Modeling of basalt tuff beneficiation by dry high-intensity magnetic separation

Purpose. To identify the regularities governing the influence of magnetic field induction and particle size distribution on the technological performance of dry high-intensity magnetic separation (DHIMS) of tuff raw materials and on the quality of the resulting magnetic product. Methodology. The study was carried out using a laboratory-scale dry magnetic separation unit. The behavior of tuff was investigated in narrow particle-size fractions of +0.63–2.5 and +0.1–0.63 mm under varying magnetic field induction (0–1.3 T), while maintaining constant separator design parameters and a fixed material feed regime. Technological performance was evaluated based on yields of magnetic and non-magnetic fractions, the calculation of distribution functions, and the construction of regression models. The chemical composition of the products was determined by X-ray fluorescence (XRF) analysis. Findings. The tuff was found to exhibit enhanced magnetic susceptibility and high friability, creating favorable conditions for efficient dry magnetic separation. The average yield of the magnetic fraction for the two samples studied was 51.8 %, with more than 98 % recovery achieved at magnetic field induction values up to B = 0.58 T. The regression models show a high degree of correlation (Pz = 0.97–0.975) within the acceptable induction range up to B = 0.9 T. XRF analysis of the magnetic product revealed an Fe2O3 content of 30.89 %, confirming effective concentration of iron-bearing components in the magnetic fraction. Originality. Mathematical modeling of the DHIMS process for tuff was performed for narrow particle-size classes, with the distribution function of magnetic separation product yield determined as a function of magnetic field induction. A consistent pattern of dependencies was established for the investigated size classes, and the statistical significance of the regression model parameters was confirmed (significance level ≤ 0.05). It was demonstrated that linear-logarithmic models provide the best description of the process up to an induction of 0.9 T. Practical value. The results enable a substantiated selection of DHIMS operating regimes for the integrated processing of tuffs, identification of rational particle-size classes for the removal of magnetic impurities, and prediction of process efficiency based on magnetic field induction. The obtained regression relationships can be applied in the design and optimization of preliminary beneficiation schemes for tuff raw materials, aimed at reducing the mass of material fed to subsequent energy-intensive operations.