2058 objective function, with the material being separated based on grade-by-size behaviour. The underlying mechanism relies on different fracture behaviour in the blasting process of the ore based on its mineralogy. This results in coarse separation of mineralized and barren fractions (Carrasco, Keeney, and Napier-Munn 2016). The use of a preconcentration facility, along with better integration of other elements of the mining complex, resulted in reduced risk of exceeding operational capacity in the leaching facilities and lowered the demand for stockpile use. Additionally, the optimized schedule yielded a $140 M increase in NPV of the project. Given the large size of the optimization problem, the function was solved using a reinforcement learning algorithm, which significantly decreased solution time and resulted in a schedule that can be further reutilized in the next planning horizons. The techniques discussed in this section, summarized in Table 2, highlight the potential of jointly optimizing multiple components of the mining complex to reduce operational risk and increase the NPV of the project. Although none of the studies quantify the impact on energy consumption directly, the integration of geometallurgical properties into the objective function can be used to leverage more selective extraction to decrease the energy demand in the downstream processes. Table 2 – Summary of methodologies connected to material considerations Author Year Objective function Planning scope Model type Solution methodology/Solver Kumar and Dimitrakopoulos 2019 Maximize DCF, penalize deviation from capacity constraints, penalize deviations from geometallurgical targets, Long-term SMIP Metaheuristics (multineighbourhood simulated annealing with adaptive neighbourhood search) Morales et al. 2019 Maximize DCF, minimize cost of deviations, optimize milling time of each block Long-term SIPDBS Heuristics (ETInc) Levinson et al. 2023 Maximize DCF, penalize deviations from production targets, minimize cost of preconcentration facility use Long-term SMIP CPLEX, metaheuristics with ML (reinforcement learning) Another potential application of geometallurgical data in mining enterprises is smart optimization of the milling circuit based on energy supply. In operations that are connected to national grids with a dynamic energy pricing policy, the feed can be adjusted to capitalize on lower electricity prices during offpeak hours to process harder material, reducing costs. Similarly, if the operation has access to renewable energy sources, such as PV panels or wind turbines, the material with less desirable grinding characteristics can be scheduled for milling in peak production periods, minimizing emissions. Pamparana et al. (2017) presents one such example. However, this approach does not acknowledge the geological uncertainty of the feed material and relies on sampling the stockpiled material. The combination of geometallurgy and stochastic optimization of mining complexes can further be developed by integrating direct feedback on material qualities from the processing stage. In light of the increasing application of real-time mining techniques, the geometallurgical data collected downstream can be used to update the extraction models (Anvari and Benndorf 2025). This would offer an opportunity to decrease uncertainty and generate more robust, dynamically adjusted short-term schedules that could better align with the feed requirements of the processing stage.
RkJQdWJsaXNoZXIy MTM0Mzk2