Modeling of geological attributes is a fundamental step in the mining process where quality resources are defined for mining and metallurgical processing. The metallurgical recovery of sulphide minerals in the flotation stage is a variable that depends not only on geological attributes such as ore type, alteration, etc., it also depends on operational parameters such as pH, quantity and quality of chemicals such as thickeners and collectors, residence time, granulometry, etc. These factors make modeling difficult, since the recovery might depend on factors external to geology.In this research we studied multivariable correlations that allow prediction of metallurgical recovery (Rec30 - percentage recovery of the ore after 30 minutes of flotation) through multivariate geostatistics: for this purpose an estimation of the recovery using co-kriging was performed taking into account variables that have high correlations. In this case study a high correlation between iron grades and recovery in potassium-rich alterations was found, which is attributed mainly to the amount of pyrite that makes the process difficult. Additionally, the incorporation of co-kriging allows increasing the estimated tonnage, when there is little information about the primary variable (but not the secondary variable).The advantage of using classical geostatistics is that recovery models can be obtained with good results in terms of cross-validation (good prediction), which overcomes the problem of non-additivity in the case of the generation of a block model for geometallurgical variables. In addition, the advantage of using co-kriging is that the information of this secondary variable is much denser, hence provides improved model resolution. The metallurgical recovery samples are usually expensive and few, the incorporation of secondary well correlated variable then generate a more robust and reliable recovery model.
展开▼