THE COUPLING OF MACROSEGREGATION WITH GRAIN NUCLEATION, GROWTH AND MOTION IN DC CAST ALUMINUM ALLOY INGOTS Miha Zaloˇ znik 1 , Arvind Kumar 1 , Herv´ e Combeau 1 , Marie Bedel 1,2 , Philippe Jarry 2 , Emmanuel Waz 2 1 Institut Jean Lamour, CNRS – Nancy-Universit´ e – UPV-Metz, Ecole des Mines de Nancy, Parc de Saurupt CS 14234, F-54042 Nancy cedex, France 2 Alcan CRV, 725 Rue Aristide Berg` es, BP 27, F-38341 Voreppe cedex, France Keywords: Direct chill casting, Aluminum alloys, Solidification, Macrosegregation, Microstructure Abstract The phenomena responsible for the formation of macrosegregations, and grain structures during solidifica- tion are closely intertwined. We present a model study of the formation of macrosegregation and grain structure in an industrial sized (350 mm thick) direct chill (DC) cast aluminum alloy slab. The modeling of these phenomena in DC casting is a challenging problem mainly due to the size of the products, the variety of the phenomena to be accounted for, and the non-linearities involved. We used a volume-averaged multiscale model that describes nucle- ation on grain refiner particles and grain growth, coupled with macroscopic transport: fluid flow driven by natu- ral convection and shrinkage, transport of free-floating globular equiaxed grains, heat transfer, and solute trans- port. We analyze the heat and mass transfer in the slurry moving-grain zone that is a result of the coupling of the fluid flow and of the grain nucleation, growth and motion. We discuss the impact of the flow structure in the slurry zone and of the grain packing fraction on the macroseg- regation. Introduction The macrosegregation in the DC casting process is gov- erned mainly by two mechanisms: by the melt flow in- duced by thermosolutal natural convection, shrinkage and pouring, and by the transport of solute-lean free-floating grains [1–6]. A commonly observed, surface-to-surface distribution of alloying elements at a transverse cross- section of a DC cast ingot reveals distinct regions of pos- itive (solute-rich) and negative (solute-depleted) segrega- tion [1]. A solute-depleted region is present in the ingot center, adjoined by a positive segregation zone spread- ing into the outward direction, an adjacent thin negative segregation zone and another positive segregation layer at the surface. Experimental investigations were pub- lished on macrosegregation and macrostructure in grain refined and non-grain refined ingots [2,7]. It was reported that macrosegregation generally increases with grain re- finement and linked this to the increased transport of free-floating coarse (slowly growing) grains, formed ei- ther by the fragmentation of dendrites or by nucleation on grain refiner particles. Eskin et al. [7] presented a sys- tematic experimental investigation of the dependence of macrosegregation and structure on process parameters. A strongly supported hypothesis states that the negative centerline segregation is caused by the transport of solute- lean free-floating grains to the center of the casting. First attempts to model the influence of free-floating grains were made by Reddy and Beckerman [3]. They considered nucleation of grains at a fixed temperature and the transport and growth of spherical globular grains in a slurry zone. The solid phase was assumed to form a con- nected rigid porous structure (packing) at a solid volume fraction of 0.637 (packing fraction). In the case of simu- lations accounting for grain motion, a significant negative segregation at the center of the billet was found. Vree- man et al. [4,5] proposed a simplified model with regard to grain nucleation and growth, assuming local equilibrium (lever rule) and a constant imposed characteristic grain diameter. They calculated the grain velocity directly from the grain diameter and the solid fraction. Macrosegrega- tion distributions in DC cast billets were calculated [5] and a parametric study of two key model parameters, the packing fraction and the grain diameter, was performed for Al-4.5 wt%Cu and Al-6 wt%Mg billets with a diame- ter of 400 mm. The results were qualitatively consistent with commonly observed macrosegregation trends. The study revealed a large degree of dependence on both the packing fraction and grain diameter. It has to be noted that their imposed grain diameter has to represent an ac- tual grain size distribution and that the packing fraction is not well known and, moreover, might not be uniform throughout the mushy zone. In a later work Vreeman et al. [4] compared model predictions to measurements on industrial-scale 450 mm diameter DC cast billets of an Al-6 wt%Cu alloy and tried to determine the value of the packing fraction to obtain the best fit. They found rea-