Melhoria do balanço de massa na planta de concentração húmida B, optimizando as amostras de alimentação em comparação com o plano da mina da KENMARE

Abstract

The province of Nampula stands out for its significant deposits of heavy mineral sands, exploited by Kenmare Resources PLC. Considering the impact of these resources on the extractive industry and revenue generation, it is essential to have reliable data on what is actually being extracted. The mine plan is based on the block model, which should reflect the real mining and processing conditions. However, discrepancies often arise between the planned grades and those obtained during operations at Kenmare Resources PLC, in the Wet Concentration Plant B (WCP-B). These inconsistencies directly affect process control and the reliability of the mass balance. This study evaluated the representativeness of the block model in relation to feed samples from the Wet Concentration Plant B, with the objective of improving the mass balance. The relevance of this topic lies in its contribution to the optimization of mineral beneficiation and decision-making based on reliable data. The methodology adopted was exploratory and quantitative, involving the collection of total heavy mineral grade data from the block model and from the beneficiation plant (WCP-B). The process included the characterization of the mined area and sampling conditions at three strategic points of the plant: the dredge, the desliming cyclone feed, and the rougher spiral feed (RSF). Statistical tools were applied, and comparative analyses were performed using graphs and recovery calculations. This approach enabled a critical analysis of sample quality in the process, focusing on evaluating total heavy mineral grades, identifying systematic deviations, and determining sample precision. The results revealed discrepancies between the total heavy mineral grades from the block model and those from the beneficiation plant (WCP-B). The dredge showed a moderate to weak correlation with the block model, while the desliming cyclone feed point showed a weak correlation. The rougher spiral feed exhibited better alignment with the block model, with a concordance index of 84.3%, a Pearson correlation coefficient of 0.754, and a determination coefficient of 0.569, indicating that, despite existing limitations, the block model can be used as a reference for mined grade control.