Search Documents

By Da-Qiang Cang, Qingde Wang, Wei Wang, Wen-Bin Dai, Min Guo, Yu Li

Utilization of both slag and its heat for hot slag is a great challenge for iron and steel industry. Many methods of hot slag modification for directly converting hot slag into value-added materials were researched in recent years, and waste heat of hot slag could thereby be transformed into inner energy or chemical energy of the materials. In this paper, an experiment of converting hot slag directly into glass ceramics by hot slag modification method was conducted. A cooled modifiers mainly composed of fly ash were firstly heated though a cyclone furnace and then fell into a molten bath to mix with hot steel slag. Half of hot slag was used as raw materials and accordingly about half energy consumption, about 5MJ/kg for melting raw materials was saved in the process compared with that of traditional producing method. Frit quenched from the modified hot slag had a higher basicity of 0.6-0.7 than traditional frits in order to reduce the needed amount of cold modifiers and accordingly utilize more hot slag and its waste heat. Glass ceramics were further prepared from the frit and a heat treatment of one-stage method is more suiter to such frit than that of two-stage method. The glass-ceramics with basicity of 0.6 and heat-treated by one stage method had the best performance with bending strength of 146.46 MPa, Vickers hardness of 6.79 GPa, water absorption rate of below 2% and shrinkage rate of around 10%.

Jan 1, 2015

By Kamilla Giló Santiago

Nesse trabalho para caracterização metalúrgica dos revestimentos compostos pela mistura das ligas AWS ERNiCrMo - 3 E AWS ERNiCrMo - 4 soldados pelo processo MIG/MAG duplo arame foram realizadas análises micrográficas por meio de microscopia eletrônica de varredura. Observou-se semelhança entre os precipitados do topo e da interface do metal de solda com o substrato e que a microestrutura obtida com a soldagem das duas ligas, simultaneamente, resultou na mistura das microestruturas, obtidas a partir das soldagens com cada uma das ligas de forma isolada. O ensaio para avaliação do desempenho da mistura quanto à resistência a corrosão foi realizado conforme a norma ASTM - G 48, método C para determinação da temperatura crítica de pite da mistura das ligas. Foram utilizadas amostras dos revestimentos com maior e menor diluição. O resultado do ensaio apontou que os revestimentos soldados apresentaram temperatura crítica de pite superior ao da liga AWS ERNiCrMo-3 e semelhante ao da liga AWS ERNiCrMo-4.

Aug 7, 2018

By Dennis Dolinar

"Tailgate entries in U.S. longwalls can be subjected to high loads and deformations. Therefore, tailgate entries are generally supported with varying types and amounts of standing support. However, even with standing support, roof falls and blockages can occur in the tailgate when the support system is not properly designed. Safe and efficient operation of a longwall face requires the selection of the appropriate type and amount of standing support. Because of the importance of maintaining a safe and functional tailgate entry, the National Institute for Occupational Safety and Health (NIOSH) is currently developing a standing support design methodology based on the ground reaction curve. Using the ground reaction curve for design requires that the ground and support interaction be quantified. This can be accomplished by the installation and analysis of data from instrumented field sites. Although necessary, the field results cannot develop the complete ground reaction curve. Numerical models, calibrated to the field results, are used to establish the complete ground reaction curve. In the present study, data was obtained and evaluated to establish points for calibration of the ground reaction curve for two western longwall operations.One test site, Mine A, was located at a depth of 1,550 ft. At this site, the floor consisted of siltstone, while the roof was an interbedded siltstone and sandstone. Because of the depth, a double row of22-in diameter CANs1 were installed to support the tailgate. Both the first and second longwall panels have mined past the instrumented site.The second site in Mine B is at a depth of only 760 ft. The floor at this mine consists of a weak shale/mudstone and the roof consists of thick, top coal overlain by weak shales. At this site, a double row of pumpable cementitious cribs was installed. The second test site has only been subjected to side abutment loading from first panel longwall mining.The measurements indicate that the support at both mines was heavily loaded when subjected to the side abutment loading from first panel mining. At the deeper site, the support began to yield when the face was more than 300 ft outby during second panel mining. In comparison to results from previous test sites in the Pittsburgh Seam and Illinois Basin, the amount of support loading and roof-to-floor convergence was an order of magnitude higher at the western sites. This has resulted in a significant difference in the position and shape of the ground reaction curves."

Jan 1, 2010

By R. K. Ham, M. L. Wayman

TD-Nickel has been broken in tension and in fatigue at voom temperature. Rod specimens failed in tension by necking, with axial cracks attributed to voids elongated in the extrusion direction. Fatigue specimens failed in shear. Thin-film electron microscopy showed that the subgrain structure of TD-Nickel was very stable, and that particle-matrix detachment was very difficult, in tension and fatigue. TD-Nickel softened slightly during fatigue but had a high fatigue ratio (0.5at 10' cycles). Fractography suggested that Stage I fatigue crack propagation is greatly extended in TD-Nickel. Td-NICKEL* is a dispersion of 2 pct by volume of thoria (ThO2) in nickel, produced by powder-metallurgical methods including compaction, sintering, and extrusion. The spherical thoria particles, which may have a mean diameter of a few hundred angstroms, are dispersed with a mean planar spacing of a few thousand angstroms. By a combination of cold working and annealing, a subgrain or cell structure of dislocations intimately associated with the dispersion can be produced.' This gives the material useful tensile properties which are extremely resistant to exposure to high temperatures.'-= At the same time, the material shows considerable ductility.3'4 At the beginning of the present work, it was considered of great interest to investigate the mechanical stability of TD-Nickel under conditions of fatigue. On the one hand, materials which derive strength from a cold-worked structure are unstable in that they are susceptible to fatigue softening: in addition, the presence of discontinuities such as the particle-matrix interfaces might be expected to assist in the initiation and propagation of fatigue cracks, in that they may provide local concentrations of internal stress and sites for the initiation of voids." On the other hand, the substantial ductility of TD-Nickel suggests that if it obeys Coffin's relation8 (or the more refined form proposed by Manson9) it should have good fatigue resistance. The initial purpose of this investigation was to assess the importance of these factors. Also, relatively little fundamental work has been done on the mechanism of fatigue in dispersion-strengthened materials. Work on overaged A1-4 pct CU10-12 revealed a very large Bauschinger effect indicative of internal stresses at the particles, and very great fatigue hardening1' presumably due to the multiple slip stimulated by these internal stresses, followed by softening which was initially due to softening in the matrix, but later might have arisen from cracking at the particle-matrix interfaces. The study of overaged Al-4 pct Cu could not settle this latter point by the electron microscopy of thin films, since the dispersion is too coarse. A previous study of the fatigue of internally oxidized copper13 was complicated by inter granular failure, attributed to oxide particles at grain bouhdaries; internal oxidation of single crystals, however, improved their fatigue properties.'3 Investigations of SAP (sintered aluminum powder) are complicated by the complex particle shapes, and the possibility of continuous internal oxide films.14 It was hoped to avoid these difficulties with TD-Nickel, which had the further advantages that it was commercially available and suitable for study by the thin-film technique. 1) MATERIALS AND EXPERIMENTAL PROCEDURE TD-Nickel was purchased from the Driver-Harris Co. as 3/9-in.-diam rod which had been extruded, swaged, and "stress-relieved" (normally at 1010°C for 1 hr). Continuous-radius fatigue specimens with a minimum diameter of -0.1 in. were ground to shape and electropolished in 40 pct phosphoric, 35 pct sul-furic, 25 pct water at 25oC, with a stainless-steel cathode, at -6 v. Fatigue tests were carried out at room temperature with a Sonntag SF-1-U machine operating in push-pull at 1800 cycles per min, and in an Instron TT-C-L modified for reversed stressing at 26 cycles per min as described elsewhere.10 Round tensile specimens were ground with 0.85-in. gage lengths of 0.135 in, diam, electropolished, and tested in a Tinius Olsen hydraulic machine. The surfaces of fatigue specimens were examined with a Reichert metallograph. Discs for electron microscopy with a Siemens Elmiskop I were spark-cut 0.01 in. thick with a modified Servomet 11 employing a tool of moving molybdenum wire, cleaned in 50 pct acetic, 30 pct nitric, 10 pct phosphoric, and 10pctsul-furic at 85oC, and thinned by the window method in the electropolishing solution at -20°C and -6 v. Two-stage replicas (parlodion, then Pt-50 pct C self-shadowed at 40 deg) were taken from fatigue fracture surfaces for electron microscopy. Plane sections parallel to the specimen axis and containing the direction of fatigue crack propagation were polished, etched with Carapella's reagent, and examined optically. Transmission Laue X-ray photographs of as-received material electropolished to a point were used to determine preferred orientation. 2) RESULTS 2.1) The Structure of As-Received Material. Grains -1 µ diam and elongated 20 to 30 u in the direction of the rod axis were observed, With' elongated "intergran-ular" voids (-4 p diam and -90 µ long and identified as such using longitudinal and transverse sections and

Jan 1, 1968

By D. W. H. Su

This paper presents the results of ill situ horizontal stress measurements using a Minifrac system in several northern Appalachian coal mines. The effect of stress magnitude and orientation on longwall gate entry stability was analyzed using a series of three-dimensional finite element analyses. The best headgate stability is achieved if the maximum horizontal stress is aligned with the direction of longwall retreat or if the panel is oriented such that the maximum stress relaxes over the headgate due to the presence of the newly formed gob. The degree of horizontal stress damage to longwall gate entries depends largely on the stress magnitude and roof geology. Accurate assessments of support or design techniques which may mitigate longwall headgate or tailgate instability in areas of thinly laminated roof and high horizontal stresses must account for the in situ horizontal stress state and site specific roof geology. The ability to measure stresses and map geology is thus essential to prevent ground control failures at the longwall headgate.

Jan 1, 1995

By Leon D. Keller

Fine screening, as discussed in this paper, pertains to the use of the Rapifine DSM Screen to make a 65 to 325-mesh separation at production tonnages. The greatest number of Rapifine Screen installations are in magnetic taconite plants where the unusual relationship between the grinding mill, cyclone classifier, magnetic separators, and fine screens makes the Rapifine Screen application unique. Since the initial fine screening installation on magnetite in 1961, (1) additional installations and further testwork have contributed to the fund of information which is the basis of this paper. The addition of fine screens to the grinding-classifying circuit of a magnetic taconite plant creates a change that effects all components of the circuit. Pine screens in a non-magnetic plant have also been investigated, but here the benefits have not' been as clearly defined. Screens classify according to the particle size. Hydrocyclones classify by a combination of particle size and particle weight. The change in the type of classification minimizes overgrinding of the mineral. The resultant savings in grinding power can be used to increase the throughput, upgrade the product, or increase recovery.

Jan 1, 1971

By H. Lawrence

In 1977, the Bureau of Mines awarded a grant to the State of South Carolina to explore the possibility of using Landsat multispectral scanner (MSS) data as an aid in monitoring surface mining activities in selected counties of the State. Work had been done in Aiken County prior to the grant by Bureau of Mines and State personnel at the EROS Data Center using both the GE Image 100 and ESL IDIMS systems. The initial work on the 1-100 consisted of locating and measuring the areas of various kaolin mines. The results were excellent on location, identification, and accurate area measurements of several mines as compared with aerial photos of each mine. The effort also included locating and measuring the size of all mines (including sand and gravel) within two U.S. Geological Survey (USGS) 7 ½ min quadrangle maps. Line printer overlays using the IDIMS equipment and software resulted in the location of all known mines within an estimated error less than ±91.4m (5300 feet). Mine acreage was calculated by counting each symbol (which represented one pixel) and multiplying by 1.1 (the number of acres per pixel). Figure 1 shows a portion of one quadrangle overlay. Table I gives a comparison of acreage measured for sever- al mines from the 1-100 and IDIMS systems vs. aerial photo measurements.

Jan 1, 1979

By A. L. Paverd

The petrology of a number of pyrometasomatic mineral deposits in north Queensland, Australia, were studied to evaluate the use of mineralogy and major element chemistry in the exploration for copper, lead and zinc-bearing skarns. Results indicate a direct correlation between the character of the mineralization and the composition and nature of the host rock. The study of garnet compositions proved the most useful. Economic mineralization was limited to limestone-derived skarns containing garnet of And60-90, whereas skarns containing garnet of less than contained less than 1And55 contained less than 1% total sulfide. Skarns containing garnet of And60-8S host the best mineralization; above And8S' copper, lead and zinc sulfides decrease and magnetite and pyrrhotite become the dominant ore minerals. Similar correlations were found for pyroxene, amphibole and 'the bulk composition of the skarns.

Jan 1, 1981

By R. L. Grant

The Bureau of Mines developed two new gap-sensitivity methods for explosives, and especially permissible explosives, to overcome certain disadvantages of the present routine half-cartridge method. The revised methods are based on the statistical up-and-down method and give more useful measurements of the 50-percent' gap values than the nonstatistical routine method. The new methods use full cartridges and incorporate certain physical improvements such as the use of cardboard tubes for containing the explosives and coal dust gaps to simulate actual mining conditions. A total of 60 samples of permissible explosives were evaluated simultaneously with the 2 improved methods and with the half-cartridge method

Jan 1, 1967

By O. Tsuge, G. Hoffman

ITmk3®(Iron Technology Mark III) is an innovative new ironmaking technology whereby high quality iron nuggets are produced directly from iron ore fines and coal in a rotary hearthfurnace1). The reaction mechanism has been investigated in detail through fundamental laboratory research2)3), and the ITmk3 process has progressed from the laboratory to commercialization by the success of the pilot plant tests initiated at Midrex’ Technical Center and at the Kakogawa Bench Plant (KBP) located at Kobe Steel Ltd. Kakogawa Steel Works4)5). Pilot operations have demonstrated rapid production of the iron nuggets with complete metal/slag separation at a relatively low operating temperature. The Kakogawa Bench Plant (KBP) in Japan has logged 21 days continuous operation during which high quality iron nuggets were produced directly from fine ore and coal. The operation technology and the engineering data obtained from these pilot tests were instrumental for the scale-up of the ITmk3 process for the Mesabi Nugget Corp. pilot demonstration plant, which is located at Cleveland Cliffs Inc. Northshore Mining iron ore pellet plant in Silver Bay, MN. The ITmk3 process is an ideally suited technology for siting at the iron ore mine, as it allows the economical production of a value-added merchant iron product.

Jan 1, 2003

By Sean Harvey

One of the manifestations of the dynamically induced failures in deep coal mines is the violent ejection of coal from the ribs. A possible protection against such failures is the use of a specifically designed dynamic support system with reinforcement units having the sufficient capacity to absorb the kinetic energy of the dynamic failure and a skin support that can adequately contain the disintegrated coal. With this method of mitigation in mind, a recently introduced yielding rock bolt, trade name Roofex, has been identified (Harvey and Ozbay, 2009) as a possible energy absorbing component of a dynamic support system. According to the manufacturer?s specification, the Roofex bolts can yield 300 mm (or more) displacement while holding 80 kN pull load. In order to assess the performance of this support unit in coal ribs, a series of in situ pull tests performed in two coal mines in Colorado. The results show that, under conventional pull test loading conditions, Roofex yielding bolts perform very close to their claimed specifications, thus showing a potential for their use as the main energy absorbing component of a dynamic support system. Further studies are required to confirm the in situ performance of these bolts under dynamic loading conditions and also to develop a containment component for a dynamic support system for the mitigation of dynamic coal rib failures.

Jan 1, 2009

By R. S. James, S. P. Goff, A. E. Armitage

"Abstract - Oxide-facies banded iron formation in the Third Portage Lake area is hosted by ultramafic and calc-alkaline volcanic rocks and related sedimentary rocks of the Archean (2.8 Ga) Woodburn Lake Group, 80 km north of Baker Lake. These rocks record at least two phases of metamorphism/ metasomatism. Greenschist facies metamorphism, in the temperature range of 370°C to 520°C and at pressures above 2.3 kbar, has affected all rocks on a regional scale. Later mineralization and alteration of banded iron formation (BIF) and associated volcanic rocks resulted from metasomatic fluids which carried Mg, K, Ca, S, As, Cu and Au. These fluids were confined to ductile to brittle, cross-cutting and layer-parallel micro-shears developed in the thickened hinge region of a shallow plunging recumbent fold. Metasomatic reactions yielded cummingtonite + biotite-bearing assemblages which are associated with strongly mineralized iron formation. Pyrrhotite is the dominant sulfide phase and occurs within ductily deformed iron formation where it pervasively replaces magnetite. Minor fine-grained chalcopyrite and arsenopyrite generally occur as inclusions within pyrrhotite and recrystallized magnetite, and rarely as individual grains. Pyrite is restricted to brittlely deformed iron formation, and to the contact between iron formation and altered ultramafic rocks, where it is associated with quartz veins. Pyrite typically replaces pyrrhotite. Native gold is associated with both pyrrhotite and pyrite, but higher gold concentrations are associated with pyrite. Grunerite/Ca-amphibole + stilpnomelane and sparse pyrrhotite and pyrite characterize weakly mineralized iron formation. Within adjacent ultramafic rocks, metasomatic reactions yielded assemblages of interlayered Ca-amphibole + biotite, and talc + Ca-amphibole + carbonate + biotite. K-Ar geochronology on metasomatic biotite from mineralized BIF yielded a 1791 ± 32 Ma age.An epigenetic model is favored for the emplacement of gold. The age of the gold mineralization in the Third Portage Lake area is uncertain, but may be Proterozoic based on the preliminary isotopic age and by analogy with Proterozoic gold deposits elsewhere in this part of the Churchill Structural Province."

Jan 1, 1996

By A. L. Field

This general subject has been so thoroughly discussed by Dr. Herty and his associates, especially with regard to the degree of oxidation of the metal at the time that the silicon pig or spiegel is added, that I am not going to spend much time on that today. I should like to propound a question, which is based on my own observations made some years ago while with a large alloy-steel company which had two open-hearth furnace plants. It is true that one of these plants was not engaged particularly in making alloy steels, but there were certain relations between slag and metal composition which appeared to me to be entirely different in the two plants. I have an explanation for it that is more or less incomplete, but possibly Dr. Herty or Mr. Reinartz or some of the others may be able to complete it. Suppose we take our three layers; the metal on the bottom, on top of that the slag, and on top of the slag, the gas. We cannot overlook the gaseous phase. Regardless of the effect of the iron oxide in the slag on the rate of carbon elimination—and that, of course, is rather important in alloy practice since it is not desirable to have the carbon dropping too rapidly at the time the deoxidizing is had in the furnace—it will probably be admitted that it depends on the iron oxide in the metal for a given carbon oxide. I have seen cases where with approximately identical metal and slag compositions one furnace would work twice as rapidly as another in removing carbon. Of course, there is some explanation of this in what others have already said, that the question of the exact lime and silica content of the slag determines the activity and viscosity of the slag; but I am convinced, without having reviewed the actual records since I was asked to make this talk, because they were not available at the time, that the slags in these two furnaces were not materially different. Putting it another way, one furnace would have

Jan 1, 1935

By A. L. Field

This general subject has been so thoroughly discussed by Dr. Herty and his associates, especially with regard to the degree of oxidation of the metal at the time that the silicon pig or spiegel is added, that I am not going to spend much time on that today. I should like to propound a question, which is based on my own observations made some years ago while with a large alloy-steel company which had two open-hearth furnace plants. It is true that one of these plants was not engaged particularly in making alloy steels, but there were certain relations between slag and metal composition which appeared to me to be entirely different in the two plants. I have an explanation for it that is more or less incomplete, but possibly Dr. Herty or Mr. Reinartz or some of the others may be able to complete it. Suppose we take our three layers; the metal on the bottom, on top of that the slag, and on top of the slag, the gas. We cannot overlook the gaseous phase. Regardless of the effect of the iron oxide in the slag on the rate of carbon elimination—and that, of course, is rather important in alloy practice since it is not desirable to have the carbon dropping too rapidly at the time the deoxidizing is had in the furnace—it will probably be admitted that it depends on the iron oxide in the metal for a given carbon oxide. I have seen cases where with approximately identical metal and slag compositions one furnace would work twice as rapidly as another in removing carbon. Of course, there is some explanation of this in what others have already said, that the question of the exact lime and silica content of the slag determines the activity and viscosity of the slag; but I am convinced, without having reviewed the actual records since I was asked to make this talk, because they were not available at the time, that the slags in these two furnaces were not materially different. Putting it another way, one furnace would have

Jan 1, 1935

By Henry Brañes

El presente trabajo describe el proceso de optimización metalúrgica de la planta concentradora “La Inmaculada” de la U.P. “El Cofre en Puno, dado que la mina no contaba con reserva suficiente para abastecer a la planta concentradora. Para ello, se ha dejado de realizar el blending de los minerales oxidados, mixtos y sulfurados y ahora se efectúan los tratamientos metalúrgicos en forma independiente para cada tipo de mineral (sulfurado, mixto y oxidado). Así también, se mejoró la calidad del producto final de molienda del orden y se ha modificado el reporte del balance metalúrgico diario de la planta concentradora, efectuando los balances por cada turno de operación, con indicación de controles de molienda, suma de recuperaciones y recuperación económica; representando este último es un índice indispensable y primordial que nos indica la eficiencia final económica con el que estamos operando.

Sep 14, 2009

By J. P. McGee

A 1,000-hour test with a gas turbine at 1,200° F showed a working fluid of 1 gr of ash per 100 scf of power gas (79 percent of the ash 7 to 35 microns in diameter) to be excessive for acceptable blade wear for commercial practicability. Blades in the turbine consisted of about 20 percent each of chromium, cobalt, and nickel; 3 percent molybdenum; 2-1/2 percent tungsten; and smaller amounts of manganese, columbium, and tantalum. Erosion of both the fronts and backs of the turbine nozzles was particularly heavy, the trailing edges being worn to a featheredge. Erosion of the buckets was less marked, although the leading edges were somewhat pitted. A refractory-lined combustor performed efficiently and with little abrasion of the lining by the ash-laden gas.

Jan 1, 1969

By Kot Unrug

For the purposes of underground mine design, twelve cores of the Springfield (V) Coal (Pennsylvanian) roof and floor were retrieved. Each core was photographed, described fresh in the barrel, and the rock quality designation (RQD) was measured. Cores were wrapped in plastic and delivered to the rock mechanics lab. At the lab RQD was remeasured and new photographs were taken before running geotechnical tests including weatherability. Significant differences in barrel RQD's vs. lab RQD's have been noticed in spite of careful handling. The drill site RQD is generally used to calculate geotechnical results, but in some types of coal measure rocks, further fragmentation of core takes place with slight changes of moisture content, which occur even in well wrapped core. This creates a dilemma which RQD should be used for support design. Rocks around excavations are exposed to the significant changes in moisture content due to the seasonal weather changes and related fluctuation of humidity of air ventilating a mine. Therefore, it appears to be more realistic using laboratory RQD which corresponds closer to the actual rock condition around the excavation. Weatherability test, developed for characterization of time dependent behavior of rock exposed to fluctuations in moisture content, is thought to be related to the above-mentioned changes of RQD. In weak shales, mudstones and claystones which weather rapidly, it is essential to characterize their future behavior in the pre-mining stage. This can allow for a realistic design of primary, and particularly secondary support in coal mines, especially in coal fields where such geologic conditions exist.

Jan 1, 2003

By Giulliane Meireles Casali Gonçalves Torres

Os arrebentamentos que ocorrem durante o processo de trefilação geram impacto na produtividade e rendimento operacional em função do retrabalho gerado na realização de solda cada vez que o material se rompe e consequentemente no tempo perdido para execução desta etapa. A metalografia é a ferramenta mais eficiente para caracterização das causas dos arrebentamentos durante a trefilação. Faz-se necessário a utilização de microscopia e aplicação dos reativos específicos destinados a revelar a origem da anomalia. O presente trabalho objetiva mostrar os principais tipos de fraturas na trefilação e suas respectivas origens, através do estudo de quatro casos distintos, são eles: fratura resultante de quebras internas (rupturas centrais), fratura dútil por tração, fraturas resultantes de quebras na solda e fraturas lascadas causadas por formação de “pé de corvo”. Os resultados das análises mostram de forma objetiva se a anomalia foi gerada a partir da matéria prima (defeitos superficiais no fio máquina) ou problemas de processo/operacionais durante a trefilação.

Aug 17, 2017

INDEX OF CONTENTSBlast Furnace SlagsForehearth AccretionsConverter SlagsConclusionsIllustrationsCOMPOSITION OF SOME COPPER SLAGS FROM PORT KEMBLASamples of copper slags from the Port Kembla Smelter of the Electrolytic Refining and Smelting Co. of Australia, Pty. Ltd., have been submitted for mineragraphic examination, with reference to the mode of occurrence of copper in them. The samples included (I) blast furnace slags, (H) a forehearth accretion, and (IH) converter slags. 1. BLAST FURNACE SLAGS Four series of samples of blast furnace slag have been submitted. The. chemical compositions of the several slags are shown in Table 1. Analyses Nos. 1 and 2 are of samples of slowly or normally cooled slag, taken from the slag dump, Analyses Nos. 3 and 4 are of samples comprising slowly or normally cooled slag, together with granulated slag from the same melt.

Jan 1, 1949

By A. Jahedsaravani, M. Massinaei, M. Zarie

Deep learning is a subset of machine learning that uses artificial neural networks for extracting high-level features from image data. In the present study, a soft sensor is proposed for the prediction of the flotation performance through froth features generated by the use of pre-trained convolutional neural networks. Several state-of-the-art convolutional neural networks (AlexNet, GoogLeNet, VGGNet, ResNet, and SqueezeNet) pre-trained on the ImageNet database are used to predict the metallurgical performance of two flotation systems. The first case study is a batch copper flotation system video-captured over a wide range of process conditions. The second case study is an industrial coal flotation column equipped with a continuous video recording system. The pre-trained networks are used to extract features from the froth images, and these features are subsequently used to predict the flotation conditions and performance. The prediction results by the pre-trained algorithms were compared with the traditional image processing algorithms. This demonstrates the ability of the pre-trained structures to generalize to images outside the ImageNet database. GoogLeNet outperforms other network architectures and provides more accurate predictions of the flotation process behavior and performance.

May 5, 2023