hammer mill, and finally pelletizing into fuel pellets using a pellet mill (Figure 1.1). The fuel pellets are cooled, screened and transported to an export port by trains. They are usually transported on conveyor belts and dropped from the height of 10 - 15 m above the storage silos for temporary storage. They are stored under a well monitored environment. Pellets are then loaded into the ocean vessel. The pellet quality needs to be maintained during transport in order to meet the European standard of the biofuel for export. Biomass preprocessing is aimed at enhancing the energy density of a bulk biomass. The biomass feedstock supply logistic cost contributes around 30 – 50 % of the total bioenergy production cost (Sokhansanj et al., 2006). An optimized pre-processing of the biomass into densified pellets is essential to achieve a cost – effective production process for bioenergy. Further optimization of the process can be achieved by enhancing the production yield and reducing the energy required for the preprocessing process. Two major technical problems during the preprocessing process need to be addressed. Poor mechanical strength of biomass pellets contributes to disintegration of pellets into fines during transportation. This usually happens for the pellets transporting on the conveyor and loading from the top of the silo into the piles when pellets break into fines due to impact. The fines cause blockage of the conveyor or hopper during processing and also lead to an occupational health problem to the workers inhaling the fines (Aleksandra et al., 2006). Moreover, the fines also lead to dust explosion problem which may cause severe fire damage to the expensive handling facilities. This is related to the lack of natural binding between the fibers of the pellets, and most biomass species including straws and stover are difficult to densify without any expensive binders (Sokhansanj et al., 2005; Kaliyan and Morey, 2006; Jannasch et al., 2004). Pellets absorb moisture easily and disintegrate into fines under high humidity conditions. High surface area of the small fines favors the susceptibility of the attack by  2  the micro-organisms during storage (Lehtikangas, 2000). Anaerobic conditions lead to local heat generation and generation of toxic off-gasing that may include terpenes (Rupar et al., 2005). Local heat generated may ignite the volatiles in the pellets to cause fire, and the off-gas accumulations inside the storage silos are toxic to the workers. High moisture content reduces pellets combustion efficiency at the power plant. The quality of the fuel pellets needs to be improved with respect to durability (i.e. better binding) and better water resistance (i.e. higher hydrophobicity). The usual practice is to apply heat and/or to add binding agents to enhance the product durability. The chemical modification of the lignocellulosic feedstock by steam explosion has been shown to improve the mechanical properties and hydrophobicity of fiberboards (Suzuki et al., 1998; Bouajila et al., 2005). The improvement was contributed to the self binding property of plasticized lignin (Bouajila et al., 2005; Startsev et al., 2000). Shaw et al. (2009) reported that the pellets made from steam treated poplar and straw feedstock had an improved tensile strength. This is due to the particle size reduction after steam treatment and that pretreated feedstock had a higher lignin content available for binding. Steam explosion has been proven as an important pretreatment technology to enhance the recovery of sugars and yield of useful chemicals and fuels (Bura et al., 2002; Bura et al., 2003; Zuo et al., 2006; Liu et al., 2002; Kobayashi et al., 2004; Maness et al., 2005). The dilute acid hydrolysis alters the chemical structure of biomass by the formation of acetic acid from the degradation of hemicelluloses during steaming. The explosion caused by a sudden release of high pressure defibrillates the cellulose bundles. Individual fibers enhance the cellulose accessible site for enzymatic hydrolysis and fermentation. Steam explosion has long been used for making good quality pulp for paper making and stronger fiberboard fabrication (Vit and Kokta, 1986; Kokta and Vit, 1987). The current potential applications of steam explosion pretreated biomass to produce chemical and fuels include ethanol from lignocellulosic or crop residue feedstock (Bura et al., 2002; Bura et al., 2003), hydrogen by fermentation (Maness et al., 2005), recovery of hemicelluloses as feedstock for microbial fuel cell (Zuo et al., 2006), and biogas  3  production (e.g. methane) from bamboo and municipal solid waste by anaerobic fermentation (Liu et al., 2002; Kobayashi et al., 2004). There is some commercial evidence that steam explosion improves the durability of pellets but limited literature to support this observation is available. This research focuses on developing and optimizing the biomass pellet durability using steam explosion pretreatment. This aims not only to produce fuel pellets for ease of chemical conversion and energy production; but also to reduce the preprocessing cost and improve the safe handling during transport and storage of pellets. 1.2 Literature Review on Steam Explosion of Biomass Granules Steam explosion, also named as Masonite technology (Delong, 1981), is usually used to yield pulp based on short time vapor-phase steam treatment at temperatures ranging from 180 – 240oC followed by explosive decompression of the biomass. It involves high pressure saturated steam ranging from 150 to 500 psi (1.034 – 3.447 MPa) to heat up biomass rapidly and with or without rapid decompression (explosion) to rupture the rigid structure of the biomass. In some cases, the use of acidic gases or dilute acid as catalyst, e.g. SO2, H2SO4, is useful for enhancing the hydrolysis rate of hemicelluloses of softwood and corn fiber during the steam explosion treatment (Boussaid et al., 2000; Shevchenko et al, 2001; Bura et al., 2002). The steam explosion aims at increasing the accessible sites for cellulose for the enzymatic hydrolysis followed by fermentation for ethanol production. (Sendelius et al., 2005; Shevchenko et al., 2001; Bura et al., 2002; Bura et al., 2003). Several patents have been granted to this process (Delong, 1983; DeLong, 1990; Foody, 1984). Batch or continuous pilot plants have been commercialized by SunOpta bioprocess Inc. (former company name: StakeTech Ltd.). SunOpta has also provided small units for research purpose in universities, like University of British Columbia, Vancouver; University of Sherbrooke, Quebec in Canada; National Renewable Energy Laboratory, NREL, Golden, CO, US; Virginia Tech, Blacksburg, VA, U.S.  4  The advantages of using steam explosion pretreatment compared to other pretreatment technologies for chemical utilization of lignocelluose (Garrote et al., 1999) are: 1. No chemicals are used except water. 2. Good yield of hemicelluloses with low degraded byproducts. 3. Equipment corrosion is minimum due to a mild pH of reaction media when compared to the acid prehydrolysis. 4. Stages of handling and acid recycling are avoided. 5. Disruption of the solid residues from bundles to individual fibers occur due to explosion effect.  1.2.1 Model of Operation A steam explosion unit can be operated in batch or in continuous mode. A batch reactor is usually used in the laboratory to pre-treat the biomass for scientific research while continuous reactor is used in the industry for biomass-pretreatment process, e.g.: SunOpta bioprocess Inc.. The commercialized continuous system is adapted for a variety of biomass feedstock including forestry and agricultural residues like wheat straw, corn stover, switchgrass and wood chip. Garrote et al. (1999) reviewed a list of different hydrothermal operating conditions of lignocelluosic materials that the previous researches had done (Garrote et al., 1999). The raw materials treated by hydrothermal treatments being studied are listed in Table 1.1. Biomass is usually pretreated after size reduction by milling or chipping. The selection of reactor type is dependent upon the particle size of the samples in order to ensure better heat transfer efficiency. Preference would be given to both batch and continuous reactors for the smaller particles as they facilitate better heat transfer of the process during pretreatment (Ballesteos et al., 2002; Brownell et al., 1986). However, finer materials less than 0.15 mm diameter are more difficult to be pretreated in batch mode. The use of plug flow reactors is required to improve the pretreatment efficiency  5  and homogeneity. Both the batch and continuous systems of the steam explosion unit used in laboratory and commercial plants are shown in Figure 1.2.  1.2.1.1 Batch System The batch system is originally manufactured by the Stake Technologies, Ontario, Canada. The left diagram of Figure 1.2 shows a modified batch system with a 10L reactor with heat jacketed and an automatic control for steam pressure and sample retention time. The biomass required to load through a 76 mm ball valve is approximately 1 kg. After loading, the valve is closed and steam begins to enter the reactor at a preset pressure. The inflowing steam passes through a trace gas injection system used in the identification and quantification of biomass-derived gas species. The information including trace gas volume, temperature and pressure were recorded for each test. After a preset reaction time, the steam saturated biomass was discharged into a 160L water cooled discharge chamber through the 38 mm ball valve and subsequently collected after cooling. The discharge chamber is vented through a water cooled condensing coil. The sample of gas emitted was collected in tedlar bags over the duration of the discharge.  1.2.1.2 Continuous System The Stake II pilot facility is a continuous system of biomass processing as shown in the right diagram of Figure 1.2 (Heitz et al., 1990). The raw material is conveyed to a storage bin before being fed into the coaxial feeder. The feeder moves and compresses the material forming a dense plug which enters the digester by exerting a pressure against the choke-cone. The back pressure of the choke cone against the plug can be adjusted hydraulically. The material is transported through the digester towards the end with supplied steam. The time of the materials exposed to the steam can be controlled by varying the speed of the auger. At the end of the digester, the auger discharges the materials through a Kamyr blow valve which has 32 cm internal diameter at the throat. The opening frequency of the valve can be adjusted by a timer or via a feedback loop from the torque generated by the accumulation of the treated material onto the last  6  transfer auger. For each valve opening and closing, the high steam pressure inside the digester generates a small detonation. The wet treated materials are disengaged from the steam by a tangential cyclone located just at the entrance of the 15.2 m3 receiving bin. The suddenly enlarged volume causes the steam explosion of the material. Finally, the wet treated material falls onto the floor of the bin and conveyed to the final discharge point. For the steam, it was recycled through an aspiration-condensation sampling train operating at 1 – 1.5 l/min of condensate. The steam is produced by a 200 hp. Clayton flash boiler having a thermal steam output of 7.0 GJ/h. Injection of steam is made through three inlet ports adjacent to the choke cone area. A vortex shedding flowmeter is attached to the steam line to measure the steam flowrate and consumption. In general, the mode of operation of steam explosion depends on whether it is for large scale production or for laboratory analytical research purpose. Batch mode is suitable for optimization of the process conditions in laboratory. The materials properties changes of the biomass by steam explosion will be discussed in the following sections. 1.2.2 Steam Treatment Severity The major operational parameters of steam explosion are particle size of the biomass feedstock (dp), the applied reaction pressure (P), the reaction temperature (T) and the residence time (t). Different combinations of the reaction parameters cause definite changes on the biomass structure and chemical composition. A severity index (Ro) developed by Overend and Chornet (1987) equation (1-1) (Overend et al., 1987) is widely used for optimizing steam treatment. The equation was developed based on modeling complex reaction systems by assuming each reaction is homogenous, and Arrhenius dependence rate law and the temperature function were linearized by Taylor series (Abatzoglou et al., 1992; Montane et al., 1998).  T  100   ' Ro   exp dt  14.75  0 t  (1-1)  7  where t is the residence time (minute), T is the reaction temperature (oC) Equation (1-1) does not include the moisture content and particle size of the feedstock that also highly affect the kinetics of physical and chemical changes of the biomass structures by steam explosion. Brownell et al. (1986) found steam is prevented from physically entering the wood, and heat transfer is by the slower process of surface condensation and heat conduction into the cold interior. This leads to a higher consumption of steam in heating wood and the moisture associated with it and in pressurizing the reaction vessel or gun. In the absence of steam recovery, approximately 0.9 kg 250oC steam is consumed in heating green aspen wood (100% MC, dry basis) of equivalent dry weight 1.0 kg (Brownell et al., 1984). Air dry wood of 10% MC consumes only half as much steam. It was found that the calculated final moisture content of steam heated aspen wood increases with initial moisture content, and that for an initial moisture content of 100% or greater (dry basis) the void volume of aspen wood is filled by condensate before steam temperature is reached (Saddler et al., 1983). Further heating is then by slow conduction. Uneven steam treatment of the wood is resulted if wood of large particle size is steam heated at high temperatures, and short times are recommended for steam explosion (DeLong, 1981). Besides, the packing of the particles (porosity) inside the reactors and materials properties include thermal conductivity of samples varies with different species of biomass feedstock at different particle size and this highly affects the steam exploded materials fuel quality which are not yet fully explored. Initial moisture content and particle size of the feedstock for steam explosion have a significant effect on the subsequent bioconversion efficiency (Cullis et al., 2004). The recovery of lignin, glucose, hemi-cellulose derived sugars, and lignin is increased by the chip size and moisture content. The range of Ro in equation (1-1) depends on the process conditions of end products. The goal of making bioethanol is to recover hemicelluloses as much in acid hydrolysates at lower pretreatment severities. At low severity (Ro < 2), the destructuring  8  of the biomass begins. If the reaction is too drastic (Ro > 4), then the dehydration and condensation reaction of the hemicellulose occurs and more soluble sugar will be degraded to side product during steaming. The degraded by-products usually inhibit the reaction rates of enzymatic hydrolysis and subsequent fermentation. The degree of polymerization of the cellulose is greatly decreased for Ro beyond 3. Therefore, optimization of the steam explosion treatment within the range of Ro of 2 – 4 is the typical objective for preparing the fuel for biochemical conversions. Zimbardi et al. (1999) compared the operations of the batch and continuous steam explosion reactors with 0.5 kg/cycle and 150 kg/h on straw’s sugar composition yield and morphology at the same treatment severity and they found that the products are macroscopically different. The difference was due to the mechanical compression acting as a source of stress in addition to the hydrothermic and explosion effects inside a continuous reactor. The mechanical compression was a competitive factor with the disruption and hydrolysis of cellulose fibers. The product collected from the continuous reactor was much dense due to the mechanical compression. The dense material was favored for high loading (200 g/L of substrate) for water or alkali extraction to recover the valuable chemicals in the downstream processes. They have also developed equation (1-2) between the 2 modes of operations and the hemicelluloses availability (Zimbardi et al., 1999). Since the product properties are not fully understood, there is a future need to develop a mathematical model of steam explosion of biomass with the consideration of all of the intrinsic (physical and thermal properties of material, e.g. bulk density, particle density, heat capacity) and extrinsic (temperature, pressure, residence time) factors. log Ro , Batch  1.50  log Ro ,Continous  1  (1-2)  where Ro is the reaction severity  1.2.3 Physical, Chemical and Morphological Changes of Biomass The original ultra-structure of a lignocellulosic biomass fiber is shown in Figure 1.3. They are divided into primary (P), secondary (S1, S2, S3) layers and middle lamella (ML). The cell wall inside these structures is made of three major components: cellulose,  9  hemicellulose and lignin. Celluloses and hemicelluloses are embedded inside the lignin which is highly concentrated in the middle lamella (Figure 1.4). Steam explosion pretreatment consists of physical, chemical and mechanical effects on modifying the biomass structure. The rigid structure of a biomass can be opened up by steam explosion pretreatment by swelling and be best described by Figure 1.5 (Hsu et al., 1980). Fiber bundles begin to separate mainly at the middle lamella due to the softening of lignin above 135oC in the presence of water (Goring, 1963). The fibers start to deconstruct to give fragments with increasing treatment severity. Fiber separation at the middle lamella of hardwood and formation of uniformly distributed lignin beads on the surface of the fibers after steam explosion were observed by Scanning Electron Microscope (SEM) and Transmission electron electroscope (TEM) (Biermann et al., 1987; Angles et al., 2001; River et al., 1991). Ramos (2003) reviewed the detailed chemistry involved in steam treatment of the lignocellulosic materials and concluded that the process is an autohydrolysis process in the absence of the acid catalyst. The role of steam and other chemicals released to open up the biomass structure during steaming reaction will be discussed in the following sections.  1.2.3.1 Hemicellulose Hemicellulose has an order of degree of polymerization (DOP) about 200 – 300, lower than cellulose. Its amorphous structure favors the OH groups to be more reactive with steam/dilute acid. Hemicellulose is first hydrolyzed among cellulose and lignin during steaming. The released hemicellulose are soluble in the condensed steam after exploding out of the reactor and can be recovered after flash cooling. Both steam and acetic acid play important roles in catalyzing the hydrolysis of the hemicellulose during steam treatment. The steam itself acts as a dilute acid in the state of hydronium ion (H3O+) to cleave the polymer chains of hemicellulose by hydrolysis to  10  allow the release of C5 (xylose) monosugars from xylan and also degraded by-products like furfural by dehydration (Figure 1.6). The kinetic model of the reaction is discussed in detail in Chapter 5. Acetic acid is released from the hydrolysis of acetyl groups of the hemicellulose by the cleavage of glycosidic linkages during steaming. This further catalyzes the hydrolysis of polymers/oligomers to monosugars: xylose. All of these chemicals composition and quantities are usually characterized using a High Pressure Liquid Chromatography (HPLC). From microscopic point of view, hemicelluloses act as an interfacial coupling agent between the highly polar surface of the microfibrils and the much less polar lignin matrix. Degradation of hemicelluloses would make wood becoming brittle and rigid. This indicates the important role of hemicellulose as imparting viscoelastic properties to the wood. The hemicelluloses form H-bonds with the surface of the microfibrils and covalent linkages with the lignin matrix. The bonding between the hemicellulose and the cellulose microfibrils are hydrogen bonding and this forms a network that provides the structural backbone of the plant cell wall.  1.2.3.2 Cellulose Cellulose is highly crystalline and it is less susceptible to hydrolysis by high pressure saturated steam compared to hemicelluloses. The structure of the biomass is destructured and results in a slight increase in crystallinity of the cellulose upon the mild steam explosion treatment. Bhuiyan et al. (2000) reported that heat was found to increase significantly the crystallinity of wood cellulose; moreover, almost twice as much crystallization was observed after heat treatment of spruce and buna under a highly moist condition than under the oven-dried condition. They also suggested that other components accompanying wood cellulose were involved in the increase of crystallinity by heat treatment, and that wood cellulose contained more quasicrystalline regions than pure cellulose. They are a result of loss of amorphous hemicelluloses due to partial  11  hydrolysis and degradation and also structural reorganization of lignin after steam treatment (Shevchenko et al., 2001). This helps to explain the pretreated materials are more rigid in mechanical strength. When steam pressure was high (3 MPa), the cellulose molecules were deconstructed and degraded to release 5-hydroxylmethylfurfural (Suzuki et al., 1998). As a result, the mechanical properties of the steam exploded fiberboards decrease. It may suggest that the concentration of the 5-hydroxylmethylfurfural in the liquid or solid product can be an indirect index to estimate the mechanical properties of pretreated product.  1.2.3.3 Lignin Lignin imparts mechanical strength to the cell wall. This helps to prevent pests attack and against diseases. The presence of lignin in cell wall inhibits the enzymatic hydrolysis of cellulose. The lignin is also a highly amorphous phenolic polymer of indeterminate molecular weight. Lignification of the wood cell wall involves the diffusion of phenylpropane monomer units and polymerization to produce a random three-dimensional network, via a free-radical mechanism. Due to the random nature of the polymerization reaction, there is no definitive structure to the lignin, although the frequency of individual bond types is well established. The random nature of the lignin polymer network is the main factor in determining the complex geometry of the cell wall micropores. Lignin is responsible for providing stiffness to the cell wall and also serves to bond individual cells together in the middle lamella region. Although lignin is relatively rigid at room temperature, it undergoes glass transition at around 140oC, and the presence of moisture in the cell wall additionally serves as a plasticizer for the lignin network. The presence of the moisture in the cell wall opens up the structure of the lignin. Lignin has a low concentration of OH groups compared to the polysaccharide components.  12  During steam explosion, the major reactions involve mainly cleavage of the β-O-4 and β-5 aryl ether linkages of the high molecular weight lignin (acid insoluble lignin) and condensation takes place to generate low molecular weight lignin (acid soluble lignin). After steam explosion, the lignin melts, redistributes, condenses and forms as beads on the surface of the cellulose micro-fibrils and thus increases the porosity of the microfibers (Donohoe et al., 2008). Shevhcenko et al. (1999, 2001) suggested that the softwood lignin transformations involve condensation reactions via benzyl cation and increase the carbonyl content after steam explosion. Miranda et al. (1978) proposed the mechanism of depolymerization/repolymerization of the lignin via carbonium ion and also their results of the apparent increase in total lignin content of the product are due to the hemicellulose degradation product, furfural and lignin polymerization. The kinetic model of the reactions was discussed in detail in Chapter 5.  1.2.4 Enhancing the Binding and Hydrophobicity by Pseudolignin Lignin According to Startsev et al. (2000), fibers of the steam exploded wood are defiberated and the surfaces are chemically modified after pretreatment. The lignincellulose (LC) chains form a new chemical bonding to give an improved binding properties and hydrophobicity of the composite after hot pressing by polycondensation. This helps to explain the improvement of mechanical properties and hydrophobicity of fiberboard with the optimized steam explosion pretreatment severity. Bouajila et al. (2005) reported that the lignin plasticization contributed to the improvement of binderless fiberboard mechanical properties. The lignin-lignin and lignin-polysaccharides cross linking actions occur at high temperatures and deformation of fibers occurs under high pressure. This is important to deform sufficiently the fibers to allow enough contact surface area to produce an intimate wood-wood contact. This allows maximum deformation under minimum pressure (i.e. rubbery state) to give the largest contact area. Besides, the polycondensated fibers are found with a lower glass transition temperature about (60 - 70oC) than the raw-materials. This is due to the contribution from the flexible fragments of the sugars that have resulted from hemicelluoses hydrolysis to  13  the molecular flexibility of condensed macromolecules (i.e. lignin-cellulose flexible chains). The binding mechanism had been proposed to that of the derivatives from the decomposed hemicelluloses liberated from biomass, furfurals, polymerize with the low molecular weight lignin liberated during steaming (autohydrolysis). The resulted resin introduces the hydrophobicity and self-bonding of the products (Suzuki et al., 1998). Miranda et al. (1978) also reported the formation of furfural/lignin by autohydrolysis of aspen to account for the polycondensate hydrophobic resin on the surface of the fibers. This increases the difficulties of the extraction of the soluble lignin. Donohoe et al. (2008) provided evidence supporting the idea that thermochemical pretreatments reaching temperatures above the range for lignin phase transition (T = 150oC) cause lignins to coalesce into larger molten bodies that migrate within and out of the cell wall, and can redeposit on the surface of plant cell walls (Figure 1.7). They also proposed a mechanism as depicted in (Figure 1.8). This newly form of coalesced lignin is called as pseudolignin, which may hypothesize as a self-binding agent during pelletization. They are hydrophobic in nature.  1.3 Biomass Densification Densification of biomass particles is achieved by first grinding the biomass then applying a mechanical force to compact the biomass to create inter-particle bonding, which form uniform shaped closely packed bulk solids such as pellets, briquettes and cubes. This process can apply heat or add external binding agent to improve the binding of the particles and obtain durable final products. Several published research explored the area of densification of woody and agricultural residues. They employed different types of processing methods and proposed different mechanisms about particles binding. Mani et al. (2004) evaluated the compaction equations applied to four agricultural crops (barley straw, wheat straw, switchgrass and corn stover) and to understand the compaction mechanism. They reported the compaction models developed for studying the compaction mechanism of pharmaceutical and cellulosic materials by Kawakita-Ludde and Cooper-Eaton which fitted well with the compression data of biomass grinds. The latter model showed the prominent compaction mechanisms for biomass grinds by  14  particle rearrangement followed by elastic and plastic deformation. Further investigation on mechanism of mechanical interlocking and the ingredient melting phenomenon during biomass compression is necessary in order to have a comprehensive understanding of the compaction. Rumpf (1962) has summarized five major mechanisms involved bonding between particles during compaction/densification: 1. Forces of attraction between solid particles 2. Interfacial forces and capillary pressure in movable liquid surfaces 3. Adhesion and cohesion forces at not freely movable binder bridges 4. Solid bridges 5. Mechanical interlocking Figure 1.9 shows the typical process of compaction. The particles rearrange themselves to form a closely packed mass which includes the release of air from void spaces (Grover and Mishra, 1996) during the first stage of the compression. The particles retain most of their properties, although most energy is dissipated due to interparticle and particle to wall friction at the densified stage. When the compaction pressure increases, the particles are forced closer to each other and undergo elastic and plastic deformations. This process gives higher particle surface areas for contact and bonding forces like Van der Waal’s forces become dominant (Rumpf, 1962; Sastry and Fuerstenau, 1973; Pietsch, 1997). The particles are under stress and they become brittle and fracture leading to mechanical interlocking (Gray, 1968). Mechanical interlocking of the particles does not contribute much to the overall strength of the pellet as they do not involve any atomic forces. However, they can provide sufficient mechanical strength to resist the disruptive forces caused by elastic recovery following compression. When the pressure keeps increasing to a higher value, the volume of the pellets continue to reduce until their densities reach the true densities of the component ingredients. If the melting points of the ingredients in a powder mix that form a eutectic mixture is favorable, the heat generated at a point of contact can lead to a local melting of materials. The molten material forms very strong solid bridges once cooled (Ghebre-Sellassie, 1989). The solid  15  bridge determines the strength of the final product and formed by different mechanisms such as: crystallization of dissolved substances, hardening binders, melting and sintering and chemical reactions.  1.4 Quality Properties of Biomass Pellets The quality of the biomass pellets affects not only the ease of handling, but also the end-user application performance. A pellet must be durable, water repellent and biologically resistant for safe handling during transporting and storage in order to ensure maintaining good quality of feedstock delivered to the end-users. Nowadays, the major use of biomass pellets is used as fuel for heating in home application and for combustion in combined heat and power (CHP) plants to generate heat and electricity. The characterization of physical properties of the pellets are critical, e.g. moisture content, dimensions of single pellet, ash content and heating value, in order to optimize the parameters and meet the heating, handling and storage systems requirement to achieve a high energy output and high processing efficiency. If pellets are prepared for biological conversion for chemical productions, then the cellulose accessible sites, sugar concentration after preprocessing, pore sizes of biomass granules/pellets are important parameters needed to be considered. However, limited research has been done on the biomass fuel pellets preparation for biological conversion. The biomass fuel can be classified as (1) woody biomass, (2) herbaceous biomass, (3) fruit biomass and (4) blend and mixtures. In this report, we will only focus on (1) woody, (2) herbaceous biomass granules and (4) blend and mixtures of woody and herbaceous biomass granules to form pellets by densification. Densification characteristics of several biomass residues have been studied in the past and this include alfalfa, wheat straw, barley straw, rice straw, rice husk, and sawdust (Mani et al., 2003). The common raw materials for making wood pellets for combustion or gasification are sawdust, planer shavings and dry chips (Lehtikangas et al., 2001). The large volume availability of barks and logging residues gained their popularity as a major  16  source. Their physical and chemical properties vary with different species of biomass. The material properties and processing conditions highly affect the quality of the pellets. The quality requirement of the wood pellets made from the chemically untreated wood without bark can be summarized as in Table 1.2. Biomass particles from agricultural crops including straws, switchgrass and stover lack natural binding ability compared to wood and alfalfa pellets (Sokhansanj et al., 2005; Kaliyan and Morey, 2006; Jannasch et al., 2004). This is directly related to the lower lignin content of most agricultural crops than the wood residues and they usually need expensive binders for densification. Besides, most lignin is presented inside the middle lamella region of the cell wall of the fiber. Most lignin cannot be activated and utilized for binding upon compaction. Therefore, there is a need to explore and develop a low cost novel pretreatment process prior to pelletizing technology to improve the quality and energy density per volume of the pellets by utilizing its binding agent. The improvement in hardness and durability of the pelletized biomass can be optimized by a combination of physio-chemical treatments (e.g. steam explosion pretreatment) of biomass before and during its densification (Sokhansanj et al., 2005).  1.4.1 Water Sorption Ability Low hydrophobicity promotes absorbing moisture and disintegration into fines under high humidity. Materials are lost as fines and less material can be transported. High surface area of the small fines also favors the susceptibility of the attack by the microbial during storage (Lehtikangas, 2000). The microbial anaerobic digestion degrades the soluble and storage carbohydrates and leads to generation of local heat and toxic offgasing, e.g. terpenes (Rupar et al., 2005). Local heat generated ignites the volatiles in the pellets to cause self-ignition during storage and the accumulations of off-gases inside the storage silos are lethal to the workers during exposure. High moisture content of the pellets also reduced burning efficiency at the power plant. Therefore, hydrophobic pellets are desired to be tailor made for safe handling, transport and storage.  17  1.4.2 Durability and Hardness Durability is a measure of the resistance of the biomass pellets to impact. Durability of pellet is defined as the percentage of mass of pellets or crumbles after tumbling divided by the mass of pellet crumbles before tumbling (ASABE S269.4, 2006). It is also an index to measure the resistance of pellets broken down into fines during transporting. Pellets with high durability are desired for safe handling and transportation as the fines may accumulate and block in conveyors and silos during processing. Besides, the fine leads to dust explosion which threatens the health of the workers in the pelleting plant and storage silos. Fines are also more susceptible for fungal attack which implies a risk of temperature increase due to microbial respiration during storage (Lehtikangas, 2000; Rupar et al., 2005). Hardness in metallurgy defines the resistance of material to permanent deformation. The deformation is associated primarily with plastic properties and also to secondary elastic properties. However, it has no general definition as a quality index for agricultural processed materials. Kaliyan and Morey (2006) reviewed major factors affecting strength and durability of densified product. It is basically affected by four major parameters, moisture content of feedstock, particle size distribution of grinds and binders (Klason lignin and extractives content) of the biomass feedstock, compaction conditions including pressure, preheating temperature and compression time. The other factors include the mixing of biomass feedstock, feeding constituents and steam conditioning.  1.4.3 Moisture Content of Feedstock The biomass feedstocks like wood residues are usually collected in wet conditions (30 – 60 % w.b.). It requires drying before making pellets and this consumes lots of energy. A long drying causes the chemical compositions changes and a change in heating value of biomass. Stahl et al. reported that long drying time will lead to the loss of the volatiles with high heating value which reduce the energy content of the sawdust and meanwhile the emissions of terpenes to the atmospheres are larger (Stahl et al., 2003).  18  However, the agricultural residues including straw and switchgrass may dry naturally in the field; hence the most energy consuming process i.e., drying can be eliminated and thereby reduce the processing cost of pellets. However, field drying is risky in humid environments. In most cases, the durability of the densified products increases with the feed material moisture content until reaching a peak value. Kaliyan and Morey (2005) found that increasing moisture content from 10 to 15 % (w.b.) increased the corn stover briquette durability from 62 to 84% at 150 MPa pressure. Colley et al. (2006) found the slight increase of durability of switchgrass pellets from 95 to 96.65 % when the moisture content increases from 5 to 8 w.b. %. However, the durability of switchgrass pellets decreased from the maximum durability of 96.65% to 78.44% when the moisture content keep increases from 8 to 16 w.b. %. An optimum moisture content of the biomass feedstock between 8 – 15 w.b. % is required for pelletization to produce durable pellets. Too much moisture makes the feedstock slippery through the dies easily during compaction, and thereby reducing the pellet quality including hardness and durability. However if the materials are too dry, they may plug the holes during pelletizing and they are required to be conditioned up to a bit higher moisture content by steam conditioning. Therefore, the optimum water content can act as both binding agent and lubricant during pelletizing to produce good quality biomass pellets. For agricultural crop residues, Kaliyan and Morey (2006) found the specific energy consumption of briquetting corn stover and switchgrass decreased by 25% and 16% with increasing moisture content from 10 to 15% at 25°C. The moisture acts as lubricant so as to reduce the wall friction and also decrease the glass transition temperature of the materials (friction exists between each fiber to resist deform). The maximum pressure takes less time to reach due to the lower resistance created by the high moisture content and thus leading to a shorter compression time by about 20% for corn stover and about 10% for switchgrass. For woody material, Li et al. (2000) reported that  19  the necessary moisture for producing quality of logs from the form of sawdust and mulches of hardwood, softwood and bark from 5 to 12 % and the optimum moisture content was about 8%.  1.4.4 Particle Size Distribution of Biomass Grinds The close packing of smaller particles after size reduction favors producing denser and durable pellets. Tabil and Sokhansanj (1996) reported the durability of Alfalfa pellets increased from 69 to 75.4 % with using the grinds produced by the hammer mill screen size of 3.2 and 2.4 mm respectively. Kaliyan and Morey (2005) reported the corn stover pellets durability increase from 61.6 to 75.2% using the grinds with 0.8 and 0.66 mm geometric mean diameter of particles produced by using hammer mill screen size of 4.6 and 3 mm respectively with 150 MPa at 25°C. The increase in durability of pellets are due to the small particles having more surface area available for more bonding between the fibers than that of larger particles. The trade-off of producing fine grinds is to consume increased energy in size reduction to produce finer particles. Finer particles usually have a higher hygroscopcity. It was found that the specific energy consumption of the corn stover grind increased from 0.8 to 1.3 MJ/t when the particle size decreased from 0.8 to 0.66 mm while that of switchgrass grinds increases from 2.5 to 4.3 MJ/t when the particle size decreased from 0.64 to 0.56 mm at 25°C (Kaliyan and Morey, 2006). A larger particle size greater than 1 mm will also act as predetermined breaking points in the pellet and therefore the optimum particle size range between 0.5 to 0.7 mm is suggested (Franke and Rey, 2006). In reality, a mixture of different particle size gives the optimum durability of the pellet as they have a better inter-particle bonding with less interspaces (MacBain, 1966; Payne, 1978; Grover and Mishra, 1966). Therefore, there is a need to study the particle size distribution on the durability of the biomass pellets.  20  1.4.5 Klason Lignin and Extractives The lignin of the biomass contributes to the bonding and stabilization during pelletizing. The lignin usually softens and sometimes melts and exhibit thermosetting properties when the biomass is heated (van Dam et al., 2004). The lignin supports the bonding of particles in high pressure and high temperature densification. The content and structural arrangement of lignin and hemicelluose affects the strength of biomass pellets (Back et al., 1982). Lehtikangas reported that the durability of the pelletized sawdust, logging residues and bark are closely related to the lignin content of the biomass (Lehtikangas, 2001). The higher the lignin contents of the biomass, the higher the durability of the pellets. The species of biomass and storage time determines the lignin content. It was found that the bark stored for 3 months increased the lignin content from 38 % to 41 % compared to the fresh bark. On the contrary, sawdust has the lowest lignin content of around 28 % for the fresh and 30% for the 6 months stored sawdust. There is a loss of organic substances during storage by leaching and this result in an increase in lignin content. Therefore, the stored biomass pellets with higher lignin content and mixing or blending the biomass species with lower lignin content (e.g. straw or switchgrass) with the one with higher lignin content (sawdust from wood) may give a higher durability of the final pellets.  1.4.6 Compaction Pressure, Holding Time and Preheat Temperature Compaction pressure and pressure holding time affect the pellet density and its quality. Li et al. (2000) and Liu et al. (2000) found sawdust from wood residues requires 100 MPa compaction pressure to produce good quality logs at room temperature, but no further improvement in quality of the logs was noted when compaction pressures were increased to 138 MPa. A pressure equal to or higher than 100 MPa is necessary to produce dense and strong sawdust logs.  21  For switchgrass and corn stover, the average time for compression of the grinds collected from the hammer mill screen size of 4.6 mm at 10 % moisture content at the maximum pressure of 100 or 150 MPa at room temperature was around 215 and 340 s (Kaliyan and Morey, 2006). The bulk density of the switchgrass grind is higher than the corn stover grind (Lam et al., 2007). The durability of corn stover briquette increases from 50 to 62 % when the applied pressure used increases from 100 to 150 MPa. The durability of switchgrass pellets was zero and could not form pellets at both pressures when densified at room temperature. This indicates switchgrass briquettes cannot be made at room temperature even under high pressure due to poor binding when compared with corn stover and wood residues. Preheating temperature prior to compaction is also another factor affecting the pellet durability. The preheating temperature used is usually within the range of the glass transition temperature (Tg) of lignin. The glass transition temperature of lignin can be determined by using differential scanning calorimeter (DSC) to study the softening temperature of the material due to onset of long range coordinated molecular motion. For the major components in the cell wall of the biomass, the lignin has the lowest Tg among the other components (e.g. hemicelluose and cellulose) presented in the biomass which is about 140oC at dry conditions and around 60 – 100oC in the presence of moisture of 8 10% (Kaliyan and Morey, 2006). When the preheating temperature used is within the range of Tg of the lignin, the lignin melts and flows. Lignin and hemicelluloses were found to be amorphous and undergo plastic deformation at the temperatures in the range of glass transition temperature (Back and Salmen, 1982). The lignin is activated and comes out from the particles as natural binding components at 75 – 100oC for corn stover and switchgrass (Kaliyan and Morey, 2006). Lignin is cooled and hardened to form solid bridges upon cooling (Rumpf, 1962) and the fibers can deform to give the best shape and orientation to give the closest packing with strong bonding. Hence, the durability of pellets increases with the preheating temperature from ambient to the glass transition temperature ranges due to the lignin plasticization of the fibers and thereby increased particle bonding. Meanwhile, the energy requirement per kg of biomass pellets formed is reduced due to increase throughput by preheating (Aqa et al., 1992).  22  For corn stover and switchgrass, the durability increased by 35.7 and 13.1 percentage points when the grinds with geometric mean particle size of 0.8 mm were preheated at the temperature from 25 to 75oC at nominal 10 and 15% initial moisture content. The durability decreased beyond an optimum peak preheating temperature (100oC) due to uneven distribution of moisture inside the particles which leads to hindered particle bonding at some less heated zones (Kaliyan and Morey, 2006). In general, the durability of a pellet made from wood and agricultural residues increases with the compaction pressure and preheating temperature until reaching an optimum. It is recommended that the heating needs to be uniform and equal among all particles; otherwise weak spots appear within the pellets. However, this also increases the energy consumption of the compaction.  1.4.7 Pellet Dimensions, Bulk Density and Single Pellet Density The mass and dimension of the pellet determines the bulk density and single pellet density. This provides packing information of the parameter for designing conveyor for transportation, silo for storage and also the reactor for combustion as well. Pellets are usually in a hardened biomass cylindrical shape made by extruding the finely ground biomass through round or other shape dies (Sokhansanj and Turhollow, 2004). The relaxed density is the term to define the density after the pellet expansion mostly in longitudinal direction after ejection from the mold. The pellets are usually with 4.8 to 19.1 mm in diameter and a length of 12.7 to 25.4 mm. The bulk densities of the biomass pellets are around 550 -700 kg/m3 depending on the size and species of the pellets. The unit particle densities of pellets are on the order of 960 to 1120 kg/m3. There is a standard of physical dimensions and bulk densities of biomass pellets set up by the European Committee for Standardization, CEN (TC335) which is listed in Table 1.2 (Alakangas et al., 2006).  23  1.5 Rationale and Objectives Durability of pellets is an important characteristic of pellets for minimizing their breakage during unavoidable frequent handlings after pellets are produced. Pellets are also highly hygroscopic and readily disintegrate when placed in a humid environment. Therefore increasing the resistance of pellets to moisture uptake is an important property of pellets during handling and storage. This research is to investigate the steam explosion of biomass and the effects of steam treatments on physical and chemical properties of steam exploded biomass. Properties of the biomass are those that influence the durability of pellets. The study of the materials properties is carried out by first analyzing the microscopic properties of biomass fibers and then relating these properties to the macroscopic properties of the whole pellet subjected to the processing conditions. All of the information is important for understanding the physical, mechanical and rheological properties (macroscopic) of the biomass pellets. Specifically, the objectives of this study are to: 1. Investigate binding characteristics of biomass particles as a function of steam treatment severity. 2. Evaluate the mechanical strength and hydrophobicity of biomass pellets made from steam treated biomass. 3. Develop kinetics of binding mechanisms for biomass vs. functionality of lignin in steam treated biomass. 4. Conduct an energy and mass balance analysis of steam treatment pelletization and compare this analysis with pelletization of untreated biomass.  24  Tables Table 1.1 Summary of Operational Conditions Used in Hydrothermal Treatments of Lignocellulosic Materials (Garrote et al., 1999) Species Woods Eucalyptus Poplar Poplar Poplulus tremoloides Populus deltoides Populus deltoides Pine (Pinus silvestris) Pine (Pinus pinaster) Mixed hardwoods (oak and gum) Oak Mixed hardwoods (oak and gum) Aspen Agricultural residues Corn cobs Corn stover Sunflower seed hulls Sugar cane bagasse Wheat Straw Vine shoots Thistle biomass Sweet sorghum bagasse Almond shells Corn stalk Sugar cane bagasse Wheat straw Bamboo grass Wheat straw Almond shells  Temp (oC)  Time (min)  Particle size  Liquid/solid ratio (g/g)  230 230 215 - 225 180-230 220 - 245 180 - 235 210 190  2 4 3-7 0.7 - 4 0.42 -1.47 2 4 8  4 mm 4 mm 1 mm 20 mm 20-40 mesh 4 mm 4 mm  10 10 7.1 - 10 10 10  160 - 280  1  6 mm  -  230  4  4 mm  10  200 - 230  2.08 - 2.23  0.5 mm  20  187 - 240  0.5 - 5  2 mm  -  180 - 223 120 -190 200 185-208 190 190 210 230 180 - 200 190 - 230 150-170 120 169.6 - 206.2 205 - 230 180 - 240  3-5 15 -120 5 20 - 29 8 8 2 0.5 1 - 68 10 - 20 15 - 300 10 2 12 - 25  35-200 mesh 1 mm 4 mm 4 mm 4 mm 4 mm 60 mesh 0.5 mm 0.8 mm 60 mesh 0.5 mm 60 mesh  2 - 10 4 10 10 10 10 3 25 5 -10 38 14.3  25  Table 1.2 Quality Requirement of the Wood Pellets (Alakangas et al., 2006; Lehtikangas et al., 2001) Parameter  Effect  CEN Standards for wood pellets as high quality for household usage (CEN/TS 14961:2005)  Moisture content  Storability, spoilage, heat calorific value, self ignition, drying cost  Moisture < 10 wt%  Ash content  Heating value, affect milling and pelleting equipment, combustion  < 0.7 wt% of dry matter  Calorific heating value  Fuel utilization, plant design  >= 4.7 kWh/kg (i.e. 16.9 MJ/kg)  Particle size distribution  Unit density of pellet, pellet quality including hydrophobicity  N/A  Toxic emission content, affects melting point of ash, formations of deposits Sulphur and (slagging) in the furnace during Chlorine content combustion, Corrosive to the furnace wall  < 0.05 wt % of dry matter for sulfur  Extractives content  Binding quality, Durability, Hardness, Ignition, Storability, Emissions  N/A  Klason lignin content  Binding quality, Durability, Hardness  N/A  -  Length (L) and diameter (D)  Bulk density  Combustion properties, specific heat conductivity, rate of gasification, Fuel feeding properties  Transport load and cost  -  D: 6±0.5 mm and L: L < 5 x diameter or D: 8±0.5 mm and L: L < 4 x diameter Maximum 20 wt% of the pellets may have a length of 7.5 x diameter  > 600 kg/m3  26  Table 1.3 Quality Requirement of the Wood Pellets (Alakangas et al., 2006; Lehtikangas et al., 2001)  Durability  Handling and transport  - 97.5 wt% of a pellet batch of 100 g shall be uncrushed after testing - Percentage of fines among pellets sieved through < 3.15 mm sieve shall not exceed 1 or 2 wt % at factory gate  Hardness  Resistance to deformation and abrasion, change of structure and properties under pressure  N/A  27  Figures  Figure 1.1 Process Flow Diagram of a Plant with the Densification of Biomass Process (Mani et al., 2006a)  28  Figure 1.2 Steam Explosion Equipment: (a.) Batch (Turn et al., 1998) and (b.) Continuous System STAKE II Pilot Plant Facility Having a Maximum Capacity of 4 t/h Located in Sherbooke, Quebec, Canada (Heitz et al., 1990).  Figure 1.3 Schematic of the Ultrastructure of the Wood Cell Wall, showing the Middle Lamella (ML), the Main Cell Wall Layers and the Associated Microfibrillar Orientation (P: Primary Cell Wall, S1, S2, S3: Secondary Cell Wall with Different Orientation of Cellulose Nanofibers) (Kettunen, 2006). 29  Figure 1.4 Microfibril Cross-Section showing Strands of Cellulose Molecules (Horizontal Line) Embedded in a Matrix of Hemicellulose (Curved Fibers) and Lignin (Outer Dark Layer) (Ramos et al., 2003).  Figure 1.5 Schematic of Goals of Deconstruction of Lignocellulosic Material (Hsu et al., 1980)  30  Figure 1.6 Hydrolysis of Xylan (Pentosan) into Xylose (Pentose) and Subsequent Degradation of Pentose to Produce Furfural (Ramos et al., 2003).  Figure 1.7 Lignin Coalescence caused by Hydrothermal Pretreatment (Donohoe et al., 2008). (A): Three Pairs of Droplets that Appear to be Fusing (Arrowheads) to Form Larger Droplets. (B): Spherical Droplets with a Rough, Coated Surface (Arrowheads). The Droplets in A and B are Naturally Redeposited on the Cell Wall Surface following Dilute Acid Pretreatment. Scale Bars = 0.5 μm.  31  Figure 1.8 Model of a Proposed Mechanism of Lignin Coalescence caused by Hydrothermal Pretreatment. Tm is Lignin Melting Temperature (Donohoe et al., 2008)  Figure 1.9 The Proposed Compaction Process (Compaction Force vs. Time) to Form a Dense Form of Loose Bulk Granular Solids (Rumpf, 1962)  32  Chapter 2 Steam Explosion Experiments 2.1  Introduction Steam explosion is an efficient and economical pre-treatment method to improve  the biomass feedstock quality for downstream ethanol conversion (Zimbardi et al., 2002). This treatment exposes biomass to steam at temperatures between 180–240oC (1.03 – 3.45 MPa) for several minutes, followed by depressurization to ambient condition. With autohydrolysis and explosive depressurization particle size distribution, shapes and chemical composition of the biomass feedstock are altered. Particle size reduction and chemical composition changes by steam explosion enhance the hydrolysis and ethanol yield (Boussaid et al., 2004; Mabee et al., 2007). The percentage of fines smaller than 0.074 mm sieve opening (# 200 mesh) increased from 5% to 90% by weight with increasing pretreatment severity of steam exploded Douglas Fir wood chips (Boussaid et al., 2000). The conventional mechanical methods were reported to require roughly 70% more energy to achieve the same size reduction as explosive depressurization (Holtzapple et al., 1989). In Chapter 1.4, the particle size, size distribution and initial packed bulk density of biomass grinds have been identified as important factors in densifying woody biomass into pellets. In general, smaller particle sizes led to higher bulk densities with closer packing (Lam et al., 2008; Mani et al., 2006). The influence of physiochemical alteration of biomass by steam explosion on pelletizing properties has been evaluated for poplar wood and steam exploded wood, which had an average pellet density of 1226 kg/m3, compared to untreated wood with an average density of 1086 kg/m3 (Shaw et al., 2009). The geometric mean particle size and bulk density of untreated wheat straw was significantly larger than steam exploded straw (Adapa et al., 2010a). However, these studies compared the properties of pellets made from the untreated materials and materials treated at one steam explosion condition.  33  Therefore, the objectives of this study are: 1. To study the effects of saturated steam temperature and exposure time on particle size, moisture content and density of a woody biomass. 2. To develop model equations of steam explosion with additional terms (particle size and moisture content) to predict bulk, tapped density, geometric mean diameter of treated samples under different severity by multiple linear regression (MLR).  2.2 Materials and Methods 2.2.1 Equipment Steam explosion pretreatments were carried out in a closed batch unit (Figure 2.1). The process flow diagram of the whole setup is depicted in Figure 2.2. The unit consists of a 2 L steam generator, generating saturated steam, and a 1 L reactor for steaming biomass samples. The heat of steam generator was supplied by a 3 zone tubular furnace (Lindberg/Blue M, STF55666C). The 1 L reactor was equipped with a 12.7 mm diameter ball valve, controlled by an electrical actuator for rapid discharging of the treated biomass into ambient pressure. A surface mounted thermocouple connected to a temperature controller regulated power input to the heater wrapped around the reactor. The details and functions of valves in the experiment unit were summarized in Table 2.1. Temperatures and pressures were measured by 1.6 mm diameter K-type thermocouples (Omega, Stamford, USA) and digital pressure transducers (Omega, Stamford, USA), respectively. Lists of thermocouples and pressure transducers are summarized in Table 2.2. Data were acquired by LabView 8.2 software (National Instruments, Austin, Texas, USA). 2.2.1.1 Operation Procedures In each experiment, 300 mL of distilled water were used to generate saturated steam. 2 L of distilled water were loaded into the water tank and pumped into the steam boiler by a water supply pump. Before loading the water, ball valves B-4 and B-6 were opened to purge the remaining water. The opening of B-4 valve facilitates faster water drainage as trapped air can vent through that valve. Ball valve (B-2) has to be always kept closed during water purging and loading. Prior to biomass loading, ball valve (B-1)  34  is closed. Biomass of 25 g powder was loaded through the ball valve B-3 into the 1 L reactor. The data were logged and saved in text file by LabView 8.2 program. The data acquisition card used is PCI DAS-08. Figure AI.1 shows the front panel data acquisition program. The data logging rate can be chosen in the delay knob on the front panel to store each data point in milli-seconds. The data logging rate of the experiments performed in this research was all with 1 data point per second. Temperature and pressure were measured based on the voltage signals and converted to degree Celsius and pound per square inch, respectively. The detail of the programming in terms of block diagram was summarized in Figure AI.2. The control of the ball valve (B-1) for rapid opening to discharge the steam cooked biomass was done by Labview control program (Figure AI.3). Two switches were used to control the signal through the digital port to open/close the ball valve (B-1). When both the output signal are high (1), the ball valves will open. When one of the output signal is high (1) and the second output signal is low (0), the ball valves will be closed. Details of the programming are summarized in the block diagram as shown in Figure AI.4.  2.2.1.2 Preliminary Run Data Analysis Figure 2.3 shows the overall temperature profile throughout an experiment at 200oC over 5 minute treatment. The steam was generated inside the boiler and the temperature (T1) kept increasing to 200oC after an hour. When the temperature of the generated steam (T1) reached 150oC, the temperature controller was switched on to provide maintenance heat to the reactor. The ramping rate of the temperature controller was set to allow the target reactor temperature (T4) of 200oC being reached simultaneously when the temperature (T1) of steam generated in the steam generator reached 200oC. This helped to prevent the prolonged heating in air to the biomass inside the reactor before steaming. When the steam reached the target reaction temperature, ball valve (B-2) was opened to allow the saturated steam transferred to the reactor chamber to  35  treat the 25 g sample for a pre-specified period. Figure 2.4 shows three experimental trials on the same steam explosion at 200oC and 5 minutes. The treatment temperature (T4) showed good agreement with three experimental trials. Similar result was also valid on the pressure before explosion (P2) throughout the 5 minute steam treatment (Figure 2.5). Treated samples were discharged into ambient conditions (101 kPa, 23oC) by opening the ball valve at 300 s. The pressure of the chamber rapidly decreased to ambient gauge pressure (0 PSI) within 3 s. Exploded samples were collected in a plastic bag inside the collection chamber and kept in sealed plastic containers at 6oC for further analysis. The preliminary run data analysis showed that the experimental unit operations were repeatable at the same processing condition.  2.2.2 Materials Douglas Fir (Pseudotsuga taxifolia) wood chips with no dirt were collected from piles of wood chips at Fiberco facilities in North Vancouver, B.C., Canada. The Douglas Fir was originated from the forest of British Columbia, Canada.  2.2.3 Sample Preparation Sampled wood chips had a moisture content of approximately 55% (w.b.). Samples were dried to 25% (w.b.) in an oven at 103oC. Dried samples were size reduced by a hammermill into two different particle size distributions (using screen openings of 1.6 mm and 3.2 mm), with geometric mean diameters of 0.421 and 0.898 mm, respectively. Samples were then further dried at 50oC to two different moisture contents (10% and 15% w.b), respectively. Samples were stored in plastic bags at 6oC before steam explosion pretreatment.  36  2.2.4 Experimental Design of Steam Explosion Pre-treatment A full factorial design, consisting of four factors (steam treatment temperature (oC), steam treatment time (min), feedstock particle size (mm), feedstock moisture content (%)) at two levels (low and high), were investigated (Table 2.3). 25 g of samples were randomly selected from the 500 g powder mixture for steam explosion pretreatment. All steam explosion pre-treatments were carried out with 3 replicates and the total number of experiments was 48.  2.2.5 Solid Yield, Moisture Content and Volatile Contents Moisture contents of the samples were analyzed according to ASABE S358.2 (2010). Around 1 g of each sample was used for moisture content measurement. Triplicate samples were oven dried at 103oC for 24 hours and moisture contents were reported in wet basis. We assume the residual moisture after 24 hours is negligible and the loss of volatiles does not affect the mass loss significantly during moisture content measurement. Solid yield recovery was calculated on dry solid basis as the percentage of recovered material after steam explosion of the initial sample weight. Volatile matter yield was analyzed according to the ASTM standard (ASTM, 2009) using a thermogravimetric analyzer (TGA-50, Kyoto, Japan). A sample amount of approximately 10 mg was put in a ceramic sample holder. The sample was treated according to the following schedule: A: 107C for 5 min in a nitrogen environment, B: 950C for 7 min in a nitrogen environment, C: increment from 600 to 750C at a rate of 15C/min in an air environment. Sample weights (mX) were recorded after each step, A to C. Ash free volatile yield (%) was calculated as:  y  m A  mB  100 m A  mC  (2-1)  where y is the ash free volatile yield (%)  37  2.2.6 Particle Size Distribution Prior to sieving analysis, samples were conditioned to 10% moisture content (w.b.) at a drying temperature of 50oC. Particle size distributions were determined according to ASABE standard S319.3 JUL97 (ASABE standard, 2006), using a Ro-Tap sieve shaker (Tyler Industrial Products, OH, USA). A sample amount of approximately 20 g was placed on top of a stack of sieves, arranged from the smallest to the largest mesh number. Sieves used for 1/8” samples were: 10, 14, 18, 25, 35, 45, 60, 80, 100, and 120, corresponding to nominal sieve openings of 2, 1.41, 1.00, 0.707, 0.500, 0.354, 0.250, 0.177, 0.149, and 0.125 mm. Sieves used for 1/16” samples were 18, 25, 35, 45, 60, 80, 100, 120, 170 and 230, corresponding to nominal sieve openings of 1.00, 0.707, 0.500, 0.354, 0.250, 0.177, 0.149, 0.125, 0.088 and 0.063 mm. Sieving time was 5 min for each sample. In total, 54 sieving trials were performed (one for each batch of steam-exploded samples, and triplicates of untreated samples for 1/8” and 1/16” grinds). The mass retained on each sieve was weighed to obtain the particle size distribution of the biomass. The geometric mean diameter (dgw) of the sample and geometric standard deviation of particle diameter (Sgw) were calculated accordingly.  2.2.7 Bulk, Tapped and Particle Density and Porosity Bulk density measurements were performed according to a slightly modified method from Mani et al. (2006). A glass cylinder with a volume of 10 mL was used. The funnel was filled with the biomass powder that was allowed to flow freely into the cylinder from a height of 20 cm and the weight/volume (loose bulk density) was determined. For tapped density, the loosely filled container was tapped on the laboratory bench 15 times before the tapped volume was recorded. Relative bulk density is defined as the ratio of the bulk density of the steam treated samples to the bulk density of untreated samples at the same moisture content. Particle density was determined by measuring the total pore volume by the adsorption of nitrogen into void spaces of the biomass powders with a Quantachrome Multipycnometer (Quantachrome, Boyton Beach, FL, USA). The measurement was done  38  by measuring the pressure difference when a known quantity of pressurized nitrogen flows from a reference volume into a sample cell with samples. Particle volume measurements were repeated five times for each sample, for determination of an average, and a reference volume of 29.42 cm3 was used in all measurements. Equation (2-2) was used to obtain the true volume of the biomass grinds deduced from the ideal gas law (Quantachrome Instruments, Multipycnometer Operating Manual). The particle density was determined by the ratio of the mass of the biomass grinds and the true volume Equation (2-3):  P  V p  VC  VR  1  1  P2   (2-2) 3  3  where: Vp = true volume of biomass grinds (m ), Vc = volume of sample cell (m ), VR = 3  reference volume (m ), P1 = pressure reading after pressurizing the reference volume (Pa) and P2 = pressure reading after including Vc (Pa)  p   m Vp  (2-3) 3  where: m = mass of the sample (kg) , Vp = true volume of biomass grinds (m ) Interparticle porosity (εo) provides packing information of the biomass grinds inside a known container and is determined by equation (2-4):  o  1  b p  (2-4)  where εo is the porosity, ρb = bulk density of biomass grinds (kg/m3), ρp = specific density of biomass grinds (kg/m3)  2.2.8 Image Analysis The images of the samples were taken by a CanoScan 4400F high resolution scanner (Canon, Lake Success, NY). The resolution of the image was scanned with 300 DPI. The particles were scattered on a transparency before images were taken by the scanner. Prior to imaging, the individual particles were guaranteed not to be in touch with each other which may affect the particle size analysis results. This was done by manual  39  separation of the particles under the observations using a magnified glass. The resulted images were analyzed by MATLAB software (The MathWorks Inc., Natick, MA) with imaging processing and statistical toolbox. The particle length and particle width were defined as an ellipsoidal major axis and ellipsoidal minor axis measured by MATLAB imaging toolbox. These two major parameters were used in toolbox to calculate individual particle’s equivalent spherical diameter and aspect ratio. The number of particles, particle length, particle width, aspect ratio and equivalent diameter in average values were reported. For each steam exploded powders at different severities, three imaging replicates were measured and appropriate measurements average values were reported.  2.2.9 Scanning Electron Microscope (SEM) SEM observations of the ultra-structure and surface were carried out on samples. The powder samples were mounted on specimen stubs and coated with gold under vacuum. All photographs were taken at 10 to 20 kV accelerating voltage by using a field emission scanning electron microscope, Hitachi S4700 (Hitachi, Japan) in UBC bioimaging facility.  2.2.10 Drying Kinetics The drying kinetics of the untreated and steam treated powder were investigated. Five samples were chosen; untreated as control, and four samples covering treatment temperatures of 200 and 220C, and treatment times of 5 and 10 minutes. The samples of 2 g each were dried in oven at 103oC for 24 hours prior to conditioning to 50% m.c. (w.b.). The conditioning was done by spraying the required amount of water to the fully dried samples to achieve the 50% m.c. (w.b.) and stored at 6oC cold room to equilibrate for 48 hours. The weight of the samples was measured every 5 minutes for the first 30 minutes and at 30 minutes intervals afterwards to develop a drying kinetic curve. The drying kinetic curve was developed based on ASABE S448.1, 2001. The moisture ratio against time was plotted to obtain the drying kinetic constant (k). Three replicates of measurement were reported. 40  M  Me  e kt Mi  Me  (2-5)  where M is the instantaneous moisture content (%, decimal dry basis), Me is the equilibrium moisture content = 0 (%, decimal dry basis), k is the drying kinetic constant (min-1), t is the time (min)  2.2.11 Moisture Sorption Isotherm Moisture equilibrium analysis was performed on the samples before pelletization. Five samples were chosen; untreated, and four samples covering steam treatment temperatures of 200 and 220C, and steam treatment times of 5 and 10 minute. A reference sample of cellulose (Avicel) was used, according to the guidelines given by Spiess and Wolf (1987), to verify conditions in each of the treatments. To reach sorption mode prior to EMC treatments, samples were dried at 103C for 24 h. After drying, samples were moved to desiccators to cool down to room temperature without absorbing air moisture. Ten different saturated salt solutions were prepared in 1.8 L glass jars (Slom, IKEA, Sweden) with silicone rubber gaskets. Salt solutions covered a range from 11 to 97% relative humidity (RH). The solutions used were: LiCl (11% RH); CH3COOK (22% RH); MgCl2 (33% RH); K2CO3 (43% RH); LiNO3 (47% RH); Mg(NO3)2 (53% RH); NaBr (57% RH);NH4NO3 (62% RH); SrCl3 (71% RH); NaCl (75% RH); (NH4)2SO4 (80% RH); KCl (84% RH); BaCl2 (88% RH); KNO3 (92% RH); Pb(NO3)2 (95% RH) and K2SO4 (97% RH). The jars were equipped with a custom-made nylon plastic sample holder stacks with six levels. In each jar, about 1.5 gram of powder from each sample was placed in weighing aluminum trays that were stacked in random order in the sample holder ladder. Samples were kept in the jars at room temperature for 45 days, during which temperature was logged (USB-502, Measurement Computing Corporation, Norton, MA, USA). After treatment, moisture contents of samples were analyzed according to ASABE S358.2 (2010).  41  Giggenheim-Anderson-deBoer (GAB) model (Chen et al., 1998) was fitted to the experimentally obtained moisture content data. Matlab curve fitting toolbox (The MathWorks Inc., Natick, MA) was used to estimate the constants of the GAB model,  M  M 0 KCAw 1  KAw 1  KAw  CKAw   (2-6)  where M is the moisture content (%), Aw is water activity. The constants M0 (monolayer moisture content), K and C are estimated from experimental isotherm data. Water activity Aw is equivalent to the relative humidity. It represents the ration of vapor pressure of a substance in a solid over the saturated vapor pressure of the substance in the solid at a given temperature. 2.2.12 Statistical Analysis Multiple linear regression was used to model responses of: relative tapped bulk density (ratio of tapped bulk density of sample before and after steam treatment), relative mean geometric diameter (ration between mean geometric diameter of sample before and after steam treatment), volatile content (%), solid yield (%), final steam treatment pressure (Pa). MLR-models were created from range-scaled factors and interactions of the four variables: steam treatment temperature (oC), steam treatment time (s), feedstock particle size (mm), and feedstock moisture content (%). By the use of range-scaled factors, differences in magnitude in factors are extinguished when values are recalculated to range from -1 to +1. Thus, the sign of modelled coefficients show if factors are negatively or positively correlated to the response, and the magnitude of the modelled coefficients will be equivalent to the impact that each factor has on the response. For a further description of range-scaling; see Myers and Montgomery (2003). Modelling and statistical evaluation were performed using the software SAS 9.1.3 (SAS Institute Inc. Cary, NC). For responses, for which MLR-models with an R2≥0.85 couldn’t be created, factors were explored with t-test analysis, to determine any significant (α=0.05) influence on the response. MLR-models with R2≥0.85 were verified by calculating the root mean squared error of cross-validation (RMSECV), using the following formula:  42  RMSECV    y   y pred   2  obs  n  (2-7)  where yobs was the observed responses, ypred was the predicted responses obtained from leave-one-out cross-validation, and n was the number of observations.  2.3 Results and Discussion 2.3.1 Solid Yield, Moisture Content and Volatile Content The darkness of the samples increased with the severity of the treatment (Figure 2.6). The untreated pellet had a light color and the rest of the pellets made by steam treated sawdust turned brownish in color. Sehistedt-Persson (2003) suggested that the darkening of color of wood treated at higher steam temperature and longer residence time resulted from hemicelluloses degradation might be due to hydrolysis by a reaction similar to Maillard reaction. Maillard reaction is a non-enzymatic reaction between an amino acid and a reducing sugar in the presence of heat. In this reaction the carbonyl group of the sugar reacts with the amino acid to form complex chemicals that could produce odors and off colors. Peterson et al. (2010) reported the presence of kinetic and mechanistic evidence of the occurrence of a Maillard-type reaction under conditions of interest to hydrothermal biomass processing. The presence of proteins and amino acids in biomass feedstock can lead to processing difficulties in terms of fouling and separation of aqueous and oil phases produced. Others reported that the cause of darkening color of the steam treated wood was due to the formation of coloured degradation produces from the extractives (McDonald et al. 1997, Sundqvist et al., 2002). An oxidized product from heat treatment, Quinones, was also suggested to contribute to the color changes (Tjeerdsma et al., 1998; Mitsui et al., 2001; Bekhta et al., 2003). Bruno et al. (2008a) reported that pine and eucalypt wood became darker with steam heat treatment similar to the treatment with hot air. The moisture content of the samples after steam explosion pretreatment ranged between 14.5% and 32.2% (Table 2.4). The mean moisture content of samples treated at 220⁰C was significantly (α=0.05) higher, compared to the mean of samples treated at 200  43  ⁰C. The solid yield of samples varied between 51% and 84% (Figure 2.7). The mean  value for solid yield was significantly (α=0.05) lower for samples treated at 220oC, compared to samples treated at 200oC. The solid yield losses of less than 50% for woody biomass from compressed hot water treatment at 240oC were reported (Kobayashi et al., 2009). Generally, losses are considered to be due to errors in materials recovery and volatile loss during steam explosion (Ibrahim et al., 1999; Emmel et al., 2003). In our experiments, the loss of fine particles adhering to the plastic bag that the sample was gathered after steam explosion was estimated to amount to no more than 1 g, or less than 5 % of the sample weight. Difficulty in determining solid yield of the steam treated biomass has been reported previously. Smaller feedstock even increased the solid yield loss of recovered samples. A 17 – 26% dry solid recovery loss after steam explosion with severity between 4.3 – 4.54 was reported for red oak chips (Ibrahim et al., 1999). It was also difficult to relate the volatile loss and solid yield of the steam treated biomass. The volatile compounds could not be quantitatively recovered due to partial loss to the atmosphere after explosive decompression (Emmel et al., 2003). Most original wood extractives were lost as volatile compounds while new extractives appear as degradation products of structural polymers from carbohydrates and lignin (Bruno et al., 2008b). From our results, the mean volatile content of samples treated at 220⁰C was significantly (α=0.05) lower compared to the mean of samples treated at 200⁰C. No other factor showed to be significant for volatile content. Mean values (with 95% confidence intervals) of volatile contents of untreated samples, samples treated at 200⁰C, and samples treated at 220⁰C were: 85.4 (±1.4)%, 84.2 (±1.4)%, and 78.9 (±1.0)%, respectively. Consequently, there were no significant difference between the mean volatile contents of untreated samples, and samples treated at 200⁰C.  44  2.3.2 Pressure and Particle Size Distribution by Sieving The pressure of steam before the samples were released from high pressure steam treatment chamber to a 50 L container ambient (i.e. explosion pressure) ranged between 1.17 – 1.32 MPa (170 – 190 psi) for samples at 200oC and between 1.73 – 1.94 MPa (250 – 280 psi) for samples treated at 220oC, respectively (Table 2.5). The steam pressure at pressure release point was modeled as a function of time and temperature, P = -4.6941+0.0008t+0.0273T R2=0.85, RMSECV = 0.13 Pa  (2-8)  where P is pressure in MPa, t is time (s) and T is temperature (oC).  A multilinear regression model with R2≥0.85 for relative geometric mean diameter could not be created when all samples were included. During sieving, it was recognized that samples ground with 1.7 mm screen were difficult to handle. Very small particles tended to stick to larger particles or to sieves due to electrostatics, and thus made consistent results hard to be obtained. The relative geometric mean diameter of samples ground with 3.2 mm screen size was modeled with R2= 0.87 from the factors of temperature, time, and moisture content. Scaled and centered coefficients of the MLR-model (Table 2.8) showed that, within the range of the design, temperature had a strong negative correlation, followed by moisture content (positive correlation) and time (negative correlation). The percentage of smaller particles and fines in each sample increased with increasing reaction severity. Figure 2.8 shows the normal size distribution of the samples of 1.7 mm ground untreated particles with 10% m.c. and respective treated particles under different severities. The fragmentation of the particles can be described by the formation of two peak regions for increasing treatment severity, i.e. fine particles and larger particles. This can be explained by the fragmentation due to pressure drop by explosion. It was speculated that the fine particles were disintegrated from the surface of the wood particles due to explosion effect and gradually formed two groups of particles. Boussaid et al., (2000) had a similar conclusion when they treated Douglas Fir with steam 45  explosion. Adapa et al. (2010a) found similar results when agricultural straws were steam exploded. In their work, the geometric mean particle size of straw particles of 0.452 and 0.997 mm was reduced to 0.309 and 0.568 mm when wheat straw particles treated at 180oC for 4 minute. Wu et al. (2000) reported the similar phenomenon that the coal and char fragmentation might have occurred during devolatilization at high pressure. The traditional sieving is not adequate to fully understand the particle size and shapes changes of the wood particles after steam explosion. This acts as a driving force to use the imaging technique to further explore the alternation of particle sizes and shapes brought by steam explosion at different treatment severities.  2.3.4 Particle Size Distribution Determined by Scanner Imaging Table 2.6 summarizes the image analysis of particle length and width for the untreated particles and the particles with steam explosion at different treatment severities. For each imaging trial, it was about 100 – 300 particles to be measured and average values of dimensions were reported. For untreated wood powder, the average length ranged between 3.14 – 4.22 mm and the average width ranged between 0.89 – 1.12 mm. The geometric mean diameter of untreated wood powder by sieving was 0.9 mm (Figure 2.8). This value was in the similar range compared to the width measured by imaging analysis. This shows that the sieving analysis was limited by the width direction. During sieving, the particle width is the determining size to pass through the sieve opening. Therefore, sieving itself is not sufficient to report the overall particle size and shape of the biomass particles. Similar results in average width for other particles treated at different severity were closely in agreement with the values of the geometric mean diameters determined by sieving. From the image analysis, we observed that both the relative length and width of the steam exploded wood particles at four different treatment severities were smaller than the untreated particles (Table 2.6). In particular, the relative length and width of the steam exploded particles increased from 0.85 at a severity of 3.64 to 0.9 at severity of 46  3.94. Both relative responses decreased with increasing the severity treatment at 4.23 and 4.53, respectively. The initial increase of the relative particle dimensions may be due to the devolatilization during steaming at the severity of 4.23. The particle swelled with the expansion of the release of volatiles during steaming. Similar observations of the coal steaming were also reported by Wu et al. (2000). The potential of a particle to swell is reduced with a reduction in volatile release due to a lower internal pressure. Further particle dimensions reduction with increasing treatment severity was due to the fines formation by fragmentation and pores shrinkage upon heating.  2.3.5 Bulk, Tapped and Specific Density Table 2.7 lists the bulk (loose fill and dense fill) and specific densities of untreated and steam treated samples under different severity. Relative bulk densities of steam treated samples ranged from 1.25 to 2.51, and relative tapped bulk densities from 1.12 to 2.23 and they increase linearly with severity with R2 ranging between 0.96 – 0.99 (Figure 2.9). Particles were found to be smaller with increasing steam explosion pretreatment severity from sieving analysis. The geometric mean diameter of the wood particles decreased from 0.83 mm to 0.74 mm with increasing steam explosion pretreatment severity from 3.64 to 4.53. The steam pressure is a function of reaction temperature. Therefore, the steam pressure at the same temperature is independent of the treatment time. However, it appears that steaming time has an effect on particle size reduction. Bulk density of the powders increased from 126 kg/m3 to 221 kg/m3 with increasing the steam explosion pretreatment severity up to 4.23. Though the bulk density of the samples treated at severity of 4.53 decreased from 221 kg/m3 to 217 kg/m3, its bulk density should be higher than that at the severity of 4.23 due to large variability during measurement. In general, the decrease in geometric mean diameter of the steam exploded particles with treatment severity is directly proportional to the increase in bulk density of the steam exploded particles with increasing treatment. A multilinear regression model for the relative tapped bulk density with R2=0.91 was created from the factors of temperature (T), time (t), particle size (d), moisture  47  content (m), and the interaction terms T×d, T×m, and t×d (Table 2.8). The absolute value of the scaled and centered coefficients was, by far, highest for temperature (T), thus being the most influential factor within the design. Temperature was positively correlated to the relative tapped bulk density. Table 2.9 lists the coefficient of the models of bulk density, tapped density and geometric mean diameter with respect to steam explosion processing conditions and feedstock parameters. They all were well correlated with R2>=0.8. There were no significant differences between mean values of particle density for any of the factors. Particle density values varied between 1.387 g/cm3 and 1.426 g/cm3. Inter particle porosity was calculated according to equation (2-2) – equation (2-4) using the measured values of both bulk density and specific density. In general, the porosity decreased with increasing severity of steam explosion. In particular, the porosity of the small particle with 10% moisture content showed the largest drop of porosity. It decreased from 0.91 to 0.79 regardless their geometric mean diameter changes from 0.38 to 0.28 mm were not the largest.  2.3.6 Particle Morphology Analysis SEM micrographs showed that the steam exploded wood particles treated at the most severity were with more fragments compared to the untreated wood particles under the same magnification (Figure 2.10). Compared to the data from scanner imaging analysis, it also agreed that both particle length and width were shortened after steam explosion. The aspect ratio of each steam exploded particle decreased after the steam explosion treatment. The particle shape changed from angular shape to a more spherical shape with increasing the steam explosion pretreatment. This increases the packing and fluidity of the particles. Figure 2.11 shows the surface of the particles with and without steam explosion under field emission scanning electron microscope at the same magnification. The untreated particle surface is smooth after grinding by hammermill. In contrast, the particle surface of the steam exploded wood particle is rough. The roughness of the particle is caused by the detachment of the surface layer of the fragments of the wood  48  fiber. This detachment process continues by layer. This suggests that the fragments formation is caused by the high pressure steam explosion. The rapid escape of steam inside the internal pores of the particles detaches the particles surface during the rapid decompression. Pores appeared on the steam exploded fiber surface. This is caused by the rupture of the fiber structure by the sudden decompression. The surface pores affect the biomass particle surface characteristics and the measured burning area under diffusion limited conditions (Bayless et al., 1997). High surface porosity increases the surface irregularity, thus enhancing the biomass particle fragmentation during burn off. This may influence the ash formation and burning rate of the steam exploded wood particles.  2.3.7 Drying Kinetics of Untreated and Steam Treated Particles Figure 2.12 shows the drying kinetics of untreated Douglas Fir and steam treated Douglas Fir particles at 103oC. The steam treated Douglas Fir particles at 220oC and 10 min gave the fastest drying rate, followed by the untreated particles. That may be contributed by small particle size, the increase in hydrophobicity of the pseudo-lignin and open pore structure of the most severely steam treated sample. The other treated materials showed a slightly slower drying kinetics rate (Table 2.10). This may be contributed by the presence of monosugars that provide available sites for forming hydrogen bonding between hydroxyl group and free water molecules. More heat is required to supply at the same drying temperature of 50oC in order to overcome the hydrogen bonding. Therefore, this results in slower drying rate constant for the less severely treated samples.  2.3.8 Equilibrium Moisture Content of Steam Treated Particles Moisture adsorption isotherms of the untreated and pretreated 1.7 mm samples treated at 200 and 220⁰C for 5 and 10 minutes show that the equilibrium moisture content is lowered with increasing treatment temperature and time. Figure 2.13 shows that the treatment temperature is more influential than treatment duration on equilibrium moisture content (EMC). The moisture content of the untreated sample increased from zero to  49  16.57% at the relative humidity (RH) of 84%. The moisture content of the treated samples at the RH of 84% were 13.21%, 12.62%, 10.47% and 9.62%, for the severities of 3.64, 3.94, 4.23 and 4.53, respectively. These results are in agreement with earlier data obtained by Bruno et al. (2007) at 190oC, who reported that the equilibrium moisture content of steam treated pine wood at 190oC for 2 hours decreased from 8.3% to 5.4% at 35% RH. The GAB equation (2-6) was found to have a good fit to the moisture adsorption data for the untreated and steam treated Douglas Fir particles. The estimated parameters were summarized in Table 2.11. Negro et al., (2003) stated that the reduction in moisture sorption is contributed by polymerization of condensed substances that results in high hydrophobicity. The hygroscopicity of the wood components is mainly due to the presence of hydroxyl (–OH) groups of hemicellulose and cellulose. These hydroxyl groups provide active sites for water molecules to form hydrogen bonds between them. Hemicellulose is usually removed completely during high severity steam treatment and as a result the number of available hydroxyl groups is decreased. Negro et al. (2003) also explained that the increased hydrophobicity of steam exploded particles can be correlated to the decrease of O/C ratio determined from elemental analysis. Less oxygen also reduces the availability of hydroxyl groups after a hydrothermal treatment. Moreover, lignin is less likely to possess free hydroxyl groups compared to cellulose and hemicellulose, which are regarded as hydrophobic. From our experimental results, the increase in lignin content was correlated with the increased hydrophobicity of the samples. Therefore, it appears that both the reduction in the number of hydroxyl sites (hemicelluloses) and/or absence of hydroxyl sites (lignin) contribute to the improved hydrophobicity.  2.4 Conclusions The effects of steam explosion processing conditions and feedstock parameters on physical and chemical properties of woody biomass were studied. The following conclusions could be derived from this study.  50  1. The moisture content of the samples after steam explosion pretreatment ranged between 14.5% and 32.2%. The mean moisture content of samples treated at 220oC was significantly higher than that of samples treated at 200oC. 2. The solid yield of samples varied between 51% and 84%. The mean value for solid yield was significantly (α=0.05) lower for samples treated at 220oC, compared to samples treated at 200oC. 3. The pressure of steam before the samples were released from high pressure steam treatment chamber to ambient (i.e. explosion pressure) ranged between 1170 – 1320 kPa for samples at 200oC and between 1730 – 1940 kPa for samples treated at 220oC, respectively. The steam pressure at pressure release point was modeled with multilinear regression with time (t [s]) and temperature (T [oC]) with R2=0.85, and a RMSECV value of 0.12 Pa. 4. The relative geometric mean diameter of samples ground with 3.175 mm screen size were modeled with R2= 0.87 from the factors of temperature, time, and moisture content. Temperature had a strong negative correlation, followed by moisture content (positive correlation) and time (negative correlation). However, it is difficult to create a multilinear regression model with R2>=0.85 for relative geometric mean diameter for all samples included. The very small particles from samples ground with 1.6875 mm screen stick to larger particles or to sieves during sieving, and thus hard to make consistent results. 5. The fragmentation of the particles could be deduced by the formation of two peak regions with increasing treatment severity from the normal distribution graph, i.e. fine particle and larger particle. It was speculated that the fine particles were fragmented out from the surface of the wood particles due to explosion effect. This gradually formed two groups of particles. 6. Relative bulk densities of steam treated samples ranged from 1.25 to 2.51, and relative tapped bulk densities from 1.12 to 2.23. A multilinear regression model for relative tapped bulk density with R2=0.91 was created from the factors of temperature (T), time (t), particle size (d), moisture content (m), and the interaction terms T×d, T×m, and t×d. The absolute value of the scaled and centered coefficients was, by far, highest for temperature (T), thus being the most  51  influential factor within the design, Temperature was positively correlated to the relative tapped bulk density. 7. Steam explosion brought a size reduction effect on wood particles and investigated by both sieving and scanner imaging analysis. Particle size and the aspect ratio of steam exploded wood particles decreased with increasing the steam explosion pretreatment severities. 8. Fragmentation of large particles during steam explosion was found by both analysis techniques. This is caused by devolatilization at high pressure of steaming. Swelling of particles with the expansion of the release of volatiles during steam was also found at a severity of 3.94. 9. Particle shape changed from angular shape to a lower energy state of spherical shape. This explains the increase in bulk density after steam explosion pretreatment. 10. Particle surface was found to be rougher after steam explosion pretreatment by scanning electron micrographs. The rapid escape of steam inside the internal pores of the particles burst the particles surface during the rapid decompression. This led to the detachment of the surface layer and formed the fragments of the wood fiber. Surface pores were found on the steam exploded wood fibers. This may enhance the combustion performance of steam exploded wood particles due to the increase of surface irregularity. 11. Steam treated particles showed a reduced moisture adsorption compared to the untreated particles. The moisture adsorption resistance of the steam treated samples increased with the treatment severity. The temperature of the steam treatment was more influential on improving the moisture sorption resistance than the treatment time. All samples moisture sorption behavior was well fitted by the GAB model.  52  Tables Table 2.1 List of Valves in the Experiment Unit Unit  Code  Operation mode  Ball valve  B-1  Electric actuated - Computer control  Ball valve  B-2  Electric actuated - switch  Ball valve  B-3  Function Close: Maintain steam in the reactor Open: Allow rapid decompression of biomass and steam and extrude out to the collection tank Close: Allow steam treatment in steam boiler Open: Allow saturated steam to pass to the reactor Close: Always close if not purging the reactor  Manual Open: Open if purging Close: Always close during steam generation and water loading in the boiler  Ball valve  B-4  Manual Open: Open if after small amount of water loading to remove trapped air Close: Always closed if not loading the water  Ball valve  Ball valve  B-5  B-6  Manual  Manual  Open: Facilitate the water loading into the boiler Close: Always closed during water loading and steam generation Open: Open only when drainage Close: Always closed  Ball valve  B-7  Manual Open: Open only for gas sampling  Pressure relief valve  PS-1  Automatic  Threshold pressure pre-set value: 750 PSI  53  Table 2.2 List of Thermocouples and Pressure Transducers in the Experiment Unit Unit  Code  Type 1/16” diameter, 36” length, K-type, Ceramic end  Thermocouple  T1  Thermocouple  T2  Surface mounted type, Ktype, Operation temperature maximum 350oC  Thermocouple  T3  Surface mounted type, Ktype, Operation temperature maximum 350oC  Thermocouple  T4  Pressure transducer  P1  Pressure transducer  P2  1/16” diameter, 18” length, K-type ¼” MPT, Digital, 0 – 1000 psi ¼” MPT, Digital, 0 – 1000 psi  Function Measure the temperature of the steam inside the generator Measure the surface temperature of the steam-line in the generator section and provide feedback control to maintain the steam-line temperature Measure the surface temperature of the steam-line in the reactor section and provide feedback control to maintain the steam-line temperature Measure the reaction temperature inside the reactor Measure the pressure of the steam generated in generator Measure the pressure of the steam in the reactor during steaming  Table 2.3 Factorial Design of Experiment consisting of Four Factors with Two Levels Temperature (oC) T Duration (min), t Geometric mean size (mm), d(gw) Moisture content (%) w.b., m  Low (-1) 200 5 0.4 10  High (+1) 220 10 0.9 15  54  Table 2.4 Solid Yield, Moisture Content and TGA Analysis of Untreated and Pretreated Feedstock Steam explosion conditions and feedstock variables Temp (oC)  Time (min)  Severity  Particle size (mm)  0.9 0.9 0.4 0.4  Untreated  0.9 5  3.64 0.4  200 0.9 10  3.94 0.4 0.9  5  4.23 0.4  220 0.9 10  4.53 0.4  1  MC (w.b.) (%)  10 15 10 15 10 15 10 15 10 15 10 15 10 15 10 15 10 15 10 15  Product characterization  TGA analysis  Moisture content1 (w.b.) (%)  Solid yield1 (%)  Avg  SD  Avg  SD  11.4 15.2 10.7 16.0 15.5 18.5 14.5 22.4 18.5 18.6 19.8 19.8 21.8 18.6 22.0 25.7 20.8 24.9 32.2 30.9  0.1 0.3 0.5 0.6 0.8 4.2 0.4 0.9 0.3 5.0 3.1 3.2 3.0 4.4 3.9 9.6 10.0 2.1 2.1 3.5  100  --  100  --  84.1 80.3 69.8 70.2 79.1 75.0 68.9 69.0 75.9 79.8 68.2 64.3 75.9 72.9 73.9 51.2  2.2 3.1 8.0 3.6 0.6 1.0 4.3 8.9 3.9 3.0 6.1 6.2 3.1 7.3 8.5 7.8  Fixed Carbon (%)  Volatiles (%)  Ash (%)  14.7  85.3  2.7  14.4 14.8 17.0 16.9 17.0 13.5 15.8 17.7 18.3 22.2 19.3 20.9 20.6 21.8 21.5 22.5 20.8  85.6 85.2 83.0 83.1 83.0 86.5 84.3 82.3 81.7 77.8 80.7 79.1 79.4 78.2 78.6 77.5 79.2  3.1 2.1 2.1 3.2 1.3 2.8 2.1 2.2 0.2 1.9 2.5 2.5 1.7 1.7 1.8 2.0 1.5  number of measurement n=3  55  Table 2.5 Size Reduction Effect of Steam Explosion at Different Processing Conditions and Feedstock Variables (n = 3) Pressure (MPa)  Steam explosion conditions and feedstock variables Temperature (oC)  Time (min)  Severity  Untreated  Particle size (mm) 0.9 0.4 0.9  5  3.64 0.4  200 0.9 10  3.94 0.4 0.9  5  4.23 0.4  220 0.9 10  4.53 0.4  Moisture content (w.b.) (%) 15 15 10 15 10 15 10 15 10 15 10 15 10 15 10 15 10 15  Geometric mean diameter (mm)  Average  SD  Average  SD  --1.17 1.18 1.24 1.21 1.18 1.31 1.24 1.32 1.73 1.81 1.78 1.75 1.91 1.84 1.94 1.92  --0.07 0.03 0.02 0.15 0.03 0.06 0.10 0.07 0.12 0.11 0.02 0.08 0.07 0.15 0.14 0.10  0.90 0.42 0.83 0.92 0.45 0.44 0.81 0.86 0.40 0.45 0.74 0.82 0.38 0.44 0.69 0.78 0.35 0.37  0.06 0.03 0.05 0.01 0.01 0.01 0.00 0.03 0.02 0.05 0.02 0.03 0.02 0.07 0.03 0.04 0.04 0.04  Table 2.6 Summary of Imaging Analysis of Particle Length and Width with and without Steam Explosion Pretreatment at Different Treatment Severities Treatment  Untreated  3.64  3.94  4.23  4.53  Trials 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3  Number of particles 131 108 128 137 302 195 179 329 205 297 367 213 196 204 217  Length (mm) Average SD 3.52 1.63 4.22 1.91 3.14 1.38 3.67 1.47 2.27 1.19 3.42 1.75 3.41 1.52 3.10 1.61 3.40 1.78 3.25 1.54 2.45 1.23 3.04 1.43 2.59 1.30 2.32 1.07 2.16 1.00  Width (mm) Average 0.9 1.12 0.89 0.93 0.62 0.85 0.91 0.84 0.91 0.85 0.68 0.78 0.71 0.70 0.66  Aspect ratio SD 0.44 0.49 0.52 0.44 0.37 0.46 0.46 0.42 0.48 0.40 0.35 0.37 0.31 0.33 0.32  Average 4.42 4.22 4.23 4.59 4.26 4.74 4.29 4.11 4.22 4.38 4.17 4.52 4.11 3.85 3.88  SD 2.12 2.11 2.31 2.38 2.40 2.83 2.22 1.97 2.08 2.41 2.16 2.73 2.66 2.44 2.46  56  Table 2.7 Bulk, Tapped and Specific Density of Untreated and Steam Treated Samples at Different Severities. (n=5) Steam explosion conditions and feedstock variables Temp (oC)  Time (min)  Severity  Untreated  Particle size (mm)  Porosity  Avg  SD  Avg  SD  Avg  SD  Avg  SD  --  144.1  --  1.37  --  0.91  --  0.4  10  118.3  --  151.4  --  1.43  --  0.92  --  10  154.2  22.0  169.8  22.1  1.40  0.01  0.89  0.02  15  154.7  17.8  169.1  17.2  1.40  0.01  0.89  0.01  10  168.9  19.6  187.4  21.6  1.43  0.00  0.88  0.01  15  159.6  11.0  185.9  11.1  1.42  0.01  0.89  0.01  10  184.7  11.4  197.4  13.3  1.41  0.02  0.87  0.01  15  177.7  10.3  189.8  9.7  1.41  0.01  0.87  0.01  10  206.7  1.5  231.3  2.0  1.41  0.03  0.85  0.00  15  183.3  11.8  209.4  8.7  1.38  0.01  0.87  0.01  10  220.7  2.9  237.6  2.3  1.39  0.03  0.84  0.00  15  197.6  19.9  214.3  17.2  1.39  0.02  0.86  0.01  10  252.6  9.3  289.6  11.3  1.40  0.03  0.82  0.01  15  214.7  11.3  246.2  18.3  1.43  0.01  0.85  0.01  10  216.8  26.3  240.1  24.3  1.39  0.02  0.84  0.02  15  210.9  9.5  228.8  18.1  1.39  0.01  0.85  0.01  10  278.2  16.6  322.0  13.7  1.41  0.02  0.80  0.01  15  251.9  9.7  289.0  10.3  1.39  0.02  0.82  0.01  3.64  0.9 3.94 0.4 0.9 4.23 0.4 220 0.9 10  Particle density (g/cm3)  125.9  200  5  Tapped density (kg/m3)  10  0.4  10  Bulk density (kg/m3)  0.9  0.9 5  MC (w.b.) (%)  Density measurement  4.53 0.4  57  Table 2.8 ANOVA Results of Bulk, Tapped Density and Geometric Mean Diameter with respect to Steam Explosion Conditions and Materials Parameters. Model components  Bulk density (kg/m3) 4  Tapped density (kg/m3) 4  Geometric mean diameter (mm) 3  Number of observations  48  48  24  R  0.9140  0.9140  0.8692  F-value  22.65  22.63  15.19  RMSEP/RMSECV  0.119444  0.103584  0.044047  Responses  2  Coeff Var.  7.2  6.8  5.9  mean  1.660037  1.523796  0.742633  F-value  Prob.  F-value  Prob.  F-value  Prob.  31.67  <0.0001  34.05  <0.0001  10.4  0.0029  Temperature, C (T)  184.48  <0.0001  220.35  <0.0001  43.15  <0.0001  Particle size, mm (d) Moisture content, % wet basis (m) txT  79.03  <0.0001  40.61  <0.0001  2.03  0.1643  15.97  0.0004  15.97  0.0004  13.59  0.0008  1.52  0.2263  0.49  0.4909  0.44  0.5129  Duration, s (t) o  txd  4  0.0541  4.17  0.0495  0.79  0.3794  txm  0.04  0.8403  0.02  0.902  0.02  0.8869  Txd  12.42  0.0013  12.54  0.0012  0.07  0.7975  Txm  2.61  0.1162  4.97  0.0329  0.87  0.357  Pxm  3.77  0.0611  2.28  0.1408  1.48  0.2332  58  Table 2.9 Multiple Linear Regression Results of Bulk Density, Tapped Density and Geometric Mean Diameter with respect to Steam Explosion Conditions and Materials Parameters Relative bulk density1  Relative tapped density1  4  4  Relative geometric mean diameter1,2 3  24  48  48  24  0.8577  0.8301  0.8320  0.8420  0.8301  35.76  34.44  32.57  41.6  30.44  32.57  17.5529  18.6294  0.0449  0.1456  0.1255  0.0449  Constant  -583.199  -1192.72  2.1086  -5.0133  -7.819  2.1086  Time, s (t)  0.1148  0.2444  -0.0003  0.0009  0.0016  -0.0003  Temperature, C (T)  3.7368  6.5072  -0.0069  0.0324  0.0425  -0.0069  Particle size, mm (d) Moisture content, % wet basis (m) txT  315.797  601.368  --  2.9054  3.7328  --  -304.238  2580.5  1.9655  -2.5337  17.0999  1.9655  --  --  --  --  --  --  txd  --  -0.1758  --  --  -0.0011  --  txm  --  --  --  --  --  --  Txd  -1.7032  -2.7996  --  -0.0164  -0.0168  --  Txm  --  -13.3005  --  --  -0.0895  --  Responses  Bulk density (kg/m3)  Tapped density (kg/m3)  Model components  4  4  Geometric mean diameter (mm)2 3  Number of observations  48  48  R  0.8098  F-value RMSEP/RMSECV  2  Coefficients  o  1  Relative responses means the measured responses under different treatment conditions over the measured responses of the untreated samples 2 Geometric mean diameter of large particles ground with 3.175 mm screen under different conditions  Table 2.10 Estimated Parameters for Drying Kinetics of Untreated and Steam Treated Douglas Fir Particles  Untreated  Drying kinetic constant K (min-1) 0.012  200oC, 5 min  0.010  0.095  0.982  0.008  0.106  0.976  0.009  0.131  0.963  0.014  0.223  0.931  o  200 C, 10 min o  220 C, 5 min o  220 C, 10 min  Intercept c  Goodness of fit R2  0.101  0.982  59  Table 2.11 Estimated Parameters for the GAB Equation fitted to Untreated and Steam Treated Equilibrium Moisture Data for Douglas Fir Particles Untreated o  200 C, 5 min o  200 C, 10 min o  220 C, 5 min o  220 C, 10 min #  Monolayer value (M0) 7.69  K  C  0.715  5.314  Goodness of fit R2 0.996  5.66  0.746  5.578  0.994  4.96  0.777  5.203  0.996  3.85  0.799  4.806  0.994  5.76  0.676  2.059  0.995  o  Measurement taken at temperature 21.3 ± 0.7 C  60  Figures  Figure 2.1 Closed System Steam Explosion Unit at the UBC Clean Energy Research Center  61  Figure 2.2 Process Flowsheet of the Closed System Steam Explosion Unit at the UBC Clean Energy Research Center (B: Ball Valve, PS: Pressure Relief Valve, T: Thermocouple, P: Digital Pressure Transducer)  62  300  250  o  Temperature ( C)  200  150  100  T1 T2 T3 T4  50  0 0  1000  2000  3000  4000  5000  Time (s)  Figure 2.3 A Typical Plot of Temperature for the Steam Explosion Experiment at 200oC for 5 minute. T1 is the Temperature of the Steam in Steam Boiler, T2 is the Surface Temperature of the Steamline before the Ball Valve (B-2), T3 is the Surface Temperature of the Steam Line after the Ball Valve 2 (B-2), T4 is the Temperature of the Reaction in the Reactor.  63  Figure 2.4 Treatment Temperature Profile (T4) of Biomass Steam Explosion Experiment at 200oC for 5 minute (n=3)  Figure 2.5 A Typical Plot of Pressure vs. Time during Steam Treatment at 200oC for 5 minute and at the Moment of Pressure Release (n=3).  64  Figure 2.6 Physical Appearance of Steam Exploded Particles Treated at Different Steam Explosion Conditions. From Left to Right: (a.) Untreated Douglas Fir, (b.) 200oC, 5 min, (c.) 200oC, 10 min, (d.) 220oC, 5 min, (e.) 220oC, 10 min.  Figure 2.7 Solid Yield of Douglas Fir Ground Particles at Different Severities of Steam Explosion Pretreatment  65  Figure 2.8 Particle Size Distribution of Untreated and Pretreated Douglas Fir Sawdust at Different Severities  Figure 2.9 Relative Tapped Density vs. Severity of Steam Explosion for Samples Treated at Different Severity of Steam Explosion Pretreatment  66  a Figure 2.10 SEM Pictures of (a.) Untreated Ground Particle by Hammermill with 1.7 mm Screen Opening and (b.) Steam Treated Particles at 4.53 Severity with x35 Magnification  a  b  Figure 2.11 SEM Pictures of (a.) Untreated Ground Particle by Hammermill with 1.7 mm Screen Opening and (b.) Steam Exploded Particles at 4.53 Severity with x500 Magnification  67  Figure 2.12 Moisture Content vs. Time for the Untreated Douglas Fir Particles and Steam Treated Particles at Oven Temperature 103oC.  Figure 2.13 Moisture Sorption Isotherms for Untreated Douglas Fir Particles and Steam Treated Particles at Different Temperature and Treatment Time (Solid Lines show GAB Model). 68  Chapter 3 Pelletization Experiments 3.1  Introduction Wood pellet is a compressed form of ground biomass. The cylindrical pellet is 6.4  mm in diameter. The length may vary from 6 to 24 mm but usually it averages around 12 mm. The moisture content of pellets produced in British Columbia is about 6% (wet mass basis). Wood pellet processing and export have experienced a rapid growth in recent years. In 2009, Canada produced around 1.7 million dry metric tons of wood pellets, 75% of which was exported to Europe for heat and power production and the remainder to USA and Japan (Swaan, 2009). Some of the pellets going to USA are used for animal bedding. The pellets industry is currently expanding the production capacity with about 25% per year and U.S. is also experiencing an even steeper growth in the production capacity and export of wood pellets. The manufacturing process of wood pellets consists of drying, size reduction, pelletizing, cooling and packaging (Mani et al., 2006a). The traditional feedstock for making pellets has been from sawdust and shavings. High lignin content in the feedstock provides adequate binding to make durable pellets. The availability of feedstock such as mill residues has been decreasing recently due to a slow housing market coupled with an increasing demand for pellets. Processors are grinding logging residues to produce chips as a feedstock for pellets. The pellets from logging residues are more durable than pellets from mill residues such as sawdust and shavings (Lehtikangas, 2001). A less durable pellet tends to disintegrate easily into fines during handling and storage. In addition to potential explosions, fines inhalation is an occupational health problem (Aleksandra et al., 2006). Agricultural grasses and mixed logging residues especially those of hardwood are difficult to form into durable pellets. A low content of lignin and unavailability of other potential natural binders (starches, proteins) within the lignocellulose matrix may require additional binders to assist in forming durable pellets (Sokhansanj et al., 2005). Donohoe et al. (2008) provided experimental evidences supporting the idea that thermochemical pretreatments reaching temperatures above the range for lignin phase transition cause lignin to coalesce into larger molten bodies that migrate within and out of the cell wall,  69  and can redeposit on the surface of plant cell walls. The reformed lignin provided solid bridges among particles after compression and cooling. Conventional wood pellets tend to absorb moisture from the surrounding humid air. Moistened pellets tend to disintegrate and also provide an ideal environment for microbial and biochemical activities (Lehtikangas et al., 2000; Rupar et al., 2005). Previous published literature has shown that wood products made from steam exploded ground wood become more dimensionally stable and less hygroscopic after steam treatment (Angles et al., 2001; Bruno et al., 2007). Angles et al. (2001) demonstrated that the strength of the compressed wood panels made from steam exploded wood particles increased as the severity of the pre-treatment increased but up to a point beyond which no gain in strength was observed. Pellets made from steam exploded poplar resisted higher breaking forces than untreated pellets (Shaw et al., 2009). While an increase in percent lignin content and a restructure of lignin after steam explosion may contribute to the increase in the breaking strength of wood pellets, the extractives prevent a close contact between bonding sites of the lignocellulose particles thus reducing the pellet strength (Nielsen et al., 2010). However, experimental data suggest that extractives may act as plasticizers and lubricants and thereby reducing the energy requirements for making pellets (Nielsen et al., 2010). Steam exploded canola, oat and wheat straw pellets were reported with significantly higher specific energy consumption during  pelletization  compared to the untreated sample, primarily due to higher extrusion specific energy (Adapa et al., 2010b; Adapa et al., 2010c). The objectives of this research were to quantify the quality attributes of pellets made from steam treated Canadian West Coast Douglas Fir (Pseudotsuga menziesii) and to quantify energy input to form the pellets from treated and untreated batches of that material. The measured quality characteristics included mechanical strength, moisture adsorption, and high heating value. The energy input was calculated from the compressive forces applied in a single piston-cylinder press arrangement to compact the loose feedstock particles to dense pellets.  70  3.2 Materials and Methods 3.2.1 Materials A detailed description of equipment for steam explosion and the test procedures were described in section 3.2.1 – 3.2.3. Briefly, the pulp quality wood chips produced from Canadian West Coast Douglas Fir (Pseudotsuga menziesii) were ground in a hammer mill. The perforations in the grinder screen were round holes with 1.6 mm in diameter. The ground biomass at about 10% m.c. was steam treated in a 1 litre high pressure steam chamber. Saturated steam with a temperature of 200oC or 220oC with a corresponding pressure of 1.6 or 2.4 MPa was applied. Treatment time was 5 or 10 minutes. Following the steam treatment, the ground biomass was decompressed by releasing it into a chamber open to atmospheric conditions. The mass of recovered solids, moisture content, density, particles size distribution and chemical composition of the exploded biomass were quantified in Chapter 2. In preparation for pelletization, batches of the ground sample before and after treatment were dried in a convection oven set at 50oC. The sample was removed from the oven when the average moisture content of the biomass was about 10% (wet mass basis). The dried samples were stored in a sealed plastic container at 4oC until pelletization.  3.2.2 Pelletization Single pellets were made in a piston-cylinder unit (Figure 3.1). The assembly consisted of 3 parts: (1) a piston with 6.30 mm in diameter and 90 mm in length, (2) a cylinder 6.35 mm inside diameter and 70 mm long, (3) a heating tape wrapped around the outer body of the cylinder (Figure 3.2). A MTI (Measurement Technology Inc., Roswell, GA) equipment was used to force piston into the cylinder. The bottom of the cylinder was open but could be closed by placing a removable block under the cylinder block during compression. The height of the block was 25.4 mm. A cylindrical rod with 6.30 mm in diameter (the same diameter of the piston) and 12 mm in length was placed in the cylinder. The rod rested on the removable block. In preparation for making a pellet, the electric power to the heater was turned on to heat the body of the cylinder to about 90oC.  71  A temperature controller maintained the set temperature. The hole in the cylinder was filled with approximately 0.8 g of ground biomass using a spatula. The MTI was preset to a maximum downward force of 4,000 N. The downward displacement speed was set at 6.67 mm/min. The movable cross head of the MTI was brought down manually to align the piston and the cylinder prior to applying force. The bulk biomass in the cylinder was compressed to the maximum force of 4,000 N and held for 30 s to arrest the spring back effect. Next, the lower block was removed. The cross head was reactivated to move downward at a speed of 13.34 mm/min. The piston pushed the formed pellets and the spacer out of the piston through the bottom of the cylinder. The force – displacement data were logged during the entire cycle of compression to form a pellet and expulsion of the pellet out of the die. The pellet was cooled to the room temperature and stored inside a sealed glass bottle for further measurements. In most cases 15 pellets were made from each steam treated and untreated biomass sample.  3.2.3 Pellet Density and Specific Density The mass, length and diameter of each pellet were measured immediately after each removal from the die. The pellets were stored inside a sealed glass bottle for one week. After one week, length and diameter were measured to evaluate the percentage expansion in diametric and longitudinal direction. Specific density was determined by measuring the total pore volume by the adsorption of nitrogen into void spaces of the biomass powders with a Quantachrome Multipycnometer (Quantachrome, Boyton Beach, FL, USA). The measurement was done by measuring the pressure difference when a known quantity of pressurized nitrogen flows from a reference volume into a sample cell with samples. Particle volume measurements were repeated five times for each sample, for determination of an average, and a reference volume of 29.42 cm3 was used in all measurements. Description of the method by equation (2-2) was in Chapter 2.  72  3.2.4 Breakage Test To measure breakage, a pellet was placed between two anvils under the MTI cross head. The compression was diametrical. The maximum load to break a pellet was recorded. The Meyer hardness (HM) is defined as the applied force divided by the projected indentation area and was calculated according to the following equation (Tabil et al., 2002): HM   F  Dh  h 2      (3-1) 2  where HM is the Meyer hardness (N/mm ) and h is the indentation depth (mm), D is the initial diameter of a pellet cross section (mm), and F is the maximum force when the pellet is crushed (N).  3.2.5 Chemical Composition Moisture contents of the samples before and after pelletization were analyzed according to ASABE S358.2 standard (ASABE S358.2, 2010). Roughly 1 g of either ground sawdust or a whole pellet was used for moisture content measurement. Triplicate samples were oven dried at 103oC for 24 h. The ash content was determined according to NREL method as the percentage of residue remaining after dry oxidation (oxidation at 550oC to 600oC) (NREL Standard, TP-510-42622, 2005). Ash content analysis was made in three replicates. Extractive contents were analyzed using modified acetone extraction method (TAPPI Standard, T280 pm-99, 1999). Measurements were made in duplicates. Klason lignin contents of extractive-free samples were measured according to NREL method (NREL Standard, TP-510-42618, 2005). Lignin was fractionated into acidinsoluble and acid-soluble material. The acid-insoluble fraction may have also included ash and protein, which were accounted for during gravimetric analysis. The acid-soluble lignin was measured by an Ultraspec 1000 UV-Vis spectroscopy (Pharmacia Biotech, Piscataway, NJ, USA). Lignin content measurements were made in triplicates. The high heating values of the pellets were determined by the oxygen bomb calorimeter (Parr 6100) from two replicates (ASTM Standard, D 2015-96, 1998).  73  3.2.6 X-ray Diffraction A Bruker D8 Advance powder X-ray diffractometer in Bragg-Brentano configuration with copper radiation and a diffracted beam graphite monochromator was used to assess the degree of crystallinity of pellets. The X-ray generator was set to 40 kV and 40 mA and data were collected from 5-90° 2-theta using a step of 0.04° at 1.5 second per step. The samples were rotated during data collection. The cellulose 1 alpha structure was identified (Nishiyama et al., 2002). The amorphous peak was constrained to be centered between 17-21 deg 2-theta. The percent crystallinity was determined by modeling with the software Topas 4.2 from Bruker. Two replicates of each sample were measured.  3.2.7 Moisture Sorption Rate Test The moisture adsorption rate of pellets was measured in a humidity chamber (ESPEC CORP, LHU-113, Japan). The chamber was set at 30oC and 90% relative humidity. Prior to an adsorption test, pellets were bone dried in a convection oven at 103oC for 24 h. A pair of pellets were placed in a petri glass dish and placed in the chamber for a minimum of 5 h. The weight of the sample was measured every 10 minutes for the first hour followed by every 30 minutes for the next 4 h. Weight measurements were made on a ALC-80.4 (Acculab, Edgewood, NY) analytical balance with 0.1 mg precision. The petri dish was covered with a glass cap during weighing to prevent moisture loss. Three replicates were measured for each set of pellets. The kinetics of moisture sorption is represented using the ASABE S448.1 formulation for thin layer drying (ASABE Standard, S448.1, 2006), M  Me  e kt Mi  Me  (3-2)  where M is the instantaneous moisture content (decimal, dry basis), Me is the equilibrium moisture content (decimal, dry basis), and Mi is the initial moisture (decimal, dry basis). The coefficient k is an adsorption constant (rate constant) and t is the exposure time (min).  74  3.2.8 Scanning Electron Microscope of Cross Section of Pellet The fibrous structure of the cross section for each set of wood pellet was studied using a field emission scanning electron microscopy (FESEM) to observe the changes as a result of the steam explosion process. The wood pellet was immersed in liquid nitrogen for 10 minute and fractured into two portions while frozen. The surface created after the fracture was mounted on the top of the specimen stubs and coated with gold under vacuum. All photographs were taken at 10 to 20 kV accelerating voltage by using a field emission scanning electron microscope Hitachi S4700 (Hitachi, Japan) in the University of British Columbia bio-imaging facility.  3.3 Results and Discussion 3.3.1 Physical Dimensions and Density of Pellets Table 3.1 lists geometric mean diameter and moisture content of untreated and treated feedstock biomass particles and the density of pellets made from these particles. Except for 200oC-5 min treatment, the average particle size of treated particles was smaller than the average size of untreated particles (0.42 mm for untreated vs. 0.35 mm for the 220oC-10 min treatment). The moisture content of the treated and untreated particles ranged from 8.6% to 11.8% (wet mass basis). The mass of individual pellets made from treated and untreated particles averaged 0.79 g with a small standard deviation (<0.01 g) except for the 200oC-5 min treatment in which the mass of a pellet was about 0.1 g less than other tests and the standard deviation was 0.03 g. The smaller mass for this treatment was incidental. The pellet diameter averaged about 6.62 mm with a relatively small standard deviation (~0.02 mm). Similar to the diameter, the length did not vary much from 20.96 to 21.80 mm except for the lower mass pellet for treatment at 200oC-5 min for which at a length of 17.58 mm, the pellet was shorter than other treatments. Table 3.1 lists the bulk density of loosely packed feedstock particles prior to pelletization. The bulk density ranged from 0.118 g/cm3 for the untreated to 0.278 g/cm3 for the treated biomass. The initial pellet density measured from mass and volume of a single pellet was 1.09 g/ cm3 for the untreated biomass. The initial pellet density of pellets  75  made from steam exploded biomass increased to 1.12 g/cm3 for the samples treated at 200oC-5 min. With increasing the severity of the treatment, the initial pellet density decreased from 1.12 to 1.07 g/cm3. Similar trend was observed for the relaxed pellet density for samples measured 7 days after pellet production. The relaxed pellet density (after 7 days of rest) of the untreated pellets decreased from 1.09 g/cm3 to 1.07 g/cm3 due to a 0.56% volumetric expansion. In particular, the diametric expansion of 0.29% was higher compared to the longitudinal expansion of 0.06%. In contrast, the steam exploded pellets showed an increase of volumetric contraction after relaxation with increasing treatment severity. This observation corroborates with a previous research that steam treatment prior to compression can increase the compressibility of wood particles and significantly reduce the buildup of internal stresses during compression (Hsu et al., 1988). Stream treatment reduces the spring back of compressed wood by irreversible swelling. The reduction in longitudinal direction between -0.24% and -0.47% was higher than diametric direction between -0.12% and -0.24% for pellets made from treated samples. Figure 3.3 shows the physical appearance of the wood pellets made using the single pellet MTI equipment. The pellets were darker with increasing severity of steam treatment. We noted that the lower part of a pellet was darker than its upper part. This was probably due to an uneven temperature distribution along the length of the pellet during pelletization. The plunger was not heated and this might create a lower temperature condition towards the upper section of the cylinder. Figure 3.4 shows the cross section of the untreated and steam exploded pellets at a low magnification of 30. For the untreated pellet, particles stacked on each other after compression. Clear boundaries and gaps between individual particles were observed. In contrast, the particles in the steam exploded pellet were more closely packed with no gaps in between.  3.3.2 Force and Energy to Make Pellets Figure 3.5 is a typical plot of force vs. displacement of the piston when compressing ground feedstock to form a pellet. The applied force was initially small for a displacement of about 30-40 mm followed by a rapid increase in force over a relatively short displacement (10-15 mm). The untreated sample experienced a larger elastic 76  compression than the treated sample did. This might be due to the broken cell structures in steam treated material and thus a lower elasticity. The difference in compression curves was also evident from the higher initial bulk density of the steam exploded particles compared to the bulk density of the untreated particles (Table 3.1). The force vs. displacement in Figure 3.5 did not change monotonically. It seems that the material consistently experiences an abrupt increase in force over a small displacement at a certain force level. It is not clear at this time whether this abrupt change in force-deformation is due to particle re-orientation or other phenomenon that may be taking place during compression. Ramos et al. (2003) postulated that upon steam explosion, hemicelluloses are hydrolyzed into simple monosugars and interfacial bonding between cellulose and lignin was destroyed. The hemicelluloses acted as a viscoelastic interface between cellulose fibers and lignin in the cell wall. Therefore, the resulting material became brittle with an increased modulus of elasticity in the elastic region. Angles et al. (2001) concluded that the steam exploded wood was brittle, having microcrystalline cellulose with re-condensed lignin droplets on fiber surface as observed by SEM photos. The restructured lignin deposited on the cellulose fibers underwent plastic deformations to form solid bridges among particles. The change in forcedeformation curve might have been the sign of any of these physiochemical processes during biomass compression. Table 3.2 lists the loading conditions and energy expenditures to form and to extrude pellets. The maximum force to compress the feedstock particles ranged from 4,079 N to 4,297 N with no clear trend in the data. The standard deviations also varied with no specific trend. The energy to compact the material that was calculated from the area under the force-deformation curve ranged from 22.3 J to 40.4 J increasing with severity of steam treatment. The energy for a displacement of 20 mm to 35 mm contributed to the breakage energy for the large particles before the elastic compression. The particles brittleness increased with the severity of treatment. Therefore, the energy required to break the large piece of particles was higher. Similarly the energy to expel pellets from the cylinder ranged from 0.049 J for extruding pellets made from untreated  77  biomass to 0.235 J to extrude pellets made from the most severely treated feedstock. This phenomonen was in line with the results of steam exploded canola, oat and wheat straw pellets reported with significantly higher specific energy consumption during pelletization compared to the untreated sample, primarily due to higher extrusion specific energy (Adapa et al., 2010b; Adapa et al., 2010c). Table 3.2 also lists the solid density of pellets measured using pycnometer. This measurement eliminates the volume of pores within the pellets – albeit those pores that the measuring gas in the pycnometer can penetrate. The measured solid density of pellets ranged from 1.42 to 1.43 g/cm3 and was slightly higher than the solid density of particles at 1.4 g/cm3 (s.d. =0.02, n=35). This slight increase may reinforce our speculation that particles experienced further disintegration under compression forces. 3.3.3 Chemical Composition Table 3.3 lists the measured chemical compositions and the heating values for pellets. The moisture content of pellet made from untreated feedstock was 6.9%, the highest among all the samples. The treated pellets at all conditions had a moisture content ranging from 5.3 to 6%. Steam exploded pellets hold less moisture than the treated pellets. The ash content of the control pellet (untreated pellet) was 0.27% by weight. There was a slight increase to 0.32% for pellets made from treatment at 200oC regardless of treatment time. The ash content of the pellets made from the 220oC samples increased further to around 0.41 – 0.52 % by weight. Pelletization itself did not affect the ash content of feedstock. The high heating value of the pellets increased from 18.82 MJ/kg for pellets made from untreated feedstock to 20.09 MJ/kg for pellets made from the most severe steam treatment. The percentage of carbon increased from 48.44% to 53.09% and the percentage of hydrogen decreased from 6.23% to 5.91% from elemental analysis. This increase of C/H ratio from 7.78% to 8.98% was due to the removal of the –OH group of hemicelluose after the hydrothermal treatment (Negro et al., 2003).  78  3.3.4 Hardness Test Figure 3.6 shows a typical force displacement graph for the hardness test on pellets. The maximum breaking force and the area under the curve (i.e., the energy absorbed to break the pellet) were higher for steam treated samples than the untreated ones. Table 3.4 summarizes the hardness test results. The maximum force required to break the untreated pellet was 18 N. The maximum breaking force required to break the pellets made from steam exploded biomass increased from 26 N to 59 N increasing with the severity of treatment. The increase in total lignin content from 30% to 43% (Table 3.3) after steam explosion contributed to the increase in the breaking strength. Lignin acted as a stiffener of the cellulose microfibrils/fibrils and an increased cross linking of this polymer appears to prevent or limit relative movement of particles within the pellet matrix. This is similar to the function of lignin in virgin wood (Sweet et al., 1999; Fengel et al., 1984). Steam explosion is an acidic autohydrolysis reaction which depolymerizes hemicellulose into monosugars, e.g. xylose (Ramos et al., 2003). Additional bonding strength of steam exploded pellets may be caused by the formation of highly branched polysaccharides by reverse reactions of monosugars during densification. The average Meyer hardness increased from 1.6 N/mm2 for untreated pellets to 6.6 N/mm2 for the steam exploded pellets treated at 220oC-5 min. The control pellets (untreated pellets) with the lowest Meyer hardness may indicate the existence of more cellular structure to allow pellets to deform. This observation is also evident from the cross section SEM photo of the untreated pellet (Figure 3.4). The hemicelluloses solubilization and the resulting increase in crystalline cellulose fiber make the pellets more brittle after hydrothermal treatment. From x-ray diffraction result, the cellulose crystallinity increased from 49.5% for the pellets made from untreated feedstock to a maximum of 67.5% for pellets made from the 220oC-5 min steam exploded feedstock. The maximum breaking force, hardness modulus and Meyer hardness of the pellets made from treated biomass at 220oC-10 min decreased from the maximum value at 220oC-5 min due to the destruction of cellulose crystalline structure by prolonged steam treatment (Bhuiyan et al., 2000). The average maximum breaking force of the  79  pellets treated at 220oC-10 minutes decreased from the maximum of 59 N to 44 N. The average Meyer hardness of steam exploded pellets treated at 220 oC-10 min decreased from 6.6 N/mm2 to 5.6 N/mm2. The crystallinity of cellulose of steam exploded pellets decreased to 61% for the most severely treated sample.  3.3.5 Moisture Adsorption Figure 3.7 is a typical plot of moisture content of pellets vs. time during moisture adsorption in the humidity chamber. The vertical axis is the percentage points for moisture increase. The pellets made from the untreated sample exhibited the most hygroscopic behavior, for which the increase in moisture content was 10.3% after 5 h of exposure to humid conditions (RH = 90% at 30oC). The moisture content of the pellets made from steam exploded biomass treated at 200oC-5 min and 200oC-10 min was close to 7.7% after 5 h exposure. The moisture content of the pellets made from 220oC-5 min and 220oC-10 min treatment was 6.0%. From a pair T-test on the effect of treatment time, the effect of treatment time was negligible at 5% confidence interval. Table 3.4 lists the estimated equilibrium moisture content (Me) read off the moisture adsorption plots. The Me for pellets made from untreated biomass approached 10% whereas the value for Me dropped to 7.5 and 5.9 for pellets made from steam treated biomass. Experimental values along with the Me (and assuming Mi=0) were used to estimate the value for adsorption rate k. Table 3.4 lists estimates for k at 0.0152 min-1 for pellets from untreated feedstock. The values for pellets made from treated biomass ranged from 0.0125 to 0.0135 min-1 without a clear trend with steam treatment severity.  3.4 Conclusions Physical and chemical properties of pellets made from untreated and steam exploded ground softwood, Canadian West Coast Douglas Fir (Pseudotsuga menziesii), were investigated. It is found that pellets made from steam exploded samples required a compression energy of 40.4 J compared to 22.3 J for pellets made from untreated biomass. Pushing pellets out of the die took about 3 orders of magnitude less energy than the 80  compression energy. It takes more energy to push out the treated pellets than untreated pellets. From the hardness test, the maximum breaking force and Meyer hardness of the steam exploded pellets treated at 220oC-5 min increased by 3.3 times and 6.6 times, respectively, when compared to the hardness of pellets made from untreated feedstock. Steam exploded pellets decreased in volume from 0.03 to 0.79% after relaxation depending upon the severity of steam explosion. This is opposite to the volume expansion of 0.56% after densification for the untreated pellets. Moisture sorption rate of pellets made from steam exploded biomass was slower than the pellet made from the untreated biomass. The equilibrium moisture content of the steam exploded pellets treated at 220oC-5 min and the untreated pellets after 5 h exposure in the humidity chamber at RH = 90% at 30oC was 6.0% and 10.3%, respectively. Although pellets made from steam exploded biomass have a superior durability than pellets made from untreated biomass, this improved durability is at the expense of expending more energy to make pellets. This research must be extended to steam treatment of mixture of bark and stem wood to assess the economic value of steam treatment in improving durability of pellets.  81  Tables Table 3.1 Moisture Content and Mean Size of Particles prior to Pelletization and Particle Moisture Content, and Pellet Mass and Dimensions after Pelletization Biomass particle before pelletizing Geometric mean size dgw1 (mm)  Bulk density (g/cm3)  Moisture content 1 (%)  Mass2 (g)  Pellet density after forming g/(cm3)  Pellet density after 1 week g/(cm3)  Avg.  8.7  0.42  0.118  6.9  0.791  1.09  1.07  Std. dev.  3.0  0.03  -  0.1  0.009  0.04  0.02  Avg.  9.4  0.45  0.169  6.0  0.680  1.12  1.11  Std. dev.  0.2  0.01  0.019  0.6  0.029  0.05  0.01  Avg.  10.0  0.40  0.207  6.0  0.786  1.09  1.09  Std. dev.  0.2  0.02  0.002  0.3  0.010  0.01  0.01  Avg.  11.8  0.38  0.253  6.0  0.789  1.08  1.08  Std. dev.  0.5  0.02  0.009  0.2  0.014  0.04  0.04  Avg.  8.6  0.35  0.278  5.3  0.791  1.07  1.06  Std. dev.  0.1  0.04  0.017  0.1  0.008  0.03  0.03  Treatment  Untreated 200oC-5 min 200oC-10 min 220oC-5 min 220oC-10 min  Pelletized biomass  Moisture content 1 (%)  1  :n=3 2 : n = 15  82  Table 3.2 Forces and Input Energy to Make Pellets and Pellet Density Max force (N)  Compression energy1 (J)  Extrusion energy2 (J)  Solid density (g/cm3)  Avg.  4290  22.3  0.049  1.43  Std dev.  70  1.3  0.003  0.00  Avg.  4190  25.0  0.118  1.42  Std dev.  145  1.0  0.030  0.01  Avg.  4120  31.8  0.117  1.43  Std dev.  39  1.6  0.014  0.00  Avg.  4297  38.8  0.178  1.42  Std dev.  54  4.4  0.045  0.00  Avg.  4079  40.4  0.235  1.42  Std dev.  27  1.6  0.003  0.00  Treatment  Untreated  200oC-5 min  200oC-10 min  220oC-5 min  220oC-10 min 1  : n = 15 :n=5  2  83  Table 3.3 Chemical Composition Analysis of Pellets Made From Untreated and Treated Feedstock Heating value1 (MJ/kg)  Ash2 (%)  C (%)  H (%)  Extractives (%)1  Acid Insoluble lignin (%)2  Acid Soluble lignin (%)2  Avg.  18.82  0.27  48.44  6.23  1.42  28.49  1.60  Std dev.  0.3  0.02  --  --  0.01  3.75  0.16  Avg.  18.94  0.32  49.14  6.08  2.22  24.23  1.65  Std dev.  0.04  0.01  --  --  0.43  0.87  0.00  Avg.  19.15  0.32  50.46  6.10  3.23  28.23  1.42  Std dev.  0.04  0.01  --  --  0.17  2.37  0.08  Avg.  19.5  0.52  52.42  5.95  8.84  36.26  1.37  Std dev.  0.05  0.09  --  --  0.17  0.15  0.05  Avg.  20.09  0.41  53.09  5.91  7.36  42.14  1.09  Std dev.  0.01  0.02  --  --  10.06  1.54  0.07  Treatment  Untreated  200oC-5 min  200oC-10 min  220oC-5 min  220oC-10 min 1 2  :n=2 :n=3  84  Table 3.4 Data on Hardness Tests, Crystallinity and Moisture Adsorption Maximum breaking force1 (N)  Meyer hardness1 (N/mm2)  Cellulose crystallinity2 (%)  Equilibrium moisture content (Me %)  Adsorption rate, k (1/min)  Avg.  18.0  1.6  49.5  10.2  0.0152  Std dev.  4.0  0.2  0.7  --  --  Avg.  25.7  2.1  51.5  7.5  0.0125  Std dev.  2.1  0.3  0.7  --  --  Avg.  34.5  2.3  57.5  7.5  0.0135  Std dev.  11.9  0.9  2.1  --  --  Avg.  59.3  6.6  67.5  6.0  0.0129  Std dev.  11.1  2.1  0.7  --  --  Avg.  44.2  5.6  61.0  5.9  0.0128  Std dev.  7.6  1.8  1.4  --  --  Treatment  Untreated  200oC-5 min  200oC-10 min  220oC-5 min  220oC-10 min 1 2  :n=5 :n=2  85  Figures  Figure 3.1 The MTI Machine with Computer Control System and Temperature Controller.  86  Figure 3.2 The Piston – Cylinder Pellet Making Assembly installed in a MTI Universal Testing Equipment. Picture shows the 6.2 mm Diameter Piston, Piston Bolted to the Top Cross Head and the Die with the Heating Tape Bolted to the Bottom Plate.  87  a  b  c  d  e  Figure 3.3 Physical Appearance of Wood Pellets Made from Untreated and Treated Douglas Fir at Different Steam Explosion Conditions. From Left to Right: (a.) Untreated, (b.) 200oC-5 min, (c.) 200oC-10 min, (d.) 220oC-5 min, (e.) 220oC-10 min. Pellets were 6.62 mm in Diameter and about 18 mm in Length.  a Figure 3.4 SEM Photos of the Cross Section of Pellets Made from (a.) Untreated and from (b.) Steam Treated (220oC-10 min) Douglas Fir at Low Magnification (x30). Untreated Pellets show the Stack of Fibrous Particles. The Fibrous Structure is not Visible in the Treated Sample.  88  Figure 3.5 A Typical Force v.s. Displacement Graph for Producing Pellets. (1: Untreated, 2: 200oC-5 min, 3: 200oC-10 min, 4: 220oC-5 min, 5: 220oC-10 min.)  Figure 3.6 Force Displacement Graph of Hardness Test of Compressing Different Pellets Made from Untreated and Steam Treated Douglas Fir Particles. (1: Untreated, 2: 200oC-5 min, 3: 200oC-10 min, 4: 220oC-5 min, 5: 220oC-10 min.)  89  Figure 3.7 Moisture Adsorption of Pellets Made from Untreated and Steam Treated Douglas Fir Particles. Both Types of Pellets were Dried to 0% Moisture Content before Placed in the Environment Chamber Set at 30oC, 90% RH  90  Chapter 4 Energetic  Analysis  of  Steam  Explosion  and  Pelletization Using Laboratory Data 4.1 Introduction Wood pellets are made by compacting ground biomass. These pellets tend to easily disintegrate during their post-production handlings. Pellets that are made from a mixture of stem wood and bark and other parts of a tree are less durable than pellets made from saw dust and shavings. Steam explosion of biomass particles prior to compaction has been suggested as a pretreatment option to increase the durability of pellets made from mixed woody materials. Steam explosion is a process that subjects a material to saturated steam at temperature 180 – 240oC (1034-3447 kPa) for 5-10 minutes followed by a rapid discharge of treated biomass to atmosphere. Steam activates and restructures the native lignin in the biomass fiber (Startsev et al., 2000; Li et al., 2007; Donohoe et al., 2008). A new chemical bonding is formed within lignin-cellulose chains to improve binding properties and hydrophobicity (Suzuki et al., 1998; Shaw et al., 2009). Negro et al. (2003) attributed an increase in hydrophobicity of steam treated biomass to pseudolignins or resins that exhibit a higher hydrophobicity than untreated biomass. Lam et al. (2010a) reported that the explosion process defibrillates the fibers and introduces more pores in the fiber matrix. Holtzapple et al. (1989) reported that the conventional mechanical methods require roughly 70% more energy to achieve the same size reduction as explosive depressurization. An increased moisture content of the treated sample is the down side if the material needs to be dried for further downstream processing. Several researchers have published data on energy use to make pellets (Polagve et al., 2007; Haase et al., 2010). The energy categories are electricity to power grinders, pelletizers, and handling equipment. Biomass and/or natural gas is used to produce heat for drying or steam treatment of biomass. In some instances, liquid fuels such as gasoline  91  and diesel are used for internal combustion engines to power mobile or stationary equipment. Although steam pretreatment is an energy intensive process, the benefits gained from enhancing the quality of biomass may offset extra energy input and associated costs. There are published literatures on techno-economic and energetic of steam pre-treated woody biomass for enzymatic hydrolysis and fermentation for bioethanol production (Shevchenko et al., 2001; Zimbardi et al., 2002; Mabee et al., 2007). These literatures do not address steam treatments for making pellets. The objective of this study was to conduct an energy balance analysis for a steam treated pelletization process. The mass balance and energy input to produce commercial pellets from steam treated and untreated feedstock were calculated. The data for the commercial pelletization was taken from a 45,000-tonne conventional pellet plant operation in Princeton, British Columbia, Canada. The commercial data were modified using laboratory steam explosion and pelletization data.  4.2 Development of Analysis Method Figure 4.1 shows a typical biomass pelleting plant consisting of seven major unit operations. The major unit operations are: receiving and sorting raw biomass, drying, grinding and blending, pelleting, cooling, screening, storing and shipping. A new steam explosion process unit is added to the system after receiving and sorting and before the drying operation. In this unit, the wood chips are subjected to steam explosion at conditions that will be specified later in this paper. Treated and untreated biomass are dried, ground and pelletized in a similar manner. Production data including power and energy input to each unit operation were taken from a commercial pelleting plant in Princeton, B.C. Canada (Mani, 2005). The pellet plant had an annual production of 45000 tonnes. The plant processed 10.35 t/h of saw dust at 45% (wet basis) moisture content to 6 t/h of pellets at 8% (wet basis)  92  moisture content. The plant operated 24 h a day 310 days per year that corresponds to an annual utilization of 85%. The calculation of the total heat required to evaporate 1 tonne of water from the biomass includes the use of pellet crumbles and saw dust as a fuel for dryer assuming the combustion efficiency at 80% and a heating value of 18 MJ/kg. Although the raw feedstock to this plant was saw dust and shavings from an adjacent saw mill, we also include debarked wood chips at 45% moisture content in calculations. We envision that a steam explosion plant would also process wood chips as the supply of saw dust dwindles. The debarked wood chips are sized to 12 mm by 12mm by 2 mm (length, width, thickness). In the following, we describe the two systems of conventional and steam explosion process in parallel.  4.2.1 Receiving 4.2.1.1 Conventional The wet debarked wood chips are unloaded and stored at the storage site at the pellet plant site. There is a 1 % mass loss of material during storage prior to drying via stream 2. The 1% mass loss were assumed according to the literature (Mani, 2005). The energy input for the conventional pre-drying process was diesel fuel consumed by the front-end loader and truck. Diesel fuel consumption rate of the machinery was estimated from the ASAE standard EP496.3 (ASABE Standard, EP496.3, 2006): Qavg  0.730.305Pm  (4-1)  where Qavg is the average fuel consumption (L/h) and Pm is the machinery power (kW). Energy input for the diesel truck and front end loader for transportation of wood chips is given by: Etp   Q1  Q2 H diesel .  m14  (4-2)  where Etp is the energy per unit mass of produced pellets to transport the wood chips from sawmill to storage site of pellet mill (MJ/kg), Q1 is the average fuel consumption for the front end loader = 16.36 L/h of diesel, Q2 is the average fuel consumption for the  93  diesel truck = 49.09 L/h, Hdiesel is the high heating value of the diesel = 36.4 MJ/L and .  m14 is the production rate of the wood pellet (kg/h).  4.2.1.2 Steam Explosion The wet wood chips stored at the pellet plant site are conveyed to steam pretreatment reactor via stream 15 (Figure 4.1). The wood chips are treated with steam explosion at 220oC for 10 min. The steam to biomass ratio was assumed to be 0.22 (mass basis). The steam is supplied to the steam explosion unit from stream 17. The moist steam treated biomass is separated into vapor and solid fractions in a cyclone. The vapor and volatiles are exhausted through stream 18. The steam exploded wood is 50% moisture content (w.b.) and is conveyed to the drying unit via stream 16. A severity index (R0) defined the severity of steam explosion treatment reactions (Overend et al., 1987), t  R0   exp   T1 T0     14.75   dt '  (4-3)  0  where t is reaction time (min), T1 is the reaction temperature (oC) and T0 is the base temperature (oC) The energy input for steam explosion is calculated based on steam temperatures, pressure and time measured from our experimental steam explosion unit available at the UBC Clean Energy Research Center (CERC) (Table 4.1). Details of measurements are reported in Chapter 2. The enthalpies of saturated vapour (hg) and enthalpies of latent heat of vaporization (hfg) at reaction temperature are used to calculate the energy input for steam generation in the boiler. The maintenance heat is calculated by estimating power input to maintain the steam reactor chamber at 260oC. The total heat input to the system is the sum of heat of steam and maintenance heat. The amount of steam generated in this system is calculated by the ideal gas law equation. The results are listed in Table 4.1.  94  Number of mole of steams generated: n  P2 V2  V1  RT  (4-4)  where n is the number of mole of steam (dimensionless), P2 is the steam pressure in the cooker (Pa), V1 is the volume of the steam boiler of 1 m3 (m3), V2 is the volume of the steam cooker and steam boiler of 3 m3 (m3), R is the universal ideal gas law of 8.314 [J/(mol K)], T is the reaction temperature (K). Energy required for saturated steam generation in the boiler: E1   nM w h g  b  (4-5)  where E1 is the energy required to generate saturated steam in the bolier (kJ), n is the number of mole of steam, Mw is the molecular weight of water of 18 (g/mol), hg is the enthalpy of saturated vapor (kJ/kg) and  b is the efficiency of boiler at 75%. Energy required to sustain the reaction temperature of the steam in the steam treatment chamber: Since the reactor is assumed to be thermally insulated perfectly, here assume the energy required to sustain the reaction temperature of the steam is equal to the heat loss of the steam to the solid particle of biomass from the plane direction by force convection:  E2   hc AT f  Ts t  e  (4-6)  where E2 is the energy required to sustain the reaction temperature T of the steam in the steam reactor (kJ), hc is the heat transfer coefficient of the saturated steam = 100 (W/m2K), A is the cross-section area (m2) of the individual powder particle assuming a slab with a length and a width of 0.3 mm respectively, Tf is the temperature of the saturated steam (K), Ts is the temperature of the solid particle assuming at 297 (K), and t is the reaction time of the steaming (s) and  e is the efficiency of electricity conversion at 30%.  95  Energy input to steam explosion unit:  E1  E 2 m p 1  mc p 1  S   E SE   (4-7)  where ESE is the energy input to steam explosion unit per unit mass of pellet (dry basis) (MJ/t), E1 is the energy required to generate saturated steam in the boiler (kJ), E2 is the energy required to sustain the reaction temperature T of the steam in the reactor (kJ), mp is the mass of the wood particles treated (25 g) (Table 4.3), mcp is the moisture content of the ground particles (10% w.b.) and S is the solid yield of the samples (%) (Table 4.3).  4.2.2 Drying The wet wood chips are dried from 45% to 10% moisture in a rotary drum dryer operating at temperatures about 300oC. In this calculation, we assume there are no extractives or volatile loss during the drying process. Only the 35% moisture loss is exhausted as steam via stream 4. Energy input to the rotary drum dryer is produced by the combustion of recycled pellet screenings and dried feedstock. The hot wood chips are then transported to the solid cyclone to separate the fines and wood chips. About 1% of material is lost as fines in stream 6 (Mani, 2005). Wood chips with 10% moisture content (w.b.) are then conveyed to hammermill via stream 7 for further size reduction. .  Determination of water evaporation rate ( m 4 ): .  .  m 3 ,bd  m3 x3,bd  mci   1  x3,bd  mc f   (4-9)  x3,bd 1  x5,bd  (4-10)  x5,bd  m 5  m3,bd 1  mc f .  (4-8)  .    (4-11)  96  .  .  .  m 4  m3  m5  (4-12)  .  where m3,bd is the dry bone biomass flow rate (kg/h), x3,bd is the solid fraction of the biomass = 55% (wet basis), mci is the initial feed moisture content = 0.82 (d.b.) (dimensionless), x5,bd is the solid fraction of the biomass = 90% (wet basis), mcf is the final product moisture content = 0.11 (d.b.) (dimensionless)  Energy input to drying operation: Total heat required for 1 kg of bone dry feed ( H d ):  H d 1  C p ,solid Td ,out  Td ,in  (4-13) H d 2  mci C p , water Tvapor  Td ,in   (4-14)  H d 3  mci  mc f H Latent (4-15)  H d 4  mci  mc f C p ,vapor Td ,out  Tvapor  (4-16)  H d 5  mc f C p , water Td ,out  Tvapor  (4-17) 5  H d   H di  (4-18)  i 1  where H d 1 is the heat required to increase the initial biomass feed from Td,in to Td,out (kJ/kg), Cp,solid is the heat capacity of the wood sawdust = 2.8 kJ/(kg˙oC), Td,in is the drying unit inlet temperature of the feed = 25oC, Td,out is the drying unit outlet temperature of the dried product = 80oC, H d 2 is the heat required to increase the moisture content of the initial biomass feed from Td,in to Tvapor (kJ/kg), Cp,water is the heat capacity of the water = 4.184 kJ/(kg˙oC), Tvapor is the temperature of the water vapor = 45oC, H d 3 is the heat required for the removal of the moisture content of the biomass  97  as the water vapor from mci to mcf (kJ/kg), H latent is the latent heat of the water vapor =2260 (kJ/kg) , Cp,vapor is the heat capacity of the water vapor = 2.00832 kJ/(kg˙oC), H d 4 is the heat required for increasing the temperature of the moisture content removed (mci – mcf) from the biomass from Tvapor to Td,out (kJ/kg), H d 5 is the heat required for increasing the temperature of the moisture content inside the wood from Tvapor to Td,out (kJ/kg), H d is the total heat required to remove moisture for 1 kg of bone dry feed (kJ/kg). Total heat required to evaporate the unit ton of water inside the biomass ( H d ,total ): H d m3,bd 1   dryer  .  H d ,total   .  m4  ce  (4-19)  where H d ,total is the total heat required to evaporate the unit ton of moisture inside the biomass (MJ/t),  dryer is the 10% heat loss assumed for this study (%),  ce is the combustion efficiency of the pellet crumbles for the pellet plant = 80% (%). Total energy input to drying operation: .  Ed   H d ,total m4 .  (4-20)  m14  where Ed is the energy input to the drying operation per unit mass of pellets (dry basis) (MJ/t) The steam treated wood will have a moisture content of 50% (w.b.) (Chapter 2). We assume the wet steam treated wood powder dried from 50% to 10% moisture in the rotary drum dryer operating at temperatures at 300oC  98  4.2.3 Size Reduction The dried wood chips with 10% moisture content are conveyed to the hammermill via stream 7 and ground through a 1.6 mm screen. There is an assumption of 1% moisture loss during the grinding process and exhaust as vapor via stream 8. The size reduced ground particles are conveyed to the densification process via stream 9. The power consumption of the commercial hammermill was estimated based on the survey of hammermills used in British Columbia pellet plants and hammer mill manufacturer (Bliss Industries, 2003) using the following correlation equation (Mani, 2005).  PH  24.44CH  4.88  (4-21)  where PH is hammermill power (kW), and CH is the mill throughput (t/h). In this case, CH .  is equal to m 7 .  ESR   3.6PH .   e m14  (4-22)  where ESR is the primary energy input to the hammermill per unit mass of pellets (dry basis) (MJ/t) and  e is the conversion efficiency of the electricity = 30% . Similar calculations of energy consumption of size reduction are done on the steam treated samples. However, there is also a size reduction effect from wood chips into powder by steam explosion. This can help to save energy in size reduction process. The net energy required for size reduction can be calculated from Kick’s law by knowing the initial size of the feedstock (Naimi, 2008; Hosseini et al., 2010) as well as the required size after steam explosion reported in Chapter 2. E  K K  Lf Lp  (4-23)  where ΔE is the energy consumption for size reduction (J/g), KK is Kick’s constant (J/g) . Lf and Lp are the initial and final sizes of the feedstock (mm), respectively. The Kick’s constant used here is 0.032 J/g reported by Esteban et al. (2006).  99  4.2.4 Pelletization The recovered finely ground sawdust is compressed to 6.2 mm diameter, 10-30 mm long pellets in a press mill. The hot pellets at 50-70oC are cooled in a pellet cooler using ambient air. This cooling dries and hardens the newly formed steamy pellets. The pellets are then screened and either placed in bags or stored in bulk in large covered warehouses or steel bins. The press mill consists of a commercial pellet mill of 220 kW in conjunction with a screw feeder of 7.5 kW without steam conditioning. The energy input for pelletization of untreated Douglas fir wood powder was estimated from the compression and extrusion data of pelletization using a single die with a heating element (Table 4.4). The die temperature used in the experiment was 70oC and assumed to be the same temperature measured for the production pellet mill. Details of experimental procedures were described in Chapter 3. The compression and extrusion energy were determined by the area under the stress strain curve. The compression energy of each sample was measured with 15 replicates. The push out energy for each pellet was averaged from 5 replicates. The total energy for pelletization (J) is defined as the sum of the compression energy (J) and the extrusion energy (J) in this case. The energy per mass (J/g) and energy per pellet density [J/(g/cm3)] of the pelletization unit operation are reported. .  0.7353.62 Etm m9 Ep  . 1000 m14 e  (4-24)  where Ep is the total energy input to the pelletization per unit mass of pellets (MJ/t) and Etm is the pelletization energy (including compression and extrusion) per mass (J/g) and   e is the efficiency of electricity conversion at 30%. Similar calculations were applied to estimate the total energy input to the pelletization using the steam exploded wood powder using data in Table 4.4.  100  4.2.5 Cooling The cooling process consists of a counterflow cooler with a feeding drive power of 15 kW, a cyclone discharger of 0.75 kW and a fan of 22.5 kW. The total energy consumption of these unit operations (Ec) are estimated by equation (4-24). Similar calculations are applied to estimate the total energy input to the cooling process using the steam exploded wood powder.  4.2.6 Screening The screener consists of a bucket elevator of 7.5 kW, a drive motor of 7.5 kW, a screener of 3.5 kW, a screener to the bin of 15 kW and a 7.5 kW crumbles conveyor. The energy consumptions of these unit operations (Ec) are estimated by equation (4-24.). Similar calculations are applied to estimate the total energy input to the screening process using the steam exploded wood powder.  4.3 Results and Discussion 4.3.1 Energy Input of Steam Explosion Table 4.1 lists the four levels of severities calculated by different combination of steam treatment time and temperature using equation (4-3). The enthalpy of steam consists of 3 parts: raising water temperature to boiling point at 100oC, evaporation of water at 100oC to steam and raising the steam temperature from 100oC to 200oC under pressure. The thermal electric energy required to generate 5.6 g saturated steam for steaming at 200oC for 5 minutes was 15.6 J. More steam around 7.7 to 8.5 g was generated at 220oC inside the closed system corresponding to the increased pressure of 1.79 and 1.93 MPa, respectively. The energy required to generate the saturated steam in the boiler depended upon temperature only. However, maintenance heat supplied to the steam chamber increased with both steaming temperature and steaming time. At the same temperature, the energy required to maintain the reaction temperature increased proportionally to the steaming time. The total energy input to steam explosion  101  pretreatments were 0.56, 0.57, 0.78 and 0.86 MJ/kg for the severity of 3.64, 3.94, 4.23 and 4.53, respectively. The energy savings of the steam explosion contributes to size reduction are calculated using the Kick’s constant of 0.032 J/g derived from the grinding energy of pine wood chips using a hammermill (Esteban et al., 2006). Assuming that initial particles size of Douglas Fir particles are 0.42 mm and size reduced to half particle size with 0.21 mm, the energy saving of the size reduction of steam exploded particles treated at 200oC and 5 min (Table 4.2) starting from 0.45 mm to 0.21 mm was - 7.2%. The negative energy saving was due to the particle swelling by volatiles loss as discussed in Chapter 2. However, energy savings from size reduction contributed by the steam explosion with increasing treatment severities from 3.94 to 4.53 were around 5 – 17% (Table 4.2).  4.3.2 Energy Input of Pelletization Table 4.4 lists mass, physical dimension and pelletization energy of the wood pellets made with untreated and steam treated ground wood particles at different severities. The pellet density of untreated sample was 1.09 g/cm3. The pellet density of steam exploded pellets at severity of 3.64 increased to 1.12 g/cm3 but decreased to 1.07 g/cm3 with increasing treatment severity to 4.53. The decrease in pellet density with increasing severity treatment was due to the volatile loss of the powder after steam treatment (Table 4.3). From proximate analysis determined by TGA, the volatile matters of steam exploded wood powder decreased from 83.1% to 77.5% by weight of the feed with increasing the severity treatment from 3.64 to 4.53. This also agrees with the increase in C/H ratio from elemental analysis and decrease in solid yield from the steam explosion experiment. This suggests that the burning characteristics of the steam exploded pellets in combustion facilities become much improved with less tar formations in exhausted gas emission.  102  The energy required to compress the steam exploded powder was higher than that of compressing the untreated sample (Table 4.4). The compression energy required to compress the untreated ground powders was 22 J. The compression energy of steam exploded powder increased from 25 J to 40 J with increasing pretreatment severity from 3.64 to 4.53. The increase in compression energy for steam exploded powder may be due to the extra energy required to break the large and hard particles into small particles and to overcome the friction between rough particles in order to fill the pores between particles. The relative compression energy of the pellet showed a second order polynomial relationship with the treatment severity and described by the following equation (Figure 4.2): Ec  0.6489 x 2  6.105 x  12.518  R2=0.99 (4-25)  where Ec is the relative compression energy per unit pellet (dimensionless) and x is the severity treatment. Similar trend was also observed with the energy used to push pellets out of die. From the observation of the pelletization experiments, the particle surfaces of the powder were sticky to the inner die wall. This attribute to the presence of higher extractives content and higher monosugars generated after steam explosion pretreatment (Chapter 3). The extractives and monosugars acted as binders between particle surfaces during pelletization (Sokhansanj et al., 2005). Meanwhile, this required a higher force to overcome the frictional force acting between the pellet surface and the inner wall of the die. Therefore, higher extrusion energy was required to extrude the pellets out of the die. The pushing energy of the untreated pellet was 49 mJ while that for steam exploded pellet increased from 118 mJ to 236 mJ with increasing treatment severity from 3.64 to 4.53. Figure 4.3 shows the relative total pelletization energy per unit mass and density including both compression and extrusion energy vs. pretreatment severity. The total energy for pelletization increased from 110% to 190% with increasing treatment severity from 3.64 to 4.53. The total pelletization energy per unit mass of the pellet showed a  103  second order polynomial relationship with the treatment severity and described by the following equation: Ertm  0.6863x 2  6.4939 x  13.463  R2=0.99 (4-26)  where Ertm is the relative total pelletization energy per unit mass (dimensionless) and x is the treatment severity.  The total pelletization energy per unit density of the pellet showed a second order polynomial relationship with the treatment severity and described by the following equation: Ertd  0.1456 x 2  1.7298 x  3.2467  R2=0.94 (4-27)  where Ertd is the relative total pelletization energy per unit density (dimensionless) and x is the treatment severity. 4.3.3 Overall Process of Wood Pellets Made from Untreated and Steam Treated Sawdust Table 4.5 lists the direct energy input to each unit operations for biomass pelletization process with steam explosion at different treatment severity with 0.22 steam to biomass ratio. The production rate of pellets using untreated biomass is 6000 kg/h. The production rate of pellets made from steam treated biomass decreased to 4487 - 4185 kg/h. Depending upon the severity of steam treatment, the solid yield and volatile loss are the contributors to reduce the steam exploded pellet production rate. The volatiles can be separated from the saturated steam, captured and recycled to be burnt as a fuel for the whole process. Future work can be considered on the study of the heating value of the volatiles and its combustion performance. The direct energy input per unit kilogram of produced pellets at 8% moisture content (w.b.) of raw material collection and transportation increased from 0.27 MJ/kg to 0.38 MJ/kg. For steam explosion pretreatment, the direct energy input per unit kilogram of pellets increased from 0.80 MJ/kg to 1.20 MJ/kg with increasing severity. For drying, the energy required to dry the sawdust per unit kilogram of pellets from 45% to 10% moisture content (w.b.) for the untreated case was 2.90 MJ/kg. The energy required to dry  104  the steam exploded sawdust with 50% moisture content to 10% moisture content decreased from 4.30 MJ/kg to 4.12 MJ/kg with increasing treatment severity. It is because some materials were lost as volatiles during the steam explosion pretreatment. The energy required to size reduce the untreated sawdust per unit pellets was 0.31 MJ/kg. The energy required to size reduce the steam treated sawdust were in the range between 0.34 MJ/kg and 0.35 MJ/kg at different treatment severity. However, the energy required for pelletization in a pellet plant per unit pellets increased with steam explosion pretreatment severity from 1.67 MJ/kg to 2.31 MJ/kg. In general, the energetic ratio of the biomass densification process with untreated woodchip based on produced pellets was 4.07. The energetic ratios of the biomass densification process with steam explosion pretreatment at 200oC for 5 and 10 minutes were both 2.54. Steam treatment Temperature is the most significant parameter to affect the energy input to the unit operation. Drying was the most energy consuming unit operation, consuming nearly 60% of the entire energy consumption for the untreated case. With steam explosion pretreatment, the energy consumption of drying was still the most energy intensive unit operation between 50% – 55% of the entire process. For all steam treatment cases, the second most energy intensive process is the pelletization which consumes 22% – 27% of the entire direct energy input. Steam explosion was the third most energy intensive process compared to other unit operations. It ranged 10% – 14% of the total primary input to the whole process. The previously reported values were based on our laboratory data of steam explosion and pelletization. In practice, the design of steam explosion can be optimized to reduce the energy consumption of the overall biomass densification process. The following are proposed 3 modifications to reduce energy input to the system. 1. The steam explosion unit used in commercial scale should be continuous with insulation against heat losses. Operational characteristics of a continuous unit would be different from the batch reactor we used in this study. Without maintenance heat, the direct energy input of steam explosion per unit feedstock was calculated to be between  105  0.79 MJ/kg and 0.80 MJ/kg for the treatment severity at 200oC for both 5 min and 10 min. For treatment severity at 220oC, the direct energy input of steam explosion per unit pellets was between 1.08 MJ/kg and 1.18 MJ/kg (Table 4.6). The direct energy input of steam explosion per unit pellets for all severities ranged between 11% and 16% of the total direct energy input to the whole process. 2. The mixture of saturated steam and steam treated material can be separated by a cyclone after steam explosion. This prevents the steam condensation on the steam treated material. This also saved some drying energy to dry the steam treated materials. The direct energy input to the drying per unit pellets is estimated to decrease from 3.63 MJ/kg to 3.49 MJ/kg with increasing the treatment severity (Table 4.6). The direct energy input of drying per unit feedstock for all severities ranged between 46% and 51% of the total direct energy input to the whole process. 3. Steam explosion had an energy saving in size reduction of steam treated samples. The particle size of steam treated material decreased with increasing treatment severity. Therefore, less energy was consumed to grind the particles to a target particle size for pelletization. When the energy saving in size reduction is considered, the direct energy input to the size reduction per unit pellets decreased from 0.36 to 0.29 MJ/kg with increasing treatment severity (Table 4.6). Furthermore, thermally treated wood particles were reported to have less energy consumption for grinding (Holtzapple et al., 1989; Govin et al., 2009). Torrefied spruce wood chips at 180oC were reported to have energy consumption of grinding dropped off by 40% compared to untreated one (Govin et al., 2009). This was due to the increase in brittleness of the materials. It was expected that steam treated materials would have similar characteristics as torrefied materials which require less grinding energy. Future work would be recommended to perform on studying the grinding energy requirement of steam treated woody materials. Figure 4.4 summarizes the energetic ratios, handling and storage quality of pellets treated at different severity of steam explosion pretreatment with steam to biomass ratio of 0.22. The handling and storage quality of the pellets was indicated by the Meyer  106  hardness and the maximum breaking force. The relative Meyer hardness and relative maximum breaking force were the responses of the steam exploded pellets over the responses of the untreated pellets. The best pellet quality for handling and transportation was the pellets treated with steam explosion at 220oC for 5 minutes (i.e., severity = 4.23). The energetic ratio of the overall optimized biomass densification process including this severity steam explosion pretreatment is 2.61. Although there is an extra energy input to the steam explosion process, the quality of pellets is improved in terms of mechanical strength and moisture sorption resistance (Chapter 3). There are benefits in reducing the operation cost associated with the storage and handling problems. Future work would be recommended on conducting an economic analysis on determining cost benefit. This analysis should consider economic gains from the quality improvement of the steam treated pellet against the extra production cost.  4.4 Conclusions The energy analysis of a biomass densification with steam explosion pretreatment at four severities was conducted. Production data from a commercial pellet plant producing 45,000 metric ton/year was used for analysis. The direct energy inputs to steam explosion and pelletization were calculated based on individual mass and energy balance of particular unit operations using the data obtained from laboratory scale experiments. Mass balance analysis showed that the production rate of pellets decreased from 6,000 kg/h to 4,185 kg/h due to the volatile loss during steam explosion pretreatment. Energy balance showed that the total direct energy input per unit kilogram of pellets to the overall biomass densification process was 4.83 MJ/kg. With steam explosion pretreatment included, the total direct energy input per unit pellets were 7.78, 7.91, 8.55 and 8.70 MJ/kg, respectively. The increase in energy use was due to increased energy input to steam explosion process for steam generation and maintenance heat, higher energy required for pelletization and extra drying energy required for removing the condensed steam on materials. The overall energetic ratio of the overall process was 4.07 107  for the conventional biomass densification process. The energetic ratios of the biomass densification process with steam explosion pretreatment were 2.54, 2.54, 2.41 and 2.38, respectively for pretreatment severities of 3.64, 3.94, 4.23 and 4.53, respectively. The energy percentage of energy consumption of steam explosion over the pellet heating value was between 4% and 6% in all treatment severity cases. The best pellet quality for handling and transportation was for pellets produced from biomass treated with steam explosion at 220oC for 5 minutes (i.e., severity = 4.23). The energetic ratio of the overall biomass densification process including this severity steam explosion pretreatment is 2.41. A commercial scale steam explosion and densification process requires a welldesigned unit for gas-solid separation (e.g. cyclone). That will provide a better heat insulation (minimize the heat loss of the saturated steam). That also prevents the steam condensation on steam treated material after the treatment, and thus resulting in reducing the extra energy required for drying. 4.5 Notation Physical properties: A  surface area of individual particle (m2)  Cp,solid  heat capacity of the wood sawdust = 2.8 kJ/(kg˙oC)  Cp,water  heat capacity of the water = 4.184 kJ/(kg˙oC)  Cp,vapor  heat capacity of the water vapor = 2.00832 kJ/(kg˙oC)  H latent  latent heat of the water vapor =2260 (kJ/kg)  Mass flow: mci  initial feed moisture content = 0.82 (d.b.) (dimensionless)  mcf  final product moisture content = 0.11 (d.b.) (dimensionless)  x3,bd  solid fraction of the biomass = 55% (wet basis)  x5,bd  solid fraction of the biomass = 90% (wet basis)  Energy flow: CH  mill throughput (t/h) 108  Ec  relative compression energy per unit mass of pellet (dimenionless)  Ed  energy input to the drying operation per unit mass of pellets (dry basis) (MJ/t)  Ep  total energy input to the pelletization per unit mass of pellets (MJ/t)  Ertd  relative total pelletization energy per unit density of pellet (dimensionless)  Ertm  relative total pelletization energy per unit mass of pellet (dimensionless)  ESE  energy input to steam explosion unit per unit mass of pellet (dry basis) (MJ/t)  ESR  energy input to the hammermill per unit mass of pellets (dry basis) (MJ/t)  Etd  total pelletization energy per unit density of pellet (J/(g/cm3))  Etm  total pelletization energy per unit mass of pellet (J/g)  Etp  energy per unit mass of pellets produced to transport the wood chips from sawmill to storage site of pellet mill (MJ/kg)  Ewc  energy content per unit time of the wood chips from sawmill to storage site of pellet mill (MJ/h)  E1  energy input to generate saturated steam in the boiler (kJ)  E2  energy required to sustain temperature of the cooker (kJ)  Hdiesel  high heating value of the diesel (MJ/L)  hc  heat transfer coefficient of saturated steam = 100 (W/m2K)  hg  enthalpies of saturated vapor (kJ/kg)  KK  Kick’s constant (J/g)  Lf  initial sizes of the feedstock (mm)  Lp  final sizes of the feedstock (mm)  mp  mass of the wood particles treated of 25 (g)  mcp  moisture content of the ground particles of 10% m.c. (w.b.).  Mw  molecular weight of water (g/mol)  n  number of moles of steam (dimensionless)  P1  steam pressure in the boiler (Pa)  P2  steam pressure in the reactor (Pa)  PH  hammermill power (kW) 109  Pm  machinery power (kW)  Q1  average fuel consumption for the front end loader (L/h)  Q2  average fuel consumption of diesel for the truck (L/h)  Qavg  average fuel consumption of diesel (L/h)  R  universal ideal gas law of 8.314 (J/mol K)  S  solid yield of samples (%)  T  reaction temperature (K)  Td,in  drying unit inlet temperature of the feed = 25oC (oC)  Td,out  drying unit outlet temperature of the dried product = 80oC (oC)  Tf  temperature of the saturated steam (K)  Ts  temperature of the solid particle assuming at 297K (K)  Tvapor  temperature of the water vapor = 45oC (oC)  t  reaction time of the steaming (s)  x  severity treatment (dimensionless)  V1  volume of the steam boiler (m3)  V2  volume of the steam reactor and boiler (m3)  ΔE  energy consumption for size reduction (J/g)  H d  total heat required to remove moisture for 1 kg of bone dry feed (kJ/kg).  H d 1  heat required to increase the initial biomass feed from Td,in to Td,out (kJ/kg)  H d 2  heat required to increase the moisture content of the initial biomass feed from Td,in to Tvapor (kJ/kg)  H d 3  heat required for the removal of the moisture content of the biomass as the water vapor from mci to mcf (kJ/kg)  H d 4  heat required for increasing the temperature of the moisture content removed (mci – mcf) from the biomass from Tvapor to Td,out (kJ/kg)  H d 5  heat required for increasing the temperature of the moisture content inside the wood from Tvapor to Td,out (kJ/kg)  H d ,total  total heat required to evaporate the unit ton of moisture inside the biomass (MJ/t) 110  Efficiency   dryer  10% heat loss (%)   ce  combustion efficiency of the pellet crumbles for the pellet plant = 80% (%)  e  conversion efficiency of the electricity = 30% (%).  b  steam boiler efficiency = 75% (%).  111  Tables Table 4.1 Direct Energy Input for Steam Explosion at Different Severity with 0.22 Steam to Biomass Ratio Severity Treatment temperature (T) Treatment time (t) Steam pressure in boiler (P1) Steam pressure in cooker (P2)  Unit C s kPa kPa  3.64 206 302 1750 1236  3.94 204 603 1762 1243  4.23 229 302 2562 1777  4.53 223 603 2804 1935  Properties of steam Saturated liquid density Saturated vapor density Specific volume - sat. liq. Specific volume - sat. vap. Enthalpy - sat. vap (hg)  kg/m3 kg/m3 m3/kg m3/kg kJ/kg  857.56 8.8637 0.001166 0.11282 2795.3  859.47 8.5869 0.001164 0.11646 2794.5  828.42 13.739 0.001207 0.072785 2802.8  836.31 12.295 0.001196 0.081334 2801.7  Number of moles (n) Mass of steam  g  310.47 5.59  313.25 5.64  425.63 7.66  469.23 8.45  Energy input Energy to generate saturated steam from boiler (E1) kJ 15.62 15.76 Energy to sustain temperature of cooker (E2) kJ 0.05 0.10 Total energy per unit mass1 (d.b.) MJ/kg 0.56 0.57 1 : refers to the mass of biomass to be treated, which is 25 g from the experiment  21.47 0.06 0.78  23.66 0.11 0.86  o  Table 4.2 Energy Saving in Size Reduction of Steam Treated Samples Conditions  0.42  0.21  Size reduction energy (J/g) 0.063  3.64  0.45  0.21  0.068  -7.14  3.94  0.40  0.21  0.060  4.76  4.23  0.38  0.21  0.057  9.52  4.53  0.35  0.21  0.053  16.67  Initial Final Severity particle size particle size (mm) (mm)  Untreated o  200 C, 5 min o  200 C, 10 min o  220 C, 5 min o  220 C, 10 min  Energy saving (%) 0  112  Table 4.3 Summary of Physical and Chemical Properties of Steam Treated Wood at Different Treatment Severity Severity Treatment temperature Steam pressure Particle size (dgw) Initial moisture content (mp) Solid yield (S) Moisture content after SE  Units o C kPa mm % % %  Untreated 0.42 10 100 10.7  3.64  3.94  4.23  4.53  1236 0.45 10 83.1 14.5  1243 0.40 10 82.3 19.8  1777 0.38 10 79.1 22  1935 0.35 10 77.5 32.2  Chemical composition Cellulose + Hemicellulose Lignin Extractives Ash content  % % % %  68.22 30.09 1.42 0.27  71.58 25.88 2.22 0.32  66.8 29.65 3.23 0.32  53.02 37.62 8.84 0.52  49.00 43.23 7.36 0.41  Elemental analysis (EA) C H N O  % % % %  48.44 6.23 0.22 45.28  49.14 6.08 0.17 44.63  50.46 6.1 0 43.12  52.42 5.95 0.18 41.29  53.09 5.91 0.17 40.76  Proximate analysis (PA) Fixed Carbon Volatile matters Ash content  % % %  14.4 85.6 3.1  16.9 83.1 3.2  17.7 82.3 2.2  20.9 79.1 2.5  22.5 77.5 2.0  %  6.89  5.96  6.03  6.03  5.25  MJ/kg MJ/kg  18.82 20.21  18.94 20.14  19.15 20.38  19.5 20.75  20.09 21.20  Moisture content @ HHV measurement HHV measured (w.b.) HHV measured (d.b.)  113  Table 4.4 Direct Energy Input of Pelletization of Untreated and Steam Treated Wood at Different Severity Conditions Untreated 200oC, 5 min 200oC, 10 min 220oC, 5 min 220oC, 10 min 1 : n = 15 2 :n=5  Severity  3.64 3.94 4.23 4.53  Mass1 (g) 0.69 0.68 0.79 0.79 0.79  Pellet density1 (g/cm3) 1.09 1.12 1.09 1.08 1.07  Compression Energy1 (J) 22.34 25.04 31.76 38.82 40.41  Extrusion Energy2 (J) 0.04886 0.11767 0.11725 0.17753 0.23539  Energy per mass (J/g) 32.48 37.06 40.54 49.59 51.40  Energy per density (J/(g cm3)) 20.47 22.39 29.38 36.07 38.09  114  Table 4.5 Direct Energy Input Per Unit Kilogram of Produced Pellets at 8 % m.c. (w.b.) to the Biomass Pelletization Process with the Laboratory Data of Steam Explosion at Different Treatment Severity and Pelletization with 0.22 Steam to Biomass Ratio. Untreated  200oC, 5 min  200oC, 10 min  220oC, 5 min  220oC, 10 min  Severity  --  3.64  3.94  4.23  4.53  Raw material collection/ Transportation  0.27  0.36  0.36  0.38  0.38  Units  Steam explosion  MJ/kg  0  0.80  0.81  1.08  1.20  Drying  MJ/kg  2.90  4.30  4.26  4.19  4.12  Size reduction MJ/kg  0.31  0.34  0.34  0.34  0.35  Pelletization MJ/kg  1.11  1.67  1.83  2.23  2.31  Cooling  MJ/kg  0.08  0.10  0.10  0.11  0.11  Screener  MJ/kg  0.08  0.11  0.11  0.11  0.12  Miscellaneous MJ/kg units  0.08  0.10  0.10  0.11  0.11  Total energy input per unit MJ/kg kg of pellets  4.83  7.78  7.91  8.55  8.70  Production kg/hr rate of pellets  6000  4487  4444  4271  4185  Energy content of MJ/kg wood pellet  19.64  19.73  20.11  20.56  20.73  Energy ratio  4.07  2.54  2.54  2.41  2.38  %  0  4  4  5  6  %  25  39  39  42  42  Energy percentage of steam explosion over pellet heating value Energy percentage of preprocessing over the pellet heating value  115  Table 4.6 Direct Energy Input Per Unit Kilogram of Produced Pellets at 8 % m.c. (w.b.) to the Biomass Pelletization Process with Optimized Steam Explosion at Different Treatment Severity and Pelletization with 0.22 Steam to Biomass Ratio. Untreated  200oC, 5 min  200oC, 10 min  220oC, 5 min  220oC, 10 min  Severity  --  3.64  3.94  4.23  4.53  Raw material collection/ Transportation  0.27  0.36  0.36  0.38  0.38  Units  Steam explosion  MJ/kg  0  0.79  0.80  1.08  1.18  Drying  MJ/kg  2.90  3.63  3.59  3.55  3.49  Size reduction MJ/kg  0.31  0.36  0.32  0.31  0.29  Pelletization MJ/kg  1.11  1.67  1.83  2.23  2.31  Cooling  MJ/kg  0.08  0.10  0.10  0.11  0.11  Screener  MJ/kg  0.08  0.11  0.11  0.11  0.12  Miscellaneous MJ/kg units  0.08  0.10  0.10  0.11  0.11  Total energy input per unit MJ/kg kg of pellets  4.83  7.12  7.22  7.88  8.00  Production kg/hr rate of pellets  6000  4487  4444  4271  4185  Energy content of MJ/kg wood pellet  19.64  19.73  20.11  20.56  20.73  Energy ratio  4.07  2.77  2.79  2.61  2.59  %  0  4  4  5  6  %  25  36  38  38  39  Energy percentage of steam explosion over pellet heating value Energy percentage of preprocessing over the pellet heating value  116  Figures  Figure 4.1 Biomass Preprocessing with and without Steam Explosion Pretreatment (Dash Line shows Steam Explosion Pretreatment, Numbers Indicate the Streamline to Indicate the Mass Balance)  Figure 4.2 Relative Compression Energy Against Steam Explosion Pretreatment Severity  117  Figure 4.3 Relative Total Pelletization Energy (Compression + Extrusion) per Mass and per Density against Steam Explosion Pretreatment Severity  Figure 4.4 The Energetic ratio, Relative Meyer Hardness and Relative Maximum Breaking Force of Pellets Treated at Different Severity of Steam Explosion Pretreatment with Steam to Biomass Ratio of 0.22  118  Chapter 5 Kinetic  Modeling  of  Polysaccharides  Depolymerization and Pseudolignin Formation of Softwood during Steam Explosion 5.1 Introduction Steam explosion is a short time vapor-phase treatment of biomass with steam followed by explosive decompression of the biomass. It involves high pressure saturated steam ranging from 1.0 to 3.5 MPa (150 to 500 PSI) to heat up biomass rapidly to 180240oC. The explosion caused by a sudden release of high pressure defibrillates the cellulose bundles. According to Delong (1981), individual fibers enhance the cellulose accessible site for enzymatic hydrolysis and fermentation. The formation of acetic acid from the degradation of hemicelluloses during steaming alters the chemical structure of biomass. The use of acidic gases (SO2, NH3) or dilute acid (H2SO4) as catalyst has enhanced the hydrolysis rate of hemicelluloses of softwood and corn fiber (Boussaid et al., 2000; Shevchenko et al, 2001; Bura et al., 2002). The steam explosion aims at increasing the accessible sites for cellulose for the enzymatic hydrolysis and subsequent fermentations (Shevchenko et al., 2001; Bura et al., 2002; Bura et al., 2003). Meanwhile, the chemical modification of the lignocellulosic feedstock by steam explosion improved the mechanical properties and hydrophobicity of fiberboards (Suzuki et al., 1998; Bouajila et al., 2005). The improvement contributed to the self-binding property of plasticized lignin (Bouajila et al., 2005; Startsev et al., 2000). In addition to fabricating stronger fiberboards, the process also has long been used as a pretreatment to remove lignin effectively for making good quality pulp for paper (Kokta, 1989; Kokta and Vit, 1998). Durable and water resistant wood pellets are desirable for safe handling during transport and storage. It is believed that the increased level of apparent lignin content, “pseudolignin”, is responsible for better durability and hydrophobicity. Pseudolignin is defined as the apparent lignin formed during steam explosion that includes the sum of the solubilized lignin, high molecular weight lignin and the condensed lignin formed from  119  lignin-lignin self-condensation and from lignin-furfural polymerization (Miranda et al., 1978). Sannigrahi (2011) also proposed the formation of pseudolignin by the combination of carbohydrate and lignin degradation products to account for the increased Klason lignin content in biomass pretreated under acidic conditions. They found the very low Klason lignin content of the starting material, together with acid catalyzed dehydration of carbohydrates is responsible for the formation of pseudolignin. Ramos (2003) reviewed the detailed chemistry involved in steam treatment of the lignocellulosic materials and concluded that the process is an autohydrolysis process in the absence of the acid catalyst. It mainly depolymerizes the hemicelluloses and lignin in the temperature range between 180 – 240oC. Cellulose is highly crystalline in structure and is less susceptible to hydrolysis by high pressure saturated steam compared to hemicelluloses. The steam hydrolyzes the acetyl group of hemicelluloses to release acetic acid. For softwood, the acid cleaves the polymer chains of hemicelluloses by autohydrolysis to release both C6 mono-sugars from galactan and mannan to form galactose and mannose, respectively. Small amounts of C5 sugars from xylan and arabinan were released to generate xylose and arabinose in softwood. Degradation of C6 monosugars (galactose) may lead to the formation of 5-Hydroxymethylfurfural (HMF) (Martinez et al.; 2010), which will react with acid soluble lignin to become part of pseudolignin (Zeitsch, 2000). During steam explosion, the major reactions involve the cleavage of the β-O-4 and β-5 aryl ether linkages of high molecular weight lignin (acid insoluble lignin). This cleavage generates a low molecular weight lignin (acid soluble lignin). Shevchenko et al. (1999, 2001) suggested that the soluble lignin redistributes, condenses and forms beads on the surface of the cellulose micro-fibrils and thereby increases the porosity of the micro-fibers. Softwood lignin transformation involves condensation reactions via benzyl cation that increases the carbonyl content after steam explosion. Miranda et al. (1979) proposed the mechanism of depolymerization/repolymerization of the lignin via carbonium ion and their results of the apparent increase in total lignin content of the  120  product are due to the hemicelluloses degradation products, furfural and lignin polymerization. Several literatures reported that the total pseudolignin content increased with the temperature (and pressure) and duration of the steaming during steam explosion of aspen wood (Miranda et al., 1979; Efremov et al., 1995). However, there is no reaction kinetic model to describe the increase in pseudolignin with the increasing severity of steam treatment condition. The objective of this work is to develop a kinetic model of sugar hydrolysis (galactan, arabinan and mannan) and lignin (acid insoluble and acid soluble lignin) reaction during steam explosion. The model is used to predict the formation of pseudolignin and to evaluate the optimized treatment condition for making durable and water repellent wood pellets.  5.2 Materials and Methods 5.2.1 Sample Preparation Sample preparation and steam explosion pretreatment were described in Chapter 2. The steam exploded materials were further ground by a Wiley Mill with 40 mesh screen to prepare the suitable particle size for Klason lignin experiment. The ground powders were dried at 105oC using the convection oven before chemical composition characterization.  5.2.2 Chemical Composition Characterization After the pretreatment, 2 g dried powder was mixed with 30 ml distilled water to recover the water soluble (WS) fraction. This simulates the C5 and C6 monosugars dissolved in the condensed steam after steam pretreatment. The resulting slurries were agitated overnight to recover the water soluble fraction. An aliquot of the water soluble fraction was also collected to analyze the sugar yield during washing. The water insoluble  121  (WI) fractions were then washed with approximately 50 ml water centrifuged at 4000 rpm for 10 minutes for four times. This helped to make sure there was no WS fraction left on the insoluble fraction. Acid post-pretreatment hydrolysis on water insoluble fractions was carried out in a final sulfuric acid concentration of 3% at 121oC for 60 min. This ensures all oligomers were converted into monomers for HPLC quantification. Furfural and HMF in the water soluble fraction were measured on Dionex (Sunnyvale, CA) HPLC (ICS-3000) equipped with an AS 50 auto sampler, ED50 electrochemical detector, GP 50 gradient pump and anion exchange column (Biorad Aminex HPX-87H). The substrates solid phase materials were analyzed for acid insoluble lignin and carbohydrates using Tappi-T-22 om-88 (TAPPI, 1994) method. The hydrolysate from this analysis was retained and analyzed for soluble lignin by reading the absorbance at 205 nm (Dence, 1992). Sugars, furfural and HMF from solid phase materials and water soluble fractions were measured on Dionex (Sunnyvale, CA) HPLC (ICS-3000) equipped with an AS 50 auto sampler, ED50 electrochemical detector, GP 50 gradient pump and anion exchange column (Dionex CarboPac PA1). All HPLC analysis was done in duplicate.  5.3 Model Development The chemical composition of Douglas Fir softwood, the Arrhenius parameters and the estimated reaction kinetic parameters were obtained from experiments. They were used to solve together with the ordinary differential equations (ODEs) of the kinetic equations to generate the reaction kinetics profiles. Matlab 6.0 software was used to solve the ODEs and an ODE solver with the built-in Runge-Kutta explicit scheme (ode45) algorithm was used. This algorithm monitors the estimate of the integration error, and reduces or increases the step size of integration in order to keep the error below a specified threshold. The accuracy is set to be that both the relative absolute (maximal) errors be less than the truncated error tolerance. The default value of this tolerance is 1.0E-6. The least sum of square error (SSE) between the experimental data and the predicted values from the model was compared.  122  The predicted data points were first generated by solving the reaction rate equations by the ode solver (ode45). Cubic spline data interpolation of the model predicted values are interpolated to obtain the exact sugar and pseudolignin content at the exact time and compared with our experimental values. Then, the least sum of square of the data is obtained by comparing the experimental value and the model predicted value. Different reaction rate constants are used for model prediction and thereby obtain different least sum of square (SSE) values. The optimized reaction rate constant value is obtained by choosing the corresponding minimum SSE values and this gives the best curve fitting of the data sets.  5.4 Results and Discussion 5.4.1 Chemical Composition of Untreated and Steam Treated Douglas Fir Table 5.1 lists the chemical composition of the solid phase of untreated and steam treated Douglas Fir. The overall mass balance of the chemical composition of Fir before and after treatment from the Klason lignin analysis is close to 100% recovery. Larger mass balance errors for the experimental results at 180oC and 3 min, and 220oC and 5 minute could be due to the experimental sampling error. From Table 5.2, the Glucan from cellulose of Douglas Fir decreased from 54.62 to 38.54 g/100g of feed after steam treatment. This indicated the steam treatment partially depolymerized cellulose to release glucose. The released glucose dissolved in condensed steam. Table 5.3 shows that the glucose percentage based on treated solid sample in the water soluble fraction increased with treatment time at 180oC and 200oC, respectively. At 220oC, the glucose percentage increased to the maximum at 5 minutes and then decreased at 10 minutes. Similar trends were also found for other C6 sugars of Mannose and Galactose formation due to steaming. This was due to the degradation of glucose into other by-products, e.g. 5-hydroxylmethylfurfural. This agreed with the formation of the 5-hydroxylmethylfurfural (HMF) at different temperatures measured by HPLC as shown in Figure 5.1. At 180oC, there was no HMF detected by HPLC. The reaction temperature 123  was not high enough to degrade the sugars into HMF. There was initially no HMF formation until 5 minutes for the treatment at 200oC. At a higher temperature of 220oC, there was a rapid increase in HMF concentration to a maximum of 0.00032 μg/ml initially at 3 minutes. HMF concentration further decreased with increasing time. This may be due to partial reaction of HMF with acid soluble lignin or further degraded into other by-products. For hemicelluloses, the C5 sugars from xylan and arabinan of the untreated Fir decreased with increasing the treatment severity. Table 5.2 shows that the xylan remaining in the solid residue decreased from 4.42 g/100g feed to the minimum of 1.52 g/100g feed treated at 220oC for 10 minutes. Similarly, the arabinan remaining in the solid residue decreased from 0.92 g/100g feed to 0.05 g/100g feed. Therefore, the summation of the total sugar dissolved into the condensed steam of the C5 sugars was 3.77 g/100g feed steam treated at 220oC for 10 minutes. The amounts of degraded product of furfural were below the detection limit of HPLC (i.e. <2 μg/mL). Table 5.2 also shows that the C6 sugars of hemicelluoses from galactan and mannan of the untreated Douglas Fir decreased with increasing the treatment severity. The galactan remaining in the solid residue decreased from 2.65 g/100g feed to the minimum of 0.21 g/100g feed treated at 220oC for 10 minutes. Similarly, the mannan remaining in the solid residue decreased from 10.02 g/100g feed to 2.81 g/100g feed. The summation of the total sugar dissolved into the condensed steam of the C5 sugars were 9.65 g/100g feed. The acid insoluble lignin content of the treated Douglas Fir by weight of the feed decreased initially and increased with time at the same temperature. The initial decrease of the acid insoluble lignin was due to the depolymerization reaction to form acid soluble lignin in solid phase. From Table 5.2, the acid soluble lignin decreased from 0.35 g/100g feed to a minimum of 0.27 g/100g feed at 3 minute and increased to 0.29 g/100g feed at 10 minute. Similar trends were also found in the reaction temperature of 200oC and 220oC, respectively. The consumption of acid soluble lignin may due to the reaction with  124  the furfural formed by degradation of xylose and arabinose and the 5hydroxylmethylfurfural (HMF) formed by degradation of mannose and galactose. This agreed with the mechanism of depolymerization/repolymerization of the lignin reported by Miranda et al. (1979).  5.4.2 Estimation of Arrhenius Parameters of Polysaccharides Depolymerization Among all of the carbohydrates in Fir, glucan and xylan composition did not change at 180oC (Table 5.2). This treatment temperature was not high enough to depolymerize the xylan and glucan. This did not allow us to have three obtained reaction rates at three different temperatures to fit the reaction rate law to obtain kinetic parameters, pre-exponential constant and activation energy. Therefore, the estimation of Arrhenius parameters was carried out on galactan, arabinan and mannan. Estimation of the Arrhenius parameters of carbohydrates depolymerization by hydrolysis was carried out based on the remaining carbohydrates remaining in the solid phase over the treatment time at three different temperatures. The reaction is assumed to be Arrhenius type temperature dependent, first order and irreversible.  k  A0  e    E act RT  (5-1)  where Ao is pre-exponential factor (min-1), Eact is the activation energy (kJ/mol), R is the universal gas constant = 8.3143x10-3 (kJ/(mol.K)), T is the reaction temperature (K).  The Arrhenius equation (5-1) can be rearranged to equation (5-2) to obtain the activation energy (kJ/mol) from the slope and the pre-exponential factor in natural logarithm of the respective sugar from the intercept. ln k  ln A0   Eact R  1   T   (5-2)  Figure 5.2 shows the Arrhenius plot of the reaction rate of arabinan, galactan and mannan in natural logarithm at different temperature. The slope of the linear fitting line  125  yields the activation energy (kJ/mol) while the intercept yields A0. The estimated Arrhenius parameters of these carbohydrates were summarized in Table 5.4. The experimental results of galactan depolymerization and lignin reaction in Table 5.2, the Arrhenius parameters of carbohydrates in Table 5.4 and the kinetic parameters of lignin in Table 5.5 were fitted with the kinetic model developed in the following section.  5.4.3 Kinetic Model of Galactan Depolymerization Martinez et al. (2010) proposed the following chain of reactions for galactan depolymerization from sugar beet pulp during hydrothermal treatment:  Galactan  kf  Galactose  k2  HMF  ks  k3  Other degraded product  (5-3)  where kf is the fast galactan depolymerization rate constant (min-1), ks is the slow galactan depolymerization rate constant (min-1), k2 is the reaction rate constant (min-1) for degradation to byproduct 5-hydroxylmethylfurfural (HMF) and k3 is the reaction rate constant (min-1) for further degradation of HMF to other product, e.g. Levulinic acid (Ulbricht et al., 1984). Our sample is softwood which is different from the sugar beet pulp used by Martinez et al. (2010). The proportion of the fast reacting and slow reacting galactan is not known. Instead, we assume the following simplified pathway:  Galactan  k1  Galactose  k2  HMF  (5-4)  where k1 is the galactan depolymerization rate constant (min-1) and k2 is the reaction rate constant (min-1) for degradation to byproduct 5-hydroxylmethylfurfural. The proposed reactions are assumed to be first order with reaction rates of Arrhenius type temperature dependence: d galactan    k1 galactan  dt  (5-5)  126  d galactoseliquid dt   k1 galactoseliquid  k 2 HMF  d HMF   k 2 galactose liquid dt  (5-6) (5-7)  Integrating equations (5-5) yields, x R  x10  e  k1t  (5-8)  where xR is the galactan remained in the solids phase at any time (g/100g feed); x10 is the initial weight of galactan remaining in the solid phase at t= 0 min (g/100g feed); k1 is the galactan depolymerization reaction rate constant (min-1) and k2 is the galactose degradation to byproduct 5-hydroxylmethylfurfural (HMF) reaction rate constant (min-1); t is time (min). Figure 5.3 shows the galactan depolymerization kinetics at three different temperatures. The model gives reasonable prediction of the experimental data points. The reaction rate constants (k1) as listed in Table 5 for this reaction were 0.0531 min-1, 0.1361 min-1 and 0.3661 min-1 for the reaction temperatures of 180oC, 200oC, 220oC, respectively. The sum of square errors (SSE) of the fitted reaction rate constants at three different temperatures were 1.32E-05, 1.28E-05 and 1.47E-05, respectively. The small SSE indicates that the reaction rates obtained are based on good fitting between model predicted values and the experimental values.  5.4.4 Net Galactose in Condensed Steam or in Liquid Prehydrolysate The following kinetic equation describes the generation rate of galactose in the condensed steam (xL):  dx L  k1 x R  k 2 x L dt  (5-9)  Combining equations (5-8) and equation (5-9) yields,  dx L k t  k 2 x L  k1 x1o e f dt  (5-10)  Solving equation (5-10) for xL yields  127   k x   x L   1 10   e  k1t  e  k2t  k 2  k1        (5-11)  where we can define k2 by: k 2  Ao 2  e    E2 RT  (5-12)  xL is the net sugar yield-percentage galactose which is soluble in liquid (g/100g of feed); x10 is the percentage of galactan at t = 0; t is time (min); k1 is the rate constants of reacting galactan hydrolysis (min-1); k2 is galactose degradation rate constant (min-1); A02 is preexponential factor for degradation reaction of galactose = 2.3x1010 (min-1) (Grohmann et al., 1985); E2 is the activation energy for the galactose degradation reaction = 100 kJ mol1  (Grohmann et al., 1985); R is the universal gas constant = 8.3143x10-3 (kJ/(mol.K)); T is  the temperature (K). Figure 5.4 shows the galactose formation in the condensed steam at 180oC, 200oC and 220oC, respectively. The rate of galactose formation increased with the reaction temperature Figure 4 shows that the production of galactose decreases with increasing temperature. At 10 min, xL = 0.012 g/100g at 180oC as compared to below 0.001 g/100g at 220oC. For 200oC reaction temperature, the model predicted that the rate of hydroxylmethylfurfural (HMF) formation was dominant over the rate of galactose generation reaction after 5 minute reaction time. Similar case was applied to the 220oC reaction temperature, and the galactose concentration increased to the peak at t = 3 minute and decreased rapidly afterwards. This was due to the consumption of galactose and degradation into HMF.  5.4.5 Hydroxylmethylfurfural Production from Galactose Degradation The amount of hydroxylmethylfurfural (HMF) is estimated as the difference between the initial amount of galactan and the undepolymerized galactan inside the solid and the water soluble galactose inside the condensed steam after steaming.  128  Combining equations (5-8) and (5-11) yields, L  1  x R  x L   or   k x   L  1  x10  e k1t   1 10   e k1t  e k2t  k 2  k1        (5-13)  where L is the 5-hydroxylmethylfurfural formation by degradation reaction of galactose relative to initial weight of galactan in starting material (g/100g feed); xL is the net sugar yield-percentage galactose which is soluble in liquid (g/100g of feed). Figure 5.5 shows the HMF formation at temperatures of 180oC, 200oC and 220oC, respectively. The model predicted higher amount of HMF formation than the experimental values. This was due to the loss of HMF in vapour form during steam explosion. A complete mass balance of HMF by analyzing the gaseous component in the steam is required for future work.  5.4.6 Kinetic Model of Lignin Solubilization and Condensation Vazquez et al. (1997) reported that the kinetics of delignification of Pinus pinaster wood followed a model of consecutive first order reactions. These reactions consisted of lignin solubilization followed by lignin condensation and precipitation onto the lignocellulosic matrix. The hydyolysis of α-aryl ether bonds was the dominant reaction of the deligninfication process.  Lignin (HMW)  kd  Lignin (LMW)  kc1  Lignin (Condensed)  (5-14)  where HMW is the high molecular weight lignin before depolymerization, LMW is the low molecular weight fraction after depolymerization and condensed lignin is the amount of lignin repolymerized from LMW lignin.  129  Vasquez et al. (1997) did not include the condensed lignin formed from the reaction of low molecular weight lignin with the furfural. This condensation can be represented by  dCl  k d Cl dt dCls  k d C l  k c1C ls dt dClc  k c1C ls dt  (5-15) (5-16) (5-17)  where Cl is the concentration of high molecular weight lignin (g/100g of feed), Cls is the concentration of the solubilized lignin (low molecular weight lignin) (g/100g of feed), Clc is the concentration of the condensed lignin (g/100g of feed), kd is the rate constants of solubilization reaction of the lignin (min-1) and kc1 is the rate constants of lignin condensation reaction (min-1). Figure 5.6 and Figure 5.7 show the acid insoluble lignin kinetics and acid soluble lignin kinetics at 180oC, 200oC and 220oC, respectively. At 180oC, the original lignin depolymerized to generate the acid soluble lignin quickly in the presence of acid at time = 3 minute. Due to the relative slow condensation/repolymerization reaction rate at 180oC, the rate of condensed lignin formation was slow. The acid insoluble lignin which comprised the condensed lignin and ash gradually increased with time. The condensation reaction rates were fast at 200oC and 220oC. Therefore, the acid soluble lignin formed throughout the reaction time at 200oC and 220oC was small compared to the low temperature at 180oC.  5.4.7 New Mathematical Kinetic Model to Predict the Pseudolignin Formation In the previous work (Lam et al., 2009), we proposed an additional reaction to the kinetic model in which the pseudolignin of hardwood aspen is formed by the process of solubilized low molecular weight lignin reacting with furfural (Miranda et al., 1979). A  130  second order reaction would be essential for quantifying the condensed lignin produced from self-condensation by reaction (5-14) and for explaining the increase in the pseudolignin content after steam treatment. The reaction mechanism of softwood we attempted here using HMF instead of furfural is due to a lower concentration of C5 sugars in softwood than in hardwood. The major degraded components of C6 sugars from softwood by hydrolysis were HMF. Therefore, the following reaction pathway is proposed (5-18):  Lignin (LMW)  + HMF  kc2  Lignin (Condensed)  (5-18)  Combining the reaction pathways (5-14) and (5-18), the reaction rates can be presented by:  dCls  k d C l  k c1Cls  k c 2 C ls C f dt  (5-19)  dClc  k c1Cls  k c 2 Cls C f dt  (5-20)  dC f dt   k 2 x L  k c 2 Cls C f  (5-21)  where Cl is the concentration of high molecular weight lignin (g/100g of feed), Cf is the concentration of hydroxylmethylfurfural (HMF) determined from the galactose degradation model (g/100g of feed), Cls is the concentration of the solubilized lignin (low molecular weight lignin) (g/100g of feed), Clc is the concentration of the condensed lignin (g/100g of feed), kd is the rate constant of solubilization reaction of the lignin, kc1 is the rate constant of lignin self-condensation reaction (min-1) and kc2 is the rate constant of lignin condensation reaction with hydroxylmethylfurfural (min-1). The pseudolignin (defined as the sum of Cl, Cls,and Clc), formation kinetics at 180oC, 200oC and 220oC is shown in Figure 5.8. The pseudolignin concentration increased slightly with reaction time at all temperatures. In general, the new kinetic  131  model showed a reasonable trend with the increase in pseudolignin formation for softwood. At 180oC, the model values were higher than the experimental data. The concentration of the pseudolignin was higher at higher reaction temperatures of 200oC and 220oC, respectively. However, the model predicted values of pseudolignin formation at reaction temperatures of 200oC and 220oC were lower than the experimental values. This could imply that the increase in lignin content by mass was also contributed by the reactions of solubilized lignin with other degraded components derived from the sugar other than galactan. Future work can be considered on identifying the multicomponent degraded products from different polysaccharides (arabinan, mannan, xylan) by experiment and modeling the increase in pseudolignin by reactions between multicomponents degraded products and the solubilized lignin.  5.5 Conclusions A kinetic model of pseudolignin formation during steam explosion was developed. The model was used to predict the formation of pseudolignin and to evaluate the optimized treatment condition for making durable and water repellent wood pellets. The pseudolignin contents predicted by the model were obtained by solving several non-linear ODEs by the Matlab explicit solver (ode45). The solutions were compared with the experimental values. The predicted galactan and pseudolignin contents were obtained by the reaction rate constant of hydroxylmethylfurfural and low molecular weight lignin condensation (kc2) by curve fitting with the least sum of square error. The additional kinetic model showed a reasonable trend in the increase in pseudolignin formation for the softwood. The model predicted values of pseudolignin formation at reaction temperatures of 200oC and 220oC were lower than the experimental values. This may imply that the increased in lignin content by mass was also contributed by the reactions of solubilized lignin with other degraded components derived from the sugar other than galactan. Future work can be considered on identifying the multicomponent degraded products from different polysaccharides (arabinan, mannan, xylan) by experiment and modeling the  132  increase in pseudolignin by reactions between multicomponent degraded products and the solubilized lignin.  5.6 Notation Ao  pre-exponential factor (min-1)  Cf  concentration of hydroxylmethylfurfural (g/ 100g of feed)  Cl  concentration of the Klason lignin (High molecular weight lignin) (g/100 g of feed)  Clc  concentration of the condensed lignin (g/100g of feed)  Cls  concentration of the solubilized lignin (low molecular weight lignin) (g/100g of feed)  Eact  activation energy (kJ/mol)  kd  rate constants of solubilization reaction of the lignin (min-1)  kc1  rate constant of lignin self-condensation reaction (min-1)  kc2  rate constant of lignin condensation reaction with 5-hydroxylmethylfurfural (min-1)  kf  fast reacting galactan depolymerization rate constant (min-1)  ks  slow reacting galactan depolymerization reaction rate constant (min-1)  k1  the galactan depolymerization reaction rate constant (min-1)  k2  reaction rate constant for degradation to byproduct 5-hydroxylmethylfurfural (HMF) (min-1)  k3  reaction rate constant for further degrading HMF to other product (min-1)  R  universal gas constant = 8.3143x10-3 (kJ(mol.K))  T  temperature (K)  t  time (min)  x1o  the initial weight of galactan remaining in the solid phase at t= 0 min (g/100g feed)  xL  the net sugar yield-percentage galactose which is soluble in liquid (g/100g of feed)  xR  the galactan remained in the solids phase at any time (g/100g feed)  133  Tables Table 5.1 Chemical Composition of the Solid Phase of the Untreated and Steam Treated Douglas Fir based on the Steam Treated Material (n=2) Acid insoluble lignin  Acid soluble lignin  Temperature  Time  Xylan  Mannan  Total  (oC)  (min)  Untreated  0  26.00  0.34  0.92  2.65  54.61  4.42  10.02  98.96  3  16.90  0.32  0.51  2.41  56.39  4.35  10.30  91.18  5  25.40  0.32  0.36  2.06  56.54  4.40  10.18  99.26  10  26.60  0.33  0.32  1.97  57.78  4.52  9.93  101.45  3  26.80  0.32  0.27  1.79  57.03  4.18  9.63  100.02  5  28.30  0.31  0.28  1.68  58.86  4.28  9.61  103.32  10  31.70  0.43  0.21  1.09  58.9  3.26  7.32  102.91  3  33.27  0.21  0.16  0.94  56.84  2.61  6.25  100.28  5  32.20  0.32  0.12  1.03  60.31  3.46  7.11  104.55  10 35.70 0.47 0.07 0.30 Untreated : original sample without steam explosion treatment; n: number of measurements.  57.34  2.14  3.94  99.96  180  200  220  Arabinan  Galactan  Glucan  (g/100 g of product)  Table 5.2 Chemical Composition of the Solid Residue Part of the Untreated and Steam Treated Douglas Fir based on the Feedstock prior to Steam Explosion (n=2) Acid insoluble lignin  Acid soluble lignin1  Temperature  Time  (oC)  (min)  Untreated  0  26.02  0.35  0.92  2.65  3  14.47  0.27  0.44  5  22.37  0.28  10  23.99  3  180  200  220  Arabinan  Galactan  Glucan  Xylan  Mannan  Total  54.62  4.42  10.02  98.98  2.06  48.22  3.72  8.81  77.99  0.32  1.82  49.87  3.88  8.98  87.52  0.29  0.28  1.77  52.02  4.07  8.94  91.36  22.14  0.26  0.22  1.48  47.04  3.44  7.94  82.53  5  22.45  0.25  0.22  1.33  46.62  3.39  7.61  81.87  10  26.26  0.36  0.18  0.90  48.82  2.71  6.07  85.29  3  22.56  0.28  0.11  0.64  38.54  1.77  4.24  68.13  5  21.29  0.22  0.08  0.68  39.90  2.29  4.71  69.16  (g/100 g of feed)  10 25.39 0.33 0.05 0.21 40.83 1.52 2.81 71.14 Untreated : original sample without steam explosion treatment; n: number of measurements. 1 : This is the acid soluble lignin left in the solid residue. The acid soluble lignin in the liquid filtrate (i.e. the difference between the untreated sample and the concentration of the acid soluble lignin in the solid residue at different treatment condition) is reported and plotted in Figure 5.7.  134  Table 5.3 Chemical Composition of the Water Soluble Fraction of the Steam Treated Douglas Fir in g/100 g Feed (n=2) Temperature (oC)  Time (min) 3  Arabinose  Galactose  Glucose  Xylose  Mannose  0.12  0.38  0.21  0.00  0.00  5  0.22  0.68  0.32  0.31  0.86  10  0.39  1.19  0.42  0.56  1.38  3  0.10  0.29  0.34  0.31  0.82  5  0.18  0.56  0.44  0.47  1.36  10  0.07  0.22  0.82  0.47  2.02  3  0.02  0.06  0.34  0.20  0.69  5  0.17  0.51  0.99  0.82  3.31  10 n: number of measurements  0.02  0.07  0.52  0.16  1.26  180  200  220  Table 5.4 Estimated Arrhenius Parameters of Carbohydrates Hydrolysis Rate Constant (n=2) Arrhenius parameters Arabinan Ai 8.23 ± 0.03 Eact 39.83 ± 0.16 Ai – pre-exponential factor in natural logarithm (min-1) Eact – Activation energy (kJ/mol) n: number of measurements  Galactan 20.02 ± 1.57 89.22 ± 6.24  Mannan 49.86 ± 19.68 212.57 ± 78.85  Table 5.5 Fitted Kinetic Parameters of Rate Constant for HMF Formation, Lignin Depolymerization and Lignin Condensation at Different Temperature Species Galactan (k1) HMF (k2) Acid insoluble Lignin (kd) Condensed lignin (kc1) Condensed lignin (kc2)  180oC 0.0531 0.0051 0.0974 0.2031 0.2053  Rate constant (min-1) 200oC 0.1361 0.3938 0.2058 0.2919 0.2972  220oC 0.3661 0.6604 0.4087 0.3457 0.4184  135  Figures  Figure 5.1 5-Hydroxylmethylfurfural Formation at Different Temperature  Figure 5.2 Arrhenius Plot of Reaction Rate of Arabinan, Galactan and Mannan in Natural Logarithm at Different Temperature  136  0.03 Model 180oC Expt. 180oC  0.025 Galactan (XR) (g/100g of feed)  Model 200oC Expt. 200oC 0.02  Model 220oC Expt. 220oC  0.015  0.01  0.005  0  0  1  2  3  4  5 6 Time (min)  7  8  9  10  Figure 5.3 Galactan Depolymerization at Different Temperature  0.012 Model 180oC Expt. 180oC  Galactose (XL) (g/100g of feed)  0.01  Model 200oC Expt. 200oC 0.008  Model 220oC Expt. 220oC  0.006  0.004  0.002  0  0  1  2  3  4  5 6 Time (min)  7  8  9  10  Figure 5.4 Galactose Formation and Degradation at Different Temperature  137  0.025 Model 180oC Expt. 180oC  HMF (L) (g/100g of feed)  0.02  Model 200oC Expt. 200oC Model 220oC  0.015  Expt. 220oC  0.01  0.005  0  0  1  2  3  4  5 6 Time (min)  7  8  9  10  Figure 5.5 Hydroxylmethylfurfural (HMF) Formation at Different Temperature  0.28  Acid Insoluble lignin (Cl) (g/100g of feed)  Model 180oC 0.26  Expt. 180oC Model 200oC Expt. 200oC  0.24  Model 220oC Expt. 220oC  0.22  0.2  0.18  0.16  0  1  2  3  4  5 6 Time (min)  7  8  9  10  Figure 5.6 Acid Insoluble Lignin Depolymerization and Condensation at Different Temperature  138  0.14 Model 180oC Acid soluble lignin (Cls ) (g/100g of feed)  0.12  Expt. 180oC Model 200oC Expt. 200oC  0.1  Model 220oC Expt. 220oC  0.08  0.06  0.04  0.02  0  0  1  2  3  4  5 6 Time (min)  7  8  9  10  Figure 5.7 Acid Soluble Lignin Formation in Liquid Phase at Different Temperature  0.4  Pseudolignin (g/g of product)  0.35  0.3  Model 180oC  0.25  Expt. 180oC Model 200oC Expt. 200oC  0.2  Model 220oC Expt. 220oC 0  1  2  3  4  5 6 Time (min)  7  8  9  10  Figure 5.8 Pseudolignin Formation at Different Temperature  139  Chapter 6 Conclusions and Future Research 6.1 Overall Conclusions Biomass preprocessing is a critical step in producing high quality feedstock economically for bioenergy. In particular, pellets made from biomass have low moisture content (around 6 – 8 % wet basis) and high bulk density (more than 650 kg/m3). The increase in bulk density is favorable for transport and safe storage. Biomass pellets are usually up to 6 mm in diameter and 12 mm long. For pellets made from wood, they are regarded as a clean fuel for combustion with low ash content (less than 0.5% by weight) and a high heating value around 18 MJ/kg. However, the pellets are easily disintegrated into fines due to impact force or moisture sorption during handling and storage. Fines aggravate accumulation of dust that may lead to explosion, off-gasing and selfcombustion. This threatens the occupational health and the workers’ safety in this industry. All of these are the huge barriers in the growth of biomass and bioenergy industry. The present study investigates the use and optimization of steam explosion pretreatment to improve the pellet quality in terms of mechanical strength and moisture sorption resistance. The effects of steam explosion processing conditions and feedstock parameters on physical and chemical properties of one species of woody biomass (Douglas Fir) were studied. The experiments were carried out with a design of experiment using saturated steam at 2 levels of temperature for 2 treatment durations and 3 replicates. A treatment severity was defined as an integral function of treatment temperature and time. It was found that the bulk density and tapped density of steam treated wood powder increased by 251% and 223%, respectively, for the most severe treatment at a severity of 4.53. Four factors significant on the density of steam treated wood powder were identified by Analysis of Variance (ANOVA) were temperature, time, particle size and moisture content. The model equations of steam explosion of woody biomass with additional terms (particle size and moisture content) predicting bulk, tapped density, geometric mean diameter of treated samples under different severity by multiple linear regression (MLR) were developed.  140  Size reduction effect of biomass particles by different severity of steam explosion was investigated in depth by conventional sieving and scanner imaging method. This helps to further understand and explain the increase in particle packing density for steam treated wood with the analysis of particle size and shape alternations after steam explosion. Particle size and the aspect ratio of steam exploded wood particles decreased with increasing the steam explosion pretreatment severities. Fragmentation of large particles during steam explosion was found by both analysis techniques. This is caused by devolatilization at high pressure of steaming. Initial increase in particle size was due to particle swelling by volatile loss during steam treatment. Particle shape changed from angular shape to a lower energy state of spherical shape. This explains the increase in bulk density after steam explosion pretreatment. Particle surface was found to be rougher after steam explosion pretreatment by scanning electron micrographs. The rapid escape of steam inside the internal pores of the particles detaches the particles surface during the rapid decompression. This led to the detachment of the surface layer and formed the fragments of the wood fiber. Surface pores were found on the steam exploded wood fibers. This may enhance the combustion performance of steam exploded wood particles due to the increase of surface irregularity. The steam exploded wood powder treated at different severities and the untreated wood powder were pelletized by a single die system with temperature and compression force control. The resulted pellets were characterized in terms of mechanical strength and moisture sorption behavior. The results demonstrated that pellets treated with steam explosion showed an improvement in mechanical strength, better hydrophobicity, less expansion in longitudinal direction after relaxation and slightly higher high heating value (HHV) than the pellets made from untreated powder. Steam treated materials became harder (i.e., higher Meyer hardness) and higher rigidity (i.e., lower asymptotic modulus) with increasing steam explosion pretreatment. However, it showed an increase in energy requirement for compression and extrusion.  141  A study was conducted to model the mass and energy balance of the overall biomass pretreatment process with steam explosion treatment based on a wood pellet plant in B.C. in Canada. The energetic data taken from experimental data of steam explosion pretreatment and pelletization using a single die system were used in this process modeling. The quality of the pellets and the energy consumption for the whole process at different scenario were discussed. In general, the best pellet quality for handling and transportation was the pellets treated with steam explosion at 220oC for 5 minutes (i.e., severity = 4.23). The energetic ratio of the overall biomass densification process including this severity steam explosion pretreatment is 2.41. A kinetic model of polysaccharides depolymerization and pseudolignin formation of Douglas Fir in the steam explosion was developed. The kinetic model can be used to evaluate the optimized pretreatment condition for the biomass fibers to make durable and water resistant biomass pellets or to minimize the pseudolignin formation, which is unfavorable for bioethanol fermentation. The proposed pseudolignin formation model is a combination of the first order reaction of galactan depolymerization in the autohydrolysis and the first order reaction of lignin solubilization and condensation in the acetic acid. An additional  reaction  pathway  including  the  hydroxylmethylfurfural  and  lignin  polymerization is introduced to form a new model. The predicted galactan remained inside the solid residue after steam explosion fitted well to the experimental values. The predicted pseudolignin content showed an increasing trend throughout the reaction time at three different temperatures. However, the values were underpredicted. This suggested that the increases in lignin content by mass were also contributed by the reactions with other degraded components derived from the sugar other than galactan. Future work could be considered to work on modeling the increase in pseudolignin by reactions between multicomponent degraded products from different polysaccharides (arabinan, mannan, xylan) and the solubilized lignin.  142  6.2 Recommendations for Future Work The current study investigated the effect of steam explosion and its mechanism on improving the wood pellet quality for handling and storage ability. Much work yet to be done in this area. The following works are recommended for future investigations derived from this study: 1. In the course of this study, it became evident that the drying rate of the steam treated particles decreased with treatment severity. Similar trend was observed for pellets made from steam treated biomass where more energy needed to form pellets. Decreased drying rate and increased energy requirement affect the economics of the steam explosion and densification. Future experiments need to be designed to understand the processes that govern the decrease in drying rate and increased energy to make pellets. 2. The effect of steam to biomass ratio on the reaction rate during the steaming process is not well understood. If the concentration of steam is too low (i.e. superheated steam), there may not be enough moisture to condense on particle surface and facilitate the hydrolysis reactions during steaming. If the steam supply is too high, the operation cost of the process will increase. There is a need to study the optimized steam to biomass ratio for the best pellet quality. 3. The current study was based on the Douglas Fir in British Columbia. Other wood species, e.g. Pine, Spruce, Poplar and Willow, would have different physical and chemical properties brought about by steam explosion. Further research towards understanding the effect of steam explosion on forming durable pellets from mixed wood species and other lignocellulosic biomass is essential. 4. The particles used in the current study were ground particles. The powder generated from wood chips after steam explosion can be made to pellets and evaluated for their properties. Larger opening of the steam treatment reactor needs to be used in order to perform the experiments. 5. More work on modeling the increase in pseudolignin by reactions between multicomponent degraded products from different polysaccharides (arabinan, mannan, xylan) and the solubilized lignin.  143  6. The effect of steam explosion at different severities and its optimization on pellet quality was investigated in the current study. However, the optimization of pelletization on steam exploded wood powder can be carried out. The proposed study can be the effect of several different die temperatures, different moisture content of feedstock and different compressive forces on pellet density. 7. Compaction behavior of the steam exploded wood pellets treated at different severities was different. Future work can be done to compare the compression data from the experiment with the predicted values with the model equation for biomass densification. 8. There is a strong odor from steam exploded wood powder and wood pellets. The most severely treated wood pellets gave out the strongest smell. It is important to evaluate the storage ability of the steam exploded pellets made at different severities by offgasing and self-heating experiments. Identification of any toxic offgases is important. 9. There is a volatile loss after steam explosion pretreatment for the wood particles by proximate analysis. It may imply that the combustion behavior of steam exploded wood pellets possess a better performance with less tar formation in the exhausts. This is beneficial to improve the product image for household heating application. Future work can be done on evaluating the combustion performance and gas emissions of steam exploded wood pellets and compared to the untreated wood pellets made at the same condition. 10. The surface structure of the steam exploded wood particles was found to have more pores in this study. High porous particles may increase the gasification efficiency. The effect of the steam exploded wood particles made at different severities on gasification can be evaluated. 11. Wood pellets have uniform physical properties. They are a good feedstock supply for biorefineries. In addition, steam explosion had been studied as an economical pretreatment prior to ethanol fermentation. The evaluation of steam exploded wood pellets made at different severities as a feedstock for ethanol production can be studied.  144  12. The amount of steam used for treating biomass was not optimized in this research. Provisions for recovering heat from wasted heat from spent steam was not considered. The numbers on energy use for steam explosion represented a worse scenario. A more accurate mass and energy balance with a sensitivity analysis on energy input to energy content of pellet is recommended.  145  References Abatzoglou N., E. Chornet, K. Belkacemi, R. P. Overend. 1992. Phenomenological kinetics of complex systems: The development of a generalized severity parameter and its application to lignocellulosics fractionation. Chemical Engineering Science. 47(5):1109-1122. Adapa P., L. Tabil, G. Schoenau. 2010a. Physical and frictional properties of non-treated and steam exploded barley, canola, oat and wheat straw grinds. Powder Technology. 201(3): 230-241. Adapa P., L. Tabil, G. Schoenau, A. Opoku. 2010b. Pelleting characteristics of selected biomass with and without steam explosion pretreatment. International Journal of Agriculture and Biological Engineering. 3(3):62-79. Adapa P., L. Tabil, G. Schoenau. 2010c. Compression characteristics of non-treated and steam exploded Barley, Canola, Oat and Wheat Straw grinds. Applied Engineering in Agriculture. 26(4):617-632. Adler P. R., M. A. Sanderson, A. A. Boateng, P. J. Weimer and G. Jung. 2006. Biomass Yield and Biofuel quality of switchgrass harvested in fall or spring. Journal of Agronomy. 98:1518-1525. Alakangas E., J. Valtanen, Levlin Jan- Erik. 2006. CEN technical specification for solid biofuels – Fuel specification and classes. Biomass and Bioenergy. 30:908 – 914 Aleksandra S., A. M. Madsen, L. Martensson, D. Pomorska, L. Larsson. 2006. Assessment of microbial exposure risks from handling of biofuel wood chips and straw – effect of outdoor storage. Annals of Agricultural and Environmental Medicine. 13:139-145. Angles M. N., F. Ferrando, X. Farriol, J. Salvado. 2001. Suitability of steam exploded residual softwood for the production of binderless panels. Effect of the pre-treatment severity and lignin addition. Biomass and Bioenergy. 21(3): 211-224 Aqa S., S. C. Bhattacharya. 1992. Densification of preheated sawdust for energy conservation. Energy. 17(6):575 – 578  146  ASABE Standards 47th ed., 2000. S368.4 - Compression test of food materials of convex shape. 679-686. St. Joseph, MI.: ASABE. ASABE Standards 47th ed., 2003b. S269.4 - Cubes, pellets and crumbles-definitions and methods for determining density, durability and moisture content. 567-569. St. Joseph, MI.: ASABE. ASAE Standards 47th ed., 2006. S448.1 - Thin layer drying of agricultural crops. 622624. St. Joseph, MI.: ASABE. ASABE Standards 49th ed., 2006. EP496.3 - Agricultural machinery management. 385390. St. Joseph, MI.: ASABE. ASABE Standards 49th ed., 2008. S319.4 - Method for determining and expressing fineness of feed materials by sieving. 1-4. St. Joseph, MI.: ASABE. ASABE Standards 47th ed., 2010. S358.2 - Moisture measurement forages. 595. St. Joseph, MI.: ASABE. ASTM. 2009. Standard test methods for proximate analysis of the analysis sample of coal and coke by instrumental procedures, Vol. ASTM D5142 - 09, ASTM International. West Conshohocken, PA, USA. Back E. L., N. L. Salmen. 1982. Glass transitions of wood components hold implications for molding and pulping processes. TAPPI. 65(7):107-110. Ballesteos J. M., M. J. Oliva, P. M. Negro, M. Ballesteros. 2002. Enzymatic hydrolysis of steam exploded herbaceous agricultural waste (Brassica carinata) at different particle sizes. Process Biochemistry. 38:187 – 192. Bari I.D., F. Nanna, G. Braccio. 2007. SO2-catalyzed steam fractionation of aspen chips for bioethanol production: optimization of the catalyst impregnation. Industrial & Engineering Chemistry Research. 46:7711-7720. Bayless D. J., A. R. Schroeder, J. E. Peters, R. O. Buckius. 1997. Effects of surface voids on burning rate measurements of pulverized coal at diffusion-limited conditions. Combustion and Flame. 108(1-2):187-198.  147  Bekhta P., P. Niemz. 2003. Effect of high temperature on the change in colour, dimensional stability and mechanical properties of spruce wood. Holzforschung. 57:539-546. Bernstein M., P. Bromley, J. Hagen, S. Hassell, R. Lempert, J. Munoz, D. Robalino. 1999. RAND, Developing Countries & Global Climate Change: Electric Power Options for Growth, Report prepared for the Pew Center on Global Climate Change Bhattacharya S. C., S. Sett, R. M. Shrestha. 1989. State of the art for biomass densification. Energy Sources. 11:161-182. Bhuiyan M. T. R., N. Hirai, N. Sobue. 2000. Changes of crystallinity in wood cellulose by heat treatment under dried and moist conditions. Journal of Wood Science. 46:431436. Biagini E., P. Narducci, L. Tognotti. 2008. Size and structural characterization of lignincelluosic fuels after the rapid devolatilization. Fuel. 87:177 – 186. Biermann C. J., D. McGinnis and P. Schultz. 1987. Scanning electron microscopy of mixed hardwoods subjected to various pretreatment process. Journal of Agricultural and Food Chemistry. 35(5):713-716. Bouajila J., A. Limare, C. Joly, P. Dole. 2005. Lignin plasticization to improve binderless fiberboard mechanical properties. Polymer Engineering and Science. 45(16):809-816. Boussaid A., R. Esteghlalian, J. Gregg, K. H. Lee, J. N. Saddler. 2000. Steam pretreatment of Douglas Fir Wood Chips. Applied Biochemistry and Biotechnology. 8486:693-705. Bouwman A. M., J. C. Bosma, P. Vonk, J. A. Wesselingh, H. W. Frijlink. 2004. Which shape factors best describe granules. Powder Technology. 146:66-72. Brownell H. H., E. K. C. Yu, J. N. Saddler. 1986. Steam explosion pretreatment of wood: Effect of chip size, acid, moisture content and pressure drop. Biotechnology and Bioengineering. 28(6):792 – 801. Brownwell H.H., J. N. Saddler. 1984. Biotechnology and Bioengineering. Symposium Series 14, 55.  148  Bruno E. A., V. Marques, I. Domingos, H. Pereira. 2007. Influence of steam heating on the properties of pine (Pinus pinaster) and eucalypt (Eucalyptus globulus) wood. Wood Science and Technology. 41:193-207. Bruno E., A. V. Marques, I. Domingos, H. Pereira. 2008a. Heat-induced colour changes of pine (Pinus pinaster) and eucalypt (Eucalyptus globulus) wood. Wood Science and Technology. 42:369-384. Bruno E., J. Graca, H. Pereira. 2008b. Extractive composition and summative chemical analysis of thermally treated eucalypt wood. Holzforschung. 62:344-351 Bura R., R. J. Bothast, S. D. Mansfield, J. N. Saddler. 2003. Optimization of SO2 catalyzed steam pretreatment of corn fiber for ethanol production. Applied Biochemistry and Biotechnology. 105-108:319-335. Bura R., S. D. Mansfield, J. N. Saddler, J. R. Bothast. 2002. SO2 catalyzed steam explosion of corn fiber for ethanol production. Applied biochemistry and biotechnology. 98-100:59-72. Canettieri E.V., G. J. M. Rocha, J. A. Carvalho, Jr. J. B. A. Silva. 2007. Evaluation of the kinetics of xylose formation from dilute sulfuric acid hydrolysis of forest residues of Eucalyptus grandis. Industrial Engineering Research Chemistry. 46:1938-1944. Carrasco F., C. Roy. 1992. Kinetic study of dilute acid prehydrolysis of xylan containing biomass. Wood Science Technology. 26:189-208. CEN Standard. CEN/TS 14961:2005 (E), Solid biofuels – Fuel Specification and Classes, CEN/TC 335 – Soild biofuels; 40 Chen C., D. S. Jayas. 1998. Evaluation of the GAB equation for the isotherms of agricultural products. Transactions of the ASAE. 41(6):1755-1760. Cherney J. H. 2007. Grass pelleting – The process, Bioenergy Information Sheet #7, Cornelll  University  Cooperation  Extension,  Available  at:  http://www.grassbioenergy.org/downloads/Bioenergy_Info_Sheet_7.pdf Accessed June 12, 2007  149  Colley Z., O. O. Fasina, D. Bransby, Y. Y. Lee. 2006. Moisture effect on the physical characteristics of switchgrass pellets. Transactions of ASABE. 49(6):1845-1851. Conner A. H. 1984. Kinetic modeling of hardwood prehydrolysis. Part I.: Xylan removal by water prehydrolysis. Wood Fiber Science. 16:268-277. Cordedo T., F. Marquez, J. M. Rodriguez, J. J. Rodriguez. 2001. Predicting heating values of carbonaceous materials from proximate analysis. Fuel. 80(11):1567-1571. Cullis I.F., J. N. Saddler, S. D. Mansfield. 2004. Effect of initial moisture content and chip size on the bioconversion efficiency of softwood lignocellulosics. Biotechnology and Bioengineering. 85(4):413-421. Delgenes J. P., M. C. Escare, J. M. Laplace, R. Moletta, J. M. Navarro. 1998. Biological production of industrial chemicals, i.e. xylitol and ethanol, from lignocelluloses by controlled mixed culture systems. Industrial Crops and Products. 7(2-3):101-111. DeLong, E.A. 1981. Method of rendering lignin separable from cellulose and hemicellulose in lignocelluosic material and the product so produced. Canadian Patent No. 1096374 DeLong, E. A. 1983. A method of rendering lignin separable from cellulose and hemicellulose in land the product so produced. Canadian Patent No. 1141376 DeLong, E. A., Delong, E.P., Ritchie, G. S., Rendall, W. A. 1990. Method for extracting the chemical components from dissociated lignocellulosic material. US Patent No. 4908098 Demirbas A. 1997. Calculation of higher heating values of biomass fuels. Fuel. 76(5):431-434. Demirbas A. 2001. Relationships between lignin contents and heating values of biomass. Energy Conversation & Management. 42(2):183-188. Dence C.W. 1992. The determination of lignin. Methods in Lignin Chemistry. 33-61. Springer, Berlin. Devore, J. L. 2004. Probability And Statistics For Engineering And The Sciences, 6th edition. Pacific Grove, CA: Thomson Books/Cole.  150  Ding T., L. Gu, T. Li. 2010. Influence of steam pressure on physical and mechanical of heat treated Mongolian pine lumber. European Journal of Wood and Wood Products. 69(1):121-126. DOE, Biomass as feedstock for a bioenergy and bioproducts industry: The technical feasibility of a billon ton annual supply. Oak Ridge, TN: U.S. Department of Energy (DOE),  Office  of  Scientific  and  Technical  Information.  Available  at:  http://www.osti.gov/bridge. Accessed 18 June 2005 Donohoe B. S., S. R. Decker, M. P. Tucker, M. E. Himmel, T. B. Vinzant. 2008. Visualizing lignin coalescence and migration through maize cell walls following thermochemical pretreatment. Biotechnology and Bioengineering. 101(5): 913-925. Efremov A. A., B. N. Kuznetsov, S. A. Kuznetsova, I. V. Krotova, V. A. Vink. 1995. Composition of the water-soluble products from the thermocatalytic activation of aspen wood. Chemistry of Natural Compounds. 31(6):746-752. Emmel A., A. L. Mathias, F. Wypych, L. P. Ramos. 2003. Fractionation of Eucalyptus grandis chips by dilute acid catalysed steam explosion. Bioresource Technology. 86(2):105-115. Esteban L. S., J. E. Carrasco. 2006. Evaluation of different strategies for pulverization of forest biomasses. Powder Technology. 166:139-151. Esteghlalian A., G. Hashimoto, J. Fenske, H. Penner. 1997. Modeling and optimization of the dilute sulfuric acid pretreatment of corn stover, poplar and switchgrass. Bioresource Technology. 59:129-136. Fengel D., G. Wegener. 1984. Wood: Chemistry, Ultrastructure, Reactions, Walter de Gruyter and Co. Berlin, New York Foody P. 1984. Method for increasing the accessibility of cellulose in lignocellulosic materials, particularly hardwoods agricultural residues and the like. US Patent No. 4461648 Forsythe G.E., M. A. Nalcolm., C. B. Moler. 1977. Computer Methods for Mathematical Computation, Prentice-Hall, Englewood Cliffs.  151  Franke M., A. Rey. 2006. Pelleting quality, World Grain. 78-79 Ghebre-Sellassie, I. 1989. Mechanism of pellet formation and growth, in pharmaceutical pelletization technology, I. Ghebre-Sellassie, ed., pp. 123- 143, New York, NY: Marcel Dekker, Inc. Goring DAI. 1963. Thermal softening of lignin, hemicelllulose and cellulose. Pulp & Paper Magazine Canada. 64(12):517-527. Gorrote G., H. Dominguez, J. C. Parajo. 1999. Hydrothermal processing of lignocellulosic materials. Holzforschung. 57:191-202. Govin A., Repellin V., Rolland M., Duplan J. 2009. Effect of torrefaction on grinding energy requirement for thin wood particle production. XII Congres de la Societe Francaise dde Genie des Procedes Poir relever les defies industriels du XXI siècle A la Croisee des Sciences et des Cultures, Marseille: France. Greene L., A. Schafer. 2003. Reducing Greenhouse Gas Emissions from US Transportation, Report prepared for the Pew Center on Global Climate Change Grethlein H.E., A.O. Converse. 1991. Common Aspects of acid prehydrolysis and steam explosion for pretreating wood. Bioresource Technology. 36:77-82. Grover, P.D., S. K. Mishra. 1996. Biomass Briquetting: Technology and Practices. Regional Wood Energy Development Program in Asia, Field Document No.46. Bangkok, Thailand: Food and Agriculture Organization of the United Nations Haase S. Assessment of Biomass Pelletization Options for Greensburg, Kansa, Technical Report NREL/TP-7A2-48073 May 2010. http:www1.eere.energy.gov/buildings/Greensburg/pdfs/48073.pdf (Accessed Jan 2011). Heitz M., E. Capek, P. Koeberie, J. Gange, E. Chornet. 1990. Section B, Biomass conversion: The integration of the stake pretreatment and the UdeS-SH process, University de Sherbrooke and Stake Technology (TP360.U582 1990) Helmisaari H.S. 1989a. Temporal variation in nutrient concentrations of Pinus sylvestris needles. Scandinavian Journal of Forest Research. 5(2):443-451.  152  Helmisaari H.S., T. Siltala. 1989b. Variation in nutrient concentrations of Pinus sylovestris stems. Scandinavian Journal of Forest Research. 4(4):443-451. Holtzapple M. T., A. E. Humphrey, J. D. Taylor. 1989. Energy Requirements for the Size Reduction of Poplar and Aspen Wood. Biotechnology and Bioengineering. 33:207-210. Hosseini S. A., R. Lambert, S. Kucherenko, N. Shah. 2010. Multiscale modeling of hydrothermal  pretreatment:  From hemicellulose  hydrolysis  to  biomass  size  optimization. Energy and Fuels. 24(9):4673-4680. Hsu, W.E, W. Schwald, J. Schwald, J. A. Shields. 1988. Chemical and physical changes required for producing dimensionally stable wood-based composites. Wood Science and Technology. 22: 281–289. Ibrahim M., W. G. Glasser. 1999. Steam assisted biomass fractionation. Part III: a quantitative evaluation of the “clean” fractionation concept. Bioresource Technology. 70(2):181-192 Igathinathaine C., L. O. Pordesimo, E. P. Columbus, W.D. Batchelor, S. Sokhansanj. 2009a. Sieveless particle size distribution analysis of particulate materials through computer vision. Computers and Electronics in Agriculture. 66: 147 – 158. Igathinathane C., S. Melin, S. Sokhansanj, X. Bi, C. J. Lim, L. O. Pordesimo, E. P. Columbus. 2009b. Machine vision based particle size and size distribution determination of wood and bark pellets. Powder Technology. 196 (2): 201-212. Jenkins B.M., L. L. Baxter, T. R. Miles Jr., T. R. Miles. 1998. Combustion properties of biomass. Fuel Processing Technology. 54(1-3):17-46. Jennings B. R., K. Parslow. 1988. Particle size measurement: the equivalent spherical diameter. Proceedings of the Royal Society of London. Series A, mathematical and Physical Sciences. 419: 137-149. Jimennez L., F. Gonzales. 1991. Study of the physical and chemical properties of lignocelluosic residues with a view to the production of fuels. Fuel. 70:947-950.  153  Kaliyan N., and Morey R.V. 2005. Densification of corn stover. ASAE 2005 Annual International Meeting, Paper No. 056134., Tampa, Florida, July 17 – July 20. St. Joseph, Mich.: ASAE. Kaliyan N. R., V. Morey. 2006. Factors affecting strength and durability of densified products. ASABE 2006 Annual International Meeting, paper number: 066077, Portland, Oregon, July 9 – July 12. St. Joseph, Mich.: ASABE. Kaliyan N., Morey, R. V. 2006. Densification characteristics of corn stover and switchgrass. ASABE paper number: 066174, Portland, Oregon, July 9 – July 12. St. Joseph, Mich.: ASABE. Kettunen, P. O. 2006. Wood: Structure and properties. Trans Tech Publications Ltd., The University of Michigan. (TA419.K43 2006) Kim S.B., Y.Y. Lee. 1987. Kinetics in acid catalyzed hydrolysis of hardwood hemicellulose. Biotechnololgy Bioengineering Symposium. 17:71-84. Klass D. L. 1998. Biomass for renewable energy, fuels, and chemicals, San Diego, Academic Press Kobayashi F., M. Take, C. Asada, Y. Nakamura. 2004. Methane production from steam exploded bamboo. Journal of Bioscience and Bioengineering. 97(6):426-428. Kobayashi N., N. Okada, A. Hirakawa, T. Sato, J. Kobayashi, S. Hatano, Y. Itaya, S. Mori. 2009. Characteristics of Solid Residues Obtained from Hot-Compressed-Water Treatment of Woody Biomass. Industrial & Engineering Chemistry Ressearch. 48:373379. Kokta B. V. 1989. Process for preparing pulp for paper making. US Patent 4798651 Kokta B. V., A. Ahmed. 1998. Chapter 6, Steam Explosion Pulping, Environmentally friendly technologies for the Pulp and Paper Industry. John Wiley & Sons, Inc. Koran, Z. 1981. Energy consumption in mechanical fiber separation as a function of temperature. Pulp & Paper Magazine Canada. 82: TR 40- TR 44 Lam P. S., S. Sokhansanj, X. Bi, S. Mani, C.J. Lim, A.R. Womac, M. Hoque, J. Peng, T. JayaShankar, L. J. Naimi, S. Nayaran. 2007. Physical Characterization of wet and dry  154  wheat straw and switchgrass – bulk and specific density, ASABE 2007 annual international meeting, paper number: 076058, Minneapolis, Minnesota, June 17 – June 20. St. Joseph, Mich.: ASABE. Lam P. S., S. Sokhansanj, X. Bi, C. J. Lim, L. J. Naimi, M. Hoque, S. Mani, A. R. Womac, S. Narayan, X. P. Ye. 2008a. Bulk Density of wet and dry wheat straw and switchgrass particles. Applied Engineering in Agriculture. 24 (3): 351 – 358. Lam P. S., S. Sokhansanj, X. Bi, C. J. Lim 2008b. Effect of particle size and shape on physical properties of biomass grinds. ASABE 2008 annual international meeting, paper number: 080014, Providence, Rhode Island, June 29 – July 2. St. Joseph, Mich.: ASABE. Lam P. S., S. Sokhansanj, C. Jim Lim, X. T. Bi and S. Melin. 2009. Kinetic Modeling of Pseudolignin Formation in Steam Exploded Woody Biomass, 8th World Congress of Chemical Engineering, Montreal, Quebec, August 23 – August 27. Lehtikangas P. 2000. Storage effects on pelletised sawdust, logging residues and bark. Biomass and Bioenergy. 19(5):287-293. Lehtikangas P. 2001. Quality properties of pelletised sawdust, logging residues and bark. Biomass and Bioenergy. 20(5):351-360. Li J., G. Henriksson, G. Gellerstedt. 2007. Lignin depolymerization/repolymerization and its critical role for deligninfication of aspen wood by steam explosion. Bioresource Technology. 98(16):3061-3068. Li Y., H. Liu. 2000. High pressure densification of wood residues to form an upgrade fuel. Biomass and Bioenergy. 19(3):177-186. Liu H. W., H. K. Walter, G. M. Vogt, H. S. Vogt, B. E. Holbein. 2002. Steam pressure disruption of municipal solid waste enhances anaerobic digestion kinetics and biogas yield. Biotechnology and Bioengineering. 77(2):121-130. Liu H., Y. Li. 2000. Compacting biomass and municipal solid wastes to form an upgraded fuel, Final Technical Report, OSTI ID: 837464  155  Mabee W. E., D. J. Gregg, C. Arato, A. Berlin, R. Bura, N. Gilkes, O. Mirochnik, W. Pan, E. K. Pye, J. N. Saddler. 2007. Updates on softwood-to-ethanol process development. Applied Biochemistry and Biotechnology. 129(1-3):55-70. MacBain, R. 1966. Pelleting Animal Feed. Chicago, IL: American Feed Manufacturing Association. MacDonald A., M. Fernandez, B. Kreber. 1997. Chemical and UV-VIS spectroscopic study on kiln brown stain formation in Radiata pine. 9th international symposium of wood and pulping chemistry, Montreal, 70:1-5. Maloney M.T., T.W. Chapman, A.J. Baker. 1985. Dilute acid hydrolysis of paper birch: kinetic studies of xylan and acetyl-group hydrolysis. Biotechnology and Bioengineering. 27:335-361. Maloney M.T. 1986. An engineering analysis of the production of xylose by dilute acid hydrolysis of hardwood hemicellulose. Biotechnology Progress. 2:192-202. Maness P., S. Czernik, S. Smolinski. 2005. Fermentation Approaches to Hydrogen Production, DOE Hydrogen program, FY 2005 Progress Report, P.230-235. Mani S., L. G. Tabil, S. Sokhansanj. 2003. An overview of compaction of biomass grinds: Powder handling and processing. The International Journal of Storing, Handling, and Processing Powder. 15(3):160– 168. Mani S., L. G. Tabil, S. Sokhansanj. 2004a. Grinding performance and physical properties of wheat and barley straws, corn stover and switchgrass. Biomass and Bioenergy. 27(4):339-352. Mani S., L. G. Tabil, S. Sokhansanj. 2004b. Evaluation of compaction equations applied to four biomass species. Canadian Biosystems Engineering. 46:(3):55-61. Mani S., 2005. A systems analysis of biomass densification process. Ph.D. dissertation. Vancouver, Canada: University of British Columbia, Chemical and Biological Engineering Mani S., S. Sokhansanj, X. Bi, A. Turhollow. 2006a. Economics of producing fuel pellets from biomass. Applied Engineering in Agriculture. 22(3): 421-426.  156  Mani S., L.G. Tabil, S. Sokhansanj. 2006b. Effects of compressive force, particle size and moisture content on mechanical properties of biomass pellets from grasses. Biomass and Bioenergy. 30:648–654. Martinez M., B. Gullon, R. Yanez, J. L. Alonso, J. C. Parajo. 2010. Kinetic assessment on the autohydrolysis of pectin rich by products. Chemical Engineering Journal. 162:480-486. Miranda G. S., M. Wayman. 1979. Characterization of autohydrolysis aspen (P. tremuloides) lignins, Part 1, Composition and molecular weight distribution of extracted autohydrolysis lignin. Canadian Journal of Chemistry. 57:1141-1149. Mitsui K., H. Takada, M. Sugiyama, R. Hasegawa. 2001. Changes in properties of lightirradiated wood with heat treatment: Part 1 Effect of treatment conditions on the change in colour. Holzforschung. 55:601-605. Montane D., R. P. Overend, E. Chornet. 1998. Kinetic models for non-homogenous complex systems with a time dependent rate constant. The Canadian Journal of Chemical Engineering. 76:58-68. Moreyra R., M. Peleg. 1980. Compressive deformation patterns of selected food powders. Journal of Food Science. 45:864–868. Mosiera N., C. Wymanb, B. Dalec, R. Elanderd, Y. Y. Lee, M. Holtzapplef, M. Ladisch. 2005. Features of promising technologies for pretreatment of lignocellulosic biomass. Bioresource Technology. 96:673 – 686. Myers, R. H., D. C. Montgomery. 2003. Response surface technology: process and product optimization using designed experiments. Second ed. John Wiley & Sons Inc. Naimi L. J. 2008. Experiments and modeling of size reduction of switchgrass in a laboratory for rotary knife mill. M.A.Sc. dissertation. Vancouver, Canada: University of British Columbia, Chemical and Biological Engineering Negro M. J., P. Manzanares, J. M. Oliva, I. Ballesteros, M. Ballesteros. 2003. Changes in various physical/chemical parameters of Pinus pinaster wood after steam explosion pretreatment. Biomass and Bioenergy. 25(3): 301-308.  157  Nguyen Q. A., F. A. Keller, M. P. Tucker, C. K. Lombard, B. M. Jenkins, D. E. Yomogida, V. M. Tiangco. 1999. Bioconversion of mixed solids waste to ethanol. Applied Biochemistry and Biotechnology. 78 (1-3), 455-472. Nielsen N. P. K., D. J. Gardner, C. Felby. 2010. Effect of extractives and storage on the pelletizing process of sawdust. Fuel. 89:94-98. Nishiyama Y., P. Langan, H. Chanzy. 2002. Crystal structure and hydrogen-bonding system in cellulose Iβ from synchrotron X-ray and neutron fiber diffraction. Journal of the American Chemical Society. 124: 9074–9082. NREL Biomass Feedstock Composition and Property Database. HTTP available: http://www1.eere.energy.gov/biomass/feedstock_databases.html. Accessed Apr 13th, 2009. Obernberger I., G. Thek. 2004. Physical characterization and chemical composition of densified biomass fuels with regards to their combustion behavior. Biomass and Bioenergy. 27:653-669. Overend R. P., E. Chornet. 1987. Fractionation of lignocellulosics by steam aqueous pretreatments. Philosophical Transactions of the Royal Society of London A. 321:523536. Payne, J. D. 1978. Improving quality of pellet feeds. Milling Feed and Fertilizer. 162:3441 Peksa B. M., P. Dolzan, J. Heinimo, M. Junginger, T. Ranta, A. Walter. 2007. Global Wood Pellets Markets and Industry: Policy Drivers, Market Status and Raw Material Potential.  IEA  Bioenergy  Task  40  technical  report.  HTTP  available:  http://www.bioenergytrade.org/downloads/ieatask40pelletandrawmaterialstudynov2007 final.pdf Peleg M. 1980. Linearization of relaxation and creep curves of solid biological materials. Journal of Rheology 24(4):451–463. Peterson, A. A., R. P. Lachance, J. W. Tester. 2010. Kinetic evidence of the maillard reaction in hydrothermal processing: Glucose-Glycine interactions in high-temp high pressure water. Industrial & Engineering Chemistry Research. 49(5):2107-2117. 158  Pietsch W. 1997. Size enlargement by agglomeration, In Handbook of Powder Science and Technology, 2nd edition, eds. M.E. Fayed and L. Otten, 202-377. Florence, KY: International Thomson Publishing Podczeck F., J. M. Newton. 1994. A shape factor to characterize the quality of spheroids. International Journal of Pharmcology. 46:82-85. Podczeck F., S. R. Rahman, J. M. Newton. 1999. Evaluation of a standardized procedure to assess the shape of pellets using image analysis. International Journal of Pharmacy. 192:123-138. Polagye B. L., K. T. Hodgson, P. C. Malte. 2007. An economic analysis of bio-energy options using thinnings from overstocked forests. Biomass and Bioenergy. 31:105-125. Quan J. R. Y., R. Samson. (2007) A process and energy analysis of pelletizing switchgrass, Ste. Anne de Bellevue, QC: Resource Efficient Agricultural Production (REAP-Canada), available at: http://www.reap-canada.com/library.htm. Accessed June 8th, 2007 Quantachrome Instruments, Multipycnometer Operating Manual, Boynton Beach, FL 33426. Ramos L. P. 2003. The chemistry involved in the steam treatment of lignocellulosic materials. Quimica Nova. 26(6):863-871. Rumpf H. 1962. The strength of granules and agglomerates, Agglomeration. Edited by W.A. Knepper, John Wiley and Sons, New York, 379-419. Rupar K., M. Sanati. 2005. The release of terpenes during storage of biomass. Biomass and Bioenergy, 28(1):29-34. Saddler J. N., M. Mes-Hartree, E.K. C. Yu, H. H. Brownell. 1983. Biotechnology and Bioengineering, Symposium Series. 13:225. Sannigrahi, P., D. H. Kim, S. Jung, A. Ragauskas. 2011. Pseudo-lignin and pretreatment chemistry. Energy & Environmental Science. 4:1306-1310. Sastry, K.V.S., D. W. Fuerstenau. 1973. Mechanisms of agglomerate growth in green pelletization. Powder Technology. 7:97-105.  159  Sawada T., Y. Nakamura. 2001. Low energy steam explosion treatment of plant biomass. Journal of Chemical Technology and Biotechnology. 76(2):139-146. Scoville E., M. Peleg. 1981. Evaluation of the effects of liquid bridges on the bulk properties of model powders. Journal of Food Science. 46(1):174–177. Sehistedt-Persson. 2003. Colour responses to heat treatment of extractives and sap from pine and spruce. 8th International IUFRO wood drying conference, Brasov, 459-464. Sendelius J. 2005. Master of Science Thesis: Steam pretreatment optimization for sugarcane bagasse in bioethenol production. Department of Chemical Engineering, Lund Institute of Technology, Sweden. Sergio A.P., B. M. Jose, J. O. E. Francisco. 2003. Image analysis of the shape of granulated powder grains. Journal of Pharmaceutical Sciences. 93(3):621-634. Shajizadeh F., W.G. Degroot. 1976. Thermal uses and properties of carbohydrates and lignins, New York: Academic Press Shaw M.D., C. Karunakaran., L.G. Tabil. 2009. Physicochemical characteristics of densified untreated and steam exploded poplar wood and wheat straw grinds. Biosystems Engineering. 103:198-207. Sheng C., J.L.T. Azevedo. 2005. Estimating the higher heating value of biomass fuels from basic analysis data. Biomass and Bioenergy. 28:499-507. Sheng C., J. L. Azevedo. 2005. Estimating the higher heating value of biomass fuels from basic analysis data. Biomass and Bioenergy. 28(5):499-507. Shevchenco S. M., R. P. Beatson, J. N. Saddler. 1999. The nature of lignin from steam explosion/enzymatic hydrolysis of softwood. Applied Biochemistry and Biotechnology. 77-79:867-876. Shevchenko S. M., K. Chang, D. G. Dick, D. J. Gregg, J. N. Saddler. 2001. Structure and properties of lignin in softwoods after SO2 catalyzed steam explosion and enzymatic hydrolysis. Cellulose Chemistry and Technology. 35(5-6):487-502. Sjostrom, E. 1993. Wood Chemistry. Fundamentals and Applications. Second edition ed. San Diego: Academic press. 292.  160  Sokhansanj S., Turhollow A.F., 2004. Biomass densification – cubing operations and costs for corn stover. Applied Engineering in Agriculture. 20(4):495-499. Sokhansanj S., S. Mani, X. Bi, P. Zaini, L. G. Tabil. 2005. Binderless Pelletization of biomass. Paper No. 056061, Portland, Oregon, July 9 – July 12. St. Joseph, Mich.: ASABE. Sokhansanj S. 2006. Cost benefit of biomass supply and pre-processing, BIOCAP research integration program synthesis paper Spiess W. E. L., W. Wolf. 1987. Critial evaluation of methods to determine moisture sorption isotherms. In: Water activity: Theory and applications to food. Eds: Rockland, L. B. & Beuchat, L. R. Institute of food technologists. Marcel Dekker, Inc., New York, USA. Stahl M., K. Granstrom, J. Berghel, R. Renstrom. 2004. Industrial processes for biomass drying and their effects on the quality properties of wood pellets. Biomass and Bioenergy. 27:621 - 628. Startsev O. V., B. N. Salin. 2000. Polycondensation of the components of the lignincarbohydrate complex of steam exploded wood. Doklady Chemical Technology. 373375(1):30-33. Sundqvist B., T. Moren. 2002. The influence of wood polymers and extractives on wood colour induced by hydrothermal treatment. Holz Roh-Werkst. 60:375-376. Suzuki S., H. Shintani, S. Y. Park, K. Saito, N. Laemasak, M. Okuma, K. Iiyama. 1998. Preparation of binderless boards from steam exploded pulps of oil palm (Elaeis guneensis Jaxq.) Fronds and structural characteristics of lignin and wall polysaccharides in steam exploded pulps to be discussed for self-binding. Holzforschung. 52(4):417-426. Swaan, J. 2009. Short Communication, Canbio Update: Challenges for Canadian pellets, Canadian Biomass Magazine. HTTP available: http://www.canadianbiomassmagazine.ca/content/view/1172/63  161  Sweet M.S., J.E. Winandy. 1999. The influence of degree of polymerization (DP) of cellulose and hemicellulose on the strength loss of fire-retardant-treated wood. Holzforschung. 53(3):311–317. Tabil, L. G. Jr., S. Sokhansanj. 1996a. Process conditions affecting the physical quality of alfalfa pellets. Applied Engineering in Agriculture. 12(3):345-350. Tabil, L.G. Jr. 1996b. Binding and pelleting characteristics of alfalfa. Ph.D. dissertation. Saskatoon, Saskatchewan, CA: University of Saskatchewan, Department of Agricultural and Bioresource Engineering Tabil L.G., S. Sokhansanj, W. J. Crerar, R. T. Patil, M. H. Khoshtaghaza, A. Opoku. 2002. Physical characterization of alfalfa cubes: I. Hardness. Canadian Biosystems Engineering. 44(3):55-63. TAPPI. 1994. Acid insoluble lignin in wood and pulp. TAPPI Test Method T222 om-88. Technical Association of the Pulp and Paper Industry. TAPPI. 1999. Acetone Extractives of Wood and Pulp. TAPPI T280 pm-99. Technical Association of the Pulp and Paper Industry. Theoretical ethanol yield calculator, US DOE, HTTP Available: http://www1.eere.energy.gov/biomass/ethanol_yield_calculator.html Accessed: July 31, 2007 Tjeerdsma B., M. Boonstra, A. Pizzi, P. Tekely, H. Militz. 1998. Characterization of thermally modified wood: molecular reasons for wood performance improvement. Holz Roh-Werkst. 56:149-153. Turn S. Q., C. M. Kinoshita, W. E. Kaar, D. M. Ishiimura. 1998. Measurements of gas phase carbon in steam explosion of biomass. Bioresource Technology. 64(1):71-75. Ulbricht, R. J., J. Sharon. J. Thomas. 1984. A review of 5-hydroxymethylfurfura HMF in parental solutions. Fundamental and Applied Toxicology. 4:843-853. Van Dam, J. E.G., M. J. A. Oever, W. Teunissen, E. R. P. Keijsers, A. G. Peralta. 2004. Process for production of high density/high performance binderless boards from whole  162  coconut husk – Part 1: Lignin as intrinsic thermosetting binder resin. Industrial Crops and Products. 19(3):207–216. Vazquez G., G. Antorrenta, J. Gonzalez, S. Freire, S. Lopez. 1997. Acetosolv pulping of pine wood. Kinetics modeling of lignin solubilization and condensation. Bioresource Technology. 59:121-127. Wu H., G. Bryant, K. Benfell, T. Wall. 2000. An experimental study on the effect of system pressure on char structure of an Australian bituminous coal. Energy & Fuels. 14: 282 - 290. Zeitsch K. 2000. The Chemistry and Technology of Furfural and its Many Byproducts, Elsevier, Amsterdam. Zimbardi F., D. Viggiano, F. Nanna, M. Demichele, D. Cuna, G. Cardinale. 1999. Steam explosion of straw in batch and continuous systems. Applied Biochemistry and Biotechnology. 77-79:117 Zimbardi F., E. Ricci, G. Braccio. 2002. Technoeconomic Study on steam explosion application in biomass processing. Applied Biochemistry and Biotechnology. 98-100 (2):89-99. Zimbardi F., E. Viola, F. Nanna, E. Larocca, M. Cardinale, D. Barisano. 2007. Acid impregnation and steam explosion of corn stover in batch processes. Industrial Crops and Products. 26:195-206. Zuo Y., P. C. Maness, B. E. Logan. 2006. Electricity production from steam exploded corn stover biomass. Energy and Fuels. 20:1716-1721.  163  Appendix I Data Acquisition of Steam Explosion Unit  Figure AI.1 Data Acquisition Front Panel Labview Program  164  Figure AI.2 Block Diagram of Temperature and Pressure Transducer Data Acquisition Labview Program  Figure AI.3 Ball Valve (B-1) Opening (Left) and Closing (Right) Controlled by Labview Program  165  Figure AI.4 Block Diagram of Ball Valve (B-1) opening Labview Program  166  

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.24.1-0059133/manifest

Comment

Related Items