UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Biomass torrefaction in slot-rectangular spouted beds Wang, Ziliang 2017

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

Item Metadata

Download

Media
24-ubc_2018_february_wang_ziliang.pdf [ 4.86MB ]
Metadata
JSON: 24-1.0362042.json
JSON-LD: 24-1.0362042-ld.json
RDF/XML (Pretty): 24-1.0362042-rdf.xml
RDF/JSON: 24-1.0362042-rdf.json
Turtle: 24-1.0362042-turtle.txt
N-Triples: 24-1.0362042-rdf-ntriples.txt
Original Record: 24-1.0362042-source.json
Full Text
24-1.0362042-fulltext.txt
Citation
24-1.0362042.ris

Full Text

BIOMASS TORREFACTION IN SLOT-RECTANGULAR SPOUTED BEDS   by   Ziliang Wang   B.Sc., China University of Mining and Technology, 2009 M.Sc., South China University of Technology, 2012   A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF   DOCTOR OF PHILOSOPHY   in    THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES   (Chemical and Biological Engineering)    THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   December 2017 © Ziliang Wang, 2017 ii  Abstract Biomass, a nearly carbon-neutral energy resource, can reduce greenhouse gas emission and replace fossil fuels, but it is characterized by heterogeneity, high moisture content, low bulk density, low calorific value, pliability and hygroscopic nature, all of which challenge its utilization. Torrefaction, a thermal pretreatment method, is capable of modifying the physical-chemical properties of biomass and enhancing its calorific value. Slot-rectangular spouted beds (SRSBs) can effectively handle biomass particles and offer a promising way to overcome the scale-up challenge of conventional spouted beds. This study explored the potential application of SRSBs to biomass torrefaction. Solids mixing in a dual-compartment slot-rectangular spouted bed (DSRSB) was first studied to address the scale-up issue, while also providing fundamental information needed for the design and operation of a DSRSB reactor. SRSB and DSRSB reactors were developed for torrefying sawdust with the semi-batch operation. Hydrodynamics of SRSB and DSRSB, torrefaction performance, torrefied product properties and torrefied product pyrolysis were subsequently investigated.  Temperature, biomass feed rate, sawdust particle size and oxygen concentration influenced torrefaction performance and torrefied product properties. Temperature was found to be the most important factor. Biomass torrefaction performed better in the DSRSB than in the SRSB. Higher temperature, lower biomass feed rate, larger sawdust particle size and greater oxygen concentration all led to increased weight loss and decreased energy yield of sawdust, and produced torrefied sawdust with higher HHV, greater atomic carbon content, lower atomic hydrogen and oxygen contents, less volatile matter, greater fixed carbon and less iii  hemicellulose. Higher temperature and greater oxygen concentration were very helpful to produce more torrefied sawdust captured by a cyclone. Performance of oxidative torrefaction was similar to that of non-oxidative torrefaction. The effect of oxygen concentration was more significant at a higher temperature. Torrefied sawdust underwent size reduction during torrefaction, with smoother and cleaner surfaces compared to raw sawdust. The activation energy for non-oxidatively torrefied sawdust was higher than for oxidatively torrefied sawdust, which was in turn greater than that of raw sawdust. Sawdust particle size affected the pressure drop across the reactor.      iv  Lay Summary Biomass has potential to replace a significant amount of greenhouse-gas-emitting fossil fuels. Torrefaction of biomass, a thermal treatment process to increase its calorific value, was carried out in a jet-type reactor of unusual geometry leading to improved performance compared to conventional reactors. Properties of the torrefied biomass were close to those of lignite, a low-rank coal.   v  Preface This dissertation is an original, independent work of the author, Ziliang Wang, under the supervision of Drs. C. Jim Lim and John R. Grace. All of the work presented was conducted in the Fluidization Research Centre of the Department of Chemical and Biological Engineering at the University of British Columbia, Vancouver campus.  A paper based on Chapter 3 entitled “Solids mixing in a dual-column slot-rectangular spouted bed with a suspended partition” has been published [Wang, Z., Lim, C.J., Grace, J.R. 2016. Powder Technology, 301, 1264-1269]. The experiments were performed by myself with advice from Drs. Lim and Grace.  Work on the effects of temperature and biomass feed rate on biomass torrefaction in a slot-rectangular spouted bed facility described in section 4.1 of Chapter 4 has been submitted as a paper [Wang, Z., Lim, C.J., Grace, J.R. Biomass torrefaction in a slot-rectangular spouted bed reactor.].  Work related to section 4.2 of Chapter 4 has been published [Wang, Z., Lim, C.J., Grace, J.R., Li, H., Regina Parise, M. 2017. Effects of temperature and particle size on biomass torrefaction in a slot-rectangular spouted bed reactor. Bioresource Technology, 244, 281-288]. I was the lead investigator, carried out all the experimental design, data collection and analysis, and paper preparation. Dr. Maria Regina Parise and Dr. Hui Li participated in the experiments. Dr. Hui Li was also responsible for fiber analysis of the raw and torrefied product. Drs. Grace and Lim assisted with the finalization of the paper. vi  The dual-compartment slot-rectangular spouted bed reactor used for the work presented in Chapter 5 is my own design and implementation. Chapter 6 is based on experimental work conducted in a TGA. Dr. Hui Li carried out most of the thermogravimetric tests of raw and torrefied product. I was responsible for the experimental design, data collection and analysis.  All elemental analyses of raw and torrefied sawdust covered in this thesis were carried out in the Department of Chemistry at the University of British Columbia.    vii  Table of Contents  Abstract ..................................................................................................................................... ii Lay Summary ........................................................................................................................... iv Preface....................................................................................................................................... v Table of Contents .................................................................................................................... vii List of Tables ......................................................................................................................... xiii List of Figures ........................................................................................................................ xvi List of Symbols and Abbreviations....................................................................................... xxv Acknowledgements .............................................................................................................. xxxi Chapter 1 Introduction .............................................................................................................. 1 1.1 Slot-rectangular spouted bed........................................................................................... 1 1.2 Biomass torrefaction ....................................................................................................... 5 1.2.1 Non-oxidative torrefaction ....................................................................................... 9 1.2.2 Oxidative torrefaction ............................................................................................ 12 1.3 Torrefaction reactor ...................................................................................................... 14 1.4 Research objectives and principal tasks in this thesis project ....................................... 15 1.5 Thesis outline ................................................................................................................ 17 viii  Chapter 2 Sample preparation, experimental set-up and methodology .................................. 19 2.1 Cold model dual-compartment slot-rectangular spouted bed ....................................... 19 2.1.1 Equipment design................................................................................................... 19 2.1.2 Material and methodology ..................................................................................... 21 2.2 Sample preparation for torrefaction experiments ......................................................... 25 2.3 Single-compartment slot-rectangular spouted bed torrefaction reactor ........................ 27 2.3.1 Equipment design................................................................................................... 27 2.3.2 Methodology .......................................................................................................... 28 2.4 Dual-compartment slot-rectangular spouted bed torrefaction reactor .......................... 32 2.4.1 Equipment design................................................................................................... 32 2.4.2 Methodology .......................................................................................................... 34 2.5 Thermogravimetric analysis for kinetic study and experimental design ...................... 37 2.6 Characterization of torrefied product ............................................................................ 39 2.6.1 Solid product .......................................................................................................... 39 2.6.2 Condensable liquid product ................................................................................... 41 2.6.3 Gas product ............................................................................................................ 41 Chapter 3 Solids mixing in a dual-compartment slot-rectangular spouted bed with a suspended partition ................................................................................................................. 42 ix  3.1 Quantitative study of solids mixing .............................................................................. 42 3.1.1 Solids exchange between compartments ............................................................... 43 3.1.2 Lacey mixing index................................................................................................ 45 3.2 Minimum spouting velocity and maximum pressure drop ........................................... 46 3.3 Image analysis method .................................................................................................. 47 3.4 Evolution of solids exchange with time ........................................................................ 48 3.5 Effect of suspended partition position .......................................................................... 50 3.6 Effect of superficial gas velocity .................................................................................. 51 3.7 Effect of static bed height ............................................................................................. 52 3.8 Effect of partition height ............................................................................................... 53 3.9 Conclusions ................................................................................................................... 55 Chapter 4 Biomass torrefaction in a single-compartment slot-rectangular spouted bed reactor................................................................................................................................................. 56 4.1 Effects of temperature and biomass feed rate ............................................................... 56 4.1.1 Pretests in TGA ...................................................................................................... 56 4.1.2 Operating conditions .............................................................................................. 58 4.1.3 Typical case ........................................................................................................... 60 4.1.4 Solid product yield ................................................................................................. 62 x  4.1.5 Properties of torrefied product ............................................................................... 66 4.1.6 Particle size distribution of torrefied product ........................................................ 73 4.2 Effects of temperature and particle size ........................................................................ 74 4.2.1 Pretest in TGA ....................................................................................................... 75 4.2.2 Operating conditions .............................................................................................. 76 4.2.3 Effect of particle size on hydrodynamics of SRSB torrefaction reactor ................ 77 4.2.4 Solid product yield ................................................................................................. 80 4.2.5 Evolution of ash content in torrefied product ........................................................ 82 4.2.6 Properties of torrefied product ............................................................................... 84 4.2.7 Particle size reduction ............................................................................................ 88 4.3 Oxidative torrefaction ................................................................................................... 90 4.3.1 Operating conditions for oxidative torrefaction ..................................................... 90 4.3.2 Typical case ........................................................................................................... 92 4.3.3 Solid product yield ................................................................................................. 95 4.3.4 Evolution of ash content in torrefied product ........................................................ 98 4.3.5 Final bed depth within SRSB reactor................................................................... 100 4.3.6 Properties of torrefied product ............................................................................. 101 4.4 Conclusions ................................................................................................................. 108 xi  Chapter 5 Biomass torrefaction in a dual-compartment slot-rectangular spouted bed reactor............................................................................................................................................... 110 5.1 Biomass torrefaction in nitrogen atmosphere ............................................................. 110 5.1.1 Operating conditions ............................................................................................ 110 5.1.2 Case study ............................................................................................................ 111 5.1.3 Evolution of ash content of torrefied product ...................................................... 114 5.1.4 Final bed height after torrefaction ........................................................................ 116 5.1.5 Solid product yield ............................................................................................... 117 5.1.6 Properties of torrefied product ............................................................................. 119 5.1.7 Liquid product ...................................................................................................... 123 5.1.8 Product gas composition ...................................................................................... 126 5.2 Biomass torrefaction in oxygen-containing atmosphere............................................. 127 5.2.1 Operating conditions for oxidative torrefaction ................................................... 128 5.2.2 Evolution of ash content of torrefied product ...................................................... 130 5.2.3 Final bed depth after torrefaction ......................................................................... 132 5.2.4 Solid product yield ............................................................................................... 133 5.2.5 Properties of torrefied product ............................................................................. 134 5.3 Conclusions ................................................................................................................. 138 xii  Chapter 6 Thermogravimetric characteristics and kinetic analysis of torrefied biomass pyrolysis ................................................................................................................................ 140 6.1 Modelling .................................................................................................................... 140 6.1.1 Kissenger-Akahira-Sunose (KAS) model ............................................................ 141 6.1.2 Coats-Redfern Method ......................................................................................... 142 6.2 Thermogravimetric analysis........................................................................................ 144 6.3 Kinetic analysis ........................................................................................................... 157 6.3.1 Kinetic parameters determined by Kissenger-Akahira-Sunose (KAS) model .... 157 6.3.2 Kinetic parameters determined by Coats-Redfern model .................................... 162 6.4 Conclusions ................................................................................................................. 170 Chapter 7 Conclusions and Recommendations ..................................................................... 172 7.1 Comparison of torrefaction reactors ........................................................................... 172 7.2 Conclusions from this thesis ....................................................................................... 178 7.3 Recommendations for future work ............................................................................. 182 References ............................................................................................................................. 184 Appendix A: Summary of previous work on slot-rectangular spouted beds ........................ 194 Appendix B: Temperature profiles ....................................................................................... 198 Appendix C: Fourier Transform Infrared Spectroscopy ....................................................... 199 xiii  List of Tables Table 1-1 Key properties of raw biomass, torrefied biomass and lignite, and energy yield and solid yield of torrefaction. ......................................................................................................... 6 Table 2-1 DSRSB column dimensions. See Figure 2-1 for definition of dimensions. ........... 20 Table 2-2 Key properties of experimental particles and gas. .................................................. 21 Table 2-3 Properties of raw SPF sawdust and glass beads. .................................................... 26 Table 3-1 Experimental minimum spouting velocities and maximum pressure drops. .......... 46 Table 4-1 Operating conditions for torrefaction experiments of 0.5-1.0 mm SPF sawdust. For each case, U = 1.2Ums. ............................................................................................................ 59 Table 4-2 Experimental solid torrefied product yield and weight loss in experiments. ......... 64 Table 4-3 Particle density, HHV and energy yield of torrefied SPF sawdust on a dry basis. 66 Table 4-4 Elemental and fiber analyses of raw SPF sawdust and torrefied SPF sawdust collected by cyclone on a dry basis. ........................................................................................ 69 Table 4-5 Proximate analysis of torrefied SPF sawdust on a dry basis. ................................. 71 Table 4-6 Operating conditions for torrefaction experiments with smallest, intermediate and coarsest sawdust. For each case, U = 1.2Ums and operating time = 50 min. .......................... 77 Table 4-7 Experimental solid torrefied product yield and weight loss. See Table 4-6 for operating conditions. ............................................................................................................... 80 Table 4-8 Proximate, ultimate and fiber analyses of torrefied sawdust on a dry basis. .......... 86 xiv  Table 4-9 HHVs, particle densities and energy yield of torrefied sawdust on a dry basis. .... 88 Table 4-10 Operating conditions for oxidative torrefaction experiments. .............................. 91 Table 4-11 Torrefied product yield and weight loss in oxidative torrefaction experiments. .. 96 Table 4-12 Properties of torrefied sawdust and energy yields on a dry basis. ...................... 105 Table 5-1 Operating conditions for biomass torrefaction experiments with nitrogen atmosphere in DSRSB facility. ............................................................................................. 111 Table 5-2 Torrefied product yield and weight loss in torrefaction experiments with nitrogen atmosphere in DSRSB facility. ............................................................................................. 117 Table 5-3 Properties of torrefied solid sawdust on a dry basis. ............................................ 121 Table 5-4 Main compounds identified in condensable liquid via GC-MS. .......................... 124 Table 5-5 Operating conditions for oxidative torrefaction experiments in DSRSB reactor. U = 1.2Ums, operating time=50 min. ............................................................................................ 128 Table 5-6 Torrefied product yield and weight loss in oxidative torrefaction experiments in DSRSB facility...................................................................................................................... 133 Table 5-7 Properties of torrefied sawdust on a dry basis. ..................................................... 137 Table 6-1 Mechanism functions for modelling pyrolysis of raw/torrefied biomass. ............ 143 Table 6-2 Kinetic parameters for raw and torrefied SPF sawdust pyrolysis obtained by Coats-Redfern model. For operating conditions, see Table 4-1, Table 4-6, and Table 4-10. ......... 163 xv  Table 7-1 Comparison of torrefaction using different reactors and operating conditions, and key properties of torrefied product (on a dry basis). ............................................................. 173 Table A-1 Previous literature on slot-rectangular spouted beds. .......................................... 194 Table C-1 Ratios of intensities of lignin-associated band with carbohydrate bands for original and cyclone-caught biomass torrefied at different temperatures and biomass feed rates in SRSB reactor. See Table 4-1 for operating conditions. ........................................................ 203 Table C-2 Ratios of relative intensity of lignin-associated band with carbohydrate bands for raw and cyclone-caught torrefied sawdust of different particle sizes. See Table 4-6 for operating conditions. ............................................................................................................. 204 Table C-3 Ratios of relative intensity of lignin-associated band with carbohydrate bands for raw and cyclone-caught torrefied biomass of oxidative torrefaction in SRSB reactor. See Table 4-10 for operating conditions. ..................................................................................... 205   xvi  List of Figures Figure 1-1 Schematic drawing of slot-rectangular spouted bed column: (a) Front view, (b) Top view. .................................................................................................................................. 2 Figure 1-2 Schematic diagram of non-oxidative biomass torrefaction. .................................... 9 Figure 1-3 Schematic diagram of oxidative biomass torrefaction. ......................................... 12 Figure 2-1 (a) Schematic and definition of symbols for dual-compartment slot-rectangular spouted bed column; (b) top view. .......................................................................................... 19 Figure 2-2 (a) Photo of cold model dual-compartment slot-rectangular spouted bed facility, (b) schematic diagram of dual-compartment slot-rectangular spouted bed with a suspended partition. .................................................................................................................................. 23 Figure 2-3 Photos of red+black particles, black particles, and red particles. ......................... 24 Figure 2-4 Operating variables and their ranges in solids mixing experiments. .................... 24 Figure 2-5 (a) Slot-rectangular spouted bed reactor; (b) plan view of base. (T1-T4: K type thermocouples, P: Pressure gauge with range 0-103.4 kPa, ΔP: differential pressure transducer with range 0-6.9 kPa.) ........................................................................................... 27 Figure 2-6 Schematic diagram of slot-rectangular spouted bed torrefaction facility. (F1, F2: Rotameter with range 0-51 m3/h, F3: Orifice flowmeter with range 0-51 m3/h, P: Pressure gauge with range 0-103.4 kPa, T1-T4, Tc and The: K-type thermocouples, V1, V2: Needle valve, V3: Ball valve, V4: Globe valve.) ................................................................................ 29 xvii  Figure 2-7 Experimental variables studied and their nominal ranges in the slot-rectangular spouted bed torrefaction facility. ............................................................................................ 30 Figure 2-8 (a) Schematic of dual-compartment slot-rectangular spouted bed reactor, (b) plan view of base, (c) photo of DSRSB column, (d) photo of W type base and windbox. (𝑇1′-𝑇7′: K type thermocouples, P: Pressure gauge with range 0-103.4 kPa, ΔP: differential pressure transducer with range 0-6.9 kPa.) ........................................................................................... 33 Figure 2-9 (a) Photo of dual-compartment slot-rectangular spouted bed torrefaction facility, (b) schematic diagram of dual-compartment slot-rectangular spouted bed torrefaction facility.................................................................................................................................................. 36 Figure 2-10 Experimental variables studied and their nominal ranges in the dual-compartment slot-rectangular spouted bed torrefaction facility. ............................................ 37 Figure 2-11 Torrefied product map and corresponding measurements. ................................. 39 Figure 3-1 Schematic configuration of solids mixing between compartments in DSRSB. .... 43 Figure 3-2 Comparison of measured and true mass fraction of tracer particles in mixtures. . 47 Figure 3-3 Tracer concentration in compartment 2 for HB,0 = 150 mm, U = 1.2Ums, ∆h = 20 mm, Hsp = 300 mm. ................................................................................................................. 48 Figure 3-4 Lacey mixing index for glass beads of three sizes with HB,0 = 150 mm, U = 1.2Ums, Hsp = 300 mm and ∆h = 20 mm; t' is the time required to reach equilibrium. ........................ 49 Figure 3-5 Effect of partition position on solids exchange coefficient and tracer concentration for 1.16 and 1.61 mm glass beads with HB,0 = 150 mm, U = 1.2Ums, t = 30s, Hsp = 300 mm. 50 xviii  Figure 3-6 Effect of superficial gas velocity on solids exchange coefficient and tracer concentration for 1.61 and 1.16 mm glass beads with HB,0 = 150 mm, ∆h = 20 mm, t = 30 s, Hsp = 300 mm. ......................................................................................................................... 51 Figure 3-7 Effect of static bed height on solids exchange coefficient and tracer concentration for 1.61 and 1.16 mm glass beads with U = 1.2Ums, ∆h = 20 mm, t = 30 s, Hsp = 300 mm. .. 52 Figure 3-8 Evolution of Lacey mixing index in the DSRSB with 100 mm and 300 mm partition heights for 1.61 mm glass beads, HB,0 = 150 mm, ∆h = 20 mm, U = 1.2Ums. .......... 53 Figure 3-9 Solids exchange coefficient for 100 mm and 300 mm partition heights for 1.61 mm glass beads, HB,0 = 150 mm, ∆h = 20 mm, U = 1.2Ums. .................................................. 54 Figure 4-1 Dynamic weight loss and devolatilization curves for SPF sawdust particles of 0.5-1.0 mm for a TGA heating rate of 1°C/min with nitrogen of 99.999% purity. ...................... 57 Figure 4-2 0.5-1.0 mm SPF sawdust torrefaction in TGA at different temperatures with nitrogen of 99.999% purity at 50°C/min heating rate. ............................................................ 58 Figure 4-3 Time variation of temperatures for SRSB reactor, case C-6 of sawdust with 0.5-1.0 mm diameter, T = 295°C, and F = 467 g/h. ...................................................................... 61 Figure 4-4 Time variation of reactor pressure drop and superficial gas velocity for case C-6 with 0.5-1.0 mm diameter sawdust particles, T = 295°C and F = 467 g/h. ............................ 62 Figure 4-5 Mass distribution map of raw and torrefied sawdust torrefied at different biomass feed rates and temperatures in SRSB facility. ........................................................................ 65 Figure 4-6 SEM images of untreated and cyclone-caught torrefied SPF sawdust. See Table 4-1 for detailed operating conditions. ..................................................................................... 72 xix  Figure 4-7 Particle size distribution of cyclone-caught sawdust torrefied at T = 240-330°C with F = 440 g/h for 50 min. See Table 4-1 for operating conditions. ................................... 74 Figure 4-8 Weight loss curves of SPF sawdust of 0.25-0.5, 0.5-1.0 and 1.0-2.0 mm diameter at different temperatures with a heating rate of 50°C/min. ..................................................... 75 Figure 4-9 Time variation of pressure drop across SRSB reactor. See Table 4-6 for operating conditions. ............................................................................................................................... 78 Figure 4-10 Final bed depth after 50 min torrefaction. See Table 4-6 for operating conditions.................................................................................................................................................. 79 Figure 4-11 Effect of particle size on mass distribution of raw and torrefied sawdust in SRSB facility. .................................................................................................................................... 81 Figure 4-12 Evolution of ash content of product captured by cyclone for (a) 0.25-0.5 mm particles; (b) 0.5-1.0 mm particles. ......................................................................................... 84 Figure 4-13 Percentage reduction of Sauter mean particle size of SF and torrefied sawdust of 240, 270 and 300°C. ............................................................................................................... 89 Figure 4-14 Temperature profiles for case OT-5, T = 267°C, with XO2 = 6 vol.% oxygen in the feed-gas and F = 407 g/h sawdust feed rate. ..................................................................... 92 Figure 4-15 Profiles of reactor pressure drop and superficial gas velocity for case OT-5, T = 267°C, with XO2 = 6 vol.% oxygen in the feed-gas and F = 407 g/h sawdust feed rate. ........ 93 Figure 4-16 Oxygen concentration time-variation for case OT-5, T = 267°C, with XO2 = 6 vol.% oxygen in the feed-gas and F = 407 g/h sawdust feed rate. .................................................... 94 xx  Figure 4-17 Mass distribution of raw and torrefied sawdust for oxidative torrefaction in SRSB facility. ......................................................................................................................... 97 Figure 4-18 Time variation of ash content of torrefied sawdust captured by cyclone at (a) T = 240°C, XO2 = 0-9 vol.%; (b) T = 270°C, XO2 = 0-9 vol.%; (c) T = 300°C, XO2 = 0-9 vol.%. .. 98 Figure 4-19 Final bed depth after 50 min of oxidative torrefaction of 0.5-1.0 mm sawdust.100 Figure 4-20 SEM images of cyclone-caught sawdust oxidatively torrefied at T = 270°C and F = 440 g/h biomass feed rate with different oxygen concentrations. ..................................... 107 Figure 5-1 Temperature time-variation for case D-C-6. See Table 5-1 for operating conditions. ............................................................................................................................. 112 Figure 5-2 Pressure drop across DSRSB reactor for case D-C-6. See Table 5-1 for operating conditions. See Figure 2-8 for definitions of ΔP1 and ΔP2. .................................................. 113 Figure 5-3  Evolution of ash content of cyclone-caught torrefied sawdust for (a) D-C-1 to D-C-3; (b) D-C-4 to D-C-6; (c) D-C-7 to D-C-9; (d) D-C-10 to D-C-12. For detailed operating conditions see Table 5-1. ...................................................................................................... 115 Figure 5-4 Final average bed depth after 50 min of torrefaction. ......................................... 116 Figure 5-5 Mass distribution of raw and torrefied sawdust for biomass torrefaction in DSRSB facility. .................................................................................................................................. 118 Figure 5-6 GC-MS spectra of condensable liquid obtained from (A) D-C-10; (B) D-C-11; (C) D-C-12. For operating conditions see Table 5-1. ................................................................. 123 xxi  Figure 5-7 Water content of condensable liquid collected for D-C-10: T = 240°C, D-C-12: T = 270°C, D-C-10: T = 300°C. For operating conditions see Table 5-1. ............................... 125 Figure 5-8 CO and CO2 concentrations in the off-gas during torrefaction at end of run D-C-10 T = 240°C, run D-C-12: T = 270°C, and run D-C-10: T = 300°C. For operating conditions see Table 5-1. ........................................................................................................................ 126 Figure 5-9 Temperature time-variation for (a) case D-OT-7; (b) case D-OT-10. See Table 5-5 for operating conditions. ....................................................................................................... 130 Figure 5-10 Evolution of ash content of torrefied product captured by the cyclone for (a) D-OT-1 to D-OT-4; (b) D-OT-5 to D-OT-7; (c) D-OT-9 and D-OT-10. For detailed operating conditions see Table 5-5. ...................................................................................................... 131 Figure 5-11 Final average bed depth after oxidative torrefaction at nominal conditions of F = 900 g/h, T = 240, 270 and 300°C with various oxygen concentrations in the feed-gas. ...... 132 Figure 5-12 Mass distribution of raw and torrefied sawdust for oxidative torrefaction in DSRSB facility...................................................................................................................... 134 Figure 6-1 Thermogravimetric (TG) and derivative thermogravimetric (DTG) curves for raw SPF sawdust at heating rates of 10, 20, 30 and 40°C/min: (a) 0.25-0.5 mm; (b) 0.5-1.0 mm; (c) 1.0-2.0 mm particles. ....................................................................................................... 146 Figure 6-2 TG and DTG curves for torrefied SPF sawdust of 0.25-0.5 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case PS-1: T = 240°C and F = 440 g/h; (b) case PS-2: T = 270°C and F = 440 g/h; (c) case PS-3: T = 300°C and F = 440 g/h. See Table 4-6 for detailed torrefaction operating conditions. ............................. 147 xxii  Figure 6-3 TG and DTG curves for torrefied SPF sawdust of 0.5-1.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case PS-4: T = 240°C and F = 440 g/h; (b) case PS-5: T = 270°C and F = 440 g/h; (c) case PS-6: T = 300°C and F = 440 g/h. See Table 4-6 for detailed torrefaction operating conditions. ............................. 148 Figure 6-4 TG and DTG curves for torrefied SPF sawdust of 1.0-2.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case PS-7: T = 240°C and F = 440 g/h; (b) case PS-8:  T = 270°C and F = 440 g/h; (c) case PS-9: T = 300°C and F = 440 g/h. See Table 4-6 for detailed torrefaction operating conditions. ............................. 149 Figure 6-5 TG and DTG curves for torrefied SPF sawdust of 0.5-1.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case C-1: T = 240°C and F = 220 g/h; (b) case C-2: T = 270°C and F = 220 g/h; (c) case C-3: T = 300°C and F = 220 g/h. See Table 4-1 for detailed torrefaction operating conditions. ................................ 151 Figure 6-6 TG and DTG curves for torrefied SPF sawdust of 0.5-1.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case C-8: T = 240°C and F = 710 g/h; (b) case C-9: T = 270°C and F = 710 g/h; (c) case C-10: T = 300°C and F = 710 g/h. See Table 4-1 for detailed torrefaction operating conditions. ................................ 152 Figure 6-7 TG and DTG curves for torrefied SPF sawdust of 0.5-1.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case OT-2: T = 240°C, F = 440 g/h and O2 conc.: 3 vol.%; (b) case OT-3: T = 240°C, F = 440 g/h and O2 conc.: 6 vol.%; (c) case OT-4: T = 240°C, F = 440 g/h and O2 conc.: 9 vol.%. See Table 4-10 for detailed torrefaction operating conditions. ........................................................................... 154 xxiii  Figure 6-8 TG and DTG curves for torrefied SPF sawdust of 0.5-1.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case OT-6: T = 270°C, F = 440 g/h and O2 conc.: 3 vol.%; (b) case OT-7: T = 270°C, F = 440 g/h and O2 conc.: 6 vol.%; (c) case OT-8: T = 270°C, F = 440 g/h and O2 conc.: 9 vol.%. See Table 4-10 for detailed torrefaction operating conditions. ........................................................................... 155 Figure 6-9 TG and DTG curves for torrefied SPF sawdust of 0.5-1.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case OT-10: T = 300°C, F = 440 g/h and O2 conc.: 3 vol.%; (b) case OT-11: T = 300°C, F = 440 g/h and O2 conc.: 6 vol.%; (c) case OT-12: T = 300°C, F = 440 g/h and O2 conc.: 9 vol.%. See Table 4-10 for detailed torrefaction operating conditions. ........................................................................... 156 Figure 6-10 (a) Activation energy from KAS method at different conversions; (b) Activation energies for 0.5-1.0 mm raw SPF sawdust at different conversions. .................................... 158 Figure 6-11 Activation energies from KAS method for torrefied SPF sawdust as a function of conversions, temperature, biomass feed rate and oxygen concentration. ............................. 161 Figure 6-12 Coats-Redfern plots for raw 0.5-1.0 mm SPF particles at different heating rates................................................................................................................................................ 162 Figure B-1 Temperature profiles in SRSB reactor for case OT-6, T = 271°C, with XO2 = 9 vol.% oxygen in the feed-gas and F = 495 g/h sawdust feed rate. ....................................... 198 Figure B-2 Temperature profiles in SRSB reactor for case OT-9, T = 318°C, with XO2 = 9 vol.% oxygen in the feed-gas and F = 469 g/h sawdust feed rate. ....................................... 198 xxiv  Figure C-1 FTIR spectra of raw and cyclone-caught torrefied sawdust: (1) Sawdust was torrefied at different temperatures and biomass feed rates in SRSB reactor. See Table 4-1 for operating conditions. (2) Different size sawdust was torrefied in SRSB reactor. See Table 4-6 for operating conditions. (3) Sawdust was torrefied at different temperature and oxygen concentration in SRSB reactor. See Table 4-10 for operating conditions. ........................... 201 xxv  List of Symbols and Abbreviations 2DSB Two-dimensional spouted bed A Pre-exponential factor, s-1 A0 Cross-section area, m2 AOR Angle of repose, degree ATR Attenuated total reflectance C Atomic carbon 𝐶̅               Average concentration of tracer particles, wt.% Ci,t Concentration of tracer particles in sample i at time t, wt.% C1 Weight fractions of tracer particles in compartment 1, wt.%   C2 Weight fractions of tracer particles in compartment 2, wt.%  CFD Computational fluid dynamics  CH4 Methane CO Carbon monoxide CO2 Carbon dioxide CSB Conventional spouted bed DSRSB Dual-compartment slot-rectangular spouted bed dp Particle diameter, mm dsv Sauter mean particle diameter, mm dsv,0 Sauter mean diameter of raw sawdust, mm dsv,i Sauter mean diameter of torrefied sawdust, mm  DTG Derivative thermogravimetric  xxvi  E Activation energy (J/mol) EFB Empty fruit bunches,  F Biomass feed rate, g/h f(α) Function of reaction mechanism, - FC Fixed carbon, wt.% FTIR Fourier transform infrared spectroscopy Gair Mass flow of air, g/h GB Glass beads GC-MS Gas Chromatograph-Mass Spectroscope H Atomic hydrogen H/C Molar ratio of hydrogen to carbon, - H2 Hydrogen  HB Final bed height after torrefaction, mm HB,0 Initial static bed height, mm HCl Hydrogen chloride Hd Divergent base height, mm HHV Higher heating value, wt.% Hsp Suspended partition height, mm Ht Overall column height, mm IL Lacey mixing index I7/I5 Relative intensity of aromatic skeletal vibration against C=O vibration I7/I10 Relative intensity of aromatic skeletal vibration against C-H deformation in cellulose and hemicellulose xxvii  I7/I12 Relative intensity of aromatic skeletal vibration against C-O-C vibration in cellulose and hemicellulose I7/I14 Relative intensity of aromatic skeletal vibration against C-H deformation in cellulose k(T) Rate constant, s-1 KAS Kissenger-Akahira-Sunose Kse Solids exchange coefficient, -  L Column thickness, mm Lslot Gas entry slot length, mm Lsp Suspended partition width, mm 𝑚0 Initial mass of the solid sample, g 𝑚𝑓 Final mass of the solid sample, g 𝑚𝑖 Actual mass of the solid sample, g ?̇?             Mass flow of between compartments, g/h  M0 Initial mass of particles in compartments 1 and 2, g Mc Mass of torrefied fine sawdust captured by cyclone, g Mf Mass of torrefied fine sawdust captured by the filter, g  Mr Mass of torrefied sawdust remaining in reactor, g Mt Total mass of sawdust fed, g n Number of particles in each sample N Atomic nitrogen N2 Nitrogen O Atomic oxygen xxviii  OT Oxidative torrefaction O/C Molar ratio of oxygen to carbon O2 Oxygen p Proportions of tracer particles in one sample P Absolute pressure, Pa PS Particle size R Universal gas constant, 8.314 J/(mol·K) SEM Scanning electron microscopy SF Screw feeder SPF Spruce, pine and fir mixture SRF Solid recovered fuel  SRSB Slot-rectangular spouted bed SV Reduction of Sauter mean diameter, %  t Time, min T Average temperature, °C T0 Gas temperature upstream of orifice flow meter, °C T1 Windbox temperature of SRSB, °C 𝑇1′ Downstream Windbox temperature of DSRSB, °C T2 Temperature at 64 mm above the base in SRSB, °C 𝑇2′ Upstream Windbox temperature of DSRSB, °C T3 Temperature 165 mm above base in SRSB, °C 𝑇3′ Temperature 64 mm above base of downstream compartment in DSRSB, °C T4 Temperature 267 mm above base in SRSB, °C xxix  𝑇4′ Temperature 64 mm above base of upstream compartment in DSRSB, °C 𝑇5′ Temperature 165 mm above base of  downstream compartment in DSRSB, °C 𝑇6′ Temperature 165 mm above base of upstream compartment in DSRSB, °C 𝑇7′ Temperature 267 mm above base of upstream compartment in DSRSB, °C Tc Surface temperature of cyclone, °C The Gas temperature at heater exchanger outlet, °C TGA Thermogravimetric analyzer U Superficial gas velocity, m/s Ums minimum spouting velocity, m/s VM Volatile matter, wt.% W Column width, mm Wb Base width, mm Wslot Gas entry slot width, mm Wsp Suspended partition thickness, mm X Weight loss of biomass, wt.% XO2 Oxygen concentration, vol.% Y Ratio of mass of torrefied sawdust captured by cyclone to total mass of torrefied sawdust, -  Z Height above entry, mm Greek symbols α Conversion of biomass pyrolysis, wt.% β Heating rate, C/min  xxx  ∆h Gap between partition and bed surface, mm ΔP Pressure drop, Pa ∆Pmax Maximum pressure drop, Pa ΔP1 Pressure drop across the compartment far from the feeding port, Pa ΔP2 Pressure drop across the compartment close to the feeding port, Pa θ Included base angle, ° ρb Bulk density of particles, kg/m3 ρg Gas density, kg/m3 ρp Particle density, kg/m3 𝜎02               Variance of completely segregated mixture, -   𝜎𝑅2               Variance of completely random mixture, - Φ Sphericity, - xxxi  Acknowledgements I would like to express my sincere gratitude to Dr. C. Jim Lim and Dr. John R. Grace for their inspiration, experienced and distinguished supervision, patience and financial assistance, which allowed me to complete this work. Financial support from the Natural Sciences and Engineering Research Council of Canada (NSERC) and the China Scholarship Council (CSC) is also acknowledged with gratitude.  I am also thankful to the rest of my thesis committee, Dr. Norman Epstein and Dr. Clive Brereton, for providing valuable suggestions and encouragement. Moreover, I would like to thank Dr. Zhiwei Chen and Dr. Yonghua Li for their important suggestions, Dr. Hui Li and Dr. Maria Regina Parise for assistance with experiments and colleagues of the Fluidization Research Centre for their suggestions and encouragement. Thanks are also due to the technical and administrative staff at the Department of Chemical and Biological Engineering. Thanks in particular to Doug Yuen, Gordon Cheng, Serge Milaire, Charles Cheung and Graham Liebelt for their invaluable assistance.  Finally, immense gratitude goes to my wife Guangxianyue, my parents and sister for their encouragement, support and understanding.     1  Chapter 1 Introduction 1.1 Slot-rectangular spouted bed Since being originated by Mathur and Gishler (1955), conical, cylindrical and conical-cylindrical spouted beds have shown versatility with respect to contacting of gases and liquids with coarse particulate solids. The solids are recirculating with a certain pattern in the spouted beds, making the spouted beds potentially perform certain useful operations more effectively than the fluidized beds with their more random solids motion (Epstein and Grace, 2011). Moreover, spouted beds have some advantages compared to fluidized beds and to other approaches of gas-solid contacting for large coarse particles (>1 mm). Conical, cylindrical and conical-cylindrical spouted beds are often referred to as conventional spouted beds (CSB). They have been used in a number of processes. Nevertheless, these studies have only been carried out in lab-scale or small units. CSB reactors are subject to well-known limitations regarding their volumetric processing capacity (Grace and Lim, 2011). Researchers have investigated several approaches in efforts to overcome these limitations by means of installing draft tube or plates (Fernandez-Akarregi et al., 2013; Luo et al., 2004; Makibar et al., 2012; Mostoufi et al., 2015; Shirvanian and Calo, 2004; Zhao et al., 2008), modifying the geometry of the fountain region (Lopez et al., 2017), and designing novel spouted beds, such as slot-rectangular spouted beds (Chen et al., 2013b; Dogan et al., 2000; Mujumdar, 1984), annular spouted beds (Hu et al., 2008), multiple-spouted beds (Ren et al., 2010), dual conical spouted beds (Fernandez-Akarregi et al., 2014), and pulsed slot-rectangular spouted beds (Parise et al., 2017; Saidi et al., 2015a).   2           Figure 1-1 Schematic drawing of slot-rectangular spouted bed column: (a) Front view, (b) Top view. In order to solve the scale-up challenge, Mujumdar (1984) promoted a modification of conventional axisymmetric spouted beds, called a “two-dimensional spouted bed” (2DSB), in which planar symmetry replaces axial symmetry. The advantages of this configuration are that the volumetric bed capacity can be easily increased by extending the bed thickness and the possibility of achieving different flow regimes by changing only the inlet nozzle dimensions (Chen, 2008; Passos et al., 1993). The configuration was renamed “slot-rectangular spouted bed” (SRSB) by Dogan et al. (2000) and Freitas et al. (2000), due to significant three-dimensional effects as the 2DSB thickness was increased. Appendix A lists previous works on slot-rectangular spouted beds. A schematic drawing of a SRSB is shown in Figure 1-1. The SRSB features a rectangular geometry with a rectangular slot opening fully or partially spanning its distributor width. In order to avoid a dead zone, the SRSB θFluid InletOutlet(a)WindboxDiverging baseRectangular columnRectangular slotRectangular slot(b)3  usually has a diverging base with an inclination angle (θ) greater than the angle of repose of the particles. Hydrodynamic studies of SRSB have been carried out in a number of previous studies, e.g. Chen et al. (2008); Chen et al. (2011b); Saidi et al. (2016). Chen (2008) comprehensively studied the hydrodynamics of SRSB and identified eight different flow patterns. Unstable spouting is a practical issue when a slot-rectangular spouted bed is scaled-up to a multiple-compartment slot-rectangular spouted bed. A multiple-compartment spouted bed with the compartments in series also provides a more favourable residence time distribution (closer to plug flow) for the solid particles. Chen et al. (2013b) investigated spouting in slot‐rectangular spouted beds with multiple compartments and slots. It was observed that the multiple-spouting flow was significantly influenced by the interaction of gas and particles in the fountain region between adjacent compartments. They suggested using a vertical suspended partition to stabilize the multiple spouting flows. They also proposed a dual-compartment slot-rectangular spouted bed (DSRSB) with a suspended partition, which was scaled-up from a single-spout SRSB configuration. Computational fluid dynamics (CFD) has also been applied to study SRSB (e.g. Golshan et al. (2017); Tabatabaei (2013); Tabatabaei et al. (2013); Zhang and Li (2017)) and DSRSB (e.g. Yang et al. (2014); Yang et al. (2015a); Yang et al. (2015b)). Wang et al. (2016c) studied experimentally and numerically three-dimensional gas-solid flow in a DSRSB with a suspended partition. They predicted that the position of the suspended partition had little effect on particle velocity and voidage under stable spouting conditions. However, raising the suspended partition benefitted solids exchange between the chambers. 4  In order to predict the performance of continuous spouted beds, comprehensive knowledge of particle residence times in the bed becomes crucial. Information concerning the gross mixing behaviour of solids can be acquired from either theoretical models or stimulus-response experiments. Saidutta and Murthy (2000) conducted stimulus-response experiments in four different rectangular columns including two and three spout cells. Beltramo et al. (2009) studied the hydrodynamics of a multiple square-based reactor for solid state polymerization. They revealed that levels in the various sectors were very hard to control, due to reciprocal interference between spouted bed cells at high solids throughout. Gan et al. (2013) studied experimentally the lateral mixing of particles in a quasi-slot-rectangular spouted bed by means of the bed collapse method with a set of partitions. The effects of particle size, superficial gas velocity, static bed height, and air inlet width on the effective lateral dispersion were tested.  Particles in spouted beds undergo vigorous movement, improving their contact with the fluid. Particle size plays an important role with respect to solids mixing (Ren et al., 2012; Wang et al., 2016b), minimum spouting velocity (Olazar et al., 2006) and residence time distribution, even affecting the discharge method of the product. Ren et al. (2012) investigated the effects of particle shape and size on solids mixing in a conical-cylindrical spouted bed. They found that the mixing quality improved for particles of narrow size distribution and sphericity. Olazar et al. (2006) identified particle size as an important property affecting the minimum spouting velocity at elevated temperatures.  5  1.2 Biomass torrefaction From a climate change perspective, biomass is considered a carbon neutral fuel (Chen et al., 2015), and a promising alternative to fossil fuels. Plants grow by absorbing CO2 from the atmosphere, as well as water and nutrients from soils, followed by converting them into hydrocarbons through photosynthesis. The carbon contained in biomass is gained from CO2; in other words, carbon is cycled in the atmosphere when biomass is consumed as a fuel. Bioenergy is the largest renewable energy source, with 14% out of 18% renewables in the energy mix and 10% of global energy supply, according to World Energy Resource (2016), who defined bioenergy to consist of traditional biomass (i.e. forestry and agricultural residues), as well as modern biomass and biofuels. In Canada, bioenergy currently accounts for approximately 6% of Canada’s total energy supply ( Natural Resources Canada (2017)).  Biomass can be utilized as a solid fuel, burned directly for the generation of heat and power, or converted into liquid or gaseous fuels via a variety of methods, such as pyrolysis, gasification, anaerobic digestion, fermentation and transesterification. Raw biomass is characterized by heterogeneity, high moisture content, low bulk density, low calorific value, pliability and a hygroscopic nature, all of which challenge its utilization, collection, grinding, storage, transportation and feeding into the reactor. In the past, a number of biomass pretreatment methods have been explored to address those disadvantages. Among the potential biomass upgrading methods, torrefaction is a promising route for solid fuel production. Torrefaction is a pretreatment process, where biomass particles are heated to modest temperatures of 200-300°C in the absence or near absence of oxygen. The benefits of torrefaction include a product with (1) higher heating value or energy density; (2) reduced 6  atomic O/C and H/C ratios; (3) lower moisture content; (4) higher water-resistivity; (5) improved grindability and reactivity; and (6) more uniform biomass properties (Chen et al., 2015). Table 1-1 lists key properties of raw biomass, torrefied biomass and lignite, a low-rank coal. After undergoing torrefaction, the properties of torrefied biomass are closer to those of coal. The solid yield of biomass from torrefaction is in the range of 24-95 wt.%, while the energy yield consequently varies from 29-99%, depending on the operating conditions.  Table 1-1 Key properties of raw biomass, torrefied biomass and lignite, and energy yield and solid yield of torrefaction.  Parameter  Raw biomass Torrefied biomass Lignite VM (wt.%) 67-88 34-85 22-46 FC (wt.%) 0.5-20 13-45 10-53 C (wt.%) 30-56 45-68 ~62.5 O (wt.%) 31-46 11-45 ~17.2 H (wt.%) 5-13 4-10 ~4.4 HHV (MJ/kg) 15-20 16-29 ~25 Activation energy (kJ/mol) 90-100 a 110-130b 80-110 a 110-200 b ~141 Energy yield (%) N/A 29-99 N/A Solid yield (wt.%) N/A 24-95 N/A a Based on Coats-Redfern model. b Based on Kissenger-Akahira-Sunose (KAS) model. VM: Volatile matter, FC: Fixed carbon, C: Carbon, O: Oxygen, H: Hydrogen, HHV: Higher heating value. Note: The data in Table 1-1 come from Arteaga-Pérez et al. (2015); Basu (2013); Cao et al. (2016); Chen et al. (2015); Colin et al. (2017); Joshi et al. (2015); Küçükbayrak et al. (2001); Li et al. (2017); Martín-Lara et al. (2017); Strandberg et al. (2015); Sukiran et al. (2017); Uemura et al. (2017). 7  In general, the constituents of biomass include hemicellulose, cellulose, lignin, organic extractives and inorganic minerals (also called ash). The first three constituents are the main components of biomass, and their weight percentages depend on the specific biomass species. In softwood, hemicellulose, cellulose and lignin typically account for ~ 27, 42, and 28 wt.%, respectively, while, in hardwood,  they comprise ~ 30, 45 and 20 wt.%, respectively (Peng et al., 2012a). Organic extractives usually account for 3% of the mass of softwood and 5% of the mass of hardwood. Inorganic minerals generally constitute less than 1% of the overall content in wood. (I) Hemicellulose is composed of various polymerized monosaccharides, mainly glucose, mannose, galactose, xylose, arabinose, 4-O-methyl glucuronic acid and galacturonic acid residues (Mohan et al., 2006). Its basic structure can be represented by (C5H8O4)m, where m is the degree of polymerization. The abundant O-acetyl groups have poor thermostability and could be dissociated at 200-300°C (Wang et al., 2013b).  (II) Cellulose is a linear homopolysaccharide composed of β-D-glucopyranose units, linked together by (1-4)-glycosidic bonds (Balat et al., 2008). Crystalline and amorphous structures are contained in cellulose and can be expressed by (C6H10O5)m, where m represents the degree of polymerization.  (III) Lignin is a three-dimensional, highly branched polyphenolic substance that consists of an irregular array of variously bonded “hydroxy-” and “methoxy-” substituted phenylpropane units (Chen and Kuo, 2011). Its chemical formula is represented by [C9H10O3·(OCH3)0.9-1.7]m (Chen et al., 2011a).  8  Due to their distinct compositions and structures, hemicellulose, cellulose and lignin differ in their thermal decomposition characteristics. Therefore, an in-depth understanding of the thermal decomposition behaviour of these constituents assists in elucidating biomass torrefaction characteristics. Yang et al. (2007) investigated the pyrolysis characteristics of hemicellulose, cellulose and lignin, and found that hemicellulose decomposition occurred in the range of 220-315°C, whereas cellulose decomposed at 315-400°C, and lignin gradually decomposed over the extensive temperature range of 160-900°C. Meanwhile, it was found that hemicellulose pyrolysis and lignin pyrolysis were exothermic, while that of cellulose was endothermic. More lignin content in biomass led to higher exothermicity in the biomass decomposition process (Gomez et al., 2009). Thermal behaviour of cellulose pyrolysis could be driven in the exothermic direction by the charring process, which competed with tar formation (Milosavljevic et al., 1996). Chen and Kuo (2011) torrefied the five basic constituents (i.e. hemicellulose, cellulose, lignin, xylan and dextran) and two pure materials (i.e. xylose and glucose) to simulate the thermal degradation characteristics of biomass. They found no interactions among the torrefaction of hemicellulose, cellulose and lignin. Therefore, the weight loss of biomass torrefaction could be predicted from linear addition of the weight losses of the individual constituents. Wang et al. (2016a) studied the effect of torrefaction on hemicellulose characteristics and found that the main reactions were dehydration of hydroxyls and dissociation of side chains at low temperature (200-275°C), while at high temperature (>275°C) depolymerization of hemicellulose and fragmentation of monosaccharide residues were of primary importance. Wang et al. (2017b) investigated the influence of torrefaction on cellulose, and proposed that the main reactions include the 9  destruction of crystalline cellulose, decomposition of amorphous regions, cleavage of b-1,4-glycosidic bonds and dehydration of hydroxyl. Based on the torrefaction atmosphere, torrefaction can be categorized into non-oxidative torrefaction, i.e. in an inert (usual nitrogen) atmosphere, and oxidative torrefaction, i.e. containing some oxygen as in a flue gas atmosphere.  1.2.1 Non-oxidative torrefaction Untreated biomassInert gas(Nitrogen)HeatVolatilesTorrefied biomass Figure 1-2 Schematic diagram of non-oxidative biomass torrefaction. Non-oxidative torrefaction is defined as occurring when the biomass is heated at modest temperatures of 200-300°C in the absence of oxygen, i.e. in a nitrogen atmosphere, as shown in Figure 1-2. The main mechanisms of non-oxidative torrefaction consist of devolatilization and thermal degradation of biomass constituents. Biomass torrefaction is generally an endothermic process, with its heat of reaction determined by the thermal degradation characteristics of the main biomass constituents (Chen and Kuo, 2011; Yang et al., 2007).    10  (i) Effect of torrefaction temperature and reaction time Temperature is the main operation parameter affecting the torrefaction process. It influences the mechanism and kinetics of reactions (Wang et al., 2017b; Wang et al., 2016a). Reaction time is another important parameter. The effects of temperature and reaction time on the chemical and physical characteristics of biomass were investigated by Peng et al. (2013b) in a bench-scale fixed bed tubular reactor for biomass torrefaction at temperatures of 240-340°C with residence times up to 60 min. Results showed that the mass loss of BC softwood mainly depended on the torrefaction temperature and reaction time. The heating value of torrefied sawdust had a close relationship with the mass loss, increasing with increasing severity of torrefaction. Severity hereafter refers to the degree of biomass degradation due to torrefaction. Considering the quality of torrefied pellets, the optimal torrefaction condition appeared to be ~30% mass loss.  Strandberg et al. (2015) studied the effects of temperature and residence time of spruce wood in a continuous torrefaction rotary drum reactor. It was found that the effect of torrefaction temperature was greater than the effect of residence time on the torrefaction process. Increased torrefaction severity led to decreased milling energy, angle of repose, mass and energy yields, contents of volatile matter, hydrogen, cellulose and hemicellulose contents, and also resulted in increased hydrophobicity, heating values, carbon and fixed carbon contents. Wang et al. (2017a) torrefied Norway spruce stem wood, stump and bark in a bench scale tubular reactor at 225, 275 and 300°C with two residence times (30 and 60 min).  They observed that an increase in torrefaction temperature and residence time led to greater heating values and higher fixed carbon contents, and decreased hemicellulose and cellulose contents of the torrefied biomass. They also concluded that mild torrefaction could 11  significantly improve the physicochemical properties and grindability of biomass. Recari et al. (2017) explored the torrefaction of a solid recovered fuel (SRF) and its influence on the fuel properties for gasification. The results showed that the torrefaction process improved the SRF gasification parameters (resulting in lower tar, higher H2/CO ratio, carbon conversion, etc.) and strongly affected the concentration of HCl in the producer gas.  (ii) Effect of particle size When torrefying biomass, particle size is an important parameter due to varying conditions experienced by reaction intermediates (Chen et al., 2015; Ciolkosz and Wallace, 2011). Peng et al. (2012b) studied the effect of particle size less than 1 mm on biomass torrefaction in a thermogravimetric analyzer (TGA) and in a tubular fixed bed reactor.  The torrefaction rate was found to be influenced by the particle size, especially at high temperature. Bates and Ghoniem (2014) compared predictions of a one-dimensional model accounting for the effects of temperature, particle size and moisture content with the experimental results of Basu et al. (2013), who examined the effects of biomass size and shape on torrefaction in a convective bed reactor. Their results showed that the energy and mass yields increased as the length-to-diameter ratio of the biomass particles increased. They also found that torrefaction conversion decreased strongly not only with increasing particle size, but also with increasing moisture content, and suggested that it may be advantageous to have a separate grinding or drying step prior to torrefaction.  12  1.2.2 Oxidative torrefaction Untreated biomassOxygen-containing gas(Flue gas or Torgas)HeatVolatilesTorrefied biomass Figure 1-3 Schematic diagram of oxidative biomass torrefaction. Oxidative torrefaction is also a pretreatment method for biomass upgrading, with biomass torrefied in the presence of some oxygen at 200-300°C, as shown schematically in Figure 1-3. As mentioned above, non-oxidative torrefaction is carried out with the inert gas, generally nitrogen, requiring a gas separation process to capture nitrogen, thereby increasing the operating cost of the torrefaction. Recycling of torrefied gas or flue gas from a combustor (Mei et al., 2015; Uemura et al., 2017) is promising as a means of reducing cost. The oxygen concentration in the flue gas generally varies from 3 to 16 vol.% (Uemura et al., 2017). The main mechanisms of oxidative torrefaction include not only devolatilization and thermal decomposition of biomass, but also oxidation reactions. These oxidation reactions are usually exothermic, providing the heat for thermal degradation of biomass and thereby reducing the heat demand of torrefaction. Moreover, the rate of oxidation reactions is generally faster than those of devolatilization and decomposition reactions, leading to shortening of the torrefaction duration. In addition, to achieve the same weight loss, a lower torrefaction temperature is required for oxidative torrefaction than for non-oxidative torrefaction. 13  Overall, an increase in the oxygen concentration leads to lower mass and energy yields of biomass, reduced H/C molar ratio, and lower HHV of the torrefied product. Rousset et al. (2012) investigated oxidative torrefaction of Eucalyptus grandis in a macro-TGA and tested the influences of two temperatures (240 and 280°C) and four gas oxygen concentrations (2, 6, 10 and 21 vol.%) on the properties of the torrefied product. They found that the oxygen concentration did not significantly influence the solid yield and composition of the biomass torrefied at 240°C. The influence of oxygen concentration on solid yield reduction and torrefied biomass properties at 280°C was more marked than at 240°C.  Uemura et al. (2013) torrefied oil palm fruit bunches of four sizes (0.375, 1.5, 3 and 6 mm) in a fixed-bed reactor in the absence and presence of oxygen (0, 3, 9, and 15 vol.%) at three temperatures (220, 250 and 300°C). The results showed that the mass and energy yields of biomass decreased with increasing temperature and oxygen concentration, but were not influenced by the biomass particle size. They reported that the oxidative torrefaction process took place in two successive steps or via two parallel reactions, where the first was ordinary (oxygen-free) torrefaction and the other was biomass oxidation. Chen et al. (2013a) explored the effect of the superficial velocity, i.e. of the flow rate of carrier gas on non-oxidative and oxidative torrefaction of biomass. They proposed that the oxidative torrefaction of biomass was dominated by surface oxidation, and that there existed an upper limit for air superficial velocity beyond which the thermal degradation of biomass was governed by internal mass transport, rather than was by surface oxidation.  Joshi et al. (2015) reported that there were temperature-specific limits beyond which an increase in oxygen concentration likely resulted in an oxidative thermal runaway in their packed bed. An increase in torrefaction temperature from 270 to 290°C for bagasse was found to reduce the combustive limits of oxygen 14  concentration from 5 vol.% to 1 vol.%. Furthermore, it was determined that adding some oxygen did not significantly reduce the mass and energy yield of the torrefied biomass, but could lead to more uniform heating of the bed. Uemura et al. (2017) torrefied biomass in combustion flue gas with oxygen contents of 8.1-8.9 vol.% and carbon dioxide content of 13.1-13.5 vol.% at 200-300°C. They reported that the mass yield of empty fruit bunches (EFB) torrefied in combustion gas was lower than that torrefied in a nitrogen atmosphere, and the solid yield decreased and the calorific value increased with increasing torrefaction temperature in a combustion gas environment. Wang et al. (2013a) investigated oxidative torrefaction of sawdust using carrier gases containing 0, 3 and 6 vol.% of oxygen at 250, 270 and 290°C in a fluidized bed reactor, followed by pelletization. They found that the properties of oxidatively torrefied sawdust particles and corresponding pellets, especially density, energy consumption for pelletization, HHV and energy yield, were similar to those of non-oxidatively torrefied sawdust. Mei et al. (2015) performed cedar wood torrefaction in a pilot-scale rotary kiln with nitrogen and flue gas atmospheres. The use of an industrial flue gas significantly influenced the behaviour of the cedar wood during torrefaction and the properties of the solid products. The samples torrefied in the flue gas environment had combustion characteristics similar to lignite.   1.3 Torrefaction reactor Torrefaction has been studied in different reactors, i.e. fixed beds (Joshi et al., 2015; Peng et al., 2013b; Wang et al., 2017a), rotary drums (Colin et al., 2017; Mei et al., 2015; Strandberg et al., 2015), screw conveyor (Nachenius et al., 2015), fluidized beds (Atienza-Martínez et al., 15  2013; Li et al., 2012b; Wang et al., 2013a), microwave reactor (Huang et al., 2017), compact moving bed (Verhoeff et al., 2011), and Torbed reactor (Janssen and Konings, 2000), etc.  Spouted beds were originally developed for gas-solids contacting of coarse particles (Mathur and Gishler, 1955), e.g. for drying of wheat. Spouted beds have found some applications in energy conversion due to their excellent gas-solids heat and mass transfer performance. Abundant research is available on pyrolysis of wood (Fernandez-Akarregi et al., 2013), scrap tyres (López et al., 2010), and sewage slurries (Alvarez et al., 2015) as well as on gasification (Erkiaga et al., 2014; Erkiaga et al., 2013a; Erkiaga et al., 2013b; Lopez et al., 2017) in spouted bed reactors. However, application of slot-rectangular spouted beds is very limited. The application has mainly focused on particle coating (Donida and Rocha, 2002; Wiriyaumpaiwong et al., 2003) and powder drying (Braga et al., 2015). Our literature review reveals that they have not been previously utilized to torrefy woody biomass. Spouted beds are capable of handling large (>1 mm) particles, such as typical biomass particles. Moreover, spouted bed reactors can provide excellent particle-particle and particle-fluid heat transfer. In addition, the slot-rectangular spouted bed reactor is relatively easy to scale up, compared to conventional spouted beds (Grace and Lim, 2011). These advantages suggest that slot-rectangular spouted beds should be investigated for their potential to serve as commercial torrefaction reactors.  1.4 Research objectives and principal tasks in this thesis project • Investigate solids mixing in a dual-compartment slot-rectangular spouted bed reactor (DSRSB) with a suspended partition at room temperature, while also providing 16  crucial information needed for the design and operation of dual-compartment slot-rectangular spouted bed reactors. • Assess the feasibility and reliability of biomass torrefaction in a single-compartment slot-rectangular spouted bed, provide data for evaluation and optimization of the performance of slot-rectangular spouted bed torrefaction reactors, and compare to other types of solid-gas reactors. • Evaluate the scale-up of a single-compartment slot-rectangular spouted bed reactor to a dual-compartment slot-rectangular spouted bed reactor.  • Test the feasibility and reliability of biomass torrefaction in a dual-compartment slot-rectangular spouted bed equipped with a suspended partition, and provide data with respect to evaluation and optimization of the performance of slot-rectangular spouted bed torrefaction reactors. • Study the effects of temperature, biomass feed rate, sawdust particle size and oxygen concentration on torrefied product properties in slot-rectangular spouted beds. To achieve the above-mentioned objectives, the principal tasks include: • Design, construct and operate a cold transparent model of a dual-compartment slot-rectangular spouted bed with a suspended partition to investigate the solids mixing between two adjacent compartments.  • Design, construct and operate an experimental platform of biomass torrefaction including preheating system, a single-compartment slot-rectangular spouted bed/a dual-compartment slot-rectangular spouted bed, feeding system, cooling system, combustor and data acquisition system.  17  • Determine the thermal degradation behaviour of sawdust by means of TGA. • Explore the effects of torrefaction temperature, biomass feed rate (reaction time), and sawdust particle size on biomass torrefaction in a semi-batch single-compartment slot-rectangular spouted bed reactor. • Investigate the influence of oxygen concentration in the feed-gas during oxidative torrefaction in a semi-batch single-compartment slot-rectangular spouted bed reactor.  • Test the effects of biomass feed rate, torrefaction temperature and oxygen concentration in the feed-gas on biomass torrefaction in a dual-compartment slot-rectangular spouted bed reactor.  • Explore the influence of torrefaction on biomass properties, such as: proximate analysis, elemental analysis, fiber analysis, higher heating value, and particle density, SEM measurement, FTIR analysis, and analysis of composition of the gas and liquid products, such as CO, CO2, H2, CH4, and water content.  • Study the chemical kinetics of pyrolysis of raw and torrefied sawdust based on TGA tests. 1.5 Thesis outline Chapter 1 reviews the literature on slot-rectangular spouted beds and biomass torrefaction, and provides an introduction to the current work.  Chapter 2 describes the detailed experimental set-up and operating procedures. It also provides the detailed information on the experimental biomass material, characterization methodology for torrefied product and the procedures used. 18  Chapter 3 presents the results of experiments which determined solids exchange between two adjacent compartments in a dual-compartment slot-rectangular spouted bed with a suspended partition.  Chapter 4 investigates biomass torrefaction in a semi-batch slot-rectangular spouted bed reactor, including the effects of torrefaction temperature, biomass feed rate, sawdust particle size and non-oxidative vs. oxidative environment on torrefaction performance and torrefied product properties.  Chapter 5 explores experimental biomass torrefaction in a semi-batch dual-compartment slot-rectangular spouted bed with a suspended partition, including effects of torrefaction temperature, biomass feed rate, non-oxidative vs. oxidative environments on torrefaction performance and torrefied product properties. Chapter 6 tests thermogravimetric characteristics and kinetics of pyrolysis of torrefied sawdust in a thermogravimetric analyzer. The torrefied biomass was obtained from the biomass torrefaction in a single-compartment slot-rectangular spouted bed reactor. Chapter 7 provides conclusions from the current work and recommendations for future studies.  19  Chapter 2 Sample preparation, experimental set-up and methodology 2.1 Cold model dual-compartment slot-rectangular spouted bed  2.1.1 Equipment design           (a)                        (b)  Figure 2-1 (a) Schematic and definition of symbols for dual-compartment slot-rectangular spouted bed column; (b) top view. The configuration of the dual-compartment slot-rectangular spouted bed (DSRSB) with a suspended partition is shown schematically in Figure 2-1(a). The column consists of a Plexiglas vessel, 300 × 100 mm in inner cross-section, 800 mm in height, supported by a W-type lower divergent section with 60° inclination angles. A thin partition of width 100 mm is 20  suspended centrally and rigidly in the upper freeboard section to create two compartments of equal volume. Two horizontal-axis ports (A and B), just above the distributor, are available for insertion of a probe from the wall to measure the bed pressure drop. Two rectangular slots shown in Figure 2-1(b) are located symmetrically at the base, each connected to an independent windbox. The DSRSB column dimensions are listed in Table 2-1. Table 2-1 DSRSB column dimensions. See Figure 2-1 for definition of dimensions. Parameter Value Overall height, Ht (mm) 800 Width, W (mm) 300 Thickness, L (mm) 100 Base width, Wb (mm) 25.4 Gas entry slot width, Wslot (mm) 4.0 Gas entry slot length, Lslot (mm) 30 Divergent base height, Hd (mm) 120 Included base angle, θ (°) 60 Suspended partition height, Hsp (mm) 100 and 300 Suspended partition width, Lsp (mm) 100 Suspended partition thickness, Wsp (mm) 6.35 Gap between partition and static bed surface, ∆h (mm) 20, 40, 60 Static bed height, HB,0 (mm) 150-210     21  2.1.2 Material and methodology Table 2-2 Key properties of experimental particles and gas.  Value Particle colour Red Black Sauter mean particle diameter, dsv (mm) 1.16, 1.61, 2.85 1.16, 1.61, 2.85 Particle density, ρp (kg/m3) 2530 2530 Sphericity, Φ (-) 1.0 1.0 Angle of repose (°) 22.0, 20.4, 22.8 19.9, 21.8, 23.3 Gas (air at 20°C and 1 atm)  Density, ρg (kg/m3) 1.2 Viscosity, μg (kg/m.s) 1.88e-5 Superficial spouting velocity, U (m/s) 0.34-1.4 The spouting fluid was air at 20°C and 1 atm. The particles studied were glass beads. The key particle properties and fluid properties are listed in Table 2-2. The measurements of angle of repose (AOR) of samples were performed on an AOR tester (Mark 4, Powder Research Ltd., developed by Professor Derek Geldart) (Geldart et al., 2006). Figure 2-2 shows a schematic diagram of the experimental dual-compartment slot-rectangular spouted bed with a suspended partition. Compressed air was first measured by a rotameter and then divided into two equal streams. Each divided line had a globe valve and an orifice flowmeter to control and measure the air flow precisely. To quantitatively analyze the exchange of particles between the chambers, half of the original glass beads were painted red and half black. The red and black solids had virtually identical density, diameter and sphericity. The red particles were assigned to be tracers. The partition was first inserted into the upper section at position (1) to separate the DSRSB column into two independent 22  compartments. The left compartment was next loaded with red tracer particles, while the right one was filled to the same height (HB,0) with black particles. The air flow rate was then gradually increased to the desired value. When both compartments reached the same stable spouting state, the partition was quickly raised to its required position (2). After each pre-determined time interval, the air flow was abruptly terminated. Next, a 4 mm thick vertical partition of width × height 100 × 300 mm was inserted in position (1) to divide the bed into two identical volumes. The particles from each compartment were then discharged by an electric vacuum cleaner, and ten 70 g samples were taken from the discharged particles for each compartment. The glass beads were in a single layer, when each sample was photographed. The number of particles of each colour in each sampled mixture was then determined by image analysis with Adobe Photoshop CS4 software, and represented by the area of each colour in one picture, as shown in Figure 2-3. The tracer mass fraction of each sample was then calculated from the ratio of the area of the tracer particles divided by the total area of the particle mixture.  Each measurement was repeated, so that each measured concentration was based on an average of 20 determinations.  23      (a) DSRSB DSRSB: Dual-compartment Slot-Rectangular Spouted BedF1: RotameterF2, F3: Orifice FlowmeterP: Pressure GaugeΔP: Differential Pressure TransducerV1: Ball ValveV2, V3: Globe ValveAirV1V2V3 F3F2F1ΔP2ΔP1PPosition (1)Position (2) (b) Figure 2-2 (a) Photo of cold model dual-compartment slot-rectangular spouted bed facility, (b) schematic diagram of dual-compartment slot-rectangular spouted bed with a suspended partition. DSRSB  Air 24     Figure 2-3 Photos of red+black particles, black particles, and red particles.  Variables studied in the solids mixing between adjacent chambers involved superficial gas velocity (U), static bed height (HB,0), particle size (dp), partition position (Δh) and partition height (Hsp). Details are shown in Figure 2-4. VariablesGas velocity (U)Static bed height (HB,0)Particle size (dp)Partition position (Δh)Partition height (Hsp)1.1-1.4Ums150, 180, and 210 mm1.16, 1.61 and 2.85 mm20, 40, and 60 mm100 and 300 mm Figure 2-4 Operating variables and their ranges in solids mixing experiments.  Black  Red+Black Red 25  2.2 Sample preparation for torrefaction experiments The experimental material, obtained from Tolko Industries Ltd., Vernon, BC, Canada, is a mixture of spruce, pine and fir (SPF) sawdust from a sawmill. The SPF sawdust was first dried in an oven (Model: LHT6, Carbolite) at 105°C for 24 h, and ground by a hammer mill (Model: 10HMBL, Glenmills Inc., USA,) with a 3.18 mm screen. The ground sawdust was subsequently sieved for 10 min in a Gilson Test-Master sieving device (Gilson Company Inc., Lewis Center, Ohio) to select the fractions to be chosen for the experiments. The 0.25-0.5, 0.5-1.0 and 1.0-2.0 mm sawdust particles are designated hereafter as the smallest, intermediate and coarsest particles. The sawdust was dried again at 105°C for 24 h prior to each torrefaction experiment.  Key properties of the three sawdust selected particles are given in Table 2-3. The bulk density was measured by half-filling a graduated cylinder with a weighed sawdust sample, inverting the covered cylinder and then quickly re-inverting it to measure the volume. The particle density was determined by a helium displacement pycnometer (Quantachrome Instruments, USA). The moisture content was measured by weighing a sample of raw sawdust before and after putting it into an oven (THELCO laboratory PRECISION oven) at 105°C for 24 h. Elemental analysis, proximate analysis, fiber analysis, and higher heating value (HHV) are described in section 2.6.1 below.     26  Table 2-3 Properties of raw SPF sawdust and glass beads.  Raw sawdust Glass beads Screen size range (mm) 0.25-0.5 (Smallest) 0.5-1.0 (Intermediate) 1.0-2.0 (Coarsest) 0.85-1.18 Sauter mean diameter, dsv (μm) 263 433 1454 1000 Bulk density, ρb (kg/m3) 87.7 137.6 220.5 1550 Moisture content wet basis (wt.%) 3.9 3.1 3.3 0 Particle density dry basis, ρp (kg/m3) 1231 1254 1239 2526 Proximate analysis d         Volatile Matter (VM) (wt.%) 84.55      Fixed Carbon (FC) (wt.%) 15.06      Ash (wt.%) 0.39  Fiber analysis d         Hemicellulose (wt.%) 14.6      Cellulose (wt.%) 49.0      Lignin (wt.%) 27.6      Extractives (wt.%) 8.4  Ultimate analysis d         C (wt.%) 46.2      H (wt.%) 6.4      O* (wt.%) 47.4      N (wt.%) <0.1  HHV (MJ/kg) 18.23  d Dry basis. * Oxygen content was determined by difference. 27  2.3 Single-compartment slot-rectangular spouted bed torrefaction reactor  2.3.1 Equipment design 60oFeed portFluid InletOutletΔPHt=1000 mm45oZ=180 mm(a)T1T2T3T4Z=165 mmZ=267 mmZ=64 mmQuartz windowP                    Figure 2-5 (a) Slot-rectangular spouted bed reactor; (b) plan view of base. (T1-T4: K type thermocouples, P: Pressure gauge with range 0-103.4 kPa, ΔP: differential pressure transducer with range 0-6.9 kPa.) The single-compartment slot-rectangular spouted bed reactor (SRSB) shown in Figure 2-5(a) was made of carbon steel, 150 × 100 mm in cross-section and 1000 mm in height, with a diverging base of 60° inclination angles. A plan view of the base is shown in Figure 2-5 (b). The width of the base (Wb) was 25.4 mm. The slot length (Lslot) was 30 mm, and its width (Wslot) was 4 mm. To observe solids motion in the reactor, five quartz windows of diameter 50.8 mm were installed on its front wall.  Thermocouples (K type) were used to measure gas WslotLslotWbW=150 mmL=100 mm(b)28  temperatures in the windbox and in the spouted bed, 64, 165 and 267 mm above the base. The latter three thermocouples measuring the spouted bed temperatures were inserted 20 mm from the inside wall of the spouted bed column. The pressure drop (ΔP) across the reactor was measured by a differential pressure transducer (Omega model: PX142 series, uncertainty: ±0.75% full span). The pressure transducer probe was just above the distributor. A feed port of diameter 25.4 mm was located on one side, 180 mm above the base, as shown in Figure 2-5(a). 2.3.2 Methodology The semi-batch slot-rectangular spouted bed torrefaction facility is shown schematically in Figure 2-6. The inlet gas flow was controlled by a needle valve upstream of the preheater, and measured by an orifice flowmeter. The flow rate measured by the orifice flowmeter was converted to standard conditions based on the fluid temperature and pressure.  3000 g of glass beads of 1.0 mm diameter were loaded into the reactor before the experiments.  At the beginning of each run, air was employed to preheat the reactor to the required temperature. Subsequently, nitrogen replaced air to purge the system for 5 min. The nitrogen or a mixture of nitrogen and air was then fed to the reactor with a pre-determined flowrate and oxygen concentration. The oxygen concentration in the feed-gas was controlled by mixing air and nitrogen at different flowrates, and the two gas flows were controlled individually, aided by two rotameters. A flue gas analyzer (PS-200, HORIBA) was connected to gas sampling port #1 to ascertain the oxygen concentration prior to biomass feeding.  29  Preheater Natural gasAirScrew-feederCycloneHeat ExchangerAfterburnerSRSB ReactorHopperHeating TapeFilterGas sampling port #1Gas sampling port #2Solid sampling portP3P4V3V4 F3TcTheN2T2T3T4T1T0AirV2F2F1N2V1P1 P2 Figure 2-6 Schematic diagram of slot-rectangular spouted bed torrefaction facility. (F1, F2: Rotameter with range 0-51 m3/h, F3: Orifice flowmeter with range 0-51 m3/h, P: Pressure gauge with range 0-103.4 kPa, T1-T4, Tc and The: K-type thermocouples, V1, V2: Needle valve, V3: Ball valve, V4: Globe valve.) After the system had stabilized for 5 min, the screw-feeder was turned on to initiate the feeding of sawdust into the reactor. The screw-feeder had been calibrated at room temperature with the SPF sawdust before being employed in the experiments, and the feed rate was calculated as the actual weight of fed biomass divided by the duration of feeding. To prevent hot gas from entering the screw-feeder, 4 L/min nitrogen was fed into it. Sawdust was fed into the reactor, contacting the heat carriers, i.e. the heated glass beads, and N2/N2+air. There was no immediate discharge of torrefied material. Instead, the sawdust particles lost density as they devolatilized, with some then being pneumatically transported out of the reactor. Most entrained particles were captured by a cyclone. The clean gas then 30  passed to a stainless-steel mesh filter of 50 μm aperture to remove the remaining particles. Finally, the off-gas entered an afterburner after being cooled to room temperature.  During the torrefaction experiments, torrefied product samples of ~2 g were taken from a solid sampling port just below the bottom of the cyclone (See Figure 2-6) at 5-10 min intervals. The cyclone was heated to the same temperature as the spouted bed reactor by a heating tape. The filter was close to the cyclone, so no condensation was observed inside it.  After cooling to less than 50°C, the mixed torrefied sawdust and glass beads were extracted from the reactor by a vacuum cleaner. Glass beads were recycled after being separated from the torrefied sawdust. Each torrefied solid sample was sealed in a zip bag, labeled and stored in a cold storage room (< -4°C) for future analysis.  SRSB ReactorTemperature (T)Feed rate (F)Particle size (dp)N2, N2+O2 (XO2)240, 270, 300, and 330oC220, 440, and 710 g/h0.25-0.5, 0.5-1.0, and 1.0-2.0 mm0, 3, 6, and 9 vol.% O2 Figure 2-7 Experimental variables studied and their nominal ranges in the slot-rectangular spouted bed torrefaction facility.  Figure 2-7 lists all experimental variables investigated in the torrefaction experiments. The average reactor temperature (T) was designed to be varied from 240 to 330°C. The biomass 31  feed rates (F) were 220, 440 and 710 g/h. The oxygen concentration (XO2) in the feed gas was controlled to be 0, 3, 6 or 9 vol.%. The mass of the inert particles was fixed to be 3000 g. The freeboard was defined as the region from the static bed surface to the top of the column; its height was 673 mm.  In order to continuously conduct a torrefaction experiment for 120 min, the superficial gas velocity was set at 1.2Ums, where Ums was the minimum spouting velocity of inert particles measured at the elevated temperature and initial bed depth at which the experiment was to be conducted.  32  2.4 Dual-compartment slot-rectangular spouted bed torrefaction reactor  2.4.1 Equipment design 60oFeed portInletOutlet(a)T'1T'3T'5T'4T'6T'7Z=180 mmHt=1000 mmZ=165 mmZ=267 mmZ=64 mmT'2ΔP2ΔP1 Inlet45oQuartz windowP            WslotLslotWbW=300 mmL=100 mmWslotLslotWb(b)  33          Figure 2-8 (a) Schematic of dual-compartment slot-rectangular spouted bed reactor, (b) plan view of base, (c) photo of DSRSB column, (d) photo of W type base and windbox. (𝑇1′-𝑇7′: K type thermocouples, P: Pressure gauge with range 0-103.4 kPa, ΔP: differential pressure transducer with range 0-6.9 kPa.) The dual-compartment slot-rectangular spouted bed reactor was made of carbon steel. Its configuration is shown in Figure 2-8(a). The DSRSB column dimensions were 1000 × 300 × 100 mm in height × width × thickness, with a W-type base of 60° inclination angles. DSRSB had twice the capacity of the SRSB. A 6.4 mm thick partition of width × height 100 × 100 mm was suspended rigidly and centrally in the freeboard. Two identical slots as shown in Figure 2-8(b) were installed, one for each compartment, each connected to an independent c d 34  windbox. Each compartment contained five quartz windows of 50.8 mm diameter for observing solids motion in the reactor. Two 𝑇1′ and 𝑇2′ (K type) thermocouples were used to measure gas temperatures in the two independent windboxes, and five thermocouples of 𝑇3′ - 𝑇7′  (K type) in the spouted bed 64, 165 and 267 mm above the top of the slot. The thermocouples measuring the spouted bed temperatures were inserted 20 mm from the inside wall of the column. A feed port of diameter 25.4 mm was located on the right side of the reactor, 180 mm above the base. The pressure drop (ΔP) across the reactor was measured by two differential pressure transducers (Omega model: PX142 series, uncertainty: ±0.75% full span). The pressure transducers probes were installed immediately above the distributor. 2.4.2 Methodology Biomass was also torrefied in the dual-compartment slot-rectangular spouted bed facility shown in Figure 2-9. The experimental operating procedures for biomass torrefaction in the dual-compartment slot-rectangular spouted bed facility were the same as those for the single-compartment slot-rectangular spouted bed reactor (see section 2.3.2). However, 8.5-13.6 m3/h of the off-gas was recycled by a blower to reduce the operating cost, as shown in Figure 2-9. The blower was modified from an air compressor (Model: 5Z28B-2, SPEEDAIRE) by adding a frequency inverter (SMVector Series, Lenze/AC Tech, USA) to control the motor speed.    35   DSRSB Screw-Feeder Cyclone Preheater Filter (a) 36  Preheater AirBlowerScrew-feederCycloneHeat ExchangerAfterburnerHopperHeating TapeDSRSB ReactorGas sampling port #2Gas sampling port #1Solid sampling portP3P4V3V7V6V5V4F5F4F3P5V8TheTcN2AirV2F2F1N2V1P1 P2T’3T’5T’4T’6T’7T’1 T’2T01 T02(b)Natural gas Figure 2-9 (a) Photo of dual-compartment slot-rectangular spouted bed torrefaction facility, (b) schematic diagram of dual-compartment slot-rectangular spouted bed torrefaction facility.  (F1, F2: Rotameter with range 0-51 m3/h, F3: Rotameter with range 0-25.5 m3/h, F4, F5: Orifice flowmeter with range 0-51 m3/h, P: Pressure gauge with range 0-103.4 kPa, T: K-type thermocouple, V1, V2 and V7: Needle valve, V3, V6 and V8: Ball valve, V4, V5: Globe valve.) The operating variables for biomass torrefaction in the dual-compartment slot-rectangular spouted bed facility included temperature (T), biomass feed rate (F), and oxygen concentration (XO2), as summarized in Figure 2-10. To ensure that the biomass torrefaction results from the SRSB and DSRSB reactors were comparable, the average bed temperature of the DSRSB reactor and oxygen content were required to be the same as those for the SRSB facility. Because the DSRSB reactor was scaled up to give twice the capacity of the SRSB 37  reactor, the biomass feed rate was set at double that of the SRSB facility. The partition was rigidly installed in the freeboard zone, 130 mm above the initial bed surface.   In order to continuously conduct a torrefaction experiment of duration 120 min, the superficial gas velocity was set at 1.2Ums, where Ums is the minimum spouting velocity of inert particles measured at the elevated temperature and initial bed depth of interest.   DSRSB ReactorTemperature (T)Feed rate (F)240, 270, and 300oC600, 900, and 1400 g/h0, 3, 6, and 9 vol.% O2N2, N2+O2 (XO2) Figure 2-10 Experimental variables studied and their nominal ranges in the dual-compartment slot-rectangular spouted bed torrefaction facility. 2.5 Thermogravimetric analysis for kinetic study and experimental design (I) Dynamic tests The raw sawdust was first tested in a thermogravimetric analyzer, TGA (TA-60WS, Shimadzu) at different temperatures and atmospheric pressure with pure nitrogen carrier gas of 50 ml/min to identify suitable temperature ranges for the spouted bed torrefaction experiments. Dynamic thermal degradation experiments were performed by: (1) heating 38  samples from room temperature to 110°C at a heating rate of 10°C/min, (2) holding at 110°C for 30 min, and (3) then heating from 110 to 800°C at a rate of 1°C/min. (II) Isothermal torrefaction experiments Isothermal experiments in the TGA were carried out to identify the weight loss characteristics of the raw sawdust at different temperatures. A ~10 mg sample was first heated to 110°C at a heating rate of 10°C/min, then held at that temperature for 30 min. Then, the sample was heated at 50°C/min to reach a pre-determined torrefaction temperature, where it was held for 8 h. After this, the sample was heated to 800°C at 50°C/min and kept there for 30 min to complete the process. The experiments were conducted in an inert atmosphere of nitrogen (99.999% purity) with a flow rate of 50 ml/min. The three torrefaction temperatures chosen for the experiments were 240, 270 and 300°C. (III) Kinetic study  The pyrolysis characteristics of the raw and torrefied SPF sawdust were investigated in the thermogravimetric analyzer (DTG-60, Shimadzu).  A ~10 mg sample was heated from room temperature to 105°C, and held there for 5 min to remove the free moisture. The sample was then heated further at a heating rate of 10, 20, 30 or 40°C/min to 800°C. Thermogravimetric data were recorded at each second during the tests. These analyses were conducted in an inert atmosphere of nitrogen (99.999% purity) with a nitrogen flow rate of 100 ml/min.    39  2.6 Characterization of torrefied product The torrefaction experiments primarily produced solid product, but also condensable liquid and gas by-products, as shown in Figure 2-11.  TorrefactionGas ProductSolid ProductLiquid ProductBiomass• Proximate analysis • Elemental analysis• HHV measurement• Particle density• Proximate analysis• Elemental analysis• Fiber analysis• HHV measurement• Particle density• Particle size distribution • FTIR analysis• SEM analysisFiltrated by FilterRemain inReactorCaptured by Cyclone• Component measurement (CO, CO2, H2, CH4 )• Component measurement• Water content Loss Figure 2-11 Torrefied product map and corresponding measurements. 2.6.1 Solid product As indicated in Figure 2-11, the torrefied solid product remained in the reactor, was captured by the cyclone, or was collected by the downstream filter. However, most solid product (> 99 wt.%) came from the cyclone and reactor. Because the torrefied sawdust from the filter constituted less than 1 wt.% of the total torrefied solid product, its properties were not analyzed.  40  The volatile matter of the solid sample was measured by the TGA (Shimadzu, TA-60WS). A dried sample was first heated to 110°C at a rate of 10°C/min, and held at that temperature for 30 min to determine the moisture content. Then, the sample was heated to 800°C at 40°C/min and kept there for 30 min to complete the process. The volatile matter was determined from the weight loss of the sample between 110 and 800°C. The ash content was determined based on the NREL/TP-510-42622 method (Sluiter et al., 2008a). The fixed carbon content was subsequently determined by difference. Ultimate analysis of raw and torrefied SPF sawdust was conducted using a Carlo Erba EA 1108 elemental analyzer, providing the carbon, hydrogen, nitrogen and oxygen elemental contents of the samples in wt.% on a dry basis, with oxygen content estimated by difference. The fiber analysis was carried out with a Gerhardt Fibretherm FT12. The higher heating value (HHV) of the samples was measured with a Parr 6100 bomb calorimeter, with the raw and torrefied SPF sawdust pelletized prior to the HHV measurements to ensure controlled combustion. The particle density was determined by a helium displacement pycnometer (Quantachrome Instruments, USA). Particle size distributions were measured by a Malvern Mastersizer 2000. The structural composition of the raw and torrefied product was analyzed by FTIR spectroscopy (Cary 600 Series, Agilent Technologies). The microscopic structure of raw and torrefied sawdust was investigated by a scanning electron microscope (FEI Quanta 650). For all property measurements, two replicates were conducted, with the average and standard deviation then calculated.  41  2.6.2 Condensable liquid product Chemical compositions of the condensable liquid were analyzed by a Gas Chromatograph (Agilent 7820 A GC system) coupled with a Mass Spectroscope (Agilent 5975 MS detector) from Agilent Technology (California, United State), with a HP-Innowax column (60 m × 0.250 mm × 0.25 µm). Prior to the GC-MS measurements, samples of the condensable liquid were diluted with acetone, with a mass ratio of acetone-to-sample of 2:1. The GC injector temperature was set at 250°C, with a split ratio of 100:1. During analysis, the GC oven temperature profile was set to follow the sequence: 50°C for 2 min, next increased to 150°C at a ramping rate of 15°C/min, and again increased to 260°C with a ramping rate of 5°C/min, then held for 5 min. The water content of the condensable liquid was determined by Karl Fischer titration (ASTM D1744). Each measurement was made twice for each sample, with the average and standard deviation then calculated.  2.6.3 Gas product The CO and oxygen concentrations in the feed-gas and off-gas were measured by a flue gas analyzer (PS-200, HORIBA), with measurement ranges of 0-25 vol.% and 0-5000 ppm for oxygen and CO, respectively. The CO2 concentration in the off-gas was measured by a CO2 detector (Model 906, Quantek Instruments) with a measurement range of 0-5000 ppm. Other compounds of the off-gas, namely H2, CH4 etc., were only present at low concentrations, outside of the ranges of our GC and flue gas analyzer (H2 > 0.5 vol.%, CH4 > 0.5 vol.%). Therefore, only O2, CO and CO2 concentration data were obtained.  42  Chapter 3 Solids mixing in a dual-compartment slot-rectangular spouted bed with a suspended partition As mentioned in Chapter 1, the slot-rectangular geometry is beneficial in overcoming the scale-up challenge of spouted beds. An internal partition is essential to minimize interference between compartments and to achieve stability in multiple spouted beds. In this chapter, a dual-compartment slot-rectangular spouted bed reactor (DSRSB) with a suspended partition was investigated to address the scale-up issue, while also providing crucial information needed for the design and operation of dual-compartment slot-rectangular spouted bed reactors, e.g. for biomass torrefaction. A solids exchange coefficient was determined to characterize the solids exchange rate between adjacent compartments. The solids mixing between adjacent compartments was studied for different operating conditions of superficial gas velocity (U), static bed height (HB,0), particle size (dp), partition position (Δh) and partition height (Hsp). See section 2.1 of Chapter 2 for the operating procedure.  3.1 Quantitative study of solids mixing  As shown in Figure 3-1, before initiating the flow, 0M g of red particles were loaded into the left compartment (compartment 1), while an equal mass, 0M  g, of black particles was added to the right compartment (compartment 2). The red particles were assigned to be tracers. Initially, the weight fractions of tracer particles in compartments 1 and 2 were 1C = 1 and 2C= 0, respectively. The variation in 1C  and  with time t depends on the lateral exchange of solids between the compartments. 2C43   DSRSBM0 g red particlesC1=1 at t=0 sCompartment 1M0 g black particlesC2=0 at t=0 sCompartment 2HB,0ΔhΔh' Figure 3-1 Schematic configuration of solids mixing between compartments in DSRSB. 3.1.1 Solids exchange between compartments  Assumptions are made as follows:  (a) Each compartment is well mixed;  (b) Each compartment maintains a total mass of 0M  g particles; (c) There is a constant mass flow of m  from compartment 1 to compartment 2 and from compartment 2 to compartment 1. Mass balance for compartment 1:  1210 )( CmCmdtCMd                         (3-1) Similarly for compartment 2:  44  2120 )( CmCmdtCMd                      (3-2) Initial conditions:  at t=0, C1=1, C2=0                   (3-3) But  121  CC , i.e.  12 1 CC                    (3-4) Substitution of Equation (3-4) into Equation (3-2) yields, )21( 220 CmdtdCM     tCdtMmCdC000 22221 After integration 022)21ln(MtmC                     (3-5)  0/22 121 MtmeC                     (3-6) For our purpose, we want to find m , which is obtained from Equation (3-6) as )21ln(220 CtMm                     (3-7) This can be expressed in dimensionless form as )21ln(4Κ 200 CtUAMGmgairse                   (3-8) 45  where Kse is a solids exchange coefficient, A0 is the cross-sectional area of each of the two equal compartments (m2), Gair is the mass flow of air (kg/h), U is the superficial gas velocity (m/s), t is time (s), and C2 is tracer concentration in the #2 compartment. A greater Kse results in a faster solid exchange rate for a given air flow rate and physical system.  3.1.2 Lacey mixing index In order to quantitatively access the solids mixing degree, the well-known Lacey mixing index (Lacey, 1954) was employed to characterize the degree of solids mixing. The DSRSB has two sampling cells, the #1 and #2 compartments shown in Figure 3-1. The variance of the concentration in each sample is expressed as  1)(12,2NCCNiti                     (3-9) NitiCNC1,1                   (3-10) where N is the number of samples, tiC , is the concentration of tracer particles in sample i at time t, and 𝐶̅ is the average concentration. The Lacey mixing index ( LI ) is then defined as 220220RLI                    (3-11) where 𝜎02  and 𝜎𝑅2  represent the variances of a completely segregated mixture and a completely random mixture, respectively. Hence )1(20 pp                   (3-12) 46  nppR)1(2                    (3-13) where p and (1-p) are the proportions of the two components determined from sampling, and n is the number of particles in each sample. 3.2 Minimum spouting velocity and maximum pressure drop The experimental minimum spouting velocities and maximum pressure drops for different operating conditions are presented in Table 3-1. As expected, these critical parameters are almost identical at the same operating conditions for the red and black particles. Furthermore, with increasing static bed height, the minimum spouting velocity and maximum pressure drop increased in a similar manner. Also as expected, larger particles gave higher minimum spouting velocities and maximum pressure drops than smaller particles for a given static bed height. Table 3-1 Experimental minimum spouting velocities and maximum pressure drops. Particles HB,0 (mm)  Ums (m/s) ∆Pmax (Pa) Black Red Black Red GB1.16* 150 0.34 0.34 911 900 180 0.38 0.40 1230 1206 210 0.42 0.44 1477 1509 GB1.61* 150 0.54 0.58 1243 1293 180 0.68 0.68 1424 1439 210 0.79 0.78 1625 1600 GB2.85* 150 1.00 1.00 1480 1495         *GB: glass beads; 1.16, 1.61 and 2.85: diameter of glass beads in mm. 47  3.3 Image analysis method 0 10 20 30 40 50 60 70 80 90 1000102030405060708090100 dp = 2.85 mm  dp = 1.61 mm  dp = 1.16 mm   Measured proportion (%)True proportion (%) Figure 3-2 Comparison of measured and true mass fraction of tracer particles in mixtures. Pretests had been conducted to assess the validity of the image analysis method. Samples of known tracer mass fraction were tested by the image analysis method. Each sample contained a 70 g mixture of red and black glass beads with different tracer mass proportions from 0 to 100%. The glass beads were in a single layer, when the sample was photographed. As shown in Figure 3-2, measured values are in good agreement with true values based on the image analysis method, with maximum relative deviations of 2.0%, 2.2%, and 1.0% for the 1.16, 1.61 and 2.85 mm glass beads, respectively. Therefore, it was verified that the image analysis method is capable of providing accurate results.  48  3.4 Evolution of solids exchange with time After a long enough time, it is expected that each compartment should approach 50% tracer particles, corresponding to a completely random mixture. This was indeed the case for the 1.61 mm and 2.85 mm particles, as shown in Figure 3-3. However, in the case of the 1.16 mm glass beads, the net transferred tracer mass fraction approached a maximum value of 42%, indicating that 8% of the tracer particles were in a dead zone for these smaller glass beads, likely due to the DSRSB base of 25.4 mm width being large enough to retain those 1.16 mm particles. Therefore, the base needs to be modified in the future, to reduce this blind zone.  0 100 200 300 400 500 600 700 80001020304050  Concentration of tracer (wt.%)Time (s) dp=2.85 mm dp=1.61 mm dp=1.16 mm Figure 3-3 Tracer concentration in compartment 2 for HB,0 = 150 mm, U = 1.2Ums, ∆h = 20 mm, Hsp = 300 mm. Figure 3-4 plots the Lacey mixing index against time for the three sizes of glass beads studied. The mixing index shows the same trend for all cases. The maximum Lacey mixing 49  indexes for the 1.61 and 2.85 mm glass beads approached unity, whereas for the 1.16 mm glass beads, the maximum was ~0.95. The elapsed times to reach the maximum mixing index were ~540, 420 and 180 s for the 1.16, 1.61 and 2.85 mm glass beads, respectively. The differences are due to different superficial velocities. A higher superficial gas velocity results in faster particle motion, which promotes particles mixing. Note that U = 1.2Ums corresponds to U = 0.41 m/s, 0.67 m/s, and 1.2 m/s for the 1.16, 1.61 and 2.85 mm glass beads, respectively, at room temperature and a static bed height of 150 mm.  0 100 200 300 400 500 600 700 8000.00.10.20.30.40.50.60.70.80.91.01.1t'3t'2t'1  I L(-)Time (s) dp = 2.85 mm  dp = 1.61 mm  dp = 1.16 mm  Figure 3-4 Lacey mixing index for glass beads of three sizes with HB,0 = 150 mm, U = 1.2Ums, Hsp = 300 mm and ∆h = 20 mm; t' is the time required to reach equilibrium.   50  3.5 Effect of suspended partition position  20 40 600.00.20.40.60.81.0 C2,t- dp = 1.61 mm  C2,t- dp = 1.16 mm se- dp = 1.61 mm se- dp = 1.16 mm  se(-)h (mm)05101520253035404550C2,t (%) Figure 3-5 Effect of partition position on solids exchange coefficient and tracer concentration for 1.16 and 1.61 mm glass beads with HB,0 = 150 mm, U = 1.2Ums, t = 30s, Hsp = 300 mm. In principle, the DSRSB does not require a partition (Chen et al., 2013b), but a partition stabilizes the flows, reduces intermingling of the fountains and prevents side-to-side percolation of gas from a spout into the corresponding annulus (Rovero et al., 2012). Figure 3-5 shows that the partition position had a significant influence on the solids exchange between the compartments. As the gap (∆h) between the bottom of the partition and the top surface of particles increased from 20 mm to 60 mm, the magnitude of the solids exchange coefficient and the concentration of tracer particles increased by a factor of 2.2. The gap is the main area available for solids exchange between the compartments. Fountain spreading provides the principal mechanism for interchange of particles between the compartments 51  (Chen et al., 2013b). When the partition was raised, the taller gap allowed more particles to pass in both directions, facilitating solids exchange.  3.6 Effect of superficial gas velocity   se- dp = 1.61 mm se- dp = 1.16 mm 1.1 1.2 1.3 1.40.00.20.40.60.81.0 C2,t- dp = 1.61 mm  C2,t- dp = 1.16 mm  se(-)U/Ums 05101520253035404550C2,t (%) Figure 3-6 Effect of superficial gas velocity on solids exchange coefficient and tracer concentration for 1.61 and 1.16 mm glass beads with HB,0 = 150 mm, ∆h = 20 mm, t = 30 s, Hsp = 300 mm. Figure 3-6 shows the effect of superficial gas velocity on the solids exchange between the compartments. The superficial gas velocity is seen to have had a secondary effect on the solids exchange. As the superficial gas velocity increased, the solids exchange coefficient increased and more solids exchanged between the compartments. This occurred because a higher superficial gas velocity resulted in faster particle motion, not only in the axial direction, but also in the lateral direction, which is helpful for particle mixing (Huang and Hu, 52  2007). Moreover, the spout diameter increased as the gas flow increased (McNab, 1972), strengthening the spout intensity, and promoting solids circulation and flow into the fountain. 3.7 Effect of static bed height  150 160 170 180 190 200 2100.00.20.40.60.81.0 se- dp = 1.61 mm  se- dp = 1.16 mm  se(-)HB,0 (mm)05101520253035404550 C2,t- dp = 1.61 mm  C2,t- dp = 1.61 mm C2,t (%) Figure 3-7 Effect of static bed height on solids exchange coefficient and tracer concentration for 1.61 and 1.16 mm glass beads with U = 1.2Ums, ∆h = 20 mm, t = 30 s, Hsp = 300 mm. Figure 3-7 presents the effect of static bed height on the exchange of solids between the compartments. The exchange coefficient increased as the static bed height increased at a given U = 1.2Ums. Slower axial mixing is expected in a deeper bed than in a shallower one (Devahastin and Mujumdar, 2001; Huang and Hu, 2007). With increasing static bed height, the gap (∆h) shown in Figure 3-1 between the top of the static bed and the bottom of the partition, available for solids exchange, decreases, and the fountain surface coverage declines. The reduced gap retards solids exchange between the compartments. Moreover, the reduction 53  in fountain particles reaching the wall retards particles mixing. However, the gap between the top of the divergent base and the surface of the static bed increased, promoting solids exchange between the compartments. Therefore, solids exchange in the lateral direction increased slightly as the static bed height increased for the cases considered here. 3.8 Effect of partition height 0 100 200 300 400 500 600 700 8000.00.20.40.60.81.0   Hsp = 300 mm Hsp = 100 mmI L (-)Time (s) Figure 3-8 Evolution of Lacey mixing index in the DSRSB with 100 mm and 300 mm partition heights for 1.61 mm glass beads, HB,0 = 150 mm, ∆h = 20 mm, U = 1.2Ums. The suspended partition prevents the fountains from merging and improves the stability of the DSRSB (Chen, 2008). However, it also impedes solids exchange between the compartments. Figure 3-8 shows the effect of the partition height on solids exchange between the compartments. Both compartments of the DSRSB contained stable overdeveloped fountains, i.e. the outermost returning particles bounced off the column wall. Less time was required to reach equilibrium when a shorter partition was present. The average fountain 54  height for the 1.61 mm glass beads with HB,0 = 150 mm, U = 1.2Ums, and ∆h = 20 mm was 175-184 mm for the partition heights of 100 and 300 mm. Since the fountain height exceeded the sum of the 20 mm gap (∆h) and the 100 mm height of the shorter partition, solids exchange occurred not only through the gap (∆h), but also over the top of partition in the case of the shorter partition.  0 100 200 300 400 5000.00.10.20.30.40.50.60.70.80.91.0   Hsp=300 mm Hsp=100 mm Hsp=300 mm Fitting line Hsp=100 mm Fitting lineKse (-)Time (s) Figure 3-9 Solids exchange coefficient for 100 mm and 300 mm partition heights for 1.61 mm glass beads, HB,0 = 150 mm, ∆h = 20 mm, U = 1.2Ums. Figure 3-9 demonstrates that a higher solids exchange coefficient was obtained with the 100 mm partition height than with the 300 mm partition height. The dependence of the solids exchange coefficient on time was the same for the short and long partitions. The magnitude of the solids exchange coefficient was less than unity, indicating that most of the energy of the air was used to accelerate particles in the vertical direction, rather than to facilitate solids exchange in the lateral direction. In addition, the solids exchange coefficient decreased 55  slightly as time elapsed because the difference between the tracer concentrations in the two compartments was becoming smaller with time, and particles which had traveled from one side to the other were then included in those transferring in the opposite direction. For a DSRSB running a continuous process with feeding on one side and discharge from the opposite side, it is seen to be possible to promote solids mixing between compartments by reducing the height of the suspended partition. 3.9 Conclusions (1) Comparison of glass beads of diameter 2.85, 1.61 and 1.16 mm shows that exchange of larger particles between compartments reached an equilibrium composition in less time in the dual-compartment slot-rectangular spouted bed than for smaller particles, in each case operating with a superficial gas velocity equal to 1.2 times the minimum spouting velocity.  (2) The effect of the partition position on solids exchange between the compartments was more important than other factors, with a larger gap leading to faster exchange.  (3) A shorter partition accelerated solids exchange between the compartments by allowing some fountain particles to pass over the partition.  (4) The static bed height and superficial gas velocity had only a slight influence on solids exchange for the range of conditions investigated.   56  Chapter 4 Biomass torrefaction in a single-compartment slot-rectangular spouted bed reactor In the present chapter, biomass was initially torrefied in a TGA, providing fundamental insights into its thermal behaviour at different operating conditions. Then, biomass was torrefied in a slot-rectangular spouted bed reactor of 1000 × 150 × 100 mm (height × width × thickness) under pure nitrogen and oxygen-containing atmospheres. Variables involved torrefaction temperature (T), biomass feed rate (F), biomass particle size (dp) and oxygen concentration (XO2) in the feed-gas. Their effects on reactor hydrodynamics, torrefaction efficiency and torrefied biomass properties are investigated. See Section 2.3 of Chapter 2 for the operating procedure. In this chapter, a mixture of spruce, pine and fir (SPF) sawdust was used as experimental material. Key properties of the sawdust are given in Table 2-3. 4.1 Effects of temperature and biomass feed rate In this section, 0.5-1.0 mm SPF sawdust and pure nitrogen were employed to investigate effects of torrefaction temperature and biomass feed rate on biomass torrefaction.  4.1.1 Pretests in TGA The SPF sawdust was pre-tested in the TGA reactor to explore its thermal behaviour under a nitrogen atmosphere. Figure 4-1 presents the dynamic weight loss and devolatilization rate curves of SPF sawdust after completion of drying at a heating rate of 1°C/min. The devolatilization rate is derived from the weight loss of biomass per unit time. It shows that the weight loss of SPF sawdust started at 172°C, and continued to 800°C. An inflection point 57  occurred at 338°C on the weight loss curve. A peak was observed at 321°C in the devolatilization rate curve. The peak point likely corresponds to the maximum decomposition rate of cellulose, while the inflection point is thought to indicate completion of hemicellulose decomposition (Chen and Kuo, 2011). 100 200 300 400 500 600 700 800 9000.00.20.40.60.81.0321C Weight loss fraction (-)Temperature (oC)172C338C0100200300Devolatilization rate (g/s.g) (x106) Figure 4-1 Dynamic weight loss and devolatilization curves for SPF sawdust particles of 0.5-1.0 mm for a TGA heating rate of 1°C/min with nitrogen of 99.999% purity. Figure 4-2 shows the weight loss curves of SPF sawdust at different temperatures obtained from isothermal experiments of duration 8 h in the TGA. It is found that temperature played a crucial role in the biomass weight loss.  Increasing the torrefaction temperature led to a greater weight loss.  SPF sawdust lost 23%, 54%, and 70% of its initial dried weight when they maintained at 240, 270 and 300°C for 8 h, respectively. The devolatilization rate was greater at a higher temperature.  58  0 60 120 180 240 300 360 420 4800102030405060708090100 240oC 270oC 300oC  Residual Weight Fraction (wt.%)Time (min) Figure 4-2 0.5-1.0 mm SPF sawdust torrefaction in TGA at different temperatures with nitrogen of 99.999% purity at 50°C/min heating rate. 4.1.2 Operating conditions The operating conditions for 0.5-1.0 mm sawdust torrefaction in the SRSB facility are listed in Table 4-1. More than half of the cases have been repeated in the present study. The screw-feeder was calibrated with the SPF sawdust tested at room temperature before the torrefaction experiments, with the actual feed rate calculated as the weight of fed biomass divided by the duration of feeding. The experiments were planned for four temperatures (240, 270, 300 and 330°C), three biomass feed rates (220, 440 and 710 g/h) and a single gas superficial gas velocity of U = 1.2Ums, but the actual operating conditions differed somewhat from these pre-set conditions. The average actual biomass feed rate was ~220, 440 or 710 g/h.  The actual temperature of the reactor was quite close to the planned value. Prior to each torrefaction experiment, the minimum spouting velocity (Ums) of nitrogen for the inert 59  particles was measured at the corresponding elevated temperatures, i.e. 240, 270, 300 and 330°C, respectively. The minimum spouting velocity (Ums) of nitrogen for the inert particles was found to be 0.30, 0.31, 0.32 and 0.33 m/s at 240, 270, 300 and 330°C, respectively. During the torrefaction process, the superficial gas velocity of nitrogen was controlled accurately at 1.2Ums. Each experiment ran for 50 min, except for case C-11 which was extended to 110 min.  Table 4-1 Operating conditions for torrefaction experiments of 0.5-1.0 mm SPF sawdust. For each case, U = 1.2Ums. Case Nominal operating conditions (T, F, t) Actual operating conditions T (°C) F (g/h) t (min) C-1 240°C, 220 g/h, 50 min 253 210 50 C-2 270°C, 220 g/h, 50 min 278±3 220±11 50 C-3 300°C, 220 g/h, 50 min 298±2 233±10 50 C-4 240°C, 440 g/h, 50 min 241±3 413±12 50 C-5 270°C, 440 g/h, 50 min 269±4 420±20 50 C-6 300°C, 440 g/h, 50 min 295±3 467±8 50 C-7 330°C, 440 g/h, 50 min 333 479 50 C-8 240°C, 710 g/h, 50 min 255 632 50 C-9 270°C, 710 g/h, 50 min 276±2 704±10 50 C-10 300°C, 710 g/h, 50 min 296 803 50 C-11 270°C, 440 g/h, 110 min 274 477 110          ±value: Value of standard deviation.    60  4.1.3 Typical case Figure 4-3 shows the temperature evolution for a typical case of torrefaction (C-6) conducted at 295°C with 467 g/h sawdust feed rate and 0.5-1.0 mm diameter sawdust. The windbox temperature (T1) and temperatures at Z = 64 mm (T2), 165 mm (T3) and 267 mm (T4) were measured at 1 s intervals. Because the initial static bed height for 3000 g of glass beads (HB,0) was 176 mm below the location of T4, the reactor temperature (T) was taken as the average of T2 and T3. As shown in Figure 4-3, the windbox temperature (T1) as expected was the highest temperature in the SRSB reactor, and it maintained a nearly constant value. The T2 and T3 thermocouples were immersed 20 mm from inside wall of the reactor, in the annulus of the SRSB (Saidi et al., 2015b), where the particles descended, resulting in decreasing in temperature with decreasing height as heat was transferred between the fluid and particles (Kmiec and Englart, 2011). Therefore, the temperature at T3 was higher than at T2. However, it was observed through the glass windows that particle movement was retarded due to more and more biomass particles packing into the vicinity of T2. The SPF sawdust feeding began at time 0 min. Initially, temperatures at T2 and T3 slightly decreased because of the biomass at room temperature being fed into the reactor. Subsequently, these two temperatures leveled off as the torrefaction continued. Affected by the sawdust feed, the temperature at T4 initially decreased, but then increased slowly during the rest of the torrefaction process, due to the gradual increase of the bed height as torrefied product accumulated in the reactor compartment. 61  0 10 20 30 40 50150200250300350 Windbox (T1) Z=64 mm (T2) Z=165 mm (T3) Z=267 mm (T4)  Average (T)Average T Z = 267 mm (T4)Z = 64 mm (T2)  T (oC)Time (min)Windbox (T1) Z = 165 mm (T3) Figure 4-3 Time variation of temperatures for SRSB reactor, case C-6 of sawdust with 0.5-1.0 mm diameter, T = 295°C, and F = 467 g/h.  Figure 4-4 shows the time variation of the pressure drop across the reactor and the superficial gas velocity for case C-6. The superficial gas velocity of nitrogen was maintained quite stable at ~0.4 m/s (1.2Ums). The reactor pressure drop (ΔP) was measured between just above the slot and the reactor top as shown in Figure 2-5 (a). For the first few minutes, the reactor pressure drop was very stable, because of the limited quantity of SPF sawdust fed into the reactor. However, the reactor pressure drop then started to oscillate with an amplitude of ~600 Pa. The irregular shape of the sawdust particles likely contributed to the oscillation of the pressure drop.  62  0 10 20 30 40 50 60-0.10.00.10.20.30.40.50.6 U (m/s)Time (min)10001200140016001800200022002400P (Pa) Figure 4-4 Time variation of reactor pressure drop and superficial gas velocity for case C-6 with 0.5-1.0 mm diameter sawdust particles, T = 295°C and F = 467 g/h. 4.1.4 Solid product yield In each experiment, mass conservation was tested for the entire system. It is found that the mass of torrefied fine sawdust captured by the filter (Mf) was much less than 1% of the total fed sawdust (Mt). The rest of the torrefied sawdust was either captured by the cyclone (Mc), or stayed in the reactor (Mr).  Table 4-2 indicates the SPF sawdust weight loss defined by  %100)1(Xtfrc MMMM                   (4-1) The weight loss increased with increasing temperature at a given biomass feed rate, consistent with the results of Figure 4-2.  A relatively high weight loss was obtained with a lower feed rate, leading to a longer residence time for biomass particles. True yield (Y) is a 63  ratio of mass of torrefied sawdust captured by the cyclone (Mc) to total mass of torrefied sawdust, defined as  %100Yrcc MMM                   (4-2) The true yield, Y, increased as the temperature increased at a given biomass feed rate, indicating that more torrefied sawdust would be captured by the cyclone at higher torrefaction temperature.  For a given temperature, Y increased as the biomass feed rate increased, suggesting that higher biomass feed rate led to more torrefied sawdust being captured by the cyclone. Comparison of runs C-5 and C-11 reveals that the weight loss of sawdust (X) slightly increased, while the mass of torrefied sawdust remaining in the reactor (Mr) increased from 134 to 254 g when the operating time was extended from 50 min to 110 min. Note that the entire system did not reach steady state within 50 min.  64  Table 4-2 Experimental solid torrefied product yield and weight loss in experiments. Case:  Nominal operating conditions* Mt (g) Mc (g) Mr (g) Mf (g) Y (%) X (%) C-1: 0.5-1.0 mm+240°C+220g/h 174.9 45.8 83.4 2.1 35.4 24.9 C-2: 0.5-1.0 mm+270°C+220g/h 182.9±4.6 54.1±1.4 65.6±1.6 2.0±0.2 45.2±1.1 33.5±0.8 C-3: 0.5-1.0 mm+300°C+220g/h 194.2±4.2 65.5±1.6 50.5±1.5 2.1±0.1 56.5±1.4 39.2±0.8 C-4: 0.5-1.0 mm+240°C+440g/h 344.0±5.0 130.0±3.0 163.9±4.9 1.0±0.1 44.2±1.9 14.3±0.3 C-5: 0.5-1.0 mm+270°C+440g/h 349.9±8.3 138.7±3.3 133.8±3.3 1.50±0.2 50.9±2.5 21.6±0.6 C-6: 0.5-1.0 mm+300°C+440g/h 389.1±3.3 150.0±3.8 136.8±3.4 1.5±0.2 52.5±2.1 25.7±0.7 C-7: 0.5-1.0 mm+330°C+440g/h 399.0 145.6 133.9 2.0 57.7 29.4 C-8: 0.5-1.0 mm+240°C+710g/h 527.1 273.2 216.5 1.9 55.8 6.7 C-9: 0.5-1.0 mm+270°C+710g/h 586.7±4.2 316.7±7.5 191.4±2.9 2.0±0.1 62.3±1.2 13.1±0.4 C-10: 0.5-1.0 mm+300°C+710g/h 668.8 333.7 163.1 2.1 67.2 25.4 C-11: 0.5-1.0 mm+270°C+440g/h+110min 873.1 384.6 253.7 2.0 60.3 26.7           *See Table 4-1 for actual operating conditions. ±value: Value of standard deviation. 65  SRSB Torrefaction Reactor(T: 240-330oC)CycloneFilterVolatiles and gases= + + +Screw-Feeder(Feed rate: 210-803 g/h)Mass (%)100Mass (%)26-48Mass (%)26-34Mass (%)<1Mass (%)26-39Feed rate (g/h)~ 220100~ 440100~ 71035-4824-4138-4050-52<1<114-267-26  Figure 4-5 Mass distribution map of raw and torrefied sawdust torrefied at different biomass feed rates and temperatures in SRSB facility. Figure 4-5 shows the mass distribution of torrefied product in the SRSB facility. At the operating conditions of 240-330°C, 210-803 g/h biomass feed rate with U = 1.2Ums nitrogen flow, 7-39% of the mass of biomass was lost in the volatiles, while the cyclone caught 26-52% of the total mass of biomass, 24-48% of the biomass remained in the reactor, and the filter captured less than 1% of the total mass of biomass.  It is observed that increasing the biomass feed rate led to a greater mass percentage of torrefied sawdust captured by the cyclone, but a smaller mass percentage of torrefied sawdust remaining in the reactor. A more important finding is that a lower weight loss of sawdust occurred at a higher feed rate.  66  4.1.5 Properties of torrefied product Table 4-3 Particle density, HHV and energy yield of torrefied SPF sawdust on a dry basis. Case: Nominal operating conditions* Torrefied product (Cyclone) Torrefied product (Reactor) Energy yield (%) ρp,c (kg/m3) HHVc (MJ/kg) ρp,r (kg/m3) HHVr (MJ/kg) C-1: 0.5-1.0 mm+240°C+220g/h 1323±68 19.68±0.02 1309±17 19.90±0.01 79.2±0.1 C-2: 0.5-1.0 mm+270°C+220g/h 1393±20 20.19±0.08 1331±6 20.46±0.15 71.8±0.4 C-3: 0.5-1.0 mm+300°C+220g/h 1345±9 20.50±0.12 1376±3 21.57±0.47 67.4±0.5 C-4: 0.5-1.0 mm+240°C+440g/h 1356±13 20.62±0.32 1336±10 21.00±0.04 97.6±0.6 C-5: 0.5-1.0 mm+270°C+440g/h 1398±53 20.20±0.08 1345±6 21.16±0.09 88.3±0.3 C-6: 0.5-1.0 mm+300°C+440g/h 1317±31 20.34±0.07 1305±17 21.95±0.10 85.6±0.2 C-7: 0.5-1.0 mm+330°C+440g/h 1367±30 20.78±0.06 1366±16 22.80±0.06 81.3±0.2 C-8: 0.5-1.0 mm+240°C+710g/h 1402±10 19.66±0.06 1363±29 20.31±0.01 98.7±0.2 C-9: 0.5-1.0 mm+270°C+710g/h 1429±50 20.04±0.05 1356±17 20.38±0.20 94.7±0.7 C-10: 0.5-1.0 mm+300°C+710g/h 1395±11 19.99±0.08 1304±10 21.45±0.29 82.9±0.6 C-11: 0.5-1.0 mm+270°C+440g/h            +110min 1415±40 20.04±0.17 1345±4 21.84±0.01 82.8±0.7 *See Table 4-1 for actual operating conditions. ± value: Value of standard deviation.  Table 4-3 shows particle densities of the torrefied sawdust collected by the cyclone and remaining in the reactor on a dry basis. The torrefied SPF sawdust consistently had a higher particle density than the untreated sawdust of 1254 kg/m3, due to the shrinkage of particles during torrefaction. The higher heating value (HHV) and the energy yield on a dry basis are also presented in Table 4-3. Torrefaction significantly increased the sawdust HHV because of the removal of low-carbon-content volatiles and moisture. However, the effects of 67  temperature and sawdust feed rate on the HHV of the torrefied sawdust collected by the cyclone were small. Comparison of the higher heating values of torrefied sawdust collected by the cyclone (HHVc) and particles remaining in the reactor (HHVr) reveals that HHVr was always higher than HHVc. This is associated with the operating procedure whereby no biomass would be extracted from the SRSB reactor until the reactor was cooled from the torrefaction temperature to less than 50°C. Furthermore, as shown in Table 4-3, the energy yield defined by  %100HHV MHHV M HHV M                   %100biomass rawin Energy product edin torrefiEnergy yieldEnergy 0trrccdrydrydry                (4-3) clearly decreased with increasing temperature. The fraction of torrefied biomass energy after treatment varied from 67.4 to 98.7%, related to temperature and biomass feed rate.  The loss of energy in volatiles increased with increasing temperature and with decreasing biomass feed rate.  One of the important advantages of torrefaction is that the energy density of the torrefied product is greater than that the untreated biomass, mainly due to the removal of oxygen from the raw biomass. Table 4-4 summarizes the composition of the raw and torrefied sawdust on a dry basis. Case C-9 was selected to test the repeatability of the elemental analysis. The results (shown in Table 4-4) indicate that the elemental analyzer was able to provide repeatable results. As shown in Table 4-4, the proportion of atomic oxygen and hydrogen were both lost during the torrefaction, so that the atomic carbon content increased. The losses of oxygen and hydrogen are likely due to decomposition of hydroxyl (-OH) and carboxyl (-68  COOH) during the torrefaction process (Wang et al., 2016a; Yang et al., 2007). An increase in temperature led to increased carbon content, but decreased hydrogen and oxygen content at a given biomass feed rate. However, the elemental compositions of torrefied sawdust were only slightly affected by the biomass feed rate. Based on fiber analysis, torrefied sawdust contained less hemicellulose compared to the raw sawdust. The proportion of hemicellulose dropped significantly with increasing temperature for a given biomass feed rate, because hemicellulose is the most readily decomposed constituent of the biomass (Chen and Kuo, 2011). Meanwhile, a decrease in the biomass feed rate resulted in less hemicellulose in the torrefied sawdust, due to a longer residence time for a smaller feed rate. Moreover, the proportion of cellulose first increased and then decreased as the temperature increased from 240 to 330°C at a 440 g/h biomass feed rate. This was likely due to the cellulose starting to decompose at ~300°C (Yang et al., 2007). For a 710 g/h biomass feed rate, the cellulose kept increasing with increasing temperature from 240 to 300°C, related to the shorter residence time at the 710 g/h feed rate.  In addition, the lignin content gradually increased as temperature increased from 240 to 300°C, due to decomposition of hemicellulose and cellulose. The lignin fraction of torrefied sawdust was only slightly affected by the biomass feed rate.  69  Table 4-4 Elemental and fiber analyses of raw SPF sawdust and torrefied SPF sawdust collected by cyclone on a dry basis. Case: Nominal operating conditions* Elemental analysis (wt.%) Fiber analysis (wt.%) C  H N  O** Hemicellulose Cellulose Lignin Extractive C-0: Raw SPF sawdust 46.2 6.4 < 0.1 47.4 14.6 49.0 27.6 8.8 C-1: 0.5-1.0 mm+240°C+220g/h 50.6 6.3 < 0.1 43.1 11.8 51.6 29.7 6.9 C-2: 0.5-1.0 mm+270°C+220g/h 51.5 6.2 < 0.1 42.3 9.6 52.6 31.3 6.5 C-3:0.5-1.0 mm+300°C+220g/h 52.0 6.1 < 0.1 41.9 6.4 51.9 36.5 5.2 C-4: 0.5-1.0 mm+240°C+440g/h 50.9 6.2 < 0.1 42.9 14.6 47.0 28.7 9.7 C-5: 0.5-1.0 mm+270°C+440g/h 51.2 6.2 < 0.1 42.6 14.1 46.5 30.1 9.3 C-6: 0.5-1.0 mm+300°C+440g/h 52.2 6.1 < 0.1 41.7 11.5 48.7 32.0 7.8 C-7: 0.5-1.0 mm+330°C+440g/h 53.3 6.1 < 0.1 40.6 3.9 46.7 46.6 2.8 C-8: 0.5-1.0 mm+240°C+710g/h 50.8 6.4 < 0.1 42.8 13.0 51.9 30.6 4.5 C-9: 0.5-1.0 mm+270°C+710g/h 51.4 ±0.1 6.3 ±0.1 < 0.1 42.3 ±0.1 9.1 55.5 30.8 4.6 C-10: 0.5-1.0 mm+300°C+710g/h 51.7 6.3 < 0.1 42.0 8.8 54.3 32.4 4.5 C-11: 0.5-1.0 mm+270°C+440g/h+110min 51.5 6.3 < 0.1 47.2 12.0 53.8 33.0 1.2    *See Table 4-1 for actual operating conditions.    ** Oxygen was determined by difference. 70  Table 4-5 shows the volatile matter (VM), fixed carbon (FC) and ash contents (AC) of the raw SPF and torrefied sawdust. It is seen that the volatile content of the torrefied sawdust was significantly less than that of the raw SPF sawdust, presumably due to hemicellulose decomposition and cellulose depolymerization during the torrefaction process (Prins et al., 2006a; Yang et al., 2007). The formation of volatiles is mainly influenced by temperature and reaction time (Prins et al., 2006b). It is also found that the volatile content decreased with increasing temperature.  Nevertheless, the HHV of the torrefied solids increased with increasing temperature as shown in Table 4-3. This means that the escaping volatiles contained less energy per unit mass than the solids left behind, especially as the temperature increased. Comparison of the torrefied products from the cyclone and reactor in Table 4-5 shows that the volatile content of torrefied product collected by the cyclone (VMc) was greater than that of the torrefied product remaining in the reactor (VMr), while the fixed carbon content (FCc) was less than FCr. As mentioned above, the explanation is that the torrefied sawdust remaining in the reactor had a longer residence time than the torrefied sawdust captured by the cyclone.  71  Table 4-5 Proximate analysis of torrefied SPF sawdust on a dry basis.  *See Table 4-1 for actual operating conditions.          VM: Volatile matter; FC: Fixed carbon; AC: Ash content.  ±value: Value of standard deviation. N/A: Not available.  Case:  Nominal operating conditions* Torrefied product (Cyclone) Torrefied product (Reactor) VMc (wt.%) FCc (wt.%) ACc (wt.%) VMr (wt.%) FCr (wt.%) ACr (wt.%) C-1: 0.5-1.0 mm+240°C+220g/h 82.67±0.83 16.99±0.77 0.34±0.03 80.74±0.40 18.81±0.41 0.45±0.01 C-2: 0.5-1.0 mm+270°C +220g/h 80.89±0.86 18.69±0.87 0.42±0.01 77.29±0.48 22.22±0.49 0.48±0.01 C-3: 0.5-1.0 mm+300°C +220g/h 78.31±1.64 21.24±1.63 0.45±0.01 69.82±0.26 29.54±0.28 0.64±0.02 C-4: 0.5-1.0 mm+240°C +440g/h 81.44±0.48 18.25±0.47 0.43±0.01 77.50±0.20 22.19±0.29 0.51±0.02 C-5: 0.5-1.0 mm+270°C +440g/h 78.75±0.79 20.78±0.75 0.47±0.04 72.52±0.42 26.91±0.39 0.57±0.02 C-6: 0.5-1.0 mm+300°C +440g/h 78.55±0.01 20.93±0.03 0.52±0.02 70.14±0.10 29.28±0.12 0.58±0.02 C-7: 0.5-1.0 mm+330°C +440g/h 73.41±0.51 26.04±0.28 0.55±0.02 N/A N/A N/A C-8: 0.5-1.0 mm+240°C +710g/h 81.08±0.51 18.54±0.4 0.38±0.02 80.00±0.16 19.55±0.28 0.46±0.12 C-9: 0.5-1.0 mm+270°C +710g/h 80.27±1.18 19.32±1.1 0.41±0.03 78.66±0.46 20.96±0.56 0.38±0.10 C-10: 0.5-1.0 mm+300°C +710g/h 78.81±0.93 20.73±1.0 0.46±0.04 72.58±0.83 26.78±0.97 0.64±0.14 C-11: 0.5-1.0 mm +270°C +440g/h           +110min 81.48±0.16 18.03±0.1 0.49±0.03 N/A N/A N/A 72  ×100 ×300 ×1000 Raw SPF sawdust    C-4: 240°C+440g/h    C-5: 270°C+440g/h    C-6: 300°C+440g/h    Figure 4-6 SEM images of untreated and cyclone-caught torrefied SPF sawdust. See Table 4-1 for detailed operating conditions. 73  Figure 4-6 shows SEM images of raw and cyclone-caught torrefied sawdust. Extractives deposited within the sieve tubes of the raw biomass can be clearly seen with 1000 magnification. It is apparent that these extractives were decomposed during the torrefaction process. After undergoing torrefaction, sieve tubes could be clearly seen. As the torrefaction temperature increased from 240 to 300°C, the cell wall was observed to become thinner, and the particle surface was smoother. 4.1.6 Particle size distribution of torrefied product Particle size distribution was determined by a sieve shaker with sieves of 125, 180, 250, 355, 425, 500, 720 and 1000 μm. The sieving time was 10 min for each test. Torrefaction had a significant impact on the particle size distribution of torrefied biomass, as shown in Figure 4-7. The particle size reduction is linked to the improved grindability of torrefied biomass (Chen and Kuo, 2011; Wang et al., 2017a). Meanwhile, spouted beds provide effective and intense particle-particle and particle-gas contact (Fernández-Akarregui et al., 2012; Haddou et al., 2013). The raw sawdust particle size was in the range of 0.5-1.0 mm. A significant increase in the number of smaller (< 0.5 mm) particles was observed for all cases, with these smaller particles approaching 36-56 wt.% of the torrefied sawdust collected by the cyclone.  It was found that the particle size distribution curve skewed towards smaller particle sizes after torrefaction. It was also seen that the particle size distribution curve tended towards smaller particles as the torrefaction temperature increased. The increase in the number of smaller (< 0.5 mm) particles was especially significant for the torrefied biomass treated at 330°C (case C-7, in Table 4-1), with a mass increase of 56%.  The reduction in particle size 74  was likely caused by partial destruction of the fibrous structure during torrefaction (Chen and Kuo, 2011).  -100 0 100 200 300 400 500 600 700 800 900 1000020406080100  Cumulative fraction undersize (%)dp (m) C-0: Raw SPF C-4: 0.5-1.0mm+240oC+440g/h C-5: 0.5-1.0mm+270oC+440g/h C-6: 0.5-1.0mm+300oC+440g/h C-7: 0.5-1.0mm+330oC+440g/h Figure 4-7 Particle size distribution of cyclone-caught sawdust torrefied at T = 240-330°C with F = 440 g/h for 50 min. See Table 4-1 for operating conditions. 4.2 Effects of temperature and particle size Within spouted beds, the particles undergo vigorous movement, improving their contact with the gas. Particle size plays an important role with respect to solids mixing (Ren et al., 2012; Wang et al., 2016b), minimum spouting velocity (Olazar et al., 2006) and residence time distribution, even affecting the discharge method of the torrefied biomass. In this section, three different sizes SPF sawdust were first torrefied in the TGA reactor to explore their thermal behaviour. The particles of different sizes were then torrefied in the SRSB facility with a nominal biomass feed rate of 440 g/h. The experimental variables include temperature within the range of 240-300°C, and particle size of 0.25-0.5, 0.5-1.0 and 1.0-2.0 mm. All 75  experiments were conducted at the same multiple of minimum spouting gas velocity, i.e. 1.2Ums. Prior to each torrefaction experiment, the minimum spouting velocity (Ums) of nitrogen for the inert particles was measured at the corresponding elevated temperatures, i.e. 240, 270 and 300°C, respectively. The minimum spouting velocity (Ums) of nitrogen for the inert particles was found to be ~0.32, 0.31 and 0.3 m/s at 240, 270 and 300°C, respectively.  Before beginning, the column was filled with 3000 g of glass beads of 1 mm diameter. Each experiment was terminated after 50 min.  4.2.1 Pretest in TGA 0 60 120 180 240 300 360 420 4800.00.10.20.30.40.50.60.70.80.91.0300oC270oC  0.25-0.5mm  0.5-1.0 mm   1.0-2.0mm    Residual Weight Fraction (wt.%)Time (min)240oC Figure 4-8 Weight loss curves of SPF sawdust of 0.25-0.5, 0.5-1.0 and 1.0-2.0 mm diameter at different temperatures with a heating rate of 50°C/min.  In pre-tests in the TGA, the smallest (0.25-0.5 mm), intermediate (0.5-1.0 mm) and coarsest (1.0-2.0 mm) sawdust particles were each torrefied for 8 h at 240, 270 and 300°C, respectively, with a heating rate of 50°C/min.  Weight loss curves of these three sizes of 76  particle at different temperatures are plotted in Figure 4-8. It is found that smaller particles resulted in lower residual weight fraction at the same torrefaction temperature, as a consequence of better mass and heat transfer within the particles (Peng et al., 2012b). It is also seen that temperature played a crucial role in the torrefaction process. After 8 h, the residual weight fraction of the 0.25-0.5, 0.5-1.0 and 1.0-2.0 mm diameter sawdust decreased from 73%, 77%, and 78% to 28%, 29%, and 31% respectively, as the temperature increased from 240 to 300°C. The weight loss curve of 0.5-1.0 mm particles was much closer to the curve of 0.25-0.5 mm sawdust at 300°C temperature. These results revealed that the torrefaction was affected by the biomass particle size.  4.2.2 Operating conditions The operating conditions for the study of the effects of temperature and particle size on torrefaction are summarized in Table 4-6. Cases PS-1 to PS-6 and PS-8 were repeated. For all experiments, the temporal variation of the average temperature (T) was less than 6°C. The sawdust feed rate (F) variation was less than 35 g/h in the repeated experiments. For each experiment, the minimum spouting velocity (Ums) of nitrogen for inert particles was measured at the corresponding elevated temperature prior to the torrefaction experiment. The minimum spouting velocity of nitrogen for the inert particles was found to be ~ 0.32 m/s over the 240-300°C range.    77  Table 4-6 Operating conditions for torrefaction experiments with smallest, intermediate and coarsest sawdust. For each case, U = 1.2Ums and operating time = 50 min. Case Nominal operating conditions (particle size range, T, F) Actual operating conditions T (°C) F (g/h) PS-1 0.25-0.5 mm, 240°C, 440 g/h 237±4 452±10 PS-2 0.25-0.5 mm, 270°C, 440 g/h 268±2 472±8 PS-3 0.25-0.5 mm, 300°C, 440 g/h 297±2 545±15 PS-4 0.5-1.0 mm, 240°C, 440 g/h 249±3 434±14 PS-5 0.5-1.0 mm, 270°C, 440 g/h 276±2 429±10 PS-6 0.5-1.0 mm, 300°C, 440 g/h 298±3 469±20 PS-7 1.0-2.0 mm, 240°C, 440 g/h 244 489 PS-8 1.0-2.0 mm, 270°C, 440 g/h 281 514 PS-9 1.0-2.0 mm, 300°C, 440 g/h 292±2 507±25 ±value: Value of standard deviation. 4.2.3 Effect of particle size on hydrodynamics of SRSB torrefaction reactor 4.2.3.1 Pressure drop across SRSB reactor The bed pressure drop reflects the degree of solids mixing in the spouted beds (Du et al., 2015). Figure 4-9 shows the bed pressure drop vs. time for the SRSB reactor during torrefaction experiments for the smallest, intermediate and coarsest sawdust.  For the 0.25-0.5 mm sawdust (PS-2), the pressure drop across the reactor oscillated within the range of 1100-1600 Pa, while the mean pressure drop decreased with time. As sawdust was continuously fed into the reactor, the proportion of the smallest particles increased. The fine particles and their lower density likely led to decreased pressure drop across the spouted bed (Du et al., 2015). For the intermediate particles (PS-5), the pressure drop varied between 78  1100 and 1850 Pa. The mean pressure drop initially increased, but slightly decreased later, because some torrefied sawdust particles were transported out of the reactor, so that the pressure drop across the reactor could not remain constant. The bed pressure drop variation of the largest sawdust particles differed significantly from the other two particle sizes. As shown in Figure 4-9, the pressure drop for case PS-8 underwent a sudden decrease during the torrefaction process. It was observed through the quartz windows that the inert and sawdust particles were stuck at the bottom of the reactor, due to the needle-like shape of the coarsest particles. In addition, the pressure drop fluctuated much less than for the smallest and intermediate particles, with an amplitude of ~300 Pa.  -5 0 5 10 15 20 25 30 35 40 45 50 551000125015001750200012501500175020001250150017502000-5 0 5 10 15 20 25 30 35 40 45 50 55 P (Pa)Time (min)PS-2:0.25-0.5mm P (Pa)PS-5:0.5-1.0mm  P (Pa)PS-8:1.0-2.0mm Figure 4-9 Time variation of pressure drop across SRSB reactor. See Table 4-6 for operating conditions. 79  4.2.3.2 Final bed depth within SRSB reactor Figure 4-10 plots the bed depth at the end of the torrefaction experiments for the three different particle sizes. Compared to the initial bed depth of 176 mm (HB,0), the final bed depth (HB) increased by 5.1-9.6%, 23.9-25.6%, and 60-67.6% for the smallest, intermediate and coarsest sawdust, respectively. The largest particles led to the deepest final bed depth, as a consequence of less entrainment. Comparison of the same size sawdust at different torrefaction temperatures reveals that the final bed height was lower at a higher temperature. For instance, the final bed depth of 0.5-1.0 mm sawdust decreased from 295 to 282 mm as the torrefaction temperature increased from 240 to 300°C, due to the higher weight loss at the higher temperature and greater entrainment as the particles lost more weight during the torrefaction process, as shown in Figure 4-8.  PS-1 PS-2 PS-3 PS-4 PS-5 PS-6 PS-7 PS-8 PS-90501001502002503003501.0-2.0 mm0.5-1.0 mm  Bed depth, HB (mm)CaseHB,0 = 176 mm0.25-0.5 mm Figure 4-10 Final bed depth after 50 min torrefaction. See Table 4-6 for operating conditions. 80  4.2.4 Solid product yield Table 4-7 Experimental solid torrefied product yield and weight loss. See Table 4-6 for operating conditions. Case:  Nominal operating conditions Mt (g) Mc (g) Mr (g) Mf (g) Y (%) X (%) PS-1:0.25-0.5mm+240°C 376.8±4.2 222.4±4.7 120±3.0 2.0±0.1 65.0±1.6 8.6±0.2 PS-2:0.25-0.5mm+270°C 393.0±3.3 283.9±5.7 68.4±1.7 2.9±0.1 80.6±2.0 9.6±0.2 PS-3:0.25-0.5mm+300°C 453.9±6.3 346.5±6.1 44.6±1.1 3.0±0.2 88.6±2.2 13.2±0.3 PS-4:0.5-1.0mm+240°C 361.5±5.8 152.1±3.8 156.4±2.7 1.5±0.2 49.3±1.2 14.2±0.3 PS-5:0.5-1.0mm +270°C 357.4±4.2 148.3±3.0 124.4±3.1 1.0±0.1 54.4±1.4 23.4±0.4 PS-6:0.5-1.0mm +300°C 390.6±8.3 157.2±3.1 123.6±2.8 1.0±0.3 56.0±1.4 27.8±0.6 PS-7:1.0-2.0mm +240°C 407.5 38.1 280.3 1.0 12.0 21.6 PS-8:1.0-2.0mm +270°C 428.0 57.1 245.2 0.8 18.9 29.2 PS-9:1.0-2.0mm+300°C 422.3±10.4 50.6±1.3 208.3±4.7 1.0±0.1 19.5±0.5 38.5±1.0 ±value: Value of standard deviation. In each experiment, the total mass of raw sawdust fed into the reactor (Mt), the mass of torrefied sawdust captured by the cyclone (Mc), the mass of torrefied sawdust left in the reactor (Mr), and mass of torrefied sawdust captured by the filter (Mf) were determined. The mass balance was then checked for the entire system. For a given size of sawdust, the weight loss increased with increasing temperature, as shown in Table 4-7. For example, the weight loss of 0.5-1.0 mm sawdust increased from 14.2 to 27.8% when the temperature increased from 240 to 300°C, consistent with Figure 4-8 and previous work (Chen et al., 2011a; Li et al., 2012b). This confirms that temperature plays an important role in the torrefaction process (Peng et al., 2012b). Comparison of the different sizes of sawdust indicates that the finest (0.25-0.5 mm) sawdust had the lowest weight loss for the same operating conditions. For 81  example, at 270°C and a 440 g/h feed rate, the weight losses after 50 min were 9.6, 23.4 and 29.2% for the smallest, intermediate and coarsest sawdust, respectively, because the smaller particles were more readily entrained and hence had shorter residence times within the reactor.  SRSB Torrefaction Reactor(T: 240-300oC)CycloneFilterVolatiles and gases= + + +Screw-Feeder(Feed rate: ~440 g/h)Size range (mm) Mass (%)0.25-0.5 1000.5-1.0 1001.0-2.0 100Mass (%)10-3232-4349-69Mass (%)59-7740-429-13Mass (%)<1<1<1Mass (%)8-1314-2822-39 Figure 4-11 Effect of particle size on mass distribution of raw and torrefied sawdust in SRSB facility. Figure 4-11 provides a mass distribution map of sawdust in the SRSB torrefaction facility. At the operating conditions of ~440 g/h sawdust feed rate and 240-300°C temperature; (i) For the 0.25-0.5 mm sawdust, 10-32 wt.% stayed in the SRSB reactor, 59-77 wt.% was captured by the cyclone, but less than 1 wt.% was captured by the filter, resulting in 8-13 wt.% weight 82  loss; (ii) For the 0.5-1.0 mm sawdust, 32-43 wt.% remained in the SRSB reactor, 40-42 wt.% was captured by the cyclone, less than 1 wt.% was captured by the filter, and the weight loss was ~ 14-28 wt.%; (iii) For the 1.0-2.0 mm sawdust, 49-69 wt.% remained in the SRSB reactor, 9-13 wt.% was captured by the cyclone, less than 1 wt.% was captured by the filter, but the weight loss was 22-39 wt.%. Because most of the 1.0-2.0 mm sawdust remained in the reactor, most of the torrefied biomass particles remained mixed with the inert particles, which would require therefore further downstream processing to separate the torrefied sawdust from the inert particles.  4.2.5 Evolution of ash content in torrefied product During the torrefaction experiments, ~2 g samples of the torrefied sawdust were taken just below the cyclone bottom at 5-10 min intervals (See Figure 2-5 for solids sample discharge location). Because the mass of torrefied sawdust captured by the cyclone (Mc) was very limited for the 1.0-2.0 mm sawdust (See Table 4-7), the ash content for the 1.0-2.0 mm torrefied sawdust could not be determined accurately. Therefore, Figure 4-12 plots the evolution of ash content in the torrefied product captured by the cyclone for the 0.25-0.5 and 0.5-1.0 mm sawdust. It is observed that the ash content first increased and then levelled off with time. The increase in the ash content of the torrefied sawdust mainly resulted from decomposition of extractives, hemicellulose and partial cellulose in the torrefaction process. The levelling off of the ash content of the torrefied sawdust indicated that the torrefied sawdust properties approached a steady state. In Figure 4-12 (a), the ash content of torrefied sawdust reached the maximum at ~20 min for all cases. In addition, the curves show that the ash content was only slightly affected by temperature in the 240-300°C range investigated, because of the brief residence time for the smallest particles. Figure 4-12 (b) indicates that 83  temperature played a more important role in the ash content evolution for the intermediate particles, with the ash content of the torrefied sawdust increasing with increasing temperature (Li et al., 2012a; Peng et al., 2012b). The greater influence was because the intermediate particles had longer residence times within the reactor than the smallest particles. The ash content of the intermediate torrefied sawdust reached a steady state at ~15 min, a little earlier than for the smallest sawdust.  0 10 20 30 40 500.250.300.350.400.450.500.550.60  Ash Content (%)Time (min) PS-1: 240oC PS-2: 270oC PS-3: 300oC(a) 0.25-0.5 mm 84  0 10 20 30 40 500.250.300.350.400.450.500.550.60(b) 0.5-1.0 mm  Ash Content (%)Time (min) PS-4: 240oC PS-5: 270oC PS-6: 300oC Figure 4-12 Evolution of ash content of product captured by cyclone for (a) 0.25-0.5 mm particles; (b) 0.5-1.0 mm particles. 4.2.6 Properties of torrefied product Table 4-8 shows the proximate, ultimate and fiber analyses of the torrefied product and raw SPF sawdust on a dry basis. The carbon content of the torrefied sawdust was much greater than that of the raw sawdust, whereas the oxygen and hydrogen contents of the torrefied sawdust were reduced after torrefaction. Oxygen is removed from the biomass, due to hydroxyl (-OH) and carboxyl (-COOH) groups decomposition in the torrefaction process (Wang et al., 2016a; Yang et al., 2007). The carbon content of the torrefied sawdust increased slightly with increasing temperature, as a consequence of the reduction of oxygen and hydrogen. For example, for the 0.25-0.5 mm sawdust, the carbon content of the torrefied sawdust increased from 49.9 to 50.4%, while the oxygen content decreased from 43.9 to 43.3%, as the temperature increased from 240 to 300°C.  Comparison of the smallest, 85  intermediate and coarsest sawdust reveals that the carbon content of the finest sawdust was lowest among the three sizes of particles at the same temperature, associated with the smaller particles having shorter residence times in the SRSB reactor.  In addition, as shown in Table 4-8, the volatile content of the torrefied solids was less than that of the raw SPF sawdust of 84.55 wt.%, due to hemicellulose decomposition and cellulose depolymerization during the torrefaction process (Yang et al., 2007).  The volatile content decreased somewhat with increasing temperature in the 240-300°C range. However, the magnitude of volatile content variation was quite limited, only varying by 1.1, 2.0 and 4.2% for the 0.25-0.5, 0.5-1.0 and 1.0-2.0 mm sawdust, respectively. Comparison of the torrefied products from the cyclone and reactor revealed that the volatile content of torrefied product collected by the cyclone (VMc) was greater than for torrefied product remaining in the reactor (VMr), while the fixed carbon content (FCc) was less than FCr. The reason is again that the torrefied sawdust remaining in the reactor had a longer residence time than for the torrefied sawdust captured by the cyclone.  Table 4-8 also presents results of fiber analysis on torrefied sawdust with different sizes. An increase in sawdust size resulted in decreasing hemicellulose of torrefied sawdust for a given temperature. Meanwhile, the hemicellulose proportion of torrefied sawdust decreased with increasing temperature for a given sawdust. In addition, the cellulose proportion of torrefied sawdust increased with increasing temperature for the 0.25-0.5 and 0.5-1.0 mm fractions. However, an increase in temperature led to decreased cellulose in the torrefied sawdust for the 1.0-2.0 mm sawdust. Furthermore, lignin of torrefied sawdust gradually increased with increasing sawdust particle size and temperature. 86  Table 4-8 Proximate, ultimate and fiber analyses of torrefied sawdust on a dry basis. Case: Nominal operating            conditions* PS-1 0.25-0.5mm 240°C PS-2 0.25-0.5mm 270°C PS-3 0.25-0.5mm 300°C PS-4 0.5-1.0mm 240°C PS-5 0.5-1.0mm 270°C PS-6 0.5-1.0mm 300°C PS-7 1.0-2.0mm 240°C PS-8 1.0-2.0mm 270°C PS-9 1.0-2.0mm 300°C  Torrefied product (Cyclone) Elemental analysis           C (wt.%) 49.9 50.1 50.4 50.6 51.1 51.2 51.1 51.4 52.2   H (wt.%) 6.3 6.3 6.3 6.3 6.2 6.1 6.2 6.0 5.9   N (wt.%) <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1   O# (wt.%) 43.8 43.6 43.3 43.1 42.7 42.7 42.7 42.6 41.9 Proximate analysis           VMc (wt.%) 83.59±0.28 82.64±0.18 82.49±0.71 83.76±0.16 82.58±0.56 81.80±0.65 82.67±0.63 79.1±0.14 78.40±0.20   FCc (wt.%) 15.9±0.28 16.84±0.16 16.98±0.70 15.81±0.14 16.95±0.50 17.68±0.65 16.64±0.61 20.05±0.11 20.55±0.13   ACc (wt.%) 0.51±0.00 0.52±0.02 0.53±0.01 0.43±0.02 0.47±0.06 0.52±0.00 0.69±0.02 0.85±0.03 1.05±0.07 Fiber analysis           Hemicellulose (wt.%) 15.5 15.0 13.2 14.5 14.2 11.3 10.3 10.0 9.5   Cellulose (wt.%) 45.1 46.4 48.7 47.3 46.8 49.6 52.3 51.1 49.8   Lignin (wt.%) 28.6 29.0 30.5 28.5 29.6 31.1 29.5 32.7 33.9   Extractive (wt.%) 10.8 9.6 7.6 9.7 9.4 8.0 7.9 6.2 6.8                                                Torrefied product (Reactor) Proximate analysis           VMr (wt.%) 81.49±1.10 75.46±0.90 72.32±1.12 78.82±0.27 73.53±0.40 70.38±0.13 72.07±1.60 71.37±1.42 65.24±1.24   FCr (wt.%) 17.99±1.04 24.0±0.88 27.13±1.11 20.69±0.26 25.9±0.44 29.09±0.19 27.53±1.59 28.12±1.41 34.13±1.19   ACr (wt.%) 0.52±0.06 0.54±0.02 0.55±0.01 0.49±0.02 0.56±0.04 0.54±0.05 0.4±0.01 0.51±0.01 0.63±0.03 * See Table 4-6 for actual operating conditions. # Oxygen was determined by difference.     ±value: Value of standard deviation. 87  Table 4-9 shows particle densities of raw and torrefied sawdust on a dry basis. The torrefied SPF sawdust consistently had a higher particle density than the untreated sawdust, due to the shrinkage of particles during torrefaction and/or reduction in internal pores. The higher heating values (HHV) and energy yields on a dry basis are also presented in Table 4-9. Torrefaction significantly increased the sawdust HHV because of the removal of low-carbon-content volatiles and moisture. As shown in Table 4-9, the energy yield, defined by Equation (4-3), clearly decreased with increasing temperature. The energy yield varied from 72.1 to 99.8%, influenced by the particle size and temperature. Because the 1.0-2.0 mm particles were much heavier than the 0.25-0.5 and 0.5-1.0 mm particles, the mean residence time of the coarsest particles was much longer than for the other two particle sizes at the same operating conditions. Therefore, the energy yield of the coarsest particles was much lower than for the smallest and intermediate sawdust. Meanwhile, a drawback was that the mass of torrefied sawdust captured by the cyclone for the 1.0-2.0 mm particles was much less than for the smallest and intermediate size particles. Recovery of the coarsest torrefied sawdust remaining in the reactor would require additional processing to separate the torrefied sawdust from the mixture of the inert particles and the torrefied sawdust. In addition, temperature had limited influence on the energy yield of the finest particles. As shown in Table 4-7, the weight loss of the smallest particles varied from 8.6 to 13.2 wt.%, as the temperature increased from 240 to 300°C. Most of energy was retained in the torrefied sawdust, with the energy yield of torrefied sawdust varying from 95.3 to 99.8% for the smallest sawdust.    88  Table 4-9 HHVs, particle densities and energy yield of torrefied sawdust on a dry basis.  Case: Nominal  operating conditions* Torrefied product (Cyclone) Torrefied product (Reactor) Energy yield (%) ρp,c (kg/m3) HHVc (MJ/kg) ρp,r (kg/m3) HHVr (MJ/kg) PS-1: 0.25-0.5mm+240°C 1429±61 19.91±0.04 1370±10 20.55±0.05 99.8±1.0 PS-2: 0.25-0.5mm+270°C 1384±27 19.92±0.01 1365±16 21.57±0.04 99.5±0.1 PS-3: 0.25-0.5mm+300°C 1379±10 19.94±0.08 1366±12 22.09±0.15 95.3±0.4 PS-4: 0.5-1.0mm+240°C 1367±21 19.95±0.01 1366±6 20.30±0.02 94.2±0.1 PS-5: 0.5-1.0mm +270°C 1336±24 20.02±0.02 1316±1 21.27±0.04 86.2±0.1 PS-6: 0.5-1.0mm +300°C 1339±18 20.28±0.09 1304±21 21.89±0.15 82.7±0.1 PS-7: 1.0-2.0mm +240°C 1460±20 20.24±0.02 1382±1 19.98±0.28 85.8±1.0 PS-8: 1.0-2.0mm +270°C 1452±38 20.72±0.11 1313±22 21.54±0.05 82.9±0.2 PS-9: 1.0-2.0mm+300°C 1400±14 20.74±0.02 1265±42 21.61±0.27 72.1±0.7 * See Table 4-6 for actual operating conditions. ±value: value of standard deviation. 4.2.7 Particle size reduction  The sawdust particles experienced size reduction during torrefaction. Figure 4-13 shows the percentage reduction of Sauter mean diameters (dsv) of sawdust in the torrefaction process, where the percentage reduction is defined by  %100SV0,,0,svisvsvddd                   (3-4) Here dsv,0 is the Sauter mean diameter of raw sawdust, dsv,i is the Sauter mean diameter of sawdust after screw feeding and then torrefying sawdust at 240, 270 or 300°C.   89  051015202530  SV (%) Finest particles (0.25-0.5 mm) Intermediate particles (0.5-1.0 mm)SF 240oC 270oC 300oC Figure 4-13 Percentage reduction of Sauter mean particle size of SF and torrefied sawdust of 240, 270 and 300°C. Note that the SPF particles were obtained by sieving for 10 min in the vibrated sieving device with 0.25, 0.5, 1.0 and 2.0 mm sieves, whereas the Sauter mean particle diameter was determined by the Malvern Mastersizer 2000. The Sauter mean diameters may well be influenced by irregular, sometimes needle-like shapes of the SPF sawdust. The particle size was reduced in two stages. In the first stage, the sawdust particles were mechanically conveyed by the screw-feeder before entering the reactor.  For the smallest sawdust, dsv decreased from 263 to 252 μm, whereas for the intermediate sawdust, dsv decreased from 433 to 413 μm as a result of the screw-feeder. The second stage was when biomass particles were torrefied in the reactor. During the torrefaction process, sawdust suffered attrition due to contact with inert particles, biomass particles, and the inside reactor wall.  For the finest sawdust particles, dsv was reduced from 252 μm to 237, 223 and 210 μm for temperatures of 240, 270 and 300°C, respectively.  For the intermediate size sawdust particles, dsv decreased 90  from 413 μm to 364, 345 and 310 μm at 240, 270 and 300°C, respectively.  For the entire process, as shown in Figure 4-13, the size reduction was 4.3, 10.2, 15.2, and 20.2% for the smallest sawdust, and 6.8, 15.8, 20.2 and 28.2% for the intermediate size sawdust after being conveyed by the screw-feeder and torrefied at 240, 270 and 300°C, respectively. The particle size reduction is related to the improved grindability of torrefied product (Phanphanich and Mani, 2011), and intense attrition in spouted beds (Fernández-Akarregui et al., 2012; Haddou et al., 2013). Comparison of the two stages reveals that most of the particle size reduction for our sawdust particles happened during the torrefaction process, i.e. in the spouted bed reactor, rather than in the screw-feeder. 4.3 Oxidative torrefaction In this section, the effects of oxygen concentration in the feed gas on the torrefied product properties and the torrefaction efficiency in the single-compartment slot-rectangular spouted bed facility are assessed. Results of the oxidative torrefaction are compared with the results of non-oxidative torrefaction, where the oxygen content in the feed-gas varied from 3 to 9 vol.%. The sawdust was intermediate SPF particles (0.5-1.0 mm). The nominal biomass feed rate was set at 440 g/h. The superficial gas velocity of nitrogen was 1.2Ums. Each experiment was conducted for 50 min.   4.3.1 Operating conditions for oxidative torrefaction Table 4-10 lists the operating conditions for the oxidative torrefaction experiments. Runs OT-2, OT-4 to OT-6 and OT-9 were repeated. The carrier gas was a mixture of nitrogen and air for the torrefaction experiments. The oxygen concentration in the carrier gas was measured by the flue gas analyzer (PS-200, HORIBA), prior to each torrefaction experiment. 91  During the torrefaction experiment, the oxygen concentration was measured at 1 s intervals. As shown in Table 4-10, the time-averaged oxygen concentrations were similar to the desired values. It is noteworthy that the temperature of the reactor experienced an abrupt increase when the 9 vol.% oxygen in the feed-gas was introduced into the reactor at a pre-set 300°C temperature. However, no temperature runaway was observed. (See Appendix B: Temperature profiles) Table 4-10 Operating conditions for oxidative torrefaction experiments.  Case Nominal operating conditions          (T, F, XO2)  Actual operating conditions T (°C) F (g/h) XO2 (vol.%) U/Ums t (min) OT-1 240°C, 440 g/h, 3vol.% O2 240 474 3.7 1.2 50 OT-2 240°C, 440 g/h, 6vol.% O2 251±3 495±15 6.2±0.1 1.2 50 OT-3 240°C, 440 g/h, 9vol.% O2 243 463 9.1 1.2 50 OT-4 270°C, 440 g/h, 3vol.% O2 275±4 477±14 3.4±0.1 1.2 50 OT-5 270°C, 440 g/h, 6vol.% O2 272±3 407±12 6.1±0.1 1.2 50 OT-6 270°C, 440 g/h, 9vol.% O2 271±3 495±14 8.8±0.2 1.2 50 OT-7 300°C, 440 g/h, 3vol.% O2 307 460 3.0 1.2 50 OT-8 300°C, 440 g/h, 6vol.% O2 307 445 6.0 1.2 50 OT-9 300°C, 440 g/h, 9vol.% O2 318±8 469±15 9.1±0.1 1.2 50      ±value: Value of standard deviation.  92  4.3.2 Typical case 0 10 20 30 40 50200220240260280300320340 Windbox (T1) Z=64mm (T2) Z=165mm (T3) Z=267mm (T4) Average (T)Average T Z = 267 mm (T4)Z = 64 mm (T2)  T (oC)Time (min)Windbox (T1) Z = 165 mm (T3) Figure 4-14 Temperature profiles for case OT-5, T = 267°C, with XO2 = 6 vol.% oxygen in the feed-gas and F = 407 g/h sawdust feed rate.  The SRSB reactor temperature was measured by four thermocouples in the windbox (T1), Z = 64 mm (T2), Z = 165 mm (T3) and Z = 267 mm (T4) above the slot. T4 was above the initial static bed height of 176 mm, and in the freeboard region. Therefore, the average reactor temperature was taken as the average of T2 and T3.  Figure 4-14 plots the temporal variation of temperature in the SRSB reactor. As expected, the highest temperature in the SRSB reactor was the windbox temperature, which was 280°C with a standard deviation of 1.2°C. This windbox temperature was 13°C higher than the average reactor temperature of 267°C. The SRSB reactor maintained a stable temperature over the entire torrefaction experiment. An evident decline was observed in temperature T2, when raw biomass was fed at room temperature into the reactor, and those biomass particles then moved downward from the 93  feed port 180 mm above the slot. After a long trajectory, the sawdust particles arrived at the T3 location. This explains why temperature T3 was so little affected by the biomass feeding. -10 0 10 20 30 40 50 600.00.10.20.30.40.5  U (m/s)Time (min)1200140016001800200022002400P (Pa) Figure 4-15 Profiles of reactor pressure drop and superficial gas velocity for case OT-5, T = 267°C, with XO2 = 6 vol.% oxygen in the feed-gas and F = 407 g/h sawdust feed rate. The carrier gas was a mixture of nitrogen and air, and its minimum spouting velocity for the inert particles was ~0.33 m/s at 270°C. Figure 4-15 shows the time-variations of the superficial gas velocity and the pressure drop across the reactor. It was found that the superficial gas velocity of the carrier gas was 0.41±0.02 m/s and ~1.2Ums at 267°C. The pressure drop across the SRSB reactor gradually increased during the first 30 min, and then kept relatively constant at ~1500 Pa for the rest of the experiment. The oscillation intensity of pressure drop increased with time, due to more irregular-shaped biomass being fed into the SRSB reactor. This oscillating pressure drop indicated that the biomass particles and inert particles were mixing well in the torrefaction experiment (Du et al., 2015). 94  -5 0 5 10 15 20 25 30 35 40 45 50 554.04.55.05.56.06.57.07.58.0  XO2 (vol.%)Time (min) Figure 4-16 Oxygen concentration time-variation for case OT-5, T = 267°C, with XO2 = 6 vol.% oxygen in the feed-gas and F = 407 g/h sawdust feed rate. Figure 4-16 shows the oxygen concentration time-variation in the feed-gas during the torrefaction experiment. The oxygen concentration was measured by the flue gas analyzer at 1 s intervals. In order to ascertain the oxygen content in the feed-gas, its concentration was measured for 5 min prior to beginning the torrefaction experiments. As shown Figure 4-16, the oxygen concentration was 6.1 vol.%, with a standard deviation of only 0.1. This reveals that the oxygen content of the feed-gas was closely controlled over the entire torrefaction experiment.   95  4.3.3 Solid product yield Table 4-11 shows the weight loss of biomass and the mass yield in the reactor, cyclone and filter during the oxidative torrefaction experiments. The weight loss of biomass is defined in Equation (4-1). It is seen that the weight loss of biomass in the oxygen-containing atmosphere was greater than in an inert atmosphere, due to biomass oxidation occurring simultaneously (Wang et al., 2013a), and that the weight loss increased with increasing oxygen concentration at a given temperature. However, this increase in weight loss was limited to 240 and 270°C rather than 300°C. As mentioned above, the temperature of case OT-9 experienced an abrupt increase at 300°C and 9 vol.% O2, possibly due to an exothermic oxidation reaction of biomass. This suggested that 9 vol.% oxygen in the feed-gas may be very close to the lower limit of oxygen concentration needed for biomass ignition in the SRSB reactor. For safety reasons, higher oxygen content in the feed-gas was not tested.  In addition, the true ratio (Y), a ratio of mass of torrefied sawdust captured by the cyclone (Mc) to total mass of torrefied sawdust, defined by Equation (4-2) increased with increasing oxygen concentration, illustrating that more torrefied biomass was transported out the reactor as more oxygen was included in the feed-gas.   96  Table 4-11 Torrefied product yield and weight loss in oxidative torrefaction experiments.  Case: Nominal operating conditions* Mt (g) Mc (g) Mr (g) Mf (g) Y (%) X (%) OT-1: 0.5-1.0mm+240°C+3vol.%O2 394.8 173.1 159.8 1.0 52.0 15.4 OT-2: 0.5-1.0mm+240°C+6vol.%O2 412.8±6.3 187.1±3.7 159.6±4.0 1.5±0.1 54.0±1.4 15.6±0.4 OT-3: 0.5-1.0mm+240°C+9vol.%O2 385.8 190.3 133.3 1.0 58.8 15.9 OT-4: 0.5-1.0mm+270°C+3vol.%O2 397.5±5.8 168.6±4.2 128.8±2.3 1.5±0.2 56.7±1.4 24.8±0.5 OT-5: 0.5-1.0mm+270°C+6vol.%O2 339.4±5.0 183.9±3.7 68.6±1.7 1.8±0.1 72.8±1.8 25.1±0.5 OT-6: 0.5-1.0mm+270°C+9vol.%O2 412.8±5.8 235.0±4.7 61.3±1.4 1.6±0.2 79.3±2.0 27.8±0.6 OT-7: 0.5-1.0mm+300°C+3vol.%O2 383.1 178.0 81.3 1.6 68.6 31.9 OT-8: 0.5-1.0mm+300°C+6vol.%O2 370.4 181.0 57.4 1.6 75.9 35.2 OT-9: 0.5-1.0mm+300°C+9vol.%O2 398.5±10.4 177.8±4.4 34.1±0.8 1.6±0.1 83.9±2.1 46.4±1.2 * See Table 4-10 for actual operating conditions. Mt: Total mass of sawdust fed; Mc: Mass of torrefied sawdust captured by cyclone;              Mr: Mass of torrefied sawdust remaining in reactor; Mf: Mass of torrefied sawdust captured by filter. ±value: Value of standard deviation. 97  SRSB Torrefaction Reactor(T: 240-300°C)(N2 (91-100vol.% )+O2 (0-9vol.%)) CycloneFilterVolatiles and gases= + + +Screw-Feeder(Feed rate: 430-500 g/h)100Mass (%)32-43Mass (%)40-42Mass (%)<1Mass (%)14-28Mass (%)0O2 conc (vol.%)1003-9 9-40 42-57 <1 15-46 Figure 4-17 Mass distribution of raw and torrefied sawdust for oxidative torrefaction in SRSB facility. Figure 4-17 summarizes the mass distribution of torrefied biomass in the SRSB facility. At 240-300°C and 3-9 vol.% O2 in the feed-gas, the biomass was fed at ~450 g/h into the SRSB facility, producing 63-86 wt.% solid product, consisting of 9-48 wt.% in the reactor, 42-57 wt.% in the cyclone and less than 1 wt.% in the filter. The biomass would lose 15-46 wt.% during the oxidative torrefaction experiment in the SRSB facility. In summary, supplying oxygen in the feed-gas increased the weight loss of biomass, decreased the mass of torrefied sawdust remaining in the reactor, and increased the mass of torrefied sawdust captured by the cyclone.  98  4.3.4 Evolution of ash content in torrefied product 0 10 20 30 40 500.200.250.300.350.400.450.500.550.600.650.70  Ash Content (%)Time (min) PS-4: 240oC+0% O2 OT-1: 240oC+3% O2 OT-2: 240oC+6% O2 OT-3: 240oC+9% O2(a) 0 10 20 30 40 500.200.250.300.350.400.450.500.550.600.650.70  Ash Content (%)Time (min) PS-5: 270oC+0% O2 OT-4: 270oC+3% O2 OT-5: 270oC+6% O2 OT-6: 270oC+9% O2(b) 0 10 20 30 40 500.200.250.300.350.400.450.500.550.600.650.70(c) PS-6: 300oC+0% O2 OT-7: 300oC+3% O2 OT-8: 300oC+6% O2 OT-9: 300oC+9% O2  Ash Content (%)Time (min)  Figure 4-18 Time variation of ash content of torrefied sawdust captured by cyclone at (a) T = 240°C, XO2 = 0-9 vol.%; (b) T = 270°C, XO2 = 0-9 vol.%; (c) T = 300°C, XO2 = 0-9 vol.%. When the sawdust was torrefied in the SRSB reactor, some torrefied particles were pneumatically entrained from the reactor and captured by the cyclone. In the torrefaction experiment, ~2 g torrefied sawdust were sampled at 5-10 min intervals from the solid discharge port (See Figure 2-5 for the port location). The time variation of the ash content of torrefied product is plotted in Figure 4-18. It is apparent that oxygen affected the ash content 99  of torrefied product in the torrefaction process. The ash content of the torrefied sawdust was significantly higher for the torrefaction with oxygen present than for inert torrefaction. At a given temperature, as shown in Figures 4-18(a)-(c), increasing the oxygen content in the feed-gas from 3 to 9 vol.% led to significantly increased ash content of the torrefied product, because decomposition and oxidation reaction of the biomass occurred simultaneously.  For a given oxygen content, the ash content of torrefied sawdust increased with increasing temperature. This is associated with the formation of volatiles during the biomass decomposition process being mainly influenced by temperature (Prins et al., 2006b). It was observed for all cases that the ash content first increased and then levelled off with time. This increase in the ash content of torrefied sawdust was mainly due to biomass oxidation reaction and decomposition during the torrefaction process. The levelling off of the ash content of the torrefied sawdust reveals that the torrefied product properties approached a steady state. In Figure 4-18(a), at 240°C torrefaction temperature, the torrefied product more quickly approached a steady state in the presence of oxygen than for the oxygen-free case. To reach steady state in the torrefied product properties, the required time was ~15 min for 3-9 vol.% O2 and 240°C, while a time of 20-25 min was required for the 240°C and 0 vol.% O2 case, showing that the torrefied sawdust properties more quickly approached constant values in the oxygen-present environment. The same was also seen for T = 300°C.  As shown in Figures 4-18(b) and (c), 9 vol.% O2 had a major influence on the ash content of torrefied sawdust. The ash content of torrefied product increased sharply for the first 5 min, much faster than for 3 and 6 vol.% O2. This suggests that the oxygen concentration in the feed-gas and torrefaction temperature may have synergistic effects on biomass torrefaction.   100  4.3.5 Final bed depth within SRSB reactor Compared to the torrefaction experiments in the absence of oxygen, the oxidative torrefaction experiments led to lower final spouted bed heights, as shown in Figure 4-19. The sawdust lost more weight and its properties changed more quickly in the presence of oxygen during the torrefaction. Furthermore, when the oxygen content increased from 3 to 9 vol.%, the final bed height kept decreasing, as a result of greater weight loss of biomass and more entrainment. It was also found that the final bed height was lowest at 300°C, compared to 240 and 270°C, at otherwise similar operating conditions.  050100150200250240o C+9%O 2  240o C+6%O 2240o C+3%O 2  Bed depth, HB(mm)HB,0 = 176 mm240o C+0%O 2270o C+0%O 2270o C+3%O 2270o C+6%O 2270o C+9%O 2300o C+9%O 2300o C+6%O 2300o C+3%O 2300o C+0%O 2 Figure 4-19 Final bed depth after 50 min of oxidative torrefaction of 0.5-1.0 mm sawdust.  101  4.3.6 Properties of torrefied product Table 4-12 shows that torrefied product generally had a lower volatile content than the raw sawdust, due to dehydration and devolatilization (Prins et al., 2006a). At a given temperature, the volatile content of the torrefied sawdust consistently decreased, whereas its fixed carbon increased as the O2 concentration increased. This indicates that it is easier for SPF sawdust to liberate small hydrocarbon molecules in the presence of oxygen in the torrefaction experiments, resulting from biomass oxidation and decomposition (Haykırı-Açma, 2003; Munir et al., 2009). The impact of the O2 concentration on the volatile content and fixed carbon was more substantial at a higher temperature. Note that the devolatilization and oxidation processes are enhanced when the temperature increases (Haykırı-Açma, 2003). Compared to the torrefied product captured by the cyclone, the torrefied product remaining in the reactor had a much lower volatile content and a higher fixed carbon content, due to the longer residence time of the torrefied sawdust remaining in the reactor. The elemental carbon content of the torrefied product was always higher than that of the raw sawdust, whereas the elemental oxygen and hydrogen contents were reduced. Torrefaction is a mild pyrolysis process, with more carbon retained in the torrefied product than hydrogen and oxygen, compared to the raw material. As shown in Table 4-12, the influences of non-oxidative and oxidative torrefaction on the carbon content were inconsistent in the 240-300°C temperature range. The carbon content of oxidatively torrefied product was found to be slightly lower than that of non-oxidatively torrefied product after processing at 240 and 270°C. In contrast, the carbon content of oxidatively torrefied product was higher than that of non-oxidatively torrefied product at 300°C. This confirms that the oxidation of hydrocarbons in the biomass also played an important role in decomposing biomass (Chen et al., 2014a; 102  Daood et al., 2010; Munir et al., 2009). It may also illustrate that oxidation was the dominant reaction at 240 and 270°C, while carbonization controlled the torrefaction process at 300°C (Daood et al., 2010). Furthermore, in the oxidative torrefaction cases, the carbon content slightly increased with increasing O2 concentration in the feed-gas.  This may be associated with the synergistic effects of the oxidation reaction and carbonization, with volatiles being released and burned (Haykırı-Açma, 2003). It is noteworthy that the elemental oxygen content in the oxidatively torrefied product was greater than in the non-oxidatively torrefied product at 240 and 270°C. This result is consistent with findings of Chen et al. (2014a), but contrary to the results of Rousset et al. (2012) and Wang et al. (2013a). However, the oxygen content of oxidative torrefied product was less than that of non-oxidative torrefied product when the reaction temperature increased to 300°C. It was also found that varying the oxygen content had a greater influence on the C, H and O contents at higher temperature. The effect of temperature on the C, H and O contents on oxidative torrefaction was the same as on the non-oxidative torrefaction. For a given oxygen content, an increase in temperature led to greater C content, but lower O and H contents. The main constituents of biomass are hemicellulose, cellulose and lignin; therefore, the thermal decomposition characteristics of these three constituents play a strong role in determining the performance of torrefaction of lignocellulosic materials. Table 4-12 shows the influence of varying the O2 concentration on the hemicellulose, cellulose and lignin constituents. Overall, whether or not oxygen was supplied in the feed-gas, the hemicellulose content was less in the torrefied product than in the raw material. It is apparent that the hemicellulose in the sawdust was influenced by oxygen, even at temperatures as low as 240°C. Furthermore, for a given temperature, the hemicellulose content slightly decreased as 103  the O2 concentration increased from 3 to 9 vol.%. This reveals that the O2 concentration variation had a slight impact on the hemicellulose content of the sawdust at 240°C, in agreement with a previous finding (Wang et al., 2013a). At temperatures of 270 and 300°C, the variation of hemicellulose content was limited as the O2 concentration increased from 3 to 6 vol.%. However, as the O2 concentration rose further to 9 vol.%, the hemicellulose content decayed abruptly. In terms of cellulose content in the torrefied sawdust, introducing the oxygen in the feed-gas caused increased the cellulose content for the 240 and 270°C cases, but decreased the cellulose content for 300°C. This is related to the decomposition temperature of cellulose, which is normally in the range of 315-400°C (Yang et al., 2007). At 240 and 270°C, the cellulose content increased with increasing oxygen content, mainly due to hemicellulose decomposition. However, at 300°C, the cellulose content decreased with increasing oxygen content, as a result of cellulose decomposition and oxidation.  Lignin, a phenolic polymer, essentially encases the polysaccharides of the cell walls, producing a strong and durable composite material (Gomez et al., 2008), resisting torrefaction, regardless of the temperature (Chen and Kuo, 2011). The lignin content of the torrefied sawdust gradually increased as the O2 concentration of the feed-gas increased from 3 to 9 vol.%, likely as a consequence of depletion of hemicellulose in the biomass (Chen and Kuo, 2011; Yang et al., 2007).  The extractives are made up of inorganic material (soil or fertilizer), non-structural sugars (chlorophyll, waxes) and nitrogenous material (Sluiter et al., 2008b). It was found that these were oxidized during oxidative torrefaction.   104  Whether the sawdust was torrefied in pure N2 or in an oxygen-present environment, the HHV of the torrefied product was higher than that of the raw sawdust. However, addition of O2 to the feed-gas led to a slight decrease in HHVc for the cyclone-caught torrefied product at otherwise similar operating conditions, presumably because more hemicellulose or/and cellulose were oxidized as the oxygen was added. However, within the 3-9 vol.% range, the HHVc and HHVr of the oxidatively torrefied product were insensitive to the oxygen variation. This may be because cellulose and lignin mainly contributed to the HHVc of torrefied sawdust (Chen and Kuo, 2011), and these two constituents were insensitive to the oxygen concentration variation in the range of 3-9 vol.%. It was also found that the HHVr of the torrefied product remaining in the reactor was higher than the HHVc of the torrefied product captured by the cyclone. This is due to the fact that the sawdust particles remaining in the reactor had longer reaction times and underwent oxidation reaction and decomposition in the oxidative torrefaction. However, an increase in temperature led to greater HHV of torrefied sawdust. At a similar temperature, it was found that varying the O2 concentration within the range of 0-9 vol.% had a pronounced effect on the energy yield. Although the energy yield of the sawdust drastically decreased with increasing O2 concentration, the energy yield of 94-61% was still in a reasonable range for making high-quality pellets (Peng et al., 2013b), leading to high-quality bio-oil or syngas (Sarkar et al., 2014).    105  Table 4-12 Properties of torrefied sawdust and energy yields on a dry basis. Parameter  PS-4 (240°C+0%O2) OT-1 (240°C+3%O2) OT-2 (240°C+6%O2) OT-3 (240°C+9%O2) PS-5 (270°C+0%O2) OT-4 (270°C+3%O2)  Torrefied product from cyclone Proximate analysis   VM (wt.%) 83.76±0.16 83.42±0.05 83.37±0.21 83.00±0.21 82.58±0.56 82.22±1.03   FC (wt.%) 15.81±0.18 16.03±0.07 16.16±0.23 16.51±0.22 16.95±0.62 17.29±1.04   AC (wt.%) 0.43±0.02 0.45±0.02 0.47±0.02 0.49±0.01 0.47±0.06 0.49±0.01 Elemental analysis   C (wt.%) 50.6 49.8 50.3 50.4 51.1 50.4   H (wt.%) 6.3 6.3 6.4 6.2 6.2 6.2   N (wt.%) <0.1 <0.1 <0.1 <0.1 <0.1 <0.1   O* (wt.%) 43.1 43.9 43.3 43.4 42.7 43.4 Fiber analysis   Hemicellulose (wt.%) 14.5 13.7 13.6 13.0 14.2 12.2   Cellulose (wt.%) 47.3 52.3 52.8 53.5 46.8 52.7   Lignin (wt.%) 28.5 29.2 27.9 28.3 29.0 29.2   Extractives (wt.%) 9.7 4.8 5.7 5.2 10.0 5.9 HHVc (MJ/kg) 19.95±0.01 19.86±0.05 19.85±0.07 19.78±0.08 20.03±0.02 19.89±0.12  Torrefied product from reactor Proximate analysis   VM (wt.%) 78.82±0.27 80.01±0.50 79.59±0.01 79.32±0.39 73.53±0.40 76.20±0.64   FC (wt.%) 20.69±0.29 19.63±0.51 20.01±0.03 20.27±0.46 25.90±0.44 23.20±0.68   AC (wt.%) 0.49±0.02 0.37±0.01 0.40±0.03 0.41±0.08 0.56±0.04 0.60±0.04 HHVr (MJ/kg) 20.30±0.02 20.06±0.07 20.08±0.05 19.98±0.06 21.27±0.04 20.10±0.04 Energy yield (%) 94.2±0.1 99.4±0.3 98.6±0.3 96.5±0.1 86.2±0.1 82.0±0.3  106  ±value: value of standard deviation.  *Oxygen was determined by difference. Parameter OT-5 (270°C+6%O2) OT-6 (270°C+9%O2) PS-6 (300°C+0%O2) OT-7 (300°C+3%O2) OT-8 (300°C+6%O2) OT-9 (300°C+9%O2)    Torrefied product from cyclone Proximate analysis   VM (wt.%) 81.78±1.75 80.71±0.22 81.80±0.65 79.79±0.30 78.44±0.51 77.80±0.66   FC (wt.%) 17.81±1.76 18.74±0.23 17.68±0.65 19.67±0.31 21.01±0.53 21.58±0.70   AC (wt.%) 0.51±0.01 0.55±0.01 0.52±0.01 0.54±0.01 0.55±0.02 0.62±0.04 Elemental analysis   C (wt.%) 50.28 50.6 51.17 51.26 52.29 52.15   H (wt.%) 6.18 6.17 6.11 6.14 6.07 6.12   N (wt.%) <0.1 <0.1 <0.1 <0.1 <0.1 <0.1   O* (wt.%) 43.54 43.23 42.72 42.6 41.64 41.73 Fiber analysis   Hemicellulose (wt.%) 12.0 11.1 11.3 9.6 9.4 8.1   Cellulose (wt.%) 52.3 51.6 49.6 47.6 47.0 46.5   Lignin (wt.%) 30.1 32.7 31.1 40.0 40.4 42.1   Extractives (wt.%) 5.6 4.6 8.0 2.8 3.2 3.3 HHVc (MJ/kg) 19.78±0.13 19.85±0.08 20.28±0.09 20.26±0.10 20.71±0.04 20.78±0.04   Torrefied product from reactor Proximate analysis   VM (wt.%) 75±0.07 68.93±1.93 70.38±0.13 48.64±6.7 57.56±0.67 55.50±2.29   FC (wt.%) 24.48±0.07 30.43±2.0 29.09±0.19 50.52±6.71 41.46±0.71 43.56±2.33   AC (wt.%) 0.52±0.01 0.64±0.06 0.54±0.05 0.84±0.01 0.98±0.04 0.94±0.04 HHVr (MJ/kg) 20.15±0.09 21.26±0.08 21.89±0.15 22.89±0.06 21.82±0.12 22.09±0.04 Energy yield (%) 81.1±0.3 79.2±0.2 82.7±0.1 78.3±0.3 74.1±0 61.2±0.1 107  Figure 4-20 shows SEM images of cyclone-caught sawdust oxidatively torrefied at different oxygen concentrations. The extractives were significantly consumed during the torrefaction as the oxygen concentration increased from 3 to 9 vol.% at 270°C. When the oxygen concentration increased from 3 to 9 vol.%, the tubes became more visible, and the surface of the torrefied sawdust particles was cleaner and smoother. Cracks were also observed on the surface of the sawdust particles torrefied with 9 vol.% oxygen. All those observations are associated with more severe torrefaction at greater oxygen concentration.  ×50 ×100 ×300 OT-4: 270°C+440g/h+3vol.% O2    OT-5: 270°C+440g/h+6vol.%O2    OT-6: 270°C+440g/h+9vol.%O2    Figure 4-20 SEM images of cyclone-caught sawdust oxidatively torrefied at T = 270°C and F = 440 g/h biomass feed rate with different oxygen concentrations. 108  4.4 Conclusions • TGA pretests showed that (i) temperature had a significant effect on weight loss of biomass in torrefaction, with higher temperature leading to a higher weight loss of biomass; (ii) biomass particle size also influenced the torrefaction performance, with larger biomass particles experiencing less weight loss at similar operating conditions.   • Higher temperature, lower biomass feed rate, smaller size biomass particle and greater oxygen concentration led to a decreased mass fraction of torrefied sawdust remaining in the SRSB reactor, but to an increased mass fraction of torrefied sawdust captured by the cyclone.  • Based on evolution of ash content in cyclone-caught torrefied sawdust, the cyclone-caught torrefied sawdust properties reached a steady state after 20 min in the torrefaction process.  • Increasing temperature, reducing the biomass feed rate, larger biomass particles, and increasing oxygen in the feed gas all resulted in increased weight loss and decreased energy yield of sawdust, and produced torrefied sawdust with higher HHV, greater atomic carbon content, lower atomic hydrogen and oxygen contents, less volatile content, greater fixed carbon content, and less hemicellulose. Torrefied sawdust had smoother and cleaner surfaces compared to raw sawdust. • The pressure drop across the SRSB reactor fluctuated significantly during biomass torrefaction. Feeding the smallest sawdust particles into the SRSB reactor led to decreased pressure drop, while feeding the coarsest sawdust particles resulted in a lower pressure drop due to inert and sawdust particles sticking at the bottom of the reactor. 109  • Sawdust particles underwent significant particle size reduction during torrefaction, with a reduction in Sauter mean particle size of up to 20.3% and 28.2% for the 0.25-0.5 mm and 0.5-1.0 mm sawdust size fractions, respectively. 110  Chapter 5 Biomass torrefaction in a dual-compartment slot-rectangular spouted bed reactor As a follow-up to Chapter 4, the single-compartment slot-rectangular spouted bed (SRSB) was scaled up to a dual-compartment slot-rectangular spouted bed (DSRSB). A partition was installed within the DSRSB to stabilize the spouting. The DSRSB provides a more uniform residence time distribution for solids, compared to the SRSB. This chapter covers the performance of the dual-compartment slot-rectangular spouted bed torrefaction reactor. Biomass was first torrefied in an oxygen-free environment, and then in an oxygen-present environment. Intermediate-size SPF sawdust particles, i.e. 0.5-1.0 mm, were employed as the experimental material. Its key properties are shown in Table 2-3. When operating the DSRSB reactor, part of the off-gas was recycled to reduce the operating cost. The schematic diagram is shown in Figure 2-9, and the operating procedure is described in section 2.4 of Chapter 2.  5.1 Biomass torrefaction in nitrogen atmosphere 5.1.1 Operating conditions Table 5-1 summarizes the operating conditions for biomass torrefaction in a nitrogen atmosphere. Cases D-C-1 to D-C-9 and D-C-11 were repeated. Temperature and biomass feed rate variations were less than 4% in the repeated experiments. Cases D-C-10 to D-C-12 were longer experiments, each lasting 120 min. Torrefaction temperatures were nominally 240, 270 and 300°C, the same as those for the single bed SRSB facility (Chapter 4).  The biomass feed rate was set at double that for the SRSB facility. These settings ensure that the results of the SRSB and DSRSB torrefaction experiments are comparable.  111  Table 5-1 Operating conditions for biomass torrefaction experiments with nitrogen atmosphere in DSRSB facility. Case Nominal operating conditions Actual operating conditions (T, F, U/Ums, t) T (°C) F (g/h) U/Ums t (min) D-C-1 240°C, 600 g/h, 1.2, 50 min 255±4 668±30 1.2 50 D-C-2 270°C, 600 g/h, 1.2, 50 min 278±4 610±10 1.2 50 D-C-3 300°C, 600 g/h, 1.2, 50 min 304±3 613±12 1.2 50 D-C-4 240°C, 900 g/h, 1.2, 50 min 242±4 930±11 1.2 50 D-C-5 270°C, 900 g/h, 1.2, 50 min 271±3 916±10 1.2 50 D-C-6 300°C, 900 g/h, 1.2, 50 min 301±3 957±10 1.2 50 D-C-7 240°C, 1400 g/h, 1.2, 50 min 244±2 1408±50 1.2 50 D-C-8 270°C, 1400 g/h, 1.2, 50 min 269±2 1334±32 1.2 50 D-C-9 300°C, 1400 g/h, 1.2, 50 min 303±3 1408±12 1.2 50 D-C-10 240°C, 900 g/h, 1.2, 120 min 249 1091 1.2 120 D-C-11 270°C, 900 g/h, 1.2, 120 min 271±2 907±10 1.2 120 D-C-12 300°C, 900 g/h, 1.2, 120 min 255±4 944 1.2 120              ±value: Value of standard deviation.  5.1.2 Case study Figure 5-1 shows the temperature time-variation for case D-C-6. The torrefaction experiment began at time 0 min after pre-heating the column. Thermocouples T′1 and T′2 measured temperatures of the carrier gas in the downstream and upstream windboxes. T′3 and T′5 measured temperatures at Z = 64 and 165 mm above the slot in the downstream compartment, whereas T′4 and T′6 portray the temperatures at Z = 64 and 165 mm above the slot in the upstream compartment. T′7 plots the freeboard temperature on the right side, because its location (Z= 276 mm) was much higher than the initial static bed height of the inert particles 112  (HB,0). The DSRSB reactor temperature was calculated as the average of thermocouples T′3 to T′6. For example, the average temperature for case D-C-6 was 299°C with a 4.9°C standard deviation. 0 10 20 30 40 50260270280290300310320330340350360 T'1(downstream windbox)  T'2(upstream windbox) T'3(Z=64mm, downstream)  T'4(Z=64mm, upstream) T'5(Z=165mm, downstream)  T'6(Z=165mm, upstream) T'7(Z=267mm, upstream)  Average  T(oC)Time (min)T'1 T'2T'7T'5T'6T'4T'3Average  Figure 5-1 Temperature time-variation for case D-C-6. See Table 5-1 for operating conditions. As shown in Figure 5-1, the inlet temperatures, T′1 and T′2, were very similar, and the highest temperatures measured within the DSRSB reactor. In the right compartment, the temperatures T′4 and T′6 first decreased and then levelled off. The decrease in temperature resulted from biomass at room temperature fed into the reactor with the feed port located on the right side. It is seen that T′3 was greater than T′4, associated with the biomass particles trajectory, with raw particles first fed into the right (upstream) chamber and then transported to the downstream chamber. This also explains why T′3 was slightly affected by the biomass feeding. It was also found that T′5 was higher than T′6, as T′6 was close to the biomass feed 113  port, where biomass at room temperature was continuously fed into the DSRSB reactor. The standard deviations of the T′1 to T′6 temperature readings were less than 3°C. Temperature T′7 increased with time, because the bed height gradually increased, causing more particles to be in contact with the T′7 thermocouple.  -5 0 5 10 15 20 25 30 35 40 45 50 551400175021002450-5 0 5 10 15 20 25 30 35 40 45 50 551400175021002450P (Pa) P (Pa)Time (min) P1 (Downstream compartment)   P2 (Upstream compartment) Figure 5-2 Pressure drop across DSRSB reactor for case D-C-6. See Table 5-1 for operating conditions. See Figure 2-8 for definitions of ΔP1 and ΔP2. Figure 5-2 shows the pressure drop across the DSRSB reactor for case D-C-6, measured at 1 s intervals. ΔP1 is the pressure drop across the downstream compartment, while ΔP2 is the pressure drop across the upstream compartment, where the feed port is located. It is seen that 114  the two pressure drops were similar for the initial ~15 min. After ~15 min, ΔP1 increased and was then higher than ΔP2. Over the rest of the experimental time, ΔP1 oscillated with an amplitude of ~700 Pa, which was about double the amplitude of ΔP2. Note that ΔP1 > ΔP2 during the torrefaction process in all cases, indicating that the downstream chamber had a higher inventory of biomass and inert particles. Note that biomass was continuously fed into the DSRSB from the right (upstream) side, disturbing solids mixing in the lateral direction between the two adjacent compartments. At similar operating conditions, ΔP1 was higher than the pressure drop across the SRSB reactor, while ΔP2 was close to the pressure drop across the SRSB reactor. The bed pressure drop oscillated during the entire period of torrefaction, while the biomass particles and inert particles mixed well in the DSRSB reactor (Du et al., 2015). 5.1.3 Evolution of ash content of torrefied product Figure 5-3 shows the evolution of the ash content of torrefied product during several torrefaction experiments in the DSRSB reactor. It is seen that the ash content of the torrefied sawdust increased initially, and then levelled off. The initial increase in the ash content resulted mainly from biomass decomposition and release of volatiles during the torrefaction process. The levelling off of the ash content indicated that the torrefied sawdust properties approached constant values. The ash content of the torrefied sawdust was found to increase as the temperature increased from 240 to 300°C at a given biomass feed rate. It has been previously reported (Prins et al., 2006b) that the formation of volatiles during the biomass decomposition process is mainly influenced by temperature. In addition, at the same temperature, the ash content was lower for a higher biomass feed rate, due to a shorter residence time at the higher biomass feed rate in the semi-batch reactor.  115  0 10 20 30 40 500.300.350.400.450.500.550.60  Ash content (%)Time (min) D-C-1: 600g/h+240oC D-C-2: 600g/h+270oC D-C-3: 600g/h+300oC (a) 0 10 20 30 40 500.300.350.400.450.500.550.60  Ash content (%)Time (min)  D-C-4: 900g/h+240oC  D-C-5: 900g/h+270oC  D-C-6: 900g/h+300oC (b) 0 10 20 30 40 500.300.350.400.450.500.550.60(c)  Ash content (%)Time (min)  D-C-7: 1400g/h+240oC  D-C-8: 1400g/h+270oC  D-C-9: 1400g/h+300oC -10 0 10 20 30 40 50 60 70 80 90 100 110 120 1300.300.350.400.450.500.550.600.650.70  Ash content (% dry)Time (min)  D-C-10: 240oC+900g/h+120min  D-C-11: 270oC+900g/h+120min  D-C-12: 300oC+900g/h+120min(d) Figure 5-3  Evolution of ash content of cyclone-caught torrefied sawdust for (a) D-C-1 to D-C-3; (b) D-C-4 to D-C-6; (c) D-C-7 to D-C-9; (d) D-C-10 to D-C-12. For detailed operating conditions see Table 5-1. 116  5.1.4 Final bed height after torrefaction D-C-1D-C-2D-C-3D-C-4D-C-5D-C-6D-C-7D-C-8D-C-9D-C-10D-C-11D-C-12050100150200250300350t=120mint=50mint=50minF=1400g/hF=900g/h  Bed depth, HB (mm)HB,0 = 176 mmF=600g/hF=900g/ht=50min Figure 5-4 Final average bed depth after 50 min of torrefaction. Figure 5-4 plots the final averaged bed depth after torrefaction experiments for three different biomass feed rates. The final bed depth was an average depth of two compartments after experiments. Compared to the initial bed depth of 176 mm (HB,0), the final bed depth (HB) increased after 50 min duration by 32.4-25.0%, 31.8-23.3%, and 39.2-25.0% at 240-300°C for 600, 900 and 1400 g/h biomass feed rates, respectively. As expected higher biomass feed rate resulted in greater final bed depth. The final bed height was lower at a higher temperature at a given biomass feed rate, due to higher weight loss at a higher temperature as the particles lost more weight during the torrefaction process and due to greater entrainment. When the experimental duration was extended from 50 to 120 min, the final bed height increased slightly at similar operating conditions, due to the semi-batch operation, with no active solids discharge port for the DSRSB reactor. 117  5.1.5 Solid product yield Table 5-2 Torrefied product yield and weight loss in torrefaction experiments with nitrogen atmosphere in DSRSB facility. Case: nominal conditions* Mt (g) Mc (g) Mr (g) Mf (g) Y (%) X (%) D-C-1: 240°C+600g/h+50min 557±13 150±3 310±8 10±0 32.6±0.8 15.5±0.3 D-C-2: 270°C+600g/h+50min 508±4 142±3 221±6 8±1 39.1±1 27.0±0.6 D-C-3: 300°C+600g/h+50min 511±5 144±3 173±4 5±0 45.5±1.1 37.1±0.7 D-C-4: 240°C+900g/h+50min 775±5 270±7 419±7 5±1 39.2±1.0 10.4±0.2 D-C-5: 270°C+900g/h+50min 763±4 290±6 300±8 5±0 49.2±1.2 22.2±0.4 D-C-6: 300°C+900g/h+50min 798±4 302±6 232±5 5±0 56.6±1.4 32.5±0.7 D-C-7: 240°C+1400g/h+50min 1173±21 572±11 473±12 10±1 54.8±1.4 10.1±0.2 D-C-8: 270°C+1400g/h+50min 1112±13 555±11 339±8 10±0 62.1±1.6 18.7±0.4 D-C-9: 300°C+1400g/h+50min 1173±5 655±16 245±6 10±1 72.8±1.8 22.4±0.6 D-C-10: 240°C+900g/h+120min 2192 1278 700 10 64.6 9.3 D-C-11: 270°C+900g/h+120min 1814±5 924±18 495±12 10±1 65.1±1.6 21.2±0.4 D-C-12: 300°C+900g/h+120min 1888 866 292 10 74.8 38.2         * See Table 5-1 for actual operating conditions. Mt: Total mass of sawdust fed; Mc: Mass of torrefied sawdust captured by cyclone; Mr: mass of torrefied sawdust remaining in reactor; Mf: Mass of torrefied sawdust captured by filter. ±value: Value of standard deviation.  Table 5-2 gives the weight loss of biomass and the mass yields in the reactor (Mr), cyclone (Mc) and filter (Mf) for torrefaction experiments with a nitrogen atmosphere in the DSRSB reactor. It was found that the weight loss of biomass, defined by Equation (4-1), increased with increasing temperature at a given biomass feed rate. On the other hand, the weight loss of biomass decreased as the biomass feed rate increased from 600 g/h to 1400 g/h at a given temperature.  118  When the experimental duration was extended from 50 to 120 min, the weight losses of biomass were very similar at 240 and 270°C, i.e. D-C-4 vs D-C-10 and D-C-5 vs D-C-12, respectively. However, the weight loss of biomass was greater for case D-C-12 (120 min duration) than for case D-C-6 (50 min duration). This is likely because the DSRSB system did not reach steady state within 50 min.  The true yield Y, defined by Equation (4-2), was found to increase with increasing temperature, due to the weight loss of biomass increasing with increasing temperature and more entrainment at a given feed rate. At a given temperature, Y increased as the biomass feed rate increased from 600 to 1400 g/h. In addition, Y was greater than 50% for a feed rate of 1400 g/h. DSRSB Torrefaction Reactor(T: 240-300°C)CycloneFilterVolatiles and gases= + + +Screw-Feeder(Feed rate: 600-1400 g/h)Mass (%)100Mass (%)34-56Mass (%)27-28Mass (%)<2Mass (%)16-37Feed rate(g/h)~ 600100~ 900100~ 140029-5421-4035-3849-56<1<111-3310-22 Figure 5-5 Mass distribution of raw and torrefied sawdust for biomass torrefaction in DSRSB facility. 119  Figure 5-5 summarizes the mass distribution of torrefied biomass in the DSRSB facility. As the biomass feed rate increased from 600 to 1400 g/h, the weight loss in the form of volatiles decreased, the mass percentage of torrefied product in the reactor decreased, that in the cyclone increased, and the mass percentage of torrefied product in the filter was hardly affected. However, as noted above, the mass of torrefied product remaining in the reactor increased when the biomass feed rate increased, as shown in Table 5-2.  5.1.6 Properties of torrefied product Table 5-3 shows relevant properties of torrefied product obtained from the cyclone and reactor. As expected the volatile content of the torrefied sawdust was generally lower than that of the raw sawdust. At a given biomass feed rate, the volatile content decreased with increasing temperature, because decomposition of the biomass is strongly affected by the torrefaction temperature (Prins et al., 2006a; Prins et al., 2006b). The fixed carbon content increased with increasing temperature, as a consequence of the decreased volatile content. The biomass feed rate had a greater impact on the carbon and volatile contents of the torrefied product at 240°C than at 270 and 300°C, illustrating that temperature was more important than biomass feed rate in the torrefaction experiments. In addition, torrefied sawdust from the cyclone had greater volatile content and lower fixed carbon content compared to torrefied sawdust from the reactor, due to longer residence times for the biomass particles remaining in the reactor. The HHV of the torrefied sawdust was found to be much greater than that of the raw sawdust, which is one of important advantages of torrefaction. As the temperature rose from 240 to 300°C, the HHV of torrefied sawdust from the cyclone increased by 1.4, 1.8 and 3.7%, while 120  the HHV of torrefied sawdust from the reactor increased by 11.6, 16.5 and 20.8% for 600, 900 and 1400 g/h feed rates, respectively. This illustrated that the temperature was more important at higher biomass feed rate. The increased HHV in the torrefied product was due to small molecules with low HHV being released in the torrefaction process (Chen and Kuo, 2011). Furthermore, the HHV of the torrefied sawdust from the reactor was higher than that of the torrefied sawdust from the cyclone, as a result of longer residence times for the biomass particles remaining the reactor.  After torrefaction, the particle density of the biomass particles increased to some extent, resulting from the shrinkage of particles during torrefaction. However, the particle densities of torrefied sawdust from the cyclone and reactor were similar. The particle density was only slightly affected by the biomass feed rate, but it was significantly affected by the temperature. As the temperature increased from 240 to 300°C, the particle density decreased gradually. It has previously been shown (Chen et al., 2014a) that the pore volume of the biomass increased after torrefaction, due to the volatiles release. It was also found that energy yield was significantly affected by the temperature and biomass feed rate. As the temperature increased from 240 to 300°C, the energy yield decreased from 91.5 to 73.3, 97.6 to 79.2 and 96.9 to 90.1% for 600, 900 and 1400 g/h, respectively. This was because a higher biomass feed rate led to shorter residence times for biomass particles in the reactor. Compared to the SRSB, the energy yield of biomass torrefied in the DSRSB was lower, possibly associated with better heat transfer in the DSRSB than in the SRSB. 121  Table 5-3 Properties of torrefied solid sawdust on a dry basis. Case D-C-1 240°C 600g/h 50 min D-C-2 270°C 600g/h 50 min D-C-3 300°C 600g/h 50 min D-C-4 240°C 900g/h 50 min D-C-5 270°C 900g/h 50 min D-C-6 300°C 900g/h 50 min D-C-7 240°C 1400g/h 50 min D-C-8 240°C 1400g/h 50 min D-C-9 240°C 1400g/h 50 min  Torrefied sawdust from cyclone accumulated over run Proximate analysis   VM (wt.%) 82.61±0.08 81.75±0.14 80.28±0.12 83.46±0.22 81.73±0.02 80.34±0.79 82.61±0.01 82.57±0.20 81.28±0.28   FC (wt.%) 16.95±0.10 17.76±0.20 19.22±0.14 16.14±0.23 17.82±0.11 19.16±0.83 16.97±0.04 17.02±0.04 18.23±0.30   AC (wt.%) 0.44±0.02 0.49±0.06 0.50±0.02 0.40±0.01 0.45±0.09 0.50±0.04 0.42±0.03 0.41±0.02 0.49±0.02 Elemental analysis   C (wt.%) 50.9 51.4 51.8 50.7 52.4 52.7 50.7 51.0 52.1   H (wt.%) 6.1 6.1 6.1 6.1 6.0 5.9 6.1 6.1 6.0   N (wt.%) <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1   O* (wt.%) 43.0 42.5 42.1 43.2 41.6 41.4 43.2 42.9 41.9 H/C 0.12 0.12 0.12 0.12 0.11 0.11 0.12 0.12 0.12 O/C 0.84 0.83 0.81 0.85 0.79 0.78 0.85 0.84 0.81 HHV (MJ/kg) 19.98±0.01 20.05±0.01 20.26±0.07 19.85±0.07 19.96±0.22 20.21±0.19 19.56±0.19 19.90±0.02 20.29±0.01 ρp (kg/m3) 1365±25 1342±14 1326±14 1386±19 1360±90 1358±42 1377±5 1370±5 1365±11  Torrefied sawdust from reactor at end of run Proximate analysis   VM (wt.%) 78.14±0.38 73.80±1.19 69.04±0.98 80.44±0.18 74.26±0.17 66.19±0.92 79.28±0.62 76.50±0.63 60.86±0.38   FC (wt.%) 21.40±0.41 25.65±1.21 30.36±0.99 19.12±0.22 25.24±0.27 33.14±0.96 20.31±0.66 22.95±0.65 38.23±0.38   AC (wt.%) 0.47±0.03 0.55±0.02 0.60±0.01 0.43±0.04 0.50±0.10 0.66±0.04 0.41±0.03 0.54±0.02 0.92±0.01 HHV (MJ/kg) 20.27±0.10 21.32±0.01 22.62±0.02 20.09±0.02 21.05±0.04 23.40±0.21 20.17±0.12 21.08±0.06 24.37±0.06 ρp (kg/m3) 1346±6 1305±2 1271±7 1383±7 1315±8 1298±1 1411±20 1435±2 1308±9 Energy yield (%) 91.5±0.3 81.6±0.0 73.3±0.3 97.6±0.2 86.9±0.4 79.2±0.7 96.9±0.8 89.7±0.2 90.1±0.1 122  Table 5-3 Properties of torrefied solid sawdust on a dry basis (continued). Parameter D-C-10 240°C 900g/h 120min D-C-11 270°C 900g/h 120min D-C-12 300°C 900g/h 120min  Torrefied sawdust from cyclone accumulated over run Proximate analysis      VM (wt.%) 82.80±0.49 80.89±1.15 74.40±1.40   FC (wt.%) 17.20±0.50 18.57±1.16 25.60±1.41   AC (wt.%) 0.37±0.01 0.43±0.01 0.58±0.01 Elemental analysis      C (wt.%) 51.21 52.15 55.28   H (wt.%) 6.14 6.11 5.72   N (wt.%) <0.1 <0.1 <0.1   O* (wt.%) 42.65 41.74 39.00 H/C 0.12 0.12 0.10 O/C 0.83 0.80 0.71 HHV (MJ/kg) 20.04±0.12 20.52±0.01 21.79±0.01 ρp (kg/m3) 1393±13 1471±30 1331±15  Torrefied sawdust from reactor at end of run Proximate analysis      VM (wt.%) 79.62±0.29 74.98±0.61 56.24±1.86   FC (wt.%) 20.38±0.31 24.18±0.68 43.76±1.87   AC (wt.%) 0.34±0.03 0.51±0.07 0.88±0.01 Elemental analysis      C (%) 52.13 55.08 55.30   H (%) 6.13 6.04 4.98   N (%) <0.1 <0.1 <0.1   O* (%) 41.74 38.88 39.72 H/C 0.12 0.11 0.09 O/C 0.80 0.71 0.72 HHV (MJ/kg) 20.52±0.11 21.39±0.13 25.95±0.03 ρp (kg/m3) 1478±4 1420±20 1220±7 Energy yield (%) 99.6 89.4±0.2 76.8 ±value: Value of standard deviation.  *Oxygen was determined by difference.                                   For properties of the raw sawdust feed, see Table 2-3.  123  5.1.7 Liquid product 0 5 10 15 20 25 30 3501E62E63E64E65E66E67E68E69E61E7AbundanceTime (min)(A) T=240oC 0 5 10 15 20 25 30 3501E62E63E64E65E66E67E68E69E61E7AbundanceTime (min)(B) T=270oC0 5 10 15 20 25 30 3501E62E63E64E65E66E67E68E69E61E7AbundanceTime (min)(C) T=300oC Figure 5-6 GC-MS spectra of condensable liquid obtained from (A) D-C-10; (B) D-C-11; (C) D-C-12. For operating conditions see Table 5-1. 124  The condensable liquid collected immediately downstream of the heat exchanger during the 2 h torrefaction experiments (D-C-10 to D-C-12) were analyzed by GC-MS. The chemical composition of the condensable liquid was determined based on the identification. Figure 5-6 shows GC-MS spectra of the condensable liquids obtained at 240, 270 and 300°C with a biomass feed rate of 900 g/h. 33, 42 and 44 chemical compounds were detected for the condensable liquid obtained at 240, 270 and 300°C, respectively. It was found that the three liquids contained ten chemical compounds in common, listed in Table 5-4. These components included acids, alcohols, ketones, phenols, aldehydes, and esters, similar to those identified in previous studies (Arteaga-Pérez et al., 2015; Chang et al., 2012). Table 5-4 Main compounds identified in condensable liquid via GC-MS.  Residence Time (min) Library/ID 1 10.45 Acetic acid  2 10.80 2-Furancarboxaldehyde  3 13.10 2-Furanmethanol  4 14.49 Cyclononasiloxane, octadecamethyl-  5 16.32 Phenol, 2-methoxy-  6 17.93 Phenol, 2-methoxy-4-methyl-  7 24.81 Phenol, 2-methoxy-4-(1-propenyl)-, (E)-  8 25.03 Tetracosamethyl-cyclododecasiloxane 9 27.28 2-Furancarboxaldehyde, 5-(hydroxymethyl)-  10 28.58 Vanillin   125  240 250 260 270 280 290 30081012141618Water content (wt.%)T(oC) Figure 5-7 Water content of condensable liquid collected for D-C-10: T = 240°C, D-C-12: T = 270°C, D-C-10: T = 300°C. For operating conditions see Table 5-1. The water in the condensable liquid mainly resulted from the absorbed water, constitutive water and dehydration of hydroxyls in the raw sawdust. The last two sources were more important in the present study, because the biomass was pre-dried before being torrefied. As shown in Figure 5-7, the water content in the condensable liquid decreased from 16.4±0.3 wt.% to 8.7±0.2 wt.%, as the torrefaction temperature increased from 240 to 300°C.  This result is consistent with previous findings (Chang et al., 2012; Chen et al., 2014b), because more organic compounds are formed due to a higher degree of decomposition of biomass when the torrefaction temperature is increased; thus, the water content of the liquid products decreased with increasing temperature. 126  5.1.8 Product gas composition 240 250 260 270 280 290 300050100150200250300 CO (ppm)T (oC) CO050100150200250300350400 CO2CO2 (ppm) Figure 5-8 CO and CO2 concentrations in the off-gas during torrefaction at end of run D-C-10 T = 240°C, run D-C-12: T = 270°C, and run D-C-10: T = 300°C. For operating conditions see Table 5-1. Off-gas compositions obtained from biomass torrefaction mainly include carbon monoxide, carbon dioxide, and traces of hydrogen and methane (Chang et al., 2012; Wang et al., 2016a). Because the hydrogen and methane concentrations were lower than the lower detectable limit of 0.5 vol.% of our GC, and could not be measured accurately, only carbon monoxide and carbon dioxide concentration data were obtained. Figure 5-8 plots these concentrations in the off-gas of the biomass torrefaction. It is seen that the CO concentration increased from 24±2 ppm to 231±5 ppm and the CO2 concentration from 129±4 ppm to 344±17 ppm when the torrefaction temperature increased from 240 to 300°C. Carbon monoxide release was mainly caused by the cracking of carbonyl (C-O-C) and carboxyl (C=O) groups and the secondary 127  reactions of carbon dioxide and steam with porous char (Wang et al., 2016a). The carbon dioxide release was mainly contributed by the cracking and reforming of carboxyl (C=O) and carboxylic acid (COOH) groups in hemicellulose of SPF sawdust (Wang et al., 2016a; Yang et al., 2007). Previous studies (Wang et al., 2016a; Yang et al., 2007) found that CO and CO2 releases at temperatures < 500°C could be attributed to the decomposition of hemicellulose rather than cellulose and lignin, and CO and CO2 concentrations increased with temperature. Moreover, due to the secondary reactions, the CO was formed more aggressively with increasing temperature, resulting in the CO release rate being higher than the CO2 release rate (Deng et al., 2009; Shen et al., 2010). Furthermore, release of these gases during torrefaction led to a decrease of elemental oxygen in the torrefied sawdust, consistent with the results in Table 5-3.  5.2 Biomass torrefaction in oxygen-containing atmosphere This section reports the effect of varying the oxygen concentration in the feed-gas on the biomass torrefaction in the dual-compartment slot-rectangular spouted bed (DSRSB) reactor. Results of the oxidative torrefaction are then compared with the results of the non-oxidative torrefaction experiments. The concentration of oxygen was controlled by varying the proportion of pure nitrogen and air flow rates in the feed-gas, with the oxygen content varied from 3 to 9 vol.%. The nominal biomass feed rate was ~900 g/h, the superficial gas velocity was 1.2Ums, and each experimental run lasted 50 min. Ums of inert particles was measured for each experiment at elevated temperature prior to initiating biomass feeding.   128  5.2.1 Operating conditions for oxidative torrefaction Table 5-5 Operating conditions for oxidative torrefaction experiments in DSRSB reactor. U = 1.2Ums, operating time=50 min. Case Nominal operating conditions (T, F, XO2) Actual operating conditions T (°C) F (g/h) XO2 (vol.%) D-OT-1 240°C, 900 g/h, 0vol.% O2 242±3 930±11 0.0 D-OT-2 240°C, 900 g/h, 3vol.% O2 241±3 908±15 3.3±0.2 D-OT-3 240°C, 900 g/h, 6vol.% O2 243±3 929±12 6.2±0.3 D-OT-4 240°C, 900 g/h, 9vol.% O2 250±2 1046±40 9.0±0.2 D-OT-5 270°C, 900 g/h, 0vol.% O2 271±3 916±10 0.0 D-OT-6 270°C, 900 g/h, 3vol.% O2 273±2 920±20 3.6±0.2 D-OT-7 270°C, 900 g/h, 6vol.% O2 282±4 946±16 6.4±0.1 D-OT-8 270°C, 900 g/h, 9vol.% O2 Not done due to safety reasons D-OT-9 300°C, 900 g/h, 0vol.% O2 299±5 957±10 0.0 D-OT-10 300°C, 900 g/h, 3vol.% O2 304 917 3.3 D-OT-11 300°C, 900 g/h, 6vol.% O2 Not done due to safety reasons D-OT-12 300°C, 900 g/h, 9vol.% O2 Not done due to safety reasons    ±value: Value of standard deviation.  Table 5-5 lists the operating conditions for the oxidative torrefaction experiments in the DSRSB reactor. Runs D-OT-2 to D-OT-4, D-OT-6 and D-OT-7 were repeated. Prior to each torrefaction experiment, the oxygen concentration in the feed-gas was measured and confirmed to be close to the planned value by the flue gas analyzer (PS-200, HORIBA). During the torrefaction experiments, the oxygen concentration was measured at 1 s intervals. As shown in Table 5-5, the time-averaged oxygen concentrations (XO2) were close to the pre-chosen values. The temperature variation was less than 4°C, and the biomass feed rate 129  variation was less than 25 g/h in the repeated experiments. The relative uncertainty of local temperature was less than 3.5°C in each experiment, excluding cases D-OT-7 and D-OT-10. The temperature of the reactor underwent an abrupt increase in cases D-OT-7 and D-OT-10, as shown in Figure 5-9. However, no biomass ignition was found during the experiments. This suggested that the oxygen concentration in the feed-gas had reached an upper limit. For safety reasons, higher oxygen concentrations were therefore not tested for temperatures of 270 and 300°C, i.e. D-OT-8, D-OT-11 and D-OT-12. Note that average reactor temperature for cases D-OT-7 and D-OT-10 is based on the average of thermocouples T′5 and T′6, whereas other averages were calculated by averaging all four thermocouples from T′3 to T′6. 0 10 20 30 40 50250300350400450500T (oC)Time (min) T'1 (downstream windbox) T'2 (upstream windbox) T'3 (Z=64mm, downstream) T'4 (Z=64mm, upstream) T'5 (Z=165mm, downstream) T'6 (Z=165mm, upstream) T'7 (Z=267mm, upstream)T'4T'3T'2T'1T'7T'5T'6(a) D-OT-7 130  0 10 20 30 40 50280300320340360380400420T'3T'4T'6T'5T'7T'2T (oC)Time (min) T'1 (downstream windbox) T'2 (upstream windbox) T'3 (Z=64mm, downstream) T'4 (Z=64mm, upstream) T'5 (Z=165mm, downstream) T'6 (Z=165mm, upstream) T'7  (Z=267mm, upstream)T'1(b) D-OT-10 Figure 5-9 Temperature time-variation for (a) case D-OT-7; (b) case D-OT-10. See Table 5-5 for operating conditions.  5.2.2 Evolution of ash content of torrefied product Figures 5-10(a)-(c) plot the evolution of ash content in the torrefied product of the oxidative torrefaction. It is observed that the ash content first increased, and then levelled off with time. The increase in ash content of the torrefied sawdust mainly resulted from biomass decomposition and oxidation. The levelling off of the ash content of the torrefied sawdust indicated that the torrefied sawdust properties approached constant values.  In Figure 5-10(a), the ash content of the torrefied sawdust reached a maximum at ~20 min for all cases. The ash content was higher after torrefaction in the oxygen-containing atmosphere than in the inert atmosphere. It was also found that the ash content increased as the oxygen concentration increased. The ash content, as shown in Figure 5-10(b), reached a maximum at ~ 10 min for all cases, except for 6 vol.% O2. The ash content for 6 vol.% O2 and 270°C 131  increased continuously after 20 min, consistent with the observed temperature variation in the DSRSB reactor shown in Figure 5-9(a). In this case, the reactor temperature increased continuously starting from ~20 min, leading to more severe biomass decomposition, oxidation, and even carbonization. In Figure 5-10(c), it is seen that the ash content kept increasing as 3 vol.% oxygen was included in the feed-gas, consistent with the reactor temperature variation shown in Figure 5-9(b). 0 10 20 30 40 500.300.350.400.450.500.550.60Ash content (%)Time (min)O2 concentration 0%  3%  6%  9%(a) T=240oC0 10 20 30 40 500.300.350.400.450.500.550.600.650.70Ash content (%)Time (min)O2 concentration 0%  3%  6%(b) T=270oC0 10 20 30 40 500.300.350.400.450.500.550.600.650.700.750.80Ash content (%)Time (min)O2 concentration 0%  3%(c) T=300oC Figure 5-10 Evolution of ash content of torrefied product captured by the cyclone for (a) D-OT-1 to D-OT-4; (b) D-OT-5 to D-OT-7; (c) D-OT-9 and D-OT-10. For detailed operating conditions see Table 5-5. 132  5.2.3 Final bed depth after torrefaction Figure 5-11 shows the final averaged bed height after 50 min of oxidative and non-oxidative torrefaction. It is seen that adding oxygen to the carrier gas had only a slight effect on the final bed height. The oxidative torrefaction led to a slightly lower final bed height, compared to the torrefaction experiments in the absence of oxygen. The sawdust lost more weight and its properties changed more quickly during oxidative torrefaction. Hence, increasing the oxygen concentration resulted in slightly lower final bed height, due to higher weight loss of biomass in the oxygen-containing atmosphere. The final bed height was lowest for 300°C, compared to 240 and 270°C, at otherwise similar operating conditions.  0 3 6 9 0 3 6 0 3050100150200250300HB,0 = 176 mm300oC270oC  Bed depth, HB (mm)Oxygen Conc. (vol.%)240oC Figure 5-11 Final average bed depth after oxidative torrefaction at nominal conditions of F = 900 g/h, T = 240, 270 and 300°C with various oxygen concentrations in the feed-gas. 133  5.2.4 Solid product yield Table 5-6 Torrefied product yield and weight loss in oxidative torrefaction experiments in DSRSB facility. Case: nominal conditions* Mt (g) Mc (g) Mr (g) Mf (g) Y (%) X (%) D-OT-1: 240°C+0vol.%O2 775±5 270±7 419±7 5±1 39.2±1.0 10.4±0.2 D-OT-2: 240°C+3vol.%O2 757±6 274±5 382±10 9±0 41.8±1.0 12.2±0.3 D-OT-3: 240°C+6vol.%O2 774±5 277±6 394±10 9±0 41.3±1.0 12.2±0.2 D-OT-4: 240°C+9vol.%O2 872±17 309±8 419±7 10±1 42.4±1.1 15.3±0.3 D-OT-5: 270°C+0vol.%O2 763±4 290±6 300±8 5±0 49.2±1.2 22.2±0.4 D-OT-6: 270°C+3vol.%O2 767±8 298±6 268±6 6±0 52.7±1.3 25.4±0.6 D-OT-7: 270°C+6vol.%O2 788±7 328±7 218±5 10±1 60.1±1.5 29.5±0.3 D-OT-9: 300°C+0vol.%O2 798±4 302±6 232±5 5±0 56.6±1.4 32.5±0.7 D-OT-10: 300°C+3vol.%O2 764 229 188 9 54.9 44.2  * See Table 5-5 for detailed operating conditions. Mt: Total mass of sawdust fed; Mc: Mass of torrefied sawdust captured by cyclone; Mr: Mass of torrefied sawdust remaining in reactor; Mf: Mass of torrefied sawdust captured by filter. ±value: Value of standard deviation. Table 5-6 shows the weight loss of biomass and the mass yield in the reactor (Mr), cyclone (Mc) and filter (Mf) during the oxidative torrefaction experiments. The weight loss of biomass is defined by Equation (4-1). Compared to the non-oxidative torrefaction, it is seen that the oxygen-containing atmosphere led to greater weight loss of biomass, due to biomass oxidation and decomposition occurring simultaneously during the oxidative torrefaction (Wang et al., 2013a).  The true yield, Y, defined by Equation (4-2) increased with increasing oxygen concentration, indicating that more torrefied biomass was entrained from the reactor when more oxygen was 134  included in the feed-gas. The greater entrainment is explained by the particles losing more weight when torrefaction was carried out in the presence of oxygen. DSRSB  Torrefaction Reactor(T: 240-300°C)(N2 (91-100 vol.%)& O2 (0-9 vol.%))CycloneFilterVolatiles and gases= + + +Screw-Feeder(Feed rate: 900 g/h)Mass (%)100Mass (%)29-54Mass (%)35-38Mass (%)<1Mass (%)11-33O2 conc. (vol.%)01003-9 25-51 35-42 <2 12-44 Figure 5-12 Mass distribution of raw and torrefied sawdust for oxidative torrefaction in DSRSB facility. Figure 5-12 summarizes the mass distribution of torrefied biomass in the oxidative torrefaction in the DSRSB facility. Comparison with Figure 5-5 shows that adding oxygen to the feed-gas led to decreased product mass in the reactor, increased product mass in the cyclone, slightly increased mass captured by the filter, and greater weight loss.  5.2.5 Properties of torrefied product Table 5-7 shows properties of the torrefied solid product obtained from the cyclone and reactor in the oxidative torrefaction. Proximate analysis showed that, as expected, the torrefied sawdust always had a lower volatile content than the raw sawdust. This is due to dehydration, devolatilization and oxidation (Prins et al., 2006a; Prins et al., 2006b). At a 135  given temperature, when oxygen was added to the feed-gas, the volatile content of the torrefied product decreased, while its fixed carbon content increased. As the oxygen concentration increased, the volatile content of the torrefied sawdust decreased, whereas its fixed carbon content increased. This illustrated that small hydrocarbon molecules were more easily liberated in the presence of oxygen in the torrefaction experiments, because of biomass oxidation and decomposition (Haykırı-Açma, 2003; Munir et al., 2009). An increase in temperature can facilitate devolatilization and oxidation processes during torrefaction (Haykırı-Açma, 2003). The impact of oxygen on the volatile and fixed carbon contents was more significant at a higher temperature.  Elemental analysis showed that the torrefied product had higher atomic carbon content than the raw sawdust, but lower atomic oxygen and hydrogen contents, probably due to the release of moisture and decomposition of –COOH and –OH groups in the biomass (Chen et al., 2014a). The atomic carbon content of oxidatively torrefied product was found to be higher than that of the non-oxidatively torrefied product at 240, 270 and 300°C. This is believed to be because the oxidation of hydrocarbons in the biomass plays an important role in decomposing biomass (Chen et al., 2014a; Daood et al., 2010; Munir et al., 2009). In addition, the atomic oxygen and hydrogen contents of the solid product decreased to some extent, when the oxygen content of the feed-gas increased. These findings are associated with oxidation of hydrogen, carbonization of biomass (Chen et al., 2014a), and combustion of volatiles released during the torrefaction process (Haykırı-Açma, 2003). Whether the sawdust was torrefied in pure nitrogen or in an oxygen-containing environment, the HHV of the torrefied product was higher than that of the raw sawdust. For operation at 240°C, introducing oxygen in the feed-gas resulted in a slight decrease in HHV for the 136  torrefied product at otherwise similar operating conditions. However, for temperatures of 270 and 300°C, as oxygen was added to the feed-gas, the HHV of the torrefied product increased. This indicated that introducing oxygen in the torrefaction experiment could improve the torrefaction performance at higher temperatures. This finding is consistent with our results obtained in the SRSB reactor (see section 4.3 of Chapter 4), as well as data reported by Almeida et al. (2010) and Wang et al. (2013a), but opposite to Chen et al. (2014a).  Furthermore, as the oxygen concentration increased from 3 to 9 vol.%, the HHV of the torrefied product increased slightly, likely because cellulose and lignin mainly contributed to the HHV of the torrefied sawdust (Chen and Kuo, 2011), and these two constituents were insensitive to the oxygen concentration in the range covered of 3-9 vol.%.  The energy yield decreased when oxygen was added to the feed-gas during the torrefaction. The effect of oxygen concentration variation was more significant at a higher temperature. This is because an increase in temperature has a positive effect on biomass oxidation and torrefaction in the oxidative torrefaction process. Although the energy yield drastically decreased with increasing oxygen content in the feed-gas,  98-66% energy yield is still in a reasonable range for making high-quality pellets (Cao et al., 2015; Li et al., 2015; Peng et al., 2013b), leading to high-quality bio-oil or syngas (Sarkar et al., 2014).   In addition, compared to the torrefied sawdust obtained from the cyclone, the torrefied product remaining in the reactor had lower volatile content, higher fixed carbon content and greater HHV, due to the longer residence time of the torrefied sawdust remaining in the reactor. 137  Table 5-7 Properties of torrefied sawdust on a dry basis.  * Oxygen was determined by difference. ±value: Value of standard deviation. Parameter D-OT-1: 240°C 0vol.%O2 D-OT-2: 240°C 3vol.%O2 D-OT-3: 240°C 6vol.%O2 D-OT-4: 240°C 9vol.%O2 D-OT-5: 270°C 0vol.%O2 D-OT-6: 270°C 3vol.%O2 D-OT-7: 270°C 6vol.%O2 D-OT-9: 300°C 0vol.%O2 D-OT-10: 300°C 3vol.%O2  Torrefied sawdust accumulated in cyclone over 50 min run Proximate analysis   VM (wt.%) 83.46±0.22 82.76±0.27 82.23±0.60 81.65±1.20 81.73±0.02 81.09±1.27 76.37±0.05 80.34±0.79 73.40±1.19   FC (wt.%) 16.14±0.23 16.83±0.29 17.35±0.60 17.94±1.22 17.82±0.11 18.40±1.30 23.10±0.08 19.16±0.83 26.04±1.21   AC (wt.%) 0.40±0.01 0.41±0.02 0.42±0 0.41±0.02 0.45±0.09 0.51±0.03 0.53±0.03 0.50±0.04 0.56±0.02 Elemental analysis   C (wt.%) 50.7 50.8 51.1 52.2 52.4 52.7 53.6 52.7 54.4   H (wt.%) 6.1 6.1 6.0 5.9 6.0 5.6 5.7 5.9 5.5   N (wt.%) <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1   O* (wt.%) 43.2 43.1 42.9 41.9 41.6 41.7 40.7 41.4 40.1 H/C 0.12 0.12 0.12 0.11 0.11 0.11 0.11 0.11 0.10 O/C 0.85 0.85 0.84 0.80 0.79 0.79 0.76 0.78 0.74 HHV (MJ/kg) 19.85±0.07 19.30±0.65 19.48±0.56 19.63±0.12 19.96±0.22 20.21±0.04 20.70±0.15 20.21±0.19 20.67±0.72 ρp (kg/m3) 1386±19 1380±41 1386±3 1381±41 1360±90 1342±51 1339±16 1398±42 1287±14  Torrefied sawdust from reactor at end of 50 min run Proximate analysis   VM (wt.%) 80.44±0.18 79.13±0.13 78.92±0.04 77.44±0.41 74.26±0.17 73.78±0.75 63.51±0.59 66.19±0.92 51.68±0.16   FC (wt.%) 19.13±0.22 20.43±0.20 20.65±0.06 22.13±0.43 25.24±0.27 25.71±0.5 35.95±0.61 33.14±0.96 47.50±0.16   AC (wt.%) 0.43±0.04 0.44±0.06 0.43±0.02 0.43±0.02 0.50±0.10 0.51±0.01 0.54±0.03 0.66±0.04 0.83±0.01 HHV (MJ/kg) 20.09±0.02 20.09±0.30 20.17±0.31 20.45±0.12 21.05±0.04 21.00±0.05 23.07±0.04 23.40±0.21 23.50±0.34 ρp (kg/m3) 1383±7 1363±60 1361±13 1357±11 1315±8 1317±19 1259±12 1248±1 1229±6 Energy yield (%) 97.6±0.2 93.9±0.5 94.5±1.9 92.1±0.1 86.9±0.4 83.3±0.2 82.3±0.3 79.2±0.7 65.7±0.7 138  5.3 Conclusions  (i) Non-oxidative torrefaction In the non-oxidative torrefaction, temperature had more influence on torrefaction performance than biomass feed rate. An increase in temperature from 240 to 300°C led to greater weight loss, lower energy yield, decreased final bed height and more torrefied sawdust captured by the cyclone. It also produced torrefied sawdust with higher HHV, greater carbon content, lower hydrogen and oxygen contents, less volatile content and increased fixed carbon content. An increase in biomass feed rate from 600 to 1400 g/h resulted in less weight loss, higher energy yield and more sawdust remaining in the reactor, while also resulting in torrefied sawdust with lower HHV, less carbon content, greater hydrogen and oxygen contents, more volatile content and reduced fixed carbon content. (ii) Oxidative torrefaction Sawdust torrefied in an oxygen-containing atmosphere had higher weight loss of biomass and lower energy yield. Compared to non-oxidatively torrefied sawdust, oxidatively torrefied sawdust has higher HHV, greater fixed carbon and lower hemicellulose at 270 and 300°C. Adding oxygen to the feed-gas decreased the product mass in the reactor, increased the product mass in the cyclone, and slightly increased the mass captured by the filter. Varying the oxygen concentration had a greater effect on biomass torrefaction at a higher temperature. An increase in oxygen concentration resulted in greater HHV, higher atomic carbon content, lower atomic hydrogen and oxygen contents, less volatile content, greater fixed carbon content, and less decomposition of hemicellulose in the torrefied sawdust.  In order to avoid thermal runaway in reactors, our experience indicates that the oxygen concentration should 139  not exceed 6 vol.% and 3 vol.% in the feed gas at 270 and 300°C, respectively, when biomass is torrefied in a DSRSB reactor.     140  Chapter 6 Thermogravimetric characteristics and kinetic analysis of torrefied biomass pyrolysis Raw SPF sawdust was torrefied at different temperatures, biomass feed rates and feed gas oxygen concentrations in the SRSB reactor, as described in Chapter 4. In this chapter, a thermogravimetric analyzer is used to assess the pyrolysis characteristics of the torrefied sawdust captured by the cyclone during biomass torrefaction in the SRSB reactor. The aim of the present chapter is to determine the values of the kinetic parameters characterizing the pyrolysis of sawdust torrefied in non-oxidative and oxidative media using data from non-isothermal thermogravimetry in an inert nitrogen atmosphere. 6.1 Modelling The overall reaction for biomass decomposition in the nitrogen atmosphere can be written as  Wood → Volatiles + Char (6-1)         For this reaction, the kinetic equation can be written as   ))f(k(d/d Ttα   (6-2) where k(T) is the rate constant, f(α) depends on the reaction mechanism, t is the reaction time, and α is the conversion defined as  f0i0m-mm-mα  (6-3) 141  Here m0, m𝑖  and m𝑓  refer to initial, actual and ultimate (final) mass of the solid sample, respectively.  k(T) is the rate constant, expressed by the Arrhenius equation as   )exp( )k( E/RTAT  (6-4) where A is the pre-exponential factor (s-1), E is the activation energy (J/mol), R is the universal gas constant of 8.314 J/(mol·K), and T is the absolute reaction temperature (K).  Substitution of Equation (6-4) into Equation (6-2) gives     )exp( )f(RTE-Adtd  (6-5) For a constant heating rate (β)  /dTdt   (6-6) Substituting Equation (6-6) into Equation (6-5) gives   dT )exp()f( RTE-Ad  (6-7) 6.1.1 Kissenger-Akahira-Sunose (KAS) model According to the KAS method, Equation (6-7) can be integrated with initial conditions α = 0, T = T0 at time 0 to give  )(dT )exp()f()(00x PRAERTE-AdgTT   (6-8) 142  where x=E/RT and 02/)(xx dxxexP  The KAS method makes no assumption concerning the kinetic model, but uses the approximation   2)(xexPx  (6-9) With this approximation taken into account, the logarithm of Equation (6-8) gives  RTEEART-)g(ln - ln)ln(2  (6-10) Therefore, the activation energy (E) can be calculated from the slope of the line of ln(β/T2) vs 1/T for a given conversion, α. To apply the KAS method, it is necessary to conduct experiments at three or more different heating rates. In the present study, four heating rates of 10, 20, 30 and 40°C/min were tested.  6.1.2 Coats-Redfern Method Integration of Equation (6-7) with α = 0, T = T0 at time 0 gives:   TT RTE-Adg0dT )exp()f()(0  (6-11) There is no exact analytical solution for the right side of Equation (6-11). Cauchy’s rule is therefore applied to estimate the integral, i.e.   TT RTEERTERTE-A0)exp()21(ARTdT )exp(2 (6-12) Since 2RT/E<<1, Equation (6-11) can be rewritten as 143   RTEEARRTEERTEARTg)ln(              ))21(ln()(ln2 (6-13) A plot of ln(g(α)/T2) vs 1/T should then result in a straight line when an appropriate mechanism (f(α) or g(α)) is selected.  The activation energy (E) can then be estimated from the slope -E/R, while the pre-exponential factor (A) can then be determined by the intercept ln(AR/βE).  The function f(α) or g(α) depends on the reaction mechanism and its mathematical model. Based on previous work (Masnadi et al., 2014; Vlaev et al., 2003), the chemical reaction is well represented by a first order reaction, and the volatile diffusion within the particles is well fitted by the Zhuravlev et al. (1948) equation for biomass pyrolysis, shown in Table 6-1. Table 6-1 Mechanism functions for modelling pyrolysis of raw/torrefied biomass. Mechanism f(α) g(α) Chemical reaction (first order) 1- α -ln(1- α) Diffusion (Zhuravlev equation) (2/3)(1- α)5/3/[1-(1- α)1/3] [1-(1- α)1/3]2 With g(α) = -ln(1- α), Equation (6-13) can be rewritten as   RTEEART)ln( α)1ln(ln2  (6-14)    144  6.2 Thermogravimetric analysis The thermogravimetric (TG) and derivative thermogravimetric (DTG) curves of raw and torrefied SPF sawdust at heating rates of 10, 20, 30 and 40°C/min in a nitrogen flow of 100 ml/min are shown in Figures 6-1 to 6-9. Weight loss mainly occurred in the temperature ranges of 220-500°C and 280-550°C for the raw and torrefied sawdust, respectively. A lower weight loss was generally found for a higher heating rate, due to inefficient heat transfer during pyrolysis (Oyedun et al., 2014). Figures 6-2, 6-3 and 6-4 show TG and DTG curves for pyrolysis of 0.25-0.5, 0.5-1.0 and 1.0-2.0 mm torrefied sawdust respectively at different heating rates in the nitrogen atmosphere. The torrefied sawdust refers to the torrefied product captured by the cyclone for cases PS-1 to PS-9, whose detailed operating conditions are provided in section 4.2.2 of Chapter 4. For sawdust of a given particle size, sawdust torrefied at a higher temperature showed a lower TG weight loss, due to the sawdust having already lost more weight during the torrefaction process. Sawdust pyrolyzed at a higher heating rate lost less weight in the TGA, due to inefficient heat transfer to the sawdust particles (Oyedun et al., 2014). Comparison of the smallest, intermediate and coarsest torrefied particles reveals that the coarsest torrefied sawdust had the lowest TG weight loss at otherwise similar operating conditions. This is consistent with the fact that the larger sawdust particles lost more weight when they were torrefied in the SRSB reactor (See Table 4-7 of Chapter 4), associated with the larger particles having longer residence times in the reactor.  It is seen that the shoulders at 300-350°C observed in the DTG curves of raw sawdust were smoothed in the DTG curves of torrefied sawdust pyrolysis. This is likely due to the removal 145  of hemicellulose and extractives during the torrefaction process, resulting from hemicellulose decomposition, mainly occurring at 220-315°C (Yang et al., 2007). For the 0.25-0.5 mm sawdust torrefied at 240 and 270°C, the maximum decomposition rates were close to that of the raw sawdust, because the torrefaction carried out on 0.25-0.5 mm sawdust was light, as described in section 4.2 of Chapter 4. As shown in Figure 6-2(c), the maximum decomposition rate of the sawdust torrefied at 300°C increased, resulting from the accumulation of cellulose in the torrefied sawdust, as described in the fiber analysis shown in Table 4-8. For the torrefied product of 0.5-1.0 mm and 1.0-2.0 mm sawdust pyrolysis, as shown in Figures 6-3 and 6-4, when the sawdust was torrefied at 240 to 300°C, the maximum decomposition rate first increased, primarily because of cellulose accumulation, and then decreased as a consequence of partial cellulose decomposition at 300°C, consistent with previous findings (Cao et al., 2015). Note that a higher activation energy is required to decompose cellulose than hemicellulose and lignin (Zhou et al., 2015). Moreover, the gaps between different heating rates in the DTG curves were found to narrow for the sawdust torrefied at a higher temperature, revealing that more severe torrefaction can improve heat transfer from/to particles for pyrolysis at a higher heating rate.    146  100 200 300 400 500 600 700 800-40-20020406080100 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)(a)0.00.20.40.60.81.01.21.41.61.8DTG359C0.226%/s372C0.421%/s388C0.620%/s DTG (%/s)395C0.804%/sTG 100 200 300 400 500 600 700 800-40-20020406080100 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)(b)0.00.20.40.60.81.01.21.41.61.8DTG360C0.183%/s378C0.403%/s388C0.602%/s DTG (%/s)398C0.825%/sTG 100 200 300 400 500 600 700 800-40-20020406080100 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)0.00.20.40.60.81.01.21.41.61.8DTG360C0.229%/s374C0.444%/s387C0.620%/s DTG (%/s)396C0.806%/sTG(c)  Figure 6-1 Thermogravimetric (TG) and derivative thermogravimetric (DTG) curves for raw SPF sawdust at heating rates of 10, 20, 30 and 40°C/min: (a) 0.25-0.5 mm; (b) 0.5-1.0 mm; (c) 1.0-2.0 mm particles.  147  100 200 300 400 500 600 700 800-40-20020406080100379C0.411%/s359C0.176%/s389C0.625%/s397C0.821%/s 40C/min 30C/min 20C/min 10C/min TG (%)T(oC)TGDTG(a)0.00.20.40.60.81.01.21.41.61.8DTG (%/s) 100 200 300 400 500 600 700 800-40-20020406080100 (b) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)0.00.20.40.60.81.01.21.41.61.8359C0.188%/s378C0.382%/s390C0.578%/sDTG394C0.805%/sTGDTG (%/s) 100 200 300 400 500 600 700 800-40-20020406080100 (c) 40C/min 30C/min 20C/min 10C/min TG (%)T(oC)399C0.887%/s379C0.418%/s390C0.656%/s359C0.193%/sTGDTG0.00.20.40.60.81.01.21.41.61.8DTG (%/s)  Figure 6-2 TG and DTG curves for torrefied SPF sawdust of 0.25-0.5 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case PS-1: T = 240°C and F = 440 g/h; (b) case PS-2: T = 270°C and F = 440 g/h; (c) case PS-3: T = 300°C and F = 440 g/h. See Table 4-6 for detailed torrefaction operating conditions.  148  100 200 300 400 500 600 700 800-40-20020406080100 (a)364C0.216%/s381C0.434%/s390C0.648%/s 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TGDTG398C0.885%/s0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s) 100 200 300 400 500 600 700 800-40-20020406080100 (b)363C0.230%/s391C0.715%/s378C0.469%/s 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TGDTG399C0.941%/s0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s) 100 200 300 400 500 600 700 800-40-20020406080100 (c) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s)TG396C0.902%/s378C0.435%/s390C0.708%/s363C0.232%/sDTG  Figure 6-3 TG and DTG curves for torrefied SPF sawdust of 0.5-1.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case PS-4: T = 240°C and F = 440 g/h; (b) case PS-5: T = 270°C and F = 440 g/h; (c) case PS-6: T = 300°C and F = 440 g/h. See Table 4-6 for detailed torrefaction operating conditions.  149  100 200 300 400 500 600 700 800-40-20020406080100 (a) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)0.00.20.40.60.81.01.21.41.61.8376C0.410%/s361C0.224%/s394C0.860%/s388C0.679%/s DTG (%/s)TGDTG 100 200 300 400 500 600 700 800-40-20020406080100 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TG389C0.916%/s372C0.459%/s383C0.707%/s349C0.189%/sDTG(b)0.00.20.40.60.81.01.21.41.61.8DTG (%/s) 100 200 300 400 500 600 700 800-40-20020406080100 (c) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)0.00.20.40.60.81.01.21.41.61.8DTG (%/s)TG386C0.835%/s379C0.628%/s354C0.214%/s370C0.421%/sDTG  Figure 6-4 TG and DTG curves for torrefied SPF sawdust of 1.0-2.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case PS-7: T = 240°C and F = 440 g/h; (b) case PS-8:  T = 270°C and F = 440 g/h; (c) case PS-9: T = 300°C and F = 440 g/h. See Table 4-6 for detailed torrefaction operating conditions.    150  In Figures 6-3, 6-5 and 6-6 TG and DTG curves are plotted for pyrolysis of SPF sawdust of 0.5-1.0 mm particle size torrefied at different biomass feed rates and temperatures in the SRSB reactor. The thermal analysis tests were conducted on the torrefied product collected by the cyclone in cases C-1 to C-6 and C-8 to C-10. Detailed operating conditions can be found in Table 4-1 of Chapter 4. Based on the TG curves, it is observed that the weight loss during thermogravimetric analysis was greater for sawdust torrefied at a higher biomass feed rate. This suggests that the weight loss of biomass decreased with increasing biomass feed rate for biomass torrefied in the slot-rectangular spouted bed reactor, as discussed in Chapters 4 and 5.  For the sawdust torrefied at a given biomass feed rate, the weight loss of sawdust torrefied at higher temperature was less than that of sawdust torrefied at low temperature. In addition, it can be seen from the DTG curves that the maximum decomposition rate increased with increasing heating rate. Sawdust torrefied at higher temperatures had greater maximum decomposition rates, as a result of cellulose accumulation. However, sawdust torrefied at a higher biomass feed rate had a lower maximum decomposition rate, due to a shorter residence time in the reactor of sawdust torrefied at a higher biomass feed rate.   151  100 200 300 400 500 600 700 800-40-20020406080100 (a) TG (%)T(C) 40C/min 30C/min 20C/min 10C/min0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s)TG397C0.855%/s380C0.430%/s387C0.648%/s364C0.216%/s DTG 100 200 300 400 500 600 700 800-40-20020406080100 (b) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TG396C0.925%/s379C0.442%/s385C0.618%/s362C0.221%/sDTG0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s) 100 200 300 400 500 600 700 800-40-20020406080100 (c) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s)TG395C0.961%/s379C0.466%/s386C0.678%/s363C0.239%/s DTG  Figure 6-5 TG and DTG curves for torrefied SPF sawdust of 0.5-1.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case C-1: T = 240°C and F = 220 g/h; (b) case C-2: T = 270°C and F = 220 g/h; (c) case C-3: T = 300°C and F = 220 g/h. See Table 4-1 for detailed torrefaction operating conditions. 152  100 200 300 400 500 600 700 800-40-20020406080100 (a) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TG397C0.890%/s374C0.381%/s388C0.658%/s363C0.226%/sDTG0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s) 200 400 600 800-40-20020406080100 (b) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s)TG397C0.911%/s372C0.408%/s389C0.687%/s364C0.239%/sDTG 100 200 300 400 500 600 700 800-40-20020406080100 (c) 40C/min 30C/min 20C/min 10C/min TG (%)T(oC)0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s)TG395C0.940%/s376C0.434%/s389C0.703%/s359C0.207%/sDTG  Figure 6-6 TG and DTG curves for torrefied SPF sawdust of 0.5-1.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case C-8: T = 240°C and F = 710 g/h; (b) case C-9: T = 270°C and F = 710 g/h; (c) case C-10: T = 300°C and F = 710 g/h. See Table 4-1 for detailed torrefaction operating conditions.    153  Figures 6-7, 6-8 and 6-9 show TG and DTG curves for sawdust oxidatively torrefied at different heating rates. These particles were torrefied with different oxygen contents in the feed-gas and at different temperatures. Detailed information on the operating conditions is provided in Table 4-10 of Chapter 4. Sawdust torrefied at a given torrefaction temperature, but with a higher oxygen content in the feed-gas, had less weight loss based on the TG curve, revealing that introducing the oxygen in the feed-gas helped liberate the volatiles. Sawdust torrefied with a given oxygen content in the feed-gas, but at a higher temperature, had less weight loss.  Compared to the raw SPF sawdust, oxidatively torrefied sawdust had higher maximum decomposition rates. However, the non-oxidatively torrefied sawdust had higher maximum decomposition rates than the oxidatively torrefied sawdust. The reason is that oxygen in the feed-gas promoted decomposition of hemicellulose/cellulose and extractive during the torrefaction process, as discussed in Chapters 4 and 5 and consistent with previous work (Chen et al., 2014a; Chen et al., 2013a; Wang et al., 2013a). For the oxidatively torrefied sawdust torrefied at the same temperature, the maximum decomposition rate decreased with increasing O2 concentration in the feed-gas during the torrefaction. For the oxidatively torrefied sawdust torrefied at a given O2 concentration in the feed-gas, the maximum decomposition rate increased with increasing temperature during the torrefaction. These findings can be explained by increases in oxygen content in the feed-gas and increasing torrefaction temperature causing hemicellulose depletion and partial cellulose decomposition, so that more char was formed in the oxidatively torrefied sawdust.   154  100 200 300 400 500 600 700 800-40-20020406080100 (a) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TG398C0.850%/s379C0.416%/s391C0.640%/s364C0.212%/sDTG0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s) 100 200 300 400 500 600 700 800-40-20020406080100 (b) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TG396C0.812%/s374C0.383%/s387C0.620%/s362C0.201%/s DTG0.00.20.40.60.81.01.21.41.61.8DTG (%/s) 100 200 300 400 500 600 700 800-40-20020406080100 (c) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TG395C0.825%/s378C0.410%/s388C0.624%/s364C0.209%/sDTG0.00.20.40.60.81.01.21.41.61.8DTG (%/s)  Figure 6-7 TG and DTG curves for torrefied SPF sawdust of 0.5-1.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case OT-2: T = 240°C, F = 440 g/h and O2 conc.: 3 vol.%; (b) case OT-3: T = 240°C, F = 440 g/h and O2 conc.: 6 vol.%; (c) case OT-4: T = 240°C, F = 440 g/h and O2 conc.: 9 vol.%. See Table 4-10 for detailed torrefaction operating conditions. 155  100 200 300 400 500 600 700 800-40-20020406080100 (a) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TG397C0.886%/s378C0.446%/s386C0.626%/s363C0.221%/sDTG0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s) 100 200 300 400 500 600 700 800-40-20020406080100 (b) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TG396C0.875%/s379C0.440%/s390C0.664%/s362C0.221%/sDTG0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s) 100 200 300 400 500 600 700 800-40-20020406080100 (c) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TG396C0.876%/s379C0.438%/s390C0.650%/s364C0.225%/sDTG0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s)  Figure 6-8 TG and DTG curves for torrefied SPF sawdust of 0.5-1.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case OT-6: T = 270°C, F = 440 g/h and O2 conc.: 3 vol.%; (b) case OT-7: T = 270°C, F = 440 g/h and O2 conc.: 6 vol.%; (c) case OT-8: T = 270°C, F = 440 g/h and O2 conc.: 9 vol.%. See Table 4-10 for detailed torrefaction operating conditions.  156  100 200 300 400 500 600 700 800-40-20020406080100 (a) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TG397C0.904%/s380C0.446%/s389C0.683%/s364C0.233%/sDTG0.00.20.40.60.81.01.21.41.61.82.0DTG (%/s) 100 200 300 400 500 600 700 800-40-20020406080100 (b) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TG396C0.848%/s379C0.425%/s390C0.640%/s364C0.220%/sDTG0.00.20.40.60.81.01.21.41.61.8DTG (%/s) 100 200 300 400 500 600 700 800-40-20020406080100 (c) 40oC/min 30oC/min 20oC/min 10oC/min TG (%)T(oC)TG394C0.890%/s380C0.447%/s388C0.664%/s365C0.228%/sDTG0.00.20.40.60.81.01.21.41.61.8DTG (%/s)  Figure 6-9 TG and DTG curves for torrefied SPF sawdust of 0.5-1.0 mm particle size at heating rates of 10, 20, 30 and 40°C/min. Torrefaction conditions: (a) case OT-10: T = 300°C, F = 440 g/h and O2 conc.: 3 vol.%; (b) case OT-11: T = 300°C, F = 440 g/h and O2 conc.: 6 vol.%; (c) case OT-12: T = 300°C, F = 440 g/h and O2 conc.: 9 vol.%. See Table 4-10 for detailed torrefaction operating conditions.  157  6.3 Kinetic analysis 6.3.1 Kinetic parameters determined by Kissenger-Akahira-Sunose (KAS) model The thermogravimetric data were analyzed by the KAS model. Linear regression lines for raw SPF determined by the KAS method are shown in Figure 6-10(a). The activation energy can be determined from the slope of the fitted lines. The correlation coefficients (R2) are greater than 0.98, suggesting that the pyrolysis process is well correlated. As shown in Figure 6-10(b), the activation energy can be divided into three stages. The activation energy of raw SPF sawdust increased from 123 kJ/mol to 128 kJ/mol, then remained at ~128 kJ/mol, before increasing to 136 kJ/mol in the conversion ranges of 5-40%, 40-70% and 70-75%, respectively. The variation of the activation energy is likely related to different decomposition temperatures of the hemicellulose, cellulose and lignin components in the SPF sawdust (Yang et al., 2007; Zhou et al., 2015). In stage I where the corresponding temperature range is from 263 to 338°C, the increase in activation energy suggests that hemicellulose had a different activation energy, due to more C-C bond decomposition with increasing temperature. In stage II with corresponding temperatures from 338 to 366°C, cellulose decomposition is dominant (Zhou et al., 2015). It can be found that the cellulose content had its activation energy mainly related to C-C bond decomposition (Cao et al., 2016; Chen et al., 2011a). In stage III, where the corresponding temperature > 366°C, the activation energy shows an increasing trend, primarily due to the decomposition of chars in the biomass (Milosavljevic et al., 1996).  Lignin decomposition was not observed in stage III. This may be related to the high activation energy and difficult decomposition for large particles, whose 158  decomposition is controlled by the kinetics and by heat transfer (Cho et al., 2010; Guo and Lua, 2001).  0.0015 0.0016 0.0017 0.0018 0.0019 0.0020-10.8-10.6-10.4-10.2-10.0-9.8-9.6-9.4-9.2-9.0-8.8Note: Solid lines are fitted lines. 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% 65% 70% 75%ln(/T2)1/T (1/K)0.5-1.0mm Raw SPF(a) 0 10 20 30 40 50 60 70 80120122124126128130132134136138140142Stage IIIStage II 0.5-1.0mm Raw SPFE (kJ/mol)Conversion (%)(b)Stage I Figure 6-10 (a) Activation energy from KAS method at different conversions; (b) Activation energies for 0.5-1.0 mm raw SPF sawdust at different conversions. 159  Activation energies for the pyrolysis of different particle sizes of sawdust torrefied at 240-300°C and 440 g/h biomass feed rate are shown in Figures 6-11 (1)-(3). For a given operating condition, it is observed that the activation energy is lowest for the smallest torrefied sawdust particles. The main reason is that less torrefaction took place for the smallest sawdust particles, compared to the intermediate and coarsest sawdust. Moreover, the activation energy of the torrefied sawdust was found to be slightly higher when the sawdust was torrefied at a higher temperature, due to the hemicellulose depletion and cellulose accumulation in the torrefied sawdust. In addition, three distinct stages of activation energy - increment, constant and increment - were found for the pyrolysis of the different sizes of torrefied sawdust.  The activation energies for the pyrolysis of the sawdust torrefied at different biomass feed rates from 220 to 710 g/h are shown in Figures 6-11 (2), (4) and (5). A small difference was observed in the activation energies of the sawdust torrefied at different biomass feed rates with otherwise same operating conditions.  The activation energies for the pyrolysis of the sawdust torrefied at different oxygen concentrations in the feed-gas are compared in Figures 6-11(6)-(8). Compared to oxygen-free torrefied sawdust, pyrolysis of the sawdust torrefied in the presence of oxygen had lower activation energy, mainly resulting from decomposition and oxidation reactions occurring simultaneously during oxidative torrefaction. Figure 6-11(6) shows that the activation energy of the sawdust torrefied at 240°C first increased, then decreased, as the oxygen concentration in the feed-gas increased from 3 to 9 vol.%. However, the hemicellulose content consistently decreased, while the cellulose content increased, based on the fiber analysis shown in Table 4-12 of Chapter 4, suggesting that the activation energy should increase with increasing 160  oxygen concentration in the feed-gas. These inconsistent results indicate that the activation energy determined by the KAS model may include energies for both the pyrolysis reaction and volatile diffusion. In Figure 6-11 (7), the activation energy of the SPF sawdust torrefied at 270°C decreased slightly with increasing oxygen concentration in the feed-gas, consistent with the fiber analysis where the cellulose content decreased slightly, as shown in Table 4-12 of Chapter 4. In Figure 6-11 (8), the activation energy of the sawdust torrefied at 300°C increased with increasing oxygen concentration, especially at 9 vol.%. The fiber analysis shows that as the cellulose decreased from 52 to 46.5%, the lignin content increased from 29.2 to 32.7% in the oxidatively torrefied sawdust. This would lead to a decrease in the activation energy of the pyrolysis of the oxidatively torrefied sawdust. A reasonable explanation could be that the charring process intensified at 300°C in an atmosphere containing oxygen.  0 10 20 30 40 50 60 70 80 9075100125150175200   0.25-0.5 mm SPF 240C-440g/h 270C-440g/h 300C-440g/hE (kJ/mol)Conversion (%)(1) 0 10 20 30 40 50 60 70 80 9075100125150175200   0.5-1.0mm SPF 240C-440g/h 270C-440g/h 300C-440g/hE (kJ/mol)Conversion (%)(2) 161  0 10 20 30 40 50 60 70 80 9075100125150175200    1.0-2.0mm SPF 240C-440g/h 270C-440g/h 300C-440g/hE (kJ/mol)Conversion (%)(3) 0 10 20 30 40 50 60 70 80 9075100125150175200   0.5-1.0mm SPF 240C-220g/h 270C-220g/h 300C-220g/hE (kJ/mol)Conversion (%)(4) 0 10 20 30 40 50 60 70 80 9075100125150175200   0.5-1.0mm SPF 240C-710g/h 270C-710g/h 300C-710g/hE (kJ/mol)Conversion (%)(5) 0 10 20 30 40 50 60 70 80 9075100125150175200    0.5-1.0 mm SPF240oC-3% O2-440g/h270oC-3% O2-440g/h300oC-3% O2-440g/hE (kJ/mol)Conversion (%)(6) 0 20 40 60 80 10075100125150175200   0.5-1.0 mm SPF 270C-3% O2-440g/h  270C-6% O2-440g/h  270C-9% O2-440g/h E (kJ/mol)Conversion (%)(7) 0 20 40 60 80 10075100125150175200    0.5-1.0 mm SPF300C-3% O2-440g/h 300C-6% O2-440g/h 300C-9% O2-440g/h E (kJ/mol)Conversion (%)(8) Figure 6-11 Activation energies from KAS method for torrefied SPF sawdust as a function of conversions, temperature, biomass feed rate and oxygen concentration. 162  6.3.2 Kinetic parameters determined by Coats-Redfern model 0.0015 0.0016 0.0017 0.0018 0.0019 0.0020 0.0021 0.0022-18-17-16-15-14-13Experimental point Linear Fity=-8353.2x+0.14819R2=0.997ln(-ln(1-α)/T^2)1/T (1/K)(A) 10C/min 0.0015 0.0016 0.0017 0.0018 0.0019 0.0020 0.0021 0.0022-18-17-16-15-14-13 Experimental point Linear Fity=-8400.8x+0.19416R2=0.998ln(-ln(1-α)/T^2)1/T (1/K)(B) 20C/min 0.0015 0.0016 0.0017 0.0018 0.0019 0.0020 0.0021 0.0022-18-17-16-15-14-13 Experimental point  Linear Fity=-8295.7x+0.61433R2=0.995ln(-ln(1-α)/T^2)1/T (1/K)(C) 30C/min 0.0014 0.0015 0.0016 0.0017 0.0018 0.0019 0.0020 0.0021 0.0022-18-17-16-15-14-13 Experimental point  Linear Fity=-8153.9x-1.00115R2=0.995ln(-ln(1-α)/T^2)1/T (1/K)(D) 40C/min Figure 6-12 Coats-Redfern plots for raw 0.5-1.0 mm SPF particles at different heating rates. The thermogravimetric data from Ti to Tpeak were also analyzed by the Coats-Redfern method. Ti, the initial decomposition temperature, corresponds to the temperature for 0.5% weight conversion of raw or torrefied sawdust on a dry basis; Tpeak corresponds to the maximum decomposition rate in the DTG curve. Figure 6-12 shows linear regression lines based on the Coats-Redfern method for raw 0.5-1.0 mm SPF sawdust at different heating rates. It is seen that R2 was higher than 0.994 for the fitted linear lines. The average activation energy and 163  pre-exponential factor are presented in Table 6-2. The activation energy based on the Coats-Redfern model is around 68-90 kJ/mol, significantly lower than determined by the KAS model (see Figure 6-10 and Figure 6-11). The reason is that the KAS model includes the activation energies for both the decomposition reaction and diffusion in the pyrolysis process, while the Coats-Redfern model only covers the decomposition reaction during the pyrolysis process.  The pre-exponential factor for the Coats-Redfern model is around 97-138 s-1. The activation energy decreased slightly as the heating rate increased from 10 to 40°C/min.  Table 6-2 Kinetic parameters for raw and torrefied SPF sawdust pyrolysis obtained by Coats-Redfern model. For operating conditions, see Table 4-1, Table 4-6, and Table 4-10.  Heating rate (β) 10 °C/min 20 °C/min 30 °C/min 40 °C/min #1 Raw SPF (0.25-0.5mm) Ti (°C) 210 210 210 210 Tpeak (°C) 359 377 388 395 E (kJ/mol) 75.1 77.4 77.5 79.8 A (s-1) 4.11E+02 9.71E+02 1.17E+03 2.23E+03 R2 0.998 0.999 0.999 0.998 #2 Raw SPF (0.5-1.0mm) Ti (°C) 210 210 210 210 Tpeak (°C) 360 378 388 398 E (kJ/mol) 69.5 69.8 69.0 67.8 A (s-1) 9.69E+01 1.38E+02 1.35E+02 1.20E+02 R2 0.997 0.998 0.995 0.995 #3 Raw SPF (1.0-2.0mm) Ti (°C) 210 210 210 210 Tpeak (°C) 360 378 387 396 E (kJ/mol) 79.8 82.3 86.4 79.2 A (s-1) 1.03E+03 2.37E+03 6.86E+03 1.98E+03 R2 0.999 1.00 0.998 0.993           164  Heating rate (β) 10 °C/min 20 °C/min 30 °C/min 40 °C/min #4 PS-1: (0.25-0.5mm)-240°C-440g/h Ti (°C) 250 250 250 250 Tpeak (°C) 359 379 389 397 E (kJ/mol) 79.9 83.3 87.5 89.0 A (s-1) 9.82E+02 2.26E+03 6.07E+03 8.99E+03 R2 1.00 0.999 0.998 0.993 #5 PS-2: (0.25-0.5mm)-270°C-440g/h Ti (°C) 250 250 250 250 Tpeak (°C) 359 378 390 394 E (kJ/mol) 84.6 81.93 83.9 84.1 A (s-1) 2.28E+03 1.71E+03 2.93E+03 3.19E+03 R2 0.999 1.00 1.00 1.00 #6 PS-3: (0.25-0.5mm)-300°C-440g/h Ti (°C) 250 250 250 250 Tpeak (°C) 359 379 390 399 E (kJ/mol) 90.2 89.5 95.7 96.4 A (s-1) 7.40E+03 7.10E+03 2.82E+04 3.28E+04 R2 0.999 1.00 0.999 0.999 #7 PS-4: (0.5-1.0mm)-240°C-440g/h Ti (°C) 250 250 250 250 Tpeak (°C) 364 381 390 398 E (kJ/mol) 87.8 90.4 93.4 95.7 A (s-1) 4.23E+03 8.97E+03 1.91E+04 2.92E+04 R2 0.999 0.999 0.999 0.999 #8 PS-5: (0.5-1.0mm)-270°C-440g/h Ti (°C) 250 250 250 250 Tpeak (°C) 363 378 391 399 E (kJ/mol) 92.0 113.3 118. 9 99.3 A (s-1) 8.37E+03 8.25E+05 2.62E+06 5.15E+04 R2 1.00 0.991 0.985 1.00 #9 PS-6: (0.5-1.0mm)-300°C-440g/h Ti (°C) 250 250 250 250 Tpeak (°C) 363 378 390 396 E (kJ/mol) 94.2 96.7 94.1 100.0 A (s-1) 1.25E+04 2.49E+04 1.61E+04 5.65E+04 R2 1.00 1.00 0.998 0.998      165  Heating rate (β) 10 °C/min 20 °C/min 30 °C/min 40 °C/min #10 PS-7: (1.0-2.0mm)-240°C-440g/h Ti (°C) 250 250 250 250 Tpeak (°C) 361 376 388 394 E (kJ/mol) 109.2 102.4 114.3 104.4 A (s-1) 3.07E+05 8.78E+04 1.05E+06 1.43E+05 R2 0.997 0.998 0.998 0.996 #11 PS-8: (1.0-2.0mm)-270°C-440g/h Ti (°C) 250 250 250 250 Tpeak (°C) 349 372 383 389 E (kJ/mol) 95.1 89.4 91.5 79.9 A (s-1) 1.24E+04 2.61E+03 6.74E+03 6.58E+02 R2 0.980 0.976 0.973 0.945 #12 PS-9: (1.0-2.0mm)-300°C-440g/h Ti (°C) 250 250 250 250 Tpeak (°C) 354 370 379 386 E (kJ/mol) 88.3 91.1 87.2 85.1 A (s-1) 2.52E+03 5.56E+03 2.14E+03 2.45E+03 R2 0.978 0.979 0.971 0.971 #13 C-1: (0.5-1.0mm)-240°C-220g/h Ti (°C) 250 250 250 250 Tpeak (°C) 364 380 387 397 E (kJ/mol) 83.9 86.5 88.5 88.7 A (s-1) 1.53E+03 3.23E+03 5.40E+03 5.91E+03 R2 1.00 0.999 0.999 1.00 #14 C-2: (0.5-1.0mm)-270°C-220g/h Ti (°C) 250 250 250 250 Tpeak (°C) 362 379 385 396 E (kJ/mol) 89.9 93.2 95.3 95.6 A (s-1) 4.96E+03 1.14E+04 2.01E+04 2.08E+04 R2 1.00 1.00 1.00 1.00 #15 C-3: (0.5-1.0mm)-300°C-220g/h Ti (°C) 250 250 250 250 Tpeak (°C) 363 379 386 395 E (kJ/mol) 87.7 102.1 85.3 93.3 A (s-1) 3.14E+03 7.29E+04 2.49E+03 1.40E+04 R2 0.996 1.00 0.986 0.997      166  Heating rate (β) 10 °C/min 20 °C/min 30 °C/min 40 °C/min #16 C-8: (0.5-1.0mm)-240°C-710g/h Ti (°C) 250 250 250 250 Tpeak (°C) 363 374 388 397 E (kJ/mol) 92.2 97.3 101. 101.3 A (s-1) 1.03E+04 3.58E+04 8.35E+04 8.89E+04 R2 1.00 0.999 0.999 0.999 #17 C-9: (0.5-1.0mm)-270°C-710g/h Ti (°C) 250 250 250 250 Tpeak (°C) 364 372 389 397 E (kJ/mol) 94.0 96.2 99.7 99.4 A (s-1) 1.33E+04 2.62E+04 5.55E+04 5.41E+04 R2 1.00 0.999 0.999 0.998 #18 C-10: (0.5-1.0mm)-300°C-710g/h Ti (°C) 250 250 250 250 Tpeak (°C) 359 376 389 395 E (kJ/mol) 97.1 108.7 101.2 97. 6 A (s-1) 2.21E+04 2.93E+05 7.35E+04 3.65E+04 R2 1.00 0.998 0.999 0.998 #19 OT-1: (0.5-1.0mm)-240°C-440g/h-3vol.% O2 Ti (°C) 250 250 250 250 Tpeak (°C) 364 379 391 398 E (kJ/mol) 84.2 96.9 88.5 87.9 A (s-1) 2.02E+03 3.53E+04 6.97E+03 6.59E+03 R2 0.999 0.993 0.999 1.00 #20 OT-2: (0.5-1.0mm)-240°C-440g/h-6vol.% O2 Ti (°C) 250 250 250 250 Tpeak (°C) 362 374 387 396 E (kJ/mol) 108.3 103.3 115.7 100.6 A (s-1) 8.97E+05 3.91E+05 5.62E+06 2.79E+05 R2 0.999 1.00 0.998 0.997 #21 OT-3: (0.5-1.0mm)-240°C-440g/h-9vol.% O2 Ti (°C) 250 250 250 250 Tpeak (°C) 364 378 388 395 E (kJ/mol) 85.4 90.8 93.8 87.2 A (s-1) 2.46E+03 9.86E+03 2.16E+04 5.53E+03 R2 0.999 0.998 0.998 1.00      167  Heating rate (β) 10 °C/min 20 °C/min 30 °C/min 40 °C/min #22 OT-4: (0.5-1.0mm)-270°C-440g/h-3vol.% O2 Ti (°C) 250 250 250 250 Tpeak (°C) 363 378 386 397 E (kJ/mol) 103.0 97.4 104.3 98.4 A (s-1) 9.41E+04 3.54E+04 1.54E+05 4.80E+04 R2 0.995 0.999 0.998 0.999 #23 OT-5: (0.5-1.0mm)-270°C-440g/h-6vol.% O2 Ti (°C) 250 250 250 250 Tpeak (°C) 362 379 390 396 E (kJ/mol) 88.8 95. 100.9 93.8 A (s-1) 5.04E+03 2.30E+04 8.14E+04 2.11E+04 R2 0.999 0.999 0.997 1.00 #24 OT-6: (0.5-1.0mm)-270°C-440g/h-9vol.% O2 Ti (°C) 250 250 250 250 Tpeak (°C) 364 379 390 396 E (kJ/mol) 90.6 106. 94.0 110.5 A (s-1) 7.21E+03 2.19E+05 2.00E+04 5.74E+05 R2 1.00 0.991 0.999 0.993 #25 OT-7: (0.5-1.0mm)-300°C-440g/h-3vol.% O2 Ti (°C) 250 250 250 250 Tpeak (°C) 364 380 389 397 E (kJ/mol) 91.1 85.9 94.0 102.6 A (s-1) 7.25E+03 3.06E+03 1.80E+04 1.04E+05 R2 0.999 0.993 0.998 1.00 #26 OT-8: (0.5-1.0mm)-300°C-440g/h-6vol.% O2 Ti (°C) 250 250 250 250 Tpeak (°C) 364 379 390 396 E (kJ/mol) 84.0 87.6 94.7 102.1 A (s-1) 1.56E+03 4.27E+03 1.96E+04 9.11E+04 R2 0.998 0.999 1.00 0.999 #27 OT-9: (0.5-1.0mm)-300°C-440g/h-6vol.% O2 Ti (°C) 250 250 250 250 Tpeak (°C) 365 380 388 394 E (kJ/mol) 91.4 95.2 93.6 100.6 A (s-1) 7.10E+03 2.04E+04 1.68E+04 6.95E+04 R2 0.998 0.999 1.00 0.997 168  Compared to the raw SPF sawdust, the torrefied sawdust had a higher activation energy and a higher pre-exponential factor, as a consequence of the removal of hemicellulose and extractives during the torrefaction process (Oyedun et al., 2014; Yang et al., 2007).  For 0.25-0.5 mm sawdust torrefied at 240-300°C and 440 g/h feed rate (#4-#6 in Table 6-2), it is seen that the activation energy of torrefied sawdust increased with increasing torrefaction temperature. For 0.5-1.0 mm sawdust torrefied at 240-300°C and 440 g/h feed rate (#7-#9 in Table 6-2), the activation energy of torrefied sawdust increased as the torrefaction temperature increased from 240 to 270°C. However, the activation energy dropped at 300°C, consistent with the decrement of cellulose content in the fiber analysis shown in Table 4-15. This may be because the kinetic parameters produced by the Coats-Redfern method are mainly determined from the biomass decomposition stage which tends to be less affected by the high chars content in stage III.  For 1.0-2.0 mm sawdust torrefied at 240-300°C and 440 g/h feed rate (#10-#12 in Table 6-2), the activation energy was observed to decrease with increasing torrefaction temperature, due to decomposition of cellulose and accumulation of lignin in the torrefied sawdust, consistent with the fiber analysis in Table 4-8.  For sawdust torrefied at the same feed rate of 440 g/h, the activation energy of the sawdust torrefied at 240 and 270°C increased with increasing particle size, because the decomposition was controlled by the kinetics and heat transfer mechanism (Guo and Lua, 2001). However, the activation energy of the coarsest sawdust torrefied at 300°C is lowest among the smallest, intermediate and coarsest torrefied sawdust particles at otherwise similar operating conditions. The main reason is believed to be that the larger particles underwent more severe torrefaction, with some cellulose decomposed, as discussed in section 4.2 of Chapter 4. 169  For 0.5-1.0 mm sawdust torrefied at different biomass feed rates (#7-#9 and #13-#18 in Table 6-2), the activation energy of sawdust torrefied at a biomass feed rate of 220 g/h was lowest, and increased with increasing biomass feed rate. There are several reasons for this finding: (i) Torrefaction was more severe at the lowest biomass feed rate, leading to hemicellulose depletion, partial cellulose decomposition and lignin accumulation. (ii) The particle size of sawdust torrefied at 200 g/h feed rate was smaller than for sawdust torrefied at 220, 440 and 710 g/h feed rates, as shown in Figure 4-8, due to improved grindability of the torrefied biomass and substantial attrition in spouted beds. The smaller particles have larger surface areas, leading to greater heat transfer. In addition, for the sawdust torrefied at 440 and 710 g/h biomass feed rates, the activation energy of the torrefied sawdust increased with increasing torrefaction temperature, revealing that modest torrefaction occurred on the SPF sawdust at 440 and 710 g/h biomass feed rate. However, the activation energy of the sawdust torrefied at 220 g/h biomass feed rate first increased and then decreased with increasing torrefaction temperature, suggesting more severe torrefaction of that sawdust.  The activation energy and pre-exponential factor of the oxidatively torrefied sawdust appear also in Table 6-2 (#19 to #27). For sawdust torrefied at 240°C with different oxygen contents in the feed-gas, the activation energy of the oxidatively torrefied sawdust first increased, and then decreased, as the feed gas oxygen concentration increased from 3 to 9 vol.%. The increase in activation energy is due to accumulation of cellulose in the torrefied sawdust, whereas its subsequent decrease is related to decomposition of cellulose and accumulation of lignin in the torrefied sawdust during oxidative torrefaction. Note that the activation energy of cellulose pyrolysis is much higher than those of hemicellulose and lignin pyrolysis (Zhou et al., 2015). For sawdust torrefied at 270 and 300°C with different oxygen contents in the 170  feed-gas, the activation energy of oxidatively torrefied sawdust pyrolysis had an opposite trend, compared to pyrolysis of sawdust torrefied at 240°C. The activation energy decreased first and then increased as the oxygen content in the feed gas increased from 3 to 9 vol.%. This suggests that the activation energy of oxidatively torrefied sawdust is sensitive to the oxygen content of the feed-gas. Meanwhile, the decrease in activation energy is caused by decomposition of cellulose due to the presence of oxygen. The increase in the activation energy may be related to char production during the oxidative torrefaction experiments, because the activation energy of char pyrolysis is much higher than for lignin pyrolysis (Peng et al., 2012a). 6.4 Conclusions (1) Torrefied sawdust lost less weight than untreated raw sawdust in the pyrolysis process. (2) The activation energy of the torrefied sawdust was higher than that of the untreated sawdust. In addition, the activation energy of non-oxidatively torrefied sawdust was higher than for oxidatively torrefied sawdust. (3) The activation energy determined by the KAS model was much higher than that derived from the Coats-Redfern model. Activation energies determined by means of the Kissenger-Akahira-Sunose (KAS) model were 122-137 kJ/mol, 95-146 kJ/mol and 110-164 kJ/mol in the conversion range of 5-75 wt.% for raw, non-oxidatively torrefied and oxidatively torrefied sawdust pyrolysis, respectively. On the other hand, activation energies from the Coats-Redfern model were 68-90 kJ/mol, 80-119 kJ/mol and 84-110 kJ/mol for raw, torrefied and oxidatively torrefied sawdust pyrolysis, respectively. 171  (4) The Coats-Redfern model gave more information on variation of the composition during the pyrolysis process than the KAS model.   172  Chapter 7 Conclusions and Recommendations 7.1 Comparison of torrefaction reactors Table 7-1 lists different types of reactor applied in torrefaction. These reactors have included fixed beds, fluidized bed, screw conveyor, rotary drum, rotary kiln, electronic furnace, microwave reactor and the single/dual-compartment slot-rectangular spouted bed from the present study. Reactors have operated in batch or continuous mode. Each reactor had its own characteristics, and different studies utilized quite different operating conditions, such as torrefaction temperature, reaction time, material feed rate, type of material, particle size and oxygen concentration, as summarized in Table 7-1. It is therefore extremely difficult to compare results and to judge whether one torrefaction technology is better than the others.  In terms of the torrefied product properties shown in Table 7-1, the previous literature studied with different reactors and operating conditions show similar trends as the present study. For example, volatile matter and oxygen content decrease, whereas fixed carbon, higher heating value (HHV) and carbon content increase to a certain extent after torrefaction in all cases. Furthermore, the torrefied product properties from the present study are similar to those from the literature, and also similar to the properties of lignite. Overall, based on the work of this thesis, the slot-rectangular spouted bed provides a potential reactor configuration to produce high-quality torrefied biomass. However, comparative studies are needed with the same fuel and feed rates to allow full comparison of this type of reactor with the other types identified in Table 7-1.  173  Table 7-1 Comparison of torrefaction using different reactors and operating conditions, and key properties of torrefied product (on a dry basis). Peng et al. (2013a) Reactor:  Fixed bed (Batch) Material:  Spruce, pine and fir (SPF) shaving pine sawdust dp range: 0.79 and 3.18 mm (for sawdust) and 4 mm (for shaving) T range: 240-300°C Reaction time range:  60 min O2 conc. Range:  0vol.% Properties  Untreated material  Torrefied product VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) 93.0 6.8 49.5-51.2 42.2-43.9 18.1 N/A N/A 63.8-84.5 15.3-35.8 52.4-71.9 22.7-41.3 20.9-25.3 63.7-90.9 49.3-86.7 Nachenius et al. (2015) Reactor: screw conveyor (Continuous)  Material:  Pine dp range:  <4 mm T range: 275-375°C (reactor wall temperature) Reaction time range:  5-15 min O2 conc. Range:  0vol.% Properties  Untreated material  Torrefied product VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) 83.6 17-34 48.8 45.0 20.37 N/A N/A 60-80 17-34 49-63 44-30 20.5-27.5 66-98 66-95 Strandberg et al. (2015) Reactor: Rotary drum (Continuous) Material:  Norway spruce chip dp range:  8x8x4.5-25x25x4.5 (width x length x thickness) mm T range: 260-310°C  Reaction time range: 8-25 min Properties  Untreated material  Torrefied product VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) 85.0 14.6 50.4 43.2 20.4 51.5-84.0 15.7-47.8 51.4-69.2 25.0-42.3 20.7-27.8 174  O2 conc. Range: 0vol.% Energy yield (%) Solid yield (wt.%) N/A N/A 62-99 46-97 Wang et al. (2017a) Reactor: Fixed bed (Batch)  Material:  Norway Spruce dp range:  10-70mm T range: 225-300°C  Reaction time range: 30-60 min O2 conc. Range: 0vol.% Properties  Untreated material  Torrefied product VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) 74.85-88.12 11.6-23.0 47.4-49.1 44.4-46.0 19.5-20.1 N/A N/A 47.9-87.9 11.8-47.9 49.2-68.9 27.2-44.5 19.8-24.4 55.6-96.9 46.2-92.7 Martín-Lara et al. (2017) Reactor: Electric muffle furnace (Batch) Material:  Olive tree pruning dp range: 0.25-1mm   T range: 200, 300°C  Reaction time range: 10, 60 min O2 conc. Range: 0vol.% Properties  Untreated material  Torrefied product VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) 79.1 18.2 45.7 46.7 17.32 N/A N/A 20-26.2 20.9-27.6 48.4-53.3 38.9-43.6 19.4-20.5 64.0-101.5 57.0-87.4 Colin et al. (2017) Reactor: Rotary kiln (Continuous) Material: Beech wood chip dp range:  5-15 mm in length, 2-7 mm in width, 1-3 mm in thickness T range: 250-300°C  Properties  Untreated material  Torrefied product VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) 85 14.5 47.2 46.7 19.5 72.2-83.4 15.8-27.1 43.4-54.0 39.7-45.7 19.8-21.6 175  Reaction time range:21.8-66.7 min O2 conc. Range: 0vol.% Energy yield (%) Solid yield (wt.%) N/A N/A 83.2-100 75-98.3 Wang et al. (2013a) Reactor:  Fluidized bed (Semi-batch) Material: Spruce, pine and fir sawdust   dp range:  250-355 μm, 0.23 mm T range: 250-290°C  Reaction time range:4-42 min O2 conc. Range: 0-6vol.% Properties  Untreated material  Torrefied product VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) N/A N/A 46.5 45.4 18.9 N/A N/A N/A N/A 50.5-53.5 40.6-44.0 20.8-21.8 79-72 64-70 Chen et al. (2014a) Reactor: Electrical furnace (Batch) Material:  Oil palm, coconut, eucalyptus, cryptomeria japonica dp range:  <30 mm (Oil palm fiber, coconut fiber), 15x10x5 mm (Eucalyptus, cryptomeria japonica) T range: 300°C  Reaction time range: 60 min O2 conc. Range: 0-21vol.% Properties  Untreated material  Torrefied product VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) 72.5-81.1 18.9-20.5 50.8-51.9 40.6-44.1 17.1-18.6 N/A N/A 33.0-56.0 44.0-62.0 66.6-76.1 20.9-28.8 18.6-28.2 20-78 18-58 Joshi et al. (2015) Reactor: packed-bed (Batch) Material:  Sugar cane bagasse dp range:  N/A T range: 250-310°C  Properties  Untreated material  Torrefied product VM (wt.%) FC (wt.%) N/A N/A N/A N/A 176  Reaction time range: 45 min O2 conc. Range: 0-10vol.% C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) N/A N/A 19.4 N/A N/A N/A N/A 18.4-21.2 37-88 37-89 Uemura et al. (2017)    Reactor: Fixed bed (Batch) Material:  oil palm empty fruit bunches dp range:  20 x 5 x 10 mm T range: 200-300°C  Reaction time range: ~30 min O2 conc. Range: 0, ~8.5vol.% (CO: ~13vol.%) Properties  Untreated material  Torrefied product VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) 84 N/A 41 37 15.2 N/A N/A 70-83 3-13 48-57 21-35 <22.6 60-91 40-72 Huang et al. (2017) Reactor: Microwave reactor (Batch) Material:  Sewage sludge, Leucaena dp range:  N/A T range: 168-404°C  Reaction time range: N/A O2 conc. Range: 0vol.% Properties  Untreated material  Torrefied product VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) 52.3 (78.8) 18.5 (18.7) 51.6 (47.9) 31.4 (38.3) 15.0 (19.3) N/A N/A 1.9-37.0 (13.7-62.8) 25.7-32.5 (35.8-81.2) 52.6-66.8 (58.7-80.9) 27.2-33.2 (15.9-35.5) 10.3-16.2 (23.5-29.7) 18-82 (23-79) 9-72 (17-59) Note: Brackets indicates properties of Leucaena. Present work  Reactor: Single-compartment slot-rectangular spouted bed Properties Untreated material  Torrefied product 177  (Semi-batch) Material:  Spruce, pine and fir mixture sawdust dp range: 0.25-2 mm T range: 240-300°C  Reaction time range: N/A (220-710g/h feed rate) O2 conc. Range: 0vol.% VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) 84.6 15.1 46.2 47.4 18.2 N/A N/A 65.2-83.6 15.9-34.1 50.0-53.2 40.7-43.9 19.7-22.8 67-99 61-93 Reactor: Single-compartment slot-rectangular spouted bed (Semi-batch) Material:  Spruce, pine and fir mixture sawdust dp range: 0.5-1.0 mm T range: 240-300°C  Reaction time range: N/A (440 g/h feed rate) O2 conc. Range: 3-9vol.% Properties Untreated material  Torrefied product VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) 84.6 15.1 46.2 47.4 18.2 N/A N/A 55.5-83.4 15.8-43.6 49.8-52.3 41.6-43.9 19.8-22.9 61.2-94.2 54-85 Reactor: Dual-compartment slot-rectangular spouted bed (Semi-batch) Material:  Spruce, pine and fir mixture sawdust dp range: 0.5-1.0 mm T range: 240-300°C  Reaction time range: N/A (900-1400 g/h) O2 conc. Range: 0 vol.% Properties Untreated material  Torrefied product VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) 84.6 15.1 46.2 47.4 18.2 N/A N/A 60.9-83.5 17.0-38.2 50.7-52.1 43.0-41.9 20.0-24.4 73.3-97.6 63-90 Reactor: Dual-compartment slot-rectangular spouted bed (Semi-Properties Untreated material  Torrefied product 178  batch) Material:  Spruce, pine and fir mixture sawdust dp range: 0.5-1.0 mm T range: 240-300°C  Reaction time range: N/A (900 g/h feed rate) O2 conc. Range: 3-9vol.% VM (wt.%) FC (wt.%) C (wt.%) O (wt.%) HHV (MJ/kg) Energy yield (%) Solid yield (wt.%) 84.6 15.1 46.2 47.4 18.2 N/A N/A 51.7-83.5 16.1-47.5 50.7-54.4 40.1-43.2 19.3-23.5 65.7-97.6 56-89 Note: dp: Particle diameter, T: temperature, VM: Volatile matter, FC: Fixed carbon, C: Carbon, O: Oxygen, HHV: Higher heating value, N/A: Not available.  7.2 Conclusions from this thesis Key findings of this work were as follows: (1) In a cold model dual-compartment slot-rectangular spouted bed (DSRSB) with a suspended partition: • Glass beads of diameter 1.16, 1.61 and 2.85 mm were used to investigate solids mixing in the DSRSB. Comparison showed that exchange of larger particles between compartments reached an equilibrium composition in less time than for smaller particles at otherwise similar operating conditions.  • The effect of the partition position on solids exchange between the compartments was very important, with a larger gap leading to faster exchange. A shorter partition also accelerated solids exchange between the compartments by allowing some fountain particles to pass over the partition.  179  • The superficial gas velocity and the static bed height had only a slight influence on the solids exchange for the range of conditions investigated.  (2) Biomass torrefaction in single and dual-compartment slot-rectangular spouted bed reactors: • The slot-rectangular spouted bed provides a promising reactor configuration to produce high-quality torrefied biomass. • Pressure drop across the bed oscillated significantly during torrefaction. Moreover, in the DSRSB reactor, the pressure drops across the two compartments were unequal during the torrefaction process in all cases, indicating that the downstream column contained a higher inventory of biomass and inert particles. At similar operating conditions, the pressure drop across the downstream column of the DSRSB reactor was higher than that across the SRSB reactor, while the pressure drop across the upstream column of the DSRSB reactor was similar to the SRSB pressure drop. • Compared to untreated SPF sawdust, torrefied SPF sawdust generally had greater higher heating value (HHV), greater carbon content, lower atomic hydrogen and oxygen contents, less volatile content, greater fixed carbon content, less hemicellulose and smaller particle size. Comparison of torrefied sawdust from the cyclone and reactor revealed that the torrefied sawdust from the reactor underwent more severe torrefaction and had higher HHV, greater fixed carbon and lower volatile content. Based on SEM images, the torrefied sawdust particles had smoother surfaces than the raw biomass. Cracks could be clearly seen on the surface of torrefied particles after undergoing severe torrefaction. 180  • Temperature was one of the most important properties affecting the torrefaction experiments. An increase in temperature from 240 to 300°C resulted in more severe torrefaction and more hemicellulose decomposition, leading to increased weight loss and decreased energy yield of sawdust particles. As temperature increased, fewer torrefied sawdust particles were left in the reactor, while more torrefied particles were captured by the cyclone of the torrefaction facility. Increasing the temperature can significantly enhance the HHV of sawdust. More carbon monoxide was formed than carbon dioxide as the temperature increased.  • In the single-compartment SRSB reactor, the biomass feed rate was tested up to 710 g/h, while it was tested up to 1400 g/h in the DSRSB reactor. The biomass feed rate was directly linked to the residence time of biomass particles, with higher feed rate leading to shorter residence time. As the biomass feed rate increased, the weight loss in the form of volatiles decreased, the proportion of torrefied sawdust remaining in the reactor decreased, the mass percentage in the cyclone increased, but the mass percentage of torrefied sawdust in a downstream filter was hardly affected. Increasing the feed rate resulted in lower HHV, lower atomic carbon content, higher atomic hydrogen and oxygen contents, more volatile content, less fixed carbon content, and less decomposition of hemicellulose in the torrefied sawdust. • When torrefying biomass in slot-rectangular spouted bed reactors, biomass particle size was an important variable due to varying conditions experienced by reaction intermediates. Sawdust particles of three size ranges (0.25-0.5, 0.5-1.0 and 1.0-2.0 mm) were torrefied in a semi-batch SRSB facility. At the same operating conditions of temperature, biomass feed rate and superficial gas velocity, the smallest particles 181  had the shortest residence time. As the particle size increased, more torrefied particles remained in the reactor while fewer torrefied particles were captured by the cyclone. For the 1.0-2.0 mm size fraction of sawdust, 49-69 wt.% of total fed biomass stayed in the SRSB reactor, requiring downstream separation processing to recover torrefied sawdust. An increase in particle size led to increased weight loss and decreased energy yield of sawdust, resulting in a torrefied sawdust with higher HHV, greater atomic carbon content, lower atomic hydrogen and oxygen contents, less volatile content, greater fixed carbon content and less hemicellulose. In addition, sawdust particles underwent significant particle size reduction during torrefaction in the SRSB facility, with a reduction in Sauter mean particle size of up to 20.3% and 28.2% for the 0.25-0.5 mm and 0.5-1.0 mm sawdust size fractions, respectively. • It was demonstrated that some oxygen can be present in the torrefaction carrier gas in slot-rectangular spouted bed reactors, significantly reducing operating and equipment costs. Performance of oxidative torrefaction was similar to that of non-oxidative torrefaction. It is seen that the oxygen-containing atmosphere led to higher weight loss of biomass, but lower energy yield, due to biomass oxidation and decomposition occurring simultaneously in the oxidative torrefaction. Compared to non-oxidatively torrefied sawdust, oxidatively torrefied sawdust has a slightly lower HHV. Adding oxygen to the feed-gas decreased the product mass in the reactor, increased the product mass in the cyclone, and slightly increased the mass captured by the filter. Variation of oxygen concentration had a greater effect on biomass torrefaction at a higher temperature. An increase in oxygen concentration resulted in greater HHV, higher atomic carbon content, lower atomic hydrogen and oxygen contents, less 182  volatile content, greater fixed carbon content, and less decomposition of hemicellulose in the torrefied sawdust. SEM images show that the oxidatively torrefied particles had cleaner and smoother surfaces compared to non-oxidatively torrefied particles.  • In order to avoid thermal runaway in reactors, the oxygen concentration should not exceed 9 vol.% in the feed gas at 300°C when biomass is torrefied in a SRSB reactor. Moreover, it should not exceed 6 vol.% and 3 vol.% in the feed gas at 270 and 300°C respectively when biomass is torrefied in a DSRSB reactor.  • Sawdust torrefaction performed better in the dual-compartment slot-rectangular spouted bed reactor than in the single-compartment slot-rectangular spouted bed reactor.  • Activation energies determined by Kissenger-Akahira-Sunose (KAS) model were 122-137 kJ/mol, 95-146 kJ/mol and 110-164 kJ/mol in the conversion range of 5-75 wt.% for raw, non-oxidatively torrefied and oxidatively torrefied sawdust pyrolysis, respectively. Activation energies determined by Coats-Redfern model were much lower: 68-90 kJ/mol, 80-119 kJ/mol and 84-110 kJ/mol for raw, non-oxidatively torrefied and oxidatively torrefied sawdust pyrolysis, respectively.  7.3 Recommendations for future work In order to reduce the carrier gas consumption, future research should focus on:  1) Employ less dense inert particles, leading to a lower minimum spouting velocity; 2) Add a mean of removing particles from the dense bed in order that continuous (better than semi-batch) operation could be investigated.  183  3) Increase the recycling of torgas as much as possible.  4) Investigate whether it is possible to terminate the use of inert particles, instead adding one or two pulsed gas lines to the current facility or replacing the current steady-flow lines with the pulsed flow lines. Benefits of those modifications should include reducing the carries gas consumption and reducing potential contamination of the torrefied product from inert particles. Future work would: • Explore how pulsations affect the hydrodynamics of slot-rectangular spouted beds. • Study the residence time distribution of sawdust particles in slot-rectangular spouted bed with pulsation(s). • Test the effect of pulsations on biomass torrefaction performance. 5) Perform energy and exergy analyses for the current torrefaction facility to provide additional insights on optimization of the operating conditions.     184  References Almeida, G., Brito, J.O., Perre, P. 2010. Alterations in energy properties of eucalyptus wood and bark subjected to torrefaction: the potential of mass loss as a synthetic indicator. Bioresource Technology, 101(24), 9778-84. Alvarez, J., Amutio, M., Lopez, G., Barbarias, I., Bilbao, J., Olazar, M. 2015. Sewage sludge valorization by flash pyrolysis in a conical spouted bed reactor. Chemical Engineering Journal, 273, 173-183. Ami, D., Posteri, R., Mereghetti, P., Porro, D., Doglia, S.M., Branduardi, P. 2014. Fourier transform infrared spectroscopy as a method to study lipid accumulation in oleaginous yeasts. Biotechnology for Biofuels, 7(1), 12. Arteaga-Pérez, L.E., Segura, C., Bustamante-García, V., Gómez Cápiro, O., Jiménez, R. 2015. Torrefaction of wood and bark from Eucalyptus globulus and Eucalyptus nitens: Focus on volatile evolution vs feasible temperatures. Energy, 93, 1731-1741. Atienza-Martínez, M., Fonts, I., Ábrego, J., Ceamanos, J., Gea, G. 2013. Sewage sludge torrefaction in a fluidized bed reactor. Chemical Engineering Journal, 222, 534-545. Balat, M., Balat, H., Öz, C. 2008. Progress in bioethanol processing. Progress in Energy and Combustion Science, 34(5), 551-573. Basu, P. 2013. Biomass characteristics, Chapter 3 in Biomass gasification, pyrolysis and torrefaction: practical design and theory. Academic press. Basu, P., Rao, S., Dhungana, A. 2013. An investigation into the effect of biomass particle size on its torrefaction. Canadian Journal of Chemical Engineering, 91(3), 466-474. Bates, R.B., Ghoniem, A.F. 2014. Modeling kinetics-transport interactions during biomass torrefaction: The effects of temperature, particle size, and moisture content. Fuel, 137, 216-229. Beltramo, C., Rovero, G., Cavaglià, G. 2009. Hydrodynamic and thermal experimentation on square‐based spouted beds for polymer upgrading and unit scale‐up. Canadian Journal of Chemical Engineering, 87(3), 394-402.  Bioenergy systems, Natural Resource Canada: http://www.nrcan.gc.ca/energy/renewable-electricity/bioenergy-systems/7311, Date Modified: 2017-07-27. Braga, M.B., Wang, Z., Grace, J.R., Lim, C.J., Rocha, S.C.S. 2015. Slot-rectangular spouted bed: hydrodynamic stability and effects of operating conditions on drying performance. Drying Technology, 33(2), 216-226. Cao, L., Yuan, X., Jiang, L., Li, C., Xiao, Z., Huang, Z., Chen, X., Zeng, G., Li, H. 2016. Thermogravimetric characteristics and kinetics analysis of oil cake and torrefied biomass blends. Fuel, 175, 129-136. 185  Cao, L., Yuan, X., Li, H., Li, C., Xiao, Z., Jiang, L., Huang, B., Xiao, Z., Chen, X., Wang, H. 2015. Complementary effects of torrefaction and co-pelletization: Energy consumption and characteristics of pellets. Bioresource Technology, 185, 254-262. Chang, S., Zhao, Z., Zheng, A., He, F., Huang, Z., Li, H. 2012. Characterization of Products from Torrefaction of Sprucewood and Bagasse in an Auger Reactor. Energy & Fuels, 26(11), 7009-7017. Chen, W.-H., Cheng, W.-Y., Lu, K.-M., Huang, Y.-P. 2011a. An evaluation on improvement of pulverized biomass property for solid fuel through torrefaction. Applied Energy, 88(11), 3636-3644. Chen, W.-H., Kuo, P.-C. 2011. Torrefaction and co-torrefaction characterization of hemicellulose, cellulose and lignin as well as torrefaction of some basic constituents in biomass. Energy, 36(2), 803-811. Chen, W.-H., Lu, K.-M., Lee, W.-J., Liu, S.-H., Lin, T.-C. 2014a. Non-oxidative and oxidative torrefaction characterization and SEM observations of fibrous and ligneous biomass. Applied Energy, 114, 104-113. Chen, W.-H., Peng, J., Bi, X.T. 2015. A state-of-the-art review of biomass torrefaction, densification and applications. Renewable and Sustainable Energy Reviews, 44, 847-866. Chen, W.H., Lu, K.M., Liu, S.H., Tsai, C.M., Lee, W.J., Lin, T.C. 2013a. Biomass torrefaction characteristics in inert and oxidative atmospheres at various superficial velocities. Bioresource Technology, 146, 152-60. Chen, Y., Yang, H., Yang, Q., Hao, H., Zhu, B., Chen, H. 2014b. Torrefaction of agriculture straws and its application on biomass pyrolysis poly-generation. Bioresource Technology, 156, 70-7. Chen, Z. 2008. Hydrodynamics, stability and scale-up of slot-rectangular spouted beds, PhD thesis, University of British Columbia. Chen, Z., Lim, C.J., Grace, J.R. 2008. Hydrodynamics of slot-rectangular spouted beds: Effect of slot configuration on the local flow structure. Canadian Journal of Chemical Engineering, 86(3), 598-604. Chen, Z., Lim, C.J., Grace, J.R. 2013b. Stability of slot-rectangular spouted beds with multiple slots. Canadian Journal of Chemical Engineering, 91, 1768-1775. Chen, Z., Lim, C.J., Grace, J.R. 2011b. Stability of slot-rectangular spouted beds: Effect of slot configuration. Canadian Journal of Chemical Engineering, 89(6), 1401-1407. Cho, J., Davis, J.M., Huber, G.W. 2010. The intrinsic kinetics and heats of reactions for cellulose pyrolysis and char formation. ChemSusChem, 3(10), 1162-5. Ciolkosz, D., Wallace, R. 2011. A review of torrefaction for bioenergy feedstock production. Biofuels, Bioproducts and Biorefining, 5(3), 317-329. 186  Colin, B., Dirion, J.L., Arlabosse, P., Salvador, S. 2017. Quantification of the torrefaction effects on the grindability and the hygroscopicity of wood chips. Fuel, 197, 232-239. Daood, S.S., Munir, S., Nimmo, W., Gibbs, B.M. 2010. Char oxidation study of sugar cane bagasse, cotton stalk and Pakistani coal under 1% and 3% oxygen concentrations. Biomass and Bioenergy, 34(3), 263-271. Deng, J., Wang, G.-j., Kuang, J.-h., Zhang, Y.-l., Luo, Y.-h. 2009. Pretreatment of agricultural residues for co-gasification via torrefaction. Journal of Analytical and Applied Pyrolysis, 86(2), 331-337. Devahastin, S., Mujumdar, A.S. 2001. Some hydrodynamic and mixing characteristics of a pulsed spouted bed dryer. Powder Technology, 117(3), 189-197. Dogan, O., Freitas, L., Lim, C.J., Grace, J.R., Luo, B. 2000. Hydrodynamics and stability slot-rectangular spouted bed. Part I: thin bed. Chemical Engineering Communications, 181(1), 225-242. Dogan, Ö.M., Uysal, B.Z., Grace, J.R. 2004. Hydrodynamic studies in a half slot-rectangular spouted bed column. Chemical Engineering Communications, 191(4), 566-579. Donida, M.W., Rocha, S.C. 2002. Coating of urea with an aqueous polymeric suspension in a two-dimensional spouted bed. Drying Technology, 20(3), 685-704. Du, W., Zhang, L., Zhang, B., Bao, S., Xu, J., Wei, W. 2015. Pressure drop and pressure fluctuations in spouted beds with binary mixtures of particles. Powder Technology, 276, 134-143. Epstein, N., Grace, J.R. 2011. Chapter 1 Introduction. in: Spouted and Spout-Fluid Beds (Eds.) N. Epstein, J.R. Grace, Cambridge Univ. Press, Cambridge, UK, pp. 1-16. Erkiaga, A., Lopez, G., Amutio, M., Bilbao, J., Olazar, M. 2014. Influence of operating conditions on the steam gasification of biomass in a conical spouted bed reactor. Chemical Engineering Journal, 237, 259-267. Erkiaga, A., Lopez, G., Amutio, M., Bilbao, J., Olazar, M. 2013a. Steam gasification of biomass in a conical spouted bed reactor with olivine and γ-alumina as primary catalysts. Fuel Processing Technology, 116, 292-299. Erkiaga, A., Lopez, G., Amutio, M., Bilbao, J., Olazar, M. 2013b. Syngas from steam gasification of polyethylene in a conical spouted bed reactor. Fuel, 109, 461-469. Fang, J., Sun, R., Tomkinson, J. 2000. Isolation and characterization of hemicelluloses and cellulose from rye straw by alkaline peroxide extraction. Cellulose, 7(1), 87-107. Fernandez-Akarregi, A.R., Makibar, J., Lopez, G., Amutio, M., Altzibar, H., Olazar, M. 2014. Development of a dual conical spouted bed system for heat integration purposes. Powder Technology, 268, 261-268. Fernandez-Akarregi, A.R., Makibar, J., Lopez, G., Amutio, M., Olazar, M. 2013. Design and operation of a conical spouted bed reactor pilot plant (25 kg/h) for biomass fast pyrolysis. Fuel Processing Technology, 112, 48-56. 187  Fernández-Akarregui, A.R., Makibar, J., Alava, I., Diaz, L., Cueva, F., Aguado, R., Lopez, G., Olazar, M. 2012. Sand attrition in conical spouted beds. Particuology, 10(5), 592-599. Freitas, L., Dogan, O., Lim, C.J., Grace, J.R., Luo, B. 2000. Hydrodynamics and stability of slot-rectangular spouted beds. Part II: increasing bed thickness. Chemical Engineering Communications, 181(1), 243-258. Freitas, L.A., Dogan, O.M., Lim, C.J., Grace, J.R., Bai, D. 2004. Identification of flow regimes in slot‐rectangular spouted beds using pressure fluctuations. Canadian Journal of Chemical Engineering, 82(1), 60-73. Gan, L., Lu, X., Wang, Q., Hu, Q., Chen, Y., Xu, J. 2013. Experimental study on lateral mixing of particles in a quasi-slot-rectangular spouted bed. Powder Technology, 243, 1-8. Geldart, D., Abdullah, E.C., Hassanpour, A., Nwoke, L.C., Wouters, I. 2006. Characterization of powder flowability using measurement of angle of repose. Particuology, 4(3-4), 104-107. Golshan, S., Zarghami, R., Mostoufi, N. 2017. Hydrodynamics of slot-rectangular spouted beds: Process intensification. Chemical Engineering Research and Design, 121, 315-328. Gomez, C., Velo, E., Barontini, F., Cozzani, V. 2009. Influence of secondary reactions on the heat of pyrolysis of biomass. Industrial & Engineering Chemistry Research, 48(23), 10222-10233. Gomez, L.D., Steele‐King, C.G., McQueen‐Mason, S.J. 2008. Sustainable liquid biofuels from biomass: the writing's on the walls. New Phytologist, 178(3), 473-485. Grace, J.R., Lim, C.J. 2011. Chapter 17  Scale-up, slot-rectangular and multiple spouting. in: Spouted and Spout-Fluid Beds, (Eds.) N. Epstein, J.R. Grace, Cambridge Univ. Press, Cambridge, UK, pp. 286-299. Guo, J., Lua, A. 2001. Kinetic study on pyrolytic process of oil-palm solid waste using two-step consecutive reaction model. Biomass and Bioenergy, 20(3), 223-233. Haddou, J.V.H., Ellis, N., Bi, X., Epstein, N. 2013. Spouting characteristics of SPF wood pellets. Canadian Journal of Chemical Engineering, 91(5), 808-813. Haykırı-Açma, H. 2003. Combustion characteristics of different biomass materials. Energy Conversion and Management, 44(1), 155-162. Hu, G., Gong, X., Wei, B., Li, Y. 2008. Flow patterns and transitions of a novel annular spouted bed with multiple air nozzles. Industrial & Engineering Chemistry Research, 47(23), 9759-9766. Huang, H., Hu, G. 2007. Mixing characteristics of a novel annular spouted bed with several angled air nozzles. Industrial & Engineering Chemistry Research, 46(24), 8248-8254. 188  Huang, Y.-F., Sung, H.-T., Chiueh, P.-T., Lo, S.-L. 2017. Microwave torrefaction of sewage sludge and leucaena. Journal of the Taiwan Institute of Chemical Engineers, 70, 236-243. Janssen, F., Konings, A. 2000. Method and installation for recovering energy from biomass and waste, U.S. Patent Application No. 09/740,007. Joshi, Y., Di Marcello, M., Krishnamurthy, E., de Jong, W. 2015. Packed-bed torrefaction of bagasse under inert and oxygenated atmospheres. Energy & Fuels, 29(8), 5078-5087. Kačuráková, M., Belton, P.S., Wilson, R.H., Hirsch, J., Ebringerová, A. 1998. Hydration properties of xylan‐type structures: an FTIR study of xylooligosaccharides. Journal of the Science of Food and Agriculture, 77(1), 38-44. Kacurakova, M., Capek, P., Sasinkova, V., Wellner, N., Ebringerova, A. 2000. FT-IR study of plant cell wall model compounds: pectic polysaccharides and hemicelluloses. Carbohydrate Polymers, 43(2), 195-203. Kmiec, A., Englart, S. 2011. Chapter 9  Heat and mass transfer. in: Spouted and spout-fluid beds, (Eds.) N. Epstein, J.R. Grace, Cambridge Univ. Press. Küçükbayrak, S., Haykırı-Açma, H., Ersoy-Meriçboyu, A., Yaman, S. 2001. Effect of lignite properties on reactivity of lignite. Energy Conversion and Management, 42(5), 613-626. Lacey, P.M.C. 1954. Developments in the theory of particle mixing. Journal of Applied Chemistry, 4(5), 257-268. Li, H., Liu, X., Legros, R., Bi, X.T., Lim, C.J., Sokhansanj, S. 2012a. Pelletization of torrefied sawdust and properties of torrefied pellets. Applied Energy, 93, 680-685. Li, H., Liu, X., Legros, R., Bi, X.T., Lim, C.J., Sokhansanj, S. 2012b. Torrefaction of sawdust in a fluidized bed reactor. Bioresource Technology, 103(1), 453-8. Li, H., Wang, S., Yuan, X., Xi, Y., Huang, Z., Tan, M., Li, C. 2017. The effects of temperature and color value on hydrochars’ properties in hydrothermal carbonization. Bioresource Technology, 10.1016/j.biortech.2017.10.046. Lopez, G., Cortazar, M., Alvarez, J., Amutio, M., Bilbao, J., Olazar, M. 2017. Assessment of a conical spouted with an enhanced fountain bed for biomass gasification. Fuel, 203, 825-831. López, G., Olazar, M., Aguado, R., Bilbao, J. 2010. Continuous pyrolysis of waste tyres in a conical spouted bed reactor. Fuel, 89(8), 1946-1952. Luo, B., Lim, C.J., Freitas, L.A., Grace, J.R. 2004. Flow characteristics in slot‐rectangular spouted beds with draft plates. Canadian Journal of Chemical Engineering, 82(1), 83-88. Makibar, J., Fernandez-Akarregi, A.R., Díaz, L., Lopez, G., Olazar, M. 2012. Pilot scale conical spouted bed pyrolysis reactor: Draft tube selection and hydrodynamic performance. Powder Technology, 219, 49-58. 189  Martín-Lara, M.A., Ronda, A., Zamora, M.C., Calero, M. 2017. Torrefaction of olive tree pruning: Effect of operating conditions on solid product properties. Fuel, 202, 109-117. Masnadi, M.S., Habibi, R., Kopyscinski, J., Hill, J.M., Bi, X., Lim, C.J., Ellis, N., Grace, J.R. 2014. Fuel characterization and co-pyrolysis kinetics of biomass and fossil fuels. Fuel, 117, 1204-1214. Mathur, K.B., Gishler, P. 1955. A technique for contacting gases with coarse solid particles. AIChE Journal, 1(2), 157-164. McNab, G. 1972. Prediction of spout diameter. British Chemical Engineering and Process Technology, 17(6), 532-&. Mei, Y., Liu, R., Yang, Q., Yang, H., Shao, J., Draper, C., Zhang, S., Chen, H. 2015. Torrefaction of cedarwood in a pilot scale rotary kiln and the influence of industrial flue gas. Bioresource Technology, 177, 355-60. Milosavljevic, I., Oja, V., Suuberg, E.M. 1996. Thermal effects in cellulose pyrolysis: relationship to char formation processes. Industrial & Engineering Chemistry Research, 35(3), 653-662. Mohan, D., Pittman, C.U., Steele, P.H. 2006. Pyrolysis of wood/biomass for bio-oil: a critical review. Energy & Fuels, 20(3), 848-889. Mostoufi, N., Kulah, G., Koksal, M. 2015. Flow structure characterization in conical spouted beds using pressure fluctuation signals. Powder Technology, 269, 392-400. Mujumdar, A. 1984. Spouted bed technology: A brief review. Drying, 84, 151-157. Munir, S., Daood, S.S., Nimmo, W., Cunliffe, A.M., Gibbs, B.M. 2009. Thermal analysis and devolatilization kinetics of cotton stalk, sugar cane bagasse and shea meal under nitrogen and air atmospheres. Bioresource Technology, 100(3), 1413-8. Nachenius, R.W., van de Wardt, T.A., Ronsse, F., Prins, W. 2015. Torrefaction of pine in a bench-scale screw conveyor reactor. Biomass and Bioenergy, 79, 96-104. Olazar, M., Aguado, R., San José, M.J., Alvarez, S., Bilbao, J. 2006. Minimum spouting velocity for the pyrolysis of scrap tyres with sand in conical spouted beds. Powder Technology, 165(3), 128-132. Oyedun, A.O., Tee, C.Z., Hanson, S., Hui, C.W. 2014. Thermogravimetric analysis of the pyrolysis characteristics and kinetics of plastics and biomass blends. Fuel Processing Technology, 128, 471-481. Parise, M.R., Wang, Z., Lim, C.J., Grace, J.R. 2017. Hydrodynamics of a slot-rectangular spouted bed of biomass particles with simultaneous injection of spouting and pulsating air streams. Chemical Engineering Journal, 330, 82-91. Passos, M., Mujumdar, A., Raghavan, V. 1993. Prediction of the maximum spoutable bed height in two-dimensional spouted beds. Powder Technology, 74(2), 97-105. 190  Peng, J., Bi, X.T., Lim, C.J., Sokhansanj, S. 2012a. Development of torrefaction kinetics for British Columbia softwoods. International  Journal of Chemical Reactor Engineering, 10(1), 1-37. Peng, J.H., Bi, H.T., Lim, C.J., Sokhansanj, S. 2013a. Study on density, hardness, and moisture uptake of torrefied wood pellets. Energy & Fuels, 27(2), 967-974. Peng, J.H., Bi, H.T., Sokhansanj, S., Lim, C.J. 2012b. A study of particle size effect on biomass torrefaction and densification. Energy & Fuels, 26(6), 3826-3839. Peng, J.H., Bi, X.T., Sokhansanj, S., Lim, C.J. 2013b. Torrefaction and densification of different species of softwood residues. Fuel, 111, 411-421. Phanphanich, M., Mani, S. 2011. Impact of torrefaction on the grindability and fuel characteristics of forest biomass. Bioresource Technology, 102(2), 1246-53. Prins, M.J., Ptasinski, K.J., Janssen, F.J.J.G. 2006a. Torrefaction of wood: Part 1. Weight loss kinetics. Journal of Analytical and Applied Pyrolysis, 77(1), 28-34. Prins, M.J., Ptasinski, K.J., Janssen, F.J.J.G. 2006b. Torrefaction of wood: Part 2. Analysis of products. Journal of Analytical and Applied Pyrolysis, 77(1), 35-40. Qiu, K., Hu, C., Yang, S., Luo, K., Zhang, K., Fan, J. 2016. Computational evaluation of depth effect on the hydrodynamics of slot-rectangular spouted bed. Powder Technology, 287, 51-60. Recari, J., Berrueco, C., Puy, N., Alier, S., Bartrolí, J., Farriol, X. 2017. Torrefaction of a solid recovered fuel (SRF) to improve the fuel properties for gasification processes. Applied Energy, 203, 177-188. Ren, B., Shao, Y., Zhong, W., Jin, B., Yuan, Z., Lu, Y. 2012. Investigation of mixing behaviors in a spouted bed with different density particles using discrete element method. Powder Technology, 222, 85-94. Ren, B., Zhong, W., Zhang, Y., Jin, B., Wang, X., Tao, H., Xiao, R. 2010. Investigation on flow patterns and transitions in a multiple-spouted bed. Energy & Fuels, 24(3), 1941-1947. Rousset, P., Macedo, L., Commandré, J.M., Moreira, A. 2012. Biomass torrefaction under different oxygen concentrations and its effect on the composition of the solid by-product. Journal of Analytical and Applied Pyrolysis, 96, 86-91. Rovero, G., Cavaglià, G., Curti, M. 2012. Optimization of spouted bed scale-up by square-based multiple unit design. INTECH Open Access Publisher. Saidi, M., Basirat Tabrizi, H., Grace, J.R., Lim, C.J., Ahmadi, G. 2015a. Hydrodynamic and Mixing Characteristics of Gas–Solid Flow in a Pulsed Spouted Bed. Industrial & Engineering Chemistry Research, 54(32), 7933-7941. Saidi, M., Wang, Z., Grace, J.R., Lim, C.J. 2015b. Numerical and experimental investigation of hydrodynamic characteristics of a slot‐rectangular spouted bed. Canadian Journal of Chemical Engineering. 191  Saidi, M., Wang, Z., Grace, J.R., Lim, C.J. 2016. Numerical and experimental investigation of hydrodynamic characteristics of a slot‐rectangular spouted bed. Canadian Journal of Chemical Engineering, 94(2), 332-339. Saidutta, M., Murthy, D. 2000. Mixing behaviour of solids in multiple spouted beds. The Canadian Journal of Chemical Engineering, 78(2), 382-385. Sarkar, M., Kumar, A., Tumuluru, J.S., Patil, K.N., Bellmer, D.D. 2014. Gasification performance of switchgrass pretreated with torrefaction and densification. Applied Energy, 127, 194-201. Shen, D., Gu, S., Bridgwater, A.V. 2010. Study on the pyrolytic behaviour of xylan-based hemicellulose using TG–FTIR and Py–GC–FTIR. Journal of Analytical and Applied Pyrolysis, 87(2), 199-206. Shirvanian, P.A., Calo, J.M. 2004. Hydrodynamic scaling of a rectangular spouted vessel with a draft duct. Chemical Engineering Journal, 103(1-3), 29-34. Sinclair, R., McKay, A., Myers, G., Jones, R.N. 1952. The infrared absorption spectra of unsaturated fatty acids and esters. Journal of the American Chemical Society, 74(10), 2578-2585. Sluiter, A., Hames, B., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D. 2008a. Determination of ash in biomass (NREL/TP-510-42622). National Renewable Energy Laboratory, Golden, 5. Sluiter, A., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D. 2008b. Determination of extractives in biomass (NREL/TP-510-42619). National Renewable Energy Laboratory, Golden. Strandberg, M., Olofsson, I., Pommer, L., Wiklund-Lindström, S., Åberg, K., Nordin, A. 2015. Effects of temperature and residence time on continuous torrefaction of spruce wood. Fuel Processing Technology, 134, 387-398. Sukhbaatar, B., Kim, M., Steele, P., Ingram, L. 2014. Optimization of hot-compressed water pretreatment of bagasse and characterization of extracted hemicelluloses. Carbohydrate Polymers, 101, 196-202. Sukiran, M.A., Abnisa, F., Wan Daud, W.M.A., Abu Bakar, N., Loh, S.K. 2017. A review of torrefaction of oil palm solid wastes for biofuel production. Energy Conversion and Management, 149, 101-120. Tabatabaei, S.A. 2013. A study on the hydrodynamic behavior of a slot-rectangular spouted bed using CFD, Master Thesis, University of Calgary. Tabatabaei, S.A., Mahinpey, N., Esmaili, E., Lim, C.J. 2013. CFD simulation of flow regime maps in a slot-rectangular spouted bed. Canadian Journal of Chemical Engineering, 91, 1856-1864. Uemura, Y., Omar, W., Othman, N.A., Yusup, S., Tsutsui, T. 2013. Torrefaction of oil palm EFB in the presence of oxygen. Fuel, 103, 156-160. 192  Uemura, Y., Sellappah, V., Hoai Thanh, T., Hassan, S., Tanoue, K.-i. 2017. Torrefaction of empty fruit bunches under biomass combustion gas atmosphere. Bioresource Technology, 243, 107-117. Verhoeff, F., Pels, J., Boersma, A., Zwart, R., Kiel, J. 2011. ECN torrefaction technology heading for demonstration. 19th European Biomass Conference and Exhibition. pp. 6-10. Vlaev, L.T., Markovska, I.G., Lyubchev, L.A. 2003. Non-isothermal kinetics of pyrolysis of rice husk. Thermochimica Acta, 406(1-2), 1-7. Wang, B., Luo, B. 2001. Minimum Spouting Velocity of Multi-component Particle Mixtures in a Slot-rectangular Spouted Bed. Chinese Journal of Process Engineering, 2, 000. Wang, C., Peng, J., Li, H., Bi, X.T., Legros, R., Lim, C.J., Sokhansanj, S. 2013a. Oxidative torrefaction of biomass residues and densification of torrefied sawdust to pellets. Bioresource Technology, 127, 318-25. Wang, L., Barta-Rajnai, E., Skreiberg, Ø., Khalil, R., Czégény, Z., Jakab, E., Barta, Z., Grønli, M. 2017a. Effect of torrefaction on physiochemical characteristics and grindability of stem wood, stump and bark. Applied Energy, doi.org/10.1016/j.apenergy.2017.07.024. Wang, S., Dai, G., Ru, B., Zhao, Y., Wang, X., Xiao, G., Luo, Z. 2017b. Influence of torrefaction on the characteristics and pyrolysis behavior of cellulose. Energy, 120, 864-871. Wang, S., Dai, G., Ru, B., Zhao, Y., Wang, X., Zhou, J., Luo, Z., Cen, K. 2016a. Effects of torrefaction on hemicellulose structural characteristics and pyrolysis behaviors. Bioresource Technology, 218, 1106-14. Wang, S., Ru, B., Lin, H., Luo, Z. 2013b. Degradation mechanism of monosaccharides and xylan under pyrolytic conditions with theoretic modeling on the energy profiles. Bioresource Technology, 143, 378-83. Wang, Z., Lim, C.J., Grace, J.R., Li, H., Regina Parise, M. 2017c. Effects of temperature and particle size on biomass torrefaction in a slot-rectangular spouted bed reactor. Bioresource Technology, 244, 281-288. Wang, Z., Lim, C.J., Grace, J.R. 2016b. Solids mixing in a dual-column slot-rectangular spouted bed. Powder Technology, 301, 1264-1269. Wang, Z., Saidi, M., Lim, C.J., Grace, J.R., Basirat Tabrizi, H., Chen, Z., Li, Y. 2016c. Comparison of DEM simulation and experiments in a dual-column slot-rectangular spouted bed with a suspended partition. Chemical Engineering Journal, 290, 63-73. Wiriyaumpaiwong, S., Soponronnarit, S., Prachayawarakorn, S. 2003. Soybean Drying by Two-Dimensional Spouted Bed. Drying Technology, 21(9), 1735-1757. Yang, H., Yan, R., Chen, H., Lee, D.H., Zheng, C. 2007. Characteristics of hemicellulose, cellulose and lignin pyrolysis. Fuel, 86(12-13), 1781-1788. 193  Yang, S., Luo, K., Fang, M., Zhang, K., Fan, J. 2014. Parallel CFD–DEM modeling of the hydrodynamics in a lab-scale double slot-rectangular spouted bed with a partition plate. Chemical Engineering Journal, 236, 158-170. Yang, S., Luo, K., Fang, M., Zhang, K., Fan, J. 2013. Three-dimensional modeling of gas–solid motion in a slot-rectangular spouted bed with the parallel framework of the computational fluid dynamics–discrete element method coupling approach. Industrial & Engineering Chemistry Research, 52(36), 13222-13231. Yang, S., Luo, K., Zhang, K., Qiu, K., Fan, J. 2015a. Numerical study of a lab-scale double slot-rectangular spouted bed with the parallel CFD–DEM coupling approach. Powder Technology, 272, 85-99. Yang, S., Zhang, K., Chew, J.W. 2015b. Computational study of spout collapse and impact of partition plate in a double slot-rectangular spouted bed. AIChE Journal, 61(12), 4087-4101. Zhang, H., Li, S. 2017. Study on drag force coefficients in modeling granular flows in a slot-rectangular spouted bed. Proceedings of the 7th International Conference on Discrete Element Methods, 188, 697-707. Zhao, X.L., Li, S.Q., Liu, G.Q., Song, Q., Yao, Q. 2008. Flow patterns of solids in a two-dimensional spouted bed with draft plates: PIV measurement and DEM simulations. Powder Technology, 183(1), 79-87. Zhou, H., Long, Y., Meng, A., Chen, S., Li, Q., Zhang, Y. 2015. A novel method for kinetics analysis of pyrolysis of hemicellulose, cellulose, and lignin in TGA and macro-TGA. RSC Advances, 5(34), 26509-26516. Zhuravlev, V., Lesokhin, I., Tempelman, R. 1948. Kinetics of reactions in the formation of aluminates and the contribution of mineralizers to the process. Journal of Applied Chemistry of the USSR, 21(09), 887-890    194  Appendix A: Summary of previous work on slot-rectangular spouted beds Table A-1 Previous literature on slot-rectangular spouted beds. Author (year) Comments  Passos et al. (1993) Predicts the maximum spoutable bed height in a slot-rectangular spouted bed. Dogan et al. (2000) Observed eight different flow regimes and studied mechanisms of spout termination. Also, renamed slot-rectangular spouted bed instead of two-dimensional spouted bed.  Freitas et al. (2000) Observed significant three-dimensional effects as the column thickness increased, leading to formation of multiple spouts.  Wang and Luo (2001) Investigated the minimum spouting velocity of multi-size glass beads mixtures, and proposed maps of minimum spouting velocity for multi-size particle mixture.  Dogan et al. (2004) Investigated effects of slot width and lower section basal angle on hydrodynamics in a half SRSB. Freitas et al. (2004) Identified flow regimes by means of pressure fluctuation with the aid of statistical, spectral and chaotic analysis in SRSBs.  Luo et al. (2004) Measured the minimum spouting velocity, maximum pressure drop and local voidage in SRSBs with draft plates.   Chen et al. (2008) Performed experiments in a slot-rectangular spouted bed of 300 mm x 100 mm cross-section. Studied effect of length-to-width ratios of the slot with a fixed cross-section area on the local flow structure, and found the effect of slot geometry was small in the upper part of the bed.    195  Author (year) Comments  Chen et al. (2011b) Found slot length, width, and depth all affect the stability of SRSBs, and proposed criteria for achieving stable spouting: √slot area𝑑𝑝< 20 ; and 𝜅/𝑑𝑝 > 25 or 𝜂/ 𝜅 ≤ 15  Chen et al. (2013b) Investigated effect of different configurations of multiple slots, including parallel and aligned, on stability of slot-rectangular spouted bed with multiple compartments and slots, and compared with a single-compartment single-slot spouted bed. They proposed to use a suspended partition to stabilize spouting in the SRSBs with multiple compartments and slots.  Gan et al. (2013) Performed experiments in a quasi-slot-rectangular spouted bed to investigate lateral mixing of particles by means of the bed collapse method, and employed the lateral dispersion coefficient of tracer particles to represent the degree of solids mixing. Tabatabaei et al. (2013) Employed computational fluid dynamics (CFD) using the two-phase Eulerian-Eulerian granular model to plot flow regime maps of SRSBs, and six distinct flow patterns were identified. Yang et al. (2013) Modelled the gas-flow in a three-dimensional SRSB under the parallel framework of the CFD-DEM coupling approach, and studied profiles of pressure drop, gas velocity and spout-annulus interface in the SRSB. Yang et al. (2014) Performed three-dimensional modeling of hydrodynamics in a DSRSB with a suspended plate using CFD-DEM approach.  Braga et al. (2015) Applied a DSRSB to milk-blackberry pulp drying, and found DSRSB allowed flow rate of paste to be tripled and the bed to be operated at a lower superficial gas velocity compared to the conventional spouted bed.   196  Author (year) Comments  Yang et al. (2015a) Conducted numerical study of a DSRSB with CFD-DEM approach, explored the start-up procedure and flow patterns.  The results showed that the minimum spouting velocity of a DSRSB was higher than that of a SRSB. Yang et al. (2015b) Numerically studied of spout collapse in a DSRSB, and suggested that inserting a vertical partition was an effective way to prevent unstable spouting in the DSRSB. Qiu et al. (2016) Studied effect of bed depth on hydrodynamics of a SRSB using a three-dimensional model with CFD-DEM approach. Saidi et al. (2016) Experimentally and numerically studied the gas-solid flow in a three-dimensional SRSB. The simulation used CFD-DEM approach in the OpenFOAM, and the experiments employed pressure transducer and optical fiber probe. The profiles of pressure, voidage and particle flux were obtained.  Wang et al. (2016b) Performed experiments on solids mixing between two adjacent compartments in a DSRSB with a suspended partition, and investigated effects of partition position, superficial gas velocity, static bed height and partition height on the solids mixing.   Wang et al. (2016c) Investigated experimentally and numerically hydrodynamics of a DSRSB with a suspended partition, i.e. particle velocity, voidage and solids mixing, and developed a CFD-DEM model for DSRSB.  Golshan et al. (2017) Studied effect of bed geometry, e.g. cone angle of 20-70° and curved cone base, on hydrodynamics of SRSBS by means of CFD-DEM approach. Parise et al. (2017) Proposed pulsation to facilitate biomass particles movement in a SRSB; investigated effects of pulsation frequency and ratio of pulsation flow rate to spouting fluid flow rate on hydrodynamics.     197  Author (year) Comments  Wang et al. (2017c) Apply SRSB to biomass torrefaction, and investigated effects of temperature an particle size on biomass torrefaction performance in a slot-rectangular spouted bed reactor. Zhang and Li (2017) Assessed effect of seven drag force coefficients on simulation of the granular flows in a SRSB; suggested Dahl and Hrenya model was the best one.   198  Appendix B: Temperature profiles 0 10 20 30 40 50200220240260280300320340 Windbox (T1 ) Z = 64 mm (T2) Z = 165 mm (T3) Z = 267 mm (T4) Average T Average T Z = 267 mm (T4)Z = 64 mm (T2)  T (oC)Time (min)Windbox (T1 )Z = 165 mm (T3) Figure B-1 Temperature profiles in SRSB reactor for case OT-6, T = 271°C, with XO2 = 9 vol.% oxygen in the feed-gas and F = 495 g/h sawdust feed rate. 0 10 20 30 40 50260280300320340360 Windbox (T1) Z = 64 mm (T2)  Z = 165 mm (T3) Z = 267 mm (T4)  Average T Average T Z = 267 mm (T4) Z = 64 mm (T2)   T (oC)Time (min)Windbox (T1) Z = 165 mm (T3)  Figure B-2 Temperature profiles in SRSB reactor for case OT-9, T = 318°C, with XO2 = 9 vol.% oxygen in the feed-gas and F = 469 g/h sawdust feed rate.199  Appendix C: Fourier Transform Infrared Spectroscopy  Fourier transform infrared spectroscopy (FTIR) was applied to access the changes in the chemical structure of the torrefied product under different operating conditions. The spectra were recorded on a spectrometer (Cary 600 Series, Agilent Technologies) in the range of 650-4000 cm-1 with a resolution of 4 cm-1. Each spectrum was accumulated from 64 scans. The measurement method is attenuated total reflectance (ATR). Figure C-1 shows the FTIR spectra of the raw and cyclone-caught biomass torrefied at different operating conditions in SRSB reactor.  It was found that the raw and torrefied sawdust had the same functional groups, but with different intensities in the range of 650-4000 cm-1. To describe important structural changes, some well-defined peaks were specifically assigned to the following structural components:  (1) 3345 cm-1 for the stretching vibration of hydroxyl, widely distributed over monosaccharide rings, hydroxymethyl, uronic acid, and absorbed water. (2) 2920 cm-1 for C-H combination tone, mainly due to stretching in -CH2- and methyl groups.  (3) 2852 cm-1 for symmetric vibration of -CH2-, likely reflects metabolic changes that accompany cell growth (Ami et al., 2014).  (4) 2360 cm-1 is a characteristic CO2 peak(L.R. Glicksman, 1993).  (5)1735 cm-1  is attributed to stretching vibration of acetyl and carboxyl in hemicellulose (Sukhbaatar et al., 2014).  (6) 1670 cm-1 may be assigned to water of hydration (Kačuráková et al., 1998).  200  (7) 1508 cm-1 is due to the aromatic skeletal vibration with C=O stretch in lignin.  (8) 1456 cm-1 corresponds to the bending and stretching vibrations of -CH2- in sugar rings.  (9) 1417 cm-1 is due to -CH2- deformation (Sinclair et al., 1952).  (10) 1373 cm-1 is characteristic of methyl groups, which mainly exists in the o-acetyl branches and 4-o-methyl-glucuronic acid in hemicellulose and cellulose.  (11) 1265 cm-1 corresponds to bending and stretching vibrations of -CH2- in sugar rings.  (12) 1157 cm-1 may result from C-O-C stretching in hemicellulose and cellulose.  (13) 1025 cm-1 is a characteristic of stretching vibration of glucosidic bonds in glucomannan. (14) 899 cm-1 corresponds to the frequency of the C1 group in the β-1, 4-glucosidic bond (Fang et al., 2000) (for C-H deformation in cellulose).  (15) 808 cm-1 implies the anomeric region of the mannose residue (Kacurakova et al., 2000).  4500 4000 3500 3000 2500 2000 1500 1000 500C-3C-2C-1  AbsorbanceWavenumbers (cm-1)RawC-6C-5C-4C-10C-9C-8 201  4500 4000 3500 3000 2500 2000 1500 1000 500PS-9PS-2PS-1  AbsorbanceWavenumbers (cm-1)RawPS-4PS-3PS-8PS-7PS-6PS-5 4500 4000 3500 3000 2500 2000 1500 1000 500OT-9OT-2OT-1  AbsorbanceWavenumbers (cm-1)RawOT-3OT-4OT-8OT-7OT-6OT-5 Figure C-1 FTIR spectra of raw and cyclone-caught torrefied sawdust: (1) Sawdust was torrefied at different temperatures and biomass feed rates in SRSB reactor. See Table 4-1 for operating conditions. (2) Different size sawdust was torrefied in SRSB reactor. See Table 4-6 for operating conditions. (3) Sawdust was torrefied at different 202  temperature and oxygen concentration in SRSB reactor. See Table 4-10 for operating conditions. The relative intensity (I7/Ii) of the lignin-associated band with carbohydrate bands for raw and torrefied sawdust was introduced to characterize the changes of the functional groups, where I7 is the intensity of 1508 cm-1 of the lignin-associated band and Ii is the intensity of the specific carbohydrate band in the spectrum (Chang et al., 2012). Specifically, I7/I5 portrays the relative intensity of aromatic skeletal vibration against C=O vibration; I7/I10 denotes the relative intensity of aromatic skeletal vibration against C-H deformation in cellulose and hemicellulose; I7/I12 represents the relative intensity of aromatic skeletal vibration against C-O-C vibration in cellulose and hemicellulose; I7/I14 is assigned to the relative intensity of aromatic skeletal vibration against C-H deformation in cellulose. Table C-1 shows ratios of intensities of lignin-associated band with carbohydrate bands for original and cyclone-caught biomass torrefied at different temperatures and biomass feed rates in SRSB reactor. The ratio I7/I5 of torrefied product was less than that of untreated sawdust, because of removal of acetyl groups in hemicellulose. Effect of temperature on I7/I5 was found to be slight. The ratios I7/I10 and I7/I12 increased with torrefaction temperature, revealing that thermal decomposition and depolymerization were dominant during torrefaction. The ratio of I7/I14 increased with torrefaction temperature, probably illustrating that intramolecular dehydration reaction took place in the cellulose during torrefaction.    203  Table C-1 Ratios of intensities of lignin-associated band with carbohydrate bands for original and cyclone-caught biomass torrefied at different temperatures and biomass feed rates in SRSB reactor. See Table 4-1 for operating conditions. Case I7/I5 I7/I10 I7/I12 I7/I14 Raw SPF 1.93±0.05 0.87±0.05 0.69±0.06 0.71±0.07 C-1: 0.5-1.0mm+240°C+200g/h 1.71±0.05 0.82±0.01 0.63±0.01 0.68±0.01 C-2: 0.5-1.0mm+270°C+200g/h 1.79±0.03 0.87±0.00 0.69±0.02 0.76±0.06 C-3: 0.5-1.0mm+300°C+200g/h 1.76±0.09 0.86±0.00 0.66±0.01 0.74±0.03 C-4: 0.5-1.0mm+240°C+400g/h 1.80±0.08 0.98±0.05 0.70±0.03 0.75±0.02 C-5: 0.5-1.0mm+270°C+400g/h 1.81±0.04 0.91±0.01 0.72±0.05 0.80±0.05 C-6: 0.5-1.0mm+300°C+400g/h 1.82±0.01 1.04±0.01 0.84±0.01 0.92±0.02 C-8: 0.5-1.0mm+240°C+600g/h 1.85±0.09 0.97±0.05 0.78±0.08 0.76±0.09 C-9: 0.5-1.0mm+270°C+600g/h 1.87±0.12 0.94±0.04 0.73±0.11 0.81±0.07 C-10: 0.5-1.0mm 300°C+600g/h 1.83±0.06 1.07±0.03 0.82±0.03 0.94±0.08 ±value: Value of standard deviation. Table C-2 shows ratios of relative intensity of lignin-associated band with carbohydrate bands for raw and cyclone-caught torrefied sawdust of different particle sizes obtained in SRSB reactor. The ratio I7/I5 decreased with increasing the sawdust size, due to longer residence time for larger particles. The ratios I7/I10, I7/I12 and I7/I14 increased with increasing particle size, because larger particles underwent more severe torrefaction.      204  Table C-2 Ratios of relative intensity of lignin-associated band with carbohydrate bands for raw and cyclone-caught torrefied sawdust of different particle sizes. See Table 4-6 for operating conditions. Case I7/I5 I7/I10 I7/I12 I7/I14 Raw SPF sawdust 1.93±0.05 0.87±0.05 0.69±0.06 0.71±0.07 PS-1: 0.25-0.5mm+240°C 1.88±0.04 0.93±0.02 0.74±0.04 0.81±0.05 PS-2: 0.25-0.5mm+270°C 1.95±0.03 1.07±0.01 0.86±0.02 0.95±0.04 PS-3: 0.25-0.5mm+300°C 1.92±0.00 0.98±0.05 0.77±0.08 0.87±0.13 PS-4: 0.5-1.0mm+240°C 1.85±0.08 1.08±0.05 0.87±0.03 0.95±0.02 PS-5: 0.5-1.0mm +270°C 1.81±0.04 0.91±0.01 0.72±0.05 0.80±0.05 PS-6: 0.5-1.0mm +300°C 1.82±0.01 1.04±0.01 0.84±0.01 0.92±0.02 PS-7: 1.0-2.0mm +240°C 1.82±0.11 0.94±0.04 0.72±0.06 0.88±0.09 PS-8: 1.0-2.0mm +270°C 1.95±0.27 0.95±0.02 0.72±0.02 0.86±0.02 PS-9: 1.0-2.0mm+300°C 1.87±0.10 0.94±0.05 0.73±0.07 0.83±0.10   ± value: Value of standard deviation. Table C-3 shows ratios of relative intensity of lignin-associated band with carbohydrate bands for raw and cyclone-caught torrefied biomass of oxidative torrefaction in SRSB reactor. It is seen that I7/I5 decreased as the oxygen concentration increased, because of removal of acetyl groups due to hemicellulose decomposition and oxidation. The I7/I10 and I7/I12 ratios increased with increasing the oxygen concentration, indicating that oxidation reaction may facilitate thermal decomposition and depolymerization during torrefaction. The ratio of I7/I14 increased with increasing oxygen concentration, probably illustrating that cellulose deformation increased as the oxygen content in the feed-gas increased during torrefaction. 205  Table C-3 Ratios of relative intensity of lignin-associated band with carbohydrate bands for raw and cyclone-caught torrefied biomass of oxidative torrefaction in SRSB reactor. See Table 4-10 for operating conditions. Case I7/I5 I7/I10 I7/I12 I7/I14 Raw SPF 1.93±0.05 0.87±0.05 0.69±0.06 0.71±0.07 OT-1: 0.5-1.0mm+240°C+3%O2 1.87±0.14 0.88±0.00 0.70±0.02 0.76±0.01 OT-2: 0.5-1.0mm+240°C+6%O2 1.58±0.00 1.02±0.02 0.61±0.01 0.80±0.09 OT-3: 0.5-1.0mm+240°C+9%O2 1.89±0.02 0.90±0.00 0.69±0.02 0.75±0.01 OT-4: 0.5-1.0mm+270°C+3%O2 1.83±0.01 0.84±0.02 0.66±0.00 0.73±0.02 OT-5: 0.5-1.0mm+270°C+6%O2 1.73±0.10 0.87±0.03 0.69±0.05 0.77±0.04 OT-6: 0.5-1.0mm+270°C+9%O2 1.67±0.15 1.09±0.04 0.63±0.04 0.81±0.06 OT-7: 0.5-1.0mm+300°C+3%O2 1.84±0.02 0.89±0.03 0.69±0.03 0.77±0.02 OT-8: 0.5-1.0mm+300°C+6%O2 1.86±0.10 1.04±0.14 0.84±0.13 0.87±0.04 OT-9: 0.5-1.0mm+300°C+9%O2 1.82±0.03 0.97±0.00 0.67±0.02 0.86±0.01        ± value: Value of standard deviation.  

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:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0362042/manifest

Comment

Related Items