AI-methods to Link Mineralogy and Core Sawing to Comminution Efficiency (eCoreX)
Important results from the project
The project demonstrated that sawing specific energy, P-wave and XRF data can be used to predict mechanical rock properties. AI models were trained to estimate strength, which was used in DEM simulations of crushers and mills. This method enables faster and lower-cost core analysis and could improve grinding circuit energy efficiency and mine planning in the future. We could also predict mineral lithology relatively well.
Expected long term effects
In the long term, the method could contribute to the development of digital twins for grinding processes, directly linked to the properties measured and predicted for the entire population of drill cores. This enables better decision-making, lower costs for physical testing, reduced energy consumption, and increased sustainability in mining operations. The framework could likely be integrated into both exploration and real-time operations.
Approach and implementation
The project was carried out by an interdisciplinary team according to plan, including drill core experiments, P-wave measurements, Schmidt hammer testing, XRF analysis, AI modelling, and DEM simulations. No major delays occurred, and the collaboration between the parties worked well. Additional data would further improve the predictive capabilities.
The text has been written by the project team. The content is copied from the funding agency’s website and has not been reviewed by the Program Office.