Multiscale modelling of twin roll casting

Abstract

Twin roll casting (TRC) is an emerging material processing technique to manufacture thin metal strips directly from liquid metal. Formation of unfavourable microstructural features during TRC, such as centreline segregation and columnar grains, prompted experimental and modelling studies on the effects of various casting conditions on the as-cast microstructures of different alloys. Previous modelling work focuses on low speed TRC with large rolling effect. By increasing nuclei density in the melt, via addition of grain refiners or melt conditioning, the effect of solidification can be significantly improved. This thesis concerns the development and application of multiscale multiphysics modelling techniques to provide an insight into the evolution of microstructure during solidification processing of metals, with a focus on twin-roll casting of thin strips of light alloys. The effects of various casting conditions on the as-cast microstructures are investigated using the multiscale model. The first part of the multiscale model is a macroscale thermal-mechanical model, which predicts the temperature and stress distributions developed in the metal strips during TRC, under the influence of various casting conditions, including casting speed, roll temperature and air convection. The theoretical maximum casting speed is calculated from a quasi-1D solidification model, which can be used to give a guideline for casting conditions used in models and experiments. The second part is a microscale model which predicts the as-cast microstructure, via a phase field-Potts model which simulates the evolution of phase field, alloy concentration and grain orientation during solidification, coupled with a Bingham lattice Boltzmann model which simulates the velocity field in fluid flow. The temperature profile obtained from the macroscale model is used as the thermal boundary condition of the microscale model. To validate the model, temperature data obtained from experiments of Al and Mg TRC is compared with the results of the macroscale model. More experimental data of the as-cast microstructure and texture is required to validate the microscale model.