Quantifying the value of geometallurgical information and optimization

Predictive Geometallurgy can use thorough quantitative characterisation of the ores microstructure and mineralogy to predict and adaptively optimize the processing and blending of the ore. Adaptive processing allows to exploit this variation to achieve higher overall recovery at lower costs, e.g. by adapting milling to mineral grain sizes or grades.

The geometallurgical properties of the ore are however typically only known with some geostatistical uncertainty. The talk demonstrates in model calculation, that naive optimization of processing parameters based on expected ore properties might lead to performance losses relative to non-adaptive processing, while overestimating its own expected performance. The contribution shows how to outperform non-adpative processsing decissions substantially based on a stochastic optimization approach and how to quantify the value of geometallurgical information in a certain setting.

Indeed in the context of predictive geometallurgy the value of a blend is no longer defined only by its acutally physical properties, but also by the information we have about it at the time of processing. This makes geometallurgical exploration activity a relevant part of the mine plan, which can actually change and on average increase the value of blocks.