NUMERICAL AND EXPERIMENTAL STUDY OF MARANGONI FLOW ON SLAG-LINE DISSOLUTION OF REFRACTORY

Abstract

<p>The local corrosion of refractories at the slag/gas interface is a serious problem that limits the life of the refractories.Although, there have been several studies focused on understanding the Marangoni effect on the refractory dissolution process, there is little quantifiable analysis available. The aim of this study is to establish a better fundamental understanding of refractory dissolution mechanisms, and develop appropriate models for predicting the extent and rate of slag-line dissolution.</p> <p>In the first part of this research, experimental studies using a high temperature dip technique were performed: MgO refractory in SiO₂-CaO-FeO_x-MgO slag and Al₂O₃- SiO₂-CaO-FeO_x-MgO. The experiments were conducted at varies temperature. There was significant evidence of a spinel phase formed at the slag/refractory interface for slags containing 20wt.% Al₂O₃. This existence of the spinel seems to have retarded the dissolution of the refractory. The decrease in erosion rate in the presence of spinel is in proportion to the decrease in the equilibrium MgO concentration at the slag/solid interface. The activation energy is calculated from the relationship of effective mass transfer coefficient vs. temperature and found in the range of mass transfer activation energy.</p> <p>The second part of this search is developing a numerical model to predict the slag-line dissolution. An effective algorithm for analysis of unsteady Marangoni convection in refractory slag line dissolution has been developed. The results show that the Marangoni effect plays a very important role in slag-line erosion at this condition; both the moving boundary condition and curved surface condition have significant effects on the slag-line erosion rate. The comparison of experimental and numerical results shows that the model can predict the refractory maximum corrosion distance caused by Marangoni flow at the slag line. However, the eroded material volume was predicted within 20~30% deviation</p>