A modified method for calculating the viscosity of multicomponent slags based on Kriging interpolation

Model prediction is an effective method to obtain the physicochemical properties data of molten slags, which is difficult to test experimentally due to their high melting points. Empirical models are prevalent at present; however, it needs large amount of experimental data to fit the empirical parameters with narrow application scope. In this paper, the Kriging interpolation method modified with oxide property weights, is firstly introduced into the viscosity prediction of multicomponent slags. The prediction results of CaO–Al2O3–SiO2, CaO–Al2O3–CaF2, CaO–Al2O3–SiO2–MgO and CaO–Al2O3–SiO2–MgO–CaF2 systems showed that the predicted errors by the Modified Kriging Interpolation (MKI) method are smaller than those by various empirical models. It is anticipated that the MKI method can be extended to predicting the continuous physicochemical properties of multicomponent slags.