MACROSEGREGATION continues to be one of the concerns in direct chill (DC) continuous casting of aluminum ingots. The variation in composition over the ingot cross section results in the need to adjust alloy composition in order to meet mechanical property specifications in finished products. The extent of macrosegregation is greatly affected by the casting conditions, such as casting speed, superheat, cooling rate, mold size, alloy composition, and grain-refining practices. Thus, it is of interest to develop a mathematical model to simulate macrosegregation in DC casting, enabling the selection of conditions that minimize macrosegregation while maintaining a high productivity. Various mechanisms for macrosegregation in aluminum DC casting have been discussed by Chu and Jacoby [1] and Yu and Granger. [2] In general, macrosegregation is caused by the relative movement of liquid and solid in the mushy zone, because the alloy constituents have different solubilities in the two phases. One cause of interdendritic liquid flow through the rigid solid structure in the mushy zone is the contraction of the liquid and the density difference between the solid and liquid phases during solidification. The commonly observed inverse segregation pattern near the outer ingot surface is known to be the result of such contraction driven backflow of solute-rich liquid. Positive segregation at the ingot surface can also be caused by exudation of interdendritic liquid through channels in the partially solidified shell into the contraction gap between the shell and the mold. This flow is driven by the metallostatic head. [3‐6] Less clear are the transport phenomena leading to either positive or negative segregation in the cen