Reducing density uncertainty in iron ore deposits: Taking advantage of the density and Fe grades correlation aiming to more accurate models

Abstract

This study addresses the critical role of density in the economic evaluation of mineral deposits and how uncertain could be the density models without considering the available secondary information. Most mining companies subjectively determine sample quantity and spatial distribution for density estimation. Usually, the density samples are sparser than the grade samples. Fewer samples tend to generate higher uncertainty as less information is available. Although the density has fewer samples, it is correlated positively with iron grades in an iron deposit. However, this correlation between density and grade is not considered to generate density models. In this context, this research aims to reduce the uncertainty associated with density in iron ore deposits by exploring the correlation between grades and density, using this information to create density models. Therefore, the iron grade was used as an auxiliary variable to create density models proposing different multivariate geostatistical techniques. Two cosimulation approaches were proposed. The first one uses simple cokriging to incorporate the auxiliary variable, while the second uses intrinsic collocated cokriging. These two approaches were compared against univariate geostatistical simulation, which ignores the correlation between density and grades. The results demonstrated the effectiveness of incorporating iron grades in density models, observing histogram matching, reproduction of spatial continuity, the possibility of using the E-Type as estimated models, conditional standard deviation reduction and better accuracy plots in ore and waste domains. Also, a mass analysis was performed in ore domains, making evident how consider Fe grades could change what to expect in future. The study concludes that leveraging iron grades significantly reduces uncertainty in density models, providing accurate and reliable models and proposing methodologies that could be extrapolated to estimates, being viable to create density estimated models for more effective mining planning and resource management in iron deposits.