Data-Driven Optimization of Energy-Efficient Comminution and Grinding Circuits Through the Integration of HPGR Operating Conditions, Grinding Media, Ore Characteristics, and Mineral Liberation

Data-Driven Optimization of Energy-Efficient Comminution and Grinding Circuits Through the Integration of HPGR Operating Conditions, Grinding Media, Ore Characteristics, and Mineral Liberation

sina soleimani1 mahmod eskandari nasab2

1) M.Sc. Student in Mineral Processing Engineering, Shahid Bahonar University of Kerman
2) Ph.D. in Mineral Processing Engineering, Shahid Bahonar University of Kerman

IOI Code : ACCD-ABKHZB
Publication : 10th International Conference on Applied Researches in Science & Engineering (carseconf.com)
Abstract :
    Comminution and grinding are the most energy-intensive stages in the mineral processing value chain, and their performance directly affects mineral liberation, the efficiency of downstream separation processes (e.g., flotation and leaching), and the overall processing cost. This paper presents an integrated data-driven framework for the optimization of comminution and grinding circuits based on five experimental and computational studies published in leading international journals. The proposed framework encompasses four major dimensions: (1) the influence of grinding environments (dry, wet, and wet-conditioned) on pulp chemistry, particle morphology and size distribution, and consequently on the flotation selectivity of semi-soluble calcium-bearing minerals (scheelite, apatite, and fluorite); (2) hybrid modeling through the integration of first-principles models and multilayer perceptron (MLP) neural networks to compensate for uncertainties associated with the feed particle size distribution in high-pressure grinding roll (HPGR) and flotation circuits; (3) unsupervised machine learning using the K-means clustering algorithm to classify semi-autogenous grinding (SAG) mills and overflow ball mills according to their dominant wear mechanisms (impact, abrasion, and corrosion), thereby enabling the targeted selection of grinding media; and (4) the design of energy-efficient comminution circuits for emerging engineered materials, such as lithium-bearing battery slags, using roller mills configured in both series and closed-circuit arrangements. In addition, Bond Work Index (BWI) and specific comminution energy data for copper slag used in alkali-activated concrete production are incorporated as comparative benchmarks for evaluating energy consumption in the processing of secondary materials. The integrated findings demonstrate that dry grinding enhances the flotation selectivity of semi-soluble minerals by reducing the concentrations of dissolved Ca²⁺ and Mg²⁺ ions in the pulp. Furthermore, hybrid models are capable of compensating for model–process mismatch with coefficients of determination (R²) exceeding 0.97. Machine learning-based clustering of grinding environments can reduce grinding media selection errors by approximately 30%, while closed-circuit grinding configurations achieve up to 20% lower specific energy consumption than equivalent series grinding arrangements. Based on these findings, an integrated decision-support framework is proposed in which operational data from HPGRs and grinding mills, ore characteristics (including abrasivity and Bond Work Index), pulp chemistry, and mineral liberation data are simultaneously processed within a data-driven decision support system. This framework enables the simultaneous optimization of energy consumption, mineral recovery, and grinding media selection, thereby improving the overall efficiency and sustainability of mineral processing operations.

Keywords : comminution grinding high pressure grinding rolls (hpgr) grinding environment hybrid modeling k means clustering bond work index (bwi) mineral liberation energy efficiency flotation of semi soluble minerals