Knowledge of the grade of valuable elements and its variation is not sufficient for geometallurgy. Minerals define not only the value of the deposit, but also the method of extraction and concentration. A number of methods for obtaining mineral grades were evaluated with a focus on geometallurgical applicability, precision and trueness. For a geometallurgical program, the number of samples to be analyzed is large, therefore a method for obtaining mineral grades needs to be cost-efficient, relatively fast, and reliable. Automated mineralogy based on scanning electron microscopy is generally regarded as the most reliable method for analyzing mineral grades. However, the method is time demanding and expensive. Quantitative X-ray diffraction has a relatively high detection limit, 0.5%, while the method is not suitable for some base and precious metal ores, it still provides significant details on gangue mineral grades. The application of the element-to-mineral conversion has been limited to the simple mineralogy because the number of elements analyzed limits the number of calculable mineral grades. This study investigates a new method for the estimation of mineral grades applicable for geometallurgy by combining both the element-to-mineral conversion method and quantitative X-ray diffraction with Rietveld refinement. The proposed method not only delivers the required turnover for geometallurgy, but also overcomes the shortcomings if quantitative X-ray diffraction or element-to-mineral is used alone

Process models in mineral processing can be classified based on the level of information required from the ore, i.e. the feed stream to the processing plant. Mineral processing models usually require information on total solid flow rate, mineralogical composition and particle size information. The most comprehensive level of mineral processing models is the particle-based one (liberation level), which gives particle-by-particle information on their mineralogical composition, size, density, shape i.e. all necessary information on the processed material for simulating unit operations. In flowsheet simulation, the major benefit of a particle-based model over other models is that it can be directly linked to any other particle-based unit models in the process simulation. This study aims to develop a unit operation model for a wet low intensity magnetic separator on particle property level. The experimental data was gathered in a plant survey of the KA3 iron ore concentrator of Luossavaara-Kiirunavaara AB in Kiruna. Corresponding feed, concentrate and tailings streams of the primary magnetic separator were sampled, assayed and mass balanced on mineral liberation level. The mass-balanced data showed that the behavior of individual particles in the magnetic separation is depending on their size and composition. The developed model involves a size and composition dependent entrapment parameter and a separation function that depends on the magnetic volume of the particle and the nature of gangue mineral. The model is capable of forecasting the behavior of particles in magnetic separation with the necessary accuracy. This study highlights the benefits that particle-based models in simulation offer whereas lower level process models fail to provide.

The ore texture and the progeny particles after a breakage in the comminution are the missing link between geology and mineral processing in the concept of geometallurgy. A new method called association indicator matrix based on co-occurrence matrix was introduced to analyze the mineral association of ore texture and particles.  The association indicator matrix can be used as a criterion to classify ore texture and analyze breakage behavior of ore texture. Within the study, the outcome of breakage analysis with association indicator matrix was used to generate particle population of iron ore texture after crushing. The particle size of forecasted particles was taken from experimental and frequency of breakage in phases was defined based on association indicator and liberation of minerals. Comparison of liberation distribution of iron oxide minerals from experimental and forecasted population shows a good agreement.

The mineral processing simulation models can be classified based on the level that the feed stream to the plant and unit operations is described. The levels of modeling in this context is as bulk, mineral or element by size, and particle. Particle level modeling and simulation utilizes liberation data in the feed stream and is more sensitive to the variations in ore feed quality. Within the paper, results of simulation for two texturally different magnetite ore is demonstrated in bulk, mineral by size and particle level. The models were calibrated for one ore, and all the modeling levels show similar results, but for the other ore, the results differ. This is because, in the bulk level, the model assumes that magnetite, as well as other minerals, do not change their behavior if ore texture and grinding fineness are changed. In the mineral by size level, the assumption is that minerals behave identically in each size fraction even the ore texture changes. In the particle level, the assumption is that similar particles behave in the same way. The particle level approach gives more realistic results, and it can also be used in optimization, thus finding the most optimal processing way for different geometallurgical domains. In iron ores where iron minerals are highly liberated the particle level shows its power in the prediction of impurity levels rather than iron grade and recovery.