Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Sorbent attrition in fluidized carbon dioxide capture systems Knight, Andrew 2013

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

Item Metadata

Download

Media
24-ubc_2013_fall_knight_andrew.pdf [ 7.47MB ]
Metadata
JSON: 24-1.0074219.json
JSON-LD: 24-1.0074219-ld.json
RDF/XML (Pretty): 24-1.0074219-rdf.xml
RDF/JSON: 24-1.0074219-rdf.json
Turtle: 24-1.0074219-turtle.txt
N-Triples: 24-1.0074219-rdf-ntriples.txt
Original Record: 24-1.0074219-source.json
Full Text
24-1.0074219-fulltext.txt
Citation
24-1.0074219.ris

Full Text

       SORBENT ATTRITION IN FLUIDIZED CARBON DIOXIDE CAPTURE SYSTEMS   by   ANDREW KNIGHT  B.A.Sc., Queen?s University, 2011    A THESIS SUBMITTED IN PARTIAL FULFIILLMENT OF THE REQUREMENTS FOR THE DEGREE OF   MASTER OF APPLIED SCIENCE   in   THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES  (Chemical and Biological Engineering)    THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)   August 2013   ? Andrew Knight, 2013      ii  Abstract  As part of a collaborative Carbon Management Canada project, the extent and causes of attrition for three proposed carbon dioxide capture sorbents for fluidized power generation systems was investigated. The sorbents are crushed limestone, lime-based pellets containing calcium aluminate cement binder, and the same pellets with a mesoporous silica coating replacing the binder. The tested sorbents were provided by the University of British Columbia, the University of Ottawa/CANMET, and Laval University, respectively.  The attrition testing equipment was an air-jet apparatus based on the ASTM D5757 standard, with several modifications to improve testing capabilities. The experiments were conducted under varying attrition test periods, gas velocities, temperatures, humidities, and initial particle sizes. Following each test cycle, particle size distributions of the produced fines and remaining bed material were analysed, and scanning electron microscopy was utilized to provide additional insight on attrition mechanism. The strengths and weaknesses of the ASTM standard were evaluated, and a new attrition testing standard is proposed to improve operability and reduce the equipment size and material requirements. The experimental results indicate that the cement-bound pellets attrit to the same extent or worse than crushed limestone, and are highly sensitive to humidity because of the hygroscopic cement and formation of calcium hydroxide. Calcination substantially increases the friability of the cement-bound pellets and limestone. Both materials approached complete degradation after 24 h during the 500?C tests. However, the silica coated pellets were found to have high attrition resistance in dry air at 500?C. After 24 hours of operation, compared to the coated sorbents, the lime mean diameter reduction and Air Jet Index were greater by factors of 2.4 and 3.4, respectively. The 1 ?m thick and 5 ?m thick-coating pellets experienced nearly the same mean reduction in diameter and production of fines, a key finding for future studies of the economic viability of the coating process. The tested materials experienced significant fragmentation, resulting in deviations from fines production models from the literature. The experimental results also displayed large variance in fines production rates during replicate trials. A novel model for mean diameter reduction rate is proposed, based on particle momentum.  iii  Preface  All of the work presented in this thesis was completed by the author under the supervision of Dr. Naoko Ellis, Dr. John Grace, and Dr. C. Jim Lim in the Department of Chemical and Biological Engineering at the University of British Columbia. Experimental material was directly supplied by Dr. Vasilije Manovic at CANMET, and Zhenkun Sun at Laval University.  Results from Section 3.1 were presented on October 15th, 2012 at the 62nd Canadian Chemical Engineering Conference in Vancouver, BC. Results from Sections 3.1 through 3.3 were presented during the poster sessions on June 3-5th at the 2013 Carbon Management Canada Annual Conference in Calgary, AB.   iv  Table of Contents  Abstract ........................................................................................................................................... ii Preface............................................................................................................................................ iii List of Tables .................................................................................................................................. v List of Figures ................................................................................................................................ vi List of Symbols and Abbreviations.............................................................................................. viii Acknowledgements ........................................................................................................................ ix Chapter 1 ? Introduction ................................................................................................................. 1 1.1 Primary Modes of Attrition .............................................................................................. 3 1.2 Material and Process Factors .......................................................................................... 3 1.3 Attrition Test Methods ...................................................................................................... 7 1.4 Attrition Results Analysis and Modeling .......................................................................... 9 1.5 Sorbent Candidates and Project Justification ................................................................ 13 Chapter 2 ? Experimental Methodology ....................................................................................... 17 2.1 ASTM Air-Jet Apparatus Test Method and Project Deviations ..................................... 17 2.2 ASTM Standard Disadvantages ..................................................................................... 21 2.3 Suggestions for ASTM Standard Improvement .............................................................. 22 2.4 Experimental Conditions ................................................................................................ 27 Chapter 3 ? Attrition Test Results ................................................................................................ 33 3.1  Uncalcined Jet Tests and Fluidized Bed Attrition Tests ................................................ 34 3.2  Calcined, Ambient Temperature Jet Attrition Tests ....................................................... 40 3.3  Calcined, High Temperature Jet Attrition Tests ............................................................ 48 3.4  Test Variance.................................................................................................................. 56 Chapter 4 ? Attrition Modeling..................................................................................................... 59 4.1  Gwyn (1969) Time vs. Fines Production Model ............................................................ 59 4.2  Jet Attrition and Bed Attrition Fines Production Models .............................................. 61 4.3  Population Momentum Mean Particle Diameter Reduction Rate Model ...................... 63 Chapter 5 ? Conclusions ............................................................................................................... 69 REFERENCES ............................................................................................................................. 71 APPENDIX A ? Introduction ....................................................................................................... 76 APPENDIX B ? Experimental Methodology ............................................................................... 78 APPENDIX C ? Attrition Test Results ......................................................................................... 84 APPENDIX D ? Attrition Modeling........................................................................................... 108  v  List of Tables  Table 2.1:  Proposed Standard vs. the ASTM D5757 Standard ................................................. 24 Table 2.2:  Attrition Test Conditions ......................................................................................... 28 Table 2.3:  Experimental Particle Densities and Minimum Fluidization Velocities .................. 32 Table 3.1:  Experimental Standard Errors .................................................................................. 58    vi  List of Figures  Figure 3.1:  Total mass mean diameter reduction vs. time for uncalcined jet and fluidized bed attrition tests for up to 24 h at 20?C.. .................................................................35 Figure 3.2:  Air Jet Index vs. time for uncalcined jet tests and fluidized bed attrition tests for up to 24 h at 20?C.. ...................................................................................................35 Figure 3.3:  SEM images of original uncalcined CANMET pellets.............................................36 Figure 3.4:  SEM images of uncalcined CANMET pellets after 5 h jet attrition in humid air at 20?C ......................................................................................................................36 Figure 3.5:  SEM images of uncalcined CANMET pellets after 5 h fluidized bed attrition in humid air at 20?C ......................................................................................................37 Figure 3.6:  SEM images of uncalcined CANMET pellets after 12 h fluidized bed attrition in humid air at 20?C ......................................................................................................37 Figure 3.7:  SEM images of the calcined CANMET pellets (500 ?m) after 0 h, 5 h, 12 h, and   24 h fluidized bed attrition in dry air. .......................................................................38 Figure 3.8:  Total mass mean diameter reduction vs. time for the calcined particle jet tests for up to 36 h at 20?C ................................................................................................40 Figure 3.9:  Air Jet Index vs. time for the calcined particle jet tests for up to 36 h at 20?C ........41 Figure 3.10:  SEM images of the lime after 0 h, 5 h, 12 h, and 36 h jet attrition in humid air at 20?C ..........................................................................................................................42 Figure 3.11:  SEM images of the calcined CANMET pellets after 0 h, 5 h, 12 h, and 36 h jet attrition in humid air at 20?C ....................................................................................43 Figure 3.12:  SEM images of the lime after 0 h, 5 h, 12 h, and 36 h jet attrition in dry air at 20?C ..........................................................................................................................44 Figure 3.13:  SEM images of the calcined CANMET pellets after 0 h, 5 h, 12 h, and 36 h jet attrition in dry air at 20?C .........................................................................................45 Figure 3.14:  SEM images of the half-calcined lime after 0 h, 5 h, 12 h, and 36 h jet attrition in dry air at 20?C. ......................................................................................................46 Figure 3.15:  SEM images of the calcined CANMET pellets (500 ?m) after 0 h, 5 h, 12 h, and   36 h jet attrition in dry air at 20?C. ...........................................................................47 Figure 3.16:  Total mass mean diameter reduction vs. time for calcined particle jet tests for up to 24 h at 500?C. .......................................................................................................48 Figure 3.17:  Air Jet Index vs. time for calcined particle jet tests for up to 24 h at 500?C. ...........49 Figure 3.18:  SEM images of the calcined CANMET pellets after 0 h, 5 h, 12 h, and 24 h jet attrition in dry air at 500?C .......................................................................................50 Figure 3.19: SEM images of the lime after 0 h, 5 h , 12 h, and 24 h jet attrition in dry air at 500?C. .......................................................................................................................51 Figure 3.20:  SEM images of the calcined pellets with a 5 ?m silica coating after 0 h, 5 h,    12 h, and 24 h jet attrition in dry air at 500?C ..........................................................53 Figure 3.21:  SEM images of the calcined  pellets with a 1 ?m silica coating after 0 h, 5 h,   12 h, and 24 h jet attrition in dry air at 500?C ..........................................................54 vii  Figure 3.22:  SEM images of coated pellet surfaces, for 1 ?m coated pellets after pre-calcination, post-calcination, and 5 ?m coated pellets pre-calcination, and post-calcination .................................................................................................................55 Figure 3.23:  SEM images of the blank calcined CANMET pellets after 0 h and 5 h jet attrition in dry air at 500?C .......................................................................................56 Figure 3.24:  Total mass mean diameter reduction vs. time for the replicate lime jet tests for up to 24 h at 20?C .....................................................................................................57 Figure 3.25:  Air Jet Index vs. time for the replicate lime jet tests for up to 24 h at 20?C.............57 Figure 4.1:  Modified fines production model..............................................................................61 Figure 4.2:  Plot of mass mean diameter reduction rate vs. population momentum fraction        (n = 2) for 20?C trials ................................................................................................66 Figure 4.3:  Plot of mean diameter reduction rate vs. population momentum fraction (n = 2) for the 500?C trials ....................................................................................................67    viii  List of Symbols and Abbreviations  A column cross-sectional area, m2 a crack length, m aor orifice area, m2 ap,i projected area of particle size i, m2 AJI Air Jet Index, % Ar Archimedes Number B(x,y) breakage function b Gwyn (1969) attrition constant C1, C2 minimum fluidization velocity correlation constants Cb bed/bubble attrition rate constant, m-2 Ccyc cyclone attrition rate constant, s2/m2 Cjet jet attrition rate constant, s2/m3 CR/K/B Rittinger/Kick/Bond comminution constants DB0 initial bubble diameter, m DSC settling chamber diameter, m dmm mass mean particle diameter, m dpb bed particle diameter, m dp particle diameter, m dpc cyclone particle diameter, m dpi particle diameter of size i, m       total mass mean diameter reduction rate at time t, %/h dsv Sauter mean particle diameter, m E Young?s modulus, Pa ER/K/B comminution energy, kJ/kg g gravitational constant, m2/s Hb bed height, m Ka Gwyn (1969) attrition constant, s-b km population momentum model attrition rate constant, -%/h Ljet vertical jet penetration length, m M total mass, kg mb bed mass, kg mbed, initial initial bed mass, kg mfines mass of fines, kg m fines,  ed bed attrition fines production rate, kg/s   m fines, cyc cyclone attrition fines production rate, kg/s m fines,  et jet attrition fines production rate, kg/s mi mass of particles of size i, kg N number of particle bin sizes n momentum model order ni number of particles of size i nor number of orifices Re Reynolds Number S(x) selection function TDH transport disengaging height, m t time, s or h U superficial gas velocity, m/s ucyc, inlet cyclone inlet velocity, m/s Umf minimum fluidization velocity, m/s uor orifice velocity, m/s u? terminal velocity, m/s x breakage function particle size x, m xi mass fraction of particle size i y reduced particle size y, m ?  gas viscosity, kg/m?s  ?c solids loading at cyclone inlet ?f fluid (gas) density, kg/m3 ?g, bed bed gas density, kg/m3 ?p particle density, kg/m3 ?cr critical stress, Pa ? particle sphericity   ASTM American Society for Testing and Materials BET Brunauer?Emmett?Teller CCS carbon capture and storage CMC Carbon Management Canada F.B fluidized bed FCC fluid catalytic cracking MEA monoethanolamine PSD particle size distribution SEM scanning electron microscope TEOS tetraethyl orthosilicate TGA thermogravimetric analysis VPO vanadium phosphorus oxide   ix  Acknowledgements  I would like to thank the following people and groups:  Dr. Naoko Ellis, Dr. John Grace, and Dr. C. Jim Lim for acting as my thesis supervisors,  Dr. Bruce Bowen and Dr. Norman Epstein for agreeing to be on my thesis committee,  Carbon Management Canada and NSERC for project funding and resources,  Dr. Arturo Macchi at the University of Ottawa, Dr. Vasilije Manovic and Dr. Edward (Ben) Anthony at CANMET, and Zhenkun Sun at Laval University for their sorbent samples and advice,  and finally, the members of the UBC Fluidization Research Centre and gasification research group for their unending support.  1 Chapter 1 ? Introduction  As the global demand for energy increases over the next century, it will become increasingly important to develop carbon emission reduction options to combat the release of greenhouse gases from the combustion of carbonaceous fuels. Fossil fuel-based power plants are a significant source of carbon dioxide emissions, with coal-based plants accounting for more than 40% of all anthropogenic carbon dioxide emissions (Fan, 2010). There are three approaches to reducing the rate of anthropogenic carbon dioxide emissions into the atmosphere: 1) reduce the rate of energy usage; 2) reduce the usage of carbon-based energy sources; and 3) provide carbon dioxide sinks or sequestration (Yang et al., 2008). This thesis project focuses on the third option, and aims to improve the cost efficiency of capturing CO2 from combustion or gasification flue gases in coal or biomass-based power plants. Carbon capture and storage (CCS) is comprised of three steps: CO2-capture at the source of emission, CO2 compression and transportation, and CO2 storage in geological formations, or possibly ocean storage or mineralization (Pires et al., 2011). The CO2-capture step represents the highest cost percentage and greatest economic challenge of the three steps, and is of special interest to those in the field of chemical engineering.  Currently, the most common method of CO2-capture in industry is via monoethanolamine (MEA) gas scrubbers. However, while this technology is mature and well-understood, the energy efficiency of the process is low and difficult to improve due to inherent limitations. Most notably, the process requires a significant amount of steam to separate the captured CO2 and recover the used MEA solution. The resultant parasitic energy loss greatly decreases the net power output from a power plant, by as much as 42% (Fan, 2010). Further taking into account the necessary environmentally-toxic chemicals, low temperatures and low pressures required for the process, MEA gas scrubbing quickly becomes an unfavourable process for widespread implementation to combat large-scale CO2 emissions.  An alternative CCS method for power plants is the use of high-temperature, dry sorbent technologies, based on sorbents derived from materials such as limestone, dolomite, and lithium silicates. The carbonation-calcination chemical looping system allows for the sorbent to react with carbon dioxide in-situ in the combustor or gasifier to form a carbonate, e.g. by reacting lime (calcium oxide) with carbon dioxide to form limestone (calcium carbonate). The metal-carbonate 2 particles are then separated and calcined at a higher temperature to remove the carbon dioxide and regenerate the original sorbent (Stanmore & Gilot, 2005), which is then returned to the combustor or gasifier for the next carbonation/calcination cycle. As long as the reactivity of the sorbent remains high over successive cycles, the efficiency of the process can be much greater than that of the MEA gas scrubbing system (Fan, 2010), although the exact parasitic energy loss is dependent on the specific sorbent chemistry.  Unfortunately, a significant issue with using such particulate sorbents in fluidized bed reactor systems is that the particles may experience substantial attrition, or mechanical breakage and degradation, over successive carbonation-calcination cycles, resulting in decreased performance, as well as sorbent loss (Lupi??ez et al., 2011). As the particles decrease in size and fines are generated, a greater fraction of the particles will be entrained in the gas stream, causing filtration/collection system issues and ultimately a gradual loss of material from the system. In addition to reducing reactor performance, replacement of the lost sorbent adds an additional economic burden onto the process, in turn increasing the cost of dry sorbent CO2-capture operations.  In order to better understand the attrition phenomena experienced by CO2-capture sorbents in fluidized combustion or gasification systems, this thesis project has studied the extent and causes of sorbent attrition due to particle impaction, surface abrasion, and thermal effects. This study is part of a national Carbon Management Canada (CMC) network project, involving several institutions, working on determining the optimal sorbent candidate for industrial CO2-capture systems  ased on the various materials? properties such as reaction kinetics and long-term performance over multiple calcination/carbonation cycles. The sorbent candidates include crushed Cadomin limestone (from Cadomin, Alberta) and novel sorbents, such as cement-bound or silica-coated lime pellets. By considering both the reaction performances and attrition-resistance characteristics of the sorbent candidates, the overall CMC project aims to select the best sorbent candidate to balance both its CO2-capture performance and the economics of its production and long term use.      3 1.1 Primary Modes of Attrition   The two primary modes of attrition are fragmentation and abrasion. Fragmentation or particle fracturing results in the original particles, often referred to as the ?mother particles?, splitting into a small number of similarly-sized particles, normally after direct impact with each other or a hard surface. Abrasion describes the rough edges or asperities of the mother particles being rounded off, producing mother particles with smoother surfaces and a significant number of very fine particles. If the particle size distribution (PSD) curve changes from a unimodal distribution to a bimodal distribution, this likely indicates that abrasion was the primary mechanism. If the distribution shows an appreciable downward-size shift, yet remains unimodal, this implies significant particle fracturing. However, most fluidized processes experience both attrition mechanisms (Pis et al., 1991), and it is important to differentiate between these two modes of attrition because they cause different issues in a fluidized system (Werther & Reppenhaggen, 2003). Abrasion is generally more responsible for fines production and economic losses from material elutriation, while fragmentation causes greater changes in the mean bed particle size, resulting in performance and control issues. By studying how a particular material attrits under various process conditions, the extent of each issue can be quantified for a particular process.    1.2 Material and Process Factors  There are numerous factors that affect the rate of attrition in fluidized processes, but they can be categorized as either being a material factor or a process factor. The simplest way to reduce the rate of attrition in a process is to use a less attrition-prone material. Of course, this is rarely a simple course of action, as the process is likely tailored for a specific material, which could be a particular fuel, catalyst, or sorbent. Nevertheless, when designing a new process or modifying an existing one, it is important to identify what material properties cause a particular material to attrit rapidly in order to minimize the short and long term effects on the process.   The most common material property used to estimate the relative rate of attrition between various materials is their hardness, or the material?s resistance to scratching, denting, or other forms of permanent deformation, depending on the specific hardness index (Bemrose & Bridgwater, 1987). The Mohs hardness is most often cited, which ranges from a value of 1 for talc to 10 for diamond. The Mohs hardness scale is an ordinal scale, with ten different minerals 4 simply ranked in order of hardness and accordingly assigned numbers from 1 to 10. This ranking approach is nearly useless on a quantitative basis, but simple for qualitative purposes. For this thesis, the most important hardness values are those of limestone, with a Mohs hardness of 3, and silica, with a Mohs hardness of 7, which correlates to a relative Turner-sclerometer hardness 11 times that of limestone (Mindat, 2013).   It is generally true that the harder a material, the more resistant it is to attrition. This is most clearly demonstrated in attrition tests involving limestone, where the rate of attrition depends on the degree of calcination and/or sulfation (Chen et al., 2008). As calcium carbonate is converted into calcium oxide, the rate of attrition increases because calcium oxide has a Mohs hardness of less than 3 (Lee et al., 1993). However, as the calcium oxide absorbs sulfur dioxide to form a calcium sulfate shell, the particle?s resistance to attrition increases, with calcium sulfate?s Mohs hardness being 3.5. The 3-species limestone system clearly demonstrates the complexity of attrition in industrial systems, as not only is the hardness of a material a function of temperature (Bemrose & Bridgwater, 1987), but the relative amounts of each species also greatly affect rates of attrition (Montagnaro et al., 2010). Furthermore, attrition issues can be exacerbated in fluidized systems with two particle species of significantly different hardnesses, as one will abrade the other, or when the particles have frequent contact with hard column walls, reactor internals or cyclone walls (Werther & Reppenhaggen, 2003).   Another material property often cited is Young?s modulus, which is most commonly used to describe the stress-strain relationship of elastic deformation in metals. However, Young?s modulus has also been used in attrition studies to quantify the minimum stress required to propagate an existing crack or flaw in a particle. Known as the Griffith relationship (Shipway & Hutchings, 1993), the minimum or critical stress obeys a relationship:   ?        (1.1)  where E is the Young?s modulus and a is the crack length. Therefore, the greater the Young?s modulus, the greater the stress required for particle fracture. However, even more significant is the crack length term. As particles continually decrease in size, so does the maximum possible 5 crack length. Therefore, there exists a minimum particle size at which the critical stress required for fracturing can no longer be obtained, and the particles cease to fragment further. A similar theory (Ray et al., 1987) describes a ?natural grain size? for amorphous or composite materials such as limestone, as they tend to break along material grain boundaries or pore structures, unlike crystalline materials. These phenomena are two important aspects of the effect of particle size on the rate of attrition. The additional effect of the change in momentum based upon particle velocity and size distribution has also been studied in this thesis, and is discussed in Section 4.3.   The particle size distribution also influences the available surface area for attrition, with an increase in the number of small particles resulting in more surface area to abrade. However, since the energy used for attrition comes from the kinetic energy of the influent gas flow, as the number of particles and surface area increase, less energy per unit area is available for attrition (Werther & Reppenhaggen, 2003). Therefore, the relationship between surface area and attrition is not well defined. The effect of porosity is clearer, with the rate of attrition increasing with porosity as a result of less internal support structures and an increase in internal flaw or crack length (Litster & Ennis, 2004). Attrition is also affected by pore sintering, either from pretreatment of the material or from the process itself. At high temperatures, sintering causes both the porosity and surface area of the particles to decrease (Stanmore & Gilot, 2005), resulting in a reduction in the rate of attrition. However, the reduction in surface area greatly reduces both the reactivity and the carrying capacity of lime-based sorbents, which is most commonly experienced through repeated calcination/carbonation cycling (Abanades & Alvarez, 2003). The effects of sintering can be studied through the use of both thermogravimetric analysis (TGA) and Brunauer-Emmett-Teller (BET) surface area tests, but it must be noted that this was not within the scope of this particular project.  Particle shape and particle surface morphology both affect the rate of attrition, and are themselves affected by it. As expected, the rougher and more jagged the surface of the particles, the greater the rate of abrasion and fines production. However, as time progresses, particles become smoother and the weakest sections of the particles will have been removed, greatly reducing the rate of abrasion. Conversely, as the particles undergo fragmentation, fresh rough surfaces are produced, which in turn are abraded to be smooth once again. This abrasion-6 fragmentation cycle is best observed using images of the particle surface from scanning electron microscopy (SEM), as utilized in this project.   Ultimately, the degree of attrition depends on the residence time of the particles in the process, with greater residence time leading to a greater extent of attrition. However, the attrition rate is generally a non-linear function of time, as will be discussed in Section 1.4. In addition to the inherent material properties of the particles, the process environment greatly affects the rate of attrition. The most important factor is the velocity of the particles, which, in gas-fluidization, is directly a function of the velocity of the influent gas stream. There are two different gas velocities to consider: the jet velocity of the gas flowing through the distributor, and the superficial gas velocity inside and above the fluidized bed, both of which are discussed in Section 1.4. As expected, the velocity of the gas stream is directly related to the rate of attrition. Furthermore, studies have shown that there exists a ?threshold velocity? for a specific set of process conditions, whereby the particles are imbued with enough kinetic energy to fracture one another instead of simply abrading the particle surfaces (Bemrose & Bridgwater, 1987).   Temperature affects the rate and extent of attrition for several reasons. As the temperature increases, there is a decrease in gas density and increase in gas viscosity, which generally results in increased threshold velocity (Chen et al., 2007), decreased particle momentum and decreased attrition rate. It is important to note that for similar reasons, high pressures are expected to increase the rate of attrition because of increased gas density (Werther & Reppenhaggen, 2003). Temperature also affects several material properties, most notably brittleness or resistance to cracking, with a reduction in temperature increasing brittleness and decreasing ductility (Schaller, 2000). On the other hand, elevated temperatures may soften the material, which can result in agglomeration and eventual melting of the particle.   Since this project involves the use of limestone, the clearest effect of temperature is on calcination, i.e. the conversion of limestone (calcium carbonate) into lime (calcium oxide) and carbon dioxide. This reversible reaction is dependent on both temperature and the partial pressure of carbon dioxide, as shown in Figure A1 in Appendix A (Stanmore & Gilot, 2005). As expected, the conversion of limestone into lime results in drastically different material properties and to increased friability, as previously discussed. However, the rapid release of carbon dioxide 7 during calcination causes internal pressure stresses which can cause thermal decrepitation (Lee et al., 1993). Since carbonation/calcination cycling can weaken or split the material and cause a decrease in particle size, it is important to consider in an industrial CO2-capture system. To perform a proper study of thermal decrepitation, a researcher would perform multiple carbonation/calcination cycles on a sorbent sample in a TGA, and use laser diffraction analysis to measure the change in particle size distribution. Unfortunately, this was outside the scope of this particular project. On the other hand, since the low and high temperature friability of different sorbents and their calcined forms was the primary focus of this project, the effects of temperature and chemical change were still extensively studied.  1.3 Attrition Test Methods  There are two main approaches for assessing the attrition of particulate materials (Yang, 2003). One approach is to subject the particles to a singular stress type: compression, impaction, or shear stress. Knowing the relative occurrences or significances of the three stresses in a process, one could then assess the material, based on its expected resistance to attrition within an expected range of operating conditions. However, it is difficult to accurately weigh the compression/impaction/shearing stresses in a process. Therefore, the second approach is to subject the particles to an environment similar to that of the process itself. While this likely results in the material being subjected to all three stress types, it provides a more realistic recreation of the intended process environment. However, since the tests then become process-specific, the attrition test results are significantly equipment-specific, and the test cannot be considered to be standardized nor the results universal.  For the first approach, the most common test method is to use a particle impactor, where a singular stress, impaction, is isolated by launching a single particle at a target plate that is at 80? to the particles? direction of motion. Alternatively, this target can be replaced with a 15? narrow cone with the aim of causing abrasion of the surface of the particles. A diagram of a UBC custom-made impactor apparatus is provided in Figure A2 of Appendix A. The primary advantage of the impactor test, other than its ability to isolate fracturing from abrasion, is that the tests are relatively simple and fast. The duration of the tests simply becomes a function of the number of repeated impacts for each sample. Also, the test is moderately reproducible at other institutions, as the simplicity of the test minimizes equipment dependency. Previous uses of impactor systems include studies on threshold velocities as a function of temperature (Chen et 8 al., 2007), sulfation effects on the rate of attrition (Lupi??ez et al., 2011), and surface morphology effects on attrition products (Cleaver et al., 1993).   However, since impactor tests isolate the fracturing or abrasion mechanisms of attrition, the results are less realistic than a test that attempts to simulate the actual environment to which particles in fluidized systems will be exposed. In fluidized systems, three primary sources of attrition have been identified: a) jet attrition from the gas distributor; b) bubbling or bed attrition; and c) cyclone attrition (Werther & Reppenhaggen, 2003). These are further discussed in Section 1.4, but for now it is important to note that most attrition tests attempt to simulate one or more of these sources of attrition. The most popular attrition test method is based on a jet test design introduced by Gwyn (1969), which was later modified and adopted as an ASTM standard (ASTM Standard D5757, 2006). Sonic jets are used to rapidly attrit a sample, followed by collection of the fines, which are weighed against the original sample mass, allowing for an easy and rapid test which produces a simple attrition index. The tests performed in this thesis are based upon the ASTM D5757 method, with alterations made to study the long term effects of temperature, velocity, humidity, and particle size. This pro ect?s experimental method is outlined in detail in Chapter 2. The ASTM D5757 method has been used to test the relative attritability of a wide variety of materials, including limestone (Xiao et al., 2012), steam methane reforming catalysts (Johnsen et al, 2006), Fischer-Tropsch catalysts (Zhao et al., 1999), and copper-based desulfurization sorbents (Slimane & Abbasian, 2000).   Bed attrition tests are best used to simulate entire fluidized process environments, and work well with chemically reactive systems (Montagnaro et al., 2010). The attrition properties are often tested in bubbling or circulating beds, depending on the intended application. Unfortunately, due to the varying experimental setups and operating conditions of each study, it is difficult to compare quantitatively bed attrition results from literature. A diagram of the UBC circulating fluidized bed attrition system appears in Figure A3 of Appendix A.  It is important to operate bed attrition tests with a low distributor orifice velocity to reduce the effects of jet attrition (Werther & Xi, 1993). In this project, a fluidizing plate with large orifices was used in the ASTM air jet apparatus for this reason, with some success. Since the superficial gas velocity is much less in the bed than jet velocities or cyclone inlet velocities, the time required to produce significant fines or change in bed PSD is considerable, making such a test method time-consuming. 9 Nevertheless, numerous bed attrition studies have been performed, especially in combination with an impactor (Lupi??ez et al., 2011) or cyclone (Chen et al., 2008). However, cyclone tests are less frequently performed, likely because specific equipment must be constructed to operate them if a full circulating system is not already available. The Grace-Davison jet cup standard was designed to reduce such issues (Zhao et al., 2000). It is operated by loading a 5 g sample in a small cylindrical cup, where the particles are circulated at high velocities to replicate a cyclone environment. Primary advantages of such a test are the low mass of material required and the rapid rate of attrition. Recently, a larger and conical shaped cup has been proposed as a replacement standard to eliminate stagnant regions in the previous design (Cocco et al., 2010). Unfortunately, due to time and scope limitations, cyclone attrition was not able to be studied in this project, although it may well be an important source of attrition in dual-bed CO2-capture systems.  1.4 Attrition Results Analysis and Modeling  Historically, numerous methods have been used to quantify the results of attrition tests. The simplest method is to use single-number metrics, or attrition indices, that measure the relative weight of fines produced after a specified amount of time. A well-known index in the comminution field is the Hardgrove Grindability Index, which measures the proportion of sub-74 ?m material produced when a sample is su  ected to 60 revolutions in a specific ball-ring pulverizer (Werther & Reppenhaggen, 2003). A Work Index or Bond Index can also be determined through similar comminution tests (Bemrose & Bridgwater, 1987). However, attrition indices determined through fluidization-related tests are best able to estimate attrition in fluidized processes. The ASTM D5757 jet plate test used in this project employs an Air Jet Index (AJI), equaling the mass of fines produced after 5 hours of operation as a fraction of the mass of original bed material (ASTM Standard D5757, 2006). The Grace-Davison jet cup test has its own Davison Index based upon the fractional weight of fines, although the test duration is only 1 hour (Zhao et al., 1999).  The most historically important attrition test metric is that of Gwyn (1969), who studied the relationship between attrition and time by measuring the production of sub-40 ?m fines in a three-jet column, which eventually became the inspiration for the ASTM D5757 standard, as 10 previously mentioned. Gwyn reported that the mass of fines produced is a non-linear function of time, in the form of:  mfinesm ed, initial  at  (1.2)  Unfortunately, both attrition constants Ka and b are found to depend not only on the material used and the process conditions, but also on the initial size distribution of the particles, reducing their applicability in different systems. However, b is found to be less than 1, meaning that the fines production rate decreases with time. This model was used to illustrate the concept of ?initial attrition?, where the originally rough surfaces of fresh particles are quickly abraded, resulting in an initial high rate of fines production. As the particle surfaces become smooth, the rate of fines production significantly decreases. However, this assumes that no particle fragmentation is occurring to produce new rough surfaces, an assumption whose validity depends on the particle velocities remaining below the threshold velocity.   Nevertheless, Gwyn?s concept of an attrition constant K, similar to a reaction rate constant, was further developed in the formulation of attrition rate equations to describe the three primary sources of attrition:  Jet attrition (Werther & Xi, 1993):   m fines,  et   C etdp nor?fdor2uor  (1.3)  Bed/bubble attrition (Ray et al., 1987):   m fines,  ed   C dp m (u umf) (1.4)  Cyclone attrition (Reppenhagen & Werther, 2000):   m fines, cyc   Ccycdpcucyc,inlet2 (  ) (1.5)  For this project, the most important source is jet attrition, as most tests were performed using a 3-orifice sonic jet plate, as described in Chapter 2. The most important aspect of Equation 1.3 is that the rate of attrition is proportional to the orifice gas velocity, raised to the power of 3. This is discussed further in Chapter 2, but it suggests that the rate of jet attrition can change drastically 11 with small changes in gas velocity. Equation 1.4 indicates that the attrition rate is only proportional to the gas velocity by a power of 1, which explains why, at high velocities, bed attrition is usually a much less important source of attrition than either jet or cyclone attrition, which has a second order velocity dependence (Werther & Reppenhaggen, 2003). Unfortunately, Equation 1.3 assumes that abrasion is the only mode of attrition, which was found to be invalid for this project. Nevertheless, the three equations provide a simple method to compare attrition test dependencies and estimate process condition effects from the relative rates of fines production of different materials.  A second approach for assessing attrition test results is to study the change in the mean particle diameter from PSD analysis. This is a much less common approach, due to the significantly increased work required to measure PSD over a single weighing of the amount of fines produced. Furthermore, attempts to model and predict the rate of change in d p have either produced complex relationships with numerous required empirical constants, or simple models with limited experimental accuracy.  Comminution theory provides three models that have been used with limited success to describe attrition phenomena (Rhodes, 2008). The Rittinger model states that the energy required to reduce a particle?s size is proportional to the amount of surface area created:   dEd(dp)   C 1dp2 (1.6)  However, the energy estimated has been found to be 200 to 300 times too small. The Kick model suggests that the energy required is proportional to the volume ratio between the mother and daughter particles:   dEd(dp)   C 1dp (1.7)  However, this model assumes that the energy requirement to reduce the size of large particles is the same as for small particles, which is conceptually invalid. The Bond model is based upon empirical tests to determine the Work and Bond indices, as previously discussed:  12  dEd(dp)   C 1dp  2 (1.8)  All three of these models have been found to only give reasonable results for limited size ranges. They generally greatly underestimate the energy required for attrition processes, especially when large threshold velocities are required (Bemrose & Bridgwater, 1987). However, another comminution model uses two different functions to estimate the change in PSD from a crushing process: a selection function S(x) that lists the probability of a particle of size x breaking, and a breakage function B(x,y), which identifies what fraction of the particle?s mass reduces to particles of size y (Werther & Reppenhaggen, 2003). The selection function is a vector, while the breakage function is a matrix. The length and dimensions of each depend on the number of particle bin sizes considered. Unfortunately, each variable contained in the functions must be determined through extensive empirical testing, and is only valid for a specific material, similar process conditions, and identical comminution equipment, making it ineffective for attrition models for fluidized processes.  Numerous models based upon particle population balances have been proposed, each with their own set of empirical correlations and constants (Ray et al., 1987). Unfortunately, numerous simplifying assumptions are often required, such as assuming that abrasion is the only mode of attrition, that all particles attrit equally, or that internal size bin transfer constants are all equal to 1 (Chen, 2011). On the other hand, some researchers use several empirical constants to model their results, making it difficult to compare attrition experiments from the literature (Xiao et al., 2012). Therefore, Section 4.3 of this thesis attempts to find a balance between the two approaches, combining a moderately simple momentum-based population model with a single empirical constant to compare the performance of several materials under different process conditions.  Lastly, it is important to acknowledge the simulation work performed at the Hamburg University of Technology (Thon et al., 2011). Industrial fines generation data for vanadium phosphorous oxide (VPO) catalyst in DuPont?s maleic anhydride production process was combined with a flow sheet model that uses Equations 1.3 through 1.5. From this simulation, it was estimated that only 16% of attrition was due to the cyclones, and that bubble attrition was a larger concern than grid jet attrition. While a similar simulation would be ideal to estimate the rate and sources of 13 attrition in an industrial CO2-capture system, the work required for such an endeavour is unfortunately outside the scope of this thesis.  1.5 Sorbent Candidates and Project Justification  As previously discussed, this study is part of a collaborative Carbon Management Canada research network project whose goal is to propose an optimal sorbent candidate for industrial CO2-capture systems. The three current sorbent candidates are crushed limestone, lime-based pellets containing a calcium aluminate cement binder, and the same pellets without the cement binder, but instead with a mesoporous silica shell. As discussed throughout this chapter, the kinetics and material characteristics of limestone have been extensively tested, and the chemistry is well understood. At UBC alone, attrition tests have been performed on limestone using a particle impactor (Chen et al., 2011), a circulating fluidized bed (Chen et al., 2008), and in the same ASTM unit as used in this study (Xiao et al., 2012). However, the ASTM tests were not performed at elevated temperatures, and the limestone was uncalcined. Furthermore, the effect of gas humidity, which is a significant consideration for steam gasification systems, was not investigated. A fluidization distributor plate was also used in this study in an attempt to estimate the relative importance of bed attrition versus jet attrition. However, the most important advantage of this study over other attrition studies is that multiple materials were tested under the same operating conditions in the same experimental unit. While the ASTM standard attempts to produce consistent results across all studies that use it, modifications are often made by researchers to tailor the tests for their own purposes, reducing the effectiveness of using a universal standard. While modifications were also used in this study, those modifications were the same for all materials, allowing for accurate comparison of attrition performances.   Currently, the cost of crushed limestone ranges from $8-12/t (USGS, 2013), resulting in an inexpensive sorbent candidate for industrial CO2-capture systems. However, the most significant advantage of limestone is that it is an environmentally benign chemical, especially relative to the chemicals used in liquid CO2-capture systems. Therefore, lime-based CO2-capture systems are potentially able to reduce the effects of global climate change without creating additional issues for the local or regional ecosystem. While raw limestone likely has a clear economic advantage, the claimed benefits of the specially prepared sorbents supplied by other institutions have made them possible candidates for industrial use. The second CMC sorbent candidate, proposed jointly 14 by CANMET and the University of Ottawa, is a lime-based pellet with a commercial calcium aluminate cement binder. Calcium aluminate cements were chosen both to minimize binder costs and to prevent sintering effects found with other binders (Manovic & Anthony, 2009a). The effects of sintering are reduced due to the production of mayenite (Ca12Al14O33) in the pellet, which increases pore sizes, thereby counteracting pore size reduction from sintering (Wu, et al., 2012). The pellets are produced by mixing calcined lime powder (< 30 ?m) with the calcium aluminate cement (71% Al2O3, 28% CaO, 1% SiO2) in a 9:1 ratio in a mechanical pelletizer. The mixture is then sprayed with water, with the amount of water and droplet size controlled to determine the final size of the pellets. It is very important to note that the addition of water to the pellets converts the calcium oxide to calcium hydroxide, which sets the cement but results in a material of lower Mohs hardness. However, the cement should help keep the pellet together during fluidization, reducing the rate of fines production. Furthermore, the pellets have a higher BET surface area and porosity than raw limestone, resulting in improved long-term carbonation/calcination cycling performance (Wu et al., 2012).  Preliminary attrition tests on the CANMET pellets in a bubbling fluidized bed indicated that the pellets have a higher attrition resistance than raw limestone, especially at temperatures nearing 800?C (Wu et al., 2012). This finding was also verified by another group using pellets with 30% aluminate cement binder (Chen et al., 2012). However, since these attrition tests were only performed over a 2 hour period, initial attrition effects would likely dominate the results. Furthermore, the researchers admitted that their experimental setup reduced their ability to collect generated fines sticking to the inner surfaces of the unit. These significant drawbacks produced an opportunity for our project to investigate the long term attrition characteristics of the pellets in direct comparison to raw limestone.  By CANMET?s own estimates (Manovic & Anthony, 2009b), the industrial cost of the pellets would be about $150-200/t, based upon a limestone price of $10/t and a calcium aluminate cement price of $1200-1300/t. Therefore, the pellets would need to be about 15 to 20 times as attrition resistant to justify their use on a purely economic basis from the standpoint of fines production and material loss. However, the kinetic and cyclic performance of the pellets would also need to be considered, which may increase their value to industry. Nevertheless, the pellets have to be considerably more attractive than limestone to replace simple limestone.  15  The third sorbent candidate was prepared by Laval University. It begins with the same lime pellets from CANMET, but without the calcium aluminate cement. Instead, the pellets have either a 1 ?m or 5 ?m-thick mesoporous outer silica shell. The primary goal of the shell is to prevent sintering and the reduction in porosity and surface area from high-temperature carbonation/calcination cycling (Sarshar et al., 2012). It is also expected that since the relative hardness of silica is 11 times that of calcium carbonate, the shelled material should attrit at a much lower rate. Initial testing by the Laval research group on this material and other silica shelled materials have shown a reduction in sintering without a decrease in the rate of reactant diffusion or rate of reaction. The sintering rate reduction results in sustained performance over multiple reaction cycles. The addition of the shell increases the surface area of the particle, but decreases the average pore size and volume.   In these particles, the silica shells are deposited using a surfactant-templating process, where the CANMET pellets are added to a solution of ethanol, surfactant P-123 or F-127, and trace amounts of hydrochloric acid and water (Sun, 2013). Tetraethyl orthosilicate (TEOS) is then added to the solution, and the thickness of the silica shell is controlled by the solution TEOS concentration. The mixture is then washed with additional water and ethanol, the water and ethanol are evaporated, and the final product is calcined in air at 550?C. Clearly, the environmentally toxic chemicals required are a concern, and this will have to be considered in a final process analysis by the CMC research network. The researchers also admit that because such a process has not been performed on an industrial scale, the production cost is difficult to estimate. Nevertheless, two different core-shell samples were provided by Laval, along with a non-shelled CANMET pellet baseline sample. Prior to this thesis project, attrition tests had not yet been performed on the silica-shelled sorbents. As will be shown in Chapter 3, the results are quite promising.   In addition to the aforementioned ASTM modifications and the unique materials studied in this project, several other factors add to the importance of the results and justify this study. The attrition tests were performed for up to 36 hours, which reduced the impact of initial attrition effects experienced in the ASTM standard?s 5-hour test period. While most studies simply measure the rate of fines production, this project also analyzes the particle size distribution of the 16 bed material, resulting in a significantly improved understanding of the material and process effects on the rate and extent of attrition. Moreover, this also allowed for the development of a simple experimentally-based, yet theoretically-derived model to describe the rate of change of the mean bed diameter size, which is unique to this study. This project also heavily uses scanning electron microscopy to explain the attrition test results. In summary, the project is an extensive study using a proven test method to gain detailed insight into the attrition characteristics and performance of several sorbent candidates for use in industrial CO2-capture systems.    17 Chapter 2 ? Experimental Methodology  As discussed in Chapter 1, a wide variety of tests have been used for both catalyst and sorbent research. For this project, the chosen attrition test method is based upon the ASTM D5757 air-jet apparatus standard (ASTM Standard, 2006), with several modifications that are outlined and justified in this chapter. The testing unit was used in previous attrition studies at UBC (Xiao et al., 2011), and therefore is well understood and shown to work as intended for this project and possibly in future studies. The following sections discuss the equipment, testing procedures used and the various advantages and disadvantages of the attrition test method itself.  2.1 ASTM Air-Jet Apparatus Test Method and Project Deviations  As previously stated, for this project the attrition tests were carried out in an air-jet apparatus based upon the ASTM D5757 method (ASTM Standard, 2006). The primary advantage of using a standardized test is that in principle it allows the attrition test results to be reproducible at other institutions. The jet test is used in an effort to replicate grid-jet attrition in a fluidized bed. The UBC test unit is located in the Clean Energy Research Centre (CERC) High Head Lab (CHBE 145). A diagram of the apparatus is provided in Figure B1 in Appendix B.   The test unit consists of the following main stainless-steel components: 1) a horizontal distributor plate with three triangularly-located 0.397 mm orifices, 2) a 710 mm long, 35 mm ID attrition tube, 3) a diverging-then-converging settling chamber, and 4) a fines collection assembly containing a ceramic filter. The entire system works by feeding air through the distributor plate into the attrition tube, where the 50 g particle sample is located. As the jets mobilize and attrit the sample, the smaller particles will be entrained into the settling chamber, from where the fines are transported into the fines collector. At the end of the 5 hour test run, the weight of the fines in the collector are compared as a fraction against the original weight of the sample, giving an attrition air jet index (AJI), or percentage of fines produced:  Air  et  ndex   Mass of fines after 5 hour test runOriginal mass of sample  100  (2.1)  A summary procedure of the standard ASTM D5757 method is described below. For full details, refer to the standard itself (ASTM, 2006).  18 1) Prepare 65 g of sample to be tested.  ? The sizes of the particles should reflect the size distribution of the material to be used in the actual industrial application, but the standard suggests that the particle size range should  e from 10 to 180 ?m. The standard also classifies ?fines? as particles less than 20 ?m. 2) Open the air supply valve to feed 10.0 L/min of air at 30-40% relative humidity into the unit. The fed air should be at standard temperature (0?C), and at standard pressure (1 atm) once the air reaches the top of the settling chamber (a total pressure drop of about 30 to 80 kPa is expected). 3) Load 50 g of the sample into the unit through the particle feeding port at the top of the settling chamber. Close the particle feeding port. 4) After running the unit for 5 hours, close the air supply valve, and collect and weigh the material in the fines collection assembly. Calculate the air jet index.  From both previous tests with the unit at UBC (Xiao et al., 2011) and the work outlined in this report, several drawbacks to the ASTM standard were identified. Most notably, the ASTM standard only uses the air jet index to determine the attritability of different samples. However, the relative mass of the fines produced provides minimal insight into the attrition process experienced by the particles, neglecting the changes in the bed particle size distribution and in the particle surface morphology. Therefore, for this study, the total particle size distribution of the sample was measured after each test run either by sieve analysis or laser diffraction, and a small sample was taken to later perform scanning electron microscopy for particle surface morphology analysis.   From simply reading the standard itself, the clearest source of experimental difficulty is that the standard directs the user to ensure that both the sample and the fed air are equilibrated at 30-40% humidity in order to prevent absorption of water as the test proceeds. The standard states that the purpose of the strict humidity control is to minimize the effect of electrostatics. While this could be important for some catalysts, it is unlikely to have a significant effect on the attrition results of most sorbents to be tested. However, the sorbents tested in this project were generally calcined calcium-based materials, which react with water to form calcium hydroxide. Calcium hydroxide has a very low attrition resistance, as shown in the literature (Montagnaro et al., 2008) and also 19 by the experiments performed in this study. Furthermore, it was found in work at the University of Ottawa (Manovic, 2012) that the cement binder used in the calcium pellets was highly sensitive to moisture, further increasing the requirement to measure and control the gas humidity, although not for the standard?s intended reason. Therefore, the relative humidity of the gas was measured using an electronic humidity meter, or minimized by using extra dry tanked air.   The standard also states that the tests should be run at a standard temperature of 0?C. However, this is not practical, as it would require that a cooling system be used to reduce the air temperature by about 20?C before entering the unit. More significantly from an application perspective, the sorbents tested are intended for high-temperature processes, at hundreds of degrees Celsius. Therefore, the attritability of the sorbents at a temperature below room temperature is not very helpful or appropriate. In fact, to study the effect of temperature on the attrition of the sorbents, the air-jet apparatus is equipped with two gas preheaters: a 750 W clamshell ceramic heater and a 470 W ceramic heating tape, in addition to the primary 1000 W clamshell ceramic heater around the attrition tube. To monitor the temperatures of the system and heaters during the experimental runs, eight K-type thermocouples were used, located at various locations in the apparatus, as shown in Figure B2 in Appendix B. A heater-control and data acquisition system had already been set up from previous catalyst attrition studies, using Omega CN76000 controllers for the heaters, and an OMEGA-DAQ-56 data acquisition card, with Personal DaqView Plus V2.0.10 software to record temperature and gas flow rate data. Note that an OMEGA FMA5526 gas flow controller was used to control the flow rate of air into the unit.   Another significant deviation from the ASTM standard is the duration of the attrition test runs. While the standard defines the run as being 5 hours long, the experiments performed in this project ranged from 5 to 36 hours in length. Historically, it has been found that the rate of attrition for a fresh sample is very high during the first few hours (Yang, 2003), as the surfaces of the particles are initially rough and jagged, with the particles themselves being irregularly shaped. These features result in significant abrasion effects (fines generation) at the beginning of the test run. Therefore, to properly study the long-term attrition performance of a material, attrition tests need to be of longer duration, ideally lasting until the rate of attrition greatly diminishes. Running the tests for up to 36 hours was found to balance the work required for each 20 test and the usefulness of the results. However, it was found that for some of the test conditions used, after 36 hours enough material had been elutriated from the system that not enough particles remained in the bed to produce useful attrition test results.  Lastly, while the ASTM standard specifies the particle size range from 10 to 180 ?m, an industrial CO2-capture system with dual circulating fluidized beds will likely use much larger particles, especially since attrition effects will decrease the mean particle size as the process proceeds. Therefore, several initial particle sizes were tested, with mean diameters ranging from approximately 500 ?m to 1000 ?m. While there was a great degree of control over the initial size of the raw Cadomin limestone particles, the initial size of the pellets and shelled sorbents was limited to the sizes that the material suppliers could produce.   The suggested particle size of 10 to 180 ?m in the ASTM standard was likely chosen to correspond to the average fluidized catalytic cracking (FCC) size range (Geldart group A). FCC is expected to be the most common material tested in such a unit due to its historical and current industrial significance. The suggested particle size range is important because it should be ensured that unattrited particles are not elutriated through the settling chamber and deposited in the fines collector. The maximum diameter of the settling chamber should be set so that the superficial gas velocity matches the terminal velocity of the intended maximum ?fines? diameter, which the ASTM standard sets to  e 20 ?m. At the standard?s 10 L min (STP), the gas velocity through the 110 mm diameter settling chamber is 0.0175 m/s. The well-known Haider and Levenspiel model for terminal velocity (equations in Appendix B) was used to calculate the corresponding maximum fines diameter for FCC, assuming a sphericity of 0.58 (Basu, 2006) and a particle density of 1750 kg/m3 (Yates, 1996). The calculations predict that the fines will have a maximum diameter of 20 ?m, which is equal to the value given in the ASTM standard.   However, the maximum fines diameter (20 ?m) overlaps with the range of the intended bed particles of 10 to 180 um. Therefore, there will be initial elutriation of the fraction between 0 to 20 ?m when the sample is loaded into the unit, resulting in inherent experimental error in the air jet index, as these elutriated particles will be interpreted as attrited fines. The magnitude of the error depends on the initial size distribution of the particles tested. To minimize this issue, the standard directs the user to measure the weight of the fines after 1 hour in addition to after 5 21 hours. However, the best way to avoid this error is to limit the initial particle size to greater than 20 ?m, as practiced during all experiments in this project.   2.2 ASTM Standard Disadvantages  The ASTM air-jet test method has important advantages and disadvantages that should be discussed to give context to the experimental results in this project, and to provide suggestions for improving the standard. Most importantly, the test requires about 50 g of sample per test run. Depending on the material, this can either be an advantage or a disadvantage. Compared to some alternative attrition test methods, such as impactor (Chen et al., 2007) or full circulating bed systems (Chen et al., 2008), 50 g is a relatively small requirement. On the other hand, other test methods such as jet cup systems (Zhao et al., 2000) require only a few grams of material. For industrially common (e.g. FCC) or inexpensive (e.g. limestone, dolomite) particulate materials, 50 g is a low requirement. However, for synthetic or novel materials (e.g. shelled pellets, lithium orthosilicate) produced by academic laboratories, 50 g is a monumental requirement, especially in comparison to the amounts required for other tests such as TGA, BET, and SEM. The ASTM standard was likely designed from an industrial perspective, where materials would be tested by organizations with ample funding and supplies of material.  Since our test procedure is derived from a defined ASTM standard, the experimental results are given extra significance because they are easier to compare to other attrition studies and should be reproducible in other labs. The test itself is also moderately easy to carry out, as the operator only needs to load the system, turn on the gas flow, and wait until the appropriate time to remove the sample and empty the fines collector. However, because of its simplicity, the test also has limitations. Since the test apparatus only uses three orifices in the distributor plate, the particles are not uniformly or truly fluidized. This design was likely chosen to maximize the jet velocity, as the small diameter of the orifices forces the air to pass through near the speed of sound, increasing the rate of attrition to allow for appreciable fines production within the 5-hour test period. The lack of a proper fluidized bed region also allows the particles to experience jet attrition without significant bed attrition, isolating the source of attrition. However, since there are only three orifices in the large plate, previous studies have shown that there is a ?stagnant region? or dead zone at the bottom of the attrition tube, where particles simply remain immobile and do not interact with the jets (Kimura et al., 1995). The size of this stagnant region seems to 22 depend inversely on the size of the particles (Xiao et al., 2011), making such a consideration very important for the standard?s intended particle size range.  Lastly, since the calculation of the air jet index only depends on the weight of the materials in the fines collector, the unit does not seem to have been designed with complete sample collection in mind. Therefore, in order to brush down and collect the residual particles in the unit for a full particle size distribution analysis, most of the unit has to be disassembled, which is especially time-consuming after high-temperature experiments. Between each attrition run cycle, the total disassembly, collection, sieve analysis, reconstruction and sample feeding requires about 2 hours of physically intensive work for room temperature experiments, or 3 hours for high temperature experiments requiring heater positioning and insulation removal/addition. Reducing the need for deconstruction would greatly decrease the amount of preparatory work between each test cycle, allowing for a greater number of tests cycles and collection of more data, especially important for attrition model generation.  2.3 Suggestions for ASTM Standard Improvement  From performing a large number of room temperature and high-temperature tests in this project, numerous drawbacks and challenges with the ASTM D5757 standard have been identified. It is the author?s opinion that several adjustments could be made in the standard methodology that would greatly improve the ease of use of the equipment. The overall focus of the suggestions is to make the unit more compact, easier to load/unload material, require less material, and make it easier to heat for high temperature tests. It is important to note that any changes to the standard would make the new results technically incomparable to previous results obtained strictly following the standard. Nevertheless, by suggesting the following modifications, a dialogue about the current standard could be opened, which may inspire others with attrition testing experience to present their own concerns and suggestions.  Firstly, the requirement for 10 L/min of air causes several issues. Most importantly, the high flow rate is demanding if cylinder gas is required, as several cylinders of air were required for each total test run in this study. The high flow rate was likely chosen to provide sonic velocities through the distributor orifices, while also allowing for a sufficient superficial gas velocity to elutriate all fines into the fines collector. The sonic velocities allow for rapid attrition and 23 therefore reduce the duration of the required test cycle. However, a short test cycle also causes issues with simply accounting for initial attrition effects without properly measuring long term attrition. Also, the same superficial gas velocity at a lower flow rate could be obtained by decreasing the diameter of the column. In addition, decreasing both the column diameter and the gas flow rate would significantly decrease the amount of power required to heat the gas to high temperatures, and would both decrease the temperature differential between the heaters and the column wall, and the column wall and the centre of the column. Decreasing the column diameter would also reduce the stagnant regions around the jets, as a fraction of the particles at the wall were found to be immobile (Xiao et al., 2011). This is especially important for this unit, as there are only three orifices near the centre of the relatively large distributor. If the diameter of the column were to be decreased, it could also be necessary to reduce the distributor orifice pitch in order to prevent erosion of the column walls from high-velocity particles from the jets. The size of the dead zones between the jets is strongly affected by both the initial bubble size, and the Geldart classification and diameter of the particles (Wen et al., 1982), which will change, not only depending on the material tested, but also as the attrition proceeds.   After considering the advantages and disadvantages of both current and historically used attrition testing units (as discussed in Chapter 1), and performing the various necessary calculations and estimations, a new standard is proposed, as outlined in Table 2.1.   24 Table 2.1: Proposed Standard vs. the ASTM D5757 Standard Specification ASTM D5757 Proposed 2013 Standard Initial Material Mass (g) 50.0 15.0 Number of Jets 3 1 Orifice Diameter (mm) 0.381 0.381 Column Inside Diameter (mm) 35 19.05 (3/4 inch) Gas Air, 30-40% relative humidity Air, < 5% relative humidity Gas Temperature (?C) 0 20 Gas Flow Rate (L/min) 10.0 3.0 Column Superficial Gas Velocity (m/s) 0.173 0.175 Settling Chamber Superficial Gas Velocity (m/s) 0.0175 0.0176 Jet Velocity (m/s) 433.1 437.5 Maximum Fines Diameter (?m) 20 20 Column Height (mm) 710 710 (tentative) Settling Chamber Middle Section Length (mm) 300 300 (tentative) Settling Chamber Diameter (m) 110 60 Settling Chamber Top Cone Length (mm) 100 35 Settling Chamber Bottom Cone Length (mm) 230 75 Test Length (h) 5 15  There are several advantages of the proposed 2013 standard that make it an improved system for attrition tests. First, it offers a much smaller and simpler design than the current ASTM standard. The column has been set to be a standard ? inch diameter tube, which would make construction of the unit a far simpler and less expensive task, and the column could be easily replaced if broken or damaged. The smaller diameter would also enable the system to be heated by a tube furnace, negating the requirement for expensive custom-made ceramic heaters and large amounts of glass wool insulation. At this diameter, a gas flow rate of only 3.0 L/min would be required to keep the column superficial gas velocity close to the ASTM standard (0.175 m/s versus 0.173 m/s). The reduction in diameter and gas flow rate should also provide a significantly more uniform temperature distribution, reducing the required temperature of the heaters. It is important to note that the standard?s gas temperature has been changed from 0?C to 20?C, allowing for a baseline attrition index that is easier to obtain. The relative humidity of the air was also set to be < 5% R.H., as it would be manageable to reduce the humidity of building air to this level with an in-line desiccator, and dry air cylinders are widely available. If higher gas humidity is required to 25 reduce electrostatic effects, then it should be noted alongside any reported results that humid air was used, and the humidifier system used should be explained in detail.  It is important to keep the initial bed height of the tested material the same for both standards to retain the same height relative to the jet penetration length. Therefore, the reduction in column diameter allows for a much smaller particle mass requirement, reduced from 50 g to 15 g. This reduction in mass is the primary motivation for the standard, as tests would henceforth require less than a third of the material for a single test. While this may not seem important for common materials such as FCC or limestone, quantity is a significant hurdle for more expensive materials such as mesoporous-shelled pellets and novel synthetic materials such as lithium orthosilicate. A reduction in sample mass would allow for a greater number of tests to be conducted, giving greater insight into the effect of various operating conditions on the attrition rate. However, the reduction in quantity of test material would likely necessitate the use of a laser particle size analyzer instead of sieve analysis, the latter requiring a minimum mass for accurate and repeatable results, dependent on the sieve opening size (ASTM Standard D6913, 2009).   A smaller gas flow rate would also allow for a much smaller settling chamber diameter. Therefore, the diameter has been reduced from 110 mm to 60 mm, retaining nearly the same velocity (0.0175 m/s versus 0.0176 m/s) for fines separation. The same convergent and divergent cone angles have been retained in the settling chamber, as the angles were likely chosen in order to minimize particle adhesion based upon the drained angle of repose of FCC. Therefore, the smaller settling chamber diameter allows for a decrease in the length of the conical sections, from 100 mm to 35 mm for the top section and 230 mm to 75 mm for the bottom section. However, the height of the column and of the settling chamber middle section have tentatively been specified to be equal to the ASTM standard, 710 mm for the column and 300 mm for the settling chamber middle section. Intuitively, the length of the column must exceed the sum of the bed height and the transport disengaging height (TDH) of the smallest non-fines bed material, and this has been shown to be true in previous attrition tests (Werther & Xi, 1993). A variety of graphical and empirical correlations are available for estimating the TDH (Yang, 2003), each providing significantly different estimated TDH values, as expected (Grace, 2012). The height of the column in the ASTM standard was likely determined through experimental testing, with the column length decreased until the amount of fines produced during an attrition test cycle 26 increased due to elutriation of non-fine material (> 20 ?m). A similar approach was likely conducted to determine the optimal settling chamber length, increasing the length until sufficient space was available for deceleration and disengagement of the non-fines. Therefore, the column height and settling chamber middle section length are initially set equal to the ASTM standard. However, tests should be performed on a constructed system in order to determine the minimum values of these specifications, which would greatly reduce the required size of any heaters and insulation, in addition to making the system easier to store and clean.   The highest priority in designing the 2013 standard was to keep the jet velocity, column velocity, and the settling chamber velocity as close as possible to the ASTM standard, in order to maintain comparability of results obtained with the two standards. The jet velocity ultimately controls the rate of attrition, as the rate has been shown to be proportional to uor3 (Werther & Xi, 1993). Therefore, in order to keep the jet velocity approximately the same while reducing the required gas flow rate, the number of jets has been reduced from three to a single jet. A reduction to a single jet allows for a reduction in gas flow rate to about a third of the ASTM flow, while reducing the rate of attrition by about a factor of three instead of by a factor of ~27 if the velocity was reduced to a third of the ASTM value. This presents the largest drawback for the proposed standard: by reducing the number of orifices, the rate of attrition is decreased to about a third of the ASTM standard. Therefore, instead of a 5 hour test, a 15 hour test would be required for the same degree of attrition. However, it is the author?s  elief that the numerous advantages described in this section outweigh this drawback.   For now, the 2013 standard will set the attrition analysis method to match that of the ASTM standard, whereby the air jet index (AJI) is defined as the mass of material accumulated in the fines collector as a fraction of the initial bed material (15 g). This allows for a simple attrition number that will not vary as a function of the method of PSD analysis used. However, it is strongly suggested that the total change in bed PSD be determined to give insight into the attrition process, and that the system be operated at various temperatures, velocities, and run times to study their effects. Nevertheless, a single attrition index at 3.0 L/min of non-humid air at 20?C allows for simple comparison with results from other institutions.  27 A variety of calculations were required in order to design the 2013 standard. These were based on 75 ?m FCC with a bulk density of 880 kg/m3 (Chen, 2003), likely the intended bed material for the ASTM standard. The equations used can be found in Appendix B, while the calculation approach is summarized as follows:  1. Based upon a ? inch (19 mm) diameter column, calculate the required gas flow rate to retain the same column gas velocity as the ASTM standard, based upon a single orifice (new gas flow rate of approximately 3.0 L/min at 20?C and 1 atm). 2. Calculate the orifice jet velocity to ensure that it remains as close to the ASTM standard as possible (new uor = 438 m/s). 3. Calculate the minimum fluidization velocity of 75 ?m FCC (umf = 0.295 cm/s). 4. Calculate the terminal velocity of 20 ?m FCC (u? = 1.74 cm/s). 5. To reduce the significance of the dead zone around the orifice, calculate the initial bubble diameter to ensure that it is similar to the tube diameter (Rees et al., 2006). Calculations show that DB0 = 16.5 mm, which is close to the column diameter of 19.1 mm. 6. Calculate the new initial mass required to retain approximately the same bed height (mb = 15 g, new Hb = 59.8 mm). 7. Calculate the jet penetration length to ensure that it is close to the ASTM value, and that the bed height is greater than the jet penetration length (new Ljet = 16.1 mm). 8. Calculate the new settling chamber diameter to keep the velocity the same as for the ASTM standard (DSC = 60 mm). 9. Using basic trigonometry, calculate the new top and bottom cone lengths to retain the same internal angles as the ASTM standard (top cone = 35 mm, bottom cone = 75 mm).  2.4 Experimental Conditions  The most significant challenge of deciding on the experimental conditions for this project was the high number of process variables that affect the rate of attrition, as discussed in Chapter 1. Further complicating the experiments was the need to test three different materials, in addition to having to consider both the pre and post-calcination species. The different experimental conditions are summarized in Table 2.2.  28 Table 2.2: Attrition Test Conditions Experimental Variable Conditions Material/Sorbent Crushed Cadomin limestone Calcined Cadomin lime CANMET pellets w/ calcium aluminate cement binder Calcined CANMET pellets w/ calcium aluminate cement binder Calcined CANMET pellets w/o binder Calcined CANMET pellets w/o binder, 5 ?m SiO2 shell Calcined CANMET pellets w/o binder, 1 ?m SiO2 shell Temperature (?C) 20?3 or 500?5 Pressure Atmospheric Initial Sample Mass (g) 55.0 Approximate Initial Mean Diameter (?m) 500, 750 or 1000 Test Run Times (h) 0, 1, 5, 12, and 24, sometimes also 36 Fluidizing Gas Building (humid) air; relative humidity = 0.0% ? 12.9% (avg. = 7.6%) Extra dry air (tanked); relative humidity = <10 ppm H2O Distributor Jet plate (3 0.397 mm (1 64?) orifices, open area 0.0386%) Fluidizing plate (31 0.794 mm (1  2?) orifices, open area 1.57%) Velocity Selection Source Jet test: ASTM Standard (10 L/min STP) Fluidization test: U ? Umf Gas Flow Rate (L/min) @ 20?C, 1 atm Jet test: 10.0 (20?C, 500?C) or 3.8 (500?C) 68.0 (limestone), 54.0 (CANMET pellets), 30.8 (lime), 29.2 (calcined CANMET pellets) Superficial Gas Velocity (m/s) Jet test: 0.173 (20?C, 500?C) or 0.457 (500?C) Fluidized test: 1.178 (limestone), 0.935 (CANMET pellets), 0.534 (lime), 0.506 (calcined CANMET pellets)  The specific conditions used for each trial are provided in Table B1 of Appendix B. As previously discussed, this study is part of a larger Carbon Management Canada (CMC) project, and thus the three main materials to be tested (crushed limestone, CANMET pellets with cement binder, silica shelled pellets) were chosen by the various participants (University of British Columbia, University of Ottawa/CANMET, and Laval University, respectively). The baseline (reference) material for the study is crushed limestone, against which all other materials are compared. The two other material groups ? CANMET lime-based pellets with a calcium aluminate cement binder, and the same pellets without the binder, but instead with a mesoporous silica shell ? are also CMC sorbent candidate, therefore requiring attrition testing. Additional information about each sorbent candidate is given in Chapter 1. 29 Ultimately, the tests were divided into two primary classifications: ambient temperature tests (20?C) and high temperature tests (500?C). The ambient temperature was chosen to be similar to the ASTM D5757 standard. However, as previously discussed, the standard of 0?C was deemed to be impractical for the required flow rate of gas, and therefore 20?C was chosen as an acceptable replacement, as the effect from such a small difference in temperature on the experimental results was likely to be small. The temperature of 500?C was selected primarily because this temperature was moderately easy to maintain stably over the long test periods. Due to the high gas flow rate required, significant power was required to heat the system, using both two preheaters and main heaters around the column. Through preliminary system testing, it was found that a significant temperature differential existed between the primary heater surfaces and the gas temperature within the column. In order to raise the gas temperature above about 550?C, the primary heater temperature would have to exceed 800?C, causing scaling of the 304 stainless steel column, which would fall off and corrupt the sample. Increasing the amount of ceramic insulation surrounding the column had minimal effect, as the temperature differential was primarily due to internal cooling from the high gas flow rate. Therefore, it was decided to run the tests at 500?C, which would prevent scale production and limit the heater/column temperature differential. Therefore, instead of calcining the material inside the unit before calcined tests, a separate Thermo Scientific Lindberg Blue M oven with a 10 mL/min nitrogen purge was used to calcine the samples at 800?C for 30 minutes, then 850?C for 4 hours, followed by cooling the material to room temperature before insertion in the attrition unit for testing. This also allowed the sample to be weighed before and after calcination to ensure that the sample was fully calcined.   During all tests, the column was at atmospheric pressure, consistent with the ASTM standard. While an increase in pressure is expected to increase the rate of attrition, as discussed in Chapter 1, the intended industrial CO2-capture system is not expected to operate at elevated pressures, and therefore the effect of pressure was not studied in this project. For almost all tests, the initial sample mass was set at 55.0 g, 10% above the ASTM standard value of 50.0 g. It was thought that since the tests were to  e operated for much longer than the standard?s 5 hour test cycle, additional material should be used to ensure that enough material is in the column to keep the rate of particle-particle collisions high, and to account for losses during the cleaning process between test cycles. 30  For the large ma ority of tests, an initial particle diameter of approximately 1000 ?m was chosen. It was thought that if the initial particle size was large, then the changes in the particle size distribution would be easier to observe and measure. Also, as discussed in Chapter 1, it was important to set the initial particle size large enough to prevent issues of ?natural grain size? significantly slowing the attrition process. However, the 1 ?m and 5 ?m-thick silica shelled pellets provided were approximately 750 ?m in diameter; thus, the tests were also performed on lime and calcined cement- ound pellets at approximately 500 ?m to study the effect of initial particle size. Note that the 500 ?m tests were performed  efore it was known that the provided silica shelled pellets would  e 750 ?m in diameter. Replicate ambient temperature tests were also conducted with lime to estimate attrition test variability.  The ASTM standard sets the total test period at 5 hours, with an additional fines collection at 1 hour to estimate initial elutriation effects. However, as discussed in Chapter 1, five hours is generally not enough time to surpass initial attrition effects. Also, the residence time of such sorbents in industrial systems is intended to be orders of magnitude greater due to multiple CO2-capture/regeneration cycles, and thus it is important to perform attrition tests for long time periods. Therefore, the samples were tested up to 36 h under ambient conditions, and 24 h for high temperature conditions. It was found that after 24 hours for some of the high temperature tests, very little sample remained in the unit, and therefore it was inappropriate to continue due to the minimal bed height. For ambient tests, it was found that 36 hours was an appropriate length of time to study the long term effects on the sorbent, balancing the amount of work required for the tests and the information obtained from the results.   nitially, the fluidizing gas was chosen to  e ?humid?  uilding air, as this matched the ASTM standard and was both inexpensive and easy to use. Using an electronic humidity meter (OMEGA RH-USB), the relative humidity of the gas was found to vary between 0.0 to 12.9%, depending on the time of day measured. However, as discussed in Section 2.1, tests at the University of Ottawa found that the CANMET pellets were very sensitive to moisture (Manovic, 2012), and it was requested that the tests be conducted in dry air. Therefore, the large majority of tests used extra dry cylinder air, classified as having < 10 ppm H2O (Praxair, 2008).  31 Two different distributors were used during the attrition tests. Most tests were performed with the three-orifice (orifice diameter = 0.397 mm) jet distributor in accordance with the ASTM standard. To study the effect of bed attrition, a 31-orifice (orifice diameter = 0.794 mm) distributor was also used to provide uniform fluidization, which also significantly decreased the jet velocity in comparison to the jet plate. For the jet tests, the gas velocity was based on the ASTM standard of 10.0 L/min. However, since a temperature of 20?C was used instead of the standard 0?C, the flow rate could have been adjusted to 10.7 L/min to match the mass flow rate of gas. However, keeping an ambient flow rate of 10.0 L/min was chosen to remain consistent with previous studies at UBC (Xiao et al., 2011). For the fluidization tests, the gas velocity was based on (U - Umf), with the gas velocity chosen to be either 25 or 50 cm/s above the Umf of the particles, based upon their initial size.   The minimum fluidization velocities were experimentally determined using the recommended method (Yang, 2003), where the pressure drop across the bed is plotted against the gas velocity, and Umf is set to equal the point of intersection of the fixed bed and fluidization portions. An example of such a Umf determination curve can be found in Figure B3 in Appendix B. The Umf tests were performed in the ASTM unit, using an OMEGA PX164 differential pressure transducer, and recording the pressure drop and gas flow data with the data acquisition system. All Umf tests were performed at room temperature, as were also all fluidization attrition tests. The pressure drop was measured as the gas flow rate was increased incrementally up to 30 L/min, and then also by decreasing the flow rate and recording the pressure drop on the way down to 0 L/min. The intersection points of both the flow-ascent and descent curves were identified, but ultimately the point on the descent curve was used to identify Umf, as interparticle forces can affect the pressure drop before fluidization is obtained on the ascending curve (Bi, 2012). For the Umf tests, it was found that the pressure drop was still a positive linear function of gas velocity even after the Umf point. However, this is expected for tests with large particles, as they result in slugging beds (Chen et al., 1997). A summary of the Umf results is given in Table 2.3. This table also includes the experimentally measured particle densities of the materials. A 25 mL liquid pycnometer with water or hexane was used to measure the particle densities. A liquid pycnometer was used instead of a gas pycnometer in order to measure the particle density, instead of the skeletal density determined by helium pycnometry.  32 Table 2.3: Experimental Particle Densities and Minimum Fluidization Velocities Material Particle Density (g/cm3) Average Particle Size in Umf Test (?m) Minimum Fluidization Velocity (Umf, m/s) Limestone 2.72?0.02 1004 0.710 CANMET Pellets 2.16?0.01 1006 0.467 Lime 1.52?0.01 490 0.284 Calcined CANMET Pellets 1.72?0.01 490 0.260  As previously stated, the jet tests were based on the 10 L/min flow rate of the ASTM standard. However, when the gas is heated to 500?C during high temperature tests, the volumetric flow rate increases to approximately 26.4 L/min. Hence, it was also important to run a high temperature test at 3.9 L/min (20?C and 1 atm), and therefore 10 L/min at 500?C, to measure the effect of keeping the velocity the same while increasing the temperature.   Most particle size distribution (PSD) measurements were made using sieve analysis, with 8 sieves plus the bottom tray, and a Fritsch Analysette 3 Pro shaker, for a shaking duration of 14 minutes with a maximum amplitude of 2.7 mm. To measure the PSD of the fines, a Mastersizer 2000 laser diffraction system was used, with a Scirocco dry feeding accessory. The Scirocco requires that all fed samples are consumed, and therefore the analyzed particles cannot be returned to the attrition unit for further testing. Since the fines are not required to be reintroduced to the batch attrition tests because they would be soon elutriated, it was appropriate to use the high-accuracy Mastersizer on the produced fines. However, to measure the PSD of samples with large particles, a significant amount of material is required for proper analysis, so that such a loss of material would greatly affect subsequent attrition test runs. Therefore, sieve analysis was deemed a more appropriate method to measure the changes in the bed PSD. To analyze the surfaces of the particles, both Hitachi S-3000N and Hitachi S570 scanning electron microscopes were used under high vacuum with 20 kV of accelerating voltage. Both microscopes are located in the Materials Engineering department at UBC.  The full test procedure used can be found in Appendix B. It is largely based on the ASTM D5757 standard method. The full list of experimental trials is outlined in Table B1, while the ASTM standard conditions are listed in Table B2 for comparison.   33 Chapter 3 ? Attrition Test Results  As shown in detail in Table B1 of Appendix B, 21 attrition tests were performed, with Trials 10 and 17 acting as replicates for Trial 9 in order to provide an estimate of the inherent variance in the particle size distribution change and the production of fines as a function of time. The trials are grouped into three categories: 1) uncalcined jet tests and fluidized bed tests, 2) calcined ambient jet tests, and 3) high temperature jet tests.   In the attrition literature, there is no general consensus as to which specific particle diameter should be used to describe the change in the PSD due to attrition. In the field of fluidization, the most popular diameter is the surface-volume mean diameter, also known as the Sauter mean or dsv:   dsv   nii dpi  nii dpi2 1 xidpii (3.1)  The Sauter mean is used because it is based upon the surface area of the particles per unit volume, which is convenient for both reaction kinetics and pressure drop calculations. It could be argued that the surface area of the particles is important for attrition studies based on the Rittinger comminution model from Section 1.4. However, as discussed in that section, the required energy estimation from the theorem is largely inaccurate. Another commonly used diameter is the mass mean diameter, also known as the deBrouckere mean or dmm:   dmm   nii dpi4 nii dpi   mii dpiM (3.2)   n the author?s opinion, this is the best diameter to use in attrition studies because it describes the relative change in the mass fraction of each particle bin size, rather than the change in surface area. The momentum of the bed particles is important in attrition, as shown in Section 4.3. Since the Sauter mean gives extra mathematical weight to small particles because of their high surface-area-to-volume ratio, it is less appropriate than the deBrouckere mean which places a higher importance on the large particles that have significantly higher momentum in the fluidized process, even with a relatively slower particle velocity. The Sauter mean tends to overlook the presence of a few large particles because of their low surface-area-to-volume ratio, but large 34 particles have a more significant influence in the production of smaller particles and fines, especially since their critical stress for fracturing is lower due to larger maximum crack lengths (Shipway & Hutchings, 1993). Therefore, in this project, the mass mean diameter is used to describe the attrition results, and also in the population momentum model in Section 4.3.  In the following sections, the Air Jet Index (AJI) from Section 2.1 is adopted to describe the production of fines as a function of time. Recall that the AJI as defined by ASTM D5757 (ASTM Standard D5757, 2006) as:   Air  et  ndex   Mass of fines after 5 hour test runOriginal mass of sample  100  (2.1)  However, in addition to the 5 hour test run AJI, AJI values for 12, 24, and 36 hour tests runs are also presented. Both the differential and cumulative particle size distributions for all experimental trials are provided in Appendix C.  3.1  Uncalcined Jet Tests and Fluidized Bed Attrition Tests  Due to their similar degrees of attrition, this section reports the results of both the jet tests on the uncalcined limestone and CANMET pellets and those of the fluidized bed tests on the same materials, both uncalcined and calcined. The results are summarized in Figures 3.1 and 3.2:   35  Figure 3.1: Total mass mean diameter reduction vs. time for uncalcined jet and fluidized bed attrition tests for up to 24 h at 20?C. For full trial details, refer to Table B1.   Figure 3.2: Air Jet Index vs. time for uncalcined jet tests and fluidized bed attrition tests for up to 24 h at 20?C. For full trial details, refer to Table B1.  In Figure 3.1, note that a high value on the y-axis refers to a greater reduction in mean diameter of the particles, including both the bed material and the fines produced. Therefore, of the six samples, the material from Trial 16 showed the greatest attrition after 24 h. Similarly, in Figure 3.2, a higher Air Jet Index value corresponds to a greater production of fines, with Trial 3 producing the greatest amount of fines after 24 h.  0.0% 1.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% 0 5 10 15 20 25 30 Total Mass Mean Diameter Reduction Time (h) CANMET, 1000 ?m, Humid, Jet (Trial 1) CANMET, 1000 ?m, Humid, F. B. (Trial 2) Limestone, 1000 ?m, Humid, F. B. (Trial 3) Limestone, 1000 ?m, Humid, Jet (Trial 4) Calcined CANMET, 500 ?m, Dry, F. B. (Trial 15) Lime, 500 ?m, Dry, F. B. (Trial 16) 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% 0 5 10 15 20 25 30 Air Jet Index Time (h) CANMET, 1000 ?m, Humid, Jet (Trial 1) CANMET, 1000 ?m, Humid, F. B. (Trial 2) Limestone, 1000 ?m, Humid, F. B. (Trial 3) Limestone, 1000 ?m, Humid, Jet (Trial 4) Calcined CANMET, 500 ?m, Dry, F. B. (Trial 15) Lime, 500 ?m, Dry, F. B. (Trial 16) 36 Comparing jet attrition Trials 1 and 4 at 20?C in Figures 3.1 and 3.2, 1000 ?m limestone and CANMET pellets have similar mean particle diameter reductions and AJI values. It was anticipated that the CANMET pellets would perform substantially better than raw limestone due to the cement binder, but the results show almost no reduction in friability. In fact, the pellets had a slightly greater diameter reduction than the limestone, although the pellet AJI was about 25% of the value for limestone. The reason for this can be seen in Figures 3.3 and 3.4:   Figure 3.3: SEM images of original uncalcined CANMET pellets.   Figure 3.4: SEM images of uncalcined CANMET pellets after 5 h jet attrition in humid air at 20?C.  It can be seen in Figure 3.3 that the pellets were initially smooth, resulting in minimal rough edges or asperities to be removed for fines production. However, comparing Figures 3.3 and 3.4, it can be seen that after 5 h, many new cracks formed in the pellets, indicating that fracturing was the primary attrition mechanism for Trial 1. While not observed in these results, it might be 37 expected that after enough time, these particles would experience sufficient fragmentation to produce rough surfaces for considerable fines production.   The uncalcined fluidized bed tests in Trials 2 and 3 show significantly different results. The limestone diameter reduction is less than for the CANMET pellets, while the AJI of limestone is nearly four times that of the pellets. Product particles are clearly seen in the SEM images in Figures 3.5 and 3.6:   Figure 3.5: SEM images of uncalcined CANMET pellets after 5 h fluidized bed attrition in humid air at 20?C.   Figure 3.6: SEM images of uncalcined CANMET pellets after 12 h fluidized bed attrition in humid air at 20?C.  After 5 hours of fluidization, the pellets appear to not only show significant cracks, but the production of ?clusters?, where constituents are only bound to the entire cluster along single edges. This effect is increasingly evident after 12 h, as Figure 3.6 shows that the clusters are breaking along the boundaries, which likely explains why the difference in diameter reduction 38 between the pellets and limestone increases with time. On the other hand, even the broken pellet clusters retain their smooth outer surface, which explains the lack of fines production.  Trials 15 and 16 used calcined material, but were fluidized in dry air to minimize the formation of calcium hydroxide. Since calcium oxide is more prone to attrition than calcium carbonate, it was expected that the rate of attrition would increase in comparison to Trials 2 and 3. However, an initial particle size of approximately 500 ?m was used to reduce the minimum fluidization velocity and the corresponding requirement for dry air, which should decrease the relative rate of fines production in accordance with Equation 1.4. Figure 3.1 shows that the extent of particle diameter reduction increased, while Figure 3.2 shows that the fines production remained about the same.    Figure 3.7: SEM images of the calcined CANMET pellets (500 ?m) after 0 h (top left), 5 h (top right), 12 h (bottom left), and 24 h (bottom right) fluidized bed attrition in dry air.  As can be seen in Figure 3.7, the CANMET particles remained smooth across all four test periods, minimizing fines production. However, clusters again formed (best seen in the 12 h 39 image), which split, causing the reduction in mean particle diameter. Note that if the test had only been performed over 5 h, Figure 3.1 indicates that the CANMET pellets are much more prone to diameter reduction than limestone. However, it is important to study the attrition process over longer periods of time to extend beyond initial attrition effects. Figure 3.1 shows calcined CANMET and lime performing about the same after 24 hours for the conditions of Trials 15 and 16.  From the particle size distributions in Appendix C (Figures C1 through C4, C15, and C16), it can be seen that the PSDs barely changed from 0 to 24 h. The PSDs also remained unimodal, consistent with the lack of abrasion and subsequent fines production. Therefore, fracturing was the primary attrition mechanism for these materials under these conditions. This is unexpected because literature sources state that the particles in the air-jet apparatus primarily experience abrasion (Werther & Xi, 1993). The reason for the discrepancy is likely due to the initial size of the particles. The abrasion-only assumption might only be valid for jet tests under the ASTM standard?s intended initial size range of less than 180 ?m. This is discussed further in Section 4.1.   After reviewing the results of these six trials, it must be emphasized that their degrees of attrition and production of fines were very small compared to the jet test results of Sections 3.2 and 3.3 below where the particles were calcined prior to attrition testing. Calcination significantly increases the rate of attrition from the high friability of calcium oxide. The fluidized bed tests, both calcined and uncalcined, likely display much greater attrition resistance due to the gas velocities used. As discussed Section 1.2, there exists a minimum threshold velocity for fragmentation for each material for a specific set of operating conditions. At room temperature, the threshold velocity for limestone is approximately 8.5 m/s (Chen et al., 2007), while separate studies indicate a threshold velocity for lime of 12-15 m/s (Scala et al., 2000). One would expect the threshold velocity for lime to be less than that of limestone; the difference in findings is likely due to the different limestones used. Nevertheless, the superficial gas velocities in my fluidized bed tests only ranged from 0.5 to 1.2 m/s, which is about a single order of magnitude too low for particle fragmentation. Even though the orifice velocities of each trial ranged from 32 to 74 m/s, only a small fraction of the material experiences this velocity at the distributor before 40 the jet expands into the bed. Therefore, the gas velocity was likely too low to cause appreciable attrition. This is supported by both the resultant PSDs and SEM imaging.  3.2  Calcined, Ambient Temperature Jet Attrition Tests  This report section analyzes the ambient temperature (20?C) jet test results for both calcined CANMET pellets and lime. Both humid and dry gas were used for initially 1000 ?m material, along with dry tests for initially 500 ?m materials. The results are summarized in Figures 3.8 and 3.9:    Figure 3.8: Total mass mean diameter reduction vs. time for the calcined particle jet tests for up to 36 h at 20?C. For full trial details, refer to Table B1.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 5 10 15 20 25 30 35 40 Total Mass Mean Diameter Reduction Time (h) Lime, 1000 ?m, Humid, Jet (Trial 6) Lime, 1000 ?m, Dry, Jet (Trial 9) Lime, 500 ?m, Dry, Jet (Trial 14) Half Calcined Lime, 1000 ?m, Dry, Jet (Trial 8) Calcined CANMET, 1000 ?m, Humid, Jet (Trial 5) Calcined CANMET, 1000 ?m, Dry, Jet (Trial 7) Calcined CANMET, 500 ?m, Dry, Jet (Trial 13) 41  Figure 3.9: Air Jet Index vs. time for the calcined particle jet tests for up to 36 h at 20?C. For full trial details, refer to Table B1.  The most significant change between the results in Sections 3.1 and 3.2 is their magnitude, the change in mean diameter and AJI even approaching 100% for some trials.   Comparing the results for Trials 5 and 6 in Figures 3.8 and 3.9, the lime and calcined CANMET pellets again have similar mean diameter reductions and AJI values. Even under calcined conditions, the cement in the pellets does not appear to have significantly reduced attrition. By the end of the 24 h test, the particle diameter reduction and AJI were only slightly lower for the calcined CANMET pellets than for lime. Figures C5 and C6 best illustrate the severe production of fines, with the PSD changing from a purely unimodal distribution initially to a bimodal distribution between 5 h and 24 h, and then almost again to a unimodal distribution, but with the peak in the fines size bin. Comparing these results to the PSD plots of Trials 1 and 4 demonstrates that calcination significantly influences the rate of attrition. Trials 5 and 6 were conducted using humid air, which would result in the formation of calcium hydroxide, which greatly increases the friability of the material. The extent of attrition is well illustrated in Figures 3.10 and 3.11.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 5 10 15 20 25 30 35 40 Air Jet Index Time (h) Lime, 1000 ?m, Humid, Jet (Trial 6) Lime, 1000 ?m, Dry, Jet (Trial 9) Lime, 500 ?m, Dry, Jet (Trial 14) Half Calcined Lime, 1000 ?m, Dry, Jet (Trial 8) Calcined CANMET, 1000 ?m, Humid, Jet (Trial 5) Calcined CANMET, 1000 ?m, Dry, Jet (Trial 7) Calcined CANMET, 500 ?m, Dry, Jet (Trial 13) 42     Figure 3.10: SEM images of the lime after 0 h (top left), 5 h (top right), 12 h (bottom left), and 36 h (bottom right) jet attrition in humid air at 20?C.     43     Figure 3.11: SEM images of the calcined CANMET pellets after 0 h (top left), 5 h (top right), 12 h (bottom left), and 36 h (bottom right) jet attrition in humid air at 20?C.  As shown in Figure 3.10, after 12 hours there was considerable fragmentation and fines production. After 36 hours the lime was nearly completely attrited. Figure 3.11 shows similar results, but with fragmentation severe earlier at 5 h. It should be noted that the particles are likely attriting so quickly that the fragmentation/abrasion cycle cannot be seen in the images, which is why the larger particles appear to remain smooth.  Similar tests were carried out in Trials 7 and 9, but with dry air instead of humid air. As expected the rate of attrition significantly decreased, but the reduction in the rate of attrition was greater for lime than for the calcined CANMET pellets. Until 12 h, the materials appear to attrit similarly, but the 24 h and 36 h tests resulted in the CANMET pellets increasing in relative friability. The cause of this could be the production of calcium hydroxide when the particles were removed from the unit for particle size analysis, as CANMET indicated that the cement is highly hygroscopic, absorbing water from ambient air even faster than lime (Manovic, 2012). 44 After 36 h, the AJI of the calcined CANMET pellets is nearly 5 times that of lime. From Figures C7 and C9, there is both the production of a bimodal distribution and a shift in the original peak, signifying both fragmentation and abrasion. SEM images of both materials are provided in Figures 3.12 and 3.13.     Figure 3.12: SEM images of the lime after 0 h (top left), 5 h (top right), 12 h (bottom left), and 36 h (bottom right) jet attrition in dry air at 20?C.  45   Figure 3.13: SEM images of the calcined CANMET pellets after 0 h (top left), 5 h (top right), 12 h (bottom left), and 36 h (bottom right) jet attrition in dry air at 20?C.  The CANMET pellets again appear to form clusters after 5 h, which clearly fracture by 12 h. However, the large particles for both lime and the calcined CANMET pellets appear to split in two after 12 h, as the critical required crack length was likely attained, resulting in complete fragmentation, especially visible in the pellets.  In addition to the lime and calcined CANMET pellets, a batch of half-calcined lime was also tested using dry air. The calcination was performed using the same method as in Appendix B, but for a duration of only 1 h at 850?C instead of 4 h. After calcination, the calcination reaction extent was calculated to be 48%. The attrition results were as expected, with partial calcination leading to less attrition, shown by a substantial decrease in both the particle diameter reduction and the AJI. Figure C8 shows that the PSD remained unimodal, clearly indicating that abrasion was not a significant attrition mechanism. The SEM images in Figure 3.14 largely support this finding. 46  Figure 3.14: SEM images of the half-calcined lime after 0 h (top left), 5 h (top right), 12 h (bottom left), and 36 h (bottom right) jet attrition in dry air at 20?C.  The SEM images do not indicate a substantial degree of abrasion, although some fines appear in the 36 h image. There does appear to be moderate fracturing, but the degree is still minimal compared to the results for fully calcined particles.   The final two calcined particle trials, Trials 13 and 14, were both in dry air at 20?C but the initial particle sizes were smaller, approximately 500 ?m. From Equation 1. , it is expected that the mass of fines produced would be approximately half of those for the 1000 ?m tests. However, the results show that both the mean diameter reduction and the AJI significantly increased. Both 500 ?m trials showed similar performance. The PSDs in Figures C13 and C14 both experienced a change from a relatively unimodal to a bimodal distribution, signifying both fragmentation and abrasion. The SEM images in Figure 3.15 indicate that both mechanisms were significant.  47   Figure 3.15: SEM images of the calcined CANMET pellets (500 ?m) after 0 h (top left), 5 h (top right), 12 h (bottom left), and 36 h (bottom right) jet attrition in dry air at 20?C.  Note that Figure 3.15 does not show cluster formation, as in the other CANMET pellet trials, although this is likely because the small initial pellet diameter was obtained from particles that were too small to form clusters. The increased fragmentation relative to the initial larger diameter samples may have resulted from increased particle interaction. After entering the jetting zone, the particles from the initially larger (1000 ?m) diameter trial would only rise a small vertical distance before falling again, while the material in the smaller diameter (500 ?m) trial would be significantly more mobilized and circulated. This would cause more collisions in the smaller trial, resulting in faster diameter reduction and fines production.   Overall, the seven trials demonstrated that humid air greatly increases the rate of mean diameter reduction and the production of fines, likely as a result of calcium hydroxide formation. The degree of calcination is also significant, with calcined material having greater friability, as 48 expected. However, the role of initial particle size is less clear. This is further discussed in Section 4.1.  3.3  Calcined, High Temperature Jet Attrition Tests  The final set of experiments consists of six trials of calcined materials tested at 500?C. Three of the trials are of the third material group, which is a pellet without the cement binder, but instead with a silica coating. The results of the six trials are summarized in Figures 3.16 and 3.17:   Figure 3.16: Total mass mean diameter reduction vs. time for calcined particle jet tests for up to 24 h at 500?C. For full trial details, refer to Table B1.    0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 5 10 15 20 25 30 Total Mass Mean Diameter Reduction Time (h) Lime, 1000 ?m, Dry, Jet (Trial 12) Lime, 1000 ?m, Dry, Low Flow Jet (Trial 18) Calcined CANMET, 1000 ?m, Dry, Jet (Trial 11) 5 ?m Shelled, 750 ?m, Dry, Jet (Trial 19) 1 ?m Shelled, 750 ?m, Dry, Jet (Trial 20) Calcined No-Binder CANMET, 750 ?m, Dry, Jet (Trial 21) 49  Figure 3.17: Air Jet Index vs. time for calcined particle jet tests for up to 24 h at 500?C. For full trial details, refer to Table B1.  Trials 11 and 12 were conducted at the same gas mass flow rate as for the ambient temperature tests. However, since the gas is at a higher temperature in the column, the inlet velocity increased by a factor of 2.64. On the other hand, it is expected that increased temperatures decrease the rate of attrition, as the gas density is lower, resulting in decreased momentum transfer. Furthermore, the higher temperatures should prevent the formation of calcium hydroxide, and also remove hydroxide formed between test cycles. However, higher temperatures increase gas viscosity, increasing the rate of momentum transfer. Therefore, there are competing mechanisms affecting the rate of attrition. The results shown in Figures C11 and C12 resemble the severe attrition from the humid low-temperature trials, where the PSDs undergo transition from a unimodal distribution to a bimodal distribution, then progress into a unimodal distribution with the peak at the collected fines bin. Comparing the mean diameter reductions of the dry low and high temperature trials, the high temperature tests experienced greater attrition by about a factor of 2 to 3 for both lime and the CANMET pellets, depending on the time frame. However, note that while lime performed better by the end of the 5 h and 12 h test periods, the rate of pellet diameter reduction significantly decreased at the end of the 24 h period, whereas it increased for lime. The AJI of both trials followed a similar pattern, with the CANMET pellets producing fewer 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 5 10 15 20 25 30 Air Jet Index Time (h) Lime, 1000 ?m, Dry, Jet (Trial 12) Lime, 1000 ?m, Dry, Low Flow Jet (Trial 18) Calcined CANMET, 1000 ?m, Dry, Jet (Trial 11) 5 ?m Shelled, 750 ?m, Dry, Jet (Trial 19) 1 ?m Shelled, 750 ?m, Dry, Jet (Trial 20) Calcined No-Binder CANMET, 750 ?m, Dry, Jet (Trial 21) 50 cumulative fines by the end of the 24 h test cycle. These results are illustrated by the SEM images of Figures 3.18 and 3.19.    Figure 3.18: SEM images of the calcined CANMET pellets after 0 h (top left), 5 h (top right), 12 h (bottom left), and 24 h (bottom right) jet attrition in dry air at 500?C.   51   Figure 3.19: SEM images of the lime after 0 h (top left), 5 h (top right), 12 h (bottom left), and 24 h (bottom right) jet attrition in dry air at 500?C.  As can be seen in the 12 h images, the lime experienced less diameter reduction than the CANMET pellets, although after 24 h the particle sizes of both materials were similar. The images also clearly show that both fragmentation and abrasion were primary attrition mechanisms. Overall, it appears that the increase in jet velocity had a greater effect on the attrition than the prevention of hydroxide formation. This is expected from Equation 1.3, as the orifice velocity term is to the power of 3, while the fluid density term is only to the first power.  To further study this effect, Trial 18 was conducted at a low gas flow rate in order to keep the gas velocities equal between the high temperature and ambient temperature trials. As shown in Figures 3.16 and 3.17, the degree of diameter reduction and AJI were the lowest of all high-temperature trials. Compared to the ambient temperature dry air lime trial (Trial 9), the mean diameter reduction decreased by an average factor of 2.1. However, the AJI was nearly equal for both trials, an unexpected finding considering that the gas density is decreased in the high 52 temperature tests by a factor of about 2.64. However, this may be a result of the large variance of the AJI results, as discussed in Section 3.4.  Trials 19 and 20 tested the friability of the pellets with the mesoporous silica coating. From Figures 3.16 and 3.17, it can be clearly seen that their attrition resistances are much greater than for either the lime or the calcined CANMET pellets. Interestingly, the 1 ?m thick-coating and 5 ?m thick-coating led to nearly the same diameter reductions and AJIs, an important finding when considering the economics of the coating process and the costs of the required chemicals. After 24 hours of operation, compared to the coated sorbents, the lime mean diameter reduction and AJI were greater by factors of 2.4 and 3.4, respectively, and by factors of 2.1 and 2.6 for the calcined pellets. From the PSD plots in Figures C19 and C20, both shelled materials seem to produce fragmentation products in the 106-250 ?m range, resulting in a nearly trimodal distribution compared to the other trials. SEM images of both materials appear in Figures 3.20 and 3.21.   53   Figure 3.20: SEM images of the calcined pellets with a 5 ?m silica coating after 0 h (top left), 5 h (top right), 12 h (bottom left), and 24 h (bottom right) jet attrition in dry air at 500?C.    54    Figure 3.21: SEM images of the calcined pellets with a 1 ?m silica coating after 0 h (top left), 5 h (top right), 12 h (bottom left), and 24 h (bottom right) jet attrition in dry air at 500?C.  The SEM images show that the fragmentation products are relatively large, which could explain the PSD peaks in the 106-250 ?m size range. The initial material is also smooth, consistent with the minimal production of fines, especially when the extent of fragmentation is relatively low. Magnified images of the sorbent surfaces before and after calcination are shown in Figure 3.22.  55   Figure 3.22: SEM images of coated pellet surfaces, for 1 ?m coated pellets after pre-calcination (top left), post-calcination (top right), and 5 ?m coated pellets pre-calcination (bottom left), and post-calcination (bottom right).  Unsurprisingly, the thin silica coats do not appear to differ significantly before and after calcination, although it was important to confirm this, as any change could affect the friability of the material over multiple carbonation/calcination cycles. The brightness of the pre-calcination image is due to charge buildup in the SEM, likely from poor gold coating distribution during imaging preparation.    The final attrition test trial (Trial 21) was performed on ?blank? CANMET pellets, which had neither the cement nor the silica coating. As shown in Figures 3.16 and 3.17, these unsupported pellets attrited so quickly that only a single 5 h test cycle could be performed, producing a mean diameter reduction of 93.4% and an AJI of 84.1%. The PSD plot in Figure C21 clearly illustrates the severe degree of attrition, further displayed in the SEM images in Figure 3.23.  56  Figure 3.23: SEM images of the blank calcined CANMET pellets after 0 h (left) and 5 h (right) jet attrition in dry air at 500?C.  As can be seen, the pellets were nearly completely attrited after 5 h. Compared to the cement-bound pellets in Trial 11, the binder effectively reduced the rate of attrition, by a factor of 1.6 for the mean diameter reduction and factor of 2.6 for the AJI. However, even more significant is how effective the silica coatings are at reducing the rate of attrition, as the coated materials also did not contain cement. The coated materials showed mean diameter reduction decreases by factors of 3.8 and 4.3, and AJI reduction decrease factors of 11.  and 9.4 for the 5 ?m and 1 ?m silica coated materials, respectively. Therefore, the silica coatings substantially increase the attrition resistance of the pellets, although further testing is needed to investigate coating performance under humid conditions.   3.4  Test Variance  In order to test the variability of the attrition test results, Trial 9 was repeated in Trials 10 and 17. The resultant PSDs can be seen in Figures C9, C10, and C17. The results are also summarized in Figures 3.24 and 3.25.  57  Figure 3.24: Total mass mean diameter reduction vs. time for the replicate lime jet tests for up to 24 h at 20?C. For full trial details, refer to Table B1.    Figure 3.25: Air Jet Index vs. time for the replicate lime jet tests for up to 24 h at 20?C. For full trial details, refer to Table B1.  Figure 3.24 shows minimal variance between the mean diameter reductions between these three trials, and the variance does not increase with time. However, Figure 3.25 illustrates that the variance between the AJI measurements is substantially larger than expected, and appears to grow with time. The high variability in the AJI values is likely because crushed limestone was used, which has a wide range of surface roughness and shapes across different particles and -10% 0% 10% 20% 30% 40% 50% 60% 0 5 10 15 20 25 30 35 40 Total Mass Mean Diameter Reduction Time (h) Lime, 1000 ?m, Dry, Jet (Trial 9) Lime,  1000 ?m, Dry, Jet, Replicate #1 (Trial 10) Lime, 1000 ?m, Dry, Jet, Replicate #2 (Trial 17) -10% 0% 10% 20% 30% 40% 50% 60% 0 5 10 15 20 25 30 35 40 Air Jet Index Time (h) Lime, 1000 ?m, Dry, Jet (Trial 9) Lime,  1000 ?m, Dry, Jet, Replicate #1 (Trial 10) Lime, 1000 ?m, Dry, Jet, Replicate #2 (Trial 17) 58 between each experimental sample. Therefore, when analyzing attrition results, the primary emphasis should be placed on the mean diameter reduction values, with the AJI results only providing secondary support. Using the replicate experimental data, the standard error for both mean diameter reduction and AJI measurements for each time period are summarized in Table 3.1:   Table 3.1: Experimental Standard Errors Time (h) Absolute Total Mass Mean Diameter Reduction Standard Error (%) Absolute Air Jet Index Standard Error (%) Fractional Mean Diameter Reduction Standard Error Fractional AJI Standard Error 5 0.64 2.03 0.044 0.542 12 1.50 1.39 0.054 0.225 24 0.56 3.55 0.014 0.312 36 1.46 4.24 0.029 0.244  The fractional standard errors are the standard errors as a fraction of the average experimental result across all three replicate trials. The fractional standard errors are incorporated into the summary plots from Sections 3.1, 3.2, and 3.3, and are shown in Figures C22-27 of Appendix C. Fractional standard errors are used instead of the absolute values because a high AJI is expected to have a greater absolute variance than a low AJI value. The summary plots show that the variance in the mean diameter reduction results is minimal, with the error bars generally being too small to be visible. However, the error bars in the AJI summary plots are large enough to significantly overlap one another, especially for the high AJI points. However, the overlap is small enough to allow low AJI results to be clearly differentiated from high AJI results.  Due to time constraints, only a single set of replicate trials was performed, and only at ambient temperature. Considering that each trial had significantly different operating conditions and materials, the variance from the replicate trials might underestimate or overestimate the errors of other trials. Therefore, the limitations of this variance analysis must be emphasized. However, it does illustrate the greater variance in AJI results compared to the mean diameter reduction values. Further replicates should be conducted, especially at high temperatures, to improve confidence in both the experimental results and to better quantify the inherent variance for all trials.   59 Chapter 4 ? Attrition Modeling   Using the mean diameter change and fines production results from Chapter 3, the data can be fitted to several empirical or semi-empirical models to determine empirical attrition constants and study the effect of operating conditions. This chapter will focus on three fines production models: the Gwyn (1969) time model, the Werther & Xi (1993) jet attrition model, and the Ray et al. (1987) bed attrition model. To study the mean diameter change data, a new model, based upon the momentum of the bed particles, is also proposed and investigated.  Regression cannot be performed for Trial 21 as there is only a single experimental point. While a theoretical point at time = 0 could be used, any resultant model coefficients would only account for initial attrition effects.  4.1  Gwyn (1969) Time vs. Fines Production Model  As discussed in Section 1.4, Gwyn (1969) studied the relationship between fines production and time by measuring the production of sub-40 ?m fines in a three-jet column, which eventually inspired the ASTM D5757 standard. Gwyn determined that the mass of fines produced is a non-linear function of time, in the form of:   mfinesm ed, initial  at  (1.2)  Extensive testing has shown that b should be less than 1, generally between 0.43 and 0.90 (Neil & Bridgwater, 1999). The constant b must be less than 1 because of ?initial attrition?, where the originally rough surfaces of fresh particles are rapidly abraded, resulting in an initial high rate of fines production that substantially decreases as particle surfaces become smooth. However, this assumes that no particle fragmentation is occurring to produce new rough surfaces, an assumption whose validity is dependent on the particle velocities remaining below the threshold velocity.  For this project, each of the experimental trials was fitted to the Gwyn model. The results are shown in Table D1 in the Appendix. Overall, the model appears to fit reasonably well, with the high AJI trials having high Ka constants, and most b constants ranging between 0.47 and 0.95. As expected, the Ka values are not constant for the same materials, varying greatly depending on 60 the operating conditions and initial particle size. The replicate Trials 9, 10, and 17 also showed greatly varying Ka and b values, which is likely due to the high variance in the AJI results, as discussed in Section 3.4. Most importantly, many of the trials produced b approaching 1, while some led to b > 1, up to a value of 2.21 for Trial 9. A b greater than 1 implies that the fines production rate increases with time, contrary to the concept of initial attrition. However, while the Gwyn model assumes that there is no fragmentation occurring in the process, this was found to be generally inaccurate in Chapter 3. If fragmentation is occurring, new rough surfaces were observed to be produced in the daughter particles, sustaining the rate of abrasion. The fragmentation/abrasion cycle continues the production of fines in a step-function-like manner. Furthermore, complete fracturing of the particles requires the formation of a crack that spans the entire particle diameter. Since the Griffith relationship states that a critical stress is required to propagate an existing crack or flaw in a particle (Shipway & Hutchings, 1993), there would likely be a required minimum time until the cracks in the particles would propagate enough to fracture the first particles. As the particles fragment and produce more particles, the number of particle collisions would likely increase, which could increase the rate of crack formation. However, it must be noted that the minimum stress to fracture a particle generally increases as the particle size decreases, due to the decrease in the maximum possible flaw length. Nevertheless, as particles continue to fragment and decrease in size, eventually the products of fragmentation become small enough to be elutriated as fines. Therefore, there is a minimum time until the attrition of initially large particles progresses enough to produce mother particles whose daughter particles are the size of fines.  In summary, there appear to be three effects considered here that affect the rate of fines production: 1) initial attrition of the rough surfaces of particles; 2) abrasion of new particle surfaces resulting from fragmentation; and 3) production of fines from the fragmentation of small particles. These three effects are illustrated in Figure 4.1:  61  Figure 4.1: Modified fines production model.  Effect #1 is drawn in red, effect #2 in green, and effect #3 in blue. The sum of the three effects is displayed by the solid purple line. Of course, the slope of the purple line and even its curvature depend on the relative significance of each of the three effects. Also, the step-shaped line for effect #2 is visually exaggerated, as the steps would be for individual particles and would individually appear miniscule, but add together to form a smooth line. It would likely be difficult to quantify the three effects, as it would be impossible to fully isolate each one. Nevertheless, the experimental trials show that Gwyn?s model is not valid for all materials, particle sizes and conditions. Gwyn?s assumed lack of fragmentation might  e valid for the small particles tested in his study, such as Group A FCC, but for larger Group B particles such as the limestone investigated here, the additional effects from the modified version of Gwyn?s model presented in this section should be considered.  4.2  Jet Attrition and Bed Attrition Fines Production Models  The three major sources of attrition in a fluidized system are the distributor grid jets, the bed, and the cyclones. In this project, both jet and bed attrition tests were explored, and the fines production rate constants have been determined for:   m fines,  et   C etdp nor?fdor2uor  (1.3)   m fines,  ed   C dp m (U Umf) (1.4) 62 The results are provided in Table D2 of Appendix D. Note that the constants were determined using the rate of fines production from the results from the 12 h test period to the end of the 24 h test period. Ideally, the constants should be determined using steady-state data to minimize the effects of initial attrition.  t is the author?s opinion that the initial attrition should  e included in the constants, since the maximum rate of fines production should be considered, but the equations are defined to be the steady-state fines production rates and have been calculated as intended.  The calculated experimental jet constants for limestone are similar to published literature values (Hartge et al., 2007). As expected, the calcined jet constants are about one or two orders of magnitude greater than for uncalcined material. The results of this study can be compared between each trial, and they consistently follow the patterns seen in the Air Jet Index results presented in Chapter 3. However, Equation 1.3 has the added advantage of taking into account the increased velocity and gas density effects from the high-temperature experiments. As expected, the shelled sorbents have the lowest jet constants of the calcined tests. The jet constants from the high temperature tests for lime and the calcined CANMET pellets are higher than those of the ambient temperature trials, but the difference is less pronounced than by simply comparing AJI results. It is expected that the tests with the initial 500 ?m particle diameters would produce jet attrition constants more consistent with the 1000 ?m initial diameter tests, although the production of fines seemed to increase instead of decrease as expected. This could be due to effect #3 discussed above in Section 4.1, as the 500 ?m particles would produce fragmentation products closer to the size of fines earlier than the originally larger (1000 ?m) particles.  Since only four bed attrition trials were conducted, there are not many values to compare. Nevertheless, the trends are consistent with the AJI results, with limestone having a greater bed attrition constant than CANMET pellets due to the increased fines production. The experimental bed attrition constants deviate from the literature values (Hartge et al., 2007), being about an order of magnitude lower. The likely reason for the low attrition constant is that the threshold velocity was not surpassed during the experimental trials, as discussed in Section 3.1. Therefore, minimal fragmentation occurred, resulting in greatly decreased attrition rates. 63 In conclusion, the determination of the jet and bed attrition constants further supports the results and analysis from Chapter 3. However, both the Gwyn equation and the jet and bed attrition equations rely on the fines production data, which was shown in Section 3.4 to vary significantly. Therefore, Section 4.3 proposes a new model based on particle momentum that uses the less-variable mass mean diameter reduction results.  4.3  Population Momentum Mean Particle Diameter Reduction Rate Model  As discussed in Section 1.4, an alternative approach for assessing attrition test results is to study the change in the mean particle diameter from PSD analysis. However, this method is less common because of the additional work needed to determine the full PSD after each test cycle. While several models to predict the rate of change in     have been proposed, they generally require numerous empirical constants, or conversely are too simple to provide accurate predictions. Therefore, the goal for suggesting a new model should be to start with a theoretical approach, and then augment the equation or relation to follow experimental results while using as few empirical constants as possible.   Dr. Grace, one of the thesis project supervisors, suggested that an attrition model based upon the momentum of the particles be constructed to describe the rate or extent of attrition in a system. From this, a simple model equation is proposed by the author of this thesis to model the total mass mean diameter reduction rate, based on the momentum of the bed particles. The model equation uses the jet attrition test results from this project, and therefore the model is applied to the rate of change of the average mass mean particle diameter due to jet attrition.  The model is based upon an analogy with an nth order chemical reaction rate equation:    ate of reaction   k A m B n (4.1)  Analogously, since there is only one ?species?, and with an empirical attrition rate constant km that considers the friability of the material:    ate of attrition   km  ed momentum n (4.2)  For this model equation, the momentum of the bed is used, which depends on the relative particle size fractions or size population. The  ed momentum will  e referred to as the ?population 64 momentum?. It is expected that the greater the momentum of the bed, the greater the severity of the collisions and the higher the rate of attrition. The bed momentum is a function of the mass of the bed, which comes from the mass of the particles and the particle size distribution, and the velocity of the particles, which is a function of  oth the gas velocity and each particles? terminal velocity. Therefore, for i number of particle bin sizes (e.g. number of sieves):     ed momentum n                 n                                 n (4.3)  where N is the number of size fractions and xi is the bed mass fraction of particle size range i. Since jet attrition is being modeled, it is necessary to use the orifice gas velocity instead of the superficial gas velocity. It could be argued that only a small fraction of the particles experience the maximum orifice velocity before the jet expands into the bed. However, the derivation of this model is similar to the derivation of Equation 1.3, whereby the rate of attrition is simply proportional to the orifice velocity, and the empirical attrition constant (Cjet or km) inherently incorporates the actual jet velocity experienced by the particles. The terminal velocity is subtracted from the orifice velocity. It is estimated by the Haider and Levenspiel correlation (1989) reproduced in Appendix B.   Three other relations should be included in the final model. The rate of attrition must increase proportionally with the number of jets (Werther & Xi, 1993). The ratio of the area of the jet orifices versus the projected area of the particles should be considered, as a greater number of smaller particles will be able to interact within a single jet as compared to larger particles. Lastly, the ratio of the open area of the jets versus the area of the column should be considered. As the column diameter increases relative to the jet area, the size of the dead zones and stagnant regions will increase, proportionally decreasing the amount of material simultaneously mobilized by the jets. However, this last consideration will not be included in the final model equation, as differing column diameters were not used in the experimental trials, and therefore such a term cannot be tested. The formulation thus far is shown in Equation 4.4:    ate of attrition                                                 n (4.4)  65 Equation 4.4 is significantly flawed because it does not consider the original size of the particles or the original momentum of the bed. A fresh sample that initially has a mean particle diameter of 1000 ?m,  ut was then reduced to 500 ?m, will not attrit the same way as a sample originally 500 ?m, as a result of initial attrition effects. Therefore, to incorporate the extent of attrition and change in momentum, the population momentum at time t (the numerator in Equation 4.5) must be divided by the original population momentum at t = 0 (the denominator in Equation 4.5), producing a population momentum fraction as follows:    ate of attrition(t)                                                                                                         (4.5)  After cancelling redundant terms and assuming a constant particle density, the final model equation is:                                                                                   (4.6)  To determine the average value of n, the total mass mean diameter reduction rate for each of the experimental jet attrition trials is plotted against the respective population momentum fraction as a function of time. Since the population momentum fraction can only be calculated at the start and end of each test cycle, the specific time period?s average population momentum fraction is used, as the momentum during the test cycle will affect the rate of attrition, instead of the population momentum at the beginning or end of the test cycle. After plotting the graphs, power law regression was performed in Excel to fit the value of n. The results are provided in Table D3 of Appendix D. It was found that the average value of n is about 2. A few of the trials produced significantly different n values because this model is inappropriate for their data, as discussed below.  Using the determined ?order? of the model equation of 2, the total mass mean diameter reduction rate was plotted against the right side of Equation 4.6 (excluding km), and the results are shown in Figures 4.2 and 4.3. 66   Figure 4.2: Plot of mass mean diameter reduction rate vs. population momentum fraction (n = 2) for 20?C trials.   R? = 0.9419 R? = 0.9221 R? = 0.9071 R? = 0.8837 R? = 0.9776 R? = 0.039 R? = 0.9258 0% 2% 4% 6% 8% 10% 12% 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 Total Mass Mean Diameter Reduction Rate (%/h) (Population Momentum Fraction)2 Lime, 1000 ?m, Humid, Jet (Trial 6) Lime, 1000 ?m, Dry, Jet (Trials 9, 10, & 17) Lime, 500 ?m, Dry, Jet (Trial 14) Half Calcined Lime, 1000 ?m, Dry, Jet (Trial 8) Calcined CANMET, 1000 ?m, Humid, Jet (Trial 5) Calcined CANMET, 1000 ?m, Dry, Jet (Trial 7) Calcined CANMET, 500 ?m, Dry, Jet (Trial 13) 67  Figure 4.3: Plot of mean diameter reduction rate vs. population momentum fraction (n = 2) for the 500?C trials.  Using Excel, linear regression was performed, and the y-intercept set equal to zero, since there is no intercept term in the proposed model equation. The slope of the resultant functions is equal to the attrition constant km for each experimental trial, and the values are shown in Table D3. The resulting attrition rate constants describe the relative friability of each material under the specific process conditions, with both their relative magnitudes and rankings matching the qualitative and quantitative results from Chapter 3.  From Figure 4.2, it appears that the model equation provides a good fit for all trials, except for Trial 7. The deviation in Trial 7 arises because the rate of the mean diameter reduction and the fines production during the first 5 h test period were minimal, while the rate of attrition quickly accelerated during subsequent test periods. The reason for the low initial attrition could be due to a minimum time required for crack formation. In any case, this trial exemplifies the importance of performing long term tests on each material. Similarly in Figure 4.3, Trial 12 has a poor fit R? = 0.3098 R? = 0.9616 R? = 0.9789 R? = 0.559 R? = 0.3383 0% 2% 4% 6% 8% 10% 12% 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 Total Mass Mean Diameter Reduction Rate (%/h) (Population Momentum Fraction)2 Lime, 1000 ?m, Dry, Jet (Trial 12) Lime, 1000 ?m, Dry, Low Flow Jet (Trial 18) Calcined CANMET, 1000 ?m, Dry, Jet (Trial 11) 5 ?m Shelled, 750 ?m, Dry, Jet (Trial 19) 1 ?m Shelled, 750 ?m, Dry, Jet (Trial 20) 68 due to a substantial increase in attrition towards the end of the 24 h test cycle, possibly due to earlier near-critical damage accumulation and rapid subsequent fracturing.   Unfortunately, due to the limited number of tests cycles during each experimental trial, the regression lines for the model are based on very few experimental data points. Ideally, each trial would have several additional points to better confirm both the order of the model equation and its long term performance. However, due to the nature of the CMC project, it was chosen to test the sorbent candidates under numerous different conditions instead of spending extra time on a select few conditions. On the other hand, Trial 9 and its replicate trials 10 and 17 are all plotted in Figure 4.2, and a single regression line is plotted for all 12 data points. The coefficient of determination of 0.922 shows that while there is a high variance in the AJI for replicate trials and a resultant high variance in empirical fines production constants, more confidence can be placed in the less variable mean diameter change fitted rate constants.   The linear fittings for Trials 1 and 4 (listed in Table D3) and for Trials 19 and 20 in Figure 4.3 all show relatively poor coefficients of determination. Also, their fitted n values are also much greater than 2. The reason for both deviations from the model is due to the lack of a y-intercept in the model equation. While some attrition models include a y-intercept value to account for initial attrition effects (Xiao et al., 2012), such an inclusion is only valid for fines production models, and not for mean diameter change models. Logically, the regression lines must pass through the origin because the mean diameter change rate is zero once the population momentum fraction reaches zero. However, Trials 1, 4, 19 and 20 all experienced very low levels of attrition, and, more importantly, experienced low levels of fragmentation. For the proposed model equation to be valid, fragmentation is required to continuously produce new daughter particles for additional fracturing and abrasion. If the attrition process slows down too early due to lack of fragmentation and the smoothing of all particle surfaces is complete, the model will greatly overestimate the future rate of particle diameter reduction. Therefore, the proposed model is only valid for systems where the particle velocity exceeds the threshold velocity, which is specific to the material and the operating conditions. However, this requirement is likely to be satisfied in high-velocity bubbling and circulating fluidized systems where attrition is of concern.    69 Chapter 5 ? Conclusions  The primary objective of this project was to evaluate the attrition resistances of three CO2-capture sorbent candidates. The experimental results show that the cement-bound pellets generally attrit to the same extent or worse than crushed limestone. For economic reasons, the pellets need to be much more attrition resistant than limestone to justify their use. The experimental results clearly show that this was not achieved. Furthermore, the pellets are highly sensitive to humidity, likely because of the hygroscopic nature of the cement and calcium hydroxide formation. Therefore, without significant improvements to the attrition and humidity resistance of the pellets, they are not a viable sorbent candidate.  Silica-coated pellets were found to have a high resistance to attrition in dry air at 500?C. After 24 hours of operation, compared to the coated sorbents, the lime mean diameter reduction and AJI were greater by factors of 2.4 and 3.4, respectively, and by factors of 2.1 and 2.6 for the calcined pellets. Pellets with 1 ?m and 5 ?m thick-coatings experienced nearly the same mean diameter reductions and production of fines. However, limited testing was performed on the silica coated pellets, and conditions such as humidity and variable particle size were not analyzed. Therefore, the silica-coated pellets should be considered as a tentative candidate, but full kinetic, economic, and environmental analyses are required to ascertain their true viability.  Both a new attrition testing standard and a mean diameter reduction model equation are proposed. The proposed testing apparatus should be constructed and used to determine the minimum column height and settling chamber length, based on the transport disengagement height of commonly tested materials such as FCC. Multiple materials should be tested and the Air Jet Index values compared to those produced in ASTM D5757 tests. The model equation should be tested with additional materials to evaluate the fragmentation requirement. Ideally, the proposed standard and the model would be evaluated simultaneously, with the results compared to previous attrition and comminution models.  Due to time and material restrictions in this project, sorbents were only subjected to a limited selection of operating conditions. Since the sorbents are intended for industrial carbonation/calcination cycling systems, they should be tested under reactive conditions to 70 include the pressure stress of the release of CO2 from within the particles. The temperature range should also be extended to 900?C, where sintering would likely affect the rate of attrition. To quantify the sintering, BET surface area tests should be conducted to relate attrition and changes in surface area. The ASTM unit is limited to small batch tests, while a large circulating system with a cyclone could be used to replicate an industrial environment. Additional attrition testing should also be conducted over a longer period of time to assess long-term attrition performance.  Overall, the goal of both this project and the entire CMC network is to reduce the emission of greenhouse gases in Canada, especially from fossil fuel power generation systems, to help in the worldwide effort to mitigate global climate change. Considering the ever-increasing role of the fossil fuel industry in Canada and Canadians? su stantial energy consumption, Canada has an obligation to perform research, develop technologies and propose solutions to reduce emission of carbon dioxide and other greenhouse gases.    71 REFERENCES  Abanades, J. C., & Alvarez, D. (2003). Conversion limits in the reaction of CO2 with lime. Energy & Fuels, 17(2), 308?315. doi:10.1021/ef020152a ASTM Standard D5757. (2006). Determination of Attrition and Abrasion of Powdered Catalysts by Air Jets. ASTM International, West Conshohocken, PA. ASTM Standard D6913. (2009). Particle-Size Distribution (Gradation) of Soils Using Sieve Analysis. ASTM International, West Conshohocken, PA. Basu, P. (2006). Combustion and Gasification in Fluidized Beds. Boca Raton, FL: CRC Press. Bemrose, C. R., & Bridgwater, J. (1987). A review of attrition and attrition test methods. Powder Technology, 49(2), 97?126. doi:16/0032-5910(87)80054-2 Bi, X. T. (2012, September 20). Onset of Fluidization. Notes in course CHBE 561 at the University of British Columbia. Chen, H., Zhao, C., Yang, Y., & Zhang, P. (2012). CO2 capture and attrition performance of CaO pellets with aluminate cement under pressurized carbonation. Applied Energy, 91(1), 334?340. doi:10.1016/j.apenergy.2011.09.032 Chen, Y.-M. (2003). Applications for Fluid Catalytic Cracking. Chapter 14 in W.-C. Yang (Ed.), Handbook of Fluidization and Fluid-Particle Systems. New York: Marcel Dekker. Chen, Z., Gibilaro, L. G., & Foscolo, P. U. (1997). Fluid pressure loss in slugging fluidised beds. Chemical Engineering Science, 52(1), 55?62. doi:10.1016/S0009-2509(96)00368-5 Chen, Z., Grace, J. R., & Lim, C. J. (2008). Limestone particle attrition and size distribution in a small circulating fluidized bed. Fuel, 87(7), 1360?1371. doi:16/j.fuel.2007.06.012 Chen, Z., Grace, J. R., & Lim, C. J. (2011). Development of particle size distribution during limestone impact attrition. Powder Technology, 207(1-3), 55?64. doi:16/j.powtec.2010.10.010 Chen, Z., Lim, C. J., & Grace, J. R. (2007). Study of limestone particle impact attrition. Chemical Engineering Science, 62(3), 867?877. doi:16/j.ces.2006.10.022 Cleaver, J. A. S., Ghadiri, M., & Rolfe, N. (1993). Impact attrition of sodium carbonate monohydrate crystals. Powder Technology, 76(1), 15?22. doi:10.1016/0032-5910(93)80036-A Cocco, R., Arrington, Y., Hays, R., Findlay, J., Karri, S. B. R., & Knowlton, T. M. (2010). Jet cup attrition testing. Powder Technology, 200(3), 224?233. doi:16/j.powtec.2010.02.029 72 Fan, L.-S. (2010). Chemical Looping Systems for Fossil Energy Conversions. Hoboken, N.J.: John Wiley & Sons, Inc. Grace, J. R. (2012, November 15). Entrainment/Carryover/Elutriation. Notes in course CHBE 561 at the University of British Columbia. Guo, X., Deng, Y., Gu, D., Che, R., & Zhao, D. (2009). Synthesis and microwave absorption of uniform hematite nanoparticles and their core-shell mesoporous silica nanocomposites. Journal of Materials Chemistry, 19(37), 6706. doi:10.1039/b910606e Gwyn, J. E. (1969). On the particle size distribution function and the attrition of cracking catalysts. AIChE Journal, 15(1), 35?39. doi:10.1002/aic.690150112 Haider, A., & Levenspiel, O. (1989). Drag coefficient and terminal velocity of spherical and nonspherical particles. Powder Technology, 58(1), 63?70. doi:10.1016/0032-5910(89)80008-7 Hartge, E.-U., Klett, C., & Werther, J. (2007). Dynamic simulation of the particle size distribution in a circulating fluidized bed combustor. Chemical Engineering Science, 62(1?2), 281?293. doi:10.1016/j.ces.2006.08.067 Johnsen, K., Ryu, H. J., Grace, J. R., & Lim, C. J. (2006). Sorption-enhanced steam reforming of methane in a fluidized bed reactor with dolomite as CO2-acceptor. Chemical Engineering Science, 61(4), 1195?1202. doi:16/j.ces.2005.08.022 Kimura, T., Horiuchi, K., Watanabe, T., Matsukata, M., & Kojima, T. (1995). Experimental study of gas and particle behavior in the grid zone of a jetting fluidized bed cold model. Powder Technology, 82(2), 135?143. doi:10.1016/0032-5910(94)02913-9 Lee, S. K., Jiang, X., Keener, T. C., & Khang, S. J. (1993). Attrition of lime sorbents during fluidization in a circulating fluidized bed absorber. Industrial & Engineering Chemistry Research, 32(11), 2758?2766. doi:10.1021/ie00023a044 Litster, J., & Ennis, B. (2004). Breakage and Attrition. Chapter 5 in The Science and Engineering of Granulation Processes. Dordrecht, Netherlands: Kluwer Academic Publishers. Lupi??ez, C., Scala, F., Salatino, P., Romeo, L. M., & D?ez, L. I. (2011). Primary fragmentation of limestone under oxy-firing conditions in a bubbling fluidized bed. Fuel Processing Technology, 92(8), 1449?1456. doi:16/j.fuproc.2011.03.005 Manovic, V. (2012, October 15). Discussion regarding initial uncalcined attrition results. Manovic, V., & Anthony, E. J. (2009a). Screening of binders for pelletization of CaO-based sorbents for CO2 capture?. Energy & Fuels, 23(10), 4797?4804. doi:10.1021/ef900266d 73 Manovic, V., & Anthony, E. J. (2009b). CaO-based pellets supported by calcium aluminate cements for high-temperature CO2 capture. Environmental Science & Technology, 43(18), 7117?7122. doi:10.1021/es901258w Merry, J. M. D. (1975). Penetration of vertical jets into fluidized beds. AIChE Journal, 21(3), 507?510. doi:10.1002/aic.690210312 Montagnaro, F., Nobili, M., Salatino, P., Telesca, A., & Valenti, G. L. (2008). Hydration products of FBC wastes as SO2 sorbents: comparison between ettringite and calcium hydroxide. Fuel Processing Technology, 89(1), 47?54. doi:16/j.fuproc.2007.06.014 Montagnaro, F., Salatino, P., & Scala, F. (2010). The influence of temperature on limestone sulfation and attrition under fluidized bed combustion conditions. Experimental Thermal and Fluid Science, 34(3), 352?358. doi:16/j.expthermflusci.2009.10.013 Mori, S., & Wen, C. Y. (1975). Estimation of bubble diameter in gaseous fluidized beds. AIChE Journal, 21(1), 109?115. doi:10.1002/aic.690210114 Neil, A. U., & Bridgwater, J. (1999). Towards a parameter characterising attrition. Powder Technology, 106(1?2), 37?44. doi:10.1016/S0032-5910(99)00064-9 Pires, J. C. M., Martins, F. G., Alvim-Ferraz, M. C. M., & Sim?es, M. (2011). Recent developments on carbon capture and storage: An overview. Chemical Engineering Research and Design, 89(9), 1446?1460. doi:10.1016/j.cherd.2011.01.028 Pis, J. J., Fuertes, A. B., Artos, V., Su?rez, A., & Rubiera, F. (1991). Attrition of coal ash particles in a fluidized bed. Powder Technology, 66(1), 41?46. doi:10.1016/0032-5910(91)80079-X Praxair. (2008). Praxair Specialty Gases References Guide. Praxair Technology, Inc. Retrieved from http://catalogs.praxairdirect.com/issue/25778 Ray, Y. C., Jiang, T. S., & Jiang, T. L. (1987). Particle population model for a fluidized bed with attrition. Powder Technology, 52(1), 35?48. doi:10.1016/0032-5910(87)80005-0 Ray, Y.-C., Jiang, T.-S., & Wen, C. Y. (1987). Particle attrition phenomena in a fluidized bed. Powder Technology, 49(3), 193?206. doi:16/0032-5910(87)80128-6 Rees, A. C., Davidson, J. F., Dennis, J. S., S Fennell, P., Gladden, L. F., Hayhurst, A. N., & Sederman, A. J. (2006). The nature of the flow just above the perforated plate distributor of a gas-fluidised bed, as imaged using magnetic resonance. Chemical Engineering Science, 61(18), 6002?6015. doi:10.1016/j.ces.2006.05.006 74 Reppenhagen, J., & Werther, J. (2000). Catalyst attrition in cyclones. Powder Technology, 113(1-2), 55?69. doi:10.1016/S0032-5910(99)00290-9 Rhodes, M. (2008). Particle Size Reduction. Chapter 12 in Introduction to Particle Technology (2nd ed.). John Wiley & Sons, Inc. Sarshar, Z., Sun, Z., Zhao, D., & Kaliaguine, S. (2012). Development of Sinter-Resistant Core?Shell LaMnxFe1?xO3@mSiO2 Oxygen Carriers for Chemical Looping Combustion. Energy & Fuels, 26(5), 3091?3102. doi:10.1021/ef3003137 Scala, F., Salatino, P., Boerefijn, R., & Ghadiri, M. (2000). Attrition of sorbents during fluidized bed calcination and sulphation. Powder Technology, 107(1-2), 153?167. doi:16/S0032-5910(99)00185-0 Schaller, R. (2000). Mechanical spectroscopy of the high-temperature brittle-to-ductile transition in ceramics and cermets. Journal of Alloys and Compounds, 310(1-2), 7?15. doi:10.1016/S0925-8388(00)01006-9 Shipway, P. H., & Hutchings, I. M. (1993). Attrition of brittle spheres by fracture under compression and impact loading. Powder Technology, 76(1), 23?30. doi:10.1016/0032-5910(93)80037-B Slimane, R. B., & Abbasian, J. (2000). Copper-based sorbents for coal gas desulfurization at moderate temperatures. Industrial & Engineering Chemistry Research, 39(5), 1338?1344. doi:10.1021/ie990877a Stanmore, B. R., & Gilot, P. (2005). Review?calcination and carbonation of limestone during thermal cycling for CO2 sequestration. Fuel Processing Technology, 86(16), 1707?1743. doi:10.1016/j.fuproc.2005.01.023 Sun, Z. (2013, August 16). E-mail describing silica sorbent preparation process. Thon, A., Puttmann, A., Hartge, E.-U., Heinrich, S., Werther, J., Patience, G. S., & Bockrath, R. E. (2011). Simulation of catalyst loss from an industrial fluidized bed reactor on the basis of labscale attrition tests. Powder Technology, 214(1), 21?30. doi:10.1016/j.powtec.2011.07.017 USGS. (2013). 2011 Minerals Yearbook: Crushed Stone. U.S. Geological Survey. Retrieved from http://minerals.usgs.gov/minerals/pubs/commodity/stone_crushed/myb1-2011-stonc.pdf Wen, C. Y., Deole, N. R., & Chen, L. H. (1982). A study of jets in a three-dimensional gas fluidized bed. Powder Technology, 31(2), 175?184. doi:10.1016/0032-5910(82)89003-7 75 Wen, C. Y., & Yu, Y. H. (1966). A generalized method for predicting the minimum fluidization velocity. AIChE Journal, 12(3), 610?612. doi:10.1002/aic.690120343 Werther, J., & Reppenhaggen, J. (2003). Attrition. Chapter 8 in W.-C. Yang (Ed.), Handbook of Fluidization and Fluid-Particle Systems. New York: Marcel Dekker. Werther, J., & Xi, W. (1993). Jet attrition of catalyst particles in gas fluidized beds. Powder Technology, 76(1), 39?46. doi:16/0032-5910(93)80039-D Wu, Y., Manovic, V., He, I., & Anthony, E. J. (2012). Modified lime-based pellet sorbents for high-temperature CO2 capture: Reactivity and attrition behavior. Fuel. doi:10.1016/j.fuel.2012.01.034 Xiao, G., Grace, J. R., & Lim, C. J. (2011). Attrition characteristics and mechanisms for limestone particles in an air-jet apparatus. Powder Technology, 207(1-3), 183?191. doi:16/j.powtec.2010.10.028 Xiao, G., Grace, J. R., & Lim, C. J. (2012). Limestone Particle Attrition in High-Velocity Air Jets. Industrial & Engineering Chemistry Research, 51(1), 556?560. doi:10.1021/ie201698r Yang, H., Xu, Z., Fan, M., Gupta, R., Slimane, R. B., Bland, A. E., & Wright, I. (2008). Progress in carbon dioxide separation and capture: A review. Journal of Environmental Sciences, 20(1), 14?27. doi:10.1016/S1001-0742(08)60002-9 Yang, W.-C. (2003). Bubbling Fluidized Beds. Chapter 3 in W.-C. Yang (Ed.), Handbook of Fluidization and Fluid-Particle Systems. New York: Marcel Dekker. Yates, J. G. (1996). Effects of temperature and pressure on gas-solid fluidization. Chemical Engineering Science, 51(2), 167?205. doi:10.1016/0009-2509(95)00212-X Zhao, R., Goodwin, J. G., & Oukaci, R. (1999). Attrition assessment for slurry bubble column reactor catalysts. Applied Catalysis A: General, 189(1), 99?116. doi:10.1016/S0926-860X(99)00261-6 Zhao, R., Goodwin, J.G., Jothimurugesan, K., Spivey, J. J., & Gangwal, S. K. (2000). Comparison of Attrition Test Methods: ASTM Standard Fluidized Bed vs Jet Cup. Industrial & Engineering Chemistry Research, 39(5), 1155?1158. doi:10.1021/ie990730j    76 APPENDIX A ? Introduction  Figures   Figure A1: Equilibrium diagram of limestone calcination/carbonation (adapted from Stanmore & Gilot, 2005).  Air SourceDrainFeed HopperFilter BagAttrition Chamber3 Preheaters (3 * 4 kW)Orifice PlateGas Superheater 0.0001 0.001 0.01 0.1 1 10 500 550 600 650 700 750 800 850 900 950 1000 Carbon Dioxide Pressure (atm) Temperature (?C) Calcination Region Carbonation Region 77 Figure A2: UBC impactor test apparatus.  CFBHeatersCFB ? eactor? ColumnCycloneAir Cooling ColumnCooling Water OutCooling Water InAir FilterExhaust3 Preheaters(3 * 4 kW)Orifice PlateAir SourceDrain Figure A3: UBC circulating fluidized bed attrition system.    78 APPENDIX B ? Experimental Methodology   Equations  Minimum Fluidization Velocity (Wen & Yu, 1966):   emf   C12 C2Ar C1 where:  C1     .7, C2   27.2, Ar   g?f(?p-?f)dp ?2 and Umf    emf??fdp  Terminal Velocity (Haider and Levenspiel, 1989):  ut     18dp 2 2.  5 1.744  dp 0.5  1  where: ut   ut     ? ?p-?f g 1   and dp    dp  ?f ?p-?f g?2 1    Initial Bubble Diameter (Mori & Wen, 1975):  DB0   1. 8g0.2 ANor U Umf  0.4  Vertical Jet Penetration Length (Merry, 1975):  L et   dor 5.2 ?g, eddor?pdp 0.  1.  uor2gdor 0.2 1    Experimental Procedure  Sample Preparation 1. Weigh out the appropriate amount of sample for the attrition test. For non-calcined tests, use 55 g. For calcined tests, ensure that there will be at least 55 g for attrition testing once all carbon dioxide and water are driven off the sample. Calcine the sample for at least 4 hours at 850?C to sure complete conversion, using a nitrogen purge to prevent hydroxide formation. Let the sample cool in the N2 purged oven until room temperature is reached to further minimize the rate of hydroxide formation. 2. Either perform sieve analysis or laser diffraction PSD analysis to determine the initial particle size distribution. 79  Unit Construction 1. Ensure that the fines collector is fixed in place and properly sealed. 2. Attach the attrition column with the appropriate distributor (jet or fluidizing) to the gas inlet tube. Initiate flow of a small amount of gas into the column and check for gas leaks. Turn off the gas. 3. Pour the sample into the attrition column and attach the settling chamber to the top of the attrition column. 4. Attach the tubing between the top of the settling chamber and the fines collector. o For high temperature tests, attach the primary clamshell heaters to the attrition column, and ensure all thermocouples are in their proper location and functional (see Figure B2). Insulate the unit as much as possible to prevent heat losses. Unit Operation 1. Turn on the main power to the unit. Set the temperature controllers to the appropriate temperatures. Turn on and set the data acquisition system to record the desired temperatures and the gas flow rate.  o For high temperature tests, turn on all three heaters, and allow 20 to 30 minutes for preheating (if heating to 500?C). 2. Open all necessary valves and set the flow controller to the desired flow rate. 3. After one hour, turn off the gas flow and remove the fines collector. Quickly collect the fines sample, record the mass and save the sample for PSD analysis. Reattach the fines collector and set the gas flow to the appropriate level. 4. After an additional 4 hours, tap on the unit for 10 minutes to dislodge any fines attached to the settling chamber and column walls. This may not be possible for high-temperature tests. Turn off the gas flow and heaters, and remove as much insulation as safely possible. Leave the unit for a few hours to cool. Do not allow additional gas flow through the unit for cooling, as this would cause further attrition of the sample. Unit Disassembly 1. Remove all insulation. Remove the tubing between the settling chamber and remove the settling chamber itself. Brush inside the tubing and chamber to dislodge and collect all adhering material. This material is considered as bed material, not fines. 2. Remove the attrition column and carefully collect the material supported by the distributor (bed material). Clean out the attrition column. Weigh the collected sample and perform sieve analysis, or save a small sample (a few grams) for laser diffraction PSD analysis. Save an additional small sample (approximately 0.5 g) for SEM tests. 3. Remove the fines collector and collect and weigh the sample. Save all material for laser diffraction PSD analysis.    80  Additional Test Cycles 1. Repeat all steps in the Unit Construction, Unit Operation, and Unit Deconstruction sections of this procedure, operating the unit for the desired length of time. The 1 hour initial fines collection is only needed for fresh samples. Once all testing is complete, save the remaining bed material for any future analysis. Ensure that all parts are fully cleaned.  Figures  Figure B1: Experimental air jet apparatus diagram.   81  Figure B2: Experimental air jet apparatus schematic.   82  Figure B3: Example experimental minimum fluidization velocity (Umf) curve, for calcined CANMET pellets (dp= 490 ?m).              -0.20 0.00 0.20 0.40 0.60 0.80 1.00 1.20 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 Pressure Drop (inches H2O) Gas Velocity (m/s) Ascending Gas Flow Descending Gas Flow 83 Table B1: Experimental Conditions Trial  Material Temperature (?C) Initial Sample Mass (g) Fluidizing Gas Gas Flow Rate (L/min)  @ 20?C, 1 atm Superficial Gas Velocity (m/s) Distributor Plate Approximate Initial Mean Diameter (?m) Total Test Run Time (h) 1 CANMET Pellets 20?3 55.0 Humid Air 10.0 0.173 Jet 1000 36 2 CANMET Pellets 20?3 55.0 Humid Air 54.0 0.935 Fluidization  1000 24 3 Limestone 20?3 55.0 Humid Air 68.0 1.178 Fluidization 1000 24 4 Limestone 20?3 55.0 Humid Air 10.0 0.173 Jet  1000 24 5 Calcined CANMET Pellets 20?3 55.0 Humid Air 10.0 0.173 Jet  1000 36 6 Calcined Lime 20?3 55.0 Humid Air 10.0 0.173 Jet  1000 36 7 Calcined CANMET Pellets 20?3 55.0 Dry Air 10.0 0.173 Jet  1000 36 8 Half-Calcined Lime 20?3 55.0 Dry Air 10.0 0.173 Jet  1000 36 9 Calcined Lime 20?3 55.0 Dry Air 10.0 0.173 Jet  1000 36 10 Calcined Lime 20?3 55.0 Dry Air 10.0 0.173 Jet  1000 24 11 Calcined CANMET Pellets 500?5 55.0 Dry Air 10.0 0.457 Jet  1000 24 12 Calcined Lime 500?5 55.0 Dry Air 10.0 0.457 Jet  1000 24 13 Calcined CANMET Pellets 20?3 55.0 Dry Air 10.0 0.173 Jet  500 36 14 Calcined Lime 20?3 55.0 Dry Air 10.0 0.173 Jet  500 36 15 Calcined CANMET Pellets 20?3 55.0 Dry Air 29.2 0.506 Fluidization 500 24 16 Calcined Lime 20?3 55.0 Dry Air 30.8 0.534 Fluidization 500 24 17 Calcined Lime 20?3 55.0 Dry Air 10.0 0.173 Jet  1000 36 18 Calcined Lime 500?5 55.0 Dry Air 3.9 0.173 Jet  1000 24 19 Calcined 5 ?m-Shelled CANMET Pellets w/o Binder 500?5 55.0 Dry Air 10.0 0.457 Jet  750 24 20 Calcined 1 ?m-Shelled CANMET Pellets w/o Binder 500?5 55.0 Dry Air 10.0 0.457 Jet 750 24 21 Calcined Pellets w/o Binder 500?5 55.0 Dry Air 10.0 0.457 Jet 750 24  Table B2: ASTM Standard Conditions Temperature (?C) Initial Sample Mass (g) Fluidizing Gas Gas Flow Rate (L/min)  @ 0?C, 1 atm Superficial Gas Velocity (m/s) Distributor Plate Initial Mean Diameter Range (?m) Total Test Run Time (h) 0 50.0 Air, 30-40% relative humidity 10.0 0.173 Jet 10 - 180 5 84 APPENDIX C ? Attrition Test Results  Particle Size Distribution Plots    Figure C1: Differential and cumulative particle size distribution plots for Trial 1: CANMET, 20?C, 1000 ?m, humid, jet. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 85   Figure C2: Differential and cumulative particle size distribution plots for Trial 2: CANMET, 20?C, 1000 ?m, humid, fluidized bed.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 86   Figure C3: Differential and cumulative particle size distribution plots for Trial 3: limestone, 20?C, 1000 ?m, humid, fluidized bed.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 87   Figure C4: Differential and cumulative particle size distribution plots for Trial 4: limestone, 20?C, 1000 ?m, humid, jet.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 88   Figure C5: Differential and cumulative particle size distribution plots for Trial 5: calcined CANMET, 20?C, 1000 ?m, humid, jet.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 89   Figure C6: Differential and cumulative particle size distribution plots for Trial 6: lime, 20?C, 1000 ?m, humid, jet.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 90   Figure C7: Differential and cumulative particle size distribution plots for Trial 7: calcined CANMET, 20?C, 1000 ?m, dry, jet.   0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 91   Figure C8: Differential and cumulative particle size distribution plots for Trial 8: half-calcined lime, 20?C, 1000 ?m, dry, jet.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 92   Figure C9: Differential and cumulative particle size distribution plots for Trial 9: lime, 20?C, 1000 ?m, dry, jet.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 93   Figure C10: Differential and cumulative particle size distribution plots for Trial 10: lime, 20?C, 1000 ?m, dry, jet, replicate #1.   0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 94   Figure C11: Differential and cumulative particle size distribution plots for Trial 11: calcined CANMET, 500?C, 1000 ?m, dry, jet.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 95   Figure C12: Differential and cumulative particle size distribution plots for Trial 12: lime, 500?C, 1000 ?m, dry, jet.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 96   Figure C13: Differential and cumulative particle size distribution plots for Trial 13: calcined CANMET, 500?C, 500 ?m, dry, jet.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 97   Figure C14: Differential and cumulative particle size distribution plots for Trial 14: lime, 500?C, 500 ?m, dry, jet.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 98   Figure C15: Differential and cumulative particle size distribution plots for Trial 15: calcined CANMET, 500?C, 500 ?m, dry, fluidized bed.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 99   Figure C16: Differential and cumulative particle size distribution plots for Trial 16: lime, 500?C, 500 ?m, dry, fluidized bed.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 100   Figure C17: Differential and cumulative particle size distribution plots for Trial 17: lime, 20?C, 1000 ?m, dry, jet, replicate #2.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 36 h 101   Figure C18: Differential and cumulative particle size distribution plots for Trial 18: lime, 500?C, 500 ?m, dry, low-flow jet.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 102   Figure C19: Differential and cumulative particle size distribution plots for Trial 19: 5 ?m shelled, 500?C, 750 ?m, dry, jet.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 103   Figure C20: Differential and cumulative particle size distribution plots for Trial 20: 1 ?m shelled, 500?C, 750 ?m, dry, jet.  0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 12 h 24 h 104   Figure C21: Differential and cumulative particle size distribution plots for Trial 21: calcined no-binder CANMET, 500?C, 750 ?m, dry, jet.    0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total Collected Fines <45 45-106 106-250 250-425 425-595 595-850 850-1180 1180-1400 >1400 Weight Percent Particle Sizes, ?m 0 h 5 h 105 Attrition Test Results Summary Plots with Standard Error Bars   Figure C22: Total mass mean diameter reduction vs. time for the uncalcined jet and fluidized bed attrition tests for up to 24 h at 20?C, with standard error bars.   Figure C23: Air Jet Index vs. time for the uncalcined jet and fluidized bed attrition tests for up to 24 h at 20?C, with standard error bars. 0.0% 1.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% 8.0% 0 5 10 15 20 25 30 Total Mass Mean Reduction Time (h) CANMET, 1000 ?m, Humid, Jet (Trial 1) CANMET, 1000 ?m, Humid, F. B. (Trial 2) Limestone, 1000 ?m, Humid, F. B. (Trial 3) Limestone, 1000 ?m, Humid, Jet (Trial 4) Calcined CANMET, 500 ?m, Dry, F. B. (Trial 15) Lime, 500 ?m, Dry, F. B. (Trial 16) 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% 3.0% 3.5% 0 5 10 15 20 25 30 Air Jet Index Time (h) CANMET, 1000 ?m, Humid, Jet (Trial 1) CANMET, 1000 ?m, Humid, F. B. (Trial 2) Limestone, 1000 ?m, Humid, F. B. (Trial 3) Limestone, 1000 ?m, Humid, Jet (Trial 4) Calcined CANMET, 500 ?m, Dry, F. B. (Trial 15) Lime, 500 ?m, Dry, F. B. (Trial 16) 106  Figure C24: Total mass mean diameter reduction vs. time for calcined particle jet tests for up to 36 h at 20?C, with standard error bars.   Figure C25: Air Jet Index vs. time for calcined particle jet tests for up to 36 h at 20?C, with standard error bars. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 5 10 15 20 25 30 35 40 Total Mass Mean Diameter Reduction Time (h) Lime, 1000 ?m, Humid, Jet (Trial 6) Lime, 1000 ?m, Dry, Jet (Trial 9) Lime, 500 ?m, Dry, Jet (Trial 14) Half Calcined Lime, 1000 ?m, Dry, Jet (Trial 8) Calcined CANMET, 1000 ?m, Humid, Jet (Trial 5) Calcined CANMET, 1000 ?m, Dry, Jet (Trial 7) Calcined CANMET, 500 ?m, Dry, Jet (Trial 13) 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 5 10 15 20 25 30 35 40 Air Jet Index Time (h) Lime, 1000 ?m, Humid, Jet (Trial 6) Lime, 1000 ?m, Dry, Jet (Trial 9) Lime, 500 ?m, Dry, Jet (Trial 14) Half Calcined Lime, 1000 ?m, Dry, Jet (Trial 8) Calcined CANMET, 1000 ?m, Humid, Jet (Trial 5) Calcined CANMET, 1000 ?m, Dry, Jet (Trial 7) Calcined CANMET, 500 ?m, Dry, Jet (Trial 13) 107  Figure C26: Total mass mean diameter reduction vs. time for calcined particle jet tests for up to 24 h at 500?C, with standard error bars.   Figure C27: Air Jet Index vs. time for calcined particle jet tests for up to 24 h at 500?C, with standard error bars. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 5 10 15 20 25 30 Total Mass Mean Diameter Reduction Time (h) Lime, 1000 ?m, Dry, Jet (Trial 12) Lime, 1000 ?m, Dry, Low Flow Jet (Trial 18) Calcined CANMET, 1000 ?m, Dry, Jet (Trial 11) 5 ?m Shelled, 750 ?m, Dry, Jet (Trial 19) 1 ?m Shelled, 750 ?m, Dry, Jet (Trial 20) Calcined No-Binder CANMET, 750 ?m, Dry, Jet (Trial 21) 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 5 10 15 20 25 30 Air Jet Index Time (h) Lime, 1000 ?m, Dry, Jet (Trial 12) Lime, 1000 ?m, Dry, Low Flow Jet (Trial 18) Calcined CANMET, 1000 ?m, Dry, Jet (Trial 11) 5 ?m Shelled, 750 ?m, Dry, Jet (Trial 19) 1 ?m Shelled, 750 ?m, Dry, Jet (Trial 20) Calcined No-Binder CANMET, 750 ?m, Dry, Jet (Trial 21) 108 APPENDIX D ? Attrition Modeling  Attrition Modeling Coefficients  Table D1: Gwyn (1969) Model Results Trial Number Ka?103 (s-b) b (unitless) R2 (Coefficient of Determination) Trial 1 0.120 0.704 0.993 Trial 2 0.549 0.675 0.999 Trial 3 2.52 0.664 0.998 Trial 4 0.965 0.477 0.959 Trial 5 54.5 0.784 0.987 Trial 6 33.8 0.949 0.978 Trial 7 0.416 2.004 0.991 Trial 8 0.556 1.306 0.922 Trial 9 0.053 2.206 0.887 Trial 10 8.29 0.944 0.989 Trial 11 158.6 0.469 0.911 Trial 12 11.7 1.340 0.999 Trial 13 64.5 0.640 0.996 Trial 14 17.7 1.071 0.967 Trial 15 0.386 0.820 0.999 Trial 16 0.278 0.900 0.954 Trial 17 30.9 0.434 0.888 Trial 18 1.32 1.141 0.999 Trial 19 25.2 0.702 0.967 Trial 20 28.9 0.725 0.982                     109 Table D2: Jet and Bed Model Results Trial Number Cjet (s2/m3) Cbed (1/m2) Trial 1 8.77E-09  Trial 2  9.15E-05 Trial 3  3.56E-04 Trial 4 1.75E-08  Trial 5 4.10E-05  Trial 6 4.29E-05  Trial 7 1.03E-05  Trial 8 5.72E-07  Trial 9 8.17E-07  Trial 10 3.99E-06  Trial 11 1.13E-05  Trial 12 7.62E-06  Trial 13 1.65E-05  Trial 14 3.87E-06  Trial 15  4.95E-04 Trial 16  2.60E-04 Trial 17 1.27E-06  Trial 18 1.97E-06  Trial 19 4.38E-07  Trial 20 6.16E-07   Table D3: Population Momentum Model Results Trial Number Non-Linear Equation Order, n km ? 10-3 (-%/h) for n = 2 Linear Model (n = 2) R2 Trial 1 6.214 1.3 0.038 Trial 4 39.27 0.9 0.078 Trial 5 1.73 148.5 0.978 Trial 6 2.11 119.2 0.942 Trial 7 0.21 36.8 0.039 Trial 8 2.75 20.3 0.884 Trials 9/10/17 1.94 33.3 0.922 Trial 11 2.83 243.8 0.979 Trial 12 0.59 102.9 0.310 Trial 13 2.35 83.4 0.926 Trial 14 1.72 80.4 0.907 Trial 18 2.31 12.6 0.962 Trial 19 7.48 49.1 0.559 Trial 20 10.86 34.5 0.338  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0074219/manifest

Comment

Related Items