Control of fluid flow, heat transfer and inclusions in continuous casting: CFD and neural network studies

Shenqiang Wang,J. Laine,P. Väyrynen,Scppo Louhenkilpi

Abstract:Fluid flow and heat transfer calculations have been carried out in tundish and mould including different kind of submerged entry nozzles. The effect of different kind of tundish dams have been studied in a bloom tundish in steady state and transient conditions. Many different CFD parameters, like turbulence models and mesh density, were tested during the studies. CFD calculations were also carried out to study the effects of swirling flow inside the SEN as well as different kind of SEN nozzles on mould flow phenomena. Different kinds of criteria for the ideal mould flow were derived. Neural network model was developed to predict and control the tundish temperature from process parameters and casting time. The model, which is based on the Bayesian MLP neural network, includes 13 inputs from ladle cycle and the output is the tundish temperature. The results were good; the mean error for temperature was 3.4 °C.

Engineering,Materials Science,Computer Science