Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Moisture-related physical properties and self-heating of wood pellets Lee, Jun Sian 2021

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2021_may_lee_junsian.pdf [ 2.9MB ]
Metadata
JSON: 24-1.0395603.json
JSON-LD: 24-1.0395603-ld.json
RDF/XML (Pretty): 24-1.0395603-rdf.xml
RDF/JSON: 24-1.0395603-rdf.json
Turtle: 24-1.0395603-turtle.txt
N-Triples: 24-1.0395603-rdf-ntriples.txt
Original Record: 24-1.0395603-source.json
Full Text
24-1.0395603-fulltext.txt
Citation
24-1.0395603.ris

Full Text

MOISTURE-RELATED PHYSICAL PROPERTIES AND  SELF-HEATING OF WOOD PELLETS  by  Jun Sian Lee  M.A.Sc., The University of British Columbia, 2015  A THESIS 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)  January 2021  © Jun Sian Lee, 2021 ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled:  Moisture-related physical properties and self-heating of wood pellets  submitted by Jun Sian Lee in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Chemical and Biological Engineering  Examining Committee: Shahab Sokhansanj, Adjunct Professor, Chemical and Biological Engineering, UBC Co-supervisor Anthony K. Lau, Associate Professor, Chemical and Biological Engineering, UBC Co-supervisor  C. Jim Lim, Professor, Chemical and Biological Engineering, UBC Supervisory Committee Member Gregory D. Smith, Professor, Wood Science, UBC University Examiner Stavros Avramidis, Professor, Wood Science, UBC University Examiner  Additional Supervisory Committee Members: Xiaotao T. Bi, Professor, Chemical and Biological Engineering, UBC Supervisory Committee Member Philip Evans, Professor, Wood Science, UBC Supervisory Committee Member     iii  Abstract Wood pellets may unintentionally be exposed to liquid water (rain) and water vapor (humid air) during their storage, handling, and transportation to markets. This dissertation aims to analyze the impact of incidental rain on the reduction of durability and the potential for self-heating of wood pellets. Initially, wood pellets spread in a pan were exposed to sprayed water to simulate rainfall. The durability values for wetted pellets were assessed using the standard tumbler test. The constant durability 96.5% as a dependent value vs. varying rain intensity (mm/h) and rain duration (h) were graphically presented on a two-dimensional (x-y) coordinate.   Subsequent tests were conducted to compare the durability of pellets exposed to liquid water and to water vapor. For liquid water, the dried pellets at 0% m.c. were immersed in water at 30 °C. For water vapor, the dried pellets were placed in a humidity chamber at 90% RH and temperature of 30 °C. When exposed to liquid water, the durability of wood pellets decreased from over 99% to below 80% when their moisture content increased to 20% (wb) within six minutes. When exposed to water vapor, the durability of wood pellets first decreased from over 99% to 95% when their moisture content increased from zero to 10.7% (wb) within four hours.  The temperature rise due to self heating of wood pellets was measured using a thermocouple placed in 200 ml liquid water and 100 g pellets. The temperature rise in humid air was measured by inserting a thermocouple inside a pellet placed in the humidity chamber at 95% RH and 33 °C. The maximum heat of wetting of wood pellets immersed in water was determined to be 66 kJ/kg dry mass. An average differential heat of water vapor adsorption of 403 kJ/kg water was calculated as a derivative of the heat of wetting data. By including the equation for heat of wetting and differential heat of adsorption, the mathematical heat and moisture transfer model quantified the contribution of moisture adsorption to the self-heating phenomenon in wood pellets.   iv  Lay Summary Wood pellets are a vital substitute for coal in power generation and heating oil in residential heating. Over three million tonnes of pellets were exported in 2018 from Canada to West Europe and East Asia to be used in power generation. During transportation, the physical quality of pellets can deteriorate due to incidental exposure to rainwater and humid air. My research aimed to analyze the impact of such exposure on the physical integrity of wood pellets. I found that the pellets maintained their physical integrity when exposed to a small amount of liquid water. More importantly, pellets’ durability stayed over 90% when stored in a highly humid environment for more than 10 hours. During the course of this research, I developed criteria to stop loading a ship when a combination of rain intensity and rain duration exceeds a certain value.   v  Preface This Ph.D. dissertation is divided into seven chapters and three appendices. The author, Jun Sian Lee, has performed the literature review, designed and conducted experiments, data processing and analysis, and preparation of manuscripts and thesis under the supervision of Professor Shahab Sokhansanj and co-supervision of Professor Anthony K. Lau and Professor C. Jim Lim. Versions of chapters 2, 3, and 5 are published in peer-reviewed journals and/or presented at conferences. Chapter 2 is a combination of the first two published articles in the list of journal publications below. I ran the experiments based on the advice of Professor Shahab Sokhansanj and wrote a manuscript draft, which was reviewed by Professors Anthony Lau, Jim Lim, and Tony Bi.  Chapters 3 and 5 are versions of the published articles in the list below (#3 and #5). I conducted testing and designed the experiments. Professor Anthony Lau suggested the experiment with wood chips. Professor Shahab Sokhansanj suggested the SEM imaging and helped with the creation of the figures. Professor Jim Lim edited the manuscript. Chapters 4 is a version of a published article and is number four in the list below. I conducted the experiments in a humidity chamber and wrote the first draft of the manuscript. Professor Philip Evans provided the micro-CT images and suggested further statistical analysis. Professors Shahab Sokhansanj, Anthony Lau, Jim Lim, and Tony Bi edited drafts of the manuscript and advised on experimental procedure. Journal publications: 1. Lee, J. S., Sokhansanj, S., Lau, A. K., & Lim, C. J. (2020). The Impact of Rain Exposure During Loading of Wood Pellets for Ocean Shipment – An Experimental Study. Transactions of the ASABE. In press. https://doi.org/10.13031/trans.13905  vi  2. Lee, J. S., Sokhansanj, S., Lau, A. K., & Lim, C. J. (2020). Physical properties of wood pellets exposed to liquid water. Biomass and Bioenergy, 142, 105748. https://doi.org/10.1016/j.biombioe.2020.105748 3. Lee, J. S., Sokhansanj, S., Lim, C. J., Lau, A., & Bi, T. (2019). Comparative Analysis of Sorption Isotherms for Wood Pellets and Solid Wood. Applied Engineering in Agriculture, 35(4), 475–479. https://doi.org/10.13031/aea.13238 4. Lee, J. S., Sokhansanj, S., Lau, A. K., Lim, J., & Bi, X. T. (2021). Moisture adsorption rate and durability of commercial softwood pellets in a humid environment. Biosystems Engineering, 203, 1–8. https://doi.org/10.1016/j.biosystemseng.2020.12.011 5. Lee, J. S., Sokhansanj, S., Lau, A. K., & Lim, C. J. (2020). Heats of wetting and sorption for wood pellets. Biomass and Bioenergy, 142, 105791. https://doi.org/10.1016/j.biombioe.2020.105791   vii  Table of Contents Abstract ......................................................................................................................................... iii Lay Summary ............................................................................................................................... iv Preface .............................................................................................................................................v Table of Contents ........................................................................................................................ vii List of Tables ............................................................................................................................... xii List of Figures ............................................................................................................................. xiv List of Symbols .............................................................................................................................xx List of Abbreviations ................................................................................................................. xxi Acknowledgements ................................................................................................................... xxii Dedication ................................................................................................................................. xxiii Chapter 1: Introduction ................................................................................................................1 1.1 Background ..................................................................................................................... 1 1.2 Literature Review............................................................................................................ 5 1.2.1 Feedstock properties ................................................................................................... 6 1.2.1.1 Moisture effect on properties .............................................................................. 7 1.2.1.2 Moisture-induced swelling in densified wood products ..................................... 8 1.2.2 Equilibrium moisture content and sorption isotherms ................................................ 9 1.2.3 Self-heating and thermodynamic properties ............................................................. 11 1.2.4 Liquid water absorption ............................................................................................ 14 1.2.5 Water vapor adsorption ............................................................................................. 15 1.2.6 Moisture and heat transfer ........................................................................................ 16 1.3 Thesis Statement and Research Objectives................................................................... 17 viii  Chapter 2: Liquid Phase Moisture Characteristics of Wood Pellets ......................................20 2.1 The Effect of Moisture on Physical Properties of Wood Pellets .................................. 20 2.1.1 Introduction ............................................................................................................... 20 2.1.2 Materials and methods .............................................................................................. 22 2.1.2.1 Sample preparation ........................................................................................... 22 2.1.2.2 Test methods ..................................................................................................... 24 2.1.3 Results ....................................................................................................................... 26 2.1.3.1 Surface structure ............................................................................................... 26 2.1.3.2 Pellet durability and fines content ..................................................................... 27 2.1.3.3 Bulk density ...................................................................................................... 30 2.1.3.4 Solid density...................................................................................................... 33 2.1.3.5 Bulk porosity ..................................................................................................... 35 2.1.4 Discussion ................................................................................................................. 36 2.2 Effects of Rain on the Durability of Pellets .................................................................. 38 2.2.1 Introduction ............................................................................................................... 38 2.2.2 Experimental procedure ............................................................................................ 41 2.2.3 Experimental Procedure ............................................................................................ 43 2.2.3.1 Set I tests ........................................................................................................... 43 2.2.3.2 Set II tests .......................................................................................................... 44 2.2.3.3 Actual rain events ............................................................................................. 45 2.2.4 Results and discussion .............................................................................................. 45 2.2.4.1 Set I tests ........................................................................................................... 45 2.2.4.2 Set II tests .......................................................................................................... 47 ix  2.2.4.3 Actual rain events ............................................................................................. 49 2.3 Conclusions ................................................................................................................... 51 Chapter 3: Sorption Isotherms of Wood Pellets .......................................................................52 3.1 Introduction ................................................................................................................... 52 3.2 Methodology ................................................................................................................. 55 3.3 Results and Discussion ................................................................................................. 57 3.4 Summary ....................................................................................................................... 62 Chapter 4: The Effects of Humidity on Single Pellets ..............................................................63 4.1 Introduction ................................................................................................................... 63 4.1.1 Objectives ................................................................................................................. 64 4.2 Materials and Methods .................................................................................................. 65 4.2.1 Materials ................................................................................................................... 65 4.2.2 Moisture adsorption .................................................................................................. 65 4.2.3 Physical dimensions, structure and appearance ........................................................ 67 4.2.4 Durability .................................................................................................................. 67 4.3 Results and Discussion ................................................................................................. 68 4.3.1 Surface conditions and internal structure .................................................................. 68 4.3.2 Kinetics of moisture adsorption ................................................................................ 70 4.3.3 Durability .................................................................................................................. 73 4.3.4 Pellet volume and density ......................................................................................... 76 4.3.5 Significance of the results ......................................................................................... 78 4.4 Conclusions ................................................................................................................... 80 Chapter 5: Heats of Wetting and Sorption for Wood Pellets ..................................................82 x  5.1 Introduction ................................................................................................................... 82 5.2 Definition of Heat of Wetting ....................................................................................... 83 5.3 Experimental Methods .................................................................................................. 86 5.3.1 Sample description .................................................................................................... 86 5.3.2 Water absorption ....................................................................................................... 87 5.3.3 Heat of sorption by calorimetry ................................................................................ 88 5.3.4 Scanning electron microscopy (SEM) ...................................................................... 90 5.4 Results ........................................................................................................................... 90 5.4.1 Water uptake ............................................................................................................. 90 5.4.2 Heat of wetting .......................................................................................................... 92 5.5 Discussion ..................................................................................................................... 99 5.6 Conclusions ................................................................................................................. 101 Chapter 6: Modeling of Moisture and Heat Transfer in Single Pellets ................................102 6.1 Introduction ................................................................................................................. 102 6.2 Methods....................................................................................................................... 104 6.2.1 Materials ................................................................................................................. 104 6.2.2 Experimental setup.................................................................................................. 104 6.2.2.1 Water vapor adsorption ................................................................................... 104 6.2.3 Heat and moisture transfer model ........................................................................... 106 6.3 Results ......................................................................................................................... 111 6.4 Discussion ................................................................................................................... 117 6.5 Conclusions ................................................................................................................. 120 Chapter 7: Conclusions and Recommendations .....................................................................121 xi  7.1 Overall Conclusions .................................................................................................... 121 7.1.1 Experimental conclusions ....................................................................................... 121 7.1.2 Modeling conclusions ............................................................................................. 122 7.1.3 Practical implications .............................................................................................. 123 7.2 Recommendations for Future Work............................................................................ 124 References ...................................................................................................................................126 Appendices ..................................................................................................................................148 Appendix A Additional Tables and Figures ........................................................................... 148 Appendix B Additional Information ....................................................................................... 154  Calculations for Maximum Temperature Rise from Heat of Wetting .................... 154  Model Development for Heat and Moisture Transfer of Single Pellet ................... 155  MATLAB m-script for the Heat and Moisture Transfer Model ............................. 158  X-ray Micro-computed Tomography of a Compressed Softwood Pellet ............... 161 Appendix C Additional Work ................................................................................................. 164  Moisture Adsorption from Ends and Circumference Sides of Single Pellets ......... 164  xii  List of Tables Table 1.1 Threshold values of the most important pellet parameters – ISO 17225-2 .................... 5 Table 1.2 Models and their coefficients representing EMC-ERH at 25°C. Equilibrium moisture content (EMC) is expressed in decimal dry basis and ERH in decimal. ...................................... 10 Table 1.3 Equilibrium moisture content (EMC) in decimal dry basis at corresponding equilibrium relative humidity in decimal for the models of the five sources. .............................. 11 Table 1.4 Differential heat of sorption at zero moisture (hs)0 from literature. Most data were adapted from Simón (2015). * The authors do not specify the species analyzed. NA is data not available. ....................................................................................................................................... 12 Table 1.5 Potential pathway to a self-ignition event in bulk wood pellets in an air-restricted ship’s hold or a silo, based on the description in Kubler (1990). ................................................. 13 Table 1.6 The saturation moisture content, the time to saturation, and the average liquid water absorption rate of several untreated materials near room temperature. ........................................ 14 Table 1.7 Moisture adsorption behaviors of several materials in 50-90% relative humidity (RH)........................................................................................................................................................ 15 Table 2.1 Canadian Standard for commercial trading of wood pellets (Wood Pellet Association of Canada, 2016). This standard complies with the ISO 17225 International Standard for wood pellets. *ar stands for as-received basis and is equivalent to wet mass basis. ............................... 21 Table 2.2 The average properties of the three types of wood pellets after oven-drying. The percentages are on a dry mass basis. For diameter and length, the number of samples (N) is 100 for each type. For fines content, durability, bulk density, N is 3. The spread of data for length and bulk density was given as the standard deviation (±SD). ............................................................. 24 xiii  Table 2.3 The mass of distilled water sprayed onto the 1500 g of wood pellets from each pellet type. ............................................................................................................................................... 24 Table 2.4 Values of the parameters of the sigmoid Eq. 2.8. ......................................................... 32 Table 2.5 Spraying rate (mL/min) converted from rain intensity (mm/h) .................................... 45 Table 3.1 Models and their coefficients representing EMC-ERH at 25C.[a] ............................... 59 Table 3.2 The computed results for a two-sample t-test assuming equal variance to compare the difference between Henderson equation’s data for wood pellets and Avramidis (1989)’s equation for solid wood. .............................................................................................................................. 61 Table 4.1 The results from two-sample t-test assuming equal variances to test the differences among the durability data for humidity chamber (HC) and saturated salt solution (SS) for three moisture content ranges: (a) 0 – 0.10 (db), (b) 0.10 – 0.15 (db), and (c) > 0.15 (db). Critical is obtained at alpha value of 0.01. .................................................................................................... 75 Table 5.1 Summary of pellets’ moisture content, initial and maximum temperatures of the pellets-and-water mixture, and the calculated heat of wetting. .................................................... 93 Table 5.2 Differential heat of sorption, hs calculated from linear (Eq. 5.14) and exponential (Eq. 5.15) using Eq. 5.10. Mi and Mf are initial and final moisture contents, respectively. ................. 96 Table 6.1 The variables as a function of the temperature of pellet T and temperature of environment Tenv and water vapor partial pressure p. ................................................................. 109 Table 6.2 Parameters for the lumped capacitance heat and moisture transfer model (Eqs. 6.3 and 6.4). ............................................................................................................................................. 110 Table 6.3 The values of moisture diffusion coefficient, DH2O-Dry Air and heat diffusion coefficient, λair in a few cases. ....................................................................................................................... 111  xiv  List of Figures Figure 1.1 Three softwood pellets (a) before and (b) after exposure to water-saturated air at 45 °C for 30 minutes. The scale on the ruler on the right has an 1 mm interval. The images are adapted from Lee and Lube (2017). ............................................................................................................. 9 Figure 1.2 Flow diagram for my research topics and their relationships. ..................................... 19 Figure 2.1 The three types of pellets used: (a) residential hardwood pellets, (b) residential softwood pellets, (c) industrial softwood pellets. The scale is in centimeters. ............................................. 23 Figure 2.2 A softwood pellet, 5 minutes after wetting by droplets of liquid water. The contrast of the image was increased to show localized swelling. ................................................................... 26 Figure 2.3 Microscopic surface photograph of the horizontal cross-section of a softwood pellet before wetting (a) and after wetting (b) by a droplet of liquid water. The swelling of the internal parts of the wetted pellet is evident............................................................................................... 27 Figure 2.4 Durability of three types of wood pellets vs. moisture content. The pellets were initially at 0% m.c. The gain in m.c. was the result of spraying water on the pellets. The lines are quadratic equations fitted to the measured data. The number of repetitions is three. The error bars represent maximum and minimum of the three repetitions. ......................................................................... 28 Figure 2.5 Fines content of three types of wood pellets vs. moisture content. The lines are quadratic equations fitted to the measured data. Only one sample was tested for fines content for every moisture content. The error bars represent maximum and minimum of the three repetitions. ..... 29 Figure 2.6 Bulk density of three types of wood pellets vs. moisture content. The lines are sigmoid equations fitted onto the measured data. The number of repetitions is three. The error bars represent maximum and minimum of the three repetitions. ......................................................................... 31 xv  Figure 2.7 Specific volume of three types of wood pellets vs. moisture content. Specific volume is calculated as one over bulk density. The error bars represent maximum and minimum of the three repetitions. ..................................................................................................................................... 32 Figure 2.8 Solid density, ρS of residential hardwood and softwood pellets in circle dots and lines for the fitted lines given by Eqs. 2.9-2.11. Hypothetical line (Eq. 2.9) assumes water with a density of 1000 kg/m3 is added into the wood with a cell-wall density of 1400 kg/m3. This test was done on whole pellets at moisture content below 20% and wet swollen pellet particles at moisture contents above 20%. ..................................................................................................................... 34 Figure 2.9 Porosity of residential hardwood and softwood pellets in circle dots and lines for the fitted line. ...................................................................................................................................... 36 Figure 2.10 Loading of wood pellets onto a bulk carrier with an open hatch (Curci, 2011) ........ 39 Figure 2.11. A rectangular tray, which is filled with pellets (left); Tumbler tester (right). .......... 43 Figure 2.12 Relationship between fines content and durability with the amount of water sprayed on 2000 g of wood pellets having an initial moisture content of 5.5% (wb). Rain intensity is 1 mm/h. This is only one test with no repetition.............................................................................. 47 Figure 2.13 Durability as a function of moisture content for the water-sprayed pellets. ............. 48 Figure 2.14 The cut-off curve of rain intensity versus exposure time to achieve 96.5% durability........................................................................................................................................................ 49 Figure 2.15 Durability as a function of moisture content for the rain-wetted pellets. .................. 50 Figure 3.1 Two of the six containers, which contained salt solutions and samples. .................... 56 Figure 3.2 Comparing adsorption (ad) and desorption (de) of equilibrium moisture content (EMC) of wood pellets at 5C, 25C, 35C after 28 days versus relative humidity (ERH). As expected, the EMC for desorption (soil fill markers) is higher than the EMC for adsorption (blank markers). xvi  Visually, the EMC for the higher temperature (35C) was slightly lower than EMC for the lower temperature (5C). ........................................................................................................................ 58 Figure 3.3 Moisture content vs. relative humidity (EMC-ERH) data points from the present study (25C adsorption tests) and the curves drawn from selected EMC-ERH equations available for solid wood (Avramidis, 1989) and wood pellets (Hartley and Wood, 2008). The solid line is Henderson’s equation fitted to my experimental data. The fitting coefficients are given in Table 3.1.................................................................................................................................................. 60 Figure 4.1 Espec LHU-113 benchtop temperature/humidity chamber. ........................................ 65 Figure 4.2 The diagram of the single pellet durability tester. A small metal box is attached to the wrist-action shaker (Schilling et al., 2015). .................................................................................. 68 Figure 4.3 (a) as-received pellets and groups of five wood pellets after exposure to the humidity of 95% RH and the temperature of 30°C, (b) for 1 hour, (c) for 2 hours, (d) for 4 hours, (e) for 7 hours, (f) for 13 hours, (g) for 18 hours, and (h) for 24 hours. ..................................................... 69 Figure 4.4 A two-dimensional image of cross-sections through an untreated wood pellet. Cracks penetrated deep from the surface into the body of the pellet. ....................................................... 70 Figure 4.5 Moisture content of wood pellets, expressed in decimal dry basis (db), at seven exposure times: 1, 2, 4, 7, 13, 18, 24 hours, showing data for 20 replicates for each time interval. The solid curve represents Page’s model (Eq. 4.2) fitted to the data. ........................................................... 71 Figure 4.6 Moisture content (db) versus relative humidity at the temperature of 25°C. The total number of data points is 36. Data were adapted from my previously published data (Lee et al., 2019). The fitting curve represents Eq. 4.3. .................................................................................. 72 Figure 4.7 Single pellet durability (Dsingle) vs. moisture content in decimal, dry basis (db) for samples exposed to 95% RH and 30°C in a humidity chamber (filled circles - HC) and exposed to xvii  atmospheres above saturated salt solutions (open circles - SS). The total number of data points is 196. The two fitting curves represent Eqs. 4.4 and 4.5 for HC and SS, respectively. .................. 74 Figure 4.8 Volumetric swelling in percent of the initial volume (%initial volume) vs. moisture content. The total number of data points is 160. The fitting curve represents the power-law equation. ........................................................................................................................................ 76 Figure 4.9 Density of wood pellets vs. moisture content. The total number of data points is 160. The fitting curve represents a 2nd order polynomial equation. .................................................... 78 Figure 4.10 The relationship between volumetric swelling (%) and durability, based on average values. ........................................................................................................................................... 80 Figure 5.1 Water in wood phase diagram, adapted from Skaar (Skaar, 1988). hv is the enthalpy of vaporization or condensation. hs is the differential heat of sorption. The shaded area above the bound water curve is the heat of wetting for the wood at a moisture content of M. Mf is the fiber saturation moisture content below which water exists only in the cell wall of wood. ................. 85 Figure 5.2 A calorimeter setup used to measure temperature rise in pellets-and-water mixtures. Thermocouples T1 and T3 measure the temperatures of pellets and water mixture in flasks 1 and 2. T2 measures the air temperature around the two flasks. The insulated box around the two flasks ensures that air temperature stays relatively constant. .................................................................. 89 Figure 5.3 The liquid water uptake of ~10 g dried wood pellets over six minutes. After six minutes, six to 10 g of water was absorbed. Six repetitions were made at each time interval. The solid curve represents the fitted Page model to the average data. ................................................................... 91 Figure 5.4 An example of temperatures of wood pellets and water mixtures in the two flasks and the air temperature near them. The initial moisture content was 0.05 (db). ................................. 92 xviii  Figure 5.5 Heat of wetting (Q) for moisture contents ranging from 0.005 to 0.158 (db). The error bars represent the minimum and maximum values at each moisture content. A sloped line, Eq. 5.13, fitted through the data has a better R2 (=0.95) than an exponential curve, Eq. 5.14 (R2=0.80). At zero moisture content, the line approaches the vertical axis at Q0=66.4 kJ/kg or the maximum heat of wetting. The line approaches the x-axis at M=0.165; that may be considered the fiber saturation moisture content for the wood pellet. ........................................................................... 95 Figure 5.6 The maximum heat of wetting of oven-dried western red cedar wood chips and wood pellets. The error bars represent the minimum and maximum values. The number of repetitions is 10 for each sample. A paired t-test (α=0.05) showed that the difference between maximum heats of wetting for wood chips and pellets was not statistically significant (P-value = 0.40). ............ 97 Figure 5.7 (a) Horizontal cross-section of the as-received wood chip. Note the large open cell lumens in earlywood (central part of the image). Voids are much smaller in latewood in the left and right of earlywood. (b) the edge of the as-received wood pellet. The wood chip contains a porous cellular structure, whereas the structure of the wood pellet is not visibly porous except having cracks. ............................................................................................................................... 98 Figure 5.8 Temperature plots of oven-dried wood pellets and oven-dried wood chips after 50 g samples of each were immersed in water. The initial moisture content was 0.0 (db). ................. 99 Figure 6.1 Schematic diagram of heat and natural convection of moisture within and around a ship’s hold full of wood pellets. .................................................................................................. 103 Figure 6.2 The experimental setup used to measure the temperature change of a wood pellet on exposure to stagnant air. ............................................................................................................. 106 Figure 6.3 Assumed configuration of a wood pellet and the distribution of temperature T, partial pressure of water vapor p, and amount of water vapor adsorbed q within the wood pellet. ...... 107 xix  Figure 6.4 Moisture adsorption rate of water vapor for dried softwood pellets measured on exposure to a humid atmosphere at a relative humidity of 0.95 (decimal) in a humidity chamber (HC) at temperatures of 30, 40, and 50 °C and in a water bath (WB) at 33 °C. The curves are calculated using the heat and moisture transfer model (Eqs. 6.3 and 6.4) at a relative humidity of 0.95 (decimal) and at temperatures of 30 °C and two different moisture transfer coefficients, kc...................................................................................................................................................... 112 Figure 6.5 Adsorption isotherm of water vapor on softwood pellets measured at 25 °C (points), adapted from Lee et al. (2019). The three curves are Henderson model and the two approximated linear relationships. ..................................................................................................................... 113 Figure 6.6 Comparison of experimentally measured and model-predicted temperature changes at 33°C and 95% RH for wood pellets. The dots are experimental data. The solid line is the predicted temperatures when hs = 403 kJ/kg (R2 = 0.9128). The dashed line is for hs = Eq. 6.8 (R2 = 0.9175)...................................................................................................................................................... 116  xx  List of Symbols ρ  density     kg/m3 ϕ  porosity    dimensionless ε  internal void fraction   dimensionless  xxi  List of Abbreviations ar  as-received basis. Practically, it is equivalent to wet mass or total mass basis. db  dry mass basis ERH  equilibrium relative humidity hs  differential heat of sorption of water hv  latent heat of vaporization of water M  moisture content in decimal dry basis m.c.  moisture content in wet basis in percentage (%) p  partial pressure of water vapor in the atmosphere Q  heat of wetting rh  relative humidity in decimal RH  relative humidity in the percentage of saturation water vapor pressure T  temperature t  time wb  wet mass basis    xxii  Acknowledgements I thank my wife, Nikki, for the unconditional support she provided for the past six years in my arduous journey to obtain a doctoral degree. I owe a debt of gratitude to my father and mother, who have encouraged me to obtain a postgraduate degree. I offer my most tremendous gratitude to my academic supervisor, Dr. Shahab Sokhansanj, whose wisdom and trust has allowed me to continue, despite my doubt about my ability to do so. I owe particular thanks to my co-supervisor, Dr. Anthony Lau, who introduced me to the Biomass and Bioenergy Research Group and the group's outstanding students. Thank you to Dr. Jim Lim, who has provided me with useful life advice when I least expected them. I am grateful to Dr. Tony Bi and Dr. Philip Evans, who provided me with critical guidance on my thesis. I would like to acknowledge the contribution of Mr. Vaughan Bassett, Senior Vice President of Pinnacle Renewable Energy Inc., and the current President of the Wood Pellet Association of Canada to my research. Mr. Bassett introduced me to the complexity of handling pellets in the presence of rain and provided the impetus for embracing the complex topic of self-heating phenomena.  xxiii  Dedication To my father, whose dream to obtain a doctoral degree is being passed onto his sons, and to my young daughter, whom I would not ask to get a doctoral degree. 1  Chapter 1: Introduction 1.1 Background The global consumption of wood pellets has grown rapidly in the past decade, reaching 35 million tonnes in 2018 (Proskurina et al., 2019). Wood pellets are primarily transported internationally using bulk carriers, which can carry up to 80,000 metric tonnes of wood pellets in a single trip (Melin, Svedberg, & Samuelsson, 2008). For a Panamax-size bulk carrier, the volume of its hold is roughly 10,000 m3, which can hold 5,000 to 6,000 tonnes of wood pellets. A Panamax-size bulk carrier usually has 7 to 10 holds (Ventura, 2008), which brings the total capacity of the ship to 60,000 to 80,000 dry tonnes.  The loading of pellets onto a bulk carrier is done using a radial loader, which drops the pellets onto any location selected by the loading operator (Fibreco Export Inc., 2016). Therefore, it can be assumed that the surface of the bulk of wood pellets in the ship’s hold is relatively uniform. In an advanced loading terminal, which is common in West European and North American ports, conveyor belt radial loaders are used to continuously load materials into the bulk carrier hold. The cargo is moved from the terminal storage facility by rotating bucket loaders and is lifted onto a moving belt, which transports the cargo into a loader and then into the ship’s holds (BIMCO, 2014). In other cases, when continuous loading facilities are not available, cranes and clams are used in place of marine loading arms. The cranes carry buckets of the materials from a storage pile into the ship’s holds.  In British Columbia, wood pellets are produced mostly in the Prince George area along the Canadian National (CN) Railway lines. Following their production, pellets are loaded onto 100-tonne covered hopper rail cars. The loaded cars travel 500 - 700 km south to either Port of Vancouver or 200-300 km west to the Port of Prince Rupert. Once at the port, the contents of the 2  cars are unloaded from the bottom of the rail cars inside a covered building (Tumuluru et al., 2010). The pellets are then transported in covered conveyors to a number of upright steel bins or flat storage structures. The pellets remain in the storage usually for less than a month. In preparation for ocean transport, the pellets are transported on a series of covered belt conveyors to ship loaders. The pellets move on an inclined conveyor to the top of a vessel when the pellets drop onto the holds of the vessel through a specially designed feeder (choke feeder) to minimize the drop height and to prevent breakage. The nominal capacity of a ship’s hold can reach 5000 tonnes of pellets. In spite of strict safe loading practices, unpredictable weather events may introduce rainwater to wood pellets during a ship loading operation. An equipment manufacturer (Boss Tek, 2013) recommends water spraying as a means of controlling dust during the loading of vessels in the Southern U.S. However, the authors are not aware of water spraying practices in BC pellet loading terminals.  The coast of BC, where two major pellet exporting ports, Port of Prince Rupert and Port of Vancouver, are located, is an area with high precipitation. Regularly wood pellets and other granular materials like grain and coal are loaded in the open and into open-vessel holds, without any protection from rain (Melin et al., 2008). Prince Rupert is known to be the municipality that receives the lowest sunshine in Canada, with only 100 days of sunshine with an annual rainfall of over 2400 mm (Environment Canada, 2001). Vancouver is also known for its high annual rainfall of over 1100 mm (Environment Canada, 2011). The annual rainfall amount on the coast of BC is expected to increase as the climate changes (Mbogga, Hamann, & Wang, 2009).   Environment Canada’s climate glossary (2015) categorizes each rain event based on precipitation rate or rain intensity, expressed in millimeters per hour. An event of rain can be categorized as “very light,” where raindrops do not completely wet a surface, “light” with a rain 3  intensity less than 2.5 mm per hour, “moderate” with a rain intensity between 2.6 mm and 7.5 mm per hour, and “heavy” with a rain intensity above 7.5 mm per hour. The quality of pellets in terms of durability would depend on the severity of the rain event during loading. No shipmaster will allow loading during moderate or heavy rain events. Due to the regular rainy conditions in the coastal region of British Columbia (BC), wood pellets periodically become wet during loading and unloading operations if the operation coincides with rainfall. When wetted with water droplets, pellets swell and disintegrate, reverting to wood fines. During the process of wetting, heat is released as heat of sorption, which may encourage a higher level of off-gassing and odor emission (Kuang et al., 2009). The oxidation of disintegrated wood pellets by atmospheric oxygen may induce self-heating reactions (Guo, 2013).   The shipmaster decides when loading operations should stop in the event of rain ((Bacchioni, 2008)Vaughan Bassett, Personal Communication 2014). Therefore, the decision to load pellets can be subjective, depending on the definition of a rain event. The responsibility for pellets falls into the hands of the shipmaster or captain until pellets are unloaded at the destination. Different ship masters allow different degrees of exposure to precipitation during loading, which will result in varying degrees of wetness and moisture content of wood pellets. Previous research at UBC (Lee et al., 2019) has demonstrated that pellets with an initial moisture content of 5 - 6% (wet mass basis) wetted to a moisture content of 8 - 9% did not lose their durability substantially. However, wetted pellets, with a moisture content of 15% or higher, swell and disintegrate due to the breakage of the physical adhesion among wood particles inside a pellet (Mantanis, 1994). A similar observation was reported by Fasina and Sokhansanj (1992) for alfalfa pellets, where the durability of the pellets increased when moisture content increased from 5% to 8% but decreased as moisture content increased further.  4  Any delay in wood pellet loading operations can cause the pellet producers over $20,000 per day in storage and handling fees at the loading port at Metro Vancouver (Vancouver Fraser Port Authority, 2020). The damage to the quality of wood pellets and relevant safety concerns (off-gassing, odor emission, and the potential of self-heating) by rain should be weighed against the costs and the loss of revenue as a result of the delay in loading. Therefore, research has to be done to analyze the relation between rain exposure and the pellet’s uptake of liquid water and predict pellet quality during subsequent marine transport. 1.1.1 Definition of durability Wood pellets experience impact fragmentation and abrasion attrition when they are handled at each step of the supply chain from production plants to ports then to the final consumers (Deng, Zhang, & Che, 2013). Pellets are loaded into freight railcars after production, and bulk carrier ships’ holds at ports through a combination of mechanical grabbing devices and pneumatic conveyors (Dafnomilis et al., 2018). Pellets are dropped from more than 10 m and are tossed around at speed more than 10 m/s in pneumatic conveying systems during handling (Oveisi et al., 2013). As a result, they may be fragmented and broken into small pieces. To quantify the pellets’ resistance to fragmentation and breakage, an indicator, named durability, is formulated as: Durability = weight of unbroken pellets after testweight of pellets before test × 100%  (1.1) ISO 17831-1 defined unbroken pellets as the pellets that stay above a 3.15 mm round sieve.  In the literature, five notable test devices, which are tumbler tester, Holman tester, Ligno tester, Dural tester, and single pellet durability tester, are designed and employed by several different groups of researchers to determine the durability pellets (Kaliyan & Vance Morey, 2009; Oveisi-Fordiie, 2011). In this thesis, the tumbler tester is used to test the durability for bulk 5  amounts of pellets (500 g) because the international quality standard ISO 17225-2 requires the use of a tumbler. Tumbler tester is pictured in Figure 2.11 in Chapter 2. When only a small amount of pellets are available, as in Chapter 4, a single pellet durability tester, which was developed at the University of British Columbia, is used (Schilling et al., 2015).  1.2 Literature Review The quality of commercially traded wood pellets is defined internationally by ENplus standards (European Pellet Council, 2015) for residential and institutional heating uses, while Industrial Wood Pellet Buyers (IWPB) standards are used for pellets destined for industrial power production and heat generation (Verhoest & Ryckmans, 2012). The equivalent Canadian standard to ENplus standards is the CANplus quality standard (Wood Pellet Association of Canada, 2016). In Canada, wood pellets are mainly made from sawdust, shavings, and logging residues (Hilton and Murray 2013). These materials are ground, compressed, and extruded through a die system, which produces a dense cylindrical-shaped pellet (Tumuluru et al., 2011).  Table 1.1 gives the ENplus or ISO 17225 Part 2: Graded wood pellets (2015) quality standard (Grades A1, A2, and B), as well as the Industrial Wood Pellet Buyers (IWPB) (2012) standards (Grades I1, I2, I3). The moisture content must be less than 10% as received (ar) or wet mass basis (wb). The mechanical durability of industrial wood pellets must be at least 96.5%.  Residential wood pellets certified by the ENplus standard must have fines content below 1%. Pellets for industrial uses can be fines content up to 6% as fines are produced in large amounts during the handling and transportation of pellets. Table 1.1 Threshold values of the most important pellet parameters – ISO 17225-2 Property Unit ENplus Certification Standard (2015) Industrial Wood Pellet Buyers (IWPB) Standard (2012) I1 I2 I3 6  A1 A2 B Diameter mm 6 ± 1 or 8±1 6 to 8 6 to 10 6 to 12 Length mm 3.15 ≤ L ≤ 40 6     L ≤ 40 6     L ≤ 40 6     L ≤ 40 8     L ≤ 50 8     L ≤ 50 8     L ≤ 50 Moisture % of weight, ar ≤ 10 Total ash % of weight, dry ≤ 0.7 ≤ 1.2 ≤ 2.0 ≤ 1.0 ≤ 1.5 ≤ 3.0 Mechanical Durability % of weight, ar ≥ 98.0 ≥ 97.5 ≥ 97.5 ≥ 96.5 Fines content  (< 3.15 mm) % of weight, ar ≤ 1.0 ≤ 4.0 ≤ 5.0 ≤ 6.0 Additives % of weight, ar ≤ 2.0 Net calorific value, ar MJ/kg ≥ 16.5 ≥ 16.5 Bulk density kg/m3, ar 600 ≤ BD ≤ 750 Temperature of pellets °C ≤ 40 ≤ 60.0 ± 1  1.2.1 Feedstock properties Typically, during pelletization, the wood particles reach temperatures around 100°C due to compression forces and friction. Thermal compression causes lignin in wood to plasticize and act like a glue that binds wood particles together (Lam, 2011; Mani et al., 2003; Stelte et al., 2011; Tooyserkani, 2013). Wood pellets produced in Canada are 6-mm in diameter, made from wood particles with a mean particle size of 1 mm to 2 mm (Jensen, Temmerman, & Westborg, 2011). Compositionally, wood pellets are essentially comprised of the same chemical components of wood: cellulose, hemicelluloses, lignin, and extractives. But mechanically, wood pellets behave differently than solid wood (Zaini et al., 2008). Lee et al. (2015) determined the calorific value of 7  pellets after a period of storage under constant temperature conditions and found that the calorific value increased by 2 to 3% mainly due to the moisture loss from the pellets.  Solid wood often refers to dimensional lumber milled from sawing off logs (Hoadley, 2000). Engineered wood, such as particleboards and wood pellets, is composed of wood particles pressed together. Solid wood contains cellular fibers throughout pieces of lumber (Ross, 2010) whereas, in engineered wood, woody elements are physically and chemically modified due to comminution, heat, and pressure applied during the compaction process of the product. Solid wood products and wood pellets are made from various hardwood (angiosperm, mostly deciduous) species and softwood (gymnosperm, conifer) tree species. In this thesis, solid wood exclusively refers to untreated lumber. 1.2.1.1 Moisture effect on properties Similar to all dried woody biomass, wood pellets are hygroscopic (Hartley & Wood, 2008). As the wood pellets attract and absorb vapor and liquid water until their moisture content exceeds 10% (wb), they swell and disintegrate into fine particles. Fine wood dust generated from the disintegration of wood pellets is considered a health hazard (Enarson & Chan-Yeung, 1990). Also, under appropriate conditions of containment, the dust may give rise to a dust explosion (Amyotte & Eckhoff, 2010). Because of this safety risk, researchers (Craven et al., 2015; Hashemi, 2013) have attempted to reduce the hygroscopicity of pellets by coating pellets with hydrophobic substances, primarily oils, with varying results. In British Columbia, the manufactured wood pellets are temporarily stored in a silo before being transported to either the Port of Vancouver (Fibreco Export Inc. Terminal) or the Port of Prince Rupert (Westview Wood Pellet Terminal) by train. During transport and at loading ports, the wood pellets may be exposed to high-humidity ambient air and precipitations. During 8  transportation to overseas markets, temperature fluctuations or swings en route cause moisture migration through biomass piles such as grain and eventual condensation on the inside walls of the hold and onto the cold grain (Khoshtaghaza et al., 1999). The hygroscopicity of biomass and food materials have been measured by its moisture uptake rate in a  humidity chamber (Igathinathane et al., 2009; Yu, et al., 2014) and by liquid water absorption rate by immersing pellets in water (Craven et al., 2015; Turhan et al., 2002).  1.2.1.2 Moisture-induced swelling in densified wood products Similar to oriented strand board and medium-density fibreboards, the wood particles in wood pellets can be compressed three to five times their initial bulk density (Lube, 2016). This compression introduces compressive strains in the finished products. When the pellets come in contact with moisture, some of these strains are released, resulting in thickness swelling. This swelling may be more pronounced at the two edges of a pellet (Carll & Wiedenhoeft, 2009).  Wood pellets are held together by natural adhesives such as lignin and sugars (Tang et al., 2018). These adhesives resist the swelling forces that lead to thickness swelling. In other words, the two forces, which are released when pellets adsorb moisture, are compressive strain and swelling stress.  In bonded wood products, such as wood pellets, Carll and Wiedenhoeft (2009) noted that for equilibrium moisture contents (EMC) corresponding to zero to 70% relative humidity, the thickness swelling of densified wood materials is roughly similar to swelling of the denser wood species in the tangential direction, and the swelling is mostly recoverable. However, at EMCs corresponding with 80% RH or higher, the swelling is more than regular wood and becomes non-recoverable. For the case of wood pellets, as they adsorb moisture, the resulting swelling slowly increases the stresses in the bonds between wood particles without breaking the bonds. At a 9  threshold moisture content value, e.g. EMCs corresponding with 80% RH, the bonds break. The built-up swelling stress and compressive strain are released in a sometimes explosive manner, as shown in Figure 1.1, when the pellets’ volume increases by two to three times its initial volume.  Figure 1.1 Three softwood pellets (a) before and (b) after exposure to water-saturated air at 45 °C for 30 minutes. The scale on the ruler on the right has an 1 mm interval. The images are adapted from Lee and Lube (2017).  1.2.2 Equilibrium moisture content and sorption isotherms Table 1.2 describes the sorption isotherm equations, which are widely used in the literature to model the woody biomass. All the equations, except the linear model, are parabolic in nature and can be reduced to a mathematically equivalent form. Zelinka et al. (2018) recommended using a so-called “ABC isotherm” to replace all these isotherms. The ABC isotherm equation is ERH/EMC = A(ERH)2 + B(ERH) + C, where ERH stands for equilibrium relative humidity in decimal, EMC stands for equilibrium moisture content in decimal dry basis, and A, B, and C are the constant parameters of the isotherm equation.  However, the different forms of equations provide different interpretations of the moisture adsorption behavior of the material at equilibrium. Henderson and Day-Nelson equations are empirical models, which are purely used for curve fitting purposes. The Guggenheim–Anderson-de Boer (GAB) equation, given in Appendix A, describes the moisture adsorption as a multi-layer phenomenon. The Hailwood–Horrobin (H-H) describes the moisture adsorption as interactions 10  between the adsorbent as a solute and water as a solvent. Several other important sorption isotherm equations are given in Appendix A. Table 1.2 Models and their coefficients representing EMC-ERH at 25°C. Equilibrium moisture content (EMC) is expressed in decimal dry basis and ERH in decimal. Material Sources Model EMC-ERH equation  Coefficients R2 Softwood Pellets (Pine-Spruce- Fir) Lee (2019) Henderson EMC= [1Kln(1-ERH)]1C K = -0.01547 C = 1.7105 0.968 Softwood Pellets  Hartley and Wood (2008) Linear EMC = K∙ERH K = 0.156 0.998 Solid Wood – Sitka Spruce Avramidis (1989) Day-Nelson EMC = [1Kln(1-ERH)]1C where K=ATB; C=DTE A = -0.34E-16  B = 5.98 D = 0.30E3 E = -0.93 0.996 Solid Wood - Sitka Spruce Simpson (1971)  H-H EMC = 18W( K1∙K2∙ERH1+K1∙K2∙ERH+K2∙ERH1-K2∙ERH )  K1 = 5.617 K2 = 0.745 W = 252 0.979 Solid Wood - Willow Krupińska et al. (2007) Peleg EMC =  A∙(ERH)B+C∙(ERH)D A = 0.1284 B = 0.4135 C = 1.4006 D = 11.1689 0.999   Table 1.3 illustrates the quantitative differences of equilibrium moisture content in decimal dry basis (db) for relative humidities ranging from 0.10 to 0.99. In terms of equilibrium moisture contents, softwood pellets behave similarly to untreated spruce wood pieces but differently compared to untreated hardwood (willow) pieces. Hartley and Wood (2008) used a linear equation to fit the equilibrium moisture content of softwood pellets, which was measured at relative humidities ranging from 0.10 to 0.80 (decimal). Their measured equilibrium moisture content was lower than that of published softwood species data because they assumed the pellets achieved equilibrium after eight days.  11  Table 1.3 Equilibrium moisture content (EMC) in decimal dry basis at corresponding equilibrium relative humidity in decimal for the models of the five sources. ERH Lee (2019) Softwood pellets Hartley and Wood (2008) Softwood pellets Avramidis (1989) Solid wood- Sitka spruce Simpson (1971) Solid wood-Sitka spruce Krupińska et al. (2007) Solid wood-willow 0.10 0.031 0.016 0.029 0.027 0.050 0.20 0.048 0.031 0.048 0.045 0.066 0.30 0.063 0.047 0.066 0.060 0.078 0.40 0.077 0.062 0.083 0.075 0.088 0.50 0.092 0.078 0.102 0.091 0.097 0.60 0.109 0.094 0.123 0.109 0.109 0.70 0.128 0.109 0.148 0.131 0.137 0.80 0.151 0.125 0.179 0.160 0.233 0.90 0.186 0.140 0.227 0.201 0.555 0.99 0.279 0.154 0.361 0.258 1.380  1.2.3 Self-heating and thermodynamic properties Guo et al. (2013) developed two equations (Eqs. 1.2 and 1.3) which relate the heat capacity, Cp, and effective thermal conductivity, λs of wood pellets to moisture content (m.c.) in percentage wet mass basis to describe the thermal characteristics of BC wood pellets. Cp = 1.01 + 0.032 (m.c.)    (1.2) λs = 0.219 + 0.01 (m.c.)    (1.3) Maximum differential heat of sorption at zero moisture content of several wood species is given in Table 1.4. For softwood species, such as Spruce (Picea) and Pine (Pinus), the maximum differential heat ranges from 1000 to 2000 kJ/kg water, with an average of ~1200 kJ/kg water.  12  Table 1.4 Differential heat of sorption at zero moisture (hs)0 from literature. Most data were adapted from Simón (2015). * The authors do not specify the species analyzed. NA is data not available.   Species  (hs)0 (kJ/kg water) Min Max Average Stamm and Loughborough (1935) Picea sitchensis 1089 1128 1108 Babbitt (1942) * NA NA 1106 Rees (1960) * NA NA 1231 Browning (1963) * 1130 1381 1256 Weichert (1963) Picea sp., Fagus sp. 804 949 877 Stamm (1964) Pinus sp., Acer sp. NA NA 1213 Nikitin (1966) * 1172 1256 1214 Peralta et al. (1997) Phyllostachys bambusoides 800 1200 1000 Lestander (2008) Pellets made from Picea Sp, Pinus sp. 808 1607 1208 Meze`e et al. (2008) Guibourtia tessmannii, G. pellegriniana NA NA 2128 Ouertani et al. (2011) Phoenix dactylifera NA NA 1500 Tagne et al. (2011) Triplochiton scleroxylon, Diospyros ebenum 1628 3489 2559 Ouertani et al. (2014) Pinus banksiana, Phoenix dactylifera 2407 2778 2592 Simón et al. (2015) Abies alba 1295 1444 1369  Larsson et al. (2012) first reported the heating response of wood pellets in a silo over a few months. They observed that the temperature of bulk wood pellets in a silo increases primarily due to heat from solar radiation. The role of the heat of condensation and sorption in inducing self-heating phenomena in materials such as coal, lignocellulosic wood, and agricultural products have long been acknowledged. A summary of the potential pathway to self-ignition events and sources of heat in bulk pellets is given in Table 1.5. The oxidative reactions in biomass pellets and coal have been extensively studied by many authors (Guo et al., 2014; Larsson et al., 2017; Nelson & Chen, 2007). However, the role of the heat of moisture adsorption as a source of heat for wood pellets has not been studied previously.  13  In his review of the processes which generate heat in organic materials, Kubler (1990) stated that low moisture lignocellulosic materials as hygroscopic substance release adsorptive heat when adsorbing water, especially in the form of water vapor with its large latent heat. Humidity in the air also catalyzes abiotic oxidation of lipids and sugars, where oxygen gas in the air oxidizes organic materials to produce various gaseous carbon compounds, primarily carbon monoxide, carbon dioxide, and water, at temperatures below 80°C (Kim, Kim, & Lee, 2014). Table 1.5 Potential pathway to a self-ignition event in bulk wood pellets in an air-restricted ship’s hold or a silo, based on the description in Kubler (1990). Temperature Room temp. – 100 °C 100 – 200 °C > 200 °C Source of heat Non-reactive • Sunlight, engine, seawater • Latent & adsorption heat of moisture from the air and water Reactive • Biotic oxidation (enzyme-catalyzed) • Abiotic oxidation (alkali metal-catalyzed) • Hydrolysis (Moisture catalyzed reaction) • Slow pyrolysis (heat generation from pyrolysis drops as temperature increases over 200 °C) Without the introduction of air • Slow pyrolysis  With the introduction of air • Ignition • Combustion • Gas (CO, CO2) absorption onto char particles Comments • The heat released from moisture adsorption is the primary reason for temperature rise at temperatures below 60°C. • Low-temperature auto-oxidation (both biotic and abiotic) leads to oxygen depletion in the air. • More heat is released by hydrolysis at the low 100 °C range. • Then, depending on reaction conditions, pyrolysis may produce or consumes heat as temperature increases. • Since pyrolysis is endothermic, the temperature may slowly decrease, given enough time. • Ignition only occurs if air is re-introduced into the system through the opening of a hatch or ventilation.   14  1.2.4 Liquid water absorption Table 1.6 lists the saturation moisture content and other associated parameters of several organic materials. In general, the maximum saturation moisture content after immersing wood pieces into liquid water is in the range of 1.09 to 1.70 for conditions close to atmospheric pressure and temperatures. Depending on species and the level of processing, it takes 200 to 336 hours for wood pieces to reach saturation in water. By comparison, agricultural products, such as chickpea and rice, takes 10 hours or less to achieve saturation, albeit at lower maximum saturation moisture content. Regenerated cellulose film (cellophane) achieved a saturation moisture content of 0.93 in minutes. Table 1.6 The saturation moisture content, the time to saturation, and the average liquid water absorption rate of several untreated materials near room temperature. Material Maximum saturation moisture content, MS (kg water/kg dry material) Time to saturation, tS (hours) Average absorption rate* (kg water/kg dry material-hour) Reference Poplar wood (Populus cathayana) 1.32 192 0.0069 Chen et al., (2020) Norway spruce (Picea abies) 1.38 300 0.0043 Sivertsen & Vestøl (2010) Norway spruce (Picea abies) 1.60 336 0.0044 Sandberg & Salin (2012) Maple (Acer) 1.09 240 0.0042 Khazaei (2008) Cork (Quercus suber) 1.70 1200 0.0014 Rosa & Fortes (2007) Regenerated cellulose film (Cellophane) 0.93 0.04 23.2500 Stamm (1956) Chickpea (Cicer arietinum) 1.24 10 0.1100 Turhan, Sayar, & Gunasekaran (2002) Rice paddy (Oryza cultivar PR116) 0.40 10 0.0260 Thakur & Gupta (2006) Rice husk (Oryza cultivar PR116) 1.05 2 0.4550 Thakur & Gupta (2006) *Average absorption rate is calculated as (M0 – MS)/tS, where M0 is the initial moisture content. 15  1.2.5 Water vapor adsorption  Table 1.7 lists the moisture adsorption behaviors of several materials near room temperature. Instead of equilibrium moisture content, the term “final moisture content” is used because, in the referred sources, the moisture content of materials is still increasing at the end of the authors’ experiments. Table 1.7 Moisture adsorption behaviors of several materials in 50-90% relative humidity (RH). Material RH (%) Temperat-ure (°C) Initial moisture content, M0 (kg water/kg dry material) Final moisture content, Mf (kg water/kg dry material) Time, t (hours) Average absorption rate* (kg water/kg dry material-hour) Reference Low rank coal, particle size < 1 mm 75 60 0.042 0.104 50 0.0012 Li et al. (2009) Scots pine sapwood (Pinus sylvestris) 0.5 cm thick blocks  80 30 0.110 0.135 12 0.0021 Droin-Josserand et al. (1988) Switchgrass (Panicum virgatum) pellets 80 25 0.067 0.126 24 0.0025 Colley et al. (2006) Alfalfa (Medicago sativa) pellets 90 30 0.081 0.307 24 0.0094 Fasina & Sokhansanj (1993) Spruce, Pine, Fir (SPF) pellets 90 30 0.000 0.202 7 0.0289 Peng et al. (2013) *Average absorption rate is calculated as (M0 – Mf)/t    The water vapor adsorption or hygroscopic moisture adsorption behavior of materials varies depending on the relative humidity and temperature of the atmosphere, the type of the material, and its initial moisture content. Coal, as a somewhat hydrophobic material, adsorbs moisture from the atmosphere at a lower rate than the wood species listed in the table above.  16  Coal has a final moisture content of 0.104 kg water/kg dry mass or dry basis (db) after 50 hours of exposure to a 75% RH and 60 °C atmosphere. Scot pine achieved a final moisture content of 0.135 (db) after 12 hours of exposure. When wood is pelletized, its rate of moisture adsorption increases (Peng et al., 2014). Compared to alfalfa pellets, wood pellets have a lower final moisture content for the same atmospheric conditions of 90% RH and 30 °C. 1.2.6 Moisture and heat transfer Sokhansanj et al. (2003) modeled the heat and moisture transfer in containerized alfalfa cubes using a one-dimensional partial differential equation (PDE) of combined conduction and natural convection. The model adequately fitted the actual recorded temperature during a container ship’s voyage but did not fit the humidity recording because it did not take into account moisture transfer from outside the container. Similar difficulties in modeling moisture transfer were encountered by Gastón et al. (2009), who modeled the heat and moisture transfer of wheat stored in silo bags. The authors explained that the high error between predicted values and measured values was due to the high variations in the measured values of moisture contents (Gastón et al., 2009). Miura (2016) was successful at modeling the heat and moisture transfer of ~100 mg of coal lumps by solving two ordinary differential equations (shown below in an unarranged form) simultaneously with respect to temperature, T, and partial pressure of water vapor, p. The re-arrangement of the two equations into the solvable simultaneous equations is given in Appendix B.2. His approach includes the contribution of the heat of water vapor condensation due to moisture adsorption from the ambient atmosphere. The modeled results gave a good fit with the measured temperature and moisture content of coal. Wpelletd(q+ερbC)/dt = kcA(Cenv-C)        (1.4) 17  where Wpellet is the weight of dried wood pellet (in kg), q is the amount of water vapor adsorbed per unit weight of dried pellet (in kg water per kg dry pellet), C and Cenv are respectively the concentrations of water vapor in the wood pellet and in the ambient atmosphere (in kg H2O(g)/m3), which are functions of p and T by assuming ideal gas law. ε is the void fraction of the wood pellet including the pore volumes within the wood particles in a pellet, ρb is the apparent density of the wood pellet (in kg/m3), t is the time (in seconds), kc is the mass transfer coefficient of water vapor in the gas film around the wood pellet (in m/s), and A is the outer surface area of the spherical particle. Wpelletddt(Hpellet+qHw(T)+ερbCaHa)=hA(Tenv-T)+kcA(Cenv-C)Hwv(Tenv) (1.5) where Ca is the total concentration of humid air in the wood pellet (in kg air/m3, which are functions of p and T by assuming ideal gas law). Hpellet is the enthalpy of the dried pellet per unit weight, in kJ/kg pellet, Hw and Hwv are respectively the enthalpies of liquid water (a function of T) and water vapor (a function of Tenv), and h is the heat transfer coefficient in the gas film around the wood pellet (in W/(m2·K)).  1.3 Thesis Statement and Research Objectives My literature review showed that there are no previous studies on the quality of wood pellets due to incidental wetting and subsequent handling of the pellets. The questions are: (i) how is the physical quality of the pellets in large mass affected when a fraction of the load mass becomes wet? (ii) will there be any further moisture redistribution and accumulation within the load? (iii) will self-heating increase when pellets are wetted? This thesis attempts to answer these important questions and fill fundamental knowledge gaps that would help to mitigate the risk in the handling 18  of commercial pellets. The end goal is to assist in developing a decision support system for the pellet handling facilities and pellet handlers to manage the safe loading of pellets. The overall objective of this thesis research is to analyze the impact of rain (liquid phase water) and humid air (vapor phase water) on the storage stability of wood pellets with respect to physical properties and self-heating. Rain and humid conditions are the sources of wetting of the pellets during handling and shipping, as mentioned above. The specific objectives are as follows: 1. To measure the rate of moisture adsorption and to define the equilibrium conditions when pellets are exposed to water in the liquid phase and the vapor phase. 2. To measure the durability and density of pellets as a function of moisture adsorption processes. 3. To develop an equation for the heat of wetting as a function of moisture adsorption and analyze the self-heating behaviors of pellets when exposed to water in the liquid phase and the vapor phase. Figure 1.2 summarizes my research topics. My research material is commercial wood pellets. After they are exposed to liquid water and water vapor, their physical properties were tested. The temperature increase in the two phases of water and the heat of wetting of pellets were measured. Finally, a heat and mass transfer model was developed. 19   Figure 1.2 Flow diagram for my research topics and their relationships.  Note on units of moisture content This thesis uses moisture content in two units: percentage wet mass basis (%, wb), or simply (wb), and decimal dry mass basis (dec, db), or simply (db). The conversion between the two units is given in Eq. 1.5. Table A.2 provides a range of values of moisture content in those two units. The unit “decimal dry basis (db)” is equivalent to kilogram moisture per kilogram dry mass of samples (kg moisture/kg dry). M (dec, db) = m.c.100 – m.c. (1.6) where M is moisture content in decimal dry mass basis (dec, db); m.c. is moisture content in percentage wet mass basis (%, wb).  20  Chapter 2: Liquid Phase Moisture Characteristics of Wood Pellets When wood pellets are exposed to liquid water, such as rainwater, their physical properties, such as durability and density, are expected to decrease. This chapter tests this hypothesis and any changes in physical properties.  2.1 The Effect of Moisture on Physical Properties of Wood Pellets 2.1.1 Introduction In the transition to renewable fuels from fossil fuels, wood pellets have emerged as the primary substitute for coal in solid-fuel power generation plants and heating oil in residential and institutional heating (Junginger et al., 2019). Compared to other biomass pellets, wood pellets are considered the best solid fuel in terms of having a low ash content, the ease to manufacture, and uniformity in quality (Sultana & Kumar, 2012). Table 2.1 lists the main physical properties of wood pellets and their values for grading pellets for international trade. Some or all of the listed quality standards are included in contractual agreements between sellers and buyers of wood pellets.  Oveisi et al. (2013) quantified the durability of wood pellets using a tumbler test (ASABE, 2015) and a drop test. The authors reported that pellets had a tumbler durability of 97.0%. Yazdanpanah et al. (2011) reported the bulk density and particle or solid density of bulk pellets as 714 - 728 kg/m3 and 1080 - 1350 kg/m3, respectively. Tumuluru et al. (2010) studied the general quality of softwood pellets made from spruce, pine, and fir in British Columbia by sampling wood pellets at a pellet handling facility in Vancouver, British Columbia, without considering the effect of moisture adsorption. The authors reported that depending on the time of sampling, the softwood pellets had a moisture content of 3.4 to 6.4% wet mass basis (wb), bulk density of 728 to 808 21  kg/m3, tumbler durability of 97.1 to 98.5%, and fines content of 0.03 to 0.87%. The tested pellets met grade A2 requirements according to Table 2.1. Table 2.1 Canadian Standard for commercial trading of wood pellets (Wood Pellet Association of Canada, 2016). This standard complies with the ISO 17225 International Standard for wood pellets. *ar stands for as-received basis and is equivalent to wet mass basis. Property Unit CANplus Certification Standard (2016) A1 A2 B Diameter mm 6 ± 1 or 8±1 Length mm 3.15 ≤ L ≤ 40 Moisture content % of weight, ar* ≤ 10 Total ash content % of weight, dry ≤ 0.7 ≤ 1.2 ≤ 2.0 Mechanical durability % of weight, ar* ≥ 98.0 ≥ 97.5 Fines content  (< 3.15 mm) % of weight, ar* ≤ 1.0 Bulk density, ar* kg/m3 600 ≤ BD ≤ 750    Net calorific value,  ar*  MJ/kg  ≥ 16.5 Temp. of pellets °C ≤ 40  Hartley and Wood (2008) are among the few researchers who investigated the moisture adsorption isotherms of wood pellets. They investigated the effect of water vapor uptake on the swelling responses of wood pellets. They found that at a moisture content of 13% (dry basis), pellet volume expanded by 12% after pellets were stored for 30 days. More recently, Deng et al. (2019) reported that after exposure to a relative humidity of 95%, the bulk density of pine pellets dropped from an initial value of 650 kg/m3 to 490 kg/m3 after 4 days. The pellet durability dropped by 1.3%. Theerarattananoon (2011) observed a consistent 20 to 30 kg/m3 drop in bulk density in several types of agricultural residue pellets exposed to moisture. Wheat straw and big bluestem pellets showed a drop of more than 2% in durability when moisture content increased from 9.7% to 15.8% (dry basis). Big bluestem (Andropogon gerardii) is a warm-season native grass that 22  comprises up to 80% of the plant biomass in the grasslands of the North American Midwest and is studied as a potential bioenergy crop (Zhang et al., 2015).   Yan et al. (2014) observed that untreated pine pellets rapidly disintegrated and lost their shape in less than one minute upon their submersion in water. This finding supported the empirical observation that reductions in durability are more severe for pellets exposed to liquid water than for pellets exposed to humid air (Lee et al., 2019); it takes two to four days for wood pellets to reach an equilibrium with a humid atmosphere (Deng et al., 2019). Although pellets can be exposed to liquid water during ship loading and unloading, to my knowledge, the effect of liquid water adsorption on durability and bulk density of Canadian softwood pellets and hardwood pellets has not been quantified. The remaining part of this chapter quantifies durability, fines content, bulk, and solid densities of three types of wood pellets for pellets before and after exposure to liquid water.   2.1.2 Materials and methods 2.1.2.1 Sample preparation Three types of Canadian wood pellets were used: (a) Groupe Savoie Canawick-brand residential hardwood pellets (made in Saint-Quentin, New Brunswick), (b) Princeton Co-Generation Corp. residential softwood pellets (made in Princeton, British Columbia), and (c) Pinnacle Renewable Energy Inc. industrial softwood pellets (made in its seven plants in British Columbia). Figure 2.1 shows the visual appearance of the three types of pellets. Compared to residential pellets (a) and (b), the industrial pellets (c) show greater variation in appearance and length (Table 2.2). The industrial pellets contain bark, whereas the residential pellets are made from white wood (no bark). The two premium-grade residential hardwood (a) and softwood pellets (b) were acquired from a 23  Canadian Tire store in Vancouver in August 2015. The industrial softwood pellets (c) were received from Pinnacle Renewable Energy Inc. at approximately the same time.  Canawick residential hardwood pellets are made from a mixture of maple (Acer), birch (Betula), aspen (Populus), and ash (Fraxinus) wood. Both residential and industrial softwood pellets were made from a mixture of pine (Pinus), spruce (Picea), and fir (Abies). Industrial softwood pellets have a high content of forest residues (10 to 50% of the raw material, depending on the location of the pellet plant).  The logging residue may increase the ash content of the pellets to more than 1% dry basis due to their higher bark content (Filbakk et al., 2011). Residential softwood and hardwood pellets, both with ash content less than 0.5%, are primarily made from sawdust and shavings from nearby sawmills. Residential pellets are burned in pellet stoves for home heating purposes. The industrial pellets are combusted in power plants for power generation.  Experiments were performed immediately after the pellet samples were obtained.  Figure 2.1 The three types of pellets used: (a) residential hardwood pellets, (b) residential softwood pellets, (c) industrial softwood pellets. The scale is in centimeters.   Table 2.2 lists the properties of pellet samples received at the laboratory. The durability of samples measured using the tumbler device (ASABE S269.5) was more than 99% for all pellet types. The measured bulk density at 684 to 718 kg/m3 met the standard in Table 2.1. The ash contents (not shown) were measured using a muffle furnace at 575 °C for three hours to be less than 1% (Sluiter et al., 2005).  24  Table 2.2 The average properties of the three types of wood pellets after oven-drying. The percentages are on a dry mass basis. For diameter and length, the number of samples (N) is 100 for each type. For fines content, durability, bulk density, N is 3. The spread of data for length and bulk density was given as the standard deviation (±SD). Type Diameter (mm) Length (mm) Fines content (%) Durability (%) Bulk density (kg/m3) Residential hardwood 6.3 14.7±4.3 0.60 99.3 713±3 Residential softwood 6.4 16.6±4.7 0.30 99.8 684±2 Industrial softwood 6.4 15.6±5.2 0.18 99.6 718±2  Wood pellets at 6% m.c. (wb) as received were dried in a convection oven set at (103 ± 2) °C for 24 hours until the pellet mass remained unchanged. The dried pellets were wetted to 1.5%, 8%, 10%, 13%, 17%, 20%, and 25% wb by mixing the pellets with pre-determined volumes of water in plastic bags to ensure uniform wetting of the sample. The mass of each sub-sample of oven-dried pellets prior to wetting was ~1500 g. Table 2.3 lists the mass of distilled water added to the pellets. The pellet samples were gently shaken in the plastic bag to ensure uniform contact with the sprayed water. The moistened pellets were left in the bag for 2 hours. Measurements of the physical qualities of the pellets are described in the next subsection. Table 2.3 The mass of distilled water sprayed onto the 1500 g of wood pellets from each pellet type. Mass of water added (g) 23 130 167 224 307 375 500 Moisture content (% wb) 1.5 8.0 10.0 13.0 17.0 20.0 25.0  2.1.2.2 Test methods First, the moisture content of three replicates of wetted pellets per moisture content was determined gravimetrically by using the oven drying method at (103 ± 2) °C for 24 hours.  The fines, generated during wetting, were removed using a 3.15 mm round sieve (ISO 17225). The fines content as a percentage of the original sample weight was calculated (Eq. 2.1). The durability of the cleaned pellets was determined as a percentage of unbroken pellets over initial mass by 25  testing 500 g pellet samples in a tumbler tester (ASABE S269.5) set at 50 rpm for 10 minutes. The unbroken pellets were those pellets that stayed on the 3.15 mm round sieve after the tumbler test.  Fines content =mfinesmsample×100       (2.1) where mfines is the mass of fines removed using a 3.15 mm round sieve, and msample is the total mass of the sample before sieving. The bulk density of pellets was determined as the ratio of mass over volume. Pellets were poured into a box (100 mm x 100 mm x 100 mm) until the box was full. Excess pellets were removed that were above the sides of the box using a plastic scraper with a straight edge. The mass of pellets in the filled container was obtained by weighing pellets on a digital balance to a precision of 0.1 g.  The solid density of pellets was determined by measuring the volume of the wood particles in pellets, using a pycnometer, model MVP-D160-E (Quantachrome Instruments, Boyton Beach, FL, USA). Nitrogen gas was used as the displacement gas. Solid density is the mass of the sample over the measured volume of wood particles in a pellet. Macro-porosity of wood pellets, ϕ was calculated by, 𝜙 =  𝜌𝑆 – 𝜌𝑏𝜌𝑆 – 𝜌𝑎𝑖𝑟           (2.2) where ρS is the solid density of wood pellets, ρb is the bulk density of wood pellets, and ρair is the air density. Olympus BX53 high-resolution light microscope was used to obtain pictures of the horizontal cross-section surface of a single pellet before and after exposure to a droplet of water. The maximum vision range of the microscope was 13 - 15 mm. The maximum magnification is 60.  26  2.1.3 Results  2.1.3.1 Surface structure Wood pellets tend to swell immediately after contact with water. Figure 2.2 shows the localized swelling of a piece of industrial softwood pellet. The swelling led to wood particles protruding from the spots where a pellet was wetted. I observed that at the point of contact between a water droplet and pellet surface, small wood particles in the form of fines became loose and separated from the pellet. I believe the swelling force due to the increasing volume of the wood particles pushed the particles away from the surface. A similar process may have occurred within the pellet matrix. Swelling causes particles to separate from each other. As I show in the next section, this separation of particles decreases the durability of moistened pellets.   Figure 2.2 A softwood pellet, 5 minutes after wetting by droplets of liquid water. The contrast of the image was increased to show localized swells.  Figure 2.3 is a light microscopy image of the internal structure of a softwood pellet before (a) and after wetting (b). Picture (b) suggests that pellets expanded substantially. This is evident from picture (b), which shows that wetting caused the pellet to lose its circular cross-section. Loss of the physical integrity of the pellet is also evident.   27   Figure 2.3 Microscopic surface photograph of the horizontal cross-section of a softwood pellet before wetting (a) and after wetting (b) by a droplet of liquid water. The swelling of the internal parts of the wetted pellet is evident.  2.1.3.2 Pellet durability and fines content Figure 2.4 and Figure 2.5 compare the durability and fines content of the three types of wood pellets. Durability is the quantification of the abrasion and impact resistance of wood pellets (Tumuluru, 2014), while fines content indicates the loose powder fraction of wood pellets. The fines agglomerate and hinder pellets from flowing freely (Kofman, 2007). Upon exposure to liquid water, the changes in durability and fines content of the three pellets were similar. Initially, the oven-dried residential hardwood had a durability of 99.3%, residential softwood had a durability of 99.6%, and the industrial softwood had a durability of 99.7% (Table 2.2). Figure 2.4 shows that the durability of wetted wood pellets remained unchanged from 6 to 9% m.c. The durability dropped to 97% and was lower for the pellet wetted to 10% m.c. At 10% m.c., the durability of residential hardwood pellets dropped from 99% to ~95%. The two softwood pellet types had a durability of ~97%. Thus, residential hardwood pellets appeared to be more susceptible to water damage. The minimum required durability to meet the CANplus standard, which is a Canadian equivalent to ISO 17225 standard (Wood Pellet Association of Canada, 2016) 28  for residential heating pellets, is 97.5% (Table 2.1). At above 10% m.c., the three types of pellets tested did not meet the minimum durability requirement for CANplus residential heating pellets.   Figure 2.4 Durability of three types of wood pellets vs. moisture content. The pellets were initially at 0% m.c. The gain in m.c. was the result of spraying water on the pellets. The lines are quadratic equations fitted to the measured data. The number of repetitions is three. The error bars represent maximum and minimum of the three repetitions.  At moisture contents above 20%, the durability of wood pellets dropped to below 80%. Industrial wood pellets performed the worst at this level of moisture content with a durability of 65.8%, while the durability of the two residential pellets remained between 70 to 80%. Theerarattananoon et al. (2011) observed a smaller drop in the durability of wheat straw at a moisture content of 14%.  60657075808590951001051100 10 20 30 40Durability  (%)Moisture content (%, w.b.)Residential HardwoodResidential SoftwoodIndustrial softwood29  Figure 2.5 Fines content of three types of wood pellets vs. moisture content. The lines are quadratic equations fitted to the measured data. Only one sample was tested for fines content for every moisture content. The error bars represent maximum and minimum of the three repetitions.  The specified fines content of CANplus standard certified pellets is < 1% (Wood Pellet Association of Canada, 2016). As shown in Figure 2.5, the fines content remained below 1% when m.c. was below 8% and exceeded 3% when m.c. surpassed 10%. Deng et al. (2019) found that at 8% and 12% m.c., the fines contents (particles < 3.15 mm) of humidified pine pellets were 3.0% and 6.2% respectively. Their results agree with my findings for fines content at similar moisture contents. The correlation between durability and fines content data was 0.995, which implies that we may use fines content of pellets as a proxy for their durability. Durability, Dtumbler and fines content, 𝐹 expressed in percentage (%) are described as a function of moisture content, 𝑚. 𝑐. in percentage (%) wet basis,  Residential hardwood: Dtumbler= -0.0301(m.c.)2 - 0.2818(m.c.) + 100, R2 = 0.950  (2.3)    F = 0.0696(m.c.)2 + 0.2878(m.c.) + 0.600, R2 = 0.971  (2.4) Residential softwood: Dtumbler = -0.0576(m.c.)2 + 0.3929(m.c.) + 100, R2 = 0.957  (2.5)    F = 0.0836(m.c.)2 - 0.5277(m.c.) + 0.300, R2 = 0.961  (2.6) 0102030405060700 10 20 30 40Fines content (%)Moisture content (%, w.b.)Residential HardwoodResidential SoftwoodIndustrial Softwood30  Industrial softwood:  Dtumbler = -0.0689m.c.2 + 0.4315m.c. + 100, R2 = 0.995  (2.7) F = 0.1420m.c.2 - 0.9912m.c. + 0.180, R2 0.982   (2.8) I note that the initial durability of the three pellet types are assumed to be 100% when pellets were in their initial dry state. The durability of dried biomass decreased with moisture content, which suggests that the adsorbed moisture affects the internal structure of the pellets.  2.1.3.3 Bulk density I measured the bulk density of the three types of wood pellets before and after wetting. Figure 2.6 plots the data (average of three measurements) for each of the three types of wood pellets. The bulk density of all three samples increased initially until the moisture content increased to about 5% wb. The bulk density then decreased as the moisture content of the pellets continued to increase. A similar trend in bulk density was observed by Colley et al. (2006) in their study of switchgrass pellets. For oven-dried samples, the bulk density varied from 682 to 730 kg/m3, depending on the pellet type. At 6%, the bulk density increased slightly to 711 - 744 kg/m3. For pellets at 10% m.c., bulk density decreased to 621 - 664 kg/m3, averaging 658 kg/m3. This observation implies that when pellets were wetted from zero moisture to 6%, their mass increased faster than the increase in volume. This resulted in a net increase in bulk density. When m.c. surpassed 10% wb, the mass increased more slowly than the increase in volume due to swelling. That resulted in a net decrease in bulk density. The spread in bulk density data at each moisture content, as represented by standard deviations, was only 2 to 10 kg/m3. Subsequently, the error bars for bulk density in Figure 2.6 are not visible for the provided scale. 31   Figure 2.6 Bulk density of three types of wood pellets vs. moisture content. The lines are sigmoid equations fitted onto the measured data. The number of repetitions is three. The error bars represent maximum and minimum of the three repetitions.  An empirical sigmoid equation (Eq. 2.9) represented the bulk density, 𝜌𝑏 as a function of moisture content. Although the measured bulk density increased from ~700 kg/m3 at m.c. of 0% to 740 kg/m3 at m.c. of 6% wb, Eq. 2.9 was not able to predict this initial increase in bulk density. However, the equation predicted that the bulk density for wood pellets reached a minimum plateau at m.c.s of 25-30% wb. Bulk density remained unchanged because volumetric swelling due to water absorption was roughly equal to the increase in the mass of water uptake. This plateau in volumetric swelling at higher moisture contents was observed in pine sawdust as well (Adhikary, Pang, & Staiger, 2008).   𝜌𝑏−𝜌𝑏0𝜌𝑏𝑓−𝜌𝑏0=11+exp[−𝑘(𝑚.𝑐.−𝑚.𝑐.0.5)]        (2.9) where 𝜌𝑏 is the bulk density (kg/m3) of the pellet sample at a certain moisture content m.c. (% wb). 𝜌𝑏0 is the starting bulk density at zero moisture. 𝜌𝑏𝑓 is the final bulk density at a very large moisture content. 𝑘 is a fitting constant. 𝑀0.5 is the middle point moisture content. Table 2.4 lists the 2003004005006007008000 10 20 30 40Bulk density (kg/m3)Moisture content (% wb)ResidentialHardwoodResidentialSoftwoodIndustrial SoftwoodA slight increase in bulk density.32  estimated 𝜌𝑏0, 𝜌𝑏𝑓, 𝑘 and 𝑀0.5 by using the Microsoft Excel’s least mean square (LMS) curve-fitting tool. Table 2.4 Values of the parameters of the sigmoid Eq. 2.9. Parameter Residential Hardwood Residential Softwood Industrial Softwood 𝜌𝑏0 746.0 702.8 743.3 𝜌𝑏𝑓 350.2 321.5 344.7 𝑘 0.2724 0.3757 0.2916 𝑀0.5 13.40 15.96 13.73 R2 0.990 0.961 0.980   Figure 2.7 Specific volume of three types of wood pellets vs. moisture content. Specific volume is calculated as one over bulk density. The error bars represent maximum and minimum of the three repetitions.  Figure 2.7 gives the calculated specific volume of the three types of pellets as a function of moisture content. The specific volume is the inverse of bulk density or 1/𝜌𝑏 as volume per unit total mass of wet pellets. The specific volumes of zero moisture samples for all three types of wood pellets were between 0.0014 to 0.0015 m3/kg. The specific volume increased slightly to 0.0015 - 0.0016 m3/kg when the moisture content increased to 10% wb. The specific volume increased linearly as moisture content approached 17%, where specific volume increased by ~30% to 0.0020 0.00000.00050.00100.00150.00200.00250.00300.00350.00400 10 20 30 40Specific volume (m3/kg total)Moisture content (% wb)Residential HardwoodResidential SoftwoodIndustrial Softwood33  - 0.0022 m3/kg. The specific volume of residential softwood pellets increased by 133% at a m.c. of 34% wb. The industrial softwood pellets and residential hardwood pellets showed an increase in their specific volumes of 81% and 93%, respectively, at a m.c. at or above 30% wb. This increase in specific volume is a safety concern for the storage of pellets in an enclosed storage vessel, as a large increase in volume can result in large pressure exerted on the vessel hold-walls. 2.1.3.4 Solid density Solid density represents the density of the solids and particles within a material and excludes any voids or air within a material, provided that these voids are accessible to the pycnometer gas (nitrogen). The solid density is generally considered as the maximum density achievable by mechanical compaction of wood particles and is used as a baseline to gauge the effectiveness of a densification process. The solid density of wood, also called “cell-wall density” in wood science literature, ranges from 1350 to 1460 kg/m3, with an average of 1400 kg/m3 (Zauer et., 2013). I assume adsorbed water molecules for a wood substrate occupies the same volume as water in its native state. A mix of wood and water makes up a density proportional to their corresponding fraction in the mix. If wood has a solid density of 1400 kg/m3, wood with 1% of water (density = 1000 kg/m3) would have a solid density of 1396 kg/m3 (0.99 × 1400 + 0.01 × 1000).  This results in a reduction in bulk density of 4 kg/m3 for a 1% increase in m.c. A hypothetical linear relation can be assumed for density as a function of moisture content,    𝜌𝑆 = -4.00(m.c.) + 1400   (2.10) where M is moisture content in % wb. Adhikary et al. (2008) used a similar linear equation to relate wood density or specific volume to moisture content.  34  Figure 2.8 shows a “hypothetical line” representing Eq. 2.10. Eqs. 2.11 and 2.12 were obtained when the density (𝜌𝑆) of moist wood pellets was fitted to m.c.  (% wb),  Residential hardwood: 𝜌𝑆 = -1.9388(m.c.) + 1426.0, R² = 0.8248   (2.11) Residential softwood: 𝜌𝑆 = -4.9704(m.c.) + 1440.1, R² = 0.7609   (2.12)  Figure 2.8 Solid density, 𝜌𝑆 of residential hardwood and softwood pellets in circle dots and lines for the fitted lines given by Eqs. 2.10-2.12. Hypothetical line (Eq. 2.10) assumes water with a density of 1000 kg/m3 is added into the wood with a cell-wall density of 1400 kg/m3. This test was done on whole pellets at moisture content below 20% and wet swollen pellet particles at moisture contents above 20%.  Siau (1984) reported that bound water density is higher at lower moisture contents, where bound water is the water adsorbed to the wood material. This may be another reason why the zero-moisture solid densities for hardwood and softwood pellets, which are the y-intercepts for Eqs. 2.11 and 2.12, are higher than the cell-wall density of 1400 kg/m3. 126012801300132013401360138014001420144014600 5 10 15 20 25 30 35 40Solid Density (kg/m3)Moisture content (% wb)Hardwood Softwood Hypothetical line35  The residential softwood had a slope similar to Eq. 2.10. The density of hardwood pellets had a smaller slope than softwood pellets. The density of hardwood pellets appeared to be less affected by liquid water exposure (rate = -1.94 kg/m3 per % m.c.) because the adsorption sites on the hardwood might bond more strongly with water and incorporate water molecules deep into the hardwood cell wall. Hence, water adsorption in hardwood pellets did not result in as large of a volume increase as softwood pellets, and consequently a lower solid density reduction. The solid density for industrial softwood pellets was not measured. 2.1.3.5 Bulk porosity Bulk porosity, ϕ, was calculated from measured solid density and average bulk density at a given moisture content using Eq. 2.2. Porosity is an important parameter used to model heat and moisture transfer in wood pellets (Mahmoudi, Hoffmann, & Peters, 2014). Figure 2.9 provides a comparison of the bulk porosity between hardwood pellets and softwood pellets. The linear Eqs. 2.13 and 2.14 represent porosity, ϕ as a percentage (%), and moisture content, M is expressed in percent wet basis (% wb).  Residential hardwood: ϕ = 1.3574(m.c.) + 41.902, R2 = 0.9839  (2.13) Residential softwood: ϕ = 1.6099(m.c.) + 36.479, R2 = 0.9507  (2.14) Note that the linear equations above are only valid for moisture contents above 7%; the porosity below 7% m.c. is almost constant with an average value of 50.7±1.3%. Guo et al. (2013) gave a range of porosity values from 42.1 to 43.4% for bulk wood pellets with moisture content of 1.7 to 9.0%. In this study, residential hardwood and softwood pellets had an average porosity of 50.7% at moisture content below 7%m.c. The porosity of softwood pellets increased to 77.7% at 23.7% m.c., whereas the porosity of hardwood pellets increased to 72.5% at 23.3% m.c.   36   Figure 2.9 Porosity of residential hardwood and softwood pellets in circle dots and lines for the fitted line.  2.1.4 Discussion Durability and fines content are two crucial quality parameters, which indicate wood pellets’ resistance to abrasion and impact. CANplus standard, the internationally recognized quality standard for wood pellets, states that durability must be above 97.5%, and the maximum fines content of wood pellets must be below 1% (Wood Pellet Association of Canada, 2016). Since wood pellets may be exposed to water during loading and unloading at ports when it rains, the equations relating durability and fines content to moisture content are useful in assessing the quality of pellets after they are exposed to rain.  404550556065707580850 2 4 6 8 10 12 14 16 18 20 22 24 26Porosity (%)Moisture content (%w.b.)Hardwood Softwood37  When wetted from zero moisture, the durability of wood pellets either stayed constant or slightly increased. This result reinforced my observations in Section 2.1 that the wood pellets may be exposed to a small amount of water up to m.c. of 8% without experiencing unacceptable loss in durability. Bulk density of pellets dictates the quantity of pellets that can be loaded onto trucks, trains, and ships to be transported to the market. The equations describing the relationships between bulk density and moisture content for the three reported types of pellets are useful in calculating the final bulk density after pellets are exposed to liquid water. My result showed an increase of bulk density from 700 kg/m3 to 740 kg/m3 when wood pellets were wetted from zero moisture to 6% wb. This increase in bulk density meant that more pellets could be loaded onto the rail cars and ship’s holds to be transported.  Section 2.1 quantifies durability, fines content, bulk and solid densities of three types of wood pellets for pellets before and after exposure to liquid water at a range of moisture content. Section 1.1 investigated the effects of weight gain from spraying water on the durability of a tray of pellets.  38  2.2 Effects of Rain on the Durability of Pellets 2.2.1 Introduction  The forest sector makes a major contribution to the economy of the province of British Columbia (BC), for example, it made up 7% of BC’s gross domestic product (GDP) in 2015 (Barnes, 2019). With the slowdown of the pulp and paper industry and demand for residues (chips and sawdust), BC’s wood pellet industry has emerged as a viable source of revenue for the forest sector, with 85% of production destined for overseas markets (Murray, 2018). In 2012, BC’s wood pellet industry generated more than $300-million in annual revenue, while providing approximately 400 direct jobs in rural BC communities (MNP Ltd., 2015). The annual production of wood pellets is expected to continue to increase significantly in response to the rising demand for pellets in the Asian Pacific and Western European countries, potentially generating more traffic to the ports in BC. The fast-growing wood pellet industry will need support in many ways; this would include the best practice during the loading of pellets when it rains.  Wood pellets are cylindrical forms of compressed wood particles, as mentioned above. They are made by extruding wood particles with particle size less than 2 mm through a perforated hard steel die (6 mm diameter holes; 40-60 mm depth). The product is a cylindrical-shaped solid fuel, which has good flowability (angle of repose < 50°), a high bulk density of 650-750 kg/m3, and a high energy density of 13-15 GJ/m3 (Rezaei et al., 2016). These desirable physical properties make the economics of transporting wood pellets over long distances feasible (Tumuluru et al., 2011). In general, wood pellets are loaded into the holds of ocean-going vessels without any protection from the rain, as shown in Figure 2.10 and described in Chapter 1. Due to the typical rainy weather conditions in the coastal region of BC, wood pellets can become wet if it rains during 39  the loading operation.  In their review of the factors that affect the durability of pellets made from woody and herbaceous biomass feedstock, Whittaker and Shield (2017) suggested that aside from feedstock characteristics and pelleting conditions, pellet durability can also be affected by post-production conditions such as the storage environment and handling frequency. Durability is a measure of the quality of wood pellets in terms of mechanical strength and physical integrity, with a typical value of 96.5% or higher (Tumuluru et al., 2010). Furthermore, according to ENplus quality standard (European Pellet Council, 2015), wood pellets with more than 1% fines content and less than 96.5% durability do not qualify as commercial-grade pellets. From the buyer’s point of view, wetted wood pellets are no longer in their desirable compact form. They may demand a refund from the pellet producers or, even worse, pursue a lawsuit for not receiving the wood pellets that comply with the quality specifications in their supply contract.   Figure 2.10 Loading of wood pellets onto a bulk carrier with an open hatch (Curci, 2011)   Obernberger and Thek (2010) suggested that the loading of pellets during rain must be avoided; however, this recommendation for rainy BC coastal regions may not always be practical. Based on records of daily precipitation during the period 2000-2010 for two major exporting ports (Vancouver and Prince Rupert) in Western Canada, on the days when there were precipitation 40  events, 40-50% had daily precipitation amounts of less than 5 mm on those days. This may imply that there are many rainy days with “very light” or “light” rain intensities, as defined in Chapter 1. Athanasatos et al. (2014) stated that weather is one of the factors that can generate risks and reduce productivity in maritime-related operations, and it can hamper actual port operations.  Different shipmasters would allow different degrees of precipitation during loading, thus leading to a varying extent of wetness and moisture contents of the wood pellets. It should be noted that any delay in loading operation can cost pellet producers more than $20,000 per day in demurrage fee (the charge paid to the port from the day that free time has been used up to leave the ship stowed at the loading yard) at loading ports such as the Port of Vancouver and the Port of Prince Rupert (Canadian Canola Growers Association, 2015). The cost induced by the damage on the quality of wood pellets by rain should be weighed against the loss of revenue due to a delay in loading.  Wood pellets, like all dried woody biomass, are known to be hygroscopic (Craven et al., 2015). When wetted with water droplets, wood pellets swell and disintegrate, reverting to wood fines. Upon exposing switchgrass pellets to humid conditions at 90% RH/30°C, Colley et al. (2006) found that durability of the pellets increased slightly to a maximum of 96.7% at 8.6% (wb) moisture content, and a further increase in moisture content to 17% (wb) reduced the durability to 78.4%. Similar observations were made by Fasina and Sokhansanj (1992) for alfalfa pellets, where the durability of the pellets increased when moisture content increased from 5% to 8% (wb) but decreased as moisture content increased further. Lee et al. (2019) demonstrated that wood pellets wetted from initial moisture content of 5-6% (wb) to final moisture content of 8-9% (wb) did not exhibit a significant drop in mechanical durability. However, pellets wetted by immersing in water to attain a moisture content of 15% (wb) or higher were observed to swell and disintegrate due to 41  loss of adhesion among the wood particles within the wood pellets (Mantanis, 1994). Graham et al. (2016) conducted laboratory experiments to study the impact of exposing wood pellets to high ambient humidity conditions (90% RH/20 °C) on their mechanical strength and found substantial degradation of the pellets after 17 days as their moisture content increased from 8.5% to beyond 20% (wb).   Currently, there is no published literature on the effect of rainwater exposure on the durability and fines content of wood pellets. Therefore, the actual change in the quality of wood pellets, in particular the mechanical durability, due to rain exposure needs to be quantified in order to accurately estimate potential damage to wood pellets when they are loaded into ships. The objectives of this exploratory study are: 1) to develop a functional relationship between the durability of pellets and the amount of rainwater to which the wood pellets are exposed to; and 2) to quantify the critical levels of rain duration and rain intensity that can significantly reduce the durability of wood pellets. 2.2.2 Experimental procedure The experimental study was primarily designed to artificially wet the wood pellets at rain intensity less than 10 mm/h and measure pellet durability as well as the fines content after wetting.   When using a rain simulator for soil erosion studies, droplet size distribution is important. Kinetic energy is a function of the drop size and impact velocity, and it is closely related to the ability of rainfall to cause erosion. These rain simulators need to generate rain intensities that can be well in excess of 10 mm/h (Mhaske, Pathak, & Basak, 2019), representing heavy rainfall to storm events. By comparison, this study aims at very-light (<0.5 mm/h), light (0.5-2.5 mm/h) and moderate (2.5-7.5 mm/h) rain intensities. Hence, a rain simulator apparatus was not set up for our experiment, and droplet size distribution was not considered to be a significant parameter.  42  Softwood pellets (primarily a mixture of pine, spruce, and fir) were used in the experiment. These pellets were obtained from Pinnacle Renewable Energy Inc, a BC wood pellet producer, in summer 2014.  As shown in Figure 2.11, the experimental apparatus consists of an 18”×25”×1” (45.7 cm × 63.5 cm × 2.54 cm) rectangular tray with 0.29 m2 surface area, 1.32 L Gilmour hand-held 7” (18 cm) sprayer with an adjustable 1.6 mm diameter nozzle, stopwatch, electronic balance (0.1 g accuracy), and the Tumbler tester for durability measurement. The nozzle is cone-type and, if pressed continuously, has a flow rate of 140 mL/min, which corresponds to a rain intensity of 29 mm/h. To achieve rain intensities less than 7.5 mm/h, the sprayer was pressed intermittently; the rain intensity represents an average of several squirts of spraying events. The average droplet size was estimated to be 0.5 mm in diameter using a broad-laser aerosol-measuring system (King, Winward, & Bjorneberg, 2010).   The ‘Tumbler’ tester is made of a rectangular stainless-steel container with inner dimensions 300 × 300 × 125 mm (Oveisi et al., 2013). One baffle is placed inside the container to enforce the tumbling effect, whereby pellets experience impact and abrasion to the container wall and among individual pellets. The durability test procedure follows the international standard ISO 17831-1 (2016). According to the operating guidelines in the Standard, the rotation speed, duration, and mass of the sample were set at 50 rpm, 10 min, and 500 g, respectively. Upon sieving (3.15 mm diameter round hole sieve, Endecotts Ltd.), the percent of pellets that remained unbroken relative to the total sample mass was reported as durability.   The rectangular tray was filled with 2 kg of sieved wood pellets (the sample), resulting in three uniform layers having a total depth of 19.5 mm. The thickness of each layer (6.5 mm) is equivalent to the pellet diameter. An appropriate amount of water equivalent to a certain rain 43  intensity and the exposure time was sprayed on the pellets from a height of approximately 100 cm (see Experimental Procedure below). The wetted sample was covered with plastic sheets and left for 2 h to allow the moisture to distribute and penetrate the pellets. Less than 1% of water was evaporated during the resting. After 2 h of resting, the sample was sieved with a 3.15 mm round sieve to remove small particles (fines). The mass of fines was recorded, and this is used to calculate the fines content after wetting the pellets (Eq. 2.1). Subsequently, the sample was tested for the durability of pellets using the Tumbler tester, and its moisture content was measured using the oven-dry method.     Figure 2.11. A rectangular tray, which is filled with pellets (left); Tumbler tester (right).  2.2.3 Experimental setup 2.2.3.1 Set I tests The first set of tests (Set I) was conducted in an attempt to obtain a preliminary quantitative relationship between the durability as well as the fines content of wood pellets and the amount of rainwater exposed (water sprayed). The basis for the amount of sprayed water was a rain intensity of 1 mm/h or a spraying rate of 4.8 mL/min. The spraying durations or exposure times were 1.0, 44  2.5, 5.0, 7.5, 10, 15, 20, 25, 30, 40, 50 and 60 min. In accord with their constituent wood species (pine, spruce, and fir), wood pellets can absorb an amount of liquid water 1.7-1.8 times their dry mass (Sandberg & Salin, 2012). Hence, for all exposure times, it is expected that the sprayed water would be completely absorbed. The loss to entrainment during spraying and evaporation to the atmosphere was less than 2% of the weight amount of water sprayed onto the pellets. The commonly used unit for rain intensity (mm/h) can also be expressed in (L/m2-h). Data obtained from the Set I tests were used as the guideline to determine the amount of water sprayed in the Set II tests (below).  2.2.3.2 Set II tests The second set of tests (Set II) was conducted with the aim of establishing the safe limit or cut-off curve of precipitation rate or rain intensity and the corresponding exposure time that would reduce the durability of pellets to 96.5%; this value represents the minimum required durability of industrial Grade B wood pellets, according to the international standard for wood pellets ISO 17225-2. From Set I tests, the amount of sprayed water that reduced the durability of pellets below 96.5% was 139 mL for 2000 g of pellets. The spraying durations and rain intensity were then determined based on the target range of final moisture contents, which is 10 to 11% (wb), when the pellets had an initial moisture content of 5.5% (wb). This range of values was selected because the allowable maximum moisture content is 10% (wb) according to the international standard (International Standards Organization, 2014). The rain intensity ranges from 0.25 to 10 mm/h, which corresponds to 1.2 to 48 mL/min of spraying, as shown in Table 2.5. Thus the spraying duration or exposure time is in the range of 1 to 78 min, and the samples on the tray were sprayed intermittently over this range to provide rain intensities below 10 mm/h. Tests were repeated until 45  I could obtain combinations of rain intensity and exposure time that gave a durability of 96.5±1.0%. Each treatment was repeated three times. Table 2.5 Spraying rate (mL/min) converted from rain intensity (mm/h) and spraying duration used in Set II tests Rain intensity (mm/h)  Spraying rate (mL/min)  Spraying duration (min)  0.25 1.2 78 66 55 0.5 2.4 36 46 41 1 4.8 18 21 23 3 14.4 11 9.3 6.1 5 24.0 5.6 3.6 3.2 7 33.6 2.6 2.3 4.0 9 43.2 2.0 2.5 3.0 10 48.0 0.4 0.9 1.8  2.2.3.3 Actual rain events Similar experimental runs were performed by exposing the wood pellets to actual rain events. Two to three trays of wood pellets were left outdoor for six times (to be wetted by the rain for 1-2 h) either when a rain event was occurring or when rain was anticipated to occur according to the weather forecast. The runs were conducted during the day times of February 9 to March 11, 2015. There were 4 light rain events and 11 very light rain events. The temperatures during those rain events ranged from 5 to 15 °C. The moisture content, durability, and fines content of the pellets were measured 2 h after the pellets were wetted.  2.2.4 Results and discussion 2.2.4.1 Set I tests Figure 2.12 depicts the relationship between the amount of water sprayed and the durability of pellets, as well as the fines that were generated during the test. As expected, an increase in the amount of water sprayed led to a decrease in the durability of the pellets and an increase in fines 46  content. Note that the durability of wood pellets was measured after the fines generated from moisture-induced pellet disintegration were removed from the tray.   The data presented in Figure 2.12 can be represented by two equations. The fines content has a quadratic relationship with the amount of water sprayed (Eq. 2.16), which is the product of rain intensity and spraying time, whereas durability has a linear relationship with the amount of water sprayed (Eq. 2.17) Fines content = 0.00013V2 + 0.00259V + 0.00887            R2 = 0.98 (2.16) Durability = -0.0251V + 100                                             R2 = 0.99 (2.17) where V is the volume of water sprayed in mL for 2000 g of pellets.  It should be noted that these data were derived from wood pellets having an initial moisture content of 5.5% (wb). In this experiment, the volume of water sprayed represents the amount of rainwater that the wood pellets were exposed to. The rainwater volume V (mL) is calculated from the rain intensity I (mm/h), duration of exposure t (h), and the surface area of exposure A (m2) such that V = (1000 I t A). These equations can be used to estimate the durability and fines content of wood pellets exposed to a certain volume of rainwater. 47   Figure 2.12 Relationship between fines content and durability with the amount of water sprayed on 2000 g of wood pellets having an initial moisture content of 5.5% (wb). Rain intensity is 1 mm/h. This is only one test with no repetition.  2.2.4.2 Set II tests The moisture content of wood pellet samples obtained from the tray was measured in conjunction with the durability test. Results shown in Figure 2.13 indicate an inverse relationship between durability and moisture content. When the pellets were at the initial moisture content of 5.5% (wb), their durability was 99.6%. But upon wetting the pellets to 12% (wb) moisture content, the durability was reduced to 95.0%. At 10% (wb) moisture content, durability of wetted pellets was 96.5±0.5%. These results indicated that there is no threshold value for wood pellet moisture content at which the durability would fall below 96.5%. The best-fit curve does not have a high regression coefficient (R2 = 0.62). The wider spread of the measured moisture content and hence the durability of the pellets may be attributed to non-uniform wetting of the pellets or variation in the physical structure of the pellets. As shown in Appendix A Figure A.4, when separated into length groups, the as-received pellets’ durability ranged from 98.8 to 99.8% regardless of their 90929496981000 50 100 150 200 250 300 350024681012Mechanical Durability (%)Amount of water sprayed (mL)Fines content (%)Fines content Mechanical Durability48  lengths. Therefore, any spread of the durability data beyond that range must be attributed to external factors such as non-uniform wetting.  A cut-off curve may be deduced from the durability of pellets vs. exposure time and rain intensity, as illustrated in Figure 2.14. The region below and above the cut-off curve represents the durability of wetted pellets that is greater than 96.5% and smaller than 96.5%, respectively. The pellet quality standard requires the durability of wood pellets to be maintained at 96.5% or higher. The results in Figure 2.14 suggested that the loading operation of wood pellets at the shipping terminal should be stopped immediately when rain intensity is 2 mm/h or higher since it will take about 15 min to halt a loading operation. For rain events with intensity less than 0.5 mm/h, loading operations must be stopped after 30 minutes. However, if the rain intensity is between 0.5 and 2 mm/h, no exact recommendations can be made for the loading operation; in this case, the control would return to the shipmaster to make a decision.     Figure 2.13 Durability as a function of moisture content for the water-sprayed pellets.  The error bars represent standard deviations of three replicates.  y = 0.0258x2 - 1.247x + 106.5R² = 0.6290929496981006.0 8.0 10.0 12.0 14.0Mechanical durability (%)Moisture content (% wb)49    Figure 2.14 The cut-off curve of rain intensity versus exposure time to achieve 96.5% durability. The shaded area represents the ±20% uncertainty zone from the curve.  2.2.4.3 Actual rain events Exposure of pellets to actual rain events, as shown in Figure 2.15, resulted in a negative correlation between moisture content and durability. The correlation is somewhat stronger with a regression coefficient R2 = 0.77. After exposure to a very-light rain event (<0.5 mm/h), the wood pellets maintained their high durability of 99.5%, although their moisture content rose to 10% (wb) from an initial moisture content of 5.5% (wb). Furthermore, when the final moisture content of pellets rose to 20% (wb), the durability dropped to 80%. These results demonstrate that if the final moisture content is greater than 10% (wb), pellet durability will fall below 96.5%.   The two mechanisms of water sorption are liquid water absorption and water vapor or hygroscopic moisture adsorption (Amer, 2015). The absorbed water can penetrate the wood pellets causing localized swelling after exposure to a rain event, as was the case for heavier rain-intensity 1101000 2 4 6 8 10Exposure  time (min)Rain intensity (mm/h)Durability < 96.5%Durability > 96.5%y = 19.337x-1.082R² = 0.9250  spraying tests (Set I and II tests). However, during very-light rain events, the dominant mechanism would be mainly hygroscopic adsorption, whereby water vapor would be slowly adsorbed and uniformly distributed onto the individual pellets. This form of moisture sorption has a lesser negative impact on the integrity of wood pellets when compared to a somewhat heavier rain event (sprayed water in Set I and II tests). Nevertheless, in all of the tests conducted, the pellets did not appear to disintegrate upon exposure to different rain intensities. By comparison with reported results in the literature, Mantanis (1994) observed the disintegration of pellets that were wetted by immersion in water to a moisture content of 15% (wb) or higher. Also, Graham et al. (2016) observed substantial disintegration of pellets after the wood pellets were exposed to high humidity conditions for 17 days, and as the moisture content increased from 8.5% to beyond 20% (wb) in their experiment.    Figure 2.15 Durability as a function of moisture content for the rain-wetted pellets.  607080901000.0 5.0 10.0 15.0 20.0 25.0Mechanical durability (%)Moisture content (% wb)Very light rain Light rainy = -1.883x + 114.1R2 = 0.7851  2.3 Conclusions In Section 2.1, residential hardwood, residential softwood, and industrial softwood pellets were exposed to liquid water. The durability, fines content, bulk density, solid density, and porosity of pellets before and after wetting were analyzed. After the different pellets were exposed to water, a substantial quantity of fines was produced; fines content increased from <1% to over 20% for moisture contents above 20%. At moisture contents greater than 10%, wood pellet durability dropped below 97.5%, which is the minimum required durability for marketable pellets according to the CANplus standard. Bulk density of wood pellets first increased from 684 to 711 kg/m3 when m.c. increased from oven-dried state (zero moisture) to 6%. This initial increase was followed by a decrease to below 500 kg/m3 when m.c. reached about 20% wb. The solid density of pellets of softwood pellets was found to decrease more rapidly than that of hardwood pellets as moisture content increased. The specific volume (expansion) and the porosity of pellets showed a steady increase when at moisture content above 6-8% wb. The results in section 1.1 produced a cut-off curve of rain intensity and the corresponding exposure time that would reduce the durability of wood pellets from 99.5% to 96.5%.  Quantitative relationships were obtained for pellet durability, their fines content, and the amount of water to which the pellets were exposed. Results suggested that a loading operation may continue for about 30 min in the event of rain with an intensity of less than 0.5 mm/h. However, the loading operation should stop immediately if the rain intensity rises to 2 mm/h or more. The best-fit functions (R2 = 0.98) were determined that related the amount of water sprayed to the durability and fines content of pellets. However, statistically, the relationship between moisture content and durability of pellets was not highly significant with a R2 value below 0.90.  52  Chapter 3: Sorption Isotherms of Wood Pellets A sorption isotherm is the relationship between the equilibrium moisture content (EMC) of a material and the equilibrium relative humidity (ERH). The EMC-ERH relationship of wood pellets is determined here using the saturated salt solution method at 5, 25, and 35 °C. Wood pellets have slightly lower equilibrium moisture content compared to published data on untreated solid wood. 3.1 Introduction Wood pellets are a popular form of solid biofuel used to reduce the use of coal for heat and power production, especially in Europe (Bioenergy Europe, 2018). Pellets have similar energy content as bituminous coal (~20 MJ/kg dry) and have a high bulk density (650-700 kg/m3). The moisture content of traded wood pellets is around 5% to 7% (wet mass basis). These characteristics allow the economical long-distance transport of wood pellets in bulk (Cocchi et al., 2011). Pellets are reduced to powder in coal pulverization systems for direct injection and combustion (Whittaker & Shield, 2017). One of the limitations of pellets is their susceptibility to breaking during handling and storage, generating small particles and fines, as described in Chapter 2 (Fasina & Sokhansanj, 1992). The presence of fines leads to health and safety complications and an increased risk of dust explosion (Samuelsson et al., 2012). Results in Chapter 2 show that for an increase in moisture content from 5% to 15%, reductions in durability are more severe for pellets exposed to liquid water than for pellets exposed to humid air, i.e., roughly 10% reduction in durability when pellets are wetted with liquid water and roughly 5% reduction in durability when pellets are exposed to humid air (Lee et al., 2019). 53  A sorption isotherm describes the thermodynamic relationship between water activity or equilibrium relative humidity (ERH) and the equilibrium moisture content (EMC) of a material at constant temperature and pressure (Andrade, Lemus, & Pérez, 2011). A knowledge of sorption isotherms of wood pellets is crucial for calculating moisture changes that may occur during storage and ventilation. The difference between the current moisture content of pellets and their equilibrium moisture content can indicate the potential for moisture adsorption or desorption. Similar to any lignocellulosic (plant-based) materials, wood is hygroscopic, which means wood particles have a propensity to exchange moisture with their surrounding environment. The densification of wood particles to produce pellets involves the application of pressure and heat, which changes the hygroscopicity and water sorption characteristics of the material (Acharjee, Coronella, & Vasquez, 2011; Simpson & Skaar, 1968; Skaar, 1988). Several published research reports have described the sorption isotherms of untreated solid wood and thermally treated solid wood (Acharjee et al., 2011; Avramidis, 1989; Rautkari et al., 2013; Simpson, 1971). A few published research papers have described the equilibrium moisture contents (EMC) for wood pellets (Hartley & Wood, 2008; Lam, 2011). Hartley and Wood (2008) reported on EMC vs. ERH for wood pellets at room temperature. They used a set of six salt solutions: lithium chloride, calcium chloride, sodium dichromate, sodium nitrate, ammonium chloride, and sodium sulfate to create six different levels of RH. They placed wood pellets in these environments and waited 200 hours (8.3 days) for the pellets to reach an EMC-ERH equilibrium. Lam (2011) determined the EMC-ERH for wood pellets made from steam-treated wood chips. Steam treatment modifies the structure of biomass and thus the steam treated pellets do not represent untreated wood pellets. Other researchers have reported EMCs of alfalfa pellets (Fasina & Sokhansanj, 1993), switchgrass pellets (Colley et al., 2006), and mixed biomass pellets 54  (Bennamoun, Harun, & Afzal, 2018). Skaar (1988) suggested that bound water in solid wood is adsorbed at or near sorption sites in the wood cell wall, which have a strong attraction for water molecules. The attractive forces are attributed mostly to the attraction of polar hydroxyl groups (partial negative charge) in the wood for the polar water molecules (partial positive charge). From NMR studies of a cellulose-water system, Peemoeller and Sharp (1985) classified adsorbed water as tightly bound and loosely bound. According to their classification, tightly bound water consists of water molecules that are adsorbed to hydroxyl groups in the wood via hydrogen bonding. Loosely bound water is equivalent to free water or bulk liquid water. More recent research (Engelund et al., 2013) has suggested that adsorbed water should only be categorized as bound water and free water. Previous research (Bennamoun et al., 2018; Colley et al., 2006; Fasina & Sokhansanj, 1993) used controlled environmental chambers to set the relative humidity in the air around samples to the desired values. The most common way is to bring the moisture content of the sample into equilibrium with the air at a set relative humidity. This method was used by Bennamoun et al. (2018). However, Fasina and Sokhansanj (1993) and Colley et al. (2006) used the equilibrium relative humidity (ERH) method where the air was brought into equilibrium with a material of fixed moisture content. The primary objective of research in this chapter is to explain the EMC-ERH relationships for wood pellets. In addition, the moisture sorption data for wood pellets is compared with the published literature values for solid wood to determine the degree of difference between the two products. ASABE Standard D245.6 (ASABE, 2012) provides extensive tabulated data for EMC-RH data for products based on the product’s major composition, such as starch, oil, protein, and fiber. 55  But there is no data on EMC-ERH of wood pellets. Traditionally the data on EMC-ERH for wood products have been published in wood-related literature. However, handling and storing chipped, ground, or pelletized wood are similar to handling and storing of grains. These products often share the same storage and handling equipment. The equipment designers and operators may find the data on EMC-ERH useful. 3.2 Methodology Wood pellets used in these experiments were manufactured in British Columbia, Canada, using a mix of softwood species of pine, spruce, and fir (SPF). I purchased the pellets from a Canadian Tire store (8277 Ontario St, Vancouver) in 40 lb (18.2 kg) bags. The average diameter of cylindrical pellets was 6.3±0.1 mm. The length was 15±5 mm (n=50). Jowitt and Wagstaffe (1989) used the following six saturated salt solutions, each in equilibrium with the target relative humidity in parenthesis: lithium chloride (11%), magnesium chloride (33%), magnesium nitrate (53%), sodium chloride (75%), potassium chloride (84%) and barium chloride (90%). The salt solutions were placed in 500 mL plastic containers. Following Jowitt and Wagstaffe (1989) method, microcrystalline cellulose (MCC), a reference material, was used to determine the relative humidity above each of the solutions. Relative humidity (RH) was back-calculated from the moisture content of MCC after two weeks, using a GAB equation (Eq. 3.1). EMC = 0.0382 9.0359(RH)[(1 - 0.7670·RH)(1 - 0.7670·RH + 9.0359·RH)] (3.1) where EMC is equilibrium moisture content in decimal dry basis. RH is relative humidity in decimal. For adsorption tests, wood pellets were oven-dried at 105°C for 24 h prior to placing the pellets in the container (Hartley & Wood, 2008). The initial moisture content of pellets for 56  adsorption tests was assumed to be zero. For desorption tests, wood pellets were placed above distilled water at 25°C for one week. The moisture content of the humidified pellets was re-measured by the oven-dry method, as above, and found to be 26±2% (wb) (n=5).  Duplicates of 2.5 to 3.5 g pellets, approximately 3 to 5 pellets, were placed on aluminum pans and placed on a platform above the surface of the salt solution. All sealed containers were placed in three temperature-controlled environments for four weeks or 28 days. One set of containers was placed in a refrigerator maintained at 5°C. The second and third sets of containers were placed in a water bath maintained at 25°C and 35°C, respectively. At the conclusion of each test, the moisture content of each sample was determined again by the oven-dry method.  Figure 3.1 Two of the six containers, which contained salt solutions and samples.   Two sample t-test assuming equal variances was used to examine whether the values measured from the saturated salt method in this chapter is statistically different from the values given in the literature. The statistical computation was performed using Microsoft Excel 2016’s data analysis tools. The difference is considered significantly different if P-value is smaller than 0.01.  57  3.3 Results and Discussion Figure 3.1 plots the EMC-ERH data for the softwood pellets tested in this research. Each data point represents the moisture content of pellets vs. the relative humidity (RH) of the air in the container after 28 days. The sealed containers were maintained at 5°C, 25°C, or 35°C. Visually, the EMC at the higher temperature (35°C) was slightly lower than EMC for the lower temperature (5°C). From 30% to 70%, the differences were small. The EMC for desorption (solid markers) was higher than the EMC for adsorption (blank markers). Comparing the two ends of the isotherm data, the moisture contents for adsorption isotherms were lower by 5% to 15% than desorption data. Hysteresis in wood, which is the observed difference in moisture contents between adsorption and desorption, is well documented (Time, 1998). For wood pellets, the mean ratio of the moisture content of adsorption to desorption (A/D) was 0.85 to 0.95, similar to the values reported by Time (1998) for solid wood. 58  The data showed that in general, the EMC for wood pellets is lower than EMC for solid wood. The difference between EMC of wood pellets and EMC of solid wood increases at ERH greater than 70%, possibly due to the compacted internal structure for wood pellets. These observations are in agreement with previous research that attributed such differences in EMC to the internal structure of wood species (Krupińska et al., 2007; Time, 1998). Avramidis (1989) developed an isotherm for solid wood, indicating a larger EMC than that predicted by Krupińska et al. (2007) and Simpson (1971). Simpson (1971) and Avramidis (1989) obtained an EMC for Sitka spruce ranging from 0.26 to 0.36 (decimal dry basis) when the RH approached 99%. My data  Figure 3.2 Comparing adsorption (ad) and desorption (de) of equilibrium moisture content (EMC) of wood pellets at 5C, 25C, 35C after 28 days versus relative humidity (ERH). As expected, the EMC for desorption (soil fill markers) is higher than the EMC for adsorption (blank markers). Visually, the EMC for the higher temperature (35C) was slightly lower than EMC for the lower temperature (5C). 59  showed an EMC for wood pellets at 0.32 to 0.34 (dry basis) when RH was 99%. Engelund et al. (2013) showed that solid wood shows a degree of cellulose and lignin softening at RH higher than 60%. This softening may have contributed to the spread in EMC values at higher RH for wood pellets. Table 3.1 Models and their coefficients representing EMC-ERH at 25C.[a] Material Sources EMC-ERH Equation  Coefficients R2 Softwood Pellets (Pine-Spruce-Fir) This work (adsorption) EMC=[ ln(1-ERH)/K]1/C K = -0.01547 C = 1.7105 0.968 This work (desorption) K = -0.01267 C = 1.6798 0.981 Softwood Pellets  Hartley and Wood (2008) EMC = K·ERH K = 0.156 0.998 Solid Wood – Sitka Spruce Avramidis (1989) EMC=[ ln(1-ERH)/K]1/C   K=-0.0212  C=1.5 0.996 [a] Equilibrium moisture content (EMC) is expressed in decimal dry basis and ERH in decimal.  K and C values for Avramidis (1989) are calculated using Day Nelson equation temperature-dependent coefficients.   As shown in Figure 3.3, the slope of EMC-ERH isotherm for wood pellets was slightly smaller than the slope of isotherm for solid wood. The flatter EMC curve indicates the pellet’s EMC is less sensitive to RH in the range of 30% to 70% than the solid wood is. A plausible reason for the smaller slope of EMC-ERH isotherm of wood pellets is that water vapor takes a longer time to penetrate the pellets as the pellets have a higher density (1.2 to 1.4 g/cm3) compared to solid wood (0.3 to 0.5 g/cm3). Hartley and Wood’s lower EMC than those obtained here might be due to their shorter conditioning time for the pellets to equilibrate with RH. 60  In a pellet die, the majority of the heating occurs on the outer sides of the pellet. This heat plasticizes lignin and binds particles to each other to create a polished outer appearance (Stelte et al., 2011). The outer layer may protect the pellet from adsorbing water vapor from ambient humidity (Whittaker & Shield, 2017). The two ends of a pellet are jagged and not plasticized. The ends provide the main entry to water vapor during moisture adsorption, where ~40% of moisture was adsorbed through the two ends, which accounted for ~20% of the total surface area of a single  Figure 3.3 Moisture content vs. relative humidity (EMC-ERH) data points from the present study (25C adsorption tests) and the curves drawn from selected EMC-ERH equations available for solid wood (Avramidis, 1989) and wood pellets (Hartley and Wood, 2008). The solid line is Henderson’s equation fitted to my experimental data. The fitting coefficients are given in Table 3.1. 0.000.050.100.150.200.250.300.350.00 0.20 0.40 0.60 0.80 1.00Moisture content (dec. db)Relative humidity (dec.)Hartley and woodAvramidis61  pellet, as shown in Appendix C.1. As a result of the limited water entry sites, it may take a longer period for a wood pellet to equilibrate with RH. This observation was tested using a t-test to compare the Henderson equation’s data for wood pellets (this chapter) versus the data for solid wood generated using Avramidis (1989)’s equation. The difference between the equilibrium moisture content data for wood pellets and solid wood (Avramidis, 1989) was statistically significant, with a P-value of 0.0008, which is less than 0.01 (Table 3.2). Table 3.2 The computed results for a two-sample t-test assuming equal variance to compare the difference between Henderson equation’s data for wood pellets and Avramidis (1989)’s equation for solid wood. Sources Avramidis (1989) for solid wood Results for wood pellets Mean 0.1262 0.1086 Variance 0.0076 0.0044 Observations 20 20 Pooled Variance 0.0061  Hypothesized Mean Difference 0 df 38 t Stat 3.6868 P(T<=t) one-tail 0.0008 t Critical one-tail 1.7291  The spread of the moisture content data at higher humidities is greater than that at lower humidities. Although previous research reported that the sorption EMC curves at lower temperatures were higher than the EMC curves at higher temperatures, the variability at higher humidity masked this distinction (Avramidis, 1989; Basu, Shivhare, & Mujumdar, 2006; Time, 1998). At 90% RH, the moisture content at 25°C (18.1% wb) was higher than the moisture content at 5°C (15% wb). The temperature gradient (2°C-5°C) existed within the containers in which pellets were brought in equilibrium with the air above the saturated salt solution. Air closer to the solution surface had a slightly higher temperature than air farther away due to the heat of sorption. As a result, relative humidity may vary due to this temperature gradient (Engelund et al., 2013). 62  3.4 Summary Saturated salt solution method was used to determine the equilibrium moisture contents (EMC) isotherms of wood pellets at 5, 25, and 35 °C, for both adsorption from zero moisture and desorption from a moisture content of 26% (wb). The equilibrium moisture contents at the three temperatures overlapped, but overall, the desorption isotherms were higher than the adsorption isotherms. The EMC isotherm for wood pellets at 25°C is on average 0.01-0.05 (decimal, db) lower than the EMC for solid wood, possibly because the surface of pellets is plasticized when they are manufactured. The values of EMC of wood pellets diverged from those of solid wood at the relative humidity (RH) larger than 70%. Henderson’s equation has the best fit for the wood pellets’ isotherm data. The slope of adsorption isotherm for wood pellets was slightly smaller than that of solid wood, indicating that pellets are less sensitive to RH in the range of 30% to 70%. 63  Chapter 4: The Effects of Humidity on Single Pellets The durability of single pellets was measured using a novel single pellet durability tester, i.e. “shaker,” after they were exposed to an environment with a temperature of 30 °C and RH of 95%. Figure A.3 in the Appendix provides this correlation equation to convert single pellet durability to durability, measured by a tumbler. Single pellet durability was strongly correlated to the volumetric swelling of the individual pellets. The variability of single pellet durability was shown to be caused by the variability of swelling behaviors of pellets. 4.1 Introduction Wood pellets are hygroscopic and adsorb moisture from a humid environment during storage and handling. The moisture sorption induces volumetric expansion that leads to a drop in durability (Colley et al., 2006; Fasina & Sokhansanj, 1992; Hartley & Wood, 2008; McMullen et al., 2005). Fasina & Sokhansanj and Colley et al. reported that the density of alfalfa and switchgrass pellets first increased then decreased as moisture content increased to 7-8% for alfalfa pellets with the adsorbed moisture. Hartley & Wood observed that the relationship between the volumetric swelling of wood pellets and the relative humidity could be described using a power-law equation. These studies were done in bulk amounts of 100 to 1000 g pellets, while the uncertainty and the applicability of their data were not discussed.  Peng et al. (2013) compared the moisture uptake of steam-treated and untreated wood pellets in an environmental chamber and showed that the steam-treated pellets had less propensity for moisture adsorption. Deng et al. (2019) measured the sorption isotherm and durability of pine wood pellets and recycled wood over a range of relative humidities and found that durability only decreased appreciably by 5% at a relative humidity (RH) above 80%. Wang et al. (2016) observed that cedar wood pellets stored at 60°C and 90% RH for five days had a moisture content of 29% 64  wb, while the hardness of pellets decreased from 5.45 to 0.15 N/mm2. During a study of the long term impact of weathering (10 to 20 months) on the mechanical and chemical properties of biomass pellets during storage, Graham et al. (2017) noted that the mechanical degradation of the pellets resulting from moisture uptake was more substantial than the chemical or biological degradation. After one month of storage at a relative humidity above 90%, untreated wood pellets had disintegrated, and their durability was essentially zero. Steam-treated wood pellets were more durable with a durability rating of around 87%. None of these studies addressed specifically the rate of moisture uptake and its effect on durability and density of softwood (spruce, pine, fir) pellets. Several factors, such as particle size distribution within individual pellets (Jensen et al., 2011), the local temperature gradient in the die, the handling history of the pellets, and varying swelling responses affect the physical qualities of a wood pellet.  4.1.1 Objectives Previous research reviewed here does not address the time-dependent nature of moisture adsorption and the effects of adsorption rate on the durability and density of wood pellets. In practice, the pellets can be exposed to humid conditions where there is airflow in a ventilated silo. Wood pellets may also be exposed to a stagnant environment in which airflow is minimal. Pellets stored in a ship hold or in a rail car are examples of the latter. The objective of the research in this chapter is to expose softwood pellets to two different sets of environmental conditions, one in a humid chamber with active air circulation and one over a number of salt solutions with no active airflow. The durability and density of the pellets were then examined.   65  4.2 Materials and Methods 4.2.1 Materials The residential softwood pellets used in these experiments were manufactured in British Columbia, Canada, using a mix of softwood species: pine, spruce, and fir (SPF). I purchased the pellets from a hardware store in 40 lbs (18.2 kg) bags as described in Chapter 3. The pellets were 6.2 mm in diameter. The length of pellets varied from 10 mm to 25 mm; 94 pellets out of 155 pellets had a length between 15-20 mm.  4.2.2 Moisture adsorption  Figure 4.1 Espec LHU-113 benchtop temperature/humidity chamber.  A benchtop humidity chamber (Model: LHU‐113; ESPEC North America, Inc., Hudsonville, Mich.), as shown in Figure 4.1, is equipped with digital control provided the desired temperature and humidity environments. Interior dimensions of the chamber were 500 mm wide × 390 mm deep × 600 mm high. For our study, the temperature and RH in the chamber were set at 30°C and 95%, respectively. Eight batches, each consisting of 20 pellets, were oven-dried at 105°C for 24 hours. For the first seven batches, one pellet in an aluminum tray was placed in the humidity 66  chamber. The pellets were left in the chamber for 1, 2, 4, 7, 13, 18, or 24 hours. The pellets in untreated batch number 8 were each tested for mass, volume, and durability. Similar measurements were made on the pellets as they were removed from the chamber.  The resulting dry basis moisture content (M) of pellets was calculated as the increase in the mass of pellets over the mass of the oven-dried sample. In the second series of moisture adsorption experiments, pellets were exposed to elevated relative humidity conditions over saturated salt solutions. Jowitt and Wagstaffe (1989) reported the use of the following six saturated salt solutions and their corresponding equilibrium relative humidity (in parenthesis): lithium chloride (11%), magnesium chloride (33%), magnesium nitrate (53%), sodium chloride (75%), potassium chloride (84%) and barium chloride (90%). Similar salt solutions were placed in 500 mL plastic containers, as described in Chapter 3. About three grams of wood pellets was oven-dried at 105°C for 24 hours. The dried pellets were placed on a platform above a salt solution. The sealed container, containing the pellet and the salt solution, was placed in a temperature-controlled water bath of 25°C for 28 days (Lee et al., 2019). At the conclusion of each test, the moisture content of each sample was determined again by the oven-dry method.  The saturated salt solution method influences the moisture content of pellets by altering the relative humidity of the atmosphere that pellets are exposed to. In the humidity chamber method, the moisture content of pellets was controlled by setting the exposure time. Because the former method takes much longer than the latter method to achieve a target moisture content (days, instead of hours), polymer relaxation (Engelund et al., 2013) can be excluded as a reason for any spread in durability data. If the large variability in durability exists in data obtained from both methods, then the variability should be caused by the inherent moisture-related properties of pelletized wood.  67  4.2.3 Physical dimensions, structure, and appearance ImageJ software (National Institutes of Health, version 1.52s 10 December 2019) was used to measure the lengths and diameters of the individual wood pellets. The dimensions were used to calculate pellet volume. Pellet density was calculated as the mass of pellets over their volume. Individual pellets were photographed before and immediately after conditioning in the humidity chamber using a hand-held digital camera with an auto-focusing function.   The samples were air-dried prior to the X-ray scanning. A softwood pellet was scanned using an X-ray micro-CT system in the Department of Applied Mathematics at the Australian National University (ANU) by Professor Philip Evans from the UBC Department of Wood Science. The pellet was placed on a rotating stage while an X-ray tube projected a cone-shaped X-ray beam onto the pellet. A CCD (charge-coupled device) was used to collect the projection data. The resolution of the CT scan was 2800 × 2800 voxels with a voxel size of 2.92 µm. The CT images could be thresholded into two phases: air and wood. The 2D visualization of the tomograms was carried using the software NCViewer. 4.2.4 Durability Durability in this study was measured using a novel single pellet durability tester “shaker” (Schilling et al., 2015). The shaker is a metal box (60 × 60 × 60 mm) installed on the arm of a laboratory shaker. The shaker box held one pellet and a piece of steel of dimensions 6.3 mm diameter and length 12 mm. During the durability test, a pellet is subjected to shaking motion in the metal box for 10 minutes. The single pellet durability, Dsingle is the ratio of the mass of pellet pieces (mf) over a 3.15 mm round sieve to initial mass (mi): Dsingle = mfmi × 100          (4.1) 68   Figure 4.2 The diagram of the single pellet durability tester. A small metal box is attached to the wrist-action shaker (Schilling et al., 2015).  Durability is considered high when its value is above 90%, medium when it is between 80% and 90%, and low when it is below 80% (Colley et al., 2006; Schilling et al., 2015). To compare the durability results obtained from the humidity chamber and saturated salt solutions, a two-sample t-test assuming equal variances was used. The statistical computation was performed using Microsoft Excel 2016’s data analysis tools. The difference is considered significantly different if P-value is smaller than 0.01. 4.3 Results and Discussion 4.3.1 Surface conditions and internal structure Figure 4.4 shows photographs of wood pellets before (oven-dried) and after they were exposed to humid air at seven different exposure times. After exposure for one hour, the surface of wood pellets became lighter in color, turning from its original yellowish-brown to a light yellow. This color change intensified gradually during their exposure to the humid air. Baar et al. (2019) suggested that the brightening upon moisture adsorption is due to the increased reflective index caused by the reactions between water-soluble extractives and adsorbed water.  69  Small cracks, not visible in Figure 4.3(e), appeared mostly along the length of the pellets after seven hours. The wood particles on the pellet’s surface swelled and this swelling pushed the particles away from the pellet matrix. The expansion of wood particles within the pellets increased with moisture adsorption causing cracks to widen. Such expansion caused small flakes to separate from the pellet surface after 13 hours of exposure. After 24 hours, some pellets disintegrated into loosely held wood particles, while others remained mostly intact.     Figure 4.3 (a) as-received pellets and groups of five wood pellets after exposure to the humidity of 95% RH and the temperature of 30°C, (b) for 1 hour, (c) for 2 hours, (d) for 4 hours, (e) for 7 hours, (f) for 13 hours, (g) for 18 hours, and (h) for 24 hours.  70  The cracks, while might not be visible on the surface of the as-received pellets, penetrate deep into the body of a pellet, as shown in the micro-CT scan images in Figure 4.4. These cracks allowed humid air to flow into a pellet. The moisture was then adsorbed into the pellet’s internal matrix and consequently causes swelling in the pellet. Additional micro-CT scan images and the procedure are given in Appendix B.4.  Figure 4.4 A two-dimensional image of cross-sections through an untreated wood pellet. Cracks penetrated deep from the surface into the body of the pellet.  4.3.2 Kinetics of moisture adsorption   Figure 4.5 plots the moisture adsorption of wood pellets at 30°C and 95% RH in a humidity chamber. The pellets took less than 10 hours to reach a moisture content above 0.14 in decimal dry basis (db). It took more than 24 hours for pellets to reach an equilibrium of ~0.195 (db). This equilibrium moisture content was estimated by using the parameters in the Page model (Eq. 4.2) (Igathinathane et al., 2009), assuming initial zero moisture content at time zero. M = Me [1- exp(-ktn) ]          (4.2) 71  where t is the exposure time (h), M is the instantaneous moisture content of a pellet (db), Me is the equilibrium moisture content of a pellet (db), k is the moisture sorption rate (h-n), n is the dimensionless exponent.    Figure 4.5 Moisture content of wood pellets, expressed in decimal dry basis (db), at seven exposure times: 1, 2, 4, 7, 13, 18, 24 hours, showing data for 20 replicates for each time interval. The solid curve represents Page’s model (Eq. 4.2) fitted to the data.   Moisture adsorption of residential hardwood, residential softwood, and industrial softwood pellets in the atmosphere at a temperature of 30 °C and 95% RH is given in Figure A.1 for comparison.   The estimated values for the moisture sorption rate constant (k) and equilibrium moisture content (Me) at 95% RH and 30°C are given in Figure 4.5. The value of R2 (>0.99) showed that Eq. 4.2 predicted the experimental data well. The coefficient of variation (CV = ratio of standard deviation over mean), an indicator of variability of data, decreased from 11% to 3.5% of the average values of moisture content at each exposure time, as shown in Figure A.5(a). This decrease 0.000.050.100.150.200.250 5 10 15 20 25Moisture content (db)Exposure time (hours)M = 0.195*[1 - exp(-0.3185t0.6553)] R² = 0.996472  in variability in moisture content suggests that increases in the variability of durability and swelling were not related to the variability in pellet’s moisture content. As previously shown in Chapter 3, the moisture content data at 25°C for softwood pellets exposed to RH over the salt solutions was sigmoidal (type II) (Jiang et al., 2019). As shown in Figure 4.6, the moisture content increased almost linearly with relative humidity until a RH of 84% and a moisture content of 0.158 were reached. Moisture content then increased sharply and approached 0.30 when RH approached 100%. Hartley and Wood (2008), as well as Sjöström and Blomqvist (2014), reported similar results on their softwood pellets.   Figure 4.6 Moisture content (db) versus relative humidity at the temperature of 25°C. The total number of data points is 36. Data were adapted from my previously published data (Lee et al., 2019). The fitting curve represents Eq. 4.3.  The relationship of moisture content vs. RH data was analyzed using the Henderson equation (ASABE D245.6): 𝑀= {-1Kln(1 - rh)1C} /100         (4.3) 0.000.050.100.150.200.250.300% 10%20%30%40%50%60%70%80%90%100%Moisture content (db)Relative humidity73  where M is the moisture content (db), rh is the relative humidity in decimal, K and C, the model coefficients are found to be 0.01547 and 1.7105, respectively. The coefficients of Eq. 4.3 were estimated using the solver program in Microsoft Excel by maximizing the coefficient of determination (R2). The value of R2 (0.9677) indicated that Eq. 4.3 fitted the moisture content vs. RH data quite well.  4.3.3 Durability The durability of the individual pellets remained almost constant with an average of 96.6% at 0.05 (db), compared to 97.5% for oven-dried samples. Further increases in moisture content initially reduced durability gradually to 94.4 at 0.108 (db), but thereafter durability abruptly decreased to 85.6% at a moisture content of 0.137 (db). A similar observation was reported by Theerarattananoon et al. (2011) for wheat straw and big bluestem pellets. This might be due to the low initial moisture-induced swelling that is insufficient to disrupt the bonds between individual particles in the pellets. At low moisture contents, the adsorbed moisture may have acted as a bridge between binding sites between particle surfaces (García et al., 2019). Further increases in moisture caused greater swelling and cracking. Such effects caused some pellets to fragment and break (Thomas & Van der Poel, 1996).  Using a similar classification system as Colley et al. (2006), wood pellets here can be classified as being of high durability (>90%) at moisture content above 0.12 (db), medium durability (80%-90%) at 0.120 to 0.145 (db) and low durability (<80%) at M > 0.145 (db). The following two-coefficient equations describe the relationship between pellet durability and moisture content well. Dsingle, HC = 1 - 6.4828E-04(1 - M)35.09 ,  R2 = 0.8364       (4.4) 74  Dsingle, SS = 1 - 3.0755E-03(1 - M)23.52 ,  R2 = 0.7140       (4.5) Eq. 4.4 is fitted to only the durability data obtained using pellets conditioned in a humidity chamber (HC) and Eq. 4.5 is fitted to only data obtained from pellets exposed above saturated salt solutions (SS). While initially below 10%, the coefficient of variation (CV) of durability increased sharply to over 20% at moisture contents above 0.15 (db) for durability data obtained from both methods as shown in Figure A.5(b).   Figure 4.7 Single pellet durability (Dsingle) vs. moisture content in decimal, dry basis (db) for samples exposed to 95% RH and 30°C in a humidity chamber (filled circles - HC) and exposed to atmospheres above saturated salt solutions (open circles - SS). The total number of data points is 196. The two fitting curves represent Eqs. 4.4 and 4.5 for HC and SS, respectively.  For the pellets exposed to different atmospheres above salt solutions, durability dropped below 90% at a moisture content above 0.14 (db). Compared to data points in the moisture content range of 0.10 and 0.15 (db) in Figure 4.7, the durability for pellets above salt solutions at a moisture 0%10%20%30%40%50%60%70%80%90%100%110%0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24Single pellet durability, DsingleMoisture content (dec, d.b.)HCSSSSHC75  content of 0.128 (db) was ~96.8%, which was higher than the HC durability data (~87%) at similar moisture content.  To compare the durability results obtained from the humidity chamber and saturated salt solutions, a two-sample t-test assuming equal variances was used. Mean durability is the average value of durability. Variance is the average value of the squared deviation of the measured value from the mean durability. Degree of freedom, df is calculated as (n1+n2-2), where n1 and n2 are the number of observations in the two sets of durability data. The difference is considered significant if P-value is smaller than 0.01. The difference in durability for 0.10 < M < 0.15 is statistically significant (P-value < 0.01) as shown in Table 1. The durability data were not statistically different for M < 0.10 (P-value = 0.086) and M > 0.15 (P-value = 0.114), as shown by the t-test in Table 4.1. The lower moisture adsorption rate of pellets above salt solution appeared to result in less damage to the pellets’ structural integrity than that of pellets in the humidity chamber. Above the salt solutions, the pellets’ matrix may have had time to relax during volumetric expansion, while in the case of humidity chamber, the pellets’ matrix may have expanded more quickly, which resulted in greater damage to the pellets’ structure.  Table 4.1 The results from two-sample t-test assuming equal variances to test the differences among the durability data for humidity chamber (HC) and saturated salt solution (SS) for three moisture content ranges: (a) 0 – 0.10 (db), (b) 0.10 – 0.15 (db), and (c) > 0.15 (db). Critical is obtained at alpha value of 0.01.  Parameter M < 0.10 0.10 < M < 0.15 M > 0.15 HC SS HC SS HC SS Mean Durability 95.7 97.1 86.8 96.3 49.2 36.5 Variance 14.4 6.8 103.4 7.9 652.8 1268.5 Number of Observations 44 18 45 10 46 8 df 60   53   52   t Stat -1.385 -2.900 1.220 P(T<=t) 0.086 0.003 0.114 t Critical 2.390 2.399 2.400 76   4.3.4 Pellet volume and density  On average, the volumetric swelling of pellet samples increased according to the power-law equation with an increase in moisture content (Figure 4.8). Hartley and Wood (2008) developed a similar power-law relationship between volumetric swelling and moisture content. The swelling behavior of each pellet varied substantially with a coefficient of variation of over 30%, regardless of its moisture content. This variability suggests that the effect of moisture adsorption on the swelling of pellets was not uniform, although the variability in moisture content data was small (2-11% in CV), as shown in Figure A.5(a & c). For example, at the moisture content of 0.18 (db), the volumetric swelling was 53% on average, but the maximum was in excess of 80% and the minimum was 24%.    Figure 4.8 Volumetric swelling in percent of the initial volume (%initial volume) vs. moisture content. The total number of data points is 160. The fitting curve represents the power-law equation.   010203040506070800.00 0.05 0.10 0.15 0.20Volumetric Swelling (% initial volume)Moisture content (db)%S = 1.3378e18.725MR² = 0.833677  Lam et al. (2014) reported that at M of 0.08-0.11 (db), the density of Douglas fir pellets was 900 to 1230 kg/m3. This matches the range of pellet densities measured at the same range of moisture content in this chapter.  The CV of the pellet density data remained relatively constant between 6 to 9% at moisture contents below 0.15. As shown in Figure A.5(d), the variability increased sharply to 16% at M of 0.18 (db). At this M, pellet density goes as low as 581 kg/m3 and as high as 1100 kg/m3. This variability is partly due to neglecting the unevenness of the two pellet ends and the difficulties in determining the volumes of the pellets. In addition, the varied effects of moisture adsorption on individual pellets probably contributed to the larger variability at higher moisture contents. A quadratic equation, given in Figure 4.9, adequately describes the relationship between pellet density and moisture content. Colley et al. (2006) used the same form of equation to describe the relationship between density and moisture content. Pellet density is inversely related to the volume of a single pellet. Its average value initially increased slightly from 1000 kg/m3 to 1059 kg/m3 at M of 0.05 (db), then decreased with further increase in moisture content to as low as 854 kg/m3 (Figure 4.9). This is a result of the increase in mass due to moisture gain being lower than the accompanying volumetric expansion of the pellet, as also described in Chapter 2.  78    Figure 4.9 Density of wood pellets vs. moisture content. The total number of data points is 160. The fitting curve represents a 2nd order polynomial equation.  4.3.5 Significance of the results Research in this chapter suggests that decreases in the durability of individual pellets as a result of moisture adsorption is caused by the volumetric swelling of pellets. The variability in the swelling behavior of pellets is one of the sources of variability in the durability of moistened pellets. In other words, individual pellets swell differently due to their differing density and other physical qualities. Therefore, their resistance to impact and abrasion, which is measured by durability, varies. The durability of zero moisture pellets varied from a minimum of 90.7% to a maximum of 99.9% (Figure 4.7). My result suggests that this heterogeneity in the durability of single pellets 50060070080090010001100120013000.00 0.05 0.10 0.15 0.20Pellet  Density (kg/m3)Moisture content (db)ρ = -14670M2 + 2068M + 1000R² = 0.761879  was magnified when their moisture contents increase. If the durability of a pellet at zero moisture was in the lower end, then moisture adsorption resulted in a greater drop in durability; if a pellet had durability near the higher end, then its durability stayed near its zero moisture durability when the pellet’s moisture content increased to over 0.15 (db). In an aeration system, which uses humid ambient air, the flowing air may cause fragmentation in pellets weakened by adsorption-induced expansion. Accordingly, if toxic off-gas is to be removed from ship holds or silos, the time of ventilation should be limited to under 1 hour. If it is necessary to cool pellets, the air must be dehumidified to a relative humidity below 50% to avoid moisture adsorption, as the cooling of bulk materials in a large storage to room temperature usually takes days to accomplish (Wang et al., 2017). Nilsson et al. (2019) reported that water vapor uptake in fuel pellets and the resulting release of the heat of sorption play a crucial role in initiating temperature rise and spontaneous combustion of pellets. Therefore, an added benefit of cooling pellets with dehumidified air is that temperature rises due to the heat of moisture sorption will be restrained by the lack of moisture adsorbed from the air.  It is clear from my results that the humid ambient air will cause losses in the durability of pellets in hours or weeks depending upon whether the air within the bulk is moving or stagnant. Therefore, it is imperative to transport pellets to their final destinations in less than four weeks to avoid any physical damage caused by moisture adsorption during storage. This advice is further supported by similar observations by Graham et al. (2017) in their study of long-term storage in which they observed that untreated pellets disintegrated after one month at high humidity (> 80% RH) indoor storage. A linear regression analysis shows that the volumetric swelling of individual pellets is negatively correlated with durability with a correlation coefficient (R) of -0.9776. This accords 80  with the general consensus in the literature (Obernberger & Thek, 2010) that volumetric swelling is associated with decreases in durability. In addition, the variability in the swelling behavior in individual pellets, as indicated by the large CV values, might be the cause of the large variability in durability (Figure A.5). Figure 4.10 shows a linear relationship between volumetric swelling and durability. As far as I am aware, no previous work has made this observation for wood pellets post-production. Previous studies, summarized by Whittaker & Shield (2017), focused on the effect on feedstock qualities, such as particle size and moisture content on pellet durability.  Figure 4.10 The relationship between volumetric swelling (%) and durability, based on average values.  4.4 Conclusions This research in this chapter examines the effect of moisture adsorption on single pellet durability, volumetric swelling, and pellet density.  At moisture contents above 0.15 (db), slower moisture adsorption by pellets above salt solutions resulted in durabilities above 96.8%, whereas the faster moisture adsorption in a humidity chamber was associated with a more severe decrease in the durability to ~87%. However, in both 010203040506040% 50% 60% 70% 80% 90% 100%Swelling (%)Durability%S = -86.77*DI + 87.88R² = 0.955781  adsorption tests, the average durability dropped below 80% at the moisture contents above 0.15. The variability in pellet durability, expressed as coefficient of variation, increased to above 20%, while the variability in moisture content data was much smaller (4%) in the same environment. In addition, the volumetric swelling of wood pellets is negatively correlated with their durability with a correlation coefficient of -0.9776. Therefore, I conclude that variability in durability data at higher moisture contents was associated with the large variability in swelling and density.   Wood pellets aerated with humid air may be more prone to degradation than pellets in a stagnant environment. Therefore, it is advisable to dehumidify humid air to below 50% RH to lessen the potential for moisture uptake. For medium-term storage, in humid environments that consistently have ambient relative humidities above 80%, the durability of pellets decreases significantly due to the uptake of moisture.  82  Chapter 5: Heats of Wetting and Sorption for Wood Pellets Wood pellets at a range of initial moisture contents were immersed in water in a calorimetry setup to measure their heat of wetting. Their heat of wetting fits a linear equation better than an exponential equation. The maximum heat of wetting of wood pellets is statistically the same as that of wood chips. However, the wood pellets release heat faster than wood chips.  5.1 Introduction The state of water in a piece of cellular wood (untreated wood) is categorized into two forms: free water that exists within a wood cell lumen and bound water or sorbed water that is in the cell wall, as mentioned above (Skaar, 1988). The free water evaporates first followed by the bound water when the wood dried. The transition of moisture content from free to bound water is called the fiber saturation moisture content. Wood does not shrink during drying as long as its moisture content remains above the fiber saturation point (Simpson, 2001). The fiber saturation moisture content for wood ranges from 0.25 to 0.40 dry mass basis (db) depending on wood species and temperature (Engelund et al., 2013; Skaar, 1988). The amount of energy applied to wood during free-water drying is close to pan evaporation, i.e. it takes roughly 2525 kJ/kg to evaporate water at 25 °C (Leuk et al., 2016). Upon further drying to a moisture content below the fiber saturation, the wood cell wall loses its moisture and this loss of moisture is accompanied by shrinkage. It takes extra energy to remove the bound water from wood. For example, drying wood to a moisture content of 0.05 (db) would take another 1000 kJ per kg of water, in addition to the heat energy required to evaporate free water (Nilsson et al., 2019). Skaar (1988) states that the heat needed to extract this bound water is analogous to the heat required to melt ice into liquid water. Differential heat of sorption (adsorption or desorption) is nearly reversible, although there is some degree of hysteresis. Approximately an equal amount of heat applied to dry a substance 83  will be released when that substance is rewetted (Bogolitsyn et al., 1995; Nopens et al., 2019). Adsorption refers to water molecules attaching to a surface of a material and potentially hydrogen-bonding with the material. Absorption refers to a bulk amount of liquid water penetrating or permeating a material (Amer, 2015).  Elevated atmospheric humidity is known to contribute to the self-heating of coal and wood (Kubler, 1990; Nelson & Chen, 2007). Moisture adsorption from the atmosphere elevates the temperature of the material and consequently increases the rate of oxidation and hydrolysis (Bhat & Agarwal, 1996). Since pellets and coal broadly have similar chemical properties, the same effects may be relevant to wood pellets.  In modeling the contribution of water vapor adsorption to self-heating of pellets, published literature (Cao & Kamdem, 2004; Lestander, 2008) assumed the differential heat of sorption or the additional bonding enthalpy of water adsorption for pellets is the same as that of untreated wood. On the other hand, Nilsson et al. (2019) ignored the differential heat in their calculation of temperature rise due to water vapor adsorption. Wood pellets are made by compressing wood particles at temperatures above 100 °C. This high-temperature compression may reduce the hygroscopicity of the pellets (Fang et al., 2012) and therefore alter the bonding enthalpy between water molecules and wood fibers. 5.2 Definition of Heat of Wetting Heat of wetting is the heat evolved when wood is mixed with water. The heat of wetting is maximum when the wood is bone dry (Skaar, 1988). The heat of wetting decreases to zero when the moisture content of wood approaches its fiber saturation point. The unit of heat of wetting is kilojoules per kilogram of the dry mass of the solid. Skaar (1988) reported the range of the maximum heat of wetting (Q0) for dry wood from 52 to 76 kJ/kg for the heartwood of a number 84  of species. Skaar (1988) mentioned that wood with higher extractive content has a lower heat of wetting. The heat of wetting for sapwood was higher than that for heartwood because of a lower content of extractives in the sapwood.  Figure 5.1 depicts the enthalpies associated with the state of water in wood. In this figure, hs (kJ/kg of water) represents the enthalpy of water below the fiber saturation point, where water exists as bound water in the wood. From Figure 5.1, the maximum heat of wetting is expressed as an integral of differential heat of sorption, as follows:  Q0 =∫ hsMf0d(M) (5.1) where hs is the differential heat of sorption, which is expressed in the unit of heat (calories or kilojoules) per unit mass of water (kg). The heat of wetting at any moisture below the fiber saturation moisture content Mf is:   Q =∫ hsMfMd(M) (5.2) It follows that the integral heat of sorption at moisture content M is defined by: (Q0 - Q) =∫ hsM0d(M)         (5.3) Skaar (1988) applied the following exponential equation to estimate the heat of wetting as a function of wood moisture content.  Q = Q0 exp(-B∙M) (5.4) where B is a constant, M is moisture content in decimal dry basis (db). Skaar (1988) reported B = 14 and Q0 = 84 kJ/kg for wood tested for heat of wetting at a temperature of 50°C. Eq. 5.3 can be differentiated with respect to M to yield the differential heat of sorption. 85   hs = d(𝑄0 - Q)dM=-dQdM (5.5) Differentiating Eq. 5.4 with respect to M and combining with Eq. 5.5 yields:  hs = BQ0 exp (-B∙M) (5.6) Q and hs vs. M equations have similar exponential forms; both curves approach zero when M increases.  Figure 5.1 Water in wood phase diagram, adapted from Skaar (1988). hv is the enthalpy of vaporization or condensation. hs is the differential heat of sorption. The shaded area above the bound water curve is the heat of wetting for the wood at a moisture content of M. Mf is the fiber saturation moisture content below which water exists only in the cell wall of wood.   A theoretical method of estimating hs for wood placed in a humid environment is the use of the Clausius-Clapeyron equation, as given in a differential form in Eq. 5.7. This equation is 86  based on the use of the equilibrium moisture content (EMC) vs. relative humidity (RH) relation for wood at several temperatures.   dPvdT=hsTv       (5.7) where Pv is vapor pressure (from EMC-RH relation). T is temperature (absolute). hs is differential heat of sorption. v is specific volume of water vapor.  A plot of Pv vs. T provides an estimate for hs. The Clausius-Clapeyron method is called the isosteric method. Fasina and Sokhansanj (1993) used the isosteric method to estimate the differential heat of sorption for alfalfa pellets.  The calorimetric method using the water soaking method, according to Skaar (1988), gives an accurate result provided the sample is small enough to become thoroughly wet and the heat loss from the mixture of water and wood particles is minimum. Recently, Leuk et al. (2016) used both the isosteric method and the direct calorimetric method to estimate the heat of sorption of small wood pulp and untreated softwood. They found good agreement between the values for the differential heat of sorption calculated from the isosteric method and the direct calorimetric method.   There is no published data on the heat of wetting of wood pellets. This chapter investigates the heat of wetting of British Columbian premium-grade wood pellets when the pellets become in contact with water in the liquid phase. My experiments used the calorimetric method. 5.3 Experimental Methods 5.3.1 Sample description The premium-grade softwood pellets were acquired from a Canadian Tire big box store in Vancouver in the summer of 2015, as described in Chapter 3. The wood pellets produced commercially in the Province of British Columbia (BC) are generally made from pine with a small portion of Douglas fir and spruce (SPF). Wood pellets had an average diameter of 6.3 mm and 87  lengths varying from 12 to 24 mm. The average density of the wood pellet was 1.2 g/cm3. A single pellet weighed 0.5-1.0 g. Western red cedar wood chips, produced in BC, are rectangular in shape, 10-20 mm in width and length, and 4-5 mm in thickness. The average density of the wood chip was 0.4 g/cm3. A wood chip weighed 0.4-0.8 g.  Samples were weighed using a digital analytical balance (0.001 g precision). The dimensions of samples were measured using a digital caliper. The density of samples was the weight of the sample divided by their volume. The formula for a cylinder and a cuboid were used to calculate apparent volumes of pellets and chips, respectively. I did not measure the extractive contents of the samples. 5.3.2 Water absorption  Prior to experimentation, the moisture contents of wood pellets were measured following a 24-hours drying in a convection oven at 105°C. Approximately 1 L of water was heated in a beaker to ~30°C. The water temperature was kept constant by placing the beaker in a water bath (Fisher Scientific™ Isotemp™ Digital-Control Water Bath model 2300). For each experiment, 10 g of dried wood pellets were placed on a tray with a wire mesh bottom (0.85 mm size hole size). The tray containing pellets was immersed in a water-filled beaker. The water-soaked pellets were removed from the beaker in 30-seconds intervals for a period of six minutes. Each sample was weighed for 12 times during this period. The surface of the pellets was gently blotted using a paper towel to remove excess water from the sieve. The blotted pellets were weighed using a digital balance (0.001 g precision). The pellets were then returned to the beaker to continue soaking in water. The experiment was terminated once the weight of the blotted pellets did not change following the three wetting and blotting cycles. I assumed that the pellets had reached their 88  maximum water-absorbing capacity at this point. Five repetitions were performed for the wood pellet samples.  5.3.3 Heat of sorption by calorimetry The heat of wetting of wood pellets was measured by mixing 100 g pellets with 200 g water. Prior to mixing, the water was kept at 20-23 °C in a 500 mL vacuum insulated flask, as shown in Figure 5.2. The initial moisture content of pellets was adjusted by drying or wetting them to moisture content ranging from 0.005 (near zero) to 0.160 (db). The moisture-adjusted pellets were mixed with the water in the flask. A second mixture of pellets and water was prepared in an identical flask to serve as the duplicate. These two flasks were placed inside a large wooden box lined with polystyrene insulation to minimize heat loss or gain. Each flask acted as a calorimeter (Lau et al., 1992; Walsh et al., 1980). Three type K thermocouples were used to sense temperatures inside the calorimeter. One thermocouple was placed within the bulk pellets in each of the two flasks. The third thermocouple sensed the air temperature within the insulated box. Thermocouples were connected to three inputs of a four-channel data logger (Omega model HH374) with a readout resolution of 0.1°C.  Similar tests were performed on a mixture of 50 g of oven-dried western red cedar (WRC) wood chips, as described above. The wood chips were held in place on the tray by wire mesh to prevent them from floating in the water.  89   Figure 5.2 A calorimeter setup used to measure temperature rise in pellets-and-water mixtures. Thermocouples T1 and T3 measure the temperatures of pellets and water mixture in flasks 1 and 2. T2 measures the air temperature around the two flasks. The insulated box around the two flasks ensures that air temperature stays relatively constant.  The heat of wetting Q was calculated as:   Q = (mwCw + mpCp)(Tmax - Ti) / [mp(1 - (m.c.)/100)] (5.8) Where mw and mp are the mass of water and the mass of pellets or wood chips (kg). Cw and Cp are the specific heat of water (4.2 kJ/kg°C) and the specific heat of moist pellets, respectively. Tmax is the maximum temperature rise of water, Ti is the initial temperature (°C). The specific heat of moist pellets varied with moisture content and was calculated as:  Cp= 4.2(m.c.)/100 + 1.01(1 – (m.c.)/100)      (5.9) 90  where Cp is the specific heat of pellets in kJ/kg·°C and m.c. is the moisture content in percent wet basis (wb). The specific heat of dry biomass was assumed to be 1.01 kJ/kg·°C (Guo et al., 2013). I assumed the same value for the specific heat of the wood chips.  The heat of sorption hi at Mi can be estimated from the equation below. hi= (Qi+1  - Qi-1 Mi+1 - Mi-1 )Mi      (5.10) where i+1 and i-1 are two adjacent intervals along the moisture content axis.   5.3.4 Scanning electron microscopy (SEM) For SEM, the pellet measuring 5 mm in length and 6 mm in diameter and wood chip measuring 5×5×5 mm3 were mounted on aluminum stubs using hot-melt glue, applied from a hot glue gun, as an adhesive. Samples were coated with a 10-nm layer of gold under vacuum using a sputter coater. Samples were imaged at an accelerating voltage of 15 kV using a field emission scanning electron microscope (Model S4700, Hitachi, Japan). Secondary electron images of samples were obtained from the microscopes and saved as TIFF files. 5.4 Results  5.4.1 Water uptake Figure 5.3 shows the water uptake of pellets at 30 °C over time. The liquid water uptake data at 30, 40, and 50 °C can be found in Appendix Figure A.2. The wood pellets reached liquid water saturation at a moisture content of 1.83 (db) after four minutes. This finding indicates that all available water adsorption sites on wood pellets were saturated when the mass of water added to the pellets was twice the total dry mass of pellets. The saturation moisture content of 1.83 (db) for wood pellets is slightly higher than the values for the liquid water saturation moisture content reported in the literature. For example, Sandberg and Salin (Sandberg & Salin, 2012) reported 91  saturation moisture contents of 1.6-1.7 (db) for Norway spruce after immersing the wood in water for 14 days.  The high variability in moisture contents at each soaking time may have been caused by the varying amounts of water adhering to but not absorbed by the pellets when they were removed from the water before weighing. Also, the differing water-absorbing capability of individual pellets may have contributed to the variability.  Figure 5.3 The liquid water uptake of ~10 g dried wood pellets over six minutes. After six minutes, six to 10 g of water was absorbed. Six repetitions were made at each time interval. The solid curve represents the fitted Page model to the average data.  I fit the following Page equation to the moisture absorption data in Figure 5.3. MR = exp(-k·tn)     (5.11) where k and n are two constants. t is time. MR is a moisture ratio defined by:  MR = 1 - (M - Mf)/(Mi - Mf)     (5.12)  0.000.501.001.502.002.500 1 2 3 4 5 6Moisture content (db)Soaking time (minutes)92  where M is the moisture content measured at each time interval, Mf is the saturation moisture content, and Mi is the initial moisture content, which is taken as zero. The estimated k and n values were 0.2436 min-1 and 1.8160 (dimensionless), respectively. 5.4.2 Heat of wetting Figure 5.4 plots a sample of data for pellets immersed in water in the two side-by-side flasks. The air temperature around the flasks is plotted for reference. It took 8 to 15 minutes for the temperature of pellets and water mix to approach a plateau. Filling the second flask took place approximately 1 minute after filling flask number 1. This delay shifted the two plots apart from each other. Otherwise, the rise in temperature for the two plots is similar. The air temperature in the insulation box was constant at about 22.5 oC.   Figure 5.4 An example of temperatures of wood pellets and water mixtures in the two flasks and the air temperature near them. The initial moisture content was 0.05 (db).   Table 5.1 lists the moisture contents of each batch of pellets (100 g) immersed in water (200 g) in the two insulated flasks. Ti represents the temperature of the mixture of pellets and water immediately after the mixture was placed into the flasks. Tmax is the maximum temperature of the 222324252627280 5 10 15 20Temperature (°C)Soaking time (minutes)T1 (Flask 1)T3 (Flask 2)T2 (air)93  mixture. The initial temperatures varied from 21 to 23°C. The maximum temperatures exceeded 30°C for one of the samples at a moisture content of 0.005 (db).  I observed that the two flasks, which were side-by-side, did not have identical temperatures. Therefore, a degree of variability in the data can be expected. Such variability may have occurred because pellets were not agitated while immersed in water. This might have resulted in heat stratification in the flasks, leading to the variation in temperatures within each flask.  Table 5.1 Summary of pellets’ moisture content, initial and maximum temperatures of the pellets-and-water mixture, and the calculated heat of wetting.  Flask 1 (Figure 5.2) Flask 2 (Figure 5.2) Moisture content  Cp Ti Tmax Heat of wetting (Q) Ti Tmax Heat of wetting (Q) decimal (db) kJ/kg-°C °C °C kJ/kg dry pellet °C °C kJ/kg dry pellet 0.005 0.943 23.2 30.8 72.00 23.0 29.1 57.79 0.050 0.956 22.2 27.5 53.35 22.1 27.0 49.32 0.064 0.960 22.6 25.7 31.80 22.5 25.4 29.75 0.081 0.965 21.0 24.2 33.60 21.3 25.2 40.95 0.110 0.973 20.4 22.5 22.95 20.3 22.1 19.67 0.120 0.975 22.6 24.6 22.16 21.9 24.1 24.38 0.130 0.978 22.7 23.9 13.49 22.7 23.4 7.87 0.140 0.980 22.7 23.7 11.40 22.6 23.4 9.12 0.158 0.985 21.6 21.9 3.51 21.6 21.8 2.34  Figure 5.5 shows the average and variations in heat of wetting (Q) for moisture contents ranging from 0 to 0.158 (db). A linear equation fits the data better than an exponential equation (R2 = 0.80), possibly because the heat and pressure treatments during the densification and pelletization process reduced wood porosity and hygroscopicity (Fang et al., 2012). The heat of wetting of wood pellets at lower moisture contents may be higher than that predicted by Skaar‘s (1988) exponential equation because during the drying, particle size reduction, and densification process in wood pellet manufacturing, previously closed hydroxyl groups were opened up 94  (Rautkari et al., 2013). As a result, more bonds between water molecules and the wood surface formed and more heat was generated. Q = 108.6 exp (-17.6M), R2=0.80   (5.13)     Q = -403.1M + 66.4,  R2=0.95   (5.14) Skaar (1988) presented several estimates for Q0 and B values for the exponential equation. One of these equations had the following form and constants:  Q = 84 exp (-14M) (5.15) For M = 0, Q0 is 84 kJ/kg, which is somewhere between 66.5 and 108.2 kJ/kg in my work, as shown in Eqs. 5.13 and 5.14. In a recent publication, Nopens et al. (2019) used a solution calorimeter to measure the heat of wetting of beech and pine. Their values for maximum heat of wetting were 66-77 kJ/kg. I cannot conclude a specific value for the heat of wetting for wood pellets until further tests are conducted on a smaller sample while also accounting for any possible heat losses. However, when I use the exponential function similar to that used by Skaar (1988), the value of Q0 becomes 108.2 kJ/kg, which is much larger than the reported values for softwood species. Skaar (1988) obtained a higher value of 84 kJ/kg for Norway spruce (Picea abies). My wood pellets were made from a mixture of spruce, pine, and fir. My maximum heat of wetting is within ±5% of Nopens’ data and ±20% of Skaar’s data. For the moisture contents from 0.05 to 0.10 (db), my values overestimated Nopens’ data by 30-40%. The values from my calorimeter experiment may have some uncertainty due to the size of the unit, especially due to heat losses. The solid line in Figure 5.5 crosses the x-axis at M = 0.165; that moisture content may be considered the fiber saturation moisture content for wood pellets.  95   Figure 5.5 Heat of wetting (Q) for moisture contents ranging from 0.005 to 0.158 (db). The error bars represent the minimum and maximum values at each moisture content. A sloped line, Eq. 5.13, fitted through the data has a better R2 (=0.95) than an exponential curve, Eq. 5.14 (R2=0.80). At zero moisture content, the line approaches the vertical axis at Q0=66.4 kJ/kg or the maximum heat of wetting. The line approaches the x-axis at M=0.165; that may be considered the fiber saturation moisture content for the wood pellet.  I used Eq. 5.10 to calculate the differential heat of sorption as a function of moisture content. I divided the range of moisture contents from 0.00 to 0.12 (db) into 5 intervals. The values of Qi in the middle of each interval were calculated using Eq. 5.13 or 5.14. Table 5.2 lists the values of the differential heat for each interval. Table 5.2 also lists the differential heats of sorption from Simón et al. (2015) and Nopens et al. (2019). As expected, the linear equation gives a constant differential heat value of 403 kJ/kg of water since the differential heat of sorption at 403 kJ/kg was calculated based on Eq. 5.14 by differentiating with respect to moisture content. This value always remains constant. Values derived from the exponential equation gradually decrease with increased moisture content. Simón et al. (2015) derived their differential heat values using the isosteric method based on adsorption isotherm data of mature silver fir (Abies alba). Nopens et al. (2019) 0204060801001200.00 0.05 0.10 0.15 0.20Heat of wetting (kJ/kg pellets)Moisture content (db)96  derived their values from solution calorimetry tests on beech and pine. Their values are consistently lower than the differential heat calculated here using the immersion method.  Table 5.2 Differential heat of sorption, hs calculated from linear (Eq. 5.14) and exponential (Eq. 5.15) using Eq. 5.10. Mi and Mf are initial and final moisture contents, respectively.   hs (kJ/kg water) Mi (db) Mf (db) Linear Exponential Simon et al. (2015) Nopens et al. (2019) 0.00 0.03 403 1474 722 917 0.03 0.06 403 872 556 567 0.06 0.09 403 516 267 400 0.09 0.12 403 305 106 246 0.12 0.15 403 180 56 185  The constant differential heat of sorption, obtained from a linear equation of heat of wetting, implies that the bonding between water molecule-wood surface releases the same amount of energy regardless of the amount of water adsorbed on the surface of wood fibers. This suggests water molecules form a monolayer on the surface of wood fibers when they are adsorbed. An exponential equation relating differential heat of sorption to moisture content suggests that the bonding energy between water molecules and wood decreases as the amount of water adsorbed on the wood surface increased. This suggests water molecules form multilayer on the surface of wood fibers when they are adsorbed. Conventionally, the differential heat of sorption is assumed to be dependent on the initial moisture content because the water molecules form layers on the surface of wood; every additional layer reduces the thermal energy released when water molecules bond with wood fibers through hydrogen bonding (Hartley, Kamke, & Peemoeller, 1992). Based on the assumption of multilayer adsorption, the curves of sorption isotherms of woody materials are type II or type IV, not type I Langmuir in shape, since Langmuir sorption isotherm assumes monolayer adsorption of vapor molecules. The sorption isotherms for wood pellets are categorized as type II, 97  as shown in Chapter 3. Therefore, the differential heat of sorption for wood pellets should be an exponential relationship with moisture content. Figure 5.6 compares the maximum heat of wetting for western red cedar wood chips and wood pellets. The average maximum heats of wetting were 60.9 kJ/kg dry mass for wood chips and 66.8 kJ/kg dry mass for wood pellets. A t-test showed that the two sets of data were not significantly different with P-value = 0.40, which is much larger than the alpha value of 0.05. This showed that although wood pellets are primarily made from pine, maximum heat of wetting of western red cedar was statistically equivalent to wood pellets. However, the average heat release rate of wood pellets was nearly twice that of wood chips: 39.5 mJ/s per g dry mass for wood chips and 76.4 mJ/s per g dry mass.   Figure 5.6 The maximum heat of wetting of oven-dried western red cedar wood chips and wood pellets. The error bars represent the minimum and maximum values. The number of repetitions is 10 for each sample. A paired t-test (α=0.05) showed that the difference between maximum heats of wetting for wood chips and pellets was not statistically significant (P-value = 0.40).  SEM pictures in Figure 5.7 show cross-sections of a wood chip and a wood pellet. Wood pellets have a higher density and a lower air void space than western red cedar wood chips. 2030405060708090100Wood chips Wood pelletsMaximum heat of wetting, Q0(kJ/kg)98  Earlywood cell lumens are open and act as conduits for liquid water. Such flow paths are compressed and destroyed during the pelletization process, but there are much larger flow paths in pellets, as could be observed in x-ray CT images in Chapter 4. Hence, the liquid water absorption into pellets was faster than that is western red cedar wood chips. In addition, western red cedar has a higher extractive content than woods (spruce and pine) used to manufacture pellets (White, 1987). As a result, the heat release rate from pellets was faster than the heat release from wood chips. The higher heat release rate is evident in the temperature versus time plot in Figure 5.8. The temperature of the wood pellets increased much faster than that of wood chips.   Figure 5.7 (a) Horizontal cross-section of the as-received wood chip. Note the large open cell lumens in earlywood (central part of the image). Voids are much smaller in latewood in the left and right of earlywood. (b) the edge of the as-received wood pellet. The wood chip contains a porous cellular structure, whereas the structure of the wood pellet is not visibly porous except having cracks.  99   Figure 5.8 Temperature plots of oven-dried wood pellets and oven-dried wood chips after 50 g samples of each were immersed in water. The initial moisture content was 0.0 (db).  5.5 Discussion The heat of wetting occurs as a result of the interaction between solid and water adsorbed onto a material. When water vapor is adsorbed by wood pellets, both heat of condensation and differential heat of adsorption are released. When considering the heat of water vapor adsorption on the self-heating of wood pellets, Nilsson et al. (2019) assumed the differential heat of adsorption is negligible compared to the heat of condensation. The heat of condensation is ~2300 kJ/kg. The differential heat is 400 to 1400 kJ/kg for the moisture content of pellets from 0.03 to 0.09 (db). When the sum of heat of condensation and differential heat is considered as the total heat released from water vapor adsorption onto wood pellets, the differential heat can contribute 10 to 40% of the total heat. The processes of condensation and heating versus evaporation and cooling are cyclical. Within a pile of pellets, these cyclical processes will take place. The question is whether the heat of adsorption and condensation is larger than the heat loss from cooling. Bhat and Agarwal (1996) reported that temperature increases due to water condensation caused an increase in the rate of 2021222324250 5 10 15 20 25 30Temperature (°C)Time (minutes)Wood pellets Wood chips100  oxidation in coal particles. In the range of 20 to 40 °C, in the absence of microbial activities, the condensation of water initiates the self-heating of pellets (Nilsson et al., 2019). As the temperature increases to near 100 °C, the moisture evaporation and convection remove heat from the pellets, whereas oxidation and hydrolysis reactions, the dominant heat-releasing processes at this elevated temperature, release more heat into the pellets (Kubler, 1990). The imbalance between these two competing forces results in either heating to the ignition temperature or cooling to the ambient temperature. During the loading operation into ships for long-distance transportation, wood pellets may be exposed to rainwater and humid environments. These exposures to water will inevitably lead to water uptake and the subsequent release of the heat of wetting. It is speculated that the heat of sorption might be the impetus for self-heating in a pile of pellets, although Guo et al. (2020; 2014) associated self-heating with auto-oxidation reactions. On the other hand, Ryckeboer et al. (2003) associated the initiation of self-heating in composting biomass with microbial activity. But the microbial activity depends on available moisture in the biomass and a favorable water activity around 0.75-0.80, which corresponds to the equilibrium moisture content of 0.15 to 0.20 (db) for wood pellets (Lee et al., 2019). Most of the chemical molecules in a wood fiber have a propensity for water adsorption that generates heat. A wood pellet’s moisture content is usually less than 0.08 (db). It is doubtful the low moisture content is conducive to microbial growth. However, Lehtikangas (2000) found microbial growth for pellets samples stored for five months. The microbial growth was attributed to a possible re-contamination of pellets during handling processes or to the condensation of moisture on pellets’ surfaces.  101  From my results, I surmise that wood pellets release heat in a matter of minutes upon water exposure. If the existing pellets in a ship’s hold were covered with a new layer of pellets, a semi-adiabatic environment would be created within the bulk pellets because of their relatively low thermal conductivity (Guo et al., 2013). Together with a fast release of heat, localized hotspots around the wetted pellets could develop in the ship’s hold. The elevated temperatures may accelerate the oxidation of pellets and possibly pyrolysis reactions, creating off-gasses (Koppejan et al., 2013). These reactions could lead to further release of heat. This positive feedback in heating reactions may result in thermal runaway. If the pellets were suddenly exposed to ambient air oxygen, they might ignite, and a fire could ensue.  5.6 Conclusions The heat of wetting of wood pellets was measured in this chapter using calorimetry. The heat of wetting at zero moisture was 66.5 kJ/kg dry mass. This value is close to the value reported for untreated wood in the literature and is similar to that of western red cedar wood chips. However, the heat release rate for pellets was twice that of wood chips. This faster heat release rate may explain the stronger propensity of pellets to self-heat in humid environments. A linear equation relating the heat of wetting of pellets to adsorbed moisture fitted the data better than an exponential equation. Using a linear fitting curve, the heat of sorption vs. moisture content curve approaches the x-axis at a moisture content of 0.165 (db). I suggest that this moisture content is the fiber saturation moisture for pellets.  For moisture contents between zero to 0.15 (db), the differential heat of adsorption was estimated to be 180 to 1474 kJ/kg water using an exponential equation. Therefore, the differential heat is significant compared to the heat of condensation of water vapor.  102  Chapter 6: Modeling of Moisture and Heat Transfer in Single Pellets The modeling of moisture and heat transfer in a single pellet was done. The model consists of simultaneous heat and moisture transfer between the pellet and its environment. The kinetic model is developed using the equilibrium moisture content -relative humidity (EMC-RH) relation and the differential heat of sorption equation developed in previous sections of this thesis. The results from modeling were validated using experimental data. 6.1 Introduction Spontaneous heating or self-heating is defined as the phenomenon of a temperature rise in a material under ambient conditions, where the heating results from some chemical and physical processes occurring within the material (Miura, 2016). In the case of bulk coal, Nelson and Chen (2007) have summarized the circumstances where self-heating and spontaneous combustion may occur: the heat generation is greater than the heat loss from air ventilation. The authors also stated that in dry coal stockpiles, hot spots frequently occur after rain; very dry coals can ignite following water sorption.  For the case of woody biomass, Krigstin et al. (2018) reported that from 2000 to 2018, there were ~30 reported incidents of self-heating and fire in the storage of woody biomass, and the number of fire incidences was increasing steadily. This trend is a cause for concern, as self-heating can lead to spontaneous ignition and fires (Koppejan et al., 2013). Most recently, Guo et al. (2020) published data about the conditions when oxidation-induced thermal runaway can occur. Using experimental measurements and empirical models, the authors predicted that the critical ambient temperature range, which is the minimum ambient temperature that will lead to a thermal runaway during the pellet storage, was 30-80 °C for a container radius larger than 5 m. This temperature 103  range is large, considering that the typical ambient temperature in storage is usually no higher than 50°C. Figure 6.1 shows a schematic diagram of the possible moisture and heat transfer within a ship’s hold. The content of a ship’s hold, in this case, wood pellets, is protected by a water-tight seal from the outside environment. Given moisture and heat gradients within pellets in a ship’s hold, the natural convection is the primary driver of moisture and heat transfers. On sunny days, the exterior of the ship is heated. This heat evaporates some moisture from the wood pellets in the hold. During cooler days, the moisture in the air may condense on the cooler hatch and steel hulls. The heat of water condensation and heat of moisture adsorption are potential sources of heat for self-heating.  Figure 6.1 Schematic diagram of heat and natural convection of moisture within and around a ship’s hold full of wood pellets.  The objectives of this chapter are to examine the role of the heat of adsorption on the temperature rise of wood pellets and to provide a model to predict the thermal behavior of a single 104  pellet upon exposure to a stagnant humid atmosphere. The model is used to estimate the diffusion coefficient for moisture content data measured above a water bath and in a humidity chamber. 6.2 Methods 6.2.1 Materials Wood pellets used in these experiments were manufactured in British Columbia, Canada, using softwood species primarily pine but mixed with spruce and fir (SPF). They were purchased from a hardware store in 40 lb (18.2 kg) bags, as described in Chapter 3.  6.2.2 Experimental setup 6.2.2.1 Water vapor adsorption Wood pellets were first oven-dried at 105 °C in a convective oven for 24 hours. To measure their moisture adsorption behavior, 8-10 grams of oven-dried pellets were then exposed to a humid atmosphere with 95% relative humidity and 30, 40, and 50 °C in a humidity chamber (Model: LHU‐113; ESPEC North America, Inc., Hudsonville, Mich.) for 24 hours. The pellets were taken out from the humidity chamber to measure their weight on an analytical balance (±0.0001 accuracy) in 10-minutes intervals for the first hour, 20-minutes intervals for the next two hours, and one-hour intervals for the final seven hours. Total equilibration time was 10 hours. A hole was drilled in the same oven-dried pellet in its center axis with a 0.8 mm bit and a benchtop drill press. A K-type thermocouple was inserted into the drilled cavity and was held in place due to friction to measure the center-point temperature, which was assumed to be the average temperature of the pellet (Miura, 2016). The pellet was placed on top of a water-bath at 35 °C to be exposed to the water-saturated stagnant atmosphere. The temperature data was logged every 15 seconds using a thermometer and data logger (Model: HH374; OMEGA Engineering, Inc., Stamford, CT) connected to a computer (Dell Inspiron 15). 105   Previously published equilibrium moisture content (EMC) – relative humidity (RH) data (Lee et al. 2019) were used for this simulation. The Henderson equation was approximated with two linear relationships. The sigmoidal adsorption isotherm was linearized to two linear equations: one for RH from 0 to 0.80, another for RH from 0.80 to 1.00.  Figure 6.2 shows the setup used to measure the temperature of a single wood pellet on exposure to a stagnant atmosphere nearing water saturation. This experiment simulates a situation where wood pellets are exposed to an atmosphere resulting from high humidity due to rain events or condensation of moisture on the sides of a ship’s hold. The water bath (Fisher Scientific™ Isotemp™ Digital-Control Water Bath model 2300) was set to 35 °C to produce water vapor. The water vapor traveled into a 2 L glass jar through an opening on its lid and saturated the atmosphere within the jar. The final temperature within the jar was regulated to 32-33°C. An oven-dried single pellet was suspended in the jar, with a type K thermocouple drilled into its center axis. The pellet weighed about 1 g. The initial moisture content of the pellet was assumed to be zero. The change of temperature during the experiment was recorded using a data logger (Omega model HH374). A water bath was used instead of a humidity chamber because the high air flow rate in the humidity chamber causes a high cooling rate in a pellet. As a result, I was unable to detect the temperature increase due to water vapor adsorption. On the other hand, the rate of the water vapor adsorption above a saturated salt solution apparatus, where the relative humidity is lower than 1.00 (decimal), was too slow to counteract the heat loss to the atmosphere. In this case, small and slow temperature increases were difficult to observe. 106   Figure 6.2 The experimental setup used to measure the temperature change of a wood pellet on exposure to stagnant air.  6.2.3 Heat and moisture transfer model The key equations of relevance to the experiment are the moisture transfer (Eq. 6.1) and the heat transfer equations (Eq. 6.2). The following assumptions and numerical data are used for this simulation.  • Pellet is long enough so that it is an infinite cylinder. The diameter of a pellet, 2R is 0.0063 m. Length L of 0.012 m is used to calculate the surface area A and volume V of a pellet. • The pellet is placed in an infinite volume of stationary air of Tenv (K) and penv water vapor pressure (Pa).  • The temperature T, the water vapor pressure p, and the amount of water vapor adsorbed q per unit weight of the dried pellet are uniform spatially throughout the pellet’s cross-section.  • q and p are in equilibrium, and their relationship obeys a linear relationship of q = a·p/ps + b, where a and b are constants. The saturated vapor pressure ps (Pa) is described by the Antoine equation, as given in Table 6.2.  107  • Adsorbed water is in a lower state than ordinary liquid water, where the heat of adsorption is the heat of condensation plus the differential heat of adsorption. • The models are applicable to the pellets on the surface of a pile of bulk wood pellets in a ship’s hold. Figure 6.3 shows the assumed configuration of a single wood pellet for the chapter’s lumped capacitance model.   Figure 6.3 Assumed configuration of a wood pellet and the distribution of temperature T, partial pressure of water vapor p, and amount of water vapor adsorbed q within the wood pellet.  Wpelletd(q+ερbC)/dt = kcA(Cenv-C)  (6.1) where Wpellet is the weight of dried wood pellet, in kg, C and Cenv are the concentrations of water vapor in the wood pellet and in the ambient atmosphere, respectively, (kg H2O(g)/m3), ε is the void fraction of the wood pellet including the pore volumes within the wood particles, which made up a pellet, ρb is the apparent density of the wood pellet (kg/m3), t is the time (s), kc is the mass transfer 108  coefficient of water vapor in the gas film around the wood pellet (m/s), and A is the outer surface area of the spherical particle (m2). Wpelletddt(Hpellet+qHw(T)+ερbCaHa)=hA(Tenv-T)+kcA(Cenv-C)Hwv(Tenv) (6.2) where Ca is the total concentration of humid air in the wood pellet (kg air/m3), Hpellet is the enthalpy of the dried pellet per unit weight (kJ/kg pellet), Hw and Hwv are respectively the enthalpies of liquid water (a function of T) and water vapor (a function of Tenv) (kJ/kg water), and h is the heat transfer coefficient in the gas film around the wood pellet (kW/(m2 ·K)). To solve the two ordinary differential equations, Eqs. 6.1 and 6.2 are rearranged into Eqs. 6.3 and 6.4 below. The detailed derivation is given in Appendix B. The two equations will be solved simultaneously as functions of dp/dt and dT/dt. (∂q∂p+ερbRgT)dpdt+(∂q∂ps∙∂ps∂T-εpρbRgT2)dTdt=kcAWpelletRg (penvTenv-pT) (6.3) Hw (∂q∂p)dpdt+ {cpellet+Hw∂q∂ps∙dpsdT+qcw+εpTρbRgT(ca-HaT )}dTdt =AWpellet(h(Tenv-T)+kcRg(penvTenv-pT)Hwv) (6.4) The values and the defining equations for all the parameters are given in Tables 6.1 and 6.2. Equations used to calculate the heat capacities and enthalpy of humid air, ca and Ha are: ca=ppTcwv+(pT - p)pTcDry air  (6.5) Ha=ppTHwv+(pT - p)pTHDry air (6.6) MATLAB software (MathWorks, version R2020a) was used to solve Eqs. 6.3 and 6.4 with the parameters from Tables 6.1-6.2. The m-scripts are given in Appendix B.3.  109  Table 6.1 The variables as a function of the temperature of pellet T and temperature of environment Tenv and water vapor partial pressure p. Variables Symbols Units Equations Amount of water adsorbed or moisture content q kg water per kg dry mass For p/ps ≤ 0.8, q = 0.1789 (p/ps), For p/ps > 0.8, q = 0.4272 (p/ps) - 0.2025 Saturated vapor pressure ps Pa water vapor ln(ps) = 23.1964 – 3816.44 / (T – 46.13), where T is in Kelvin. Concentrations of water vapor in the wood pellet C kg H2O(g)/m3 C = p/(RgT) Concentrations of water vapor in the ambient atmosphere Cenv kg H2O(g)/m3 Cenv = penv/(RgTenv) Enthalpy of water vapor Hwv kJ/kg water Hwv = hs + 2500 + cwv (T-273) hs is a constant value or the heat of moisture adsorption given in Eq. 6.8. 2500 is the heat of condensation in kJ/kg at zero degree Celsius. Enthalpy of dry air HDry air kJ/kg dry air HDry air = ca (T – 273) Enthalpy of liquid water Hw kJ/kg water Hw = cw (T – 273) Moisture diffusion coefficient DH2O-Dry Air m2/s DH2O-Dry Air = 1E-10*exp(0.011*(T - 273)), T in Kelvin. The diffusion coefficient was taken from Rezaei et al. (2017) Heat diffusion coefficient for humid air λair (kW/m·K) λair = 1E-03*(9.0E-05*T), T in Kelvin.  The equation is for natural convection in stagnant air (Miura, 2016; Perry & Green, 1985). Moisture transfer coefficient kc m/s 2*DH2O-Dry Air /2R   Heat transfer coefficient h kW/(m2·K) 2*λair/2R  110  Table 6.2 Parameters for the lumped capacitance heat and moisture transfer model (Eqs. 6.3 and 6.4). Parameters Symbols Units Values Internal void fraction of a pellet 𝜀 dimensionless 0.3 Apparent density of a pellet ρb kg/m3 -14670*q2 + 2068*q + 1000, where q in kg water / kg dry mass, taken from Chapter 4. Gas constant Rg kJ/(kg water· K) 0.4619 Surface area of a pellet A m2 πDL = π(0.0063)(0.012) = 0.0002375 Volume V m3 π (D2/4) L = π (0.00632/4)(0.12) = 3.7407E-07 Weight of an oven-dried pellet Wpellet kg V* ρb  Specific heat of a pellet cpellet kJ/(kg·K) 1.2 Specific heat of liquid water cw kJ/(kg·K) 4.186 Specific heat of water vapor cwv kJ/(kg·K) 2.0 Specific heat of dry air cDry air kJ/(kg·K) 1.0 Atmospheric pressure pT Pa 101,325 Air temperature in the environment Tenv °C K 33 306.15 Air pressure in the environment penv Pa rhenv*ps  111  6.3 Results  Table 6.3 The values of moisture diffusion coefficient, DH2O-Dry Air and heat diffusion coefficient, λair in a few cases. Coefficients Symbols Cases Values Moisture diffusion coefficient DH2O-Dry Air In a humidity chamber at RH of 0.95 and T of 30 °C 3.2E-09 m2/s Above a water bath at T of 30 °C  7.0E-10 m2/s Rezaei et al. (2017) 0.6E-10 – 7.6E-10 m2/s Heat diffusion coefficient λair All cases 2.7E-05 kW/(m·K)  Figure 6.4 shows the experimentally measured moisture content data over time in a 95% relative humidity. At higher temperatures, the rate of moisture adsorption was slightly higher. Igathinathane et al. (2009) reported a similar temperature effect for corn stover. The moisture content of pellets increased quickly from zero to 0.13 (db) in the first five hours. Then, the rate of increase decreased as the moisture content approached the equilibrium moisture content of 0.21 to 0.23 (db).   The modeled curve with kc equals to 1.0E-06 shows a slight drop in moisture content just after 6 hours because the model switched from one linear equation for the equilibrium moisture content to another as the RH rose above 0.80. 112   Figure 6.4 Moisture adsorption rate of water vapor for dried softwood pellets measured on exposure to a humid atmosphere at a relative humidity of 0.95 (decimal) in a humidity chamber (HC) at temperatures of 30, 40, and 50 °C and in a water bath (WB) at 33 °C. The curves are calculated using the heat and moisture transfer model (Eqs. 6.3 and 6.4) at a relative humidity of 0.95 (decimal) and at temperatures of 30 °C and two different moisture transfer coefficients, kc.  For humidity chamber, the moisture transfer coefficient, kc is 4.5 times larger than that of the case of water bath (Figure 6.4). These values are within the ranges of the diffusion coefficient values estimated by Rezaei et al. (2017). In a stagnant atmosphere (above the water bath), the moisture content increases at a much lower rate than that for a humidity chamber forced circulated environment. Correspondingly, the diffusion coefficient is 3.2E-09 m2/s for the case of the humidity chamber, which is higher than the case of the water bath, which has a diffusion coefficient of 7.0E-10 m2/s (Table 6.3).  The higher diffusion coefficient is caused by the higher air velocity in humidity chamber. 0.000.020.040.060.080.100.120.140.160.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00Moisture content (db)Exposure time (hours)Data - WB Data - HC (30 °C) Data - HC (40 °C) Data - HC (50 °C)kc = 2.2E-07 m/skc = 1.0E-06 m/s113  The Page equation, which is an exponential equation with an exponent on the time variable, t, was used to describe the relationship between moisture content and time for biomass drying and wetting systems (Igathinathane et al., 2009). The Page equation was not shown in Figure 6.4, but it closely follows the measured data points for the humidity chamber (HC). M = Mf[1− exp (-k∙t0.7233)]        (6.7) Where M is the moisture content in decimal dry basis (db). Mf is the equilibrium moisture content, which is represented by the adsorption isotherm. t is time in hours. k = 0.2668·exp (2.0129T+273.15) is the rate constant. T is temperature in °C. The correlation coefficient, R2 is 0.99.  Figure 6.5 Adsorption isotherm of water vapor on softwood pellets measured at 25 °C (points), adapted from Lee et al. (2019). The three curves are Henderson model and the two approximated linear relationships.   As shown in Figure 6.5, the sigmoidal adsorption isotherm was linearized to two linear equations to simplify the process of solving the two ordinary Eqs. 6.3 and 6.4. The first equation 0.000.050.100.150.200.250.300.0 0.2 0.4 0.6 0.8 1.0EMC (db)ERH = p/psData Henderson model (Lee et al., 2019) Linear 1 Linear 2y = 0.1789xR² = 0.9751y = 0.4272x - 0.2025R² = 0.9787Measured at 25°C114  fits data from zero to 0.80 relative humidity. The second equation fits data from 0.80 to 1.00 relative humidity. The Henderson equation published in Lee et al. (2019) was compared favorably to the two linear equations.  The equation below gives the differential heat of sorption of wood pellets, hs in kJ/kg water as a function of moisture content, M in decimal dry basis (db). Two differential heat of sorption equations were tested: Eq. 6.8, which was derived from Eq. 5.13, or was assumed to be a constant 403 kJ/kg water (Eq. 5.14). hs = 1917*exp(-17.51M)        (6.8)  This exponential equation implies that the differential heat released from moisture adsorption is large at zero moisture with a value of 1917 kJ/kg water, compared to the heat of condensation of water (2400 kJ/kg water near room temperatures). The differential heat rapidly drops in magnitude with an increase in moisture content. At a moisture content of 0.10 (db), the differential heat is 333 kJ/kg water. At a moisture content of 0.20 (db), the differential heat is 58 kJ/kg water, which is negligible compared to the heat of condensation.  The differential heat of sorption, hs is used to calculate the enthalpy of water vapor, Hwv in Eq. 6.4. The enthalpy of water vapor is defined in Eq. 6.9 below. Hwv = hs + 2500 + cwv (T-273)         (6.9) where 2500, in kJ/kg, is the latent heat of condensation at zero degrees Celsius. My modeling calculations were performed with the physical properties and the rate parameters given in Table 6.1. The internal porosity of a pellet ε is 0.30. Since the density of a single wood pellet changes with moisture content, a quadratic equation was used to estimate the density of a pellet as given in Chapter 4. As the oven-dried pellet adsorbs water, its mass increases faster than the increase in volume, therefore, the density of the pellet first increases. Above a 115  moisture content of ~0.10 (db), the volume increases faster than the increase in mass, therefore the pellet density decreases. Compared to the kc equation in Miura (2016), I found that the kc equation must be modified using the diffusion coefficient correlation equation of Rezaei et al. (2017) for ground pellet particles to fit the data. The diffusion coefficient, DH2O-Dry Air calculated using the correlation equation of Rezaei et al. (2017) is smaller than the diffusion coefficient used by Miura (2016). The moisture transfer in wood pellets is therefore much slower compared to coal. One possible reason for the slower moisture transfer in wood pellets compared to coal is that woody material diffuses moisture slower than more porous carbonized material such as charcoal and coal (Assis et al., 2016). Miura (2016) also used a smaller amount of samples (0.1 to 0.4 g) than the amount of pellet samples (1 g) used in this thesis. To better fit the measured data, the kc value was further adjusted by a multiply of 5. This adjustment may mean that the diffusion coefficient equation obtained by Rezaei et al. (2017) by drying ground pellets differed slightly from the diffusion coefficient when single whole pellets were adsorbing moisture from a humid atmosphere. The rate of moisture adsorption is usually higher than the rate of moisture desorption or drying due to hysteresis (Time, 1998). Therefore, the moisture diffusion coefficient for the case of moisture adsorption is higher than that of moisture desorption. 116   Figure 6.6 Comparison of experimentally measured and model-predicted temperature changes at 33°C and 95% RH for wood pellets. The dots are experimental data. The solid line is the predicted temperatures when hs = 403 kJ/kg (R2 = 0.9128). The dashed line is for hs = Eq. 6.8 (R2 = 0.9175).   Figure 6.6 compares the experimentally measured temperature data over time in a water-saturated atmosphere at 33 °C on top of a water bath over 30 minutes and model-predicted temperatures using two different values for differential heat of sorption (constant value of 403 kJ/kg and an exponential equation, Eq. 6.8). When a constant value for differential heat of sorption was used, the predicted temperature matched the rate of increase of measured temperatures from zero to 5 minutes. Although it was ahead of the increase in the measured temperature initially, the model using Eq. 6.8 adequately fits the measured temperature data (R2 = 0.9175). Although both model curves have similar R-squared values, the model assuming an exponential equation for 117  differential heat biased towards higher temperature values, whereas the model assuming a constant value for differential heat predicted values on both sides of the measured data. Compared to the initial temperature of the single pellet at 25 °C, the model using a constant value shows a temperature rise of 10 °C, whereas the model using Eq. 6.8 shows a temperature rise of 11.5 °C. The measured temperature increased by 11.3 °C. This increase in temperature is caused by convective heat transfer from a hotter atmosphere, and the water vapor adsorption onto the pellet. 6.4 Discussion  Between the years 2000 to 2017, ~30 cases of self-heating and fire in woody biomass storage were reported in the media, as mentioned above (Krigstin et al., 2018). A report on a fire on a wood-pellet-carrying bulk carrier in 2016 (Transport Malta, 2017) has also added to the urgency of tackling the self-heating phenomenon in storage and in ship’s holds. The investigation of the fire on the Maltese registered bulk carrier has listed the following potential factors: moisture content, the temperature of the biomass, length of time in storage, condensation heat, and age of the biomass. In addition to the condensation heat, the heat of moisture adsorption adds to the total amount of heat released when wood pellets adsorb water vapor from the ambient atmosphere. Since port terminals are situated near bodies of water, the ambient atmosphere at a loading or unloading port often has high humidity of over 80%. Relatively dry wood pellets in such a high humidity atmosphere can rapidly adsorb a large amount of moisture. I observed that the temperature of a wood pellet increased by 10 to 11 °C when the pellet was exposed to humid air at an ambient temperature of 33 °C. A maximum temperature of 36 °C was reached after 1800 seconds or 30 minutes, as shown in Figure 6.7. Both the heat of condensation and heat of wetting are involved in the temperature. On the other hand, the 118  temperature increased by 5 to 7 °C when bulk amounts (100 g) of pellets were immersed in liquid water, as I demonstrated in the previous chapter. The maximum temperature of 29-31 °C was reached after 5 minutes. Wood pellets release heat faster in liquid water. Only heat of wetting is involved in this temperature rise due to exposure to liquid water.  Since both heat of condensation and heat of wetting are released when water vapor is adsorbed and only heat of wetting is released when liquid water is absorbed, the amount of heat release per kilogram of water vapor adsorbed is larger than the heat release per kilogram of liquid water absorbed. However, compared to the case with water vapor exposure, the magnitude of temperature rise in the case of liquid water exposure is larger. This higher temperature rise was due to the larger amount of pellets exposed to water. At room conditions, liquid water has a higher heat capacity (4.186 kJ/kg.°C) compared to ambient air (1.004 kJ/kg.°C). The higher heat capacity of liquid water meant that more heat is needed to increase one centigrade of liquid water than ambient air. However, liquid water also has a higher thermal conductivity (598 mW/m.K) than ambient air (26 mW/m.K). This meant almost all the heat of wetting was adsorbed by liquid water. For the case of water vapor, due to the low thermal conductivity of air, minimal heat is transferred to the ambient air. As a result, the heat of wetting and the heat of condensation transferred to the pellet and caused substantial temperature increase, even though only one pellet was exposed to water vapor. For grain stored in large bins, there are seasonal changes in moisture content in the stored grain due to natural convection and ambient temperature variation (Jian, Jayas, & White, 2009). Similar to stored grains, when a large quantity of wood pellets was stored in a confined space, such as a ship’s hold, exposure to water during loading will result in a temporary wet and hot zone. This wet zone migrates due to natural convection and temperature variation of the seawater and the 119  outside atmosphere. The hatch of a ship’s hold is usually sealed to prevent exposure to rainwater. However, wear and tear on seals and the insulation of the hatch may leak air into a ship’s hold. As a result, during the voyage to the destination port, the moisture content of wood pellets towards the top of the ship’s hold changes, and moisture migrates to the rest of the bulk. There are observations that wood pellets on the top of a hold are wet and swollen due to the formation of condensed water (Melin et al., 2008).  Modeling to predict the temperature response of a single wood pellet when it is exposed to a high humidity environment shows a potential contribution of moisture adsorption to self-heating of pellets. A number of researchers (Koppejan et al., 2013; Miura, 2016; Nilsson et al., 2019) have suggested the initial contribution of moisture adsorption to self-heating and spontaneous ignition of wood pellets and coal. Spontaneous ignition can only occur at pellet temperatures above 200°C. A critical temperature of between 80 to 100 °C must be archived to cause a chain of positive feedback reactions, a thermal runaway, to occur to increase the temperature of pellets above their ignition temperature. Guo et al. (2020) showed that the amount of wood pellets, which can occupy a bin with a radius larger than 5 m, provided good heat insulation, and heat sources other than auto-oxidation are required to create a thermal runaway. Guo et al. (2020) suggest that the heat of moisture adsorption contributes to the initial temperature increase requested to hit the critical temperature for thermal runaway. A typical pellet-carrying bulk carrier has holds, each with a volume over 10,000 m3 and a typical dimension of 28 m × 30 m × 17 m. The dimensions and volume of a ship’s hold are much larger than the minimum volume of wood pellets to sustain a thermal runaway. Therefore, a ship’s hold provides a conducive environment for wood pellets to reach their critical temperature for thermal runaway. 120  6.5 Conclusions With the inclusion of differential heat of moisture sorption and heat of water condensation, the temperature results from modeling using two ordinary differential equations of heat and moisture balance fit measured temperature data adequately. The inclusion of differential heat of moisture adsorption allowed for more accurate modeling of the heat and moisture transfer phenomenon in a humid atmosphere.  Compared to the initial temperature of 26 °C, the temperature rise of 10.3 °C in measured data is mirrored by the temperature rise of 10.5 °C in the modeling results. This temperature increase is caused by convective heat transfer and water vapor adsorption. Simplified ordinary differential equations are therefore adequate to predict the temperature behavior of pellets when the heat of adsorption is released as a result of moisture adsorption.  The heat released from the condensation and adsorption of water vapor onto a single pellet was enough to increase the pellet’s temperature above the ambient temperature of 33 °C. In a bulk volume of wood pellets, such as in a silo or a ship’s hold, the heat released from water vapor adsorption must be larger in magnitude and may be enough to elevate the pellets’ temperature above the critical thermal runaway temperature. In other words, the water vapor adsorption contributes to the initiation of self-heating and self-ignition phenomena in wood pellets.      121  Chapter 7: Conclusions and Recommendations 7.1 Overall Conclusions 7.1.1 Experimental conclusions The heat of wetting of whole wood pellets has been measured using the calorimetry method for the first time. This approach ensures that heat of wetting values more accurately represents the amount of heat released when pellets are exposed to liquid water. The EMC of wood chips was compared to that of wood pellets because the former is more commonly used in combined heat and power systems in North America than wood pellets. The equilibrium moisture content (EMC) of wood pellets was determined using the saturated salt method. When compared with literature data, the wood pellets had a slightly lower EMC than wood chips by 0.01-0.05 (db). When immersed in liquid water, the temperature of wood pellets increased by 5 to 7 °C in 5 minutes, whereas the temperature of wood chips increased by 2 to 3 °C over 30 minutes. An explanation for these differences is provided in Chapter 5. As moisture content increases, exposure to liquid water causes a faster decrease in durability than exposure to water vapor. The durability at 13% moisture content was ~95% for pellets exposed to liquid water and was ~97% for pellets exposed to water vapor in the humidity chamber at 95% RH (if the single pellet durability was converted to tumbler durability using the equation in Figure A.3).   From the investigation of the effect of water vapor adsorption on durability (Chapter 4), I found that single pellet durability decreased very gradually from almost 100% to 95% when moisture content increased from 0 to 0.12 (db). When moisture content increased above 0.15 (db), durability dropped abruptly from 90% to 0% at 0.19 (db) moisture content. This pattern of decrease in single pellet durability was strongly negatively correlated with volumetric swelling of single 122  pellets with a R2 value of 0.956. The high variability in single pellet durability was linked to the high variability in volumetric swelling and pellet density, as well as the heterogeneity in the durability of single pellets. 7.1.2 Modeling conclusions The Henderson equation, instead of a linear or polynomial equation, was used to describe the moisture adsorption isotherm of wood pellets at 25 °C (Basu et al., 2006; Hartley & Wood, 2008). Although the Henderson equation is commonly used to describe sorption isotherms of food products and wood substrates, the work here is the first time that a sigmoidal equation has been used to describe a sorption isotherm of wood pellets. Comparing the sigmoidal curves with similar equations published for solid wood, the pelletization and densification reduced the equilibrium moisture contents of wood by up to 0.05 (db) for relative humidity above 0.70 (decimal). A novel empirical equation was used to describe the relationship between single pellet durability and moisture content when the pellets were exposed to a humid atmosphere in humidity chamber. The equation successfully depicted the initial slow decrease in single pellet durability and the following abrupt drop at moisture contents above 0.15 (db).   The inclusion of heat of moisture adsorption as a source of heat, which contributes to the self-heating phenomenon, is a novel contribution in the field of self-heating of wood pellets. The moisture adsorption contributes to temperature rise and the self-heating of wood pellets. A lumped capacitance heat and moisture transfer model was used to predict the temperature response to moisture adsorption with a correlation coefficient of 0.97. This lumped model could be used in future work to show the combined contribution of auto-oxidation and moisture adsorption to self-heating. However, this model does not take into account the moisture transfer between two or more 123  pellets. In storage and transportation, pellets will transfer moisture among themselves and with the atmosphere.  7.1.3 Practical implications Loading a bulk carrier with wood pellets is a multi-day operation. The current practice is to stop loading when rain is forecasted to avoid damaging the cargo. However, this procedure causes long and costly delays during the autumn and winter seasons on the West Coast of Canada. Based on my experiments in Chapter 2, a curve was developed to specify the rain intensity threshold given a range of exposure durations to the minimum acceptable durability of 96.5%. The latter durability value is the minimum durability required by the Industrial Wood Pellet Buyers (IWPB) trade group. According to this curve, during a rain event with an intensity of 1 mm/hr, the loading could continue for up to 20 minutes before the pellets’ durability drops below 96.5%. The cut-off curves allow the loading operators to decide when to stop loading wood pellets. This additional information will reduce delays when bulk carriers are waiting to be loaded with pellets. A system to predict the durability of pellets from an instantaneous record of exposure of pellets to rainfall should then be implemented at the port.  Furthermore, the relationships established between physical properties, such as density and durability, and moisture content in Chapter 2 allows the prediction of the properties of wood pellets after they are exposed to liquid water and water vapor. My research shows that compared to exposure to water vapor in humid air, the durability of wood pellets decreases more rapidly with the increase of moisture content when they are exposed to liquid water. Therefore, practices to reduce condensation and wetting of wood pellets in storage, such as dehumidifying the air in the storage and loading pellets under cover, will help to reduce exposure to liquid water.    124  7.2 Recommendations for Future Work A study on the effects of rainfall on wood pellets should be conducted by measuring and simulating the various metrological variables for weather conditions, such as air salinity, wind speed, temperatures, and humidity. The durability and density of wood pellets should be tested after they are exposed to a range of different rain patterns in each of the four seasons. Weekly, daily and hourly weather forecasting should be used to better predict the weather and to plan the course of action. In my study of the effects of liquid water exposure to durability and density of wood pellets, pellets were oven-dried prior to spraying them with water. In future work, pellets with different initial moisture contents should be exposed to water to investigate the effect of the initial moisture content of pellets on their final durability and density. The durability and density of pellets stay almost constant at moisture content from 0 to 10% (wb). Therefore, this range of initial moisture content can be used in future work. To find out the reasons behind the varied swelling responses of individual pellets, pellets at a range of moisture content can be examined using X-ray micro-computed tomography. Pellets made from a narrow particle size distribution can be exposed to water vapor to investigate the relationship between particle size and the swelling behavior of pellets. The single-pellet lumped capacitance heat and moisture transfer model demonstrated the contribution of moisture adsorption to the self-heating of wood pellets. In practice, the change of moisture content of a large quantity of pellets, in the ballpark of over 5000 tonnes, will be of interest to the suppliers and shippers of industrial and residential wood pellets. Although the lumped model predicts single pellet behaviors well, to better predict the temperature and moisture responses of a bulk amount of pellets, a distributed model could more accurately describe their 125  internal temperature distribution. Therefore, a temperature and moisture gradient within a ship hold should be modeled using 1-D or 2-D partial differential equations in future work. A larger-scale experiment should be conducted to show the contribution of the moisture adsorption process to the self-heating phenomenon in bulk wood pellets. Larsson et al. (2020) conducted a medium-scale self-heating experiment with biomass pellets to look at the temperature rise and off-gassing emissions. They heated the cylindrical test container using six heating mats and controlled the airflow within the container using a fan. Future experiments on self-heating are recommended to be based on the setup by Larsson et al. (2020).   126  References Acharjee, T. C., Coronella, C. J., & Vasquez, V. R. (2011). Effect of thermal pretreatment on equilibrium moisture content of lignocellulosic biomass. Bioresource Technology, 102(7), 4849–4854. https://doi.org/10.1016/j.biortech.2011.01.018 Adhikary, K. B., Pang, S., & Staiger, M. P. (2008). Long-term moisture absorption and thickness swelling behaviour of recycled thermoplastics reinforced with Pinus radiata sawdust. Chemical Engineering Journal, 142(2), 190–198. https://doi.org/10.1016/j.cej.2007.11.024 Amer, A. M. (2015). Chapter 1 Water vapor adsorption and soil wetting. In M. Aliofkhazraei (Ed.), Wetting and Wettability (pp. 1–14). InTech. Retrieved from http://dx.doi.org/10.5772/60953 Amyotte, P. R., & Eckhoff, R. K. (2010). Dust explosion causation, prevention and mitigation: An overview. Journal of Chemical Health and Safety, 17(1), 15–28. https://doi.org/10.1016/j.jchas.2009.05.002 Andrade P., R. D., Lemus M., R., & Pérez C., C. E. (2011). Models of sorption isotherms for food: Uses and limitations. Vitae, 18(3), 325–334. ASABE. (2012). ASAE D245.6 Moisture Relationships of Plant-based Agricultural Products (pp. 1–40). ASABE. (2015). Standard S269.5. Cubes, Pellets and Crumbles—Definitions and Methods for Determining Density, Durability and Moisture Content. St. Joseph, MI, USA: American Society of Agricultural and Biological Engineers. Assis, M. R., Brancheriau, L., Napoli, A., & Trugilho, P. F. (2016). Factors affecting the mechanics of carbonized wood: Literature review. Wood Science and Technology, 50(3), 519–536. 127  https://doi.org/10.1007/s00226-016-0812-6 Athanasatos, S., Michaelides, S., & Papadakis, M. (2014). Identification of weather trends for use as a component of risk management for port operations. Natural Hazards, 72(1), 41–61. https://doi.org/10.1007/s11069-012-0491-z Avramidis, S. (1989). Evaluation of three-variable models for the prediction of equilibrium moisture content in wood. Wood Science and Technology, 23(3), 251–257. https://doi.org/10.1007/BF00367738 Baar, J., Paschová, Z., Čermák, P., & Wimmer, R. (2019). Color changes of various wood species in response to moisture. Wood and Fiber Science, 51(2), 119–131. https://doi.org/10.22382/wfs-2019-014 Babbitt, J. D. (1942). On the adsorption of water vapour by cellulose. Canadian Journal of Research, 20a(9), 143–172. https://doi.org/10.1139/cjr42a-012 Bacchioni, A. (2008). Best Practices in Port Management: An Assessment of Two Ports (M.Sc. Thesis). Duke University, North Carolina, USA. Barnes, A. (2019). 2018 Economic State of the B.C. Forest Sector (pp. 1–86). Ministry of Forests, Lands, Natural Resource Operations and Rural Development. Retrieved from Ministry of Forests, Lands, Natural Resource Operations and Rural Development website: https://www2.gov.bc.ca/assets/gov/farming-natural-resources-and-industry/forestry/forest-industry-economics/economic-state/2018_economic_state_of_bc_forest_sector-with_appendix.pdf Basu, S., Shivhare, U. S., & Mujumdar, A. S. (2006). Models for sorption isotherms for foods: A review. Drying Technology, 24(8), 917–930. https://doi.org/10.1080/07373930600775979 Bennamoun, L., Harun, N. Y., & Afzal, M. T. (2018). Effect of storage conditions on moisture 128  sorption of mixed biomass pellets. Arabian Journal for Science and Engineering, 43, 1195–1203. https://doi.org/10.1007/s13369-017-2808-4 Bhat, S., & Agarwal, P. K. (1996). The effect of moisture condensation on the spontaneous combustibility of coal. Fuel, 75(13), 1523–1532. https://doi.org/10.1016/0016-2361(96)00121-4 BIMCO. (2014). How ports work – bulk terminals. Retrieved from https://www.bimco.org/Education/Seascapes/Maritime_Matters/2014_07_02_Bulk_terminals.aspx. Bioenergy Europe. (2018). Statistical Report, 2018 Edition. Retrieved from https://bioenergyeurope.org/statistical-report.html Bogolitsyn, K., Volkova, N., & Wadso, I. (1995). A microcalorimetric technique for the study of water vapour sorption on cellulose materials. In Cellulose and Cellulose Derivatives: Cellucon’93 Proceedings: Physico-Chemical Aspects and Industrial Applications (pp. 101–106). Cambridge, UK: Woodhead Publishing. Boss Tek. (2013). Case Study: Port Controls Wood Pellet Dust without Oversaturating Cargo. Retrieved from https://bosstek.com/high-powered-suppression-system-clobbers-dust-during-ship-loading/ Browning, B. L. (1963). The Chemistry of Wood. New York: Interscience (Wiley). Canadian Canola Growers Association. (2015). Infographic: Agriculture can grow our economy but rail service must respond to market demand. Retrieved from http://albertacanola.com/wp-content/uploads/2015/06/Rail-Service-Must-Meet-Demand-July-2015.pdf Cao, J., & Kamdem, D. P. (2004). Moisture adsorption thermodynamics of wood from fractal-129  geometry approach. Holzforschung, 58(3), 274–279. https://doi.org/0.1515/HF.2004.042 Carll, C., & Wiedenhoeft, A. C. (2009). Moisture-related properties of wood and the effects of moisture on wood and wood products. In Moisture Control in Buildings: The Key Factor in Mold Prevention (2nd ed., pp. 54–79). West Conshohocken, PA: ASTM International. Chen, J., Wang, Y., Cao, J., & Wang, W. (2020). Improved water repellency and dimensional stability of wood via impregnation with an epoxidized linseed oil and carnauba wax complex emulsion. Forests, 11(3), 271–284. https://doi.org/10.3390/f11030271 Cocchi, M., Nikolaisen, L., Junginger, M., Goh, C. S., Heinimö, J., Bradley, D., … Thrän, D. (2011). Global wood pellet industry market and trade study. 40, 190. Retrieved from https://www.ieabioenergy.com/wp-content/uploads/2017/11/Two-page-summary-%E2%80%93-Global-Wood-Pellet-Industry-and-Trade-Study-2017.pdf Colley, Z., Fasina, O. O., Bransby, D., & Lee, Y. Y. (2006). Moisture effect on the physical characteristics of switchgrass pellets. Transactions of the ASABE, 49(6), 1845–1851. https://doi.org/10.13031/2013.22271 Craven, J. M., Swithenbank, J., Sharifi, V. N., Peralta-Solorio, D., Kelsall, G., & Sage, P. (2015). Hydrophobic coatings for moisture stable wood pellets. Biomass and Bioenergy, 80, 278–285. https://doi.org/10.1016/j.biombioe.2015.06.004 Curci, M. (2011, April 1). Bulk Pellet Behaviour. Canadian Biomass Magazine. Retrieved from http://www.canadianbiomassmagazine.ca/news/bulk-pellet-behaviour-2399 Deng, T., Alzahrani, A. M., & Bradley, M. S. (2019). Influences of environmental humidity on physical properties and attrition of wood pellets. Fuel Processing Technology, 185, 126–138. https://doi.org/10.1016/j.fuproc.2018.12.010 Droin-Josserand, A., Taverdet, J. L., & Vergnaud, J. M. (1988). Modelling the absorption and 130  desorption of moisture by wood in an atmosphere of constant and programmed relative humidity. Wood Science and Technology, 22(4), 299–310. https://doi.org/10.1007/BF00353320 Enarson, D. A., & Chan-Yeung, M. (1990). Characterization of health effects of wood dust exposures. American Journal of Industrial Medicine, 17(1), 33–38. https://doi.org/10.1002/ajim.4700170107 Engelund, E. T., Thygesen, L. G., Svensson, S., & Hill, C. A. (2013). A critical discussion of the physics of wood–water interactions. Wood Science and Technology, 47(1), 141–161. https://doi.org/10.1007/s00226-012-0514-7 Environment Canada. (2001). Monthly Data Report for 2000 – Prince Rupert. Environment Canada. (2011). Canadian Climate Normals 1981-2010 Station Data – Vancouver International Airport. Environment Canada. (2015). Climate: Glossary. Retrieved from https://climate.weather.gc.ca/glossary_e.html#r European Pellet Council. (2015). Quality Certification Scheme For Wood Pellets—Part 3: Pellet Quality Requirements (Vol. 2). Retrieved from https://enplus-pellets.eu/en-in/component/attachments/?task=download&id=103 Fang, C., Mariotti, N., Cloutier, A., Koubaa, A., & Blanchet, P. (2012). Densification of wood veneers by compression combined with heat and steam. European Journal of Wood and Wood Products, 70, 155–163. Fasina, O. O., & Sokhansanj, S. (1992). Hygroscopic moisture absorption by alfalfa cubes and pellets. Trans.ASABE, 35(5), 1615–1619. Fasina, O. O., & Sokhansanj, S. (1993). Equilibrium moisture relations and heat of sorption of 131  alfalfa pellets. J. Agric. Engng. Res., 56, 51–63. Fibreco Export Inc. (2016). Personal Communication. Filbakk, T., Jirjis, R., Nurmi, J., & Høibø, O. (2011). The effect of bark content on quality parameters of Scots pine (Pinus sylvestris L.) pellets. Biomass and Bioenergy, 35(8), 3342–3349. https://doi.org/10.1016/j.biombioe.2010.09.011 García, R., Gil, M. V., Rubiera, F., & Pevida, C. (2019). Pelletization of wood and alternative residual biomass blends for producing industrial quality pellets. Fuel, 251, 739–753. https://doi.org/10.1016/j.fuel.2019.03.141 Gastón, A., Abalone, R., Bartosik, R. E., & Rodríguez, J. C. (2009). Mathematical modelling of heat and moisture transfer of wheat stored in plastic bags (silobags). Biosystems Engineering, 104(1), 72–85. https://doi.org/10.1016/j.biosystemseng.2009.06.012 Graham, S., Eastwick, C., Snape, C., & Quick, W. (2017). Mechanical degradation of biomass wood pellets during long term stockpile storage. Fuel Processing Technology, 160, 143–151. https://doi.org/10.1016/j.fuproc.2017.02.017 Graham, S., Ogunfayo, I., Hall, M. R., Snape, C., Quick, W., Weatherstone, S., & Eastwick, C. (2016). Changes in mechanical properties of wood pellets during artificial degradation in a laboratory environment. Fuel Processing Technology, 148, 395–402. https://doi.org/10.1016/j.fuproc.2016.03.020 Guo, W. (2013). Self-heating and spontaneous combustion of wood pellets during storage. University of British Columbia. Guo, W., Bi, X., Lim, C. J., & Sokhansanj, S. (2020). Investigation of wood pellets self-heating kinetics and thermal runaway phenomena using lab scale experiments. The Canadian Journal of Chemical Engineering, 98(1), 127–137. https://doi.org/10.1002/cjce.23592 132  Guo, W., Lim, C. J., Bi, X., Sokhansanj, S., & Melin, S. (2013). Determination of effective thermal conductivity and specific heat capacity of wood pellets. Fuel, 103, 347–355. https://doi.org/10.1016/j.fuel.2012.08.037 Guo, W., Trischuk, K., Bi, X., Lim, C. J., & Sokhansanj, S. (2014). Measurements of wood pellets self-heating kinetic parameters using isothermal calorimetry. Biomass and Bioenergy, 63, 1–9. https://doi.org/10.1016/j.biombioe.2014.02.022 Hartley, I. D., Kamke, F. A., & Peemoeller, H. (1992). Cluster theory for water sorption in wood. Wood Science and Technology, 26(2), 83–99. https://doi.org/10.1007/BF00194465 Hartley, Ian D., & Wood, L. J. (2008). Hygroscopic properties of densified softwood pellets. Biomass and Bioenergy, 32(1), 90–93. https://doi.org/10.1016/j.biombioe.2007 Hashemi, S. Z. (2013). Studies on encapsulation of pelletized biomass (M.A.Sc. Thesis). University of British Columbia, Vancouver, BC. Hoadley, R. B. (2000). Understanding wood: A craftsman’s guide to wood technology. Taunton Press. Igathinathane, C., Pordesimo, L. O., Womac, A. R., & Sokhansanj, S. (2009). Hygroscopic moisture sorption kinetics modeling of corn stover and its fractions. Applied Engineering in Agriculture, 25(1), 65–73. https://doi.org/10.13031/2013.25420 International Standards Organization. (2014). ISO 17225-2: Solid biofuels—Fuel specifications and classes—Part 2: Graded wood pellets. In International Standard. Jensen, P. D., Temmerman, M., & Westborg, S. (2011). Internal particle size distribution of biofuel pellets. Fuel, 90(3), 980–986. https://doi.org/10.1016/j.fuel.2010.11.029 Jian, F., Jayas, D. S., & White, N. D. (2009). Temperature fluctuations and moisture migration in wheat stored for 15 months in a metal silo in Canada. Journal of Stored Products Research, 133  45(2), 82–90. https://doi.org/10.1016/j.jspr.2008.09.004 Jiang, Y., Lawrence, M., Hussain, A., Ansell, M., & Walker, P. (2019). Comparative moisture and heat sorption properties of fibre and shiv-derived from hemp and flax. Cellulose, 26(2), 823–843. https://doi.org/10.1007/s10570-018-2145-0 Jowitt, R., & Wagstaffe, P. J. (1989). The certification of the water content of microcrystalline cellulose (MCC) at 10 water activities. In EUR 12429. Luxembourg: Commission of the European Communities (EC). Junginger, H. M., Mai‐Moulin, T., Daioglou, V., Fritsche, U., Guisson, R., Hennig, C., … Lamers, P. (2019). The future of biomass and bioenergy deployment and trade: A synthesis of 15 years IEA Bioenergy Task 40 on sustainable bioenergy trade. Biofuels, Bioproducts and Biorefining, 13(2), 247–266. https://doi.org/10.1002/bbb.1993 Kaliyan, N., & Vance Morey, R. (2009). Factors affecting strength and durability of densified biomass products. Biomass and Bioenergy, 33(3), 337–359. https://doi.org/10.1016/j.biombioe.2008.08.005 Katsevich, A. (2002). Theoretically Exact Filtered Backprojection-Type Inversion Algorithm for Spiral CT. SIAM Journal on Applied Mathematics, 62(6), 2012–2026. https://doi.org/10.1137/S0036139901387186 Khazaei, J. (2008). Water absorption characteristics of three wood varieties. Cercetări Agronomice În Moldova, 41(2), 134. Retrieved from http://www.uaiasi.ro/CERCET_AGROMOLD/ro/fisiere/2008/V2/CA2-08-01.pdf Khoshtaghaza, M., Sokhansanj, S., & Gossen, B. (1999). Quality of alfalfa cubes during shipping and storage. Applied Engineering in Agriculture, 15(6), 671–675. Kim, J. Y., Kim, M., & Lee, J. (2014). Role of moisture on the lipid oxidation determined by D2O 134  in a linoleic acid model system. Food Chemistry, 146, 134–140. https://doi.org/10.1016/j.foodchem.2013.09.025 King, B. A., Winward, T. W., & Bjorneberg, D. L. (2010). Laser precipitation monitor for measurement of drop size and velocity of moving spray-plate sprinklers. Applied Engineering in Agriculture, 26(2), 263–271. https://doi.org/10.13031/2013.29551 Kingston, A. M., Sakellariou, A., Sheppard, A. P., Varslot, T. K., & Latham, S. J. (2010, September 1). An auto-focus method for generating sharp 3D tomographic images. 7804. https://doi.org/10.1117/12.860285 Kofman, P. D. (2007). Simple ways to check wood pellet quality. Coford Connect, 11(11), 8–9. Retrieved from http://www.coford.ie/media/coford/content/publications/projectreports/cofordconnects/ccncheckpelletquality.pdf Koppejan, J., Lönnermark, A., Persson, H., Larsson, I., Blomqvist, P., Arshadi, M., … Wheeler, P. (2013). Health and safety aspects of solid biomass storage, transportation and feeding. IEA Bioenergy Task 32. Retrieved from IEA Bioenergy Task 32 website: https://www.ieabioenergy.com/wp-content/uploads/2013/10/Health-and-Safety-Aspects-of-Solid-Biomass-Storage-Transportation-and-Feeding.pdf Krigstin, S., Wetzel, S., Jayabala, N., Helmeste, C., Madrali, S., Agnew, J., & Volpe, S. (2018). Recent Health and Safety Incident Trends Related to the Storage of Woody Biomass: A Need for Improved Monitoring Strategies. Forests, 9(9), 1–24. https://doi.org/10.3390/f9090538 Krupińska, B., Strømmen, I., Pakowski, Z., & Eikevik, T. M. (2007). Modeling of sorption isotherms of various kinds of wood at different temperature conditions. Drying Technology, 135  25(9), 1463–1470. https://doi.org/10.1080/07373930701537062 Kuang, X., Shankar, T. J., Bi, X. T., Lim, C. J., Sokhansanj, S., & Melin, S. (2009). Rate and peak concentrations of off-gas emissions in stored wood pellets—Sensitivities to temperature, relative humidity, and headspace volume. The Annals of Occupational Hygiene, 53(8), 789–796. https://doi.org/10.1093/annhyg/mep049 Kubler, H. (1990). Self-Heating of Lignocellulosic Materials. In G. Nelson (Ed.), Fire and Polymers (pp. 429–449). American Chemical Society. https://doi.org/10.1021/bk-1990-0425.ch026 Lam, P. S. (2011). Steam explosion of biomass to produce durable wood pellets. University of British Columbia, Vancouver, Canada. Lam, P. Y., Lam, P. S., Sokhansanj, S., Bi, X. T., Lim, C. J., & Melin, S. (2014). Effects of pelletization conditions on breaking strength and dimensional stability of Douglas fir pellet. Fuel, 117, 1085–1092. https://doi.org/10.1016/j.fuel.2013.10.033 Larsson, I., Lönnermark, A., Blomqvist, P., & Persson, H. (2017). Measurement of self-heating potential of biomass pellets with isothermal calorimetry. Fire and Materials, 41(8), 1007–1015. https://doi.org/10.1002/fam.2441 Larsson, I., Lönnermark, A., Blomqvist, P., Zimmermann, F., & Dahlbom, S. (2020). Medium-scale self-heating tests with biomass pellets (No. RISE Report 2020:51; pp. 1–58). RISE Research Institutes of Sweden. Retrieved from RISE Research Institutes of Sweden website: https://www.diva-portal.org/smash/get/diva2:1432456/FULLTEXT01.pdf Larsson, S. H., Lestander, T. A., Crompton, D., Melin, S., & Sokhansanj, S. (2012). Temperature patterns in large scale wood pellet silo storage. Applied Energy, 92, 322–327. https://doi.org/10.1016/j.apenergy.2011.11.012 136  Lau, A. K., Lo, K. V., Liao, P. H., & Yu, J. C. (1992). Aeration experiments for swine waste composting. Bioresource Technology, 41(2), 145–152. https://doi.org/10.1016/0960-8524(92)90185-Z Lee, J. S., & Lube, V. M. (2017). Wetting pellets with moist air [YouTube]. Vancouver, B.C.: UBC Wood Science Department. Retrieved from https://youtu.be/bxc9MQXCgko Lee, J. S., Sokhansanj, S., Lau, A. K., & Lim, C. J. (2019, July 11). Moisture characteristics of British Columbia (BC) wood pellets. Presented at the The Canadian Society for Bioengineering Annual Conference, Vancouver, BC. Lee, J. S., Sokhansanj, S., Lau, A. K., Lim, C. J., Bi, X. T., Basset, V., … Melin, S. O. (2015). The effects of storage on the net calorific value of wood pellets. Canadian Biosystems Engineering, 57(1). https://doi.org/10.7451/CBE.2015.57.8.5 Lee, J. S., Sokhansanj, S., Lim, C. J., Lau, A., & Bi, T. (2019). Comparative analysis of sorption isotherms for wood pellets and solid wood. Applied Engineering in Agriculture, 35(4), 475–479. https://doi.org/10.13031/aea.13238 Lehtikangas, P. (2000). Storage effects on pelletised sawdust, logging residues and bark. Biomass and Bioenergy, 19(5), 287–293. https://doi.org/10.1016/S0961-9534(00)00046-5 Lestander, T. A. (2008). Water absorption thermodynamics in single wood pellets modelled by multivariate near-infrared spectroscopy. Holzforschung, 62(4), 429–434. https://doi.org/10.1515/HF.2008.071 Leuk, P., Schneeberger, M., Hirn, U., & Bauer, W. (2016). Heat of sorption: A comparison between isotherm models and calorimeter measurements of wood pulp. Drying Technology, 34(5), 563–573. https://doi.org/10.1080/07373937.2015.1062391 Li, X., Song, H., Wang, Q., Meesri, C., Wall, T., & Yu, J. (2009). Experimental study on drying 137  and moisture re-adsorption kinetics of an Indonesian low rank coal. Journal of Environmental Sciences, 21, S127–S130. https://doi.org/10.1016/S1001-0742(09)60055-3 Lube, V. M. (2016). Effects of moisture-induced thickness swelling on the microstructure of oriented strand board (MSc Thesis). University of British Columbia, Vancouver, Canada, Wood Science department. Mahmoudi, A. H., Hoffmann, F., & Peters, B. (2014). Application of XDEM as a novel approach to predict drying of a packed bed. International Journal of Thermal Sciences, 75, 65–75. https://doi.org/10.1016/j.ijthermalsci.2013.07.016 Mani, S., Tabil, L., & Sokhansanj, S. (2003). An overview of compaction of biomass grinds. Powder Handling and Processing, 15(3), 160–168. Mantanis, G. I. (1994). Swelling of Lignocellulosic Materials in Water and Organic Liquids (University of Wisconsin-Madison; p. 56). University of Wisconsin-Madison, Madison. Retrieved from http://users.teilar.gr/~mantanis/PhD-thesis.pdf Mbogga, M. S., Hamann, A., & Wang, T. (2009). Historical and projected climate data for natural resource management in western Canada. Agricultural and Forest Meteorology, 149(5), 881–890. https://doi.org/10.1016/j.agrformet.2008.11.009 McMullen, J., Fasina, O. O., Wood, C. W., & Feng, Y. (2005). Storage and handling characteristics of pellets from poultry litter. Applied Engineering in Agriculture, 21(4), 645. https://doi.org/10.13031/2013.18553 Melin, S., Svedberg, U., & Samuelsson, J. (2008). Emissions from Woodpellets During Ocean Transportation (EWDOT)—Research Report. In Wood Pellet Association of Canada. Meze’e, Y. N., Noah Ngamveng, J., & Bardet, S. (2008). Effect of enthalpy-entropy compensation during sorption of water vapour in tropical woods: The case of bubinga (Guibourtia 138  Tessmanii J. Léonard; G. Pellegriniana JL). Thermochimica Acta, 468(1), 1–5. Mhaske, S. N., Pathak, K., & Basak, A. (2019). A comprehensive design of rainfall simulator for the assessment of soil erosion in the laboratory. CATENA, 172, 408–420. https://doi.org/10.1016/j.catena.2018.08.039 Miura, K. (2016). Adsorption of water vapor from ambient atmosphere onto coal fines leading to spontaneous heating of coal stockpile. Energy & Fuels, 30(1), 219–229. https://doi.org/10.1021/acs.energyfuels.5b02324 MNP Ltd. (2015). BC Forest Industry—Economic Impact Study. January 2015. BC Council of Forest Industries. Retrieved from BC Council of Forest Industries website: http://www.cofi.org/wp-content/uploads/2015/01/bc_industry_impact_01-2015.pdf Murray, G. (2018, June). Canadian wood pellet update. Presented at the Canadian Bioeconomy Conference, Prince George, BC. Nelson, M. I., & Chen, X. D. (2007). Survey of experimental work on the self-heating and spontaneous combustion of coal. In G. B. Stracher (Ed.), Geology of coal fires: Case studies from around the world: Reviews in Engineering Geology XVIII (pp. 31–83). Boulder, CO: The Geological Society of America. Nikitin, N. l. (1966). The Chemistry of Cellulose and Wood. Schmorak, Israel Program for Scientific Translations, Jerusalem. Nilsson, C., Ramebäck, H., Hill, C., & Arshadi, M. (2019). Water up-take in fuel pellets studied by Dynamic Vapour Sorption (DVS) analysis and its potential role in self-heating during storage. European Journal of Wood and Wood Products, 77(1), 5–14. https://doi.org/10.1007/s00107-018-1359-z Nopens, M., Wadsö, L., Ortmann, C., & Fröba, M. (2019). Measuring the heat of interaction 139  between lignocellulosic materials and water. Forests, 10(8), 674–695. https://doi.org/10.3390/f10080674 Obernberger, I., & Thek, G. (2010). Chapter 5 Safety considerations and health concerns relating to pellets during storage, handling and transportation. In The pellet handbook: The production and thermal utilisation of biomass pellets (Vols. 1–3, pp. 47–84). London, UK: Earthscan Ltd. Ouertani, S., Azzouz, S., Hassini, L., & Belghith, A. (2011). Palm wood drying and optimization of the processing parameters. Wood Material Science and Engineering, 6(1–2), 75–90. Ouertani, S., Azzouz, S., Hassini, L., Koubaa, A., & Belghith, A. (2014). Moisture sorption isotherms and thermodynamic properties of Jack pine and palm wood: Comparative study. Industrial Crops and Products, 56, 200–210. Oveisi, E., Lau, A., Sokhansanj, S., Lim, C. J., Bi, X., Larsson, S. H., & Melin, S. (2013). Breakage behavior of wood pellets due to free fall. Powder Technology, 235, 493–499. https://doi.org/10.1016/j.powtec.2012.10.022 Peemoeller, H., & Sharp, A. R. (1985). NMR study of cellulose-water systems: Water proton spin-lattice relaxation in the rotating reference frame. Polymer, 26(6), 859–864. https://doi.org/10.1016/0032-3861(85)90128-4 Peng, J. H., Bi, H. T., Lim, C. J., & Sokhansanj, S. (2013). Study on Density, Hardness, and Moisture Uptake of Torrefied Wood Pellets. Energy & Fuels, 27(2), 967–974. https://doi.org/10.1021/ef301928q Peralta, P. N., Bangi, A. P., & Lee, A. W. (1997). Thermodynamics of moisture sorption by the giant-timber bamboo. Holzforschung-International Journal of the Biology, Chemistry, Physics and Technology of Wood, 51(2), 177–182. 140  Perry, R. H., & Green, D. (1985). Chemical engineers’ handbook (9th Edition). New York: MacGraw-Hill Inc. Proskurina, S., Junginger, M., Heinimö, J., Tekinel, B., & Vakkilainen, E. (2019). Global biomass trade for energy—Part 2: Production and trade streams of wood pellets, liquid biofuels, charcoal, industrial roundwood and emerging energy biomass. Biofuels, Bioproducts and Biorefining, 13(2), 371–387. https://doi.org/10.1002/bbb.1858 Rautkari, L., Hill, C. A. S., Curling, S., Jalaludin, Z., & Ormondroyd, G. (2013). What is the role of the accessibility of wood hydroxyl groups in controlling moisture content? Journal of Materials Science, 48(18), 6352–6356. https://doi.org/10.1007/s10853-013-7434-2 Rees, W., Hearle, J., & Peters, R. (1960). Heat of absorption. Moisture in Textiles. Wiley Interscience, New York, 33–58. Rezaei, H., Lim, C. J., Lau, A., Bi, X., & Sokhansanj, S. (2017). Development of empirical drying correlations for ground wood chip and ground wood pellet particles. Drying Technology, 35(12), 1423–1432. https://doi.org/10.1080/07373937.2016.1198912 Rezaei, H., Lim, C. J., Lau, A., & Sokhansanj, S. (2016). Size, shape and flow characterization of ground wood chip and ground wood pellet particles. Powder Technology, 301, 737–746. https://doi.org/10.1016/j.powtec.2016.07.016 Rosa, M. E., & Fortes, M. A. (2007). Water absorption by cork. Wood and Fiber Science, 25(4), 339–348. Retrieved from https://wfs.swst.org/index.php/wfs/article/view/1245/1245 Ross, R. J. (2010). Wood Handbook: Wood as an Engineering Material (General Technical Report No. FPL-GTR-190, 2010). USDA Forest Service, Forest Products Laboratory. Ryckeboer, J., Mergaert, J., Vaes, K., Klammer, S., De Clercq, D., Coosemans, J., … Swings, J. (2003). A survey of bacteria and fungi occurring during composting and self-heating 141  processes. Annals of Microbiology, 53(4), 349–410. Samuelsson, R., Larsson, S. H., Thyrel, M., & Lestander, T. A. (2012). Moisture content and storage time influence the binding mechanisms in biofuel wood pellets. Applied Energy, 99, 109–115. https://doi.org/10.1016/j.apenergy.2012.05.004 Sandberg, K., & Salin, J. G. (2012). Liquid water absorption in dried Norway spruce timber measured with CT scanning and viewed as a percolation process. Wood Science and Technology, 46(1), 207–219. https://doi.org/10.1007/s00226-010-0371-1 Schilling, C., Wöhler, M., Yazdanpanah, F., Bi, X., Lau, A., Lim, C. J., … Pelz, S. (2015). Development of a novel wood pellet durability tester for small samples. In World Sustainable Energy Days. Siau, J. F. (1984). Basic Wood-Moisture Relationships. In J. F. Siau (Ed.), Transport Processes in Wood (pp. 1–34). Berlin, Heidelberg: Springer Berlin Heidelberg. https://doi.org/10.1007/978-3-642-69213-0_1 Simón, C., Esteban, L. G., de Palacios, P., Fernández, F. G., Martín-Sampedro, R., & Eugenio, M. E. (2015). Thermodynamic analysis of water vapour sorption behaviour of juvenile and mature wood of Abies alba Mill. Journal of Materials Science, 50(22), 7282–7292. https://doi.org/10.1007/s10853-015-9283-7 Simpson, William T. (1971). Equilibrium moisture content prediction for wood. Forest Products Journal, 21(5), 48–49. Retrieved from https://www.fpl.fs.fed.us/documnts/pdf1971/simps71a.pdf Simpson, William T., & Skaar, C. (1968). Effect of restrained swelling on wood moisture content. In United States Department of Agriculture (pp. 1–4). US Department of Agriculture, Forest Service, Forest Products Laboratory. Retrieved from US Department of Agriculture, 142  Forest Service, Forest Products Laboratory website: https://www.fpl.fs.fed.us/documnts/fplrn/fplrn0196.pdf Simpson, W.T. (2001). Wood: Dimensional Change from Moisture. In K. H. J. Buschow, R. W. Cahn, M. C. Flemings, B. Ilschner, E. J. Kramer, S. Mahajan, & P. Veyssière (Eds.), Encyclopedia of Materials: Science and Technology (pp. 9627–9629). Oxford: Elsevier. https://doi.org/10.1016/B0-08-043152-6/01743-5 Sivertsen, M. S., & Vestøl, G. I. (2010). Liquid water absorption in uncoated Norway spruce (Picea abies) claddings as affected by origin and wood properties. Wood Material Science and Engineering, 5(3–4), 181–193. https://doi.org/10.1080/17480272.2010.503939 Sjöström, J., & Blomqvist, P. (2014). Direct measurements of thermal properties of wood pellets: Elevated temperatures, fine fractions and moisture content. Fuel, 134, 460–466. https://doi.org/10.1016/j.fuel.2014.05.088 Skaar, C. (1988). Wood-Water Relations (1st ed.). Berlin/Heidelberg: Springer Berlin Heidelberg. https://doi.org/10.1007/978-3-642-73683-4 Sluiter, A., Ruiz, R., Scarlata, C., Sluiter, J., & Templeton, D. (2005). Determination of Ash in Biomass—Laboratory Analytical Procedure (LAP) (No. NREL/TP-510-42622). National Renewable Energy Laboratory. Retrieved from National Renewable Energy Laboratory website: http://www.nrel.gov/docs/gen/fy08/42619.pdf Sokhansanj, S., Khoshtaghaza, H., Schoenau, G., Arinze, E., & Tabil, L. (2003). Heat and moisture transfer and quality changes in containerized alfalfa cubes during transport. Transactions of the ASAE, 46(2), 423. Stamm, A. J. (1956). Diffusion of water into uncoated cellophane. I. From rates of water vapor adsorption, and liquid water absorption. The Journal of Physical Chemistry, 60(1), 76–82. 143  https://doi.org/10.1021/j150535a019 Stamm, A. J. (1964). Wood and cellulose science. New York: Ronald Press Co. Stamm, A. J., & Loughborough, W. K. (1935). Thermodynamics of the swelling of wood. The Journal of Physical Chemistry, 39(1), 121–132. Retrieved from https://www.fpl.fs.fed.us/documnts/pdf1935/stamm35a.pdf Stelte, W., Holm, J. K., Sanadi, A. R., Barsberg, S., Ahrenfeldt, J., & Henriksen, U. B. (2011). A study of bonding and failure mechanisms in fuel pellets from different biomass resources. Biomass and Bioenergy, 35(2), 910–918. https://doi.org/10.1016/j.biombioe.2010.11.003 Sultana, A., & Kumar, A. (2012). Ranking of biomass pellets by integration of economic, environmental and technical factors. Biomass and Bioenergy, 39, 344–355. https://doi.org/10.1016/j.biombioe.2012.01.027 Tagne, M. S., Zoulalian, A., Njomo, D., & Bonoma, B. (2011). Modelisation of desorption isotherms and estimation of the thermophysic and thermodynamic properties of tropical woods in Cameroon: The case of Ayous and Ebony woods. Revue Des Energies Renouvelables, 14(3), 487–500. Tang, Y., Chandra, R. P., Sokhansanj, S., & Saddler, J. N. (2018). The Role of Biomass Composition and Steam Treatment on Durability of Pellets. BioEnergy Research, 11(2), 341–350. https://doi.org/10.1007/s12155-018-9900-9 Thakur, A. K., & Gupta, A. K. (2006). Water absorption characteristics of paddy, brown rice and husk during soaking. Journal of Food Engineering, 75(2), 252–257. https://doi.org/10.1016/j.jfoodeng.2005.04.014 Theerarattananoon, K., Xu, F., Wilson, J., Ballard, R., Mckinney, L., Staggenborg, S., … Wang, D. (2011). Physical properties of pellets made from sorghum stalk, corn stover, wheat 144  straw, and big bluestem. Industrial Crops and Products, 33(2), 325–332. https://doi.org/10.1016/j.indcrop.2010.11.014 Thomas, M., & Van der Poel, A. (1996). Physical quality of pelleted animal feed 1. Criteria for pellet quality. Animal Feed Science and Technology, 61, 89–112. https://doi.org/10.1016/0377-8401(96)00949-2 Time, B. (1998). Hygroscopic Moisture Transport in Wood (Norwegian University of Science and Technology). Norwegian University of Science and Technology, Trondheim, Norway. Retrieved from http://www.ivt.ntnu.no/docs/bat/bm/phd/AvhandlingBeritTime.pdf Tooyserkani, Z. (2013). Hydrothermal pretreatment of softwood biomass and bark for pelletization (PhD Thesis). University of British Columbia, Vancouver, BC. Transport Malta. (2017). Safety investigation into the self-combustion of wood pellets in cargo hold no. 5 on board the Maltese registered bulk carrier V DUE at Point Lynas Anchorage, Liverpool in position 53° 23.70’ N 004° 09.90’ W on 29 November 2016 (Marine Safety Investigation Report No. 26/2017; pp. 1–50). Tumuluru, J. S. (2014). Effect of process variables on the density and durability of the pellets made from high moisture corn stover. Biosystems Engineering, 119, 44–57. https://doi.org/10.1016/j.biosystemseng.2013.11.012 Tumuluru, J. S., Sokhansanj, S., Lim, C. J., Bi, T., Lau, A., Melin, S., … Oveisi, E. (2010). Quality of wood pellets produced in British Columbia for export. Applied Engineering in Agriculture, 26(6), 1013–1020. https://doi.org/0.13031/2013.35902 Tumuluru, J. S., Wright, C. T., Hess, J. R., & Kenney, K. L. (2011). A review of biomass densification systems to develop uniform feedstock commodities for bioenergy application. Biofuels Bioproducts & Biorefining-Biofpr, 5(6), 683–707. 145  https://doi.org/10.1002/bbb.324 Turhan, M., Sayar, S., & Gunasekaran, S. (2002). Application of Peleg model to study water absorption in chickpea during soaking. Journal of Food Engineering, 53(2), 153–159. https://doi.org/10.1016/S0260-8774(01)00152-2 Vancouver Fraser Port Authority. (2020, August 17). Fee Document. Retrieved from https://www.portvancouver.com/wp-content/uploads/2020/06/2020-08-17-FEE-Document-for-2020.pdf Ventura, M. (2008). Bulk Carriers – Ship Design I for MSc in Marine Engineering and Naval Architecture. Retrieved from http://www.mar.ist.utl.pt/mventura/Projecto-Navios-I/EN/SD-1.4.3-Bulk%20Carriers.pdf Verhoest, C., & Ryckmans, Y. (2012). Industrial Wood Pellets Report. Laborelec. Walsh, B. L., Chow, J. C., Watson, E. L., & Townsley, P. M. (1980). Computerized estimation of bacterial heat production in a laboratory fermenter. Canadian Agricultural Engineering, 22(1), 93–95. Wang, R., Liu, S., Li, H., & Ma, H. (2017). Study of temperature changes in wheat grain heap during cooling process. Cereals & Oils, (5), 23–27. Retrieved from http://en.cnki.com.cn/Article_en/CJFDTotal-LSYY201705007.htm Wang, S., Yuan, X., Li, C., Huang, Z., Leng, L., Zeng, G., & Li, H. (2016). Variation in the physical properties of wood pellets and emission of aldehyde/ketone under different storage conditions. Fuel, 183, 314–321. https://doi.org/10.1016/j.fuel.2016.06.083 Weichert, L. (1963). Investigations on sorption and swelling of spruce, beech and compressed beech wood at temperatures between 20 °C and 100 °C. Holz Roh Werkst, 21(8), 290–300. White, R. H. (1987). Effect of lignin content and extractives on the higher heating value of wood. 146  Wood and Fiber Science, 19(4), 446–452. Retrieved from http://www.fpl.fs.fed.us/documnts/pdf1987/white87a.pdf Whittaker, C., & Shield, I. (2017). Factors affecting wood, energy grass and straw pellet durability–A review. Renewable and Sustainable Energy Reviews, 71, 1–11. https://doi.org/10.1016/j.rser.2016.12.119 Wood Pellet Association of Canada. (2016). CANplus handbook – Quality control certification for wood pellet. Retrieved from https://www.pellet.org/CANplus/downloadcp.html Yan, W., Hoekman, S. K., Broch, A., & Coronella, C. J. (2014). Effect of hydrothermal carbonization reaction parameters on the properties of hydrochar and pellets. Environmental Progress & Sustainable Energy, 33(3), 676–680. https://doi.org/10.1002/ep.11974 Yazdanpanah, F., Sokhansanj, S., Lau, A. K., Lim, C. J., Bi, X., & Melin, S. (2011). Airflow versus pressure drop for bulk wood pellets. Biomass and Bioenergy, 35(5), 1960–1966. https://doi.org/10.1016/j.biombioe.2011.01.042 Yu, M., Igathinathane, C., Hendrickson, J., & Sanderson, M. (2014). Moisture sorption kinetics of switchgrass, big bluestem, and bromegrass biomass. Trans.ASABE, 57(4), 1231–1242. https://doi.org/10.13031/trans.57.10130 Zaini, P., Sokansanj, S., Bi, X., Lim, C., Mani, S., Melin, S., & Kadla, J. (2008). Density, heating value, and composition of pellets made from lodgepole pine (pinus concorta douglas) infested with mountain pine beetle (dendroctonus ponderosae hopkins). Canadian Biosystems Engineering, 50, 47–55. Zauer, M., Pfriem, A., & Wagenführ, A. (2013). Toward improved understanding of the cell-wall density and porosity of wood determined by gas pycnometry. Wood Science and 147  Technology, 47(6), 1197–1211. https://doi.org/10.1007/s00226-013-0568-1 Zelinka, S. L., Glass, S. V., & Thybring, E. E. (2018). Myth versus reality: Do parabolic sorption isotherm models reflect actual wood–water thermodynamics? Wood Science and Technology, 52(6), 1701–1706. https://doi.org/10.1007/s00226-018-1035-9 Zhang, K., Johnson, L., Vara Prasad, P. V., Pei, Z., & Wang, D. (2015). Big bluestem as a bioenergy crop: A review. Renewable and Sustainable Energy Reviews, 52, 740–756. https://doi.org/10.1016/j.rser.2015.07.144 148  Appendices Appendix A  Additional Tables and Figures Table A.1 Sorption isotherm equations and their number of parameters.   Model Name Model Equation Variable number Reference 1.  Henry’s Equation or Linear Equation a  (Empirical) 𝐸𝑀𝐶 = 𝐾 ∙ 𝐸𝑅𝐻 1 Ng et al. (2001) 2.  Langmuir Equation a,b (theoretical) 𝐸𝑀𝐶 = 𝑀0𝐾(𝐸𝑅𝐻)1 + 𝐾(𝐸𝑅𝐻) 2 Times (1998) 3.  Brunauer–Emmett–Teller (BET) equation a,b (theoretical) 𝐸𝑀𝐶 = 𝑀0𝐾(𝐸𝑅𝐻)[1 − 𝐸𝑅𝐻][1 − 𝐸𝑅𝐻 + 𝐾(𝐸𝑅𝐻)] 2 Times (1998) 4. Modified Chung-Pfost equation 𝐸𝑀𝐶 = −1𝐶ln (−1𝐾𝑙𝑛(𝐸𝑅𝐻)) where 𝐾 = 𝐴/(𝑇 + 𝐵) 2 ASABE (2007)  5.  Modified Henderson equation (Semi-empirical) 𝐸𝑀𝐶 = [−1𝐾ln(1 − 𝐸𝑅𝐻)]1𝐶 where 𝐾 = 𝐴(𝑇 + 𝐵) 2 ASABE (2007) He (2013) 6. Day-Nelson Equation (Semi-empirical) 𝐸𝑀𝐶 = [1𝐾ln(1 − 𝐸𝑅𝐻)]1𝐶 where 𝐾 = 𝐴𝑇𝐵; 𝐶 = 𝐷𝑇𝐸 2 Avramidis (1989) 7. Modified Hasley equation (Semi-empirical) 𝐸𝑀𝐶 = [−𝐾ln(𝐸𝑅𝐻)]1𝐶 where 𝐾 = exp(𝐴 + 𝐵 ∙ 𝑇) 2 ASABE (2007) He (2013) 8. Modified Oswin equation (Empirical) 𝐸𝑀𝐶 = 𝐾 [𝐸𝑅𝐻1 − 𝐸𝑅𝐻]1𝐶 where 𝐾 = 𝐴 + 𝐵 ∙ 𝑇 2 ASABE (2007) He (2013) 9.  Guggenheim, Anderson and de Boer (GAB) equationb (theoretical) 𝐸𝑀𝐶= 𝑀0𝐾 ∙ 𝐶(𝐸𝑅𝐻)[1 − 𝐾 ∙ 𝐸𝑅𝐻][1 − 𝐾 ∙ 𝐸𝑅𝐻 + 𝐾 ∙ 𝐶(𝐸𝑅𝐻)] where 𝐶 is equilibrium constant related to the monolayer sorption and 𝐾 is equilibrium constant related to the multilayer sorption. 3 ASABE (2007); Olek, Majka & Czajkowski (2013)  149  10.  Hailwood-Horrobin (H-H) equation (theoretical) 𝐸𝑀𝐶 =18𝑊(𝐾1 ∙ 𝐾2 ∙ 𝐸𝑅𝐻1 + 𝐾1 ∙ 𝐾2 ∙ 𝐸𝑅𝐻+𝐾2 ∙ 𝐸𝑅𝐻1 − 𝐾2 ∙ 𝐸𝑅𝐻) where 𝐾1 is the equilibrium constant between the hydrate and the dissolved water, 𝐾2 is the equilibrium constant between the dissolved water and water vapor in moist air, and 𝑊 (kg/kmol), apparent molecular mass of the dry wood per sorption sites. 3 Simpson (1971); Lenth & Kamke (2007) 11.  generalized D’Arcy and Watt (GDW) equation (Theoretical, expanded on GAB theory) 𝐸𝑀𝐶 = 𝑀0  (𝐾1 ∙ 𝐸𝑅𝐻1 + 𝐾1 ∙ 𝐸𝑅𝐻)∙ (1 − 𝐾2 ∙ (1 − 𝑤) ∙ 𝐸𝑅𝐻1 − 𝐾2 ∙ 𝐸𝑅𝐻) where 𝐾1 is the kinetic constant related to sorption of the primary sites, 𝐾2 is the kinetic constant related to sorption on the secondary sites, and 𝑤 is the ratio of water molecules bound to the primary sites and converted into the secondary sites. 4 Olek, Majka & Czajkowski (2013) 12. Peleg equation (purely empirical) 𝐸𝑀𝐶 = 𝐴 ∙ (𝐸𝑅𝐻)𝐵 + 𝐶 ∙ (𝐸𝑅𝐻)𝐷 4 Krupińska et al (2007) Note: EMC is the equilibrium moisture content in decimal dry basis (db), e.g. kg water per kg dry matter. ERH is equilibrium relative humidity of water vapor in air in decimal, e.g. Pa water per Pa saturation pressure. T is temperature in Kelvin. a. 𝐾 = 𝐾0 exp (𝑄𝑠𝑡𝑅𝑇). 𝑄𝑠𝑡 is the isosteric heat of adsorption in kJ/kg, 𝑅 is the gas constant in J/kg-K, 𝑇 is temperature in K. 𝑏. 𝑀0 is the moisture content (db) corresponding to complete monolayer coverage of all available sorption sites.    Table A.2 Conversion between moisture content in percent wet mass basis (%, wb) to moisture content in decimal dry mass basis (dec, db). Moisture content, m.c.  (%, wet mass basis) 5 10 15 20 25 30 35 40 45 50 Moisture content, M (decimal dry basis) 0.053 0.111 0.176 0.250 0.333 0.429 0.538 0.667 0.818 1.000 The equation used for conversion between the two units is M (dec, db) = m.c. / (100 – m.c.). 150   Figure A.1 Moisture adsorption of three types of pellets: residential hardwood, residential softwood, and industrial pellets in the atmosphere with a temperature of 30 °C and an RH of 95%. The experiment was done in a humidity chamber (ESPEC LHU-113) with 10-20 g samples on aluminum trays. The results showed that the moisture adsorption behaviors of three distinct types of pellets were very similar, with an average absolute difference of ~0.005 (db). 151   Figure A.2 Water absorption of softwood heating pellets at (a) 30°C, (b) 40°C, and (c) 50°C. S are pellets with less than 0.5 g each; M is between 0.5 to 1.0 g each; L is larger than 1.0 g each. The lines represent the Page model fitting curves. The experiment was done in a water bath (Fisher Scientific™ Isotemp™ Digital-Control Water Bath model 2300) using 10-20 g samples. The results showed that the wood pellets equilibrated to the saturation moisture content of 1.50-2.40 (db) in 3-5 minutes when immersing in water. 152    Figure A.3 Correlation between the mechanical durability (%), measured by a tumbler (ISO 17225-2), and the single pellet durability (%), measured by a shaker (Schilling et al., 2015). The number of repetitions is five for each data point. The pellets were wetted by a pre-determined volume of liquid water to reduce the pellets’ durability.  Figure A.4 The tumbler durability of softwood pellets versus their lengths. Each length group consists of six repetitions. Unsorted pellets are pellets that have the as-received size distribution (40% is less than 10 mm, 30% is 10-15 mm, 20% 15-20 mm, 10% is more than 20 mm). The figure shows little difference in the tumbler durability among the five length groups of pellets.  60%70%80%90%100%0% 20% 40% 60% 80% 100%Mechanical Durability, DtumblerSingle pellet durability, DsingleDtumbler = 0.002089(Dsingle)2 + 0.076317(Dsingle) + 70.333148R² = 0.96169790.091.092.093.094.095.096.097.098.099.0100.0Tunbler durability (%)Pellet length (mm)153    Figure A.5 The percentage of coefficient of variation (CV) of (a) moisture content vs. exposure time, (b) durability, (c) pellet density, and (d) percentage volumetric swelling (%swelling) vs. moisture content (dec, db). CV, which represents the variability in a set of data, is calculated as mean over standard deviation times 100. Each data point represents 20 repetitions. The variability of moisture content, as shown by their CVs, decreased with exposure time. This decrease in variability implied that the increase in variability of durability at moisture contents above 0.15 (db) was related to the variability in moisture content and might be caused by the high variability in pellet density and volumetric swelling. 154   Appendix B  Additional Information  Calculations for Maximum Temperature Rise from Heat of Wetting Heat of wetting is the total amount of heat released when wood pellets are wetted in excess water from an initial starting moisture content to moisture content beyond its fiber saturation point. At moisture content beyond the fiber saturation point, wood no longer releases heat because the available sites for hydrogen bonding with water molecules are fully occupied. For example, the heat of wetting, Q in J/g dry mas of a starting moisture contents, Mi, of wood pellets may be defined by the exponential equation below: Q = 81.82exp(-13.28Mi) (1) where Mi is the starting moisture content (dec, db) When moisture content is increased from 5% to 10%, the heat released ∆𝑸 can be calculated as ∆Q = Q(0.05) - Q(0.10) = 20.4 J/g (2) Assume the system is insulated and no heat is dissipated from the pellets, I can define: ∆Q = cpellet∆T (3) where cpellet=1.01+0.032Mavg*100 is the heat capacity of wood pellets in J/g-K,  ∆T is the temperature rise due to heat of sorption, Mavg, is the average moisture content between initial and final moisture content, in this case, is 0.75 dec, db Plugging in the values for heat capacity, I arrive at the temperature rise, ∆T, which is 16°C.   155   Model Development for Heat and Moisture Transfer of Single Pellet  The model is developed based on Miura (2016)’s model for coal with modifications to suit the case of a single wood pellet. General assumptions:  • Pellet is a cylinder. The diameter of pellets, 2R, is 0.0063 m. A length of 0.0012 m is used to calculate the surface area and volume of a pellet, assuming it is a cylinder.  • The pellet was placed in an infinite volume of stationary air of Tenv of temperature in K and penv of water vapor pressure, in Pa.  • The temperature T, the water vapor pressure p, and the amount of water vapor adsorbed q per unit weight of dried pellet were uniform throughout the pellet. • q and p were in equilibrium and their relationship obeys a linear relationship of q = a·p/ps Mass balance of water vapor around a wood pellet Wpelletd(q+ερbC)/dt = kcA(Cenv-C)  (B.1) where Wpellet is the weight of dried wood pellet, in kg, C and Cenv are respectively the concentrations of water vapor in the wood pellet and in the ambient atmosphere, in kg H2O(g)/m3, ε is the void fraction of the wood pellet including the pore volumes within the wood particles, which made up a pellet, ρb is the apparent density of the wood pellet, in kg/m3, t is the time, in seconds, kc is the mass transfer coefficient of water vapor in the gas film around the wood pellet, in m/s, and A is the outer surface area of the spherical particle. Eq. B.1 can be expressed as, Wpellet (ddtq+ερbdCdt) = kcA(Cenv-C)  (B.1b) 156  Since q is a function of both water vapor pressure, p and saturation pressure, ps, i.e. q = f(p, ps), by chain rule, dqdt=∂q∂pdpdt+∂q∂psdpsdt , while dpsdt=∂ps∂tdTdt, then dqdt=∂q∂pdpdt+∂q∂ps∙dpsdT∙dTdt   (B.2) Since C = p/(RgT), where Rg (= 0.4619 kJ/kg) is the gas constant for water, C = f(p, T), by chain rule, dCdt=∂C∂T∙dpdt+∂C∂T∙dTdt, while ∂C∂P=1RgT , ∂C∂T=-pRgT2 , then dCdt=1RgTdpdt-pRgT2dTdt  (B.3) So, now, plugging in Eqs. B.2 and B.3 into Eq. B.1b, and Cenv = penv/(RgTenv), we get, dqdpdpdt+dqdps∙dpsdT∙dTdt+ερb(1Rgdpdt-pRgT2dTdt)=kcAWpellet kcA (penvRgTenv-pRgT)    (B.4) Finally, we organize Eq. B.4 into terms with its two key variables, dp/dt and dT/dt, (∂q∂p+ερbRgT)dpdt+(∂q∂ps∙∂ps∂T-εpρbRgT2)dTdt=kcAWpelletRg (penvTenv-pT) (B.5) Enthalpy Balance around a wood pellet Additional assumptions • Adsorbed water is in a lower state than ordinary liquid water, where heat of adsorption is heat of condensation (=2500 kJ/kg water) plus differential heat of adsorption, hs. Wpelletddt(Hpellet+qHw(T)+ερbCaHa)=hA(Tenv-T)+kcA(Cenv-C)Hwv(Tenv) (B.6) where Ca is the total concentration of humid air in the wood pellet, in kg air/m3, Hpellet is the enthalpy of the dried pellet per unit weight, in kJ/kg pellet, Hw and Hwv are respectively the enthalpy of liquid water (a function of T) and the enthalpy of water vapor (a function of Tenv), and h is the heat transfer coefficient in the gas film around the wood pellet.  Eq. B.6 can be expressed as, 157  Wpellet {ddthpellet+ddt(qHw)+ερbddt(CaHa)}= hA(Tenv-T)+kcA(Cenv-C)Hwv(Tenv) (B.7) On the left-hand side of Eq. B.7, we start to break down the equation to quantifiable parameters. If the reference temperature is 0 K, we know, hpellet = cpelletT,  therefore d/dt hpellet = cpellet dT/dt     (B.8) By chain rule, ddt(qHw)=dqdtHw+qdHwdt , while dHwdt=dhwdT∙dTdt=cwdTdt , and we know dq/dt from Eq. B.2, we express,  ddt(qHw)=Hw [dqdpdpdt+dqdps∙dpsdT∙dTdt]+qcwdTdt   (B.9) By chain rule, ddt(CaHa)=dCadtHa+CadHadt, further break down into dCadTdTdTHa+cadHadT∙dTdt Given Ca = pT/(RgT), now ddt(CaHa)=-pTRgT2  HadTdt+pTRgTcadTdt (B.10) Plugging Eq. B.8 to B.10 to left-hand side of Eq. B.7, we get Wpellet {cpelletdTdt+Hw (dqdpdpdt+dqdps∙dpsdTdTdt)+qcwdTdt+ερb(-pTRgT2 HadTdt+pTRgT cadTdt)} (B.11) On the right hand side of Eq. B.6, we replace Cenv = penv/(RgTenv) and C = p/(RgT), we have hA(Tenv-T)+kcA (penvRgTenv-pRgT)Hwv,  Rearranging, A (h(Tenv-T)+kcRg(penvTenv-pT)Hwv)      (B.12) Finally, combining RHS (Eq. B.11) and LHS (Eq. B.12) and then reorganizing the equation, we get,  Hw (dqdp)dpdt+ {cpellet+Hwdqdps∙dpsdT+qcw+εpTρbRgT(ca-HaT )}dTdt=AWpellet(h(Tenv-T)+kcRg(penvTenv-pT)Hwv) (B.13) Miura (2016) converted A/Wpellet to 3/(ρbR) by using the formulas for area, A (=4πR2) and volume, V (=4/3πR3) of a sphere and the fact that density, ρb = Wpellet/V 158  Equations to calculate the heat capacities of humid air, ca Ca=ppTcwv+(pT-p)pTCDry air Ha=ppTHwv+(pT-p)pTHDry air Note: kc = 2E-05*2*DH2O-Dry Air /2R, where DH2O-Dry Air = 1E-10*exp(0.011*(T - 273)), h = 5E-04*2.0*λair/2R, where λair = (9.0E-05*T), T in Kelvin, Rg = 0.4619 kJ/kg water. Hw = cw (T – 273) Hwv = hs + 2500 + cwv (T-273), where hs = 1916.97*exp(-17.51*q).    MATLAB m-script for the Heat and Moisture Transfer Model  %       Parameters          % epsil = 0.3;  % decimal internal porosity of a pellet rhob = 1000;  % kg/m3 unit density of a pellet rg = 0.4619;  %kJ/kg water.K gas constant diam = 0.0063;  % m diameter of pellet length = 0.012; % m length of pellet area = pi*diam*length;  %surface area of a single pellet m2 volume = pi*diam^2/4*length;  %volume of a pellet m3 c_pellet = 1.2;  % specific heat of pelles kJ/kg-K c_water = 4.186; %specific heat of water kJ/kg-K c_watervapor = 2.0; %specific heat of water vapor kJ/kg-K c_dryair = 1.0; % specific heat of dry air kJ/kg-K patm = 101325.;  %Pa atmospheric pressure  ti = 25.;  %initial pellet temperature C top = ti+273.15; %top is initial temperature in K        %%      Initialization              %% pps = 0.0; % pps = p/ps = rh, water vapor pressure ratio - water activity stime = 0; %time counter in seconds dtime = 0.1; % time interval in seconds length_loop = 360000; T_op = zeros(length_loop,1); % vector to store temperature MC = zeros(length_loop,1);   % vector to store moisture content 159  p_H2O_g = zeros(length_loop,1); % vector to store partial pressure of water vapour Psat = zeros(length_loop,1); %vector to store saturation pressure of water vapour time = zeros(length_loop,1); %vector to store time for i = 1:length_loop      %%       Equations to be solved      %% %%       Setting up equation 4 coefficients     %% if pps > 0.8     % IF rh > 0.8, q = 0.4272*rh - 0.2025.     s = 0.4272;     q = 0.4272*pps - 0.2025; else     s = 0.178868; % slope of the emc (or q)  vs. pps (q = 0.178858*p/ps, valid unitl rh = 0.8)     q = s*pps; %equilibrium water weight fraction, kg water / kg dry material end   ps = satp(top);% saturation water vapor pressure, Pa p = ps*pps; %water vapor pressure, Pa dqdp = s/ps; %derivative dq/dp kg water/kg dry coal * Pa saturation pressure / Pa water x1 =  dqdp;      x2 = epsil/(rhob*rg*top); x11 = x1+x2;  %first LHS coefficient for dp/dt (eq 4 Miura)   dqdps = -s*p/ps^2; % derivative dq/dps = dq/dp*dp/dps dpsdT = ps*(3816.44)/(top-46.13)^2; % derivative dps/dT x12 = dqdps*dpsdT - epsil*p/(rhob*rg*top^2); % coefficient for dT/dt (eq 4 Miura)   t_env = 33;  %air temperature C t_envs = t_env+273;  %air temperature K ps_env = satp(t_envs); %saturation pressure at T_envs rh_env = 0.95; %0.80; % ambient relative humidity kc = 2*( 1E-10*exp(0.011*(top-273)))/diam; % mass transfer coeffiecient m/s Wpellet = volume*rhob; % mass of coal kg p_env = rh_env*ps_env; % ambient vapor pressure Pa y1 = kc*area/(Wpellet*rg);  y2 = p_env/t_envs - p/top;  k1 = y1*y2; % Forcing function (eq 6.3)   %%       Setting up Equation 6.4 coefficents       %% Qs = 403; %kJ/kg, constant differential heat on Sept 17, 2020 %Qs = 1916.97*exp(-17.51*q); % exponential differential heat of water sorption, kJ/kg water h_dryair = c_dryair*(top-273); % enthalpy of dry air, set 273K is ref. temp.  160  h_waterliquid=c_water*(top-273); %  enthalpy of water liquid  h_watervapor= Qs + 2500 + c_watervapor*(top-273); % enthalpy of water vapor in the pellet x21=h_waterliquid*dqdp; % coefficient for dp/dt in eq (6.4)   z1=c_pellet; z3= q*c_water; z4=h_waterliquid*dqdps*dpsdT; z5=epsil*patm/(rhob*rg*top); c_air=p/patm*c_watervapor+(patm-p)/patm*c_dryair; h_air=p/patm*h_watervapor+(patm-p)/patm*h_dryair; z6=c_air - h_air/top;  x22=z1+z3+z4+z5*z6; % coefficient for dT/dt in eq (6.4)   u1=3/(diam/2*rhob);  hc= 2*((9.0E-5*top)/1000)/diam; % convective heat transfer coefficient, kW/m^2.K u2=hc*(t_envs - top); h_watervapor= Qs + 2500 + c_watervapor*(t_envs-273); % enthalpy of water vapor in the environment u3=(kc/rg)*(p_env/t_envs - p/top)*h_watervapor; k2=u1*(u2+u3); % Forcing function (eq 5)   %%       Maxtrix to solve for the moisture and temperature profile      %% A=[x11, x12;x21, x22]; B=[k1;k2]; C=A\B*dtime; po=p; topo=top; p=po+C(1); top=topo+C(2);   ps = satp(top); pps = p/ps;   % Update non-constant variables rhob = -14670*q^2 + 2068*q + 1000;  % kg/m^3 unit density of a pellet  % Store values to plot in excel T_op(i) = top; MC(i) = q; p_H2O_g(i) = p; Psat(i) = ps; time(i) = stime;  % Next time step stime = stime + dtime;   end   161  RH = sprintf('%s dec of rh', num2str(p/ps)); Temp= sprintf('%s degC', num2str(top-273)); Moisture = sprintf('%s dec db', num2str(q)); disp(RH) disp(Temp) disp(Moisture)  function [ps] = satp (x)  ps=exp(23.1964-3816.44/(x-46.13));  end   X-ray Micro-computed Tomography of a Compressed Softwood Pellet The X-ray micro-computed tomography images (CT) and the write-up below were provided by Professor Philip Evans from the UBC Wood Science department. Methods The samples were air-dried prior to the X-ray scanning. A softwood pellet was scanned using an X-ray micro-CT system in the Department of Applied Mathematics at the Australian National University (ANU). The pellet was placed on a rotating stage while an X-ray tube projected a cone-shaped X-ray beam onto the pellet. A CCD (charge-coupled device) was used to collect the projection data. The resolution of the CT scan was 2800 × 2800 voxels with a voxel size of 2.92 µm. Projection data were sent to the Raijin supercomputer (National Computational Infrastructure, Australia) at The ANU to reconstruct the tomographic data set using an auto-focus aligning method (Kingston et al., 2010) and the Katservich (2002) formula. The tomographic data set was analyzed on a desktop workstation equipped with a discrete graphics accelerator (Nvidia RTX Titan 24GB). The CT images could be thresholded into two phases: air and wood. The 2D visualization of the tomograms was carried using the software NCViewer. The 3D structure of the sample was rendered using the open-sourced volume exploration software Drishti. Information on Drishti can be found in Limaye (2012).  162  Results Two-dimensional cross-sections of the pellet show numerous fine cracks representing the interfaces between particles within untreated softwood pellets (Figure B.1). Three-D images in Figure B.2 also show these cracks but also reveal damage (voids) to the interior and the outer surface of the pellet. Such voids and cracks would provide flow paths for water to penetrate the interior of pellets.    Figure B.1 Two-dimensional images of cross-sections through an untreated softwood pellet. Wood is grey and void (cracks) are black.  163   Figure B.2 Three-dimensional images of an untreated softwood pellet. Wood is brown, and void (cracks) are black/grey. A virtual section has been made through the center of the interior of the pellet on the left. Note internal voids in the pellet (left) and damage to the outer surface (right).   164  Appendix C  Additional Work  Moisture Adsorption from Ends and Circumference Sides of Single Pellets  Methods Set-1 tests - The objective is to compare the moisture content from the end pieces to the center pieces of single pellets. 20 single pellets with weights between 0.8-1.2 g were placed either in a humidity chamber at 95% RH and 30 °C. The samples were taken out to be weighed in one-hour intervals for a total 4-hour period. Each conditioned pellet was split into three parts by using a scissor. The two cuts from two ends yielded approximately 0.15 g pieces of a pellet. The center pieces weighed 0.5-0.7 g. The moisture content of the pieces from left and right of the pellets plus that of the remaining center pieces was measured using the A&D MF-50 moisture analyzer. Set-2 tests – To examine the hypothesis that the unplasticized ends of pellets are the main entrance of moisture, single pellets with weights between 0.8-1.2 g were either covered entirely with octadecyltriethoxysilane (OTS), a hydrophobic coating (Hashemi, 2013), covered only on two ends or coated on the circumference surface only. Uncovered pellets serve as an upper limit of total moisture absorbed. The pellets were 15 to 20 mm in length. Five samples were used for each treatment. Then, the samples were placed in a humidity chamber at 95% RH and 30 °C. The samples were taken out to be weighed in one-hour intervals for a total 4-hour period.  Figure C.1 The diagram of a single pellet. The two ends and the circumference sides were labeled.  165  Results The results from Set-1 tests showed that the moisture content for the end pieces was higher than the center pieces (Figure C.2). This fact suggests that the adsorbed moisture primarily accumulated near the two ends of the pellets.  Table C.1 (Set-2 tests results) showed that the weight of uncovered pellets increased by ~5%, while completely covered pellets increased by ~1%. This observation meant that the OTS blocked almost 80% of the moisture adsorption. When the two ends were covered with OTS, the moisture was stopped from adsorbing through the ends; the circumference sides were expected to be the primary entrance to adsorb most, if not all, of the moisture. When the circumference sides were covered with OTS, the two ends were expected to adsorb most of the moisture.  Assuming that the moisture adsorbed through the covered portion of a pellet can be neglected, moisture adsorbed by the ends or the circumference side is calculated as a percentage of the moisture adsorbed by the uncovered pellet (Eq. C.1).  % moisture adsorbed = % wt change of a treatment / % wt change of the uncovered pellet  (C.1) where % wt change is the difference in the weight of pellets after 4 hours in the humidity chamber over initial weight times 100. The two ends absorbed 40% of the total moisture adsorbed by the entire pellet, whereas the circumference side adsorbed 60%. The two ends adsorbed less moisture because of their smaller surface area. However, although the two ends have a surface area a factor of four smaller than that of the circumference side, as indicated by the length to diameter ratio of ~4, the two ends adsorbed disproportionally larger amount of moisture compared to their surface area. This evidence suggests the ends are the primary entrance of water vapor during moisture adsorption. 166   Figure C.2 Moisture content of end pieces (N = 40) and center pieces (N = 20) after exposing single pellets to an atmosphere with a relative humidity of 95% and a temperature of 30 °C. N is the number of repetitions. The end pieces had moisture contents which were ~5% (wb) higher than the center pieces.  Table C.1 percentage increase over initial weight (%wt change) of a single pellet after 4 hours in an atmosphere of RH of 95% and 30 °C in the humidity chamber.     %wt change after 4 hours Treatment L/D* Average Range (min-max) Uncovered 4.45 4.95 4.72-5.11 End covered 4.01 3.31 2.97-3.53 Circumference covered 4.46 2.00 1.66-2.35 Completely covered 4.34 0.95 0.68-1.18 *L/D is the average length to diameter ratio.  0.02.04.06.08.010.012.014.016.018.020.0End pieces  Center piecesMoisture content (%, wb)

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}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            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.24.1-0395603/manifest

Comment

Related Items