Abstract

Chilean mining must reduce its high energy consumption to remain sustainable amid declining porphyry copper grades. Early gangue rejection via a coarser classification cut, directing the undersize fraction to coarse flotation-regrinding, is adopted as mitigation measure. However, clay-rich porphyry ores, owing to their low Mohs hardness, generate ultrafine and fine particles during comminution. These fines impair flotation recovery, destabilize tailings, and hinder water recovery in settlers. Conventional corrective measures, such as chemical additives, raise operational costs and induce side-effects like particle aggregation and ion precipitation. Yet reducing fines can compromise mineral liberation, thereby degrading coarse flotation performance. To balance this trade-off, we formulate a surrogate-assisted multi-objective optimization problem that minimizes ultrafine generation while preserving an optimal particle-size distribution for early gangue rejection. Surrogate models, derived from simulated comminution systems, facilitate uncertainty quantification and reduce the dimensionality of the operational space, enabling algorithms to approximate the Pareto front effectively. The sensitivity analysis reveals that the most influential factors in generating ultrafine and fine particles include the ore feed rate to the plant, the Bond work index, the critical speed, and the volume of ball loading in ball mills. The influential input variables affecting coarse particles are the fraction of solids fed to hydrocyclones, the ore feed rate to the plant, the Bond work index, the critical speed, and the volume of ball loading in ball mills. Evaluation of seven multi-objective evolutionary algorithms in this reduced space shows that CNSGAII, SMS-EMOA, NSGAIII, and AGEMOEA strike the best compromise between minimizing fines and retaining coarse particles. © 2025 Taylor & Francis Group, LLC.