PREDICTING THE VISCOSITY OF CHROMIUM-CONTAINING SLAGS USING PARAMETERS OF INTERATOMIC INTERACTIONS IN MELTS

Abstract

The aim of this work is to develop predictive models of the viscosity of multicomponent CaO–SiO₂–Al₂O₃–MgO–FeO/Fe₂O₃–MnO–Cr₂O₃ slag systems using SNIR interatomic interaction parameters (System of nonpolar ionic radii). During the study, expert filtering and systematization of the experimental data set (over 490 values) were performed, which allowed the identification of the region of stable homogeneous melts. Data clustering was performed using the K-Means and DP-GMM methods. It was established that the NBO/T structural descriptor provides the most physically sound classification of slags into three groups depending on their degree of polymerization. For each cluster, a mathematical model of viscosity was constructed using integral parameters of interatomic interaction—the electronic chemical equivalent ∆e and the stoichiometry parameter ρ. It has been proven that local models better account for the specific influence of amphoteric chromium oxide, which is negated in the overall array due to the dominant influence of the main components. The results obtained provide a basis for optimizing the compositions of slag-forming mixtures during the smelting of ferrochromium steels, which contributes to reducing the consumption of alloying elements and increasing the energy efficiency of the process.