Determinação de curvas de lavabilidade utilizando análise de imagens digitais - Estudo de caso: minério de ferro

• Ano de defesa: 2016
• Autor(a) principal: Josue Mesquita de Souza Junior
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Curvas de Lavabilidade; Análise de Imagens Digitais; Separação Gravítica; Beneficiamento mineral; Engenharia metalúrgica
• Link de acesso: http://hdl.handle.net/1843/BUBD-AEAH62

The gravity concentration corresponds to one of the oldest and most important techniques of mineral processing and it is used to process a large variety of ores. In this context, the best way of investigating and predicting results of gravitational processes are called washability curves. These curves show the distribution of particles of a given material according to their densities and they are constructed from experimental data obtained under ideal conditions in so-called float-sink tests. For peforming these tests dense liquids with different densities are used, which correspond to very toxic, volatile, unwieldy and, in most cases, very expensive substances. Furthermore, because there are no dense liquids with very high densities, they can not be used in fractionation of sulfides or metal oxides, which restricts the use of washability curves to lightweight materials, such as coal. With this in mind, it proceeded to the development of this work, whose goal was to present a new methodology for the construction of the washability curves from digital image analysis. With this method it will be possible to build washability curves more quickly and safely, making possible the application of this tool for ores having particles with higher densities. In the proposed methodology, the proportion in area of the minerals which comprise a given sample is calculated through digital image analysis using the software OptLib. Knowing the density of the mineral and the proportion of each particle in the material, it is possible to estimate the behavior of the ore in the float-sink tests and build the washability curve thereof. To prove the applicability of the technique a case study was developed with itabiritic iron ore samples from a rougher stage of a spiral separator test. The washability curves of products were built using the proposed methodology and a comparison between the mass recovery results estimated by these curves and those observed in dense liquid testing and cleaner stage of spiral separator was done. In the case of the dense liquid, there was a difference of 7.5 percentage points between real data and the one estimated by washability curves. In the case of cleaner stage of spiral separator, this difference fell to 2.1 percentage points. The difference found in the case of dense liquid is explained by the fact that analyzes are made considering size ranges slightly different. Already the results observed in the case of spiral separator cleaner stage were satisfactory, indicating that the proposed methodology fits well to reality when the original parameters are respected. Furthermore, an attempt was made to determine the material grade curves using the results generated in OptLib software, but a comparison with real results obtained in laboratory showed differences of 8.3 percentage points, which shows that efforts should be made in order to perfect the technique. Through thei work developed it was concluded that there is a great potential for the determination of washability curves using digital image analysis, including ores with high densities, such as iron ore.