Numerical Analysis of the Influence of Melting and Application of Electromagnetic Stirring Prior to Solidification on Macrosegregation Formation during Casting of a Binary Alloy

"Numerical simulations of macrosegregation formation during horizontal solidification of a hypoeutectic Sn-Pb alloy in a rectangular cavity is presented and compared to experimental results. The experiment involves a melting phase, a holding stage at constant temperature with/without application of electromagnetic stirring, and finally solidification. We simulate all stages of the experiment and study the segregation of Pb throughout the experiment. A twophase volume averaged model is used for the numerical simulations, accounting for thermosolutal convection and assuming perfect microscopic mixing (lever rule). The influence of the melting phase and the electromagnetic stirring is studied with different cases to illustrate the consequences on the flow pattern and Pb segregation. The numerical simulations are compared to the experimental data and the impact of chemical inhomogeneities prior to solidification on macrosegregation formation is discussed.IntroductionMany extensive investigations have been performed so far on the solidification of metallic alloys. However, the melting process of metallic alloy has not attracted attention in a same manner. Nevertheless, melting of alloy may put forth peculiar effects as described in the present paper. It has been observed from a benchmark experiment [1, 2] that significant stratification might arise during the melting process of Sn-Pb alloy. That effect was put forth thanks to the analysis of the experimental temperature distribution which indicated that the convective fluid flow was almost zero. The effect was enhanced when the lead composition was higher (above 6wt.%). The present paper is aimed at confirming that the melting of an alloy containing a heavier solute may be accompanied with a stratification of the solute which exerts a stabilizing effect on the convective fluid flow. Results obtained from numerical modeling are compared with experimental data obtained from the benchmark facility."