Applied Earth Science: Transactions of the Institutions of Mining and Metallurgy

This article details two geometallurgical case studies using classification schemes developed for Mn fine and lump ores. For the first, the relative abundance of 15 material types was compared to chemistry by size fraction. Positive correlations were evident between the proportion of aluminosilicate-bearing ore groups and Al₂O₃ content, the ratio of Mn oxide/Fe oxide ore groups and the Mn/Fe content, the proportion of cryptomelane-bearing groups and the K₂O and BaO contents, and the ratio of hard to moderately hard + friable particles and the K₂O + Ba + Na content. For the second, agreement was observed between the types of predominant material types in the two different ores and their mass distributions, major and trace element chemistries. Different material types had clear variances in their envelope and apparent particle densities. These two case studies support the expanded use of particle-based ore classification schemes for the characterisation of Mn ores to better predict their downstream processing performance.

The Fongo-Tongo's bauxites were investigated to characterise them. Their texture varies from massive, vesicular, alveolar, conglomeratic to nodular with dominantly red colour, reddish-brown and yellow. The main minerals identified by X-ray diffraction are gibbsite and goethite with subordinate quartz, anatase, hematite, magnetite, and traces of kaolinite. The abundance of gibbsite and goethite suggested intense weathering during the formation of the bauxite deposits. Chemical data of the bauxite showed high Al₂O₃ (37.4–57.5 wt-%) with varied Fe₂O₃ (3.97–29.5 wt-%), TiO₂ (0.57–7.5 wt-%) and SiO₂ (0.48–3.21 wt-%) contents, while other oxides are generally less than 0.6 wt-% indicating high bauxite quality with low impurities. The wide range of trace and REE concentrations of Zr, Nb, Sr, V, Ce, La, Nd and the presence of both positive and negative Eu anomalies suggested and acid igneous source with mafic input. These bauxites could serve as raw material for the aluminium industry.

Debre Tabor kaolin deposit is located around Debre Tabor town in Amhara region of northwestern Ethiopia. The kaolin deposit in the study area needs a detailed study to evaluate the geological, mineralogical, physical, and geochemical conditions. For this purpose, detailed geological, physical, mineralogical, and geochemical laboratory tests were performed. XRD and petrographic analysis were used to study the mineralogical composition. Geochemical analysis was determined using ICP-MS and ICP-AES. The Debre Tabor kaolin deposit is exposed along riverbanks, road cuts, hillside, and quarry sites. The laboratory results reveal that the deposit is formed from the weathering of felsic rocks mainly trachyte and tuff units. From the laboratory analysis, we found that quartz is the dominant impurities. The geological, mineralogical, and geochemical studies indicate that in situ weathering followed by leaching played a major role in the formation of the Debre Tabor kaolin deposit. The average Chemical Index of Alteration (77.98%) shows that the host rock has experienced moderate weathering and alteration under the hot and humid climatic conditions of the subtropical zone. The low bulk density (1.54 g cm⁻³) and Ce + Y + La vs. Ba + Sr plot correspond to the supergene type of formation. Based on the mineralogical, geochemical, and physical properties, Debre Tabor kaolin deposit can be used for refractory, ceramics, paper coating, and filler (paint) industries, but treatment is important to remove the existing impurities such as iron, quartz, and feldspars.

Machine learning (ML) models provide useful tools to generate spatial estimations of geological features, but they do not consider the spatial dependence among the observations and they primarily use coordinates as predictors. Thus, many ML models produce visible artifacts in the resulting estimates along the coordinate directions. To overcome this significant problem, this paper presents an ensemble super learner (ESL) model which uses the super learner (SL) model as the ML model. In the ESL model, numerous training sets are created from the original dataset by a coordinate rotation strategy and then the estimates obtained from the fitted SL models are ensembled to produce a final estimate. A dataset from a high-grade gold deposit demonstrates the approach and compares the results to kriging and the SL model. The results demonstrate that the ESL model manages artifacts in ML spatial estimation. It also provides better results than the kriging and SL model in terms of estimation accuracy.