Geometallurgical model of the work index for a porphyry copper deposit by machine learning

The development of mining projects in the exploration and pre-feasibility stages involves challenges related to geological and processing heterogeneity and uncertainty due to variability in their primary and response attributes. Therefore, the objective of this research is to develop the geometallurgical model of the Bond work index. For this purpose, linear regression models were developed in Jupyter Notebook considering 790 samples of uniaxial compressive strength of rock (UCS), lithologies, mineral zones and Bond work index (BWi) in a porphyry copper deposit. The results indicate a directly proportional linear relationship between BWi and UCS, generating modeling with acceptable R2 performances between 0.76 and 0.90. In addition, the lithologies and ore zones in the deposit studied are relevant characteristics related to comminution. Finally, according to rock competence, the rock is classified as medium to hard, hard and very hard in the oxide, mixed and primary sulfide zones, respectively.