A multiscale slice model for continuous casting of steel

Abstract

A simple Lagrange-an traveling slice model has been applied for the prediction of the relations between process parameters, macrosegregation and solidification grain structure formation (equiaxed to columnar and columnar to equiaxed transition) during the continuous casting process of steel billets. The main advantage of the slice model is its very fast calculation time in comparison with the complete 3D heat and fluid flow model which might need calculation time, measured in days. The slice models thus allows for fast optimisation and even for on-line simulation. The heat and species transfer models are based on the mixture continuum assumptions with Lever solidification rule and enhanced thermal and solutal diffusivities for heuristic accounting of fluid flow effects. The grain structure evolution model is based on the Gaussian nucleation rule, and KGT growth model, coupled to the macroscopic heat and species transfer models. The heat and species transfer models are solved by the meshless technique by using local collocation with radial basis functions. The grain structure evolution model is solved by the point automata technique, a novel meshless variant of the cellular automata method. A comparison of the results with the experimental data for steel grade 51CrV4 is shown in terms of macrosegregation and grain structure across the billet. Simulations and comparisons have been carried out for nominal casting conditions, reduced casting temperature, and reduced casting speed. The model predicts surprisingly well the qualitative features of the macrosegregation and grain structure patterns. Possible refinements of the model with respect to other physical mechanisms are discussed.