Rem: Revista Escola de Minas
Print version ISSN 0370-4467
On-line version ISSN 1807-0353
COSTA, João Felipe Coimbra Leite et al. Improving blending strategies in chevron piles using geostatistical simulation. Rem: Rev. Esc. Minas [online]. 2008, vol.61, n.3, pp.291-296. ISSN 0370-4467. http://dx.doi.org/10.1590/S0370-44672008000300005.
Blending and homogenization piles are usually employed for reducing grade fluctuation from the ore feeding a processing plant. These piles are characterized by their form, size (length and height), construction lay-out and number of layers. Various methods are found for blending piles design and most fail to incorporate the in situ grade variability intrinsic of a mineral deposit. The methodology suggested in this study is able to quantify the variability of the homogenization system using multiple equally probable realizations derived from a geostatistical simulation for the grade block model. The variable tested was silica (SiO2) in a grannulometry below 0,15mm (SI3), an erratic contaminant in iron deposits. The results demonstrated a reduction exceeding 90% in the original variability (in situ compared to grades feeding the processing plant) of the system in favorable conditions. Finally, the reduction in the uncertainty in the head grades feeding the processing plant, leads to cost savings and lower economical risks.
Keywords : Homogenization; Piles; Sequential Gaussian Simulation (SGS).