Análise comparativa de métodos de estimativa do teor de ferro em uma mina no Quadrilátero Ferrífero

Abstract

This study aimed to evaluate the performance of different grade estimation methods in the construction of block models for an iron deposit located in the Quadrilátero Ferrífero region. Based on a classification key, the lithologies present in the database were grouped into two domains to facilitate analysis: rich itabirites (ITR) and poor itabirites (ITP). Implicit geological solids were modeled using radial basis functions. This technique allows the modeling of complex geological bodies in an automated way, considering the spatial relationships between the different lithologies. The generated solids were classified according to the probability of belonging to each of the lithologies defined in the classification step. The block model was estimated using three interpolation techniques: nearest neighbor, inverse distance squared, and ordinary kriging. The results were compared and validated in terms of accuracy, considering the variability and correlation between the estimated values and the database values. The results showed that ordinary kriging performed better than the other methods. The choice of the interpolation method directly influences the quality of the estimated values for the block model.