Electrical Conductivity Calculation of Molten Multicomponent Slag by Neural Network Analysis

Abstract

Only a few models for estimating electrical conductivity of molten slag have been developed, which are limited to systems with certain types of components. In this study, a new method for estimating the electrical conductivity of molten slag through neural network calculations is proposed and is compared with previous estimation approaches. The present estimation approach can reproduce the electrical conductivity of molten slag composed of SiO₂, CaO, MgO, MnO, Al₂O₃, FeO, Fe₂O₃, and Na₂O to an uncertainty of 30%. We found that the neural network calculation is applicable to various kinds of molten slag over wider mole fraction and temperature ranges than conventional models.