2006 Volume 46 Issue 3 Pages 346-351
A new estimation method of viscosity or solidification temperature of mold fluxes was proposed by applying the neural network computation. In this evaluation system, the viscosity and the solidification temperature of mold fluxes can be evaluated from the analytical compositions in multi-component systems of SiO2–Al2O3–CaO–MgO–Na2O–F–T.Fe–ZrO2–TiO2–BaO–MnO–B2O3–S–C without any conversion of S or F to sulphide or fluoride. It was found that the calculated results of the dependence of viscosity on temperature and composition agree with the experimental results more precisely than some conventional physical models for viscosity. Furthermore, viscosity of mold fluxes can be estimated precisely in the wide range of SiO2 content.
