中文摘要：

基于Eulerian-Eulerian方法,阐述了简化枝晶状等轴晶、柱状晶以及金属液三相完全混合的凝固模型.模型考虑了等轴晶的移动及柱状晶对等轴晶的捕获,跟踪了柱状晶尖端的位置并考虑了等轴晶和柱状晶的相互竞争生长,因此该模型具备了预测柱状晶向等轴晶转变（CET）的能力;为了在不过量增加计算量的前提下提高模型的精度,模型对等轴晶采取了简单的枝晶化处理,即采用简化方法描述等轴晶包络线内固相分数.分别模拟了3.25和25 t钢锭的凝固过程,成功预测了大型钢锭凝固过程所形成的底部锥形负偏析、“类-A型”偏析以及CET等现象.分析认为长细形状铸锭中出现的顶部负偏析区,是由于凝固后期所形成的局部小钢锭及等轴晶在其内部的沉积聚集而成.

英文摘要：

The prediction of the macrosegregation in large ingot is a challenging issue due to the size of the ingots and the variety of the phenomena to be accounted for, such as thermal-solutal convection of the liquid,equiaxed grain motion, evolution of grain morphology by suitably considering a coupled grain growth model in the macroscopic solidification model, the coltmmar-to-equiaxed transition （CET）, and shrinkage, etc.. Each of these phenomena is very important to the solidification pattern, while it is impossible for one model to consider all the phenomena together until now due to the computation power limited. Thus, the model capability and computational cost should be counterpoised for the simulation of large ingot. In this work, a mixed three-phase （simplified dendrit- ic-equiaxed, columnar and liquid） solidification model is described based on Eulerian-Eulerian approach and vol- ume average method. The model considers the thermosolutal buoyancy flow, the movement of equiaxed crystal, and the capture of the equiaxed crystals by growing columnar tree trunks. The mechanical interaction and impinge- ment between columnar and equiaxed crystals are considered which give the capability to predict CET. In order to enhance the model capability without increasing the computational cost significantly, a simplified method is pro- posed to consider the dendritic of equiaxed crystal. This model is employed to simulate the formation process of macrosegregation for two different steel ingots （3.25 and 25 t）. The general macrosegregation pattern predicted by this model includes the cone of negative segregation in the bottom of ingot, quasi-A-segregation in the columnar zone, and positive segregation in the top region, which are quite similar to the classic knowledge. The CET zones are also predicted. Although there is still some quantitative discrepancy, the macrosegregation distribution predict- ed by this model is quite similar to the experimental measurements. The non-globular equiaxed three-phase mixed model re