UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The kinetics of the flash converting reaction of MK (chalcocite) concentrate Morgan, Grant John 1994

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

Item Metadata


831-ubc_1994-0566.pdf [ 2.95MB ]
JSON: 831-1.0078613.json
JSON-LD: 831-1.0078613-ld.json
RDF/XML (Pretty): 831-1.0078613-rdf.xml
RDF/JSON: 831-1.0078613-rdf.json
Turtle: 831-1.0078613-turtle.txt
N-Triples: 831-1.0078613-rdf-ntriples.txt
Original Record: 831-1.0078613-source.json
Full Text

Full Text

THE KINETICS OF THE FLASH CONVERTING REACTIONOF MK (CHALCOC1TE) CONCENTRATEByGrant John MorganB.Sc.(Eng.), Queen’s University at Kingston, 1987A THESIS SUBMITED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF APPLIED SCIENCEinTHE FACULTY OF GRADUATE STUDIES(Department of Metals and Materials Engineering)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAOctober, 1994© Grant John Morgan, 1994In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of cLycttsThe University of British ColumbiaVancouver, CanadaDate H /gI’DE-6 (2/88)ABSTRACTABSTRACTThe chemical kinetics and mechanism of the flash converting reaction of MK concentrate(approximately 93% chalcocite and 7% pentlandite) in oxygen containing atmospheres weredetermined. The reaction profile of individual particles undergoing flash oxidation in alaminar flow furnace was followed by observing the particle temperature and changes inapparent particle diameter with a fast response two-wavelength pyrometer; and theseobservations were compared to the predictions of a mathematical model of a single reactingparticle. Flash reaction products were examined under a scanning electron microscope todetermine chemical composition and particle morphology. A campaign of pilot plant trialswas also conducted to further study the flash converting process.The initial oxidation of chalcocite at lower particle temperatures was determined to bechemically rate-controlled, to follow the Arrhenius rate law and to be first order with respectto oxygen partial pressure. It was established, by comparison of experimental particleignitions results with the predictions of a mathematical model, that the activation energy ofthe low-temperature chemically rate-controlled reaction was 460 Id mol’.A dust generation mechanism was elucidated whereby preferential evaporation of copperfrom the surface of the particle leaves a nickel-rich shell encasing the particle. Continuedoxidation of sulphides and metallic copper leads to ongoing particle heating and, uponreaching the boiling point of copper, the generation of copper vapour which either ventsthrough a rupture or causes catastrophic fragmentation of the nickel-rich shell.A pilot plant campaign, conducted at the UBC mini pilot plant facility, tested variousconfigurations of the concentrate-oxygen injector, and it was determined that the flash flame11ABSTRACTcould be manipulated to decrease the dust generation rate. A vertical injector design thatintroduced the oxygen as a high velocity axial jet surrounded by freely falling MKconcentrate gave dust generation rates that were as low as one-third of the dust generationrates produced in trials utilizing injector designs that introduced the concentrate and oxygenin a single stream.111TABLE OF CONTENTSTABLE OF CONTENTSABSTRACT iiTABLE OF CONTENTS ivLIST OF FIGURES viLIST OF TABLES viiiLIST OF NOMENCLATURE ixACKNOWLEDGEMENTS xi1 INTRODUCTION 11.1 Flash Converting of Chalcocite 11.2 Two-Wavelength Pyrometry 42 LiTERATURE REVIEW 72.1 Flash Smelting and Converting Processes 72.2 Flash Reaction Studies 82.2.1 Flash Smelting Furnace Models 82.2.2 Experimental Studies of Sulphide Oxidation 112.3 Two-Wavelength Pyrometry Measurements 213 SCOPE AN]) OBJECTIVES 344 EXPERIMENTAL 364.1 MK Concentrate Characterization 364.1.1 Chemical and Mineralogical Composition 364.1.2 Particle Size Classification 374.2 Combusting Particle Temperature Measurements 404.2.1 Laminar Flow Furnace 414.2.2 Two-Wavelength Pyrometer 44ivTABLE OF CONTENTS4.3 Photomicrographic Analysis of Reaction Products 464.4 Pilot Plant Studies 474.4.1 UBC Flash Smelting Pilot Plant Facility 474.4.2 MK Flash Converting Pilot Study 505 RESULTS 655.1 Particle Combustion Temperatures 655.2 Mathematical Modeling of Reaction Kinetics 685.3 Photomicrographic Analysis of Reaction Products 715.4 Pilot Plant Results 745.4.1 Steady-State Furnace Temperatures 745.4.2 Desuiphurization and Nickel Deportment 775.4.3 Reactor Gas Analyses 815.4.4 Dust Generation Rate 836 DISCUSSION 1016.1 Particle Combustion Temperatures 1016.2 Mathematical Modeling of Reaction Kinetics 1046.3 Photomicrographic Analysis of Reaction Products 1066.4 Pilot Plant Studies 1097 SUMMARY AND CONCLUSIONS 115REFERENCES 117APPENDIX A.I: TWO-WAVELENGTH PYROMETRY 122A.I. 1 Particle Temperature Measurement 122A.I.2 Apparent Particle Diameter Measurement 125APPENDIX A.ll: HEATING OF CARRIER GAS STREAM 127VUST OF FIGURESLIST OF FIGURES1.1: Inco Copper Cliff Operations as of 1991 51.2: Proposed Inco Copper Cliff Operations 62.1: Schematic Diagram of Inco Flash Furnace 232.2: Schematic Diagram of Outokumpu Flash Furnace 242.3: Schematic Diagram of Kivcet Flash Smelting Process 252.4: Predicted Convective and Radiative Heat Transfer To/From 25 .tm Particle 262.5: Schematic Diagram of Experimental Apparatus of Kim and Themelis 272.6: Arrhenius Plot of Chemical and Mass Transfer Controlled Reaction Rates 282.7: Copper-Oxygen Phase Diagram 292.8: Typical Type Al Particle Combustion Pulse 302.9: Typical Type A2 Particle Combustion Pulse 312.10: Typical Type A3 Particle Combustion Pulse 322.11: Typical Type A4 Particle Combustion Pulse 334.1: MK Particle Size Analysis by Ortech Ltd 544.2: MK Particle Size Distribution by Cyclosizing 554.3: Photomicrograph of 24-34p.m Fraction of MK Concentrate 564.4: Photomicrograph of 24-331.Lm MK Partially Reacted Under Near-IgnitionConditions 574.5: Schematic Diagram of Laminar Flow Furnace 584.6: Schematic Diagram of Pyrometer and Data Acquisition System 594.7: Type Al Combustion Pulse for Pyrrhotite, from Tuffrey 604.8: Type Al Combustion Pulse for Pyrrhotite Obtained in This Work 614.9: Schematic Diagram of Pilot Plant Flash Reactor 62viUST OF FIGURES4.10: Pilot Plant Burner Configurations.634.11: Schematic Diagram of Gas Sampling Apparatus 645.1: Type Al Combustion Pulse, Maximum at Chalcocite Melting Point 855.2: Type Al Combustion Pulse, Maximum at Copper Boiling Point 865.3: Type A2 Combustion Pulse for MX Concentrate 875.4: Type A3 Combustion Pulse Without Temperature Increase 885.5: Type A3 Combustion Pulse With Temperature Increase 895.6: Type A4 Combustion Pulse for MK Concentrate 905.7: Combustion Pulse Under Highly Oxidizing Conditions 915.8: Predicted Ignition Threshold Conditions Using Data From Kim and Themelisand Experimental Data 925.9: Predicted and Mean Experimental Ignition Threshold Conditions 935.10: Particle Temperature Profile Predicted by Model of Shook 945.11: Extent of Desulphurization by Flash Oxidation 955.12: Photomicrograph of Sub-Ignition Reaction Product 965.13: Photomicrograph of Perforated Cenosphere 975.14: Photomicrograph of Perforated Cenosphere 985.15: Depletion of Copper by Flash Oxidation 995.16: Photomicrograph of Collapsed Cenosphere 1006.1: Peak Particle Temperatures Determined from Pyrometer Measurements 1136.2: Calculated Copper-to-Nickel Vapour Pressure Ratio 114viiUST OF TABLESLIST OF TABLES2.1: Combustion Pulse Characterization 224.1: Selected MK Concentrate Head Assays 374.2: MK Concentrate Size Assay 384.3: Particle Size Distribution by Cyclosizing 394.4: Pilot Plant Trial Conditions 525.1: Energy Dispersive Analysis of MK Flash Reaction Products 725.2: Steady State Pilot Furnace Temperature 765.3: Furnace Gas Temperatures by Suction Thermocouple 765.4: Desulphurization Along Pilot Plant Reactor Shaft 785.5: Nickel Deportment in Pilot Plant Reactor 805.6: Oxygen Injection Velocity and Mass Deportment to Wall 805.7: Gas Chromatographic Analyses of Furnace Gas 825.8: Dust Generation Rates for Pilot Plant Trials 84viiiUST OF NOMENCLATURELIST OF NOMENCLATURESymbol DefinitionA Arrhenius pre-exponential constant (mol m2 atm1 s’)A area of particle (m2)surface area (m2)B interaction parameter for Chapman-Enskog relationc velocity of light (3x108m s’)C1 first radiation constant (3.74xlO6W m2)C2 second radiation constant (1.438x102m K)d. diameter of particle (m)diffusivity of oxygen in bulk gas (m2 sjEa activation energy (3 mold)detector-particle view factorpyrometer signal amplifier gain settingh Planck constant (6.6262xlO J s)I collision integral for Chapman-Enskog relationk Boltzmann constant (l.3806x1023J K.!)icr chemical reaction rate constant (m2 atm s moI1)M12 mean molecular weight (g mold)partial pressure of oxygen (atm)radius of particle (m)R gas constant (8.3143 mor1K4)ixUST OF NOMENCLATURER chemical reaction rate (mol s1)mass transfer reaction rate (mol s1)Re Reynolds numberSc Schmidt numberSh Sherwood numberT temperature (K)Tf gas film temperature (K)V, pyrometer output voltage (Volts)thermal diffusivityWb intensity of monochromatic radiation from a blackbody (W m3)W,, intensity of monochromatic radiation from a non-blackbody (W m3)mono-chromatic emissivity of a non-blackbodywavelength (m)Pauling electronegativityxACKNOWLEDGEMENTSACKNOWLEDGEMENTSI thank my advisors, Dr. J. Keith Brimacombe and Dr. Greg G. Richards, for their guidanceand insights, as well as my colleagues, Dr. Andrew A. Shook and Dr. Nigel E. Tuffrey, fortheir ground-breaking work and advancement of the science of flash smelting. I must alsothank Mr. Pat Wenman and Mr. Rudy Cardeno for their expertise in the design, maintenanceand operation of the UBC flash converting pilot plant. Dr. Carlos Diaz of Inco offeredinvaluable direction to the project, as well as continuing interest in the results.My greatest debt of gratitude is to my wife, Shannon Gillin, for her perseverance and supportthroughout my research.Financial support for this work was provided by Inco Ltd., the Natural Sciences andEngineering Research Council, and CANMET.x1.1 Flash Converting of Chalcocite1 INTRODUCTION1.1 Flash Converting of ChalcociteThe process of flash smelting for the treatment of sulphide ores and concentrates waspioneered in the 1940’s by the International Nickel Company of Canada and Outokumpu Oyof Finland. These two groups developed the flash smelting process independently, both fortreatment of copper concentrates (chalcopyrite). The process is still used primarily fortreatment of copper concentrates; indeed more copper concentrates are treated in flashfurnaces than by any other process [1]. Flash smelting is also used in the processing ofnickel concentrates at Kalgoorlie Nickel Mines in Australia [2] and other sites, and in theprocessing of lead concentrates via the Kivcet Process [3-5] at Portovesme, Sardinia, and aproposed facility at the Cominco operations in Trail, British Columbia [6].The copper and nickel production process at the Inco Copper Cliff smelter (the process flowsheet as of 1991 is shown schematically in Figure 1.1) treated the metal sulphidesindividually after an initial separation of the bulk concentrate by flotation [7]. Suchseparations are invariably incomplete, so that the nickel concentrate contains significantlevels of copper. The nickel concentrate was smelted progressively in multiple-hearthroasters, reverberatory furnaces and Pierce-Smith converters. The product, a Bessemermatte, was slow cooled, crushed and ground then again processed by flotation to separatecopper and nickel suiphides. The resultant copper sulphide concentrate, a chalcociteconcentrate termed “MK” in the industry, was flash converted to semi-blister copper andfinally oxidized in Pierce-Smith converters.11.1 Flash Converting of ChalcociteRevisions to the process were examined and evaluated in order to meet the more stringentsulphur dioxide gas emissions regulations laid out in the Ontario Countdown to Acid RainProgram. One such process revision was proposed by Landolt et al [7], shown in Figure 1.2,which called for elimination of the initial copper-nickel separation by flotation; rather thebulk concentrate would be flash smelted and converted to Bessemer matte. After subsequentseparation, the nickel concentrate would be treated as before, but the MK concentrate wouldbe flash converted to semi-blister copper containing approximately 1% sulphur, and fmallyreduced in a novel top blown finishing converter developed by Inco.Flash converting was viewed as a potential alternative to Pierce-Smith converting becausethe off-gas generated in a flash reaction has a more consistent and substantially moreenriched SO2 content, thereby facilitating sulphur dioxide scrubbing. The consistent andenriched SO2 production is due to the fact that flash converting is a continuous process runwith tonnage oxygen. In contrast, Pierce-Smith. converters run on a batch cycle with air oroxygen enriched air as the oxidant [8].The major problem encountered in applying flash smelting technology to the converting ofchalcocite is the generation of considerable amounts of finely divided dust [9, 10]. The dustmust be separated from the reactor off-gas prior to sulphur dioxide scrubbing, requiring anadditional process step and handling equipment. Therefore, if excessive dust generation ratescannot be avoided, there is a significant increase in the capital cost of the process plant.Indeed, the problem of dust generation led to the abandonment of flash converting in favourof submerged injection of chalcocite and oxygen in a conventional Pierce-Smith converter;the process currently used at Inco Copper Cliff.21.1 Flash Convening of ChalcociteAs part of the evaluation of the flash converting option, an extensive research project wasundertaken, specifically to characterize the flash “flame” and the reaction products. Shook[11] developed a mathematical model of the flash converting furnace to study the dustgeneration mechanism within the flame. A major component of that model was a calculationof the reaction kinetics-- based largely on data generated by Otero, Brimacombe andRichards [12]. The latter researchers studied the reaction of chalcocite and copperconcentrate (chalcopyrite) in simulated flash conditions.The work described in this thesis was undertaken to determine the kinetics governing theflash reaction of chalcocite by studying the reaction in an environment virtually identical tothat of a flash furnace. A two-wavelength pyrometer was used to observe the thresholdconditions for particle ignition and the combustion temperature of particles reacting in alaminar flow furnace; a range of furnace temperatures and oxygen concentrations wasapplied. These results were compared to predictions made by the kinetic model developedby Shook [11] in order to determine the activation energy and pre-exponential constant in theArrhenius expression for the low temperature, chemically controlled oxidation of chaicocite.The reaction products were examined in a scanning electron microscope to determinechemical composition and particle morphology. This information, together with thechemical kinetics, was used to confirm, and refine, the reaction mechanism proposed byShook.Concurrent to this work, a pilot plant campaign was run to investigate alternate burnerconfigurations and their effect on the flash reaction; particularly the extent ofdesuiphurization and the dust generation rate.312 Two-Wavelength Pyrometry1.2 Two-Wavelength PyrometryThe measurement of the combustion temperatures of very small particles in free fail, as is ina flash furnace, must be done in a non-invasive, non-contact manner. This can only beaccomplished through the use of some form of radiation pyrometry. Of the three generalforms of radiation pyrometry -- total radiation, brightness and two-wavelength pyrometry[12] -- only the last is applicable to particles in a flash smelting furnace. Total radiationpyrometry is not applicable to the study of flash reacting particles as it would requireconfinement of the reacting particle within a photosensitive sphere and brightness pyrometryis not applicable, as it would require knowledge of the emissivity and surface area of theparticle.Two-wavelength pyrometry (also called two-colour or ratio pyrometry) works on theprinciple that the intensity of monochromatic radiation emitted by an object is a function ofboth the temperature of the object and of the monochromatic wavelength considered. Atwo-wavelength radiation pyrometer collects two narrow bands of virtually monochromaticradiation emitted by an object, and the temperature of the object is determined as a functionof the ratio of the intensity of radiation at the two wavelengths; independent of the particlesize. The only incalculable assumption is that the ratio of the emissivity of the particle at thetwo wavelengths chosen is temperature independent. An explanation of the theory oftwo-wavelength pyrometry is given in Appendix I.Two-wavelength pyrometry has become the preferred method for studying the combustion ofsmall particles, and instruments have been constructed, calibrated and successfullyimplemented by researchers studying the combustion of coal particles and suiphide oreparticles [13-25].41 INTRODUCTIONBulk concentrate Sulfur dioxide______plant________I iFlash furnaceSSulfuric addLiquid SOconverters i Market— ,.Bessemer matteI Copper sulfide I I Flash BlisterConverter capperIMatte___ ___ __Finish toSeparation1__________ingI [L converter RefineryNickel sulfides & metaltics- MarketLoasters Nickel oxide and- refineriesFIGURE 1.1: Inco Copper Cliff Operations as of 1991(from Landolt et al [7])51 INTRODUCTIONNI concentrate Sutfur dioxide Co concentrateTo liquidJ411Rash furnace —Reverbs .101411111 COflVrte(’S BlisterConverters 111+” Flash toi I converter ()ppeRefinerySeParatL:LJ‘I I Nickel sulfides &fl allicsandI JflRoasters__________refineriesFIGURE 1.2: Proposed Inco Copper Cliff Operations(from Landolt et al [71)62.1 Flash Smelting and Converting Processes2 LITERATURE REVIEW2.1 Flash Smelting and Converting ProcessesThere are three different configurations of flash furnace used in the smelting of sulphide oresand concentrates; namely the Inco, Outokumpu and Kivcet designs.The Inco Flash Smelting furnace design employs horizontal injection, feeding concentrateand oxygen into both ends of a long furnace. Slag and matte are tapped from the reactor andthe off-gas is vented from a centrally located stack. The reactor configuration is shown inFigure 2.1. The reactor has been used by Inco for over forty years to smelt chalcopyrite andpyrite ores, and has been extensively investigated by pilot plant and modeling studies[1,6,10].The first copper smelter put into production using the Outokumpu flash smelting process wasat Harjavalta, Finland in 1949 [26]. Since then, the Outokumpu flash smelting process hasbecome the most widely used process for the smelting of copper concentrates (chalcopyrite).The process was subsequently adapted to treat nickel concentrates, with the first Outokumpunickel flash smelter put into production in 1959, also at Harjavalta, Finland. As of 1993,there were 38 licensees world wide using Outokumpu flash smelters in the production ofcopper and nickel. A schematic diagram of the Outokumpu flash reactor is presented inFigure 2.2.The Outokumpu flash furnace design uses a top entry burner, as opposed to the Inco flashfurnace which uses horizontal burners. In the Outokumpu flash furnace, the solids aredispersed at the point of entry by a double-entry burner design such that the concentrate72.2.1 Flash Smelting Furnace Modelsdescends the furnace shaft as a virtual rain of particles with little variation in the oxidizingenvironment (temperature and partial pressure of oxygen) across the shaft. This mode ofentry facilitates the combustion of copper and nickel concentrates, but leads to excessive dustgeneration when applied to the flash smelting of chalcocite and lead concentrates. The Incoburner design generates a flash reaction region that more closely resembles a true flame, withthe most oxidizing conditions at the periphery of the flame [11].The Kivcet process [3-5] is a viable single vessel lead production process developed in theSoviet Union and used successfully at Portovesme, Sardinia. A Kivcet reactor is also beingconstructed at the Trail, British Columbia operations of Cominco Ltd. [6]. A schematicdiagram of the Kivcet process is shown in Figure 2.3. In the Kivcet process, lead sulphideconcentrate is flash smelled with a stoichiometric excess of oxygen and a molten lead/leadoxide mixture is produced. Reduction to lead metal is then performed in an adjacent electricarc furnace. The use of excess oxygen leads to greater dust generation by fuming in the flashsmelting of some materials, but in the Kivcet Process, the recycle of non-reactive lead oxidedust indroduces a significant heat sink, such that the flame temperature is lowered, thereaction rate is retarded and excessive dust generation is not observed in the flash flame [3].Moreover, the entrainment of furnace gas, principally SO2, by the jet of feed oxygen lowersthe oxygen partial pressure in the vicinity of the reacting particles, thus lowering the reactionrate and the dust generation rate.2.2 Flash Reaction Studies2.2.1 Flash Smelting Furnace ModelsMathematical models of the different flash smelting processes have been published.822.1 Flash Smelting Furnace ModelsThese models encompass a wide range of approaches in terms of the assumptions andsimplification made regarding the fluid flow within the reactor and the chemical reactionkinetics.The Outokumpu chalcopyrite flash smelter is the most thoroughly studied flash reactor,having been modelled by Ruotto [27], Hahn and Sohn [28-30] as well as by Kim andThemelis [31].The model of Ruotto [27] went to great lengths to predict the flow of the gas and particleswithin the reactor accurately. However, the treatment of thermodynamics, temperaturegradients and increasing particle temperatures was comparatively weak because of theassumptions that the gas was incompressible and that the particles and gas were always atthermal equilibrium. Consequently, numerous empirical parameters had to be introducedto fit the model to experimental data; this approach diminishes the validity of the modeLHahn and Sohn [28,29] and Seo and Sohn [30] developed a model of the Outokumpureactor that was mathematically complex in its treatment of the fluid flow and heattransfer. One assumption made was that radiation was the dominant mode of heattransfer to the particles, an assumption that may not be correct. The work of Shook [11],modelling the chalcocite flash reaction, showed that convective/conductive heat transferbetween the gas and particles dominates, as shown in Figure 2.4, which shows thepredicted convective and radiative heat fluxes to/from a 25j.i.m particle undergoing flashreaction with a furnace temperature of 1250K. Shook also cited the work of Walsh et al[32] and Kolb et al [33] who determined that convective/conductive heat transfer is the92.2.1 Flash Smelting Furnace Modelsdominant mode of heat transfer to coal-water slurry droplets in a test chamber.Moreover, the Hahn and Sohn model does not consider the phenomena of particledisintegration, agglomeration or dust generation.Kim and Themelis [31] developed a one-dimensional model of the flash reaction shaft.The apparently simple model was strengthened by valid assumptions, and the modelsuccessfully predicted the particle and gas compositions and temperatures. Kim andThemelis combined their furnace model and reaction kinetics findings to develop amechanism for particle fragmentation.Jorgensen and Elliot [2] used the PHOENICS fluid dynamics calculation package tomodel the burner and shaft of the Kalgoorlie Nickel Smelter. The objective of theirmodelling was to examine the effect of burner configuration on the distribution ofconcentrate and gas, and the resulting effect on the combustion within the furnace shaft.Successful predictions of furnace temperature and particle dispersion were made; andperhaps the most valuable result of the modeling was the ability to predict the effects ofchanges in feed concentrate composition.The model developed by Shook [11] was specific to the flash converting of chalcocite(MK concentrate) as proposed by Inco. This model utilized the boundary layer form ofthe gas phase conservation equations. Model calculations included particle trajectoriesand gas phase fluid flow (including gas recirculation), radiative and convective heattransfer to and from the particles, temperature gradients within the gas phase andassociated gas properties, increasing particle temperature and the resultant variations in102.2.2 Experimental Smdies of Suiphide Oxidationchemical reaction kinetics and products. The model was verified by comparison to theresults of MK flash converting runs in the Inco, Port Colborne pilot reactor as well as inthe UBC mini pilot plant.Assumptions were made by Shook regarding the chemical reaction mechanism and thereaction kinetics; assumptions based largely on the work of Otero, Brimacombe andRichards [12] and Kim and Themelis [34]. Otero et al used a stagnant gas reactor toexamine the weight loss on ignition of chalcocite particles in various furnace atmospheresover a range of temperatures; and the experiments of Kim and Themelis looked at theoxidation kinetics of compressed chalcocite pellets. As part of the development of thefurnace model, a kinetic model of a single particle falling through a stagnant gas furnacewas written to model the work of Otero et al. This model is also applicable, with minormodifications, to individual particles reacting in a laminar flow furnace.2.2.2 Experimental Studies of Suiphide OxidationMost studies of copper sulphide oxidation kinetics have examined the reactions at thesurface of molten copper suiphides and copper oxides (as is the case in Pierce-Smithconverting) or have attempted to simulate flash reaction conditions with the combustionof a compressed pellet of chalcocite in a heated gas stream. These experiments, by theirvery nature, depart from the real conditions of individual reacting particles experienced inthe flash furnace, but provide some valuable insights.Glen and Richardson [35] studied the oxidation of magnetically levitated liquid copperdroplets, and among their findings was the effect of surface active species, such as silica,112.2.2 Experimental Studies of Suiphide Oxidationin retarding the oxidation reaction without reducing the rate of copper evaporation. Theyobserved and quantified the preferential evaporation of copper metal, leaving the dropletsenriched in the minor species present.Ajersch and Toguri [36] and the subsequent related work of Ajersch and Benlyamani [37]investigated the oxidation rates of liquid copper and liquid copper suiphide in a capillarytube sample holder, with argon gas containing set oxygen concentrations flushed acrossthe open end of the capillary. They were able to determine the diffusion coefficients forsulphur dioxide and oxygen in argon, and to establish that the oxidation of chalcociteproceeds through the intennediate State of metallic copper before complete oxidation tocopper oxide. Their attempt to evaluate the chemical kinetics of the lower temperaturereaction was less successful. Their value of the activation energy for the reactionCu2S+0—2Cu+S0 [2.1]was found to be 657 kJ molt,which is significantly higher than other values in theliterature. Ajersch and Toguri acknowledged that their silica crucible was dissolving intothe molten copper and molten copper suiphide samples, but there was no recognition thattheir experiments were thus biased by the effect of the silica retarding the oxidationreaction, as was proven by Glen and Richardson [35] for molten copper.Alyaser [3 8,39] studied the kinetics of the oxidation of molten chalcocite baths duringtop-lancing. The reaction occurred in two stages. The first stage involved partialdesulphurization accompanied by the dissolution of oxygen into the liquid sulphide, andthe rate was oxygen mass transport controlled. Upon saturation with dissolved oxygen,metaffic copper and sulphur dioxide gas were produced by bath and surface reactions that122.2.2 Experimental Studies of Suiphide Oxidationwere electrochemical in nature; sulphur dioxide bubbles were observed to erupt from thesurface of the melt during the second phase. The conditions at the point of transitionfrom oxygen dissolution to electrochemical oxidation were characterized by the sulphurcontent falling below approximately 18 weight percent with an oxygen content ofapproximately 1.3 to 1.5 weight percent; the precise conditions dependent upon melttemperature and oxygen partial pressure. The difference of the Paulingelectronegativities (ItXAO) for the copper-sulphur couple (0.68) indicates that thecopper-sulphur bond is approximately 12% ionic in character [40], indicating that it ispossible for electrochemical reactions to occur in copper suiphide melts.Henderson [411, Lewis et al [42], Wadsworth et at [43], Rao and Abraham [441 and Asakiet at [45] roasted compressed pellets of chalcocite and attempted to determine the reactionkinetics by the pellet temperature increase or the sulphur dioxide generation rate. Theirexperiments were largely inconclusive as a result of undependable temperaturemeasurements due to rapid heating and large temperature gradients within the pellet.Moreover, the effect of the growing copper oxide product layer, impeding oxygentransport to the chalcocite core, caused oxygen mass transport to be the rate determiningfactor rather than chemical kinetics.Studies of the ignition temperature and the extent of desulphurization during the flashreaction of sulphide minerals were reported by Jokilaakso et al [46], Sohn et al [47] andDunn and Mackey [48], though chalcocite was not one of the minerals tested. Jokilaaksoet at and Sohn et al each used a laminar flow furnace to flash sulphide mineral particles;the reaction products were collected and analyzed to determine the extent of reaction andignition temperature as functions of the furnace temperature and furnace gas composition.132.2.2 Experimental Studies of Suiphide OxidaticeWhile this method is useful in that it simulates a flash furnace and may indicate thecomposition of the concentrate stream along the flash furnace shaft, it cannot be used todetermine single particle reaction characteristics. The cold nitrogen carrier gas used tofacilitate concentrate feeding (1.4% of the total volumetric gas flow rate) would surroundthe particle during the initial particle heating phase, the temperature range in which theflash reaction rate is normally controlled by chemical kinetics. This invalidates anyattempt to calculate particle temperature and near-particle gas composition, and thusnegates any prediction of single particle chemical reaction kinetics or gas phase oxygenmass transfer rate.Kim and Themelis [34] conducted one of the best studies of the rate of oxidation ofcompressed pellets of copper and other metal sulphides in an air stream. The chalcocitesample was ground to -200 mesh (<74p..m) and formed into compressed cylindricalpellets of various porosities. The pellets were then suspended in a nitrogen flushedfurnace and the gas stream flowing through the furnace was switched to air to initiate theexperiment. The apparatus of Kim and Themelis is shown in Figure 2.5. The change insulphur dioxide content of the off-gas during the initial few seconds of the experimentwas determined by differential thermal conductivity; thus the initial rate of oxidation wasestablished.Thermocouples embedded in the centre of the pellets registered temperature increases onthe order of 1 Kelvin per second due to the exothermicity of the reaction. Thetemperature increase suggests that the rate governing phenomenon would at some pointchange from chemical kinetics to mass transfer control, since as the particle temperatureincreases, the rate of the chemical reaction becomes more rapid and the rate determining1422.2 Experimental Studies of Suiphide Oxidationstep in the process becomes the transport of oxygen to the reacting surface. Indeed, thetransition was observed to occur between 1068 and 1097 K. Kim and Themelis statedthat the mass transfer resistance was diffusion of oxygen to the pellet through a diffusionboundary layer in the gas phase. However, as the experiments progressed to theirduration of 15 minutes, transport of oxygen through the reaction products would also beof considerable importance.The initial oxidation rate data follows first order kinetics with respect to oxygen partialpressure.R = k,A5P02 [2.2]The chemical reaction rate constant is described by the Arrhenius equation.kr=Aexp(_4) [2.3]An Arrhenius plot of the initial reaction rate for the oxidation of chalcocite versus thereciprocal temperature gave an activation energy for the chemically rate controlled region(T < 1070 K) of 525 Id mor1,and an effective activation energy for the mass transfer ratecontrolled region of 48 kJ mol’. Kim and Themelis did not state a value for theArrhenius pre-exponential constant (A) for the rate expression in the chemicallycontrolled region, but Shook [11] used the data of Kim and Themelis to estimate thepre-exponential constant at 2.52(l0’) mol m2 atm’ s1. The Arrhenius plot of thechemically controlled and mass transfer controlled reaction rates produced by Kim andThemelis is recreated in Figure 2.6.1522.2 Experimental Studies of Suiphide OxidationIn the calculation of the pre-exponential constant using Equation [2.2], based onexperimental data such as that of Kim and Themelis, the reacting surface area ofchalcocite must be estimated. In the case ofcompressed cylindrical pellets the value ofthe surface area is not precisely known, because there is an effective reacting surface areasomewhat greater than the surface area of the cylinder, due to the rapid transport ofoxygen into the surface pores.Moreover, initial reaction rate data reflect the concurrent oxidation of the sub-surfacechalcocite material, which would be mass-transfer rate controlled by the rate of ingress ofoxygen into the porous compressed pellet. Figure 2.6 is an Arrhenius plot showing thetemperature dependence of the chemically controlled and mass transfer controlledreaction rates determined by Kim and Themelis, as well as the mass transfer controlledreaction rate predicted by the model of Shook. At the initial pellet temperatures of theexperiments in the work of Kim and Themelis (963-1216 K), the role of the mass transferrate controlled reaction would be secondary, but not trivial, in determining the rate ofsulphur dioxide production, and thus the empirically determined reaction rate.It is useful to note that the apparent mass transfer activation energy obtained by theexperimental work of Kim and Themelis (48 U mol’) is greater than that predicted by themodel of Shook (2.6 U molt). Shook’s calculation of the mass transfer rate was basedon the Ranz-Marshall expression [49,50] to determine the mass transfer rate constant fortransport of oxygen to the particle (k0).Sh= k0d=2 + O.6ReSc”3 [2.4]02,b1622.2 Experimental Swdies of Suiphide OxidationThe diffusivity of oxygen was predicted by the Chapman-Enskog relation [511.B7MD02b = ,212 [2.51Another examination of chalcocite flash reaction kinetics was by Otero, Brimacombe andRichards [12], in which pulses of chalcocite particles were released into a stagnant gasreactor at set temperature and furnace gas composition. In this work, MK concentrateparticles were observed to ignite in different nitrogen-oxygen gas mixtures at furnacetemperatures of 1073 to 1173 K with variations in the particle size showing no effect.Combustion of copper concentrate (chalcopyrite) and MK particles was photographed,and the MK particles were seen to explode more violently, generating small particlefragments which continued to react. Photomicrographs of the flash reaction productsshowed that both chalcocite and chalcopyrite concentrate particles melted during reaction,but there was a greater tendency toward the formation of hollow cenospheres with theejection of material during the combustion of chalcopyrite; the cenospheres were high iniron content, indicating that perhaps the iron in the chalcopyrite, when oxidized, held theparticle together.Severe dust generation in the combustion of MK concentrate was observed as a mass lossbeyond the expected stoichiometric desulphurization, especially with furnace gas oxygenconcentrations in excess of 50%. The dust generation was explained as the result ofparticle explosion fragments being ejected outside the boundaries of the sample collector.Another likely source of mass loss is the evaporation of molten copper as described byGlen and Richardson [35]. The evaporation of copper from the hot particle into therelatively cooler furnace gas would be enhanced by condensation in the steep temperature172.2.2 Experimental Studies of Suiphide Oxidationgradient near the particle surface, thus reducing the near-surface copper partial pressureand generating a greater driving force for evaporation. Jorgensen determined that coppervapourized during flash reactions can be in the form of monomeric copper, cupric orcuprous oxide [15], and will readily oxidize as finely divided Cu20 so that it may nothave been collected by the sampler used in the work of Otero et al.The mechanism of the flash reaction of chalcocite was elucidated by Shook [11]. Theproposed mechanism is:1. At temperatures less than the melting point of Cu2S, the dominant oxidationreaction is:Cu2S()+ l.5O) = Cu2O()+ SO) [2.6]The reaction rate is governed by chemical kinetics at particle temperatures belowapproximately 1400K, and by oxygen mass transport above temperatures of 1400K.This reaction is highly exothermic, so the particle temperature rises rapidly and thechemically controlled reaction rate accelerates.2. At the melting point of Cu2S, the mobility of the sulphide increases, as does itsreactivity so that the heterogeneous reaction of the oxide with the remainingsulphide can occur according to the reaction:Cu2S(,) + 2CuO(3)= 6Cu(,) + SO) [2.7]182.22 Experimental Studies of Suiphide OxidationThis reaction has a finite rate at the melting point of Cu2S, as determined byByerley et al [52], but the oxide would be rapidly consumed due to the fact that lessthan 5% of the Cu2S would have reacted to form Cu20, compounded by the everincreasing particle temperature which would further accelerate the reaction rate.3. Molten copper suiphide reacts with oxygen to form metallic copper directly by thereaction:Cu2S(I) + O) = 2Cu(,) + SO) [2.8]The rate of this reaction is assumed to be controlled by the rate of oxygen masstransport to the surface until the point when all suiphide is consumed.4. The molten copper reacts with oxygen by one of two possible routes according tothe analysis of the Cu-O system done by Schmid [53], and the Cu-O phase diagramgiven as Figure 2.7, showing the miscibility limit of oxygen in liquid copper as afunction of oxygen partial pressure.(i) at furnace gas oxygen concentrations below the miscibility limit, oxygendissolves in the liquid copper:O) =2[°lCu [2.9](ii) at higher oxygen concentrations, the oxygen reacts with the copper to formCu20via Equation [2.10]:2Cu(,) + 0.5O) = Cu2O() [2.10]192.22 Experimental Studies of Sulpliide OxidationThe chemical reaction in Equation [2.101 is highly exothermic, and it proceeds untilall liquid copper is consumed either by oxidation or by evaporation due to boiling.The rate determining step of the chemically rate controlled reaction between molten Cu2Sand oxygen (Equation [2.8]) was established by the work of Morland et al [54], whostudied the kinetics of the exchange of radioactive S35 between sulphur dioxide gas indynamic equilibrium with mixed copper sulphide melts. The rate determining step in thereaction could be any one of the following processes:(i) adsorption of oxygen onto the sulphide surface as monatomic oxygen(ii) formation of the adsorbed intermediate complex SO(iii) reaction of the intermediate SO complex with another adsorbedmonatomic oxygen to form adsorbed sulphur dioxide(iv) desorption of SO2 gas.Morland et al showed that the rate limiting step was the formation of the surface-adsorbedactivated complex, SO.Shook proposed two mechanisms of dust generation [11]. The first mechanism consistsof the processes of particle heating due to the exothermic oxidation reactions, theproduction of metallic copper that subsequently reaches its boiling point, vapourization ofcopper then direct oxidation of the vapour or recondensation followed by oxidation. Thismechanism accounts for the generation of very finely divided copper oxide dustsuspended in the off gas stream.The non-volatile components of the MK concentrate (e.g., nickel oxide) were alsodetected in the dust, although in greatly depleted quantities with respect to theirconcentration in the feed material. Therefore, a second dust generation method was202.3 Two-Wavelength Pyrometiy Measurementsproposed, that being the production of finely divided dust by particle fragmentation. Itwas speculated that the rapid generation of copper vapour within the particle would resultin a catastrophic explosion producing many small fragments. This hypothesis issupported by the photographic evidence of particle explosions obtained by Otero [12].Otero used high-speed photography to capture images of particle explosions in whichparticle fragments were seen radiating from a point source.2.3 Two-Wavelength Pyrometry MeasurementsThe application of two-wavelength pyrometry to combusting particles was pioneered in thestudy of coal burners [20], and first applied to flash reactions by Jorgensen [14,15].Jorgensen measured the average temperature of masses of combusting particles, rather thanthe temperatures of individual combusting particles. Jorgensen observed reaction mechanicssuch as cenosphere formation, particle fragmentation by explosion and agglomeration ofmolten particles in a concentrate stream.A sensitive two-wavelength pyrometer, integrated with a fast data acquisition system, wasdesigned and built by Tuffrey [22,23] to study flash smelting of lead and iron suiphide oresand concentrates. In his work, the combustion of individual particles in a laminar flowfurnace was observed; and enough detail was seen to discern and interpret the combustioncharacteristics of various suiphidic materials.Tuffrey identified four distinct and characteristic combustion pulse types. These are definedin Table 2.1; and examples taken from Tuffrey of the temperature profiles giving rise to thefour types of combustion pulses are given as Figures 2.8 through 2.11. All combustionpulses presented in the literature by Tuffrey were obtained at a furnace temperature of212.3 Two-Wavelength Pyrometry Measurements1130K.Tuffrey used the measured particle ignition temperature, maximum combustion temperatureand apparent area calculations to speculate on the reaction mechanisms and mechanicalphenomena occurring during the combustion of lead and iron suiphide particles.Table 2.1: Combustion Pulse Characterization (after Tuffrey [22])Pulse CharacteristicsPulse Energy Temperature • MineralType • Physical PhenomenonAl • constant and low • 1400- 1700 K • typical of galenaintensity • near furnace • reaction products trailing• low level “tail” temperature particleA2 • 1st stage: <1800-2350 K, heating • pyrite and pyrrhotiteslow rise rate 2-6(10) K s• 2nd stage: 2500-2600 K peak • 5-10 fold increase insudden 5-10 fold increase temperature apparent area• 3rd stage: 2600-1600 K, cooling • particle coolingrapid decrease rate 8-20( 10) K s1A3 very short, high intensity 1800-2500 K, rapid • lead suiphide concentrates,pulse increase and decrease in pyrite and pyrrhotitemay be superimposed on temperature • material ejection, orother type pulses particle disintegrationA4 periodic variation, 1600-2 100 K, similar to • lead sulphide concentratesrelatively constant Al, temperature • periodic variation inperiodicity variations may or may particle area and/ornot be present temperature due to:- area pulsation- particle rotation- pulsating flame front222 LITERATURE REVIEWCHALCOPYRITE$ AN0C0 14 CEN TRATECHALCOPYRITE-- CONCENTRATES c4 SANOCONSTAW OXYGENWEIGHTFEEDERS- - —CONVERTEROXYG8(-SLAG RECYCLESLAG •• .-•. —SLAG MATTE MATTEFIGURE 2.1: Schematic Diagram of Inco Flash Furnace, from Landolt [7]232 LITERATURE REVIEWCon cenkateburnersUp take)shaftReoctaonshaftI / /1/ / / / / /1/ /Gas SettlerJL/L/...L/Lstg /L//JL/JLJL1J L’J L/JMatte Tappkghole/‘Ii’’’ / / / / / / / / / / / 1/ / 1/17/I/I /Topping holeFIGURE 2.2: Schematic Diagram of Outokumpu Flash Furnace, from Shook [11]242 LITERATURE REVIEWDRYINGFIGURE 2.3: Schematic Diagram of Kivcet Flash Smelting Process, from Reimers [5]GASLEAD CONCENTRATEAND FLUXESSMELT) NGGASCLEANINGSLAGREDUCTIONANDZINC FUMING252 LITERATURE REVIEW0.0140.012I0.0100.0080.006a,0.004a,g 0.002• 0.000a,-0.002-0.004-0.0062500 3000FIGURE 24: Predicted Convective and Radiative Heat Transfer To/From 25.tm Particle0 500 1000 1500 2000Particle Temperature (K)262 LITERATURE REVIEWFIGURE 2.5: Schematic Diagram of Experimental Apparatus of Kim and Themelis [34]Ar02N2so2FLow meter272 LiTERATURE REVIEW40Kim & Themetis - Chem20 Kim & Themelis - M.T.Shook - M.T.U,0E.S -20.2 -400C). -60C-J-80-1000 0.0005 0.001 0.0015 0.002Reciprocal Temperature (Ilk)FIGURE 2.6: Arrhenius Plot of Chemical and Mass Transfer Controlled Reaction Rates, asdetermined by Kim and Themelis [34] and predicted by Shook [11]282 LiTERATURE REVIEW4 4-5 - -5 -3 4444 4 I 44 4 4 — 4 4 4:———.44 4eRet-2813 1343 44 I‘ 44 I‘R(3J4‘ (I QRe(31A2i II4 4R132I 4, /- I jRe(33D / 4444a 4444 0t34I -- ai lI 44444 O1(4 L III II 4,. / 14414 4 I 7.OlolmlRe(27144/‘ 0 \ )j1OtW441I / ‘ ‘ 44 I4j ‘ ‘a 404 0 44I Is_P1O k4€ I I Ii - - 4l$4 I0‘ o0 031t II1 4‘III10 4I I4- II 4 otp24.4&m-012038I3t. Cd3.O2(g -65fbnvo4u (Eu) tu0Cu 1)1 02 03 (ijt) 04 (ii)l4e tro1gy OXy.ljFIGURE 2.7: Copper-Oxygen Phase Diagram (from Schmid [531) showing oxygenmiscibility limit as function ofP02 and temperature.292 LITERATURE REVIEW2.‘ 1400C•0C120010000 5 10 15 20Time (ms)(b)FIGURE 2.8: Typical Type Al Particle Combustion Pulse, from Tuffrey [22](a) Pyrometer Output Voltage(b) Calculated Particle TemperatureSample: Galena, Furnace Temp = 1130K, P0 = 0.21 atm,Particle Size = 63-74 jim0 5 10 15Time (ms)(a)302 LiTERATURE REVIEWI(V5 •10 15Thie (ms)2086420C180016001400120010000 5 10 20Time (ms)FIGURE 2.9: Typical Type A2 Particle Combustion Pulse, from Tuffrey [22](a) Pyrometer Output Voltage(b) Calculated Particle TemperatureSample: Sullivan Mine Lead Concentrate, Furnace Temp = 1130K,P02 = 0.21 atm, Particle Size = 105-125 urn1531‘52 LiTERATURE REVIEW10864200 5 10lime (ms)20EFIGURE 2.10: Typical Type A3 Particle Combustion Pulse, from Tuffrey [22](a) Pyrometer Output Voltage(b) Calculated Particle TemperatureSample: Pyrrhotite, Furnace Temp = 1130K, P02 = 0.21 atm,Particle Size = 74-88 tm0 5 10Time (ms)15 2032(Ua£(U(UV2 LITERATURE REVIEW108642000 5 10 15 2025 30Time (ms)0 5 10 15 20 25 30Time (ms)FIGURE 2.11: Typical Type A4 Particle Combustion Pulse, from Tuffrey [22](a) Pyrometer Output Voltage(b) Calculated Particle TemperatureSample: Sullivan Mine Lead Concentrate, Furnace Temp = 1130K,P02 = 0.21 atm, Particle Size = 63-74 .tm333 SCOPE AND OBJECTIVES3 SCOPE AND OBJECTIVESIt has been shown in published work that flash converting of chalcocite using existing flashreactor technology results in the generation of excessive amounts of finely divided dust[9,10]. Shook [11] speculated on the mechanism of dust generation by modeling the flashconverting flame, basing the reaction rates on kinetic data published in the technical literatureby Otero, Brimacombe and Richards [12] and Kim and Themelis [34].Shook was not able to present authoritative physical evidence to support his proposed dustgeneration mechanism. Moreover, the model calculations were based on the kineticparameters determined by Kim and Themelis [34] from experiments that reacted compressedpellets of chalcocite in a heated gas stream, rather than using true flash conditions.Therefore, the objective of this study was to develop a mechanistic model of the combustionof chalcocite particles in a flash reaction, by defining the reaction mechanism, kineticparameters and mechanical phenomena regulating the reaction; thus a better understanding ofthe mechanism of dust generation in the flash converting of chalcocite would be achieved.A two-wavelength pyrometer was used to observe chalcocite particles undergoing flashoxidation in a laminar flow furnace. The reaction progress was followed by observing andrecording the particle temperature during combustion, changes in apparent area and thethreshold conditions for particle ignition. The effects of concentrate particle size, furnace gastemperature and oxygen partial pressure were observed. Reaction products were examined ina scanning electron microscope, to determine product morphology and composition.343 SCOPE AND OBJECTIVESA mathematical model that predicted the rate and extent of combustion for a singleconcentrate particle reacting in a laminar flow furnace with known gas composition andfurnace temperature was employed to determine the constants for an Arrhenius rateexpression and the order of the chemical kinetics with respect to oxygen.A pilot plant campaign of tests was conducted to evaluate the influence on the flashconverting reaction of the oxygen-to-concentrate ratio, furnace temperature and theconcentrate and oxygen dispersion patterns. Primary attention was paid to the resultingdesuiphurization and dust generation rates. Various burner designs were examined, eachintended to offer different configurations of reactant injection and thereby to alter the flashflame characteristics and dust generation rate.354.1.1 Chemical and Mineralogical Composition4 EXPERIMENTAL4.1 MK Concentrate CharacterizationMK concentrate is the copper bearing fraction of crushed and ground Bessemer matte whichis separated from the nickel bearing fraction by flotation. The origin of the name MK isunclear even to those in the industry, but it is believed to be an acronym of a Germandesignation.The MK concentrate used in all laminar flow reactor experiments and pilot plant trials wassupplied by Inco Copper Cliff Operations. Chemical analyses of the concentrate and sizedistribution data were provided by Inco Research Labs, Mississauga Ontario.4.1.1 Chemical and Mineralogical CompositionPrior to each pilot plant run, the concentrate was re-assayed to ensure consistencybetween barrels. The Inco MK concentrate is typically 70% copper, 21% sulphur (assuiphide), 3-4% nickel, and less than 1% iron. Other elements are also present in traceamounts. Selected MK concentrate head assays are presented in Table 4.1, with theassays reported in mass percent. The variation in sulphur content (a gauge of chemicalreactivity) from run to run is not appreciable; generally within 1%. The concentrate usedfor all of the particle combustion tests was from Sample 1, for which the composition islisted in Table 4.1.364.1.2 Particle Size ClassificationTABLE 4.1 Selected MK Concentrate Head Assays (mass percent)Sample Cu S Ni 0 Fe(%) (%) (%) (%) (%)1 72.7 20.3 4.80 1.73 0.302 67.9 19.9 4.58 1.21 0.243 71.2 19.2 5.80 1.07 0.27The calculated mineralogical composition of MK concentrate is over 93% chalcocite(Cu2S), with approximately 5 to 7% pentlandite ({ Fe,Ni }9S8). The oxygen content of theconcentrate can be attributed to oxides contained in the Bessemer matte and subsequentoxidation of the chalcocite and pentlandite during flotation or storage.4.1.2 Particle Size ClassificationA size analysis of the MK concentrate was commissioned by Inco and performed byOrtech Ltd, employing a slurry turbidity technique. This analysis, given as Figure 4.1,shows an 80% passing size (P80) of 39p.m, and a mean particle size of approximately1 8p.m. Size assay of the feed shows a reasonable consistency of composition between the+30 and -30p.m size fractions. The size assay is given in Table 4.2, with compositionstated as mass percent.374.1.2 Particle Size ClassificationTABLE 4.2 MK Concentrate Size Assay (mass percent)Sample Cu S Fe(%) (%) (%)Unsized 68-73 19-21 <0.50over 30pm 73.4 21.3 0.03under 30p.m 71.1 18.6 0.32For the purpose of this investigation, it was not only necessary to determine the sizedisthbution of the material, but also to separate the feed into different size fractions. Sizeclassification of micron-sized particles can be done either by wet seiving or bycyclosizing. The cyclosizing technique was adopted.The cyclosizer consists of a series of five hyrdocyclones of progressively increasingpitch. The particulate material to be processed is slurried as a low density suspension inwater and slowly released into the chain of the cyclones, carded in a stream of water at ahigh volumetric flow rate. The five cyclones retain progressively finer size fractions ofthe particulate material, with a sixth and finest fraction passing in the effluent water. Thecyclones are calibrated with a material of known size distribution and specific gravity;and four multiplication factors, based on the water temperature, water flow rate, particlespecific gravity and the duration of the test, are employed to determine the particle size ofthe fractions retained. The specffic gravity of the MK concentrate was determined bywater displacement to be 5518 kg m3, and the other factors were determined or set so thatthe particle size fractions retained were as reported in Table 4.3. The size distribution by384.1.2 Pafficle Size Classificationcyclosizing is given in Figure 4.2.TABLE 4.3: Particle Size Distribution by CyclosizingSize % Retained Cumulative %_Passing(pm) Test 1 Test 2 Test 3 Test 4 Test I Test 2 Test 3 Test 4>33 26 24 23 26 ----24-33 33 31 32 33 74 76 77 7417-24 12 12 13 13 41 44 44 4112-17 10 11 11 10 30 32 32 289-12 5 6 6 6 19 21 21 189* 14 15 15 12 14 15 15 12*smallest fraction not collected, mass calculated by difference.The particle size analysis by cyclosizing is comparable to that provided by Ortech Ltd.with a similar P80, but shows a somewhat tighter disthbution. Particle shape factors playa large role in slurry turbidity size analyses, like that performed by Ortech; thus theparticle size distribution tends to be smeared. Figure 4.3, a photomicrograph of the 24-33tm fraction of the feed material shows that the MK concentrate particles are irregular inshape, which would affect a particle size analysis by slurry turbidity. Size classificationby cyclosizing, however, is by hydrodynamic drag, thus particle shape has a lesssignificant role than particle size. By allowing sufficient test duration, particles areseparated into discrete size fractions.For the purpose of process modeling, it is often assumed that particles are spherical. Themodel of Shook [11] makes this assumption. While the particles, which are formed bygrinding the Bessemer matte, are in fact angular (as shown in the photomicrograph of the24-33p.m fraction shown in Figure 4.3) the spherical assumption is justified. Calculation394.2 Combusting Particle Temperature Measurementsof the mass transfer of oxygen to the angular particle is approximated by the mass transferto a spherical particle, since the shape of the gas phase diffusion boundary layer canreasonably be approximated as a sphere. Moreover, as the flash reaction progresses, theparticles reach the melting point of chalcocite (1403 K), and become truly spherical.Figure 4.4 shows a photomicrograph of 24-33.tm particles after passing through thelaminar flow furnace under conditions near the threshold for particle ignition; this figureshows the partially reacted particles as spheres.The combustion characteristics of all of the size fractions collected by cyclosizing of theMK concentrate were examined in the laminar flow furnace. The discussion in this thesiscentres primarily on the analysis of the 24-33,tm fraction, because it is this fraction whichbest represents the mean particle size of the concentrate as determined by cyclosizing,25pm (Figure 4.2). Moreover, the combustion pulses generated by the 24-33p.m particlesdisplayed the most favourable signal-to-noise ratio, due to the greater total radiativeintensity offered by the larger particles.4.2 Combusting Particle Temperature MeasurementsParticle reactions were observed in a laminar flow furnace designed and built by Tuffrey[22,23] to study the flash smelting of sulphide concentrate. A two-wavelength pyrometerwas used to record combustion pulses, and the reaction temperature and change in apparentdiameter were calculated from the intensity of the detected radiation.404.2.1 Laminar Flow Furnace4.2.1 Laminar Flow FurnaceA schematic diagram of the laminar flow furnace is shown in Figure 4.5. The furnaceconsists of an insulated mild steel shell enclosing two annular compartments separated bya 438mm x 50mm i.d. axial reactor tube. Located in the outer annulus are three siliconcarbide globar heating elements. An additional heat source was required at the furnaceexit to avoid axial temperature gradients, so a 50mm i.d. chromel-wound resistancefurnace was installed. Three Pt/Pt- 10% Rh (type S) thermocouples, positioned axially,confirmed a uniform furnace temperature; measurements of the reactor tube walltemperature varied by no more than 15 K along the reaction zone.The oxygen/nitrogen reactant gas from separate compressed gas cylinders was meteredthrough rotameters and combined prior to entry into the furnace. The outer annulus actsas a pre-mix and pre-heat region, so that the gas entered the inner reaction tube at thedesired furnace temperature. The gas entered the inner reaction tube flowing downward,and although the furnace is not sufficiently long to allow laminar flow to develop fully,the gas flow along the axis of the reactor, where the particles are introduced,approximates laminar flow.The concentrate was introduced by a vibratory feeder into a water cooled feed tube. Noeffort was made to meter the concentrate feed rate, but the vibratory feeder was adjustedto establish a feed rate on the order of ten particles per second. The feed tube itself wasalso vibrated to avoid hold-up of particles and potential blockage. A low volumetric flowof reactant gas was passed through the feed tube to further ensure that the particles did414.2.1 Laminar Flow Furnacenot agglomerate or adhere to the inner wall of the feed tube. Exiting the feed tube theparticles entered along the axis of the reactor tube into a stream of preheated gas undervirtually laminar flow conditions.The pyrometer view port is situated at the bottom of the inner reaction tube, above thebottom resistance furnace. The height of the feed tube is adjustable, such that the reactionzone or “flame front” can be adjusted to be in front of the pyrometer view port. Thelift-off distance, or the distance between the feed tube exit and the pyrometer view port, isa function of the reaction kinetics and reactor geometry, but cannot be readily used tointerpret the reaction kinetics because of the effect of the carrier gas surrounding theparticles which extends the particle heating time, or incubation time.The experimental procedure involved first equilibrating the furnace and reactant gas at aset temperature, thereby establishing the oxidizing conditions for the test. The sizedconcentrate was then introduced and the resulting combustion pulses obtained by thepyrometer were recorded. This procedure was repeated for reactant gas compositionsranging from 5 to 100% oxygen, for all of the particle size fractions collected bycyclosizing, and through a range of furnace temperatures giving sub-ignition to veryreactive conditions.An early series of experiments attempted to determine the incubation time prior toparticle ignition by observing the lift-off distance as a function of particle size and reactorconditions. The intent was to compare the observed lift-off distance to that predicted bythe model. This series of experiments was fruitless. Just as with the work of Jokilaaksoet al [46], which was discussed in Section 2.2.2, the observed lift-off distance proved tobe unreliable response data. The introduction of an inert carrier gas to prevent feed tube424.2.1 Laminar Flow Furnaceblockage biased the lift-off distance by two modes: heat transfer to the particle byconvection/conduction was impeded by the presence of the stream of cold carrier gas, andthe oxidation reaction was impeded by the inert carrier gas which induced low oxygenpartial pressures surrounding the particles during the heat-up phase.Although the flow of carrier gas was low relative to the reactant gas feed rate(approximately 1 molar percent) the carrier gas flow rate showed a significant effect onthe observed lift-off distance. In one series of tests, increasing the carrier gas flow ratefrom 150 to 250 mL mm4 (at standard temperature and pressure) increased the particleignition lift-off distance from 20 to approximately 30 cm. The rate of heat transfer to thestream of cold carrier gas was estimated assuming laminar flow, with no mixing, andconductive heat transfer only, from the furnace gas through to the axis of the stream ofcold carrier gas; this calculation would give a minimum rate of heat transport. Thepredicted minimum rate of heat transport indicated that the carrier gas would reach thefurnace temperature in approximately 10-15 cm, with the carrier gas flow rate used.The results of this early series of experiments show that the effect of an inert carrier gasmust be accounted for in any calculation of a predicted lift-off distance; both in terms ofthe rate of heating of the carrier gas and the rate of oxygen diffusion through the carriergas stream to the particle surface. The calculation of a predicted lift-off distance isfurther frustrated by the possibility of various particle trajectories that may displace theparticles from the axial carrier gas stream into the laminar reactant gas stream. Suchdisplacement is not predictable, and would result in more rapid oxidation.After recognizing the problems created by using an inert carrier gas, the apparatus wasmodified such that the stream of reactant gas was split, introducing the majority into the434.2.2 Two-Wavelength Pyrometerpre-heat zone of the furnace, and diverting a minor amount (approximately 1 molarpercent) to the concentrate feed tube to act as a carrier gas. This modification ensuredthat the prescribed oxygen partial pressure was encountered by the particles, but did notcorrect the delay in particle heating.4.2.2 Two-Wavelength PyrometerThe two-wavelength pyrometer, designed and built by Tuffrey [22,23] for the study ofsuiphide particle flash reactions, is unique in its high sensitivity and fast response time. Aschematic diagram of the pyrometer and data acquisition system is shown in Figure 4.6;and a complete description of the design, construction and calibration of the pyrometer isgiven by Tuffrey.The 300mm long lens tube is inserted into the pyrometer view port in the laminar flowfurnace. The lens tube is water cooled to protect the optics, and the quartz windowexposed to the furnace is flushed with a low flow of inert gas to protect it from reactionproducts. The line-of-sight of the pyrometer is directed through the reaction zone in theaxis of the furnace, and into a water-cooled black body cavity opposite to the pyrometerview port to provide a low radiation background. The black body cavity is a 25.4mm o.d.cone tapering to a 10mm i.d., 150mm long cavity coated with a flat black paint.The beam collected by the pyrometer is split by a partially silvered mirror beam splitter;and the split beams are then filtered by monochromatic interference filters at wavelengthsof 7 lOnm and 8 lOnm. The filters each have an acceptably narrow band width of lOnm.Two photo-multiplier tubes with extended red sensitivity are used as the photo-detectorsto quantify the intensity of the monochromatic beams.444.3 Phctomicrographic Analysis of Reaction ProductsThe photo-multiplier output voltages are directed through a DT2828 DMA DataAcquisition Board to an IBM AT personal computer using ASYST data acquisitionsoftware. The DT2828 DMA board is capable of sampling at 50kHz per channel andallows direct memory access data storage. The maximum data collection is limited to 16k-bytes per channel, which at 50kHz corresponds to 0.32 seconds of data collection. Thetypical viewing time of a particle in front of the pyrometer view port is on the order of 10to 20 ms. The detection of particle combustion pulses is indicated by the occurrence ofprolonged pyrometer voltages above the background or dark signal level.Extensive validation and calibration of the pyrometer was conducted by Tuffrey [22,23],who also showed that the reliability of the pyrometer was excellent by checking thecalibration over the course of a long testing program and fmding no drift in readings.Confirmation of pyrometer calibration for this chalcocite study consisted of duplicatingthe combustion pulses obtained by Tuffrey for the flash oxidation of pyrrhotite in air.Figure 4.7(b) and Figure 4.8(a) show the observed particle temperature of Type Al(non-explosive) pyrrhotite combustion pulses obtained by Tuffrey and in this work,respectively. Typically, the maximum particle temperature is in the range of2200-2550K, and changes in apparent particle diameter are small, on the order of 2 fold(as is seen in Figure 4.8(b)), for these non-explosive combustion pulses.4.3 Photomicrographic Analysis of Reaction ProductsA series of combustion tests was run by releasing a stream of particles into the laminar flowfurnace with a constant furnace length and varying the furnace temperature and oxygenpartial pressure. This series of tests was run to replicate the method of Jokilaakso et a! [24]454.3 Photomicrographic Analysis of Reaction Productsto examine the extent of reaction as a function of furnace conditions, and to reproduce theignition temperature and dust generation data obtained by Otero et a! [12] using a stagnantgas reactor.The furnace length was fixed at 50 cm, by positioning the feed tube. A furnace length of 50cm is more than sufficient to allow for particle heating to progress to the extent that particleignition can occur, if the furnace conditions dictate that ignition will take place. The reactorconditions for the series of tests were furnace temperatures of 1023, 1123 and 1223 K withoxygen partial pressures of 0.10, 0.50 and 1.00 atm. All tests were performed using the24-3 .tm fraction of the MK concentrate feed. The conditions were chosen to demonstrate arange of combustion characteristics from sub-ignition through to highly oxidizing. The24-33 p.m size fraction was used because it approximates the mean MX concentrate particlesize, as discussed in Section 4.1.2.The reacted particles were collected as they exited the furnace (at a distance of 50 cm frompoint of entry) and the reaction was quenched by flushing with cold argon gas. The particleswere then mounted on a graphite specimen pedestal for examination under a scanningelectron microscope. The proportional content of the major constiment elements (copper,nickel, iron and sulphur) were determined by energy dispersive analysis, andphotomicrographs were obtained to document the morphology of reacted particles in order tosubstantiate the proposed reaction mechanism.464.4.1 UBC Flash Smelting Pilot Plant Facility4.4 Pilot Plant StudiesIn addition to the kinetics study reported in this work, a program of pilot plant MK flashconverting tests was conducted. Several important discoveries were made in the course ofthe pilot plant campaign which complement the kinetics study; and thus the pilot plant resultswarrant some discussion in this thesis.The MK flash converting pilot plant trials were conducted at the UBC Flash Smelting MiniPilot Plant facility in the Centre for Metallurgical Process Engineering at the University ofBritish Columbia.4.4.1 UBC Flash Smelting Pilot Plant FacilityThe UBC flash smelting pilot plant facility was originally designed to study the flashsmelting of lead concentrates, and was commissioned in 1986. Modifications were madeto the plant to accommodate the testing of chalcocite concentrates, including changes tothe concentrate feeder, the reactor wall lining and the off-gas handling system. The fullyintegrated and instrumented pilot plant was designed for up to 2 kg min’ solids feed rate,which is approximately 1/250 of the proposed full scale MK flash converting plantcapacity.The individual unit operations incorporated in the pilot plant, shown in Figure 4.9, are thegas and concentrate top-entry delivery systems, the vertical reactor shaft, post reactor dustrecovery equipment and the sulphur dioxide scrubber. The pilot plant was instrumentedand connected to a data acquisition system to facilitate the recording of pertinent data.474.4.1 UBC Flash Smelting Pilot Plant FacilityOxygen was supplied to the reactor from a liquid oxygen storage facility. Oxygen flowrate control was performed by a PT flow controller connected to a pneumatic valve. Theset-point of the controller was automatically fixed by the main computer system. Oxygenflow measurement was accomplished by a Micro-Motion mass flow-meter which gives adirect reading of oxygen mass flow once calibrated; no compensation for gas temperatureor pressure is required. The gas flow was concurrently gauged with one or morerotameters in pilot plant runs which required apportioning the oxygen to two separatestreams.The concentrate was conveyed from a feed hopper to the burner head by a variable speedscrew-feeder followed by a vibratory feeder. The screw-feeder was controlled by adedicated computer system. The screw feeder and feed hopper were mounted oncalibrated load cells which were continuously read by the controlling computer. Acomputer-controlled bucket elevator system reloaded the hopper as necessary. Theconcentrate was transported from the screw feeder to the burner head (a horizontaldistance of O.45m) by a vibratory feeder.The configuration of the burner was changed throughout the pilot plant campaign, but inall cases the oxygen and concentrate feeds were introduced simultaneously into the top ofthe reactor shaft. The various burner configurations are discussed further in Section4.4.2.The flash reaction shaft is 1. 8m tall and the inside diameter of the refractory lining isapproximately O.5m. The shaft is vertically segmented into four sections (denoted A, B,C and D in Figure 4.9) each section having two ports for solids and gas sampling; andthree Pt/Pt-lO% Rh (type S) thermocouples were mounted vertically equidistant at the484.4.1 UBC Flash Smelting Pilot Plant Facilityinner surface of the refractory wall. The reactor shaft consists of an outer steel shellhousing a refractory lining in two concentric layers; 17.8 cm of Claycast 60 refractorybacked by 5.1 cm of Kfac- 19 alumina wool block. This combination of refractorymaterials was used to provide sufficient heat retention in the shaft to sustain reactiontemperatures, while giving a chemically inert and relatively non-adherent inner surface.The hearth at the base of the reactor shaft was lined similarly to the shaft, with theaddition of a bed of alumina powder to allow separation of the semi-blister copperproduced from the hearth lining after the reactor was cooled. A Pt/Pt- 10% Rhthermocouple was inserted through the bottom of the hearth into the “bath” of reactionproducts.The reaction off-gas containing suspended dust generated in the flash reaction passedfrom the hearth through an adjacent settling chamber where the largest suspendedparticles were gravity separated. The off-gas with remaining suspended fines was thenquenched by rapid cooling with approximately 1.3 m3 mint of cold air and piped to acyclone. The refractory-lined cyclone recovered dust particles greater than 5-10.tmquantitatively, and was emptied periodically to follow the dust generation rate. The fineparticles not separated from the gas stream by the cyclone were finally captured by animpingement separator and bag-house.With solids separation complete, the off-gas was passed in series through twinpacked-bed towers where it was contacted with caustic soda (a strong sodium hydroxidesolution) to scrub out sulphur dioxide gas, then the remaining inert gas was discharged.The entire circuit was kept under approximately 12-25 mm Hg vacuum by a tail-endexhaust fan.494.4.2 MK Flash Converting Pilot StudyThe reactor was instrumented to record reactant feed rates, inside reactor walltemperatures, heat flux through the reactor wall, outer steel shell temperatures, hearthtemperature and bag-house temperature. The data acquisition was carried outautomatically using a PC connected to a Keithley Series 500 inteffigent front-end,providing for 16 single-ended high level analog inputs, 48 thermocouple inputs, 32 digital1/0 lines and 4 0-2OmA analog outputs. The thermocouple boards used an isothermalblock with a calibrated thermistor to provide on-board cold junction compensation. Themain system computer communicated with the concentrate feeder control computer via anRS-232 line.The Series 500 data acquisition system and RS-232 line were driven by the maincomputer using a program written to operate in the Keithley Viewdac environment.4.4.2 MK Flash Converting Pilot StudyA total of 18 pilot plant trial runs were performed, with the test program encompassing arange of reactor conditions and a variety configurations of the burner head component ofthe reactant feed injector system. The test conditions are summarized in Table 4.4, andthe various burner head designs tested are depicted in Figure 4.10.The stoichiometric ratio of MK concentrate-to-oxygen feed stated in Table 4.4 wasdetermined based on the ideal complete desulphurization of chalcocite and production ofmetallic copper, as in Equation [4.1].Cu2S() + O) = 2CU(S) + SO) [4.1]504.4.2 MK Flash Converting Pilot StudyWhile this stoichiometry does not reflect the true nature of oxygen utilization in the flashreaction, it has been the basis used for pilot plant testing at the UBC mini-pilot plant, aswell as at the Inco pilot flash reactor at Port Colborne, Ontario.The particle and gas velocities cited in Table 4.4 are the calculated velocities of thereactants as they exit the burner head. The calculation was based on the volumetric flowrates and the cross-sectional area of the exits, assuming that the particles entrained by theoxygen stream are accelerated to terminal velocity as calculated by a balancing the forcesacting on the particle; aerodynamic drag due to the relative velocity of the particle andlaminar flowing gas, and the force of gravity. The variation in oxygen injection velocityis a result of the modifications to the ejector tube diameter and oxygen flow-rate appliedduring the campaign of pilot plant trials.The burner designs labeled 4 through 6 incorporated a split of the oxygen feed, with oneportion propelling the concentrate out of the burner head, and the secondary portion (2°Gas in Table 4.4) not contacting the concentrate before entry into the reactor.The procedure for all pilot plant trials was consistent. The furnace was preheated with anatural gas flame to attain an initial furnace temperature on the order of 1200 K. Thenatural gas flame was then extinguished, the oxygen feed rate established and theconcentrate feeder activated. The changes in furnace temperatures were observed andsteady state was generally reached within 30 minutes. The reaction progress wasfollowed by observation of the collected data, and by periodic sampling.Material was sampled whenever possible during the trials for analysis, including thereacting solids in the flash flame, the reactor gas, and the dust collected in the cyclone.514.4.2 MK Flash Converting Pilot StudyTABLE 4.4: Pilot Plant Trial ConditionsRun No. Burner MKIO2 Particle 1° Gas 2° GasType Stoich. Velocity Velocity Proportion(m s’) (m Sl)(a) (%)1 1 1.00 26 26-2 2 0.82 29 29-3 2 1.00 26 26 7.34 1 0.80 26 26-5 1 1.00 26 26-6 3 1.00 13 13-7 1 0.86 24 24 -8 3 1.00 nil 46-9 4 0.96 3.7 110 2310 4 0.98 4.8 110 2811 4 1.00 6.8 40 2712 5 0.82 22 22 4713 6 1.00 7.3 34 2814 6 0.98 nil 180 2515 6 1.00 nil 270 3516 7 1.00 ni1 240-17 7 1.12 nil 270-18 7 1.10 ni1 240-(a) calculated velocity of primary oxygen feed jet1 weight % petroleum coke added to concentrateCc) particles introduced falling freely with gravity(d) 0.032 kg min’ nitrogen gas shrouding flameSolids samples were obtained from four levels of the reacting flash flame with a watercooled impingement sampler and were later dissected and analyzed. Gas samples wereobtained from point sources within the reactor at each of the four levels and at variousradial positions across the shaft; the gas sampling apparatus is shown in Figure 4.11. Thegas samples were pumped from the water-cooled sampler, filtered and immediatelyanalyzed by two techniques: infrared transmittance for SO2 content and gaschromatography (G.C.) for the proportionate gas composition in terms of SO2, 02 and N2524.4.2 MK Flash Converting Pilot Studygases. Reactor gas temperature was determined by suction thermocouple; a type Sthermocouple enclosed in an insulating tube through which reactor gas was pumped at asufficient rate to give an indication of true furnace gas temperature. The cyclone dustwas collected and weighed at set intervals to follow the dust generation rate.At the conclusion of each run, the reactor was allowed to cool (with the system exhaustfan continuing to draw air through the reactor) then the hearth was detached. Thereaction products captured in the reactor and by the dust collection equipment wereweighed to determine the mass deportment for the trial. Solid samples were collected forchemical analysis from points throughout the pilot plant, including the hearth material,reactor wall, settling chamber, cyclone, impingement separator and bag house. Allchemical analyses were performed by Inco Research Labs in Mississauga, Ontario.534 EXPERIMENTALParticle Size (pm)FIGURE 4.1: MK Particle Size Analysis Performed by Ortech Ltd.10090C)CU, 80U)702 60C)C,U)40I100. - — —— . I - — — — — —.%-- -\--“___\__L100 10 1544 EXPERIMENTALParticle Size (pm)FIGURE 4.2: MK Particle Size Disthbution by Cyclosizing100900)C(o 80U)702 60,50U)(U40I ::100.—EE___\\_HE__100 10 1554 EXPERIMENTALFIGURE 4.3: Photomicrograph of 24-3311m Fraction of MK Concentrate564 EXPERIMENTALFIGURE 4.4: Photomicrograph of 24-33pm MX Partially Reactedunder Near-IgnitionConditions. Furnace Temp = 1123K, P0 = 0.50 atm574 EXPERIMENTALReactorInletConccntrate and/Purge Gas InletSteel ShellInsulationThermocouplesPrometerView PortFIGURE 4.5: Schematic Diagram of Laminar Flow FurnaceBlack BodyView PortWater CooledFeed TubeGLOBARHeating Element(One of Three)AluminaTubeI I ResistanceHeated Furnace58DT 2820Data Acquisition To Computer4 EXPERIMENTALFIGURE 4.6: Schematic Diagram of Pyrometer and Data Acquisition System, Constrctedby Tuffrey [22]SignalAmplifierWater—Cooled CasingTransferLens594 EXPERIMENTAL1.5. I0t 0.5C002024002200C2 2000C1800I1600C140012001000Time (ms)(b)FIGURE 4.7: Type Al Combustion Pulse for Pyrrhotite, from Tuffrey [22]Conditions: Furnace Temp =1130 K, P02 = 0.21 atm,Particle Size = 74-8 8 pm0 5 10 15Time (ms)(a)0 5 10 15 20604 EXPERIMENTAL3000 3000> 2500 - 25002000 - 20004-a)C.I-,15000a-C.I-, 1000 Ea)a)E___________I-2______500500>-0.Temp00.0 2 4 6 8 10 12 14 16Time (ms)3000 40> 250032000 -a-4-a)C E4-2 .1500 -0 0a-4-,a)C4- 1000 - a)a)aE_-1C-0I- C>.-_500_OIDo.00 2 4 6 8 10 12 14 16Time (ms)FIGURE 4.8: Type Al Combustion Pulse for Pyrrhotite Obtained in This Work(a) Pyrometer Output Voltage and Calculated Temperature(b) Pyrometer Output Voltage and Apparent DiameterConditions: Furnace Temp = 1127 K, P0 = 0.20 atm,Particle Size = 74-88 pm61C,,C) I I—c•-I00E..10øx4 EXPERIMENTAL02 02 02MK MK+02 MK+02 MK 4002Z02 /MK100EE0’2004,CD3000— —000215.9mm 15.9mm 22mm 15.9mm 12.7mm400- 402O2FO2 I—159mm 22mmBurnerType 1 2 3 4 5 6 7FIGURE 4.10: Pilot Plant Burner Configurations63tf)0 CDr0C-)1 0 “1CD11C) I IOh00c—a CD1jCD 1(I)5.1 Particle Combustion Temperatuis5 RESULTS5.1 Particle Combustion TemperaturesParticles were observed throughout the reaction process: the heat-up phase, the melting ofchalcocite, temperature peaks at the boiling point of copper, and the oxidation of coppervapour (seen in highly oxidizing conditions). Observations were made of the particletemperature profile, changes in apparent diameter during combustion and the combustionpulse type as established by Tuifrey [22-251 and discussed in Section 2.3. Over two hundredparticle combustion pulses were recorded.The resolution of the calculated particle temperature and apparent particle diameter obtainedin this work, examining the combustion of small (less than 4Opm) chalcocite particles wasnot as good as the resolution of the calculated data in the work of Tuffrey [22-25j, whostudied the flash oxidation of pyrite, pyrrhotite and commercial lead concentrates. Theparticle size of the concentrates in the Tuffrey study was in the range of 63-125p.m; three tofive times the mean particle diameter of MX concentrate. Consequently, the total intensity ofradiation from MK particles combusting at similar temperatures is significantly less, and thesignal-to-noise ratio is correspondingly higher. While the calculated particle temperature andapparent diameter data exhibits considerable noise, temperature plateaus were discemablefrom steadily increasing or decreasing temperatures, and from sudden temperature peaks.Particles in the heat-up phase that precedes flash reaction were observed to exhibit Type Alcombustion behaviour (Figure 5.1): a low and constant radiation intensity with a calculatedparticle temperature profile that plateaued slightly above the set furnace temperature, at themelting point of chalcocite (1403K). These characteristic pulses were observed in low655.1 Particle Combustion Temperaturesoxidizing conditions below the threshold conditions for particle ignition-- lower furnacetemperature and low oxygen partial pressure. Moreover, these pulses were only detected intests of the larger particle size fractions, since the sensitivity of the pyrometer was notsufficient to detect the low total radiated energy of the smaller particles at lowertemperatures.The particle ignition threshold is defined as the combination of furnace conditions andparticle size that give rise to a self-sustaining combustion reaction, whereby heat generationby exothermic oxidation exceeds the rate of heat transfer from the particle, thus generating anincrease in particle temperature and an acceleration of the oxidation reaction rate. Particleignition is discussed further in Section 5.2.A second example of the Type Al combustion pulse was obtained with the plateautemperature at the boiling point of copper (2840 K) and an example is shown in Figure 5.2(a). This form of pulse was obtained under reactor conditions at the threshold for particleignition.There was little change in the apparent diameter of particles undergoing Type Alcombustion, Figures 5.1(b) and 5.2(b). This is consistent with the observed phenomenon,which is the steady increase in particle temperature due to the generation of heat by theoxidation reaction at a rate comparable to the rate of heat transfer from the particle with theresulting temperature increase progressing either to the melting point of chalcocite(sub-ignition conditions) or to the boiling point of copper (ignition conditions).Combustion pulses obtained at oxidizing conditions at, or slightly above, the threshold forparticle ignition demonstrated peak particle temperatures reaching the boiling point of665.1 Particle Combustion Temperawresmetallic copper and followed Type A2 or Type A3 behaviour. Type A2 behaviour pulseswere obtained when the particle temperature reached the copper boiling point (2840K). Anexample is given as Figure 5.3. The associated change in apparent diameter was seen to bean increase on the order of 5 to 10 fold, which suggests particle expansion or the existence ofa photosphere surrounding the particle in which copper vapour was oxidized according toEquation [5.1]2Cu)+O) = Cu20(,) [5.1]The diameter of the radiant oxidation photosphere surrounding the particle would be greaterthan the diameter of the particle, thus observation of the photosphere would be manifest as anincrease in the apparent diameter of the particle.The Type A3 combustion pulse is characterized by a very short duration, high intensity pulsethat may be superimposed on Type A2 pulses. The Type A3 pulse displays a dramaticincrease in the apparent diameter of the particle which may be associated with a temperatureincrease. This pulse behaviour is indicative of the ejection of material or the disintegration ofthe particle by fragmentation or explosion. Examples of Type A3 combustion pulses, withand without an associated temperature increase above the boiling of copper, are given inFigures 5.4 and 5.5 respectively.Many combustion pulses obtained at, or slightly above, the threshold conditions displayedType A4 behaviour, which further suggested material ejection. The Type A4 combustionpulse is characterized by a Type Al or A2 pulse upon which is superimposed a rapid periodicfluctuation in radiative energy from the particle. The energy fluctuation may beaccompanied by a temperature and apparent diameter fluctuation. An example of a Type A467i2 Mathematical Modeling of Reaction Kineticscombustion pulse is given in Figure 5.6. Tuffrey [22-25] observed Type A4 combustionpulses in his study of iron and lead suiphide flash smelting, and stated that the observedphenomenon could be an diameter pulsation, a pulsating flame front or particle rotation.Under the most oxidizing furnace conditions of higher temperature and high oxygen partialpressure, the combustion pulses observed were Type A2 or A3 with exceedingly highmaximum particle temperature; often over 4000 K. Apparent particle diameter changes of upto 50 fold were also observed. An example of a combustion pulse obtained under highlyoxidizing conditions is given in Figure Mathematical Modeling of Reaction KineticsA mathematical model of the kinetic processes occurring during the flash reaction ofchalcocite concentrate was developed and validated by Shook [11]. The most suspectassumption made in the development of that model related to the kinetics of the chemicallyrate controlled, low temperature reaction of chalcocite with oxygen, producing copper oxideand sulphur dioxide gas according to the reaction in Equation [2.6].Other suspect computational factors included the heat transfer rate to/from the particle andthe rate of mass transfer of oxygen to the particle. Shook also assumed that the particle iscomposed of 100% chalcocite. Sensitivity tests determined the effect of varying physicalparameters such as the oxygen diffusion rate and rate of heat transfer. These tests showedthat the low temperature reaction kinetics played the greatest role in determining thepredicted ignition temperature. The rates of transfer of oxygen and heat to/from the reactingparticle were estimated using well established engineering calculations. Inaccuracies of as6852 Mathematical Modeling of Reaction Kineticsmuch as ten percent in the estimation of the rates of either heat or mass transport resulted inchanges to the predicted particle ignition threshold temperatures of only one to three Kelvin,for a set furnace gas composition and particle size.Shook based his calculation of the rate of the chemical reaction on the best information fromthe technical literature, which was the work of Kim and Themelis [34], who studied theoxidation of compressed chalcocite pellets. Kim and Themelis state that the reaction followsfirst order kinetics with respect to oxygen partial pressure, Equation [2.2], and that thechemical reaction rate constant can be defined by the Arrhenius expression, Equation [2.3]The experiments of Kim and Themelis produced the following values for the kineticparameters (the pre-exponential constant and the activation energy) in Equation [2.3]:A = 2.52 x 1019 mol m2 atm s4Ea = 525 U mot’The threshold conditions for particle ignition predicted by the model using the data of Kimand Themelis and a mean particle size (25j.im) are plotted in Figure 5.8, along with thethreshold conditions obtained by experiment in this work for the 24-33E.tm fraction. Thecorrelation is poor. Model calculations made using the kinetics parameters of Kim andThemelis predict that particle ignition does not occur at furnace temperatures below 1350 K,regardless of oxygen partial pressure.The experimental data obtained in this work from an actual flash reaction process were usedin conjunction with the model developed by Shook to determine more precisely the flashreaction kinetic parameters in the Arrhenius expression, Equation [2.3]. By examining theeffect of changing the value for the kinetic parameters in the model on the predictedthreshold for particle ignition, it was determined that a more correct value for the activation695.2 Mathematical Modeling of Reaction Kineticsenergy of the chemically rate controlled, low temperature oxidation reaction of chalcocitewas 460 kJ mot1. Figure 5.9 shows a plot of the predicted threshold for particle ignitionusing a chemical activation energy of 460 U molt compared to the mean experimentallydetermined threshold.The threshold conditions for particle ignition range in furnace temperature fromapproximately 1200 to 1300 K, varying according to oxygen partial pressure and particlesize. The ignition threshold temperature increases with lower oxygen partial pressure anddecreasing particle size. The work of Otero [12], who flash reacted MK concentrate in astagnant gas reactor, showed that weight loss from the concentrate (interpreted as particleignition) occurred at temperatures of 1050 to 1150 K, depending on oxygen partial pressure.The experiments of Otero involved releasing unsized MK concentrate into the stagnant gasreactor in pulses.The pre-exponential constant (or frequency factor) determined by the experiments of Kimand Themelis appears to be satisfactory. Model calculations were performed using alternatevalues for the pre exponential constant, and this factor proved to have less effect on theparticle ignition threshold conditions than did the activation energy.An example of the results of the model calculations is given in Figure 5.10, which shows thepredicted particle temperature profile for a single particle reacting in the laminar flowfurnace at a set of furnace conditions just above and just below those that define a thresholdfor particle ignition (1210 and 1200 K respectively, for a 25tm particle reacting in 0.50 atmoxygen).705.3 Photomicrographic Analysis of Reaction ProductsEvidence of the occurrence of different phenomena are visible in Figure 5.10. The modelpredicts that the particle heats up to the furnace temperature in the first 20 to 40 millimetersof the furnace. The retarding effect of the cold carrier gas was not included in the modelpredictions. The cold carrier gas would have the effect of moving the ignition point muchfarther from the burner tip, but would not influence the predicted extent of reaction (i.e.,whether the reaction progresses to ignition) because of the existence of the incubation periodwhere the particle temperature plateaus at the furnace temperature prior to ignition. Theparticle ignites when the rate of heat generation from the exothermic oxidation reactionexceeds the rate of heat transfer from the particle, and rapid heating of the particle results.Upon ignition, the predicted temperature profile rises rapidly and continuously, with nodiscontinuity at the melting point of chalcocite (1403K). This differs from the experimentalobservations, where some particles were seen to plateau at the melting point of chalcocitewhile heat generated by the reaction was consumed by providing the latent heat required tomelt the chalcocite, rather than to increase the temperature of the particle. The copper oxideproduced by the low temperature oxidation is reacted with chalcocite at this point byEquation [2.7]. The particle temperature then continues to increase, and a plateau isencountered at the boiling point of copper (2840K). The model continues to calculatereaction progress until all of the metallic copper produced by the oxidation of chalcocite isdetermined to have either boiled-off or been oxidized to Cu20by Equation [2.10].5.3 Photomicrographic Analysis of Reaction ProductsThe extent of the flash converting reaction as a function of furnace conditions was observedin a series of tests using a fixed furnace length and varying the furnace temperature andoxygen partial pressure.715.3 Phctosnicrographtc Analysis of Reaction ProductsThe extent of desuiphurization of the products (and resultant copper enrichment) wasdetermined by energy dispersive analysis (EDA) of a wide view-field of the particulatesample. The EDA results are summarized in Table 5.1 and shown graphically in Figure 5.11.The chemical analyses are nonnalized to show only the detected elements; for exampleoxygen content cannot be determined by EDA. The extent of desuiphurization increases withincreasing furnace temperature and oxygen concentration, that is, with increases in theoxidizing strength of the reaction environment.TABLE 5.1 Energy Dispersive Analysis of MK Flash Reaction Products (Normalized)Furnace Conditions Mass PercentTemperature P02(K) (atm) Cu S Ni Fe24-331.LmMKFeed 71.61 22.78 5.07 0.551023 0.10 80.87 12.69 5.81 0.621023 0.50 80.56 12.37 6.17 0.901023 1.00 82.95 9.08 6.94 1.021123 0.10 82.14 10.98 6.27 0.621123 0.50 73.10 6.29 19.48 1.131123 1.00 70.64 8.50 19.61 1.261223 0.10 76.01 8.31 12.45 3.241223 0.50 73.80 3.69 18.53 3.991223 1.00 68.04 2.50 26.70 2.761223 0.50 19.44 * 16.87 * 63.25 * 0.44 ** EDA of the individual collapsed cenosphere shown in Figure 5.16Spherical particles were formed due to the melting of the chalcocite component of the MKconcentrate, even at lower temperatures. Figure 5.12 shows a photomicrograph of theproduct from the test run at 1023 K with 0.50 atm oxygen and reveals that even at low725.3 Photomicrographic Analysis of Reaction Pixductsoxidizing conditions, some particles exhibit sufficient reaction progress, with a subsequentincrease in particle temperature, to reach the melting point of chalcocite-- in this case 380 Kabove the furnace temperature. Those particles that retain the angular shape of the feedconcentrate do show significant rounding of the edges as compared to the feed concentrateshown in Figure 4.2.There was also photomicrographic evidence of dust generation by the action of boilingcopper. Under more oxidizing conditions (at or above the threshold conditions for particleignition) perforated cenospheres were observed indicating that the generation of gas withinthe particle led to a build-up of pressure and the ejection of a gas jet (either SO2 or coppervapour) through a puncture in the surface. Figures 5.13 and 5.14 show perforatedcenospheres. The chemical composition of these ruptured cenospheres, determined bynarrow view-field EDA, revealed that the cenospheres had a similar composition to that ofthe surrounding particles, but showed a modest increase in nickel-to-copper ratio.A depletion of copper during the flash reaction was noted in all samples, as evidenced by anincrease in the ratio of nickel-to-copper content in the products. Figure 5.15 shows thedepletion of copper as an increase in the nickel-to-copper ratio.The preferential depletion of copper is dramatically illustrated by the existence of collapsedcenospheres with high nickel content seen in the products of flash reactions at high oxidizingconditions. Figure 5.16 presents an example of these collapsed cenospheres, produced at1223 K in 0.50 atm oxygen. The nickel content of this cenosphere was enriched from 5.07mass percent in the feed MK to 63.3 mass percent in the collapsed cenosphere--an almost50 fold increase in the nickel-to-copper ratio.735.4.1 Steady-State Furnace Temperatures5.4 Pilot Plant ResultsThe pilot plant campaign yielded a vast array of information, only some of which wasapplicable to this kinetic study. The data examined, as pertinent to this work, were thesteady-state furnace temperatures obtained, the desulphurization rate along the flash flame,the deportment of nickel in the reactor, the gas analyses taken during the trials and the dustgeneration rate.Some of the results of the earlier pilot plant trials were summarized and discussed by Shook[11], who modeled the MK flash flame and reaction shaft, using the pilot plant trials tovalidate his model.5.4.1 Steady-State Furnace TemperaturesThe temperature of the furnace at steady state is taken as an indication of the reactiontemperature and is determined by the rate of heat generation within the flash flame, heatpropagation to the wall and heat flux through the wall. The reactor wall temperature isanalogous to the set furnace temperature in the laminar flow furnace used in thepyrometer study in that it establishes the rate of heat transfer to the particles bywall-particle radiation and it is similar to the furnace gas temperature, which establishesthe particle heating rate by convection/conduction.The inside wall temperature of the reactor was recorded at five locations in each of thefour vertical sections: three thermocouples were mounted vertically equidistant along theshaft to record the temperature profile and two other thermocouples were situated torecord any variation in temperature around the circumference of the shaft. The axial745.4.1 Steady-State Furnace Temperaturestemperatures are summarized in Table 5.2, showing the mean steady-state temperature ofeach section in the pilot plant trials. Circumferential temperature variations were onlyobserved during upset conditions when the flash flame was perturbed from radialsymmetry.The data show a moderate trend indicating that the higher reactor temperatures weregenerated by the trials which were conducted with lower MX/02stoichiometric ratios;that is, with an excess of oxygen. Specifically, Trials 2 and 12 were run with a 20%stoichiometric excess of oxygen, and these trials show higher reactor wall temperatures.Trials 17 and 18 were deficient in oxygen, yet show similar reactor wall temperatures tothose trials with an exact stoichiometric MK/02ratio.Gas temperature readings were not attempted in every trial, and the reliability of the gassuction thermocouple was poor -- suffering from frequent blockage by ingested solids --but gas temperatures were successfully obtained during trials 14, 16 and 18. Theobserved reactor gas temperatures at the axis of the reactor in the active flame regionwere up to 150 K greater than the recorded reactor wall temperature. Table 5.3 lists thesuction thermocouple temperatures obtained.755.4.1 Steady-State Furnace TemperaturesTABLE 5.2: Steady-State Pilot Furnace TemperaturesTrial No. MK/02 Mean Section Temperature (K)Stoich. A B C D1 1.00 1084 1130 1192 12832 0.82 1203 1243 1351 14543 1.00 1062 1101 1170 12744 0.80 1103 1161 1224 12925 1.00 1087 1137 1180 12466 1.00 1108 1198 1306 13487 0.86(a) 1114 1173 1252 13128 1.00 1135 1242 1350 13439 0.96 1143 1245 1331 135210 0.98 1123 1203 1272 139511 1.00 1187 1124 1216 120312 0.82 1195 1239 1285 133413 1.00 1035 1159 1231 132014 0.98 1160 1232 1306 134415 1.00 1232 1247 1284 132716 1.00 1201 1251 1259 131317 1.12 1193 1237 1246 130218 1.10 1215 1247 1287 1263(a) 1 weight % petroleum coke added to concentrateTABLE 5.3: Furnace Gas Temperatures by Suction ThermocoupleTrial No. Section Wall Temp. Gas Temp.(K) (K)14 D 1344 136816 C 1259 140716 D 1313 138318 D 1263 1368765.4.2 Desulphurization and Nickel DeportmentThe reactor wall temperature profile in all trials shows that the reaction temperatureincreased as the flame traveled down the shaft. Most trials demonstrate a maximumtemperature at the fourth (lowest) section. This phenomenon was confirmed byqualitative observations of the appearance of the flash flame in the reactor. The viewports in each section of the reactor were used to confirm that the reaction was mostviolent, as demonstrated by the intensity of radiation and the appearance of “sparks”indicative of particle ignition or fragmentation, at the lowest levels.5.4.2 Desuiphurization and Nickel DeportmentThe desulphurization of MK concentrate along the pilot plant flash reactor shaft , asdetermined by analysis of solids samples collected during the trials and hearth samplesacquired after each trial, is summarized in Table 5.4 for some selected trials, along with atypical MK feed assay. The design of the solids sampler was not ideal, in that thecollection of solids from the flame was not quantitative due to the scouring action of thehigh velocity gas jet; but the samples obtained did give an indication of the radialdistribution of the concentrate and its chemical composition. The analyses show that theextent of desuiphurization increased with extended reaction time, that is, as the particlestraveled down the shaft. Accompanying the desuiphurization was an increase in coppercontent due to the decreased mass of the particles, and a marked enrichment of nickel inthe particles beyond the increase due to mass loss. The nickel-to-copper ratio in thesamples increases along the reactor shaft.775.4.2 Desuiphurization and Nickel DeportmentTABLE 5.4 Desulphurization Along Pilot Plant Reactor ShaftTrial No. Assay (mass %)Section Cu S Ni Fe 0MK Feed 72.7 20.3 4.8 0.3 1.739 A 73.0 16.2 4.2 0.3 4.9B 76.5 13.5 3.8 0.3 4.8C 77.7 13.6 4.9 0.2 2.7D 80.5 10.1 5.8 0.3 2.7Hearth 80.8 12.4 4.4 0.2 2.410 A 73.6 18.3 4.6 0.2 1.3B 79.2 9.7 3.8 0.3 6.8C 75.8 14.3 4.8 0.2 4.2D 78.9 9.1 6.1 0.3 4.1Hearth 87.6 5.7 5.2 0.3 0.914 A n.a. n.a. n.a. n.a. n.a.B 77.1 13.4 4.3 0.3 4.2C 77.6 13.6 5.2 0.3 3.4D 78.4 9.7 6.1 0.4 3.8Hearth 89.1 3.8 4.2 0.2 1.217 A 72.0 16.7 4.3 0.4 5.8B 76.7 14.3 4.8 0.3 3.2C 77.8 13.9 4.8 0.2 3.0D 78.0 12.6 5.5 0.3 3.9Hearth 82.6 7.5 7.8 0.4 4.2* cyclone dust assaysThe composition of the hearth material (Table 5.4) represents the result of a partialthermodynamic equilibration between chalcocite and copper oxide, producing metalliccopper and sulphur dioxide via Equation [2.7]; suiphide and oxide were both found in thehearth material. The hearth temperature in the pilot flash reactor (the measured hearthtemperature never exceeded 1300 K) cannot be maintained as high as that in the bath of a785.4.2 Desu1phurizatii and Nickel Deportmentproduction scale facility, that is, above the melting point of copper (1356 K). Therefore,the flash reaction products in the pilot plant reactor hearth are not completely moltenwhich decreases their mobility and subsequent reactivity. The desired semi-blister copperproduct from flash conversion of MK concentrate typically contains 1 weight percentsulphur, and virtually no oxygen other than that contained as sulphates or refractoryoxides of the minor constituent elements in the MK feed [7].The enrichment of nickel in the samples collected from the pilot plant is most remarkablein the samples collected from the reactor wall after the reactor was cooled. Nickelcontents of 13% or more were common in the shaft wall material. Conversely, the assaysof the dust collected in the cyclone show a marked depletion in nickel content comparedto the feed concentrate. Table 5.5 presents the deportment of elements to the reactor walland cyclone dust.The mass of material collected on the wall of the reactor during pilot plant trials with aprimary oxygen injection velocity below 50 m s’ (Table 4.4) was generally on the orderof 6-16%, calculated as the percentage of the MK feed used in the trial. In trials 16through 18, however, the proportion of material collected on the wall was significantlyhigher-- 50 % or greater, as a proportion of the mass of MK feed. There was acorrelation between the calculated oxygen injection velocity and the mass of materialcollected on the wall of the reactor shaft. Table 5.6 lists selected data showing thedeportment of material to the reactor shaft wall and the oxygen injection velocity.795.4.2 Desulphurization and Nickel DeportmentTABLE 5.5 Nickel Deportment in Pilot Plant ReactorTiial No. Assay (mass %)Sample Cu S Ni Fe 0MK Feed 72.7 20.3 4.8 0.3 1.739 Wall 75.8 2.4 12.9 0.8 7.5Cyclone 69.1 7.1 1.2 0.2 20.610 Wall 87.5 0.1 6.2 0.3 8.3Cyclone 69.5 5.4 1.2 0.1 23.614 Wall 78.5 1.3 13.0 0.9 4.9Cyclone 57.4 8.5 0.9 0.1 32.017 Wall 84.6 2.7 5.5 0.3 8.8Cyclone 73.1 3.4 1.2 0.1 22.6TABLE 5.6: Oxygen Injection Velocity and Mass Deportment to WallRun No. Gas Mass onVelocity Shaft Wall(m sW’) (%) (a)5 26 15.96 13 8.28 46 11.39 110 15.311 40 10.012 22 8.213 34 6.514 180 16.216 240 51.617 270 48.818 240 69.2(a) total mass of material on wall as percent of total mass of MK feed805.4.3 Reactor Gas Analyses5.4.3 Reactor Gas AnalysesThe gas analyses taken during the pilot plant trials reveal that the concentration of oxygenwas highest at the centre-line (along the axis) of the reactor shaft, and decreased radially.Moreover, the centre-line 02 concentration decreased along the shaft as oxygen wasconsumed in the flash reaction and the furnace gases, SO2 and N2, were entrained into theflame.Selected chromatographic analyses are shown in Table 5.7, demonstrating that thedynamic flash flame exposes reacting particles to a range of oxygen concentrations.The presence of N2 gas in the reactor was due to infiltration of air into the shaft; thesystem is kept under a moderate vacuum (12-25 mm Hg) by a tail-end exhaust fan.Shook [11) determined that the rate of air infiltration into the UBC mini-pilot plant wasup to 42 litres mm4 due to air leakage around fittings and to cooling the air which wassupplied to sensitive instruments such as the natural gas flame detector. This level of airinfiltration corresponds to an excess oxygen inlet of up to 3% of the stoichiometricoxygen requirement for a 2.0 kg miii’ concentrate feed rate.The gas analyses do not present a consistent description of the cross-section of the flashflame, because of the variety of burner headconfigurations used. The object of changingthe burner head design was to introduce novel flow patterns that would change the gasand particle distribution in the flash flame, and thus effect the reaction progress. Thisobjective being achieved makes it difficult to draw comparisons between the gas analysisresults of different trials.815.4.3 Reactor Gas AnalysesTABLE 5.7 Gas Chromatographic Analyses of Furnace GasGas Sample Analysis (volume %)Trial Burner Section Sample Location SO2 02 N2No. Type*2 2 A Centre-line 18 68 14B Centre-line 30 49 19C Centre-line 62 26 12D Centre-line 80 6 145 1 A Centre-line 48.0 45.0 4.5A Centre-line 80.0 12.3 7.0D 10 cm off centre 89.0 2.0 8.76 3 A Centre-line 48.5 45.0 6.5A 10 cm off centre 88.0 0.6 10.5C Centre-line 79.0 11.0 8.2C 10 cm off centre 90.5 3.1 7.117 7 A l0cmoff centre 77.5 1.5 19.5B Centre-line 73.0 7.7 18.3B 10 cm off centre 77.0 3.0 19.5D 10 cm off centre 76.0 <1.0 22.0* The various burner designs tested are shown Figure 4.10.However, it is apparent that in Trial 2, which was run with a stoichiometric excess ofoxygen, the oxygen content of the gas in the upper sections along the axis of the reactorshaft is higher than in the other runs. The oxygen is largely depleted at the lowestsection, indicating either entrainment of SO2 gas into the flame or consumption of theexcess oxygen through oxidation of copper to copper oxide. In all runs, the oxygencontent of the recirculating gas was very low -- not exceeding approximately 4 volumepercent -- even in those trials testing a burner design which imparted high gas injectionvelocity and thus high recirculation rates (as discussed in Section 5.4.2).825.4.4 Dust Generation RateThe problem of sampler blockage was often encountered during gas sampling,particularly when drawing gas from the most reactive regions of the flame. The tip of thesampler tube was beveled (Figure 4.11) in an attempt to exclude solids, but moltenmaterial agglomerated at the tip and was often drawn into the sampler with the extractedgas, thus blocking the sampler. A considerable volume of reactor gas was required topurge the gas sampling apparatus prior to initiation of the G.C. analysis; the gas samplingapparatus was considered purged when the I.R. analyzer showed a constant SO2 level.Sampling times were often in excess of two to three minutes, and the probability ofsampler blockage increased with extended sampling time. However, sufficient gassamples were collected to provide a clear depiction of the gas composition throughout thereactor.5.4.4 Dust Generation RateFine particulate dust generated in the flash reaction was collected in the settling chamber,cyclone, impingement separator and bag house: the dust collection equipment is shownin Figure 4.9. The dust generation rate for each trial was calculated as the cumulativemass of dust collected relative to the mass of MK concentrate introduced to the reactor.Typical dust generation rates for the Inco Port Colborne pilot plant and the UBCmini-pilot plant are on the order of 10 to 15 % of the mass of MK feed [11]; the dustgeneration rates for the UBC mini-pilot plant trials are listed in Table 5.8. High dustgeneration rates were obtained when the MK/02ratio was reduced either by anover-supply of oxygen or a low MK feed rate. Trials 2 and 4, both run withapproximately 20% excess oxygen, demonstrated some of the higher dust generationrates, at 14.8 and 14.0% respectively.835.4.4 Dust Geiieration RateTABLE 5.8: Dust Generation Rates for Pilot Plant TrialsRun No. MK/02 Dust Rate Run No. MK/02 Dust RateStoich. (%) (a) Stoich. (%) (a)1 1.00 n.a. 10 0.98 12.22 0.82 14.8 11 1.00 16.43 1.00 13.6 12 0.82 14.84 0.80 14.0 13 1.00 17.05 1.00 8.8 14 0.98 10.56 1.00 16.6 15 1.00 12.37 0.86 12.1 16 1.00 4.98 1.00 13.3 17 1.12 4.69 0.96 9.7 18 1.10 4.8(a) total mass of dust collected as percent of total mass of MK feed1 weight % petroleum coke added to concentrate(c) commissioning trial, dust generation rate not determinedThe lowest dust generation rates were obtained in Trials 16 through 18, at 4.6 to 4.9%.These trials were run using the burner design noted as Type 7 in Figure 4.10. This designinjected the oxygen into the reactor through a small diameter tube, thus imparting a veryhigh gas velocity (240 to 270 m s1). The concentrate fell under gravity in an annulussurrounding the stream of high velocity oxygen.In addition to the change of burner design, Trials 17 and 18 were run with a 10% excessof MK with respect to the stoichiometric requirement of oxygen. The desulphurizationrate in these runs was not correspondingly low. The assays of the hearth material, shownin Table 5.4, indicate that the desulphurization rate was not significantly different fromtrials run at the calculated stoichiometric MK/02ratio.845 RESULTS250020001500II.fl.RJE50004IC,C,E2IC,I C.0.00 5 10 15Time (ms)02058007006008007006004-5000.4-0IC,4-20FIGURE 5.1: Type Al Combustion Pulse, Maximum at Chalcocite Melting Point(a) Pyrometer Output Voltage and Calculated Temperature(b) Pyrometer Output Voltage and Apparent DiameterConditions: Furnace Temp = 1170 K, P = 0.20 atm,Particle Size = 24-33 jim30 5 10 15Time (ms)85S R.ESULTh1400 4500- 40005 1200 -35001000 3000II25000 2000_•1I.- 600a, CEa,E 400__________I—0I71Onm-500> 200 -___Temp0 I 00 5 10 15Time (ms)1400 58lOnmiiOnmI 05 1200 - 0,001-4I-a,4-4-z a,C.4- E10000a, 4-CC, a,E a,0 C-C.>..a. 2000 5 10 15Time (ms)FIGURE 5.2: Type Al Combustion Pulse, Maximum at Copper Boiling Point(a) Pyrometer Output Voltage and Calculated Temperature(b) Pyrometer Output Voltage and Apparent DiameterConditions: Furnace Temp = 1173 K, P02 = 0.50 atm,Particle Size = 24-33 im865 RESULTh500 4500- SlOnm450-7lOnm4000400-_____& 3500 —3000 —a.25000-2000, 200-1500G) 150 -C,E0 100- “ 1000k->°- 50-- 5000 00 5 10 15Time (ms)500___ ____9SlOnm45071Onm8 o400- 716 ezI-a. C,-5 EDC,0-4 .E 150C)4-Ca)C,2 .o 100-1>- a.1 <a..00 5 10 15Time (ms)FIGURE 5.3: Type A2 Combustion Pulse for MK Concentrate(a) Pyrometer Output Voltage and Calculated Temperature(b) Pyrometer Output Voltage and Apparent DiameterConditions: Furnace Temp = 1273 K, P02 = 0.50 atm,Particle Size = 24-33 jim875 RESULTS1400__________40008lOiun____35005 1200 -3000 g10004--25000. nrn4- OiJU- 20000.. 600a)- 15004-a)2 40- 1000 i0a> 200 -- 5000 I 00 2 4 6 8 10Time (ms)1400 8- —— 8lOnm5 1200 -_-61000 I-a,4- 5 ,C. 800 - E4--4 .0600-a- 2•.14-aa,400a, a-20.0a 0.> 200 I <00 1 I I 00 2 4 6 8 10Time (ms)FIGURE 5.4: Type A3 Combustion Pulse Without Temperature Increase(a) Pyrometer Output Voltage and Calculated Temperature(b) Pyrometer Output Voltage and Apparent DiameterConditions: Furnace Temp = 1243 K, P0 = 1.0 atm,Particle Size = 24-33 p.m885 RESULTS250020001500>E4-4-0a,4-a)201>‘100050000 2 4 6 8Time (ms)10 12 14700060005000400030000200010000162500 1202000 1004- 80 4-c. 1500Eo 60 .h- 4j’fja) Iu,Jv 14- AfI C.tuC h.o rrin20 oa-0 0Time (ms)FIGURE 5.5: Type A3 Combustion Pulse With Temperature Increase(a) Pyrometer Output Voltage and Calculated Temperature(b) Pyrometer Output Voltage and Apparent DiameterConditions: Furnace Temp = 1253 K, P0 = 0.50 atm,Particle Size 24-33 I.Lm0 2 4 6 8 10 12 14 16895 RESULTS800 45008lOnm7lOnm 4000700 -_____5;3000 —tempcf13500 —.4-25004-4-0 2000I 0.C,4- 1500 EC)C,E- 1000 ‘—0a>.a- 100 5000 I I 00 2 4 6 8 10Time (ms)800 118lOamI - 10 oI 7lOnml 9700__I oiJ600 Ia,4- 7 4-a,0. 6 E4-.C,0 5 EI-C, 44- C,a, 2C,E 200I100 10 I I I 00 2 4 6 8 10Time (ms)FIGURE 5.6: Type A4 Combustion Pulse for MK Concentrate(a) Pyrometer Output Voltage and Calculated Temperature(b) Pyrometer Output Voltage and Apparent DiameterConditions: Furnace Temp = 1250 K, P02 = 0.50 atm,Particle Size = 24-33 pm905 RESULTS600 7000> 500--- 8lOnrnTemJ { 5000I-1 40004-z I016000I-I- 3000I- I4- 200a,-1 2000EI I-0 1- 1000>..100 2 4 6 8 10 12 14Time (ms)-600 87.25 500 4-400H H a,z 5C.4-4o6EI-3 4-a, C4-4) 4)..E0 C.I-C.100> I <0.0 I I .00 2 4 6 8 10 12 14Time (ms)FIGURE 5.7: Combustion Pulse Under Highly Oxidizing Conditions(a) Pyrometer Output Voltage and Calculated Temperature(b) Pyrometer Output Voltage and Apparent DiameterConditions: Furnace Temp = 1260 K, P0 = 1.0 atm,Particle Size = 24-33 im915 RESULTS14001350Kim and ThemeUsType A3, A4 PulsesType P.21300 10. 0E AA A(3E125o I I/TYPeAlIL I A A• Aa01200 a a I °a a aaaa a a11500 0.2 0.4 0.6 0.8Oxygen PartaI Pressure (atm)FIGURE 5.8: Predicted Ignition Threshold Conditions Using Data from Kim and Themelis[34] (25.tm particle) and Experimental Data (24-33irn size fraction)925 RESULTS1300• 401Jm1280 a 30pmx20pmC,a 15pmC,a lOiim1240 Decreasing1! 0 ParticleaE 1220 Size0C, • a1200 8CI...11801160 I0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9Oxygen Partial Pressure (atm)FIGURE 5.9: Predicted and Mean Experimental Ignition Threshold Conditions935 RESULTS30002500-2000 ------Sub-IgnitionIgnition1500-I____Distance from Burner Tip (mm)FIGURE 5.10: Particle Temperature Proffle Predicted by Model of Shook [111Sub-Ignition Conditions: Furnace Temp = 1200 K, P02 = 0.50 atm,Particle Size = 25jimIgnition Conditions: Furnace Temp = 1210 K, P02 = 0.50 atm,Particle Size = 25tm945 RESULTS2520—A— 1223K I0.—.—1123K1o15—.—1023K1010C,50 0.2 0.4 0.6 0.8Oxygen Partial Pressure (atm)FIGURE 5.11: Extent of Desuiphurization by Flash Oxidation955 RESULThFIGURE 5.12: Photomicrograph of Sub-Ignition Flash Reaction ProductConditions: Furnace Temp= 1023 K, P = 0.50 atm,Particle Size = 24-33 pm965 RESULTSFIGURE 5.13: Photomicrograph of Perforated CenosphereConditions: Furnace Temp= 1123 K, P02 = 0.50 atm,Particle Size = 24-33 pm975 RESULThFIGURE 5.14: Photomicrograph of Perforated CenosphereConditions: Furnace Temp= 1223 K, P02 = 0.10 atm,Particle Size = 24-33 I.tm985 RESULTS0.400.3500.30 V0.25U,0.20 I—--1223KC.—.—1123KC.0.15 1023K010z0.050.00 I0 02 0.4 0.6 0.8Oxygen Partial Pressure (atm)FIGURE 5.15: Depletion of Copper by Flash Oxidation995 RESULTSFIGURE 5.16: Photomicrograph of Collapsed CenosphereConditions: Furnace Temp= 1223 K, P0 = 0.50 atm,Particle Size = 24-33 p.m1006.1 Particle Combustion Temperatures6 DISCUSSION6.1 Particle Combustion TemperaturesSeveral important inferences can be drawn from inspection of the characteristics of thecombustion pulses observed, and the effect that the reaction conditions had upon the reactionprogress.Firstly, a plateau temperature was maintained at the melting point of chalcocite in some ofthe Type Al combustion pulses obtained at furnace conditions less oxidizing than thoserequired for ignition, Figure 5.1. This experimcntal observation substantiates the assumptionof Shook [11] (discussed in Section 2.2) that the consumption of copper oxide by reactionwith molten chalcocite may be approximated as being instantaneous.It is thermodynamically unfavourable for chalcocite and copper oxide to co-exist, but thereaction by which the suiphide and oxide compounds form SO2 and metallic copper(Equation [2.7]) does have a finite rate, as was shown by Byerley et al [52]. However, asustained plateau temperature was observed at the melting point of chalcocite (Figure 5.1)which is indicative of a required elapsed time for the exothermic reactions (Equation [2.7] aswell as the continuing oxidation of chalcocite, Equation [2.8]) to discharge the quantity oflatent heat required to melt the chalcocite remaining in the particle. Therefore, theassumption that thermodynamic equilibration, Equation [2.7], occurs instantaneously isjustified. While the reaction is not truly instantaneous, it is likely to be virtually completebefore particle heating progresses to the next stage of the reaction; the oxidation of metalliccopper.1016.1 Particle Combustion TemperaturesThe model predictions of the particle temperature profile of MK particles undergoing flashoxidation at conditions which define an ignition threshold (Figure 5.10) do not show aplateau at the melting point of chalcocite. Therefore, it is possible, that in some cases, thereaction may be arrested at this point. The reaction may be arrested by an unpredictableoccurrence such as venting of the sulphur dioxide gas produced within the particle as aproduct of the reaction of copper oxide with chalcocite (Equation [2.7]). This venting wouldincrease the rate of heat transfer from the particle, and thus influence the balance of heatgeneration and heat transfer which determines particle ignition.The combustion pulses also revealed that the boiling point of copper was reached (Figures5.2 through 5.6), and that material ejection and/or particle disintegration occurred at thecopper boiling point. These incidents were demonstrated by the appearance of Type Alpulses at the boiling point of copper, and Type A3 and A4 pulses showing dramatic apparentdiameter changes (Figures 5.5 and 5.6). Shook had speculated that dust generation in theflash converting of MK concentrate was the result of material ejection and particledisintegration due to the action of the volume of gas generated by the boiling of metalliccopper. This hypothesis is substantiated by evidence obtained in this work.Tuifrey [22-25] explained the fluctuating energy profile of Type A4 combustion pulses asevidence of three possible physical phenomena: area pulsation, a pulsating flame front (orphotosphere) or particle rotation. Only the last is a reasonable explanation for thephenomenon observed in the flash reaction of chalcocite.Particle rotation is evidenced by the frequency of the fluctuations observed in Type A4combustion pulses, Figure 5.6, which was on the order of 500Hz. Moreover, the period ofthe fluctuation was seen to change slightly from cycle to cycle, by about 5 to 10%, in all1026.1 Particle Combustion TemperaturesType A4 combustion pulses. The rate of data acquisition (50kHz) was sufficiently high toestablish that this observation was not an experimental artifact, but that the speed of rotationwas indeed varying.The fluctuating change in calculated apparent particle diameter in the case of chalcocite TypeA4 combustion pulses is on the order of 5 to 10 fold. If this observation represented a truechange in particle size with a frequency of 500Hz, the resultant destructive force wouldlikely fracture the particle after very few cycles. Similarly, the fluctuation may beattributable to a pulsating flame front or photosphere due to the ejection of surges of materialin the flash smelting of larger particles (the work of Tuffrey involved 65 to 125pm sizedparticles) but this would quickly result in disintegration of smaller (less than 30p.m) particles.Moreover, the venting of pulses of gas with a frequency of 500Hz would require the cyclicopening and closing of a surface orifice at an impossibly high rate.The periodicity and frequency of the energy, temperature and apparent diameter fluctuationin the Type A4 combustion pulses can be best explained as evidence of particle spin. Theparticles are induced to spin by the ejection of a jet of gas (copper vapour or sulphur dioxide)from a surface rupture with a tangential component to the jet trajectory. The variability inthe period of rotation is consistent with the phenomenon of continued discharge of a gas jetwith an unsteady volumetric flow rate.Exceedingly high temperatures were noted in combustion pulses obtained under the mosthighly oxidizing conditions. The high observed temperature represents the temperature ofthe products of the oxidation of copper vapour in the photosphere surrounding the particle. Acalculation of the temperature that would be reached by a chalcocite particle reactingadiabatically to produce a contiguous particle of Cu20 shows that the maximum temperature1036.2 Mathematical Modeling of Reaction Kineticscannot exceed approximately 3700 K. This calculation was performed using publishedthermodynamic data [56] for the enthalpies of reactants and products, accounting for thelatent heats of fusion of chalcocite and copper oxide.Copper vapour, however, when ejected from the particle due to boiling, would oxidizeaccording to the photosphere reaction in Equation [5.1], for which the maximum allowabletemperature of the product Cu20, as determined by the enthalpy released in the exothermicreaction, would be much higher. Calculation of the change in Gibb’s free energy forEquation [5.1] from published enthalpy and entropy data [56] indicates that the reaction isthermodynamically unfavourable at temperatures in excess of 25 80K.It must be noted, however, that observations of the maximum calculated particle temperatureshowed a distinct gap between peak temperatures of less than 3700 K, and peak temperatureswell in excess of 3700 K. This discontinuity, illustrated in Figure 6.1, represents thedifference between measurements of the particle temperature and measurements of thetemperature of the reacting photosphere generated by highly oxidizing furnace conditions.6.2 Mathematical Modeling of Reaction KineticsThe activation energy of the low temperature, chemically controlled oxidation reaction ofchalcocite was adjusted in the model of the flash reaction of single particles, developed byShook [11], such that the model predictions of particle ignition temperatures closely fit theexperimental data, Figure 5.9.The good correlation between the model predictions and the experimental data (Figure 5.9)confirmed that the kinetics of the low temperature, chemically controlled oxidation reactionof chalcocite follow Arrhenius type behaviour, with the reaction kinetics first order with1046.2 Mathematical Modeling of Reaction Kineticsrespect to oxygen partial pressure. The predicted effect of oxygen partial pressure comparesfavourably to the observations. The fact that the reaction is first order with respect to oxygenimplies that the rate determining step in the overall reaction mechanism is an intermediatereaction involving molecular oxygen. Morland [54] determined that the formation of a S-Oactivated complex from adsorbed monatomic oxygen is the rate determining step in theexchange of radioactive S35 between SO2 gas and chalcocite melts, as discussed in Section2.2.2. If this were the case in the low temperature chemically controlled oxidation reactionof chalcocite then the kinetics observed would be one-half order with respect to oxygen,according to the atomic collision theory of chemical reaction kinetics [55].The value of the activation energy of the chemical reaction determines the temperaturedependence of the reaction rate -- the activation energy is the slope of the Arrhenius plot ofthe log of the reaction rate versus reciprocal temperature (Figure 2.1)-- thus varying thevalue of the activation energy used in the model calculations showed a marked effect on thetemperature dependence of the predicted particle ignition threshold. Conversely, sensitivitytests showed that the inherent inaccuracies in the estimated heat and mass transfer ratesshowed only minor effect on the temperature dependence of the ignition threshold. Whilethe rates of heat and mass transfer are also temperature dependent functions, their effect onthe predicted ignition threshold is dwarfed by the effect of the reaction kinetics; the mostimportant unknown quantity in the development of the model.The experimental data showed that the particle ignition temperature increased withdecreasing oxygen partial pressure and increasing particle size. These are two factors whichimpede chemical reaction rate-- the chemical reaction rate is first order with respect tooxygen, so low oxygen partial pressure impedes the reaction, and larger particles have a1056.3 Photomicrographic Analysis of Reaction Productslower reacting surface area to particle volume ratio, thus impeding the reaction.6.3 Photomicrographic Analysis of Reaction ProductsThe photomicrographs of the particles reacted at low oxidizing conditions showed that theparticles behaved independently, in that some exhibited reaction progress to the extent thatthe released heat of reaction exceeded the heat transfer from the particle, thus enabling someparticles to reach the chalcocite melting point, while others merely show rounding at theedges due to oxidation occurring only at the most accessible sites, Figure 5.12. Moreimportantly, the chemical analysis and photomicrographic evidence obtained further verifiesthe bi-modal dust generation mechanism proposed by Shook and discussed in Section 2.2.2.The evidence also suggests that the phenomenon of particle fragmentation by production ofcopper vapour within the particle is enhanced by the formation of a nickel enriched shell atthe outer surface of the particle, Figure 5.16. The origin of the nickel enriched outer shell isexplained thus:• oxidation progresses beyond the melting point of chalcocite, and to a point wherethere is a substantial metallic copper content in the particle• molten copper preferentially evaporates from the surface of the particle• the particle is subsequently left encapsulated by an outer layer enriched in nickel• the nickel oxide content of the outer layer imparts a degree of structural integrity tothe periphery of the particle so that it resists expansion1066.3 Photoinicrographic Analysis of Reaction Productscontinued heat generation by oxidation of the encapsulated molten copper (bydiffusion of oxygen through the shell, or by electrochemical oxidation as described byAlyaser [3 8,39]) leads to boiling and the generation of copper vapour within theparticle and a subsequent increase in internal pressurethe built-up pressure is relieved through the sudden release of copper vapour eitherthrough a perforation in the outer shell or by a catastrophic explosion resulting inparticle fragmentation.Figure 5.16 is an example of a collapsed cenosphere with a greatly enriched nickel content.Also seen is the perforation in the shell through which the copper vapour escaped. It isevident that the shell first expanded beyond the original particle size (24-33tm) before theperforation and material ejection occurred.The preferential evaporation of copper over nickel is supported by published thermodynamicdata [56]. Assuming unit activity for both copper and nickel in the particle, the ratio ofcopper vapour to nickel vapour pressure as a function of particle temperature, Figure 6.2,shows that copper evaporates preferentially, at all particle temperatures. Thisthermodynamic precept, when compounded with the recognition that the concentration ofcopper in the particle (and hence its activity) is much greater than that of nickel-- the flashoxidation of chalcocite produces metallic copper by Equation [2.8]-- suggests that thetendency for the preferential evaporation of copper is seen to be even stronger for flashreacting MK particles than for the pure substances.The low concentrations of nickel detected in the dust collected during the pilot plant trials,Table 5.5, could originate from two phenomena. Firstly, catastrophic fragmentation of the1076.3 Photomicrographic Analysis of Reaction Productsnickel-rich shell due to rapid generation of copper vapour in the core of the particle wouldgenerate small nickel-rich particles which may be carried in the off-gas as a suspended solidand report as dust. Moreover, the jet of copper vapour venting from the core of the particlethrough a surface rupture would entrain molten core material, including nickel at theconcentration occurring in MK, which would rapidly oxidize to form finely divided particlesand thus report as dust. In this way, a minor amount of nickel would be carried into the duststream.The oxidation of chalcocite via an electrochemical reaction, as described by Alyaser [38,39]and discussed in Section 2.2.2, is also consistent with the observed phenomenon of dustgeneration by material ejection. Saturation of the molten chalcocite with dissolved oxygen isrequired to establish the conditions which Alyaser found to be conducive to electrochemicaloxidation. The initial oxygen content of the MK feed is approximately 1 to 2 weight percent(Table 4.1), and as desulphurization of the concentrate progresses, the transition point atwhich electrochemical oxidation begins would be encountered. Alyaser observed thegeneration of sulphur dioxide bubbles, the product of the oxidation half-reaction, within themelt. The reduction half-reaction was the reduction of cuprous ion (Cu) to metallic copper.In the case of the flash oxidation of MK particles, the coupled electrochemical reactionswould be the oxidation of suiphide (S2) by oxygen gas to form SO2 at, or near, the surface ofthe particle (where the local oxygen potential would be greatest), and the reduction ofcuprous ion to metallic copper in a region of lower local oxygen potential; that is, within theparticle. Therefore, gaseous material ejected from the particle to form dust may be eithersulphur dioxide or copper vapour. However, the fact that material ejection and particlefragmentation was observed in conjunction with particle temperatures at, or above, the1086.4 Pilot Plant Studiesboiling point of copper indicates that the generation of copper vapour is responsible for dustgeneration. The change in apparent particle diameter under non-dusting conditions may beattributed to particle expansion due to the formation of SO2bubbles near the surface of theparticle.6.4 Pilot Plant StudiesThe pilot plant results demonstrated that the dust generation rate was increased in those trialsthat were run with oxygen feed rates in excess of the calculated stoichiometric requirement.This fact was discussed by Shook [11] who cited an induced increase in the dust generationrate as corroboration of the model predictions. Increased dust generation due to a decrease inthe MK/02feed ratio is also supported by anecdotal observations during pilot plant trialsreported in this work, where the cyclone dust collection rate was seen to increase duringupset conditions in which the MK concentrate feed rate was reduced.Shook stated that dust generation in a well-mixed, two-phase flash flame occurred in twolocations in the flame: at the periphery of the jet where gas and particles temperatures werehighest, and near the centre of the jet where oxygen concentrations were highest. Byincreasing the oxygen feed rate in the pilot plant trials, both mechanisms of dust generationwere enhanced. Increased oxygen generated a more energetic and thus higher temperaturereaction at the periphery of the jet, as well as increasing the oxygen concentration at thenear-centre dust generation region.The pilot plant campaign went on to demonstrate that the flash flame could also bemanipulated to decrease the dust generation rate, as discussed in Section 5.4.4. Pilot planttrials 16 through 18 were performed with a burner design that introduced the oxygen feed as1096.4 Pilot Plant Studiesa very high gas velocity jet (240 to 270 m s1) surrounded by an annulus of freely fallingconcentrate, burner Type 7 in Figure 4.10. The success of this burner design was attributedto the flow patterns established due to the high energy axial oxygen jet. The tendency of thejet to entrain furnace gas was enhanced by its high velocity; thus the jet drew hot SO2through the stream of MK. This led to heating of the MK particles prior to their exposure tooxygen, and rapid dilution of the oxygen feed. These two factors produced a flash flamewith a reduced tendency towards dust generation. Dust generation at the periphery of the jetwas reduced due to the lack of oxygen in this region; the only oxygen present at theperiphery of the jet was that which was recirculated vertically up the reactor shaft.Moreover, dust generation at the near-centre region was reduced by rapid entrainment of SO2into the oxygen jet (lowering the oxygen partial pressure) and by lowering the location of thenear-centre dust generation region, likely to an imaginary point beyond the hearth.Pilot plant thals using burner designs that employed high velocity oxygen injection showedthat lowering the location of the near-centre dust generation region increased the inducedrecirculation in the furnace shaft. Table 5.6 shows that the amount of material collected onthe wall of the reactor shaft is greater in the trials using high velocity oxygen injection. Anincrease in the amount of material adhering to the wall of the reactor can be seen to be adirect consequence of higher recirculation, which would carry suspended solids up the shaft,rather than allowing them to settle into the hearth. The high recirculation rate, in turn,indicates that entrainment of furnace gases (primarily SO2) is increased by high velocityoxygen injection. Recirculation of furnace gases up the shaft is driven by a pressuredifferential between the bottom and top of the reactor shaft induced by the entrainment anddownward transport of furnace gas by the oxygen jet.1106.4 Pilot Plant StudiesThe practicality of the high velocity oxygen injection burner design and the subsequent flashflame configuration as applied to a full scale production facility is questionable, since theInco flash converter geometry differs because it depends on horizontal injection; and thescaled-up mass feed rates of a production plant would offer different flow patterns.However, the trials did establish that dust generation rates could be reduced by manipulationof the flash flame.The high nickel content of the material that collected on the reactor shaft wall, as comparedto the material that settled in the hearth, (Table 5.4) could be the result of the preferentialcollection of nickel-rich material (collapsed cenospheres like that shown in Figure 5.16) orthe depletion of copper from the material adhered to the shaft wall. There is no reason tobelieve that the nickel-rich material would preferentially collect on the reactor shaft wall -- itwould be no more adhesive than molten copper or molten unreacted chalcocite, and it wouldbe no more prevalent as a suspended solid in the recirculating gas than it would be as asuspended solid in the material collected in the settling chamber (Figure 4.8). The nickelcontent of the material collected in the settling chamber was similar to that of the cyclonedust.The enrichment of nickel in the shaft wall material is therefore a result of the preferentialevaporation of copper over nickel. As discussed in Section 6.3, and demonstrated by Figure6.2, the vapour pressure of copper is much greater than that of nickel, by more than threeorders of magnitude at the lower temperatures experienced by the material adhering to thereactor shaft wall, 1100 to 1400K (Table 5.2). Over the course of the trial, and thesubsequent cooling period prior to obtaining a sample for analysis, copper would be depletedfrom the material adhering to the wall by evaporation. Glen and Richardson [351 showed1116.4 Pilot Plant Studiesthat the evaporation of copper is enhanced by condensation (or gas-phase oxidation) in thesteep temperature gradient near the particle surface, thus reducing the near-surface copperpartial pressure and generating a greater driving force for evaporation. Indeed, in theexperiments of Glen and Richardson the evaporation rate of copper was increased by a factorof three over the predicted evaporation rate. Jorgensen [15] determined that coppervapourized during flash reactions can be in the form of monomeric copper, cupric or cuprousoxide, and will readily oxidize to form finely divided Cu20dust.1126 DISCUSSION7000— 6000-C,14-IC,04000-. B3000- 200010001160 1180 1200 1220 1240 1260 1280 1300Furnace Temperature (K)FIGURE 6.1: Peak Particle Temperatures Determined from Pyrometer MeasurementsConditions: P02 0.50 atm, Particle Size = 24-33 urn1136 DISCUSSION4 0->--2__-4Cu/NiU)Cu --6- Ni 01--‘-o I-oI..--10 az00--140 : : : -161000 1500 2000 2500 3000 3500Temperature (K)FIGURE 6.2: Calculated Copper-to-Nickel Vapour Pressure Ratio1147 SUMMARY AND CONCLUSIONS7 SUMMARY AND CONCLUSIONSThe important findings of this work are:The conditions defining the MK particle ignition threshold were discerned, and thedependence of the particle ignition threshold on the furnace temperature, oxygen partialpressure and particle size was established. This information was compared to theignition threshold conditions predicted by a mathematical model to evaluate thekinetics of the initial low temperature oxidation of chalcocite, Equation [2.61. Thereaction followed Arrhenius type behaviour, was first order with respect to oxygenpartial pressure, and the activation energy was determined to be 460 U mo11.2. The depletion of copper from particles during flash oxidation, and from the materialadhering to the reactor shaft wall, was explained as being the result of preferentialevaporation of copper. The particle combustion pulses also indicated the loss of copperfrom the particles due to boiling, and subsequent oxidation in a reacting photospheresurrounding the particle.3. The bi-modal mechanism of dust generation in the flash converting of MK concentratethat was proposed by Shook [11] was confirmed and refined. It was established thatthe preferential evaporation of copper led to the formation of a nickel-rich shellencasing the particle, and subsequently to dust generation. Using the terms defined byShook; “mechanical” dust was generated by fragmentation of the nickel-rich shell dueto violent release of vapour generated by copper boiling within the shell, and finelydivided copper oxide, or “chemical” dust, was generated by the oxidation of the ventedcopper vapour and the copper vapour in the photosphere surrounding the reacting1157 SUMMARY AND CONCLUSIONSparticles.4. It was shown that the flash converting flame can be manipulated by altering the burnerdesign in such a way as to alter the gas and particle distribution patterns, and thusreduce dust generation. Three pilot plant trials using a novel burner design, employinga high velocity oxygen jet along the axis of the reactor surrounded by freely falling MXconcentrate, showed increased furnace gas recirculation and reduced dust generationrates -- to approximately one-third of the usual dust generation level.116REFERENCESREFERENCES1 Matousek J.W., “The Influence of Concentrate Grade on the Performance of a CopperFlash Smelting Furnace”, CIM Bulletin, 86 (971), pp 126-129 (1993)2 Jorgensen F.R.A. and Elliot B.J., “Flash Furnace Shaft Evaluation Through Simulation”,Pubi. Australas. Inst. Mm. Metal!, 9/92, pp 387-94 (1992)3 Perillo A., Carminati A., Carlini G. and Thba R., “The Kivcet Lead Smelter atPortovesme. Comnilsioning and Operating Results”, Met. Soc. AIME, TMS TechnicalPaper #A88-2 (1987)4 Fern P.F. and Perillo A., “The New Lead Smelter at Portovesme”, in ExtractionMetallurgy ‘85,1MM, London, 1985, pp 891-9035 Reimers J.H. and Taylor J.C., “The Future of Lead Smelting”, in Advances in SuiphideSmelting. Vol. 1, Eds. H.Y. Sohn, D.B.George and A.D. Zunkel, Met. Soc. AIME, 1983,pp 529-5516 “NDP, Cominco Forge Deal for Trail Smelter: New Plant will use RussianTechnology”, The Vancouver Sun, March 5, 1994, p C77 Landolt C.A., Fritz A., Marcuson S.W., Cowx R.B. and Miszczak J., “Copper Making atInco’s Copper Cliff Smelter”, in Copper 91 - Cobre 91, Diaz C., Landolt C., Luraschi A.and Newman C.J. editors, Pergamon Press, Elmsford, NY, 1991, pp 15-298 Bustos A.A., Brimacombe J.K. and Richards G.G., “Heat Flow in Copper Converters”,Met. Trans. B, 17B (4), pp 677-685 (1986)9 Asteljoki J.A. and Mueller H.B., “Direct Smelting of Blister Copper -- Pilot FlashSmelting Tests of Olympic Dam Concentrate”, in Pvrometallurgv 87, Inst. Mining andMetallurgy, 1987, pp 19-5210 Smith T.J.A, Poesner I. and Williams C.J., “Oxygen Smelting and the Olympic DamProject”, International Svmvosium on the Impact of Oxygen on the Productivity ofNon-Ferrous Processes, Winnipeg, Aug 23-26, 1987, pp 49-5911 Shook A.A., “Flash Converting of Chalcocite Concentrate: A Study of the Flame”,Ph.D. Thesis, The University of British Columbia, 199212 Otero A., Brimacombe J.K. and Richards G.G., “Kinetics of the Flash Reaction OfCopper Concentrates”, in Copper 91- Cobre 91, Diaz C., Landolt C., Luraschi A. andNewman C.J. editors, Pergamon Press, Elmsford, NY, 1991, pp 459-47313 Kostkowski H.J., “The Accuracy and Precision of Measuring Temperatures Above10000K”, Procedings of the International Symposium on High Temperature Technology,McGraw-Hill, New York (1959) pp 33-44117REFERENCES14 Jorgensen F.R.A, and Zuiderwyk M.A., “Some Applications of Two-Colour Pyrometryto Pyrometallurgical Research”, The Aus. I.M.M. Melbourne Branch, Symposium onExtractive Metallurgy, November, 1984, pp 3 17-32115 Jorgensen F.R.A, and Zuiderwyk M., “Two-Colour Pyrometer Measurement of theTemperature of Individual Combusting Particles”, J. Phys. E: Sci. Instrum., l8pp486-491 (1985)16 Mishin J., Vardalle M., Lesinski 3. and Fauchais P., “Two-colour Pyrometer for theStatistical Measurement of the Surface Temperature of Particles Under Thermal PlasmaConditions”, 3. Phys. E: Sci. Instrum., 20 pp 620-625 (1987)17 Hahn J.W. and Rhee C., “Reference Wavelength Method for Two-Colour Pyrometer”,Applied Optics 26 (24) pp 5276-5279 (1987)18 Andreic Z, Svenda K. and Persin A., “Two-Year Experience With MultiwavelengthOptical Pyrometry”, Proc. SPIE- mt. Soc. Optical Engineering, 807, pp 79-85 (1987)19 Hahn J.W. and Rhee C., “Calculation of Temperature Error in a Two-Colour PyrometerDesigned With the Reference Wavelength Method”, Applied Optics 25 (10) pp1916-1918 (1988)20 Lafollette R.M., Hedman P.O. and Smith P.J., “An Analysis of Coal ParticleTemperature Measurements with Two-Colour Optical Pyrometers”, Combust. Sci. andTech., 66, pp 93- 105 (1989)21 Luther-Davies B., Radlinski A.P. and Cailca A., “Two-Channel Optical Pyrometry ofMetals Irradiated by Picosecond Laser Pulses”, 3. Appl. Phys. 66 (7), pp 3293-3297(1989)22 Tuifrey N.E., “Pyrometry Studies of the Combustion of Lead Concentrate Particlesunder Controlled Conditions”, Ph.D. Thesis, The University of British Columbia, 198923 Tuffrey N.E., Richards G.G. and Brimacombe J.K., “Two-Wavelemgth Pyromerty Studyof the Combustion of Sulphide Minerals. Part I: Apparatus and General Observations”,Accepted by Met.Trans. B, 199424 Tuffrey N.E., Richards G.G. and Brimacombe J.K., “Two-Wavelemgth Pyromerty Studyof the Combustion of Suiphide Minerals. Part II: Galena snd Commercial LeadConcentrates”, Accepted by Met.Trans. B, 199425 Tuffrey N.E., Richards G.G. and Brimacombe J.K., “Two-Wavelemgth Pyromerty Studyof the Combustion of Sulphide Minerals. Part Ill: The Influence of OxygenConcentration on Pyrite Combuston”, Accepted by Met.Trans. B, 199426 Jokilaakso A., Yang Y., Teppo 0. and Ahokainen T., “Recent Research Activities on theOutokumpu Flash Smelting Process for Suiphidic Non-Ferrous Concentrates at HelsinkiUniversity of Technology”, in Proceedings of the International Conference on Mining118REFERENCESand Metallurgy of Complex Nickel Ores, Chongyue F., Huanhua H. and Chuanfu Z.editors, Jingehang, China, September 5-8, 1993, International Academic Publishers,Beijing, P.R. China, pp 33-4227 Ruotto S., “The Description of a Mathematical Model for the Flash Melting ofCu-Concentrates”, Combustion and Flame, 34 (1979), pp 1-1128 Hahn Y.P. and Sohn H.Y., “Mathematical Modeling of Sulphide Flash SmeltingProcess: Part I. Model Development and Verification with Laboratory and Pilot PlantMeasurements for Chalcopyrite Concentrate Smelting”, Met. Trans. B, 21B (12), pp945-958 (1990)29 Hahn Y.P. and Sohn H.Y., “Mathematical Modeling of Sulphide Flash SmeltingProcess: Part II. Quantitative Analysis of Radiative Heat Transfer”, Met. Trans. B, 21B(12), pp 959-966 (1990)30 Seo K.W. and Sohn H.Y., “Mathematical Modeling of Sulphide Flash Smelting Process:Part Ill. Volatilization of Minor Elements”, Met. Trans. B, 22B (12), pp 79 1-799 (1991)31 Kim Y.H. and Themelis N.J., “Effect of Phase Transformation and ParticleFragmentation on the Flash Reaction of Complex Metal Sulfides”, in InnovativeTechnology and Reactor Design in Extraction Metallurgy, Proceedings of the R.Schuhmann Symposium, eds. D.R. Gaskell, J.P. Hager, J.E. Hoffmann and P.J. Mackey,TMS/AIME, pp 349-369 (1986)32 Walsh P.M., Zhang M., Farmayan W.F. and Beer J.M., “Ignition and Combustion ofCoal-Water Slurry in a Confined Turbulent Diffusion Flame”, in Proceedings of theTwentieth (International) Symposium on Combustion, The Combustion Institute, 1984,pp 1401-140733 Kolb T., Farmayan W.F., Walsh P.M. and Beer J.M., “The Contribution of Radiation tothe Ignition of Coal-Water Slurry Diffusion Flame”, Combust. Sci. and Tech., 58, pp77-95 (1988)34 Kim Y.H. and Themelis N.J., “Rate Phenomena in the Oxidation of Zinc, fron andCopper Sulphide Pellets”, Can. Met. Quart., 26 (4), pp 341-349 (1987)35 Glen C.G. and Richardson F.D., “Kinetics of the Oxidation of Liquid Copper and theEffects of Interfacial Silica”, in Heterogeneous Kinetics at Elevated Temperatures:Proceedings of an International Conference in Metallurgy and Materials Science, TheUniversity of Pennsylvania, September 8-10, 1969, Belton G.R. and Worrell W.L.editors, Plenum Press, New York, 1970, pp 369-39 136 Ajersch F. and Toguri J.M., “Oxidation Rates of Liquid Copper and Liquid CopperSulphide”, Met. Trans. 3, pp 2187-2193, (1972)37 Ajersch F. and Benlyamani M., “Thermogravimetric Identification and Analysis ofReaction Products During Oxidation of Solid or Liquid Sulfides”, Thermochimica Acta,143, pp 221-237 (1989)119REFERENCES38 Alyaser A.H., “Oxidation Kinetics of Molten Copper Suiphide”, M.A.Sc. Thesis, TheUniversity of British Columbia, 199339 Alayser A.H. and Brimacombe J.K.B., “Oxidation Kinetics of Molten Copper Sulphide”,Accepted by Met. Trans. B, 199440 Atkins P.W., Physical Chemistry, Second Edition, W.H. Freeman and Company, SanFrancisco, 198241 Henderson T.A., “The Oxidation of Powder Compacts of Copper-lion Suiphides”, Bull.Inst. Mm. and Met., 620, pp 497-520 (1957)42 Lewis J.R., Hamilton J.H., Nixon J.C. and Graverson C.L., “The Oxidation ofChalcocite in Air Compared with its Oxidation in Pure Oxygen”, Trans AIME, 1948, pp177- 18543 Wadsworth M.E., Leiter K.L. Porter W.H. and Lewis J.R., “Suiphating of CuprousSuiphide and Cuprous Oxide”, Trans. Met. Soc. AIME, 218, pp 5 19-525 (1960)44 Rao V.V.V.N.S.R. and Abraham K.P., “Kinetics of Oxidation of Copper Sulfide”, Met.Trans 2 (9), pp 2463-2470 (1971)45 Asaki Z., Ueguichi A., Tanabe T. and Kondo Y., “Oxidation of Cu2S Pellet”, Trans. J.Inst. Metals, 27 (5), pp 361-37 1 (1986)46 Jokilaakso A.T., Suominen R.O., Taskinen P.A. and Lilius K.R., “Oxidation ofChalcopyrite in Simulated Suspension Smelting”, Trans. Inst. Mi Mettall. C, 100, ppC79-C90 (1991)47 Sohn H.S., Kumazawa H., Fukunaka Y. and Asaki Z., “Non-Isothermal Oxidation ofCopper Concentrate Particles Falling in a Vertical Tube: Effect of Composition on theInterfacial Reaction Rate of Copper Concentrate”, Metall. Review of MMII, 8 (2), pp34-52 (1992)48 Dunn J.G. and Mackey L.C., “The Measurement of Ignition Temperatures and Extentsof Reaction on lion and lion-Nickel Suiphides”, J. Thermal Analysis, 37, pp 2143-2164(1991)49 Ranz W.E. and Marshall W.R.Jr., “Evaporation from Drops, Part I”, Chem. Eng.Progress, 48 (3), pp 141-146 (1952)50 Ranz W.E. and Marshall W.R.Jr., “Evaporation from Drops, Part II”, Chem. Eng.Progress, 48 (4), pp 173-180 (1952)51 Chemical Engineer’s Handbook, Sixth Edition, R.H. Perry ed., McGraw-Hill, New York(1991)52 Byerley J.J., Rempel G.L. and Takebe N., “Interaction of Copper Sulfides with CopperOxides in the Molten State”, Met. Trans. 3 pp 2501-2506 (1974)120REFERENCES53 Schmid R., “A Thermodynamic Analysis of the Cu-O System with an AssociatedSolution Model”, Met. Trans. B, 14B, pp 473-481 (1983)54 Morland P.T., Matthew S.P. and Hayes P.C., “Kinetics of S35 Exchange betweenS02/CO/C0 Mixtures and Copper Suiphide Melts at 1523 K”, Met. Trans. B, 22B, pp21 1-217 (1991)55 Avery H.E., Basic Reaction Kinetics and Mechanisms, MacMillan Press Ltd., London,198256 Rao Y.K., Stoichiometrv and Thermodynamics of Metal1urical Processes, CambridgeUniversity Press, Cambridge, 198557 Whitaker S., Fundamental Principles of Heat Transfer, Krieger Publishing, Malabar Ha.(1985)121Al 1 Particle Temperature MeasurementAPPENDIX A.I: TWO-WAVELENGTH PYROMETRYA.I.1 Particle Temperature MeasurementThe mathematical derivation which facilitates the measurement of particle temperatures bytwo-wavelength pyrometry (also called two-colour or ratio pyrometry) is based onfundamental thermal radiation theory.Planck’s law of radiation states the relationship between the temperature of an ideallyemitting blackbody and the intensity of monochromatic radiation that is emitted:2thc5 C1)Wbb==[Al.l1‘rThis equation can be simplified by the Wien’s Law approximation, which assumes that(c2expjJ>> 1With the pyrometer used in this work, employing 710 and 810 nm filters, the Wien’s Lawapproximation is valid for measured temperatures less than approximately 20 000 K.Equation [Al.1] then becomes:Wbb = Ci25exp() [A1.21122Al. 1 Particle Temperature MeasurementIn a real experiment, the emitter must be treated as a non-blackbody, and the mono-chromaticemissivity (defmed as the ratio of the intensity of radiation from a non-blackbody to that ofan ideal blackbody at a given wavelength and temperature) must be factored into Equation[A 1.2], yielding:W,,= eCi25exp() {A1.3]Taking the ratio of the radiative intensities at two different wavelengths from a singlenon-blackbocly at a uniform temperature gives:(w’ (ri , C2 1lnJ=ln_J J—i-(3c--5) [A1.4]If grey body behaviour is assumed, that is, the ratio of the emissivity at the two wavelengthsis assumed to be independent of temperature, then Equation [A1.4] reduces to a simplerelation with the ratio of the radiative intensities at the two wavelengths inverselyproportional to the temperature of the observed body. The equation takes the form:Aln-1J=j+B [A1.5]The quantity of energy measured by the pyrometer is a fraction total monochromaticradiative intensity; as defined by instrument dependent and object dependent factors. Thesignificant object dependent factor is the particle area (Ar). The pyrometer dependent factors123All Particle Temperature Measurementinclude the particle-detector view factor (FdP), the overall pyrometer gain including alloptical losses and electronic amplification (G) and the filter band-width of the opticalmonochromater (tSX). Thus the pyrometer output voltage on each monochromatic channel is:V,, =AFd..PGt\XW. [A1.61These parameters cannot be determined independently, but by combining several of theminto a single empirical constant (K), the Wien’s Law approximation (Equation [A 1.3]) can berewritten as:[A1.7]with the empirical constant determined by calibrationK=FdPG,ClThis expression shows that the pyrometer output voltage at the two separate wavelengths isstill dependent on the particle area and emissivity. Taking the ratio of the pyrometer outputvoltages at two wavelengths removes these factors and gives the useful form of Equation[A1.5]:(v” (K c C2[A1.8]Equation [A1.8] incorporates the assumption that the ratio of the emissivities of the particleat the two wavelengths is temperature independent.124A.L2 Apparent Particle Diameter MeasurementA.I.2 Apparent Particle Diameter MeasurementThe change in the apparent diameter of the particle is also calculated from the recordedpyrometer output voltages. While the true diameter of the particle is not calculable, thechange relative to the initial particle diameter can be derived.It is assumed that throughout a combustion pulse, changes in the pyrometer output voltageare independent of the properties of the pyrometer (e.g. signal gain and pyrometer-particleview factor) and are dependant only on the properties of the object observed (e.g.temperature, surface area and emissivity). From Equation [A 1.7], the ratio of the pyrometeroutput voltage at time t, to that at an arbitrary reference time t0, can be written for both of thewavelengths sampled:(-cv’’ ‘‘ , expi__.!— ph ‘).).T(t)— A() e(r0) ( -Cexp—-)As in the calculation of the particle temperature, it is necessary to assume that the emissivityof the particle is not a function of particle temperature. Thus, the change in apparentdiameter of the particle can be calculated.dQ) = dQ0) (•exp((-)_—)))) [A1.lOj125A12 Apparent Particle Diameter MeasurementThe change in the apparent diameter of the particle is presented as a ratio of the instantaneousdiameter to the diameter at an arbitrary time zero in the first few milliseconds of thecombustion pulse.126All Heating of Carrier Gas StreamAPPENDIX A.JJ HEATING OF CARRIER GAS STREAMAn estimation of the time required to heat the stream of cold carrier gas to the furnacetemperature was calculated assuming worst-case conditions for heat transfer. Grossassumptions were made concerning the nature of the system in order to simplify thetemperature determination, since the calculation was performed only to present a gauge ofthe expected lag in particle heating. The presence of the cold carrier gas caused a lag in theheating of the particles, and thus increased the ignition lift-off distance, while the particleignition threshold conditions were not effected, as discussed in Section 4.2.1 and 5.2.It was assumed that the carrier gas flowed as a cylindrical stream with no mixing of thecarrier gas and the surrounding furnace gas. The Biot modulus for the system was assumedto be much greater than unity; reflecting the assumption that the resistance to heat transfer isby conduction only. Moreover, it was assumed that the carrier gas was at a uniform initialtemperature of 298 K, and that the boundary between the carrier and furnace gases wasconstant, and equal to the furnace gas temperature.The axial temperature history of the system, as defined by Whitaker [57], is given as adimensionless temperature driving force which is a function of the thermal diffusivity of thegas (a), the radius of the stream (r) and the elapsed time (t):(Ta_T Ic,.i’[A2.1]The relationship between the dimensionless temperature driving force and the heat flux ispresented graphically by Whitaker.127All Heating of Canier Gas StreamThe axial temperature of the carrier gas stream was calculated by iteration, and the resultsshowed that the cold carrier gas stream is expected to reach the furnace temperature withinthe first 10 to 15 cm of the furnace.128


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items