This thesis presents a robust nonlinear model predictive controller (RNMPC), nominal nonlinear model predictive controller (NMPC) and single-loop proportional-integral-derivative (PID) controllers that are applied to a nonlinear model of a run-of-mine (ROM) ore milling circuit. The model consists of nonlinear modules for the individual process units of the milling circuit (such as the mill, sump and cyclone), which allow arbitrary milling circuit configurations to be modelled easily. This study aims to cast a complex problem of a ROM ore milling circuit into an RNMPC framework without losing the flexibility of the modularised nonlinear model and implement the RNMPC using open-source software modules. The three controllers are compared in a simulations study to determine the performance of the controllers subject to severe disturbances and model parameter variations. The disturbances include changes to the feed ore hardness, changes in the feed ore size distributions and spillage water being added to the sump. The simulations show that the RNMPC and NMPC perform better than the PID controllers with regard to the economic objectives, assuming full-state feedback is available, especially when actuator constraints become active. The execution time of the RNMPC, however, is much too long for real-time implementation and would require further research to improve the efficiency of the implementation. / Thesis (PhD)--University of Pretoria, 2009. / Electrical, Electronic and Computer Engineering / unrestricted

The Rävliden North deposit (Rävliden N) is a volcanogenic massive sulphide (VMS) deposit in the western part of the Skellefte district, northern Sweden. The district is one of Sweden’s major metallogenic provinces with a significant amount of VMS deposits. The Rävliden N deposit, discovered in 2011, contains copper, zinc, lead, silver and subordinate gold and occurs close to the largest VMS deposit in the district, the Kristineberg deposit, which has been mined for more than 70 years. The purpose of this master thesis is to study the composition, mineralogy and paragenetic relationships in different types of sulphide mineralization from the Rävliden N deposit. Emphasis is placed on characterizing the distribution and paragenetic relationships of silver-bearing minerals. The methods include core logging, sampling and mineralogical studies through light optical microscopy (LOM), scanning electron microscopy (SEM) and quantitative evaluation of mineralogy by scanning electron microscopy (QEMSCAN). Lastly, electron microprobe analysis (EMPA) was used to determine the chemical composition of silver-bearing minerals and sulphides. Mineralization types studied include 1: the main massive to semi-massive sulphide mineralization, 2: stratigraphically underlying stringer mineralization and 3: local, vein- and/or fault-hosted silver-rich mineralization in the stratigraphic hanging wall. The massive to semi-massive sulphide mineralization is dominated by sphalerite with lesser galena and pyrrhotite. In contrast, the stringer mineralization is dominated by chalcopyrite and pyrrhotite. The major minerals show evidence of a coeval formation and textural as well as structural evidence suggest that ductile deformation has affected the mineralization types. Notable evidence includes ball-ore textures, accumulation of minerals in pressure shadows and brittle fracturing of competent arsenopyrite and pyrite porphyroblasts and infilling by more incompetent sulphide minerals. The silver-bearing minerals identified are commonly spatially associated with galena and the major species is freibergite ((Ag,Cu,Fe)12(Sb,As)4S13), which also occur as inclusions in chalcopyrite mainly in the stringer mineralization. The stringer mineralization also contains notable amounts of hessite (Ag2Te). Notably, galena, pyrrhotite, freibergite and other sulphosalt minerals are commonly accumulated in pressure shadows near host rock fragments in the massive to semi-massive sulphide mineralization. The only gold-bearing mineral identified in this study is electrum (Au, Ag) in the stringer mineralization. The hanging wall mineralization locally comprises faulted and/or sheared massive sulphide mineralization which is compositionally similar to the main massive to semi-massive sulphide mineralization, besides a significantly higher content of freibergite. However, parts of the hanging wall mineralization are entirely dominated by sulphides and sulphosalts of silver, such as pyrargyrite (Ag3SbS3), pyrostilpnite (Ag3SbS3), argentopyrite (AgFe2S4), sternbergite (AgFe2S3) and stephanite (Ag5SbS4). These occur in structurally late settings, which along with consideration of their temperature stabilities suggest a late origin. Since the silver-bearing minerals in the massive to semi-massive sulphide mineralization and the two varieties of hanging wall mineralization contains the same metals, the mineralization in the hanging wall may have formed by late-stage remobilization of ore components from the underlying Rävliden N deposit. This negates the need for multiple mineralization events to explain the local silver-enriched zones in the hanging wall. The paragenetically late mineralization types contains high content of Ag-bearing minerals in relation to base metal sulphides. This suggests that remobilisation processes were important for locally upgrading the Ag-content.

Lead isotope ratios of ores and igneous rocks in the Central and Southern Andes show a large-scale geographic pattern related to magmatic source processes. This pattern changes in the Northern Andes for reasons that are not well understood; this study is an investigation of potential causes of this change. Deep ocean sediment samples from the Nazca Plate were analyzed for 206Pb/204Pb, 207Pb/204Pb and 208Pb/204Pb, and the data were compared with published data on central Andean ores and ores and igneous rocks from Ecuador. Lead isotopic compositions of the Nazca Plate sediments are quite homogenous and are a close match with Andean ore lead in the coastal arc from central Perú through south-central Chile. However, the lead isotope ratios of the sediment samples are much lower than northern Perú and Ecuador ores. Variations in sediment composition are probably not the source of the northern Andean ore lead isotope pattern.

This study involves the application, interpretation and utilization of regional geochemical data for target generation in the North West Province, South Africa. A regional soil geochemical survey programme has been carried out by the Council of Geoscience South Africa since 1973. A number of 1:250 000 sheet areas have been completed, but there are no interpretative maps which could aid in mineral exploration and other purposes. In order to utilize the valuable and expensive data, the project was motivated through data acquisition and interpretation to generate exploration targets. The study area is confined to Mafikeng, Vryburg, Kuruman and Christiana in the Northwest Province, where potential exploration and mining opportunities exist in areas of great geological interest. These include geological events such as the Bushveld Complex, the Kalahari manganese field and the Kraaipan greenstone belts. The aim of this project was to utilize geochemical data together with geophysical and geological information to verify and identification of possible obscured ore bodies or zones of mineralization, and to generate targets. Another objective was the author to be trained in the techniques of geochemical data processing, interpretation and integration of techniques such as geophysics, in the understanding of the geology and economic geology of the areas. Approximately 5 kg of surface soil was collected per 1 km2 by CGS from foot traversing. Pellets of the samples were prepared and analyzed for TiO2, MnO and Fe2O3, Sc, V, Cr, Ni, Co, Cu, Zn, As, Y, Ba, Nb, Rb, Th, W, Zr, Pb, Sr and U using the simultaneous wavelengthdispersive X-ray fluorescence spectrometer technique at the Council for Geoscience, South Africa. For each element the mean +2 standard deviations were used as a threshold value to separate the negative from the positive anomalies. The integration of geological, geophysical and geochemical information was used to analyze and understand the areas of interest. A number of computer programmes were extensively used for data processing, manipulation, and presentation. These include Golden Software Surfer 8®, Arc-View 3.2a®, TNT-Mips®, JMP 8 ®, and Microsoft Excel®. Through geochemical data processing and interpretation, together with the low resolution aeromagnetic data, gravity data and geological data, seven (7) exploration target areas have been generated: These have been numbered A to G. It is concluded that there is good potential for Cr, PGMs, vanadium, nickel, iron, copper, manganese, uranium and niobium in the targets generated. The results provide some indication and guide for exploration in the target areas. In Target A, Cu, Cr, Fe, Ni and V anomalies from the lower chromitite zone of far western zone of the Bushveld Complex, which has be overlain buy the thick surface sand of the Gordonia Formation. Target B occurs over the diabase, norite, andesitic lava and andalusite muscovite hornfels of the Magaliesberg Formation. This target has the potential for Cu, Fe and Ni mineralization. The felsic rocks of the Kanye Formation and the Gaborone Granite in target C have shown some positive anomalies of niobium, uranium, yttrium and rubidium which give the area potential for Nb, REE and U exploration. Target D is located on the Allanridge Formation, and has significant potential for Ni-Cu mineralization, and is associated with the komatiitic lava at the base of the Allanridge Formation in the Christiana Area. The light green tholeiitic, calc-alkali basalt and andesitic rocks of the Rietgat Formation are characterized by a north-south trending yttrium anomaly with supporting Ba and Y anomalies (Target E). This makes the area a potential target for rare earth elements. Calcrete on the west of the Kuruman has a low b potential target for vanadium. It is believed that the area might be potential for potassium-uranium vanadate minerals, carnotite which is mostly found in calcrete deposits. This study has proved to be a useful and approach in utilizing the valuable geochemical data for exploration and future mining, generated by Council for Geoscience Science. It is recommended that further detailed soil, rock and geochemical surveys and ultimately diamond drilling be carried out in the exploration target areas generated by this study.

The Atlantis II Deep is a location of modern submarine hydrothermal activity along the slowly spreading Red Sea rift axis. Venting of high-temperature hydrothermal fluids, similar to those associated with black smokers, takes place in a brine pool and has led to the accumulation of 90 Mt (dry, salt free) of stratiform, metalliferous sediment. The conditions of mineralization are unique in the modern oceans, but have been widely suggested as a possible analog of some important ancient stratiform base metal ore deposits. This study shows that many of the proposed genetic models for these ancient deposits may be highly simplified and do not take into account rapid diagenetic transformations, widespread non-equilibrium processes, and many other aspects of metal deposition. Sediment cores of the Atlantis II muds were last studied more than 30 years ago. High-resolution sampling and careful re-examination of the mineralogy and geochemistry of the sediments, using modern analytical techniques has significantly improved the understanding of the different processes responsible for the formation of the finely layered metallifeous sediments. The geochemistry of the individual layers is controlled by highly variable detrital, hydrogenic and hydrothermal input. Primary depositional pathways from the brine pool are the main control on basin-wide metal distribution (e.g., increasing Cu/Zn away from the vents) including variable enrichment in trace metals via scavenging from the brine pool and from the enriched pore waters. Cu and Zn have been deposited not only as sulfides but also with poorly crystalline Si-Fe-(oxy)hydroxides. A significant proportion of the original non-sulfide Cu and Zn are diagenetically transformed into sulfides, but also carbonates and clays, in large part reflecting sulfide deficient pore waters. Negative δ34S values, previously unrecognized in the sulfide- and metal-rich units, indicate a source of bacteriogenic sulfide. Syn-diagenetic processes also appear to have been responsible for the sharp laminations in the sediments, as well as distinctive zoning of carbonate and clay minerals around the vent source. The early diagenetic transformations observed in the Atlantis II Deep may not be preserved in the ancient rock record but nevertheless have important implications for metal deposition in some of the world’s largest and richest base metal ore deposits.

The Silver Queen polymetallic vein system south of Houston, B.C., can be treated as a 2-dimensional problem for purposes of reserve/resource estimation. Complexities in obtaining reserve/resource estimates arise from (i) uncertainties in geological interpolation and extrapolation of the vein system, (ii) uncertainties in the distinction between vein and highly altered wallrock in some old drill logs, (iii) complex and multivariable geological character of the vein, (iv) a limited number of exploration drill holes and (v) a different support for drift and drill hole data. Each of these problems has been considered in detail. The study comprises geological analysis, data analysis, point kriging (analysis of thickness and metal distribution) block kriging and comparison of reserve/resource estimation by various procedures including ordinary kriging, inverse squared distance weighting, and polygonal methods. A novel component of the investigation is the use of correlograms (in reality, 1 minus the correlogram) as a substitute for the variogram in geostatistical estimates. This procedure was tested as a means of defining continuity of DDH and Drift assay data of differing support. Ordinary kriging of large polygonal blocks provides metal contents more or less comparable to but locally more conservative than polygonal results reported in a recent feasibility study. Differences are in part due to the use of somewhat different data for the two procedures. The effect of the volume of the selective mining unit on the recovered tonnage and grade is described and limitations of the indirect lognormal method are presented. / Applied Science, Faculty of / Mining Engineering, Keevil Institute of / Graduate

The Dolly Varden camp, Alice Arm area, northwestern British Columbia, is characterized by stratiform and volcanogenic silver-lead-zinc-barite deposits in Early to Middle Jurassic calc-alkaline volcanic rocks of the Hazelton Group. These deposits, containing exceptional silver and significant base metal values, are in andesitic tuffaceous rocks, and occur typically as layers of quartz, carbonate, barite and jasper, with lesser amounts of pyrite, sphalerite and galena, and sparse chalcopyrite. Production from three deposits, the Dolly Varden, Northstar and Torbrit mines, totaled 1,284,902 tonnes of ore that averaged 484g silver per tonne, 0.38 percent lead and 0.02 percent zinc. The Hazelton Group is a thick, widespread assemblage of basaltic to rhyolitic volcanic flow rocks, their tuffaceous equivalents, and derived sedimentary rocks. Dolly Varden camp is underlain by more than 3,000m of Hazelton Group rocks comprised of one major volcanic and one major sedimentary formation. Volcanic rocks underlie sedimentary rocks and have been subdivided into footwall and hangingwall units based on stratigraphic position relative to the mineralized stratiform horizon. Footwall volcanic rocks consist of green ± maroon basaltic-andesite tuff, green ± maroon porphyritic andesite and green andesite shard tuff. Stratiform mineralization rests conformably upon the underlying green andesite shard tuff. Hangingwall volcanic rocks above the stratiform layer consist of pale grey basaltic-andesite ash tuff, maroon basaltic-andesite ash-lapilli tuff, grey-green porphyritic andesite, and pale green andesite ash tuff. Hangingwall volcanics are unconformably capped by sedimentary rocks consisting of maroon siltstone, calcareous and fossiliferous wacke, and black siltstone and shale; black siltstone and shale form the youngest rock unit of the Hazelton Group in the Dolly Varden area. Basalt and lamprophyre dykes intrude all rocks of the Hazelton Group. The rocks of the Hazelton group exposed in the Dolly Varden camp are folded into a series of anticlines and synclines with gentle, northwestern plunges. Two major sets of nearly vertical block faults cut all rock units; earlier faults trend northwest and younger faults trend north-northeast. Geological mapping, combined with petrologic, petrographic and isotopic data, indicate that the stratiform deposits probably formed as submarine exhalative deposits associated with andesitic volcanism of the Hazelton Group during the Early to Middle Jurassic. Evidence for a volcanogenic origin is the conformity of layered mineralization with stratigraphy, lateral and vertical mineral zonation patterns, consistent hangingwall versus footwall contact relationships, fragments of stratiform ore within tuffaceous volcanic rocks of the hangingwall, consistent differences in the stable isotopic compositions between the sulfides versus barite, quartz and carbonate gangue, and the Jurassic "fingerprint" for the lead-bearing deposits of the Dolly Varden camp. The Dolly Varden deposits display criteria for classification of a new, previously unrecognized, stratiform and volcanogenic, deposit type, named here, the "Dolly Varden type", and is characterized by silver-rich, low sulfide and high oxide stratiform mineralization within andesitic volcanic rocks. / Science, Faculty of / Earth, Ocean and Atmospheric Sciences, Department of / Graduate

Ph.D. (Geology) / Shear zone controlled hydrothermal alteration zones in the northern Kaapvaal craton (NKC) are developed in host rocks of vastly different chemical composition and metamorphic grade. Some carry appreciable Au and base metals and some are barren. Alteration zones in three different distinctive crustal zones were examined in detail to determine the controls of these two types of alteration. 1. The Matok Complex is situated in the southern marginal zone (SMZ) of the Limpopo Belt (LB), close to the zone of rehydration. Two major stages of hydrothermal alteration could be identified in local shear zones, a pervasive propylitization and a subsequent vein controlled quartzalbite alteration. The two-stage alteration occurred sometimes between the emplacement of the Matok Complex (2670 Ma) and the intrusion of unaltered mafic dykes (1900 Ma). Calculated isotopic compositions of the hydrothermal fluids indicate that magmatic ± meteoric waters as well as juvenile C02 were responsible for the establishment of the alteration zones. The fluids most probably were late magmatic fluids associated with the Matok magmatism. The propylitic alteration was accompanied by introduction of small amounts of CU + Au and represents an alteration type identical to that developed in porphyry copper deposits. The subsequent quartz-albite alteration was caused by extremely saline fluids which depleted the rocks of all the major and trace elements with exception of Si, Al, Na and Zr. 2. This chemical alteration pattern' contrasts with those developed in two alteration zones associated with economic gold mineralization in greenstone belts of the NKC (Sutherland and Pietersburg belts). At the Birthday and Eersteling gold mines, a biotite-calcite-quartz alteration is developed. The chemical pattern of the alteration is...

The Nkout (Cameroon) and Putu (Liberia) oxide facies iron ore deposits comprise fresh magnetite banded iron formation (BIF) at depth, which weathers towards the surface, forming high grade martite–goethite ores. This study aimed to improve the mineralogical understanding of these deposits in order to predict their metallurgical responses. It concentrated on developing the QEMSCAN® technique and testing its application to these ore types, but also used a variety of other analysis methods. The QEMSCAN® species identification protocol was developed to include three goethite entries: goethite/limonite, phosphorus-bearing and aluminium-bearing goethite. QEMSCAN® was also used to distinguish between the iron oxides using their backscattered electron signals. To test the correlation between the mineralogy and metallurgical characteristics, magnetic separations were carried out. The samples were divided into 4 main groups based on their whole rock Fe content, determined by XRF analysis, and their degree of weathering: enriched material, weathered magnetite itabirite, transitional magnetite itabirite and magnetite itabirite. Quartz and Al oxide and hydroxide minerals such as gibbsite are the major gangue minerals in the magnetite BIF and martite–goethite ores respectively. From the QEMSCAN® analysis it was concluded that the iron oxides are closely associated and liberation of them individually is poor. Liberation increases when they are grouped together as iron oxide. Chamosite concentrations > 6 wt. % significantly lower liberation of the iron oxides. From the metallurgical testing, it was concluded that iron oxide modal mineralogy gives an indication of iron recovery but other QEMSCAN® data such as mineral association and liberation could be important especially if the iron oxide minerals are not liberated. Grain size and instrument characteristics also affect recovery of iron minerals. There is no evidence to show that there is any structural control on the BIF mineralisation at Nkout because metamorphism has significantly affected the lithological characteristics. The BIF mineralised zones occur as stacks with no particular stratigraphic relationship. Alteration and stratigraphy are the main controls on the martite–goethite ores. These results are applicable to most other BIFs so that as direct shipping ores are exhausted, the approach used here can help to develop the lower grade portions of the deposits.

Extensive work is reported in literature on the reduction of iron oxides with carbonaceous reductants. Most of this work considered isothermal reaction of the material mixture, although as shown in some studies, isothermal reaction conditions are not often the norm because of sample size and heating arrangement in the experiment. In industrial processes, such as the rotary hearth type processes and the IFCON® process for iron ore reduction, the norm is non-isothermal reaction. Simulation of industrial processes should take non-isothermal reaction into account if the heat transfer effects within the process are to be investigated. To avoid the complications of coal volatiles in the experimental set-up, few studies were done with coal as reductant. The primary aim of the work presented here is to quantify radiation heat transfer to the surface of an iron ore and coal mixture heated uni-directionally from the sample surface to show the importance of heat transfer in the IFCON® process. Secondary aim of this work are to show the effects of layer thickness, coal volatiles, phase chemistry and particle size in this reaction system. The experimental set-up consists of a tube furnace modified to transport the sample into and out of the experimental tube furnace heating zone under a protected atmosphere, whilst the product gas is analysed throughout the experiment by quadropole mass spectrometer. The sample surface temperature, heating zone temperatures and material bed temperatures were measured throughout the experiment. A sample cutter-splitter was developed to divide the reacted sample into three horizontal segments for chemical analyses. The sample surface temperature and the heating zone temperatures were used as inputs to a radiation network calculation to quantify radiation heat transferred to the sample surface. The radiation network calculation was calibrated against heat-mass balance calculations for pre-reduced ore and graphite samples reacted at furnace temperatures of 1300, 1400 and 1500°C. The results show that radiative and conduction heat transfer control prevails for 16 mm to 40 mm material layers heated uni-directionally from the material layer surface. It is shown that coal volatiles contribute to reduction in the stagnant material layer. Also, smaller particle sizes result in increased reaction rates because of a decrease in the diffusion limited effects which were seen in reaction of the base size of coal and ore particles. / Thesis (PhD (Metallurgical Engineering))--University of Pretoria, 2009. / Materials Science and Metallurgical Engineering / unrestricted