No

This thesis work aimed to contribute to the development of online characterization techniques for non-metallic inclusions in steelmaking. The study focused on assessing the precision and accuracy of the optical emission spectroscopy pulse distribution analysis (OES/PDA) technique using asreference light optical microscope (LOM) and scanning electron microscope(SEM). Steel samples were collected from three different routes at SSAB Oxelösund steel mill, including vacuum degassing with aluminium or silicon as deoxidizers, and a direct route using magnetic stirring. The precision study compared the inclusion index obtained from each burn spot within the same sample. The precision study includes two parts, the first one compared the dispersion of a set of samples when testing four and eight burn spots, and the second part all the samples were tested to determine the overall dispersion using four burn spots. This study revealed that samples with eight burn spots exhibited higher dispersion and higher average inclusion index, indicating potentially higher inclusion content in the middle section of the sample. Overall, the precision of PDA/OES was found to be satisfactory, with the possibility of further improvement by eliminating outliers. The particle size distribution (PSD) and volume fraction of inclusions obtained from OES/PDA and LOM was compared, and it was observed that OES/PDA yielded higher values in most samples, suggesting limited agreement between the two techniques. Comparing the results of OES/PDA with scanning electron microscopy (SEM) for PSD and volume fraction of inclusions showed relatively better agreement, although accuracy could not be described as entirely accurate. The comparison of the B-factor indicated that OES/PDA generally provided lower values than SEM. SEM was more effective in capturing additions during the steelmaking process, while OES/PDA exhibited lower sensitivity. Finally, the comparison of inclusion chemical composition between OES/PDA and SEM indicated significant disagreement, emphasizing that OES/PDA should not be utilized for this purpose. The limited gathered data from the seven sampled heats indicate that OES/PDA is a precise tool that need more development to be consider accurate to be consider as a monitor tool during secondary metallurgy. / Detta examensarbete syftade till att bidra till utvecklingen av online-karakteriseringstekniker för icke-metalliska inneslutningar i ståltillverkning. Studien fokuserade på att bedöma precisionen och noggrannheten hos OES/PDA-tekniken (Optical Emission Spectroscopy Pulse Distribution Analysis) med hjälp av ljusoptiskt mikroskop (LOM) och svepelektronmikroskop (SEM)som referens. Stålprover samlades in från tre olika produktionsvägar vid SSAB Oxelösundsstålverk, inklusive vakuumavgasning med aluminium eller kisel som deoxidationsmedel, och en direkt produktionsväg med magnetisk omrörning. Precisionsstudien omfattar två delar, den första jämförde spridningen av en uppsättning prover vid testning av 4 och 8 brännfläckar, och den andra delen testades alla prover för att bestämma den totala spridningen med 4 brännfläckar. Denna studie visade att prover med 8 brännfläckar uppvisade högre dispersion och högre genomsnittligt inkluderingsindex, vilket tyder på potentiellt högre inkluderingsinnehåll i den mellersta delen av provet. Precisionen för PDA/OES var överlag tillfredsställande, med möjlighet till ytterligare förbättringar genom att eliminera avvikande värden. Partikelstorleksfördelningen (PSD) och volymfraktionen av inneslutningar som erhållits från OES/PDA och LOM jämfördes, och det observerades att OES/PDA gav högre värden i de flesta prover, vilket tyder på begränsadöverensstämmelse mellan de två teknikerna. Jämförelsen av resultaten från OES/PDA med svepelektronmikroskopi (SEM) för PSD och volymfraktion av inneslutningar visade relativt bra överensstämmelse, även om noggrannheten inte kunde beskrivas som bra. Jämförelsen av B-faktorn visade att OES/PDA i allmänhet gav lägre värden än SEM. SEM var mer effektivt för att fånga upp tillskott underståltillverkningsprocessen, medan OES/PDA uppvisade lägre känslighet. Slutligen visade jämförelsen av den kemiska sammansättningen mellan OES/PDA och SEM på betydande oenighet, vilket understryker att OES/PDA inte bör användas för detta ändamål. De begränsade data som samlats in från dem sju utvalda härdarna indikeraratt OES/PDA är ett exakt verktyg som behöver utvecklas ytterligare för att ansesvara korrekt som ett övervakningsverktyg under sekundärmetallurgi.

Chalcopyrite (CuFeS2) is the most abundant and the most economically important copper mineral. Increasing worldwide demand for copper accompanied by exhaustion of copper resources necessitate the development of new processes for treating lower-grade copper ores. Heap (bio)leaching of copper oxides and secondary sulphides (covellite (CuS) and chalcocite (Cu2S)) is a proven technology and a convenient process due to its simplicity and relatively low capital investment requirement. As a result, the technology appears to be an attractive process option for treatment of low-grade chalcopyrite ores as well. However, chalcopyrite is recalcitrant to leaching and bioleaching in conventional heaps: its dissolution is slow and it halts on a low level of copper extraction usually unacceptable for a commercial practice. Slow dissolution of chalcopyrite is attributed to the formation of compounds on the surface of the mineral during its oxidative dissolution, often termed “passivation”. There is still no consensus about the nature of the passivation layer. There are, however, four proposed candidates suggested in the literature: metal deficient sulphides, polysulphides, jarosite and elemental sulphur. This project was aimed to further investigate the chalcopyrite dissolution and its passivation.In this thesis, dissolution of a pyritic and a pure chalcopyrite concentrate was studied in stirred tank reactors in the absence and presence of moderately thermophilic microorganisms. The abiotic experiments were performed under accurately controlled redox potential conditions to reproduce the same oxidising conditions recorded during the bioleaching experiments. The results showed that the microorganisms did not have any effect in the copper leaching efficiency other than oxidation of ferrous ions to ferric ions. Comparing the copper dissolution rates in the experiments where bulk elemental sulphur was formed with those experiments where the elemental sulphur was oxidised to sulphate due to microbial activity showed that the bulk elemental sulphur did not hinder the dissolution. The same phenomenon was observed in case of bulk jarosite. Under otherwise identical leaching conditions, the presence of bulk jarosite did not decrease the leaching efficiency compared to those experiments were bulk jarosite was not formed. It was also shown that surface spectroscopic methodologies such as X-ray photoelectron spectroscopy (XPS) cannot be applied on powder leached samples due to interfering data from the bulk precipitated species. As a result, massive natural chalcopyrite samples were II prepared and used in the leaching experiments for XPS measurements. Different samples in different stages of leaching were extracted from the biotic and abiotic experiments and analysed by XPS. Results indicated that the surface elemental sulphur was not oxidised by bacterial activity. The data revealed that the common phases on the surface of the samples leached biotically and abiotically for different durations were elemental sulphur and iron-(oxy)hydroxides. The surface elemental sulphur was rigidly bound to the surface and did not volatise in the room temperature XPS measurements. Jarosite was observed in only one sample from the abiotic experiment, but no correlation between its presence and the slow dissolution could be made. Other minor surface compounds such as iron-sulphate was also observed in some samples with no contribution to the leaching behaviour. It was concluded that a multi-component surface phase consisting of mainly sulphur and iron-hydroxides were responsible for the slow chalcopyrite dissolution.

Recycling of secondary raw materials by non-ferrous metal producers is steadily increasing. These materials inevitably introduce various impurities to the production furnaces, which can change the chemical composition, and thereafter, the physical and chemical properties of the products and by-products.Alumina is one of the most common and significant impurities that enters the furnaces in different ways. Finding a balance between the recycling advantages and possible adverse effects of the increased alumina content in the process is of great interest to the metals industry.In the current work, effects of alumina on two of important properties (melting and leaching behaviours) of an actual industrial fayalite slag, from a Swedish copper producer, is investigated.A reference slag sample, devoid of alumina addition, and three different mixtures of slag with 5, 10, and 15 wt-% alumina addition were prepared. After homogenization (re-melting) of the mixtures, in an induction furnace, the materials were characterized using scanning electron microscopy (SEM) and X-ray diffraction (XRD). A standard single stage leaching test was used to study the leaching behaviour of the samples. Differential scanning calorimetry (DSC), Pt-Rh pans, was applied to investigate the melting and solidification properties of the mixtures.Results show that, olivine, spinel, and augite based solid solutions are the three main constituents of the samples. Addition of alumina to the slag increases the amount of spinel solid solution and changes its composition from iron rich to aluminum rich. The slag shows an alumina saturation limit between 5 and 10 wt-% alumina additions. Chemical composition of the major phases does not change above the saturation and excess amount of aluminum is consumed for formation of a new aluminum containing phase, anorthite. Below the saturation limit, alumina addition causes a relatively sharp increase in the melting temperature of the slag, and at the same time, an increase in the leached amounts of the elements is seen. Alumina addition, close to the saturation limit of the slag, lowers the leachability considerably, while the melting characteristics are not changed dramatically. However, above the saturation limit, a more complex transition occurs as the partial melting and leached amounts of most of the elements increase. In order to achieve an appropriate melting and leaching characteristics, the total alumina content of the slag must be close to its saturation limit. The grounds and influential parameters on such behaviours are thoroughly discussed.

Comminution test methods used within mineral processing have mainly been developed for selecting the most appropriate comminution technology for a given ore, designing a grinding circuit as well as sizing the equipment needed. Existing test methods usually require comparatively large sample amounts and are time-consuming to conduct. This makes comprehensive testing of ore comminution behavior – as required in the geometallurgical context – quite expensive. Currently the main interest in the conduct of comminution test lies in the determination of particle size reduction and related energy consumption by grindability test methods, which provide the necessary information about mill throughput. In this procedure mineral liberation is regarded as a fixed parameter due to missing this information in ore characterization as well as a lack of suitable comminution models. However, ignoring the connection between particle size and mineral liberation prevents the scheduling and controlling of the production process from being optimal.For these reasons new comminution tests need to be developed or alternatively the existing test methods need to be suited to geometallurgical testing where the aim is to map the variation of processing properties of an entire ore body. The objective of this research work is on the one hand to develop small-scale comminution test methods that allow linking comminution behavior and liberation characteristics to mineralogical parameters, and on the other hand establish a modeling framework including mineral liberation information.Within the first stage of the study the comminution of drill cores from Malmberget’s magnetite ore, classified by modal mineralogy and texture information, have been investigated. It was found that there is a direct correlation between the mechanical strength of the rock, as received from unconfined compressive or point load tests, and the crusher reduction ratio as a measure for crushability. However, a negative correlation was found between crushability and grindability for the same samples. The grindability showed inverse correlation with both magnetite grade and the magnetite’s mineral grain size. The preliminary conclusion is that modal mineralogy and micro-texture (grain size) can be used to quantitatively describe the ore comminution behavior although the applied fracture mechanism of the mill cannot be excluded.With crushed ore samples from Malmberget also grindability tests and mineral liberation analyses were conducted using laboratory tumbling mills of different size. Starting from the dimensions of the Bond ball mill a modified test method was developed where small size samples of approximately 220 g were pre-crushed and ground in a down-scaled one-stage grindability test. Down-scaling was done by keeping similar impact effects between the mills. Mill speed and grinding time were used for adjusting the number of fracture events in order to receive similar particle size distributions and specific grinding energy when decreasing mill size by the factor 1.63. A detailed description of the novel geometallurgical comminution test (GCT) is given.With respect to ore crushability and autogenous and semiautogenous grinding (AG/SAG) also drop weight tests were conducted. For a more accurate and precise measurement of the energy transferred to the sample a novel instrumented drop weight was used. Initial tests with fractions of drill cores and pre-crushed ore particles showed that the simple energy calculation based on potential energy needs to be corrected. For the future work these tests will be extended to other ore types in order to investigate the effects of mineralogy and to include mineral liberation in comminution models suitable for geometallurgy.

Geometallurgy is a team-based multidisciplinary approach aimed at integrating geological, mineralogical and metallurgical information and yielding a spatial quantitative predictive model for production management. Production management includes forecast, control and optimization of the product quality (concentrates and tailings) and metallurgical performance (e.g. recoveries and throughput); and minimization of the environmental impact. Favourable characteristics of an ore body calling for geometallurgical model are high variability, low mineral grades, complex mineralogy and several alternative processing routes or beneficiation methods.Industrial application of geometallurgy is called a geometallurgical program. This study undertook a critical review and evaluation of methods and techniques used in geometallurgical programs. This evaluation aimed at defining how geometallurgical program should be carried out for different kinds of ore bodies. Methods applied here were an industry survey (questionnaire) along with development and use of a synthetic ore body build-up of geometallurgical modules. Survey on geometallurgical programs included fifty two case studies from both industry professionals and comprehensive literature studies. Focus in the survey was on answering why and how geometallurgical programs are built. This resulted in a two-dimensional classification system where geometallurgical program depth of application was presented in six levels. Geometallurgical methods and techniques were summarised accordingly under three approaches: traditional, proxy and mineralogical. Through the classification it was established that due to similar geometallurgical reasoning and methodologies the deposit and process data could be organized in a common way. Thus, a uniform data structure (Papers I, II) was proposed.Traditionally the scientific development in geometallurgy takes place through case studies. This is slow and results are often confidential. Therefore, an alternative way is needed; here a synthetic testing framework for geometallurgy was established and used as such alternative. The synthetic testing framework for geometallurgy consists of synthetic ore body and a mineral processing circuit. The generated digital ore body of a kind is sampled through a synthetic sampling module, followed by chemical and mineralogical analyses, and by geometallurgical and metallurgical testing conducted in a synthetic laboratory. The synthetic testing framework aims at being so realistic that an expert could not identify it from a true one while studying data it offers. Important and unique aspect here is that the geological ore body model is based on minerals. This means that synthetic ore body has full mineralogical composition and properties information at any point of the ore body. This makes it possible to run different characterisation techniques in synthetic analysis laboratory.The first framework built was based on Malmberget iron ore mine (LKAB). Two aspects were studied: sampling density required for a geometallurgical program and difference in the prediction capabilities between different geometallurgical approaches. As a result of applying synthetic testing framework, it was confirmed that metallurgical approach presents clear advantage in product quality prediction for production planning purposes. Another conclusion was that optimising the production based solely on head grade without application of variability in the processing properties gives significantly less reliable forecast and optimisation information for the mining value chain.For the iron ore case study it was concluded that the number of samples required for a geometallurgical program must vary based on the parameters to be forecasted. Reliable recovery model could be established based on some tens of samples whereas the reliable concentrate quality prediction (e.g metal grade, penalty elements) required more than 100 samples. In the latter the mineralogical approach proved to be significantly better in the quality of prediction in comparison to the traditional approach based on elemental grades. Model based on proxy approach could forecast well the response in magnetic separation performance with the help of Davis tube test. But the lack of geometallurgical test for flotation and gravity separation caused that in total the proxy approach forecast capability was worse than in mineralogical approach. This study is a part of a larger research program, PREP (Primary resource efficiency by enhanced prediction), and the results will be applied to on-going industrial case studies.

Coal used in metallurgical processes can participate in reduction reactions to produce metals and alloys from oxides. Base metals production leads to generation of slag, which contains valuable metals that can be recovered and recycled. There are several options to treat the slag, depending on the metal content in the slag. One is slag fuming, which is a well-established process that is traditionally used to vaporize zinc from zinc containing slags, mainly lead blast furnace slag, but is applied in a few plants for copper smelting slags. In this process reduction is achieved using pulverized coal, lump coal or natural gas. Conventionally pulverized coal injected to the furnace is utilized both to participate in reduction reactions and also to supply heat. On the other hand, the amount of possible alternative reducing agents such as residue plastic material is increasing steadily and the issue of sustainable disposal management of these materials has arisen. As carbon and hydrogen are major constituents of the residue plastic-containing materials, they have the potential to be an auxiliary source of reducing agents, to partially replace conventional sources such as coal. Shredder Residue Material (SRM) is a plastic-containing residue material after separation of main metals. Utilization of SRM as an alternative reducing agent, would lead to not only decreased dependency on primary sources such as coal but also to an increase in the efficiency of utilization of secondary sources. This calls for systematic scientific investigations, wherein these secondary sources are compared with primary sources with respect to e.g., devolatilization characteristics, gasification characteristics and reactivity. As a first step, devolatilization characteristics of SRM are compared with those of coal using thermogravimetric analysis. To study the reduction potential of the evolved materials, composition of evolved off-gas was continuously monitored using quadrupole mass spectroscopy. To gain a better understanding of possible interaction of plastics in a mixture, the devolatilization mechanisms and the volatile composition of three common plastics; polyethylene, polyurethane and polyvinylchloride and their mixture have been studied. Furthermore, gasification characteristics and reactivity of char produced from SRM is compared with coal char. The effect of devolatilization heating rate on gasification rate of char was investigated. Proximate analysis has shown that SRM mainly decomposes by release of volatiles, while coal shows high fixed carbon content, which is reported to contribute to reduction reactions. The composition of volatiles shows H2, CO and hydrocarbons which are known to have reduction potential. Therefore, it is essential that SRM be used in a process that could utilize the evolved volatiles for reduction. The results confirm the interaction between the plastics within the binary and ternary mixtures, which suggests that similar phenomenon may occur during devolatilization of SRM. Although the char produced from SRM contains lower amounts of fixed carbon compared to coal char, it has a porous structure and high surface area, which makes it highly reactive during gasification experiments. In addition to physiochemical properties, the catalytic effect of ash content of SRM char contributes to its higher reactivity and lower activation energy value compared to coal char. Moreover, the gasification reactivity of char produced at fast devolatilization heating rate was highest, due to less crystalline structure of produced char.

New technologies are continuously considered for the handling and deposition of tailings. With thickened tailings only a fraction of water used conventionally reaches the disposal area, thus limiting sizes of water holding dam facilities and areas to be rehabilitated. Heat may also be recovered with direct water circulation from the thickening. Technical-economical feasibility considerations are here coupled to the Svappavaara iron ore operations by the Luossavaara-Kiirunavaara AB (LKAB) Company in northern Sweden. Local conditions here favour placement with an elevated thickener location close to the disposal area.The thickened tailings studied here have typically average particle sizes of 20 to 100 µm with maximum sizes of up to about 500 µm. The corresponding slurries cover an intermediate area between homogeneously and heterogeneously flowing mixtures. Various design features are often refereed to in rheological terms, i.e. homogeneous flow. Here, these terms are considered with some caution due to the notable amount of particles larger than about 40 µm.The objective is to form a background for feasibility considerations based on laboratory and pilot-scale experiments in pipelines and flumes together with viscometric measurements and depositional slope observations. Pipeline diameters covered the full-scale range, up to 0.15 m.It was found that a solids concentration by volume of about 48 % characterizes the pipeline pumping, here estimated to be on the conservative side with respect to thickened product quality requirement. The overall pipeline loop results were evaluated in terms of a representative pipe wall stress of about 200 Pa. Approximate values of the Bingham yield stress and plastic viscosity were estimated to 185 Pa and 0.03 Pas, respectively. Depositional slopes have been estimated to about 4 % for slurry properties that gives a conceptually even inclination with no segregation of particles and virtually no drainage of water.With an elevated thickener location close to the disposal area a new high pressure pumping system for about 10 MPa is avoided while the thickener underflow slurry is delivered with a centrifugal pump a short distance during the first years of operation. In order to meet the most effective long-term solution for pipeline distribution and placement, continued use of centrifugal pumps in series installations is considered as the discharge point advances during the 20-year life.The performance of a centrifugal pump is affected when handling highly concentrated slurries which shows up as reductions of the clear water head and efficiency curves. It was found that maximum reductions in head and efficiency were about 10 and 15 %, respectively, when operating at the design flow rate, QBEP, in the best efficiency region. As opposed to Newtonian fluids, a diverging tendency from the water head curve for decreasing flow rates has sometimes been reported below about 0.5QBEP for slurries that show a highly non-Newtonian behaviour. Various mechanisms related to the diverging tendency are discussed. The suction side pressure requirement is normally fulfilled with a large margin for a thickener underflow pump. It was observed in an installation how a standard pump operating at 0.5QBEP under a thickener showed no unstable head tendencies for a tailings slurry with an yield stress of nearly 200 Pa. However, pumps in series installation should be arranged with sufficient margins for an increased suction side pressure requirement.Keywords: Tailings slurry, highly concentrated, non-Newtonian behaviour, centrifugal pump performance, tailings depositional conditions.

In blast furnace ironmaking, efforts are made to decrease the coke consumption mainly by increasing the pulverized coal injection rate. This will cause changes in in-furnace reduction conditions, burden distribution and demands on raw material strength, etc. In order to maintain stable operation and minimize material losses through the off-gas, it is important to understand fines generation and behaviour in the blast furnace. The strength and reactivity of coke at high temperature, measured by the Coke Strength after Reaction (CSR) and Coke Reactivity Index (CRI), have been studied. Mechanisms of disintegration were evaluated using basket samples charged into the LKAB Experimental Blast Furnace (EBF) prior to quenching and dissection. Coke charged into basket samples was analysed with CSR/CRI tests and compared with treated coke from the blast furnace. Results from tumbling tests, chemical analyses of coarse and fine material, as well as Light Optical Microscope (LOM) studies of original and treated coke have been combined and evaluated. The results indicate a correlation between the ash composition and CSR values. Differences in the texture of the coke were determined with LOM, and a change in the coke texture during the CSR/CRI test conditions was found. The results suggest that the main reaction between coke and CO2 during the solution loss reaction took place in isotropic areas, which was especially pronounced in coke with a low CSR. Signs of degradation were apparent throughout the coke pieces that have undergone CSR/CRI testing, but were less observable in coke reacted in the blast furnace. The results indicate that the solution loss reaction was generally limited by the chemical reaction rate in the CSR/CRI test, while in the blast furnace the reaction is limited by the diffusion rate. Coke degradation is therefore mostly restricted to the coke surface in the blast furnace. At a later EBF campaign, off-gas dust and shaft fines were sampled during operation with different iron-bearing materials. EBF process data were used to evaluate the relationship between off-gas dust amounts and furnace conditions. Characterization was focused on fines from coke, iron-bearing materials and slag formers. The graphitization degree (Lc value) of coke taken out of the EBF shaft and coke in flue dust was determined in order to trace the fines generation position. The results showed that flue dust, mainly <0.5 mm, was mechanically formed and created in the same manner for all investigated samples. Carbon-containing particles dominated in the fractions >0.075 mm and consisted mainly of coke particles originating from the shaft. Solution loss in the shaft had a negligible effect on coke degradation and the coke particles which ended up in the flue dust were mainly derived from abrasion at low temperatures. Sludge consisted mainly of chemically formed spherical particles <1μm formed in the blast furnace high-temperature area and then precipitated from the ascending gas as the temperature decreased. The amount of alkali and SiO2 in sludge increased with higher pulverized coal injection rates and flame temperatures, which confirmed that submicron spherical particles in sludge originated from the high-temperature area around the raceway. Theoretical critical particle diameters of materials, which could be blown out with the off-gas, were estimated. Flow conditions in the top of the shaft as well as the properties of fine particles in terms of size and density are important when outflow of mechanical dust, such as flue dust, is concerned. Low off-gas temperatures, and thus lower off-gas velocities, are desirable for blast furnace operation with low amounts of flue dust.

After injection of calcium carbide and magnesium during desulphurisation of hot metal, the slag is normally solid and contains large amounts of iron. Besides the enclosed iron droplets in the slag, drawn-off hot metal during slag skimming also accounts for iron losses during desulphurisation of hot metal. Iron losses during hot metal desulphurisation using both calcium carbide (mono-injection), and calcium carbide and magnesium (co-injection), have been studied by large-scale investigations of slag from the slag pit as well as slag sampling during the desulphurisation process at SSAB EMEA in Luleå.An alkali-containing mineral, nepheline syenite, was mixed together with thecalcium carbide to facilitate slag formation during desulphurisation and to enhance the separation of iron from the slag. Even though the addition of nepheline syenite resulted in a more fine-grained slag, no reduction in the iron content in the slag before slag skimming was observed. The addition of 5 wt-% of nepheline syenite to the calcium carbide during mono-injection decreased the magnetic fraction of the slag from the slag pit from 2.5 wt-% to 1.9 wt-%. The decrease in the larger magnetic fractions combined with the improved iron yield may arise from a more effective slag skimming due to the change in slag consistency.During co-injection with magnesium and calcium carbide, the magnetic fraction of the slag from the slag pit decreased from 3.2 wt-% to 3.1 wt-% with the addition of 10 wt-% of nepheline syenite. Even though nepheline syenite has a positive effect on the slag skimming, due to a decrease in larger magnetic fractions, the total magnetic iron losses have not decreased significantly. A plausible explanation for the lack of improved iron yield during co-injection in the present study could be that the increased turbulence during injection of magnesium counteracts any positive effect from addition of nepheline syenite on the slag properties. Finally, no negative effect on the reagent efficiency was observed during the trials with addition of nepheline syenite.