Multiphysics Modeling of Steel Continuous Casting

Brian Thomas, University of Illinois at Urbana-Champaign

Usage Details

This project aims to develop computationally-intensive multiphysics models to predict turbulent fluid flow, MagnetoHydroDynamics (MHD), bubble behavior (formation, coalescence, and breakup), particle transport, two-phase interfacial flow, heat transfer, solidification, and thermal-mechanical behavior in the continuous casting of steel. These models will be applied to gain practical insights into defect formation mechanisms, and to improve this important manufacturing process for better final-product quality. Specifically, advanced thermal multiphase-flow simulations with the simulated gas bubble size distribution, particle capture model, MHD, and heat transfer will be conducted to get insights into transient particle-defect formation and effect of Electro-Magnetic Stirring on reduction of the defect, and to find safe operating windows of adjustable casting conditions. The multi-GPU code CUFLOW has been developed and tested on the Blue Waters XK node, with NVIDIA K20x GPU co-processors, and shows good speed up. Recent testing of commercial code, ANSYS FLUENT also shows good scaling on the Blue Waters XE node. Blue Waters is needed to shorten simulation times (currently several months on workstations), in order to enable better mesh refinement for the high-resolution analysis, which is important for accurate thermal multiphase-flow simulation, needed for realistic investigation of the defect formation. Transient thermal-stress models of the solidifying steel shell will be applied to investigate strain concentration and the formation of longitudinal cracks and depression problems, again requiring very fine computational meshes, which are only possible on Blue Waters.