Geometallurgy of iron ores aims at linking geological variability and responses in the beneficiation process by a wide usage of automated mineralogy, proxy metallurgical tests (e.g., Davis tube) and process simulation.

In this study several patterns from iron ore processing, in context of their textural description, are revealed and modelling is attempted. The first one is an ore classification method with a novel quality estimator X_LTU. The second one is an algorithm for predicting iron recovery in wet low intensity magnetic separator (WLIMS). The last one is predicting liberation distribution of iron oxides. Process variables are generated with a Davis tube and a WLIMS test. All streams are chemically characterised. Ore feed was studied for modal mineralogy and liberation distribution with QEMSCAN (Quantitative Evaluation of Minerals by Scanning Electron Microscopy). Post-processing for detecting the patterns and data dependencies was done with multivariate statistics: principal component analysis (PCA), and projection to latent structures regression (PLS). The study was done on 13 apatite iron ore type samples from the Leveäniemi mine (LKAB).

It is concluded that Davis tube test is better used only for marginal ores, since iron oxide minerals tend to be fully liberated in high grade magnetite massive ores after grinding. In addition, they will give high recovery and high mass pull in a Davis tube test.

The developed models can be used in populating a production block model in the future. Furthermore, future work should cover larger variability of marginal ores in terms of Fe grades and modal mineralogy (e.g. feldspar, amphibole, apatite dominated). Variability of grain size distribution ought to be included into future studies.