Modeling transient multiphase flow and mold top surface behavior in steel continuous casting

This thesis develops, validates and applies a system of computational models to investigate transient multiphase turbulent flow physics in the mold region and mold top surface behavior during continuous casting of steel slabs. Each model can be used independently or combined together as a comprehensive system to gain insights into the inter-related multiphase fluid dynamics in the caster mold during both quasi-steady-state and essentially transient events in practical casting operations. Argon gas commonly is injected into the liquid metal stream through porous refractory walls in many metallurgical processes. Modeling multiphase flows in caster molds is of great significance to understanding of inclusion transport and defect formation mechanisms and to improving the quality of the final products. To better understand the gas injection process, a new model is developed to investigate gas permeating through heated upper tundish nozzle (UTN) porous refractory, including the effects of nozzle geometry, gas thermal expansion, temperature-dependent gas viscosity, and possible gas leakages into unsealed joints. Furthermore, a procedure to predict initial bubble size is established. Two (semi-) analytical models, a stopper-position- and a gate-position- based model, predict liquid steel flow rate histories during the transient events and serve as a first step of this comprehensive model system. Argon-steel two-phase flow during a transient “declogging” event with multiple stopper-rod movements is simulated. The flow rate history during stopper rod movements is obtained from the analytical model, and the hot argon flow rate calculated using the porous-flow model and the initial bubble size estimated. Nail board experiments are also conducted to measure steel surface velocities and mold level profiles. A correlation for surface iii velocity prediction is proposed based on previous modeling results and validated by another set of measurements using a sub-meniscus velocity control (SVC) device. To further understand particle transport and deposition in wall bounded turbulent flows, direct numerical simulations (DNS) are performed in the continuous phase and a Lagrangian particle tracking algorithm was developed into the in-house code, CU-FLOW, to investigate dispersion and deposition of particles with different Stokes numbers in a square duct flow with and without the effect of imposed magnetic field. A new free-surface tracking model with a moving-grid technique is developed and integrated in the commercial computational fluid dynamics (CFD) package of ANSYS Fluent (v14.5) based on its dynamic mesh feature, which naturally combines with multiphase flow models. This model is validated and adopted to simulate the dynamic responses of mold top surface to flow rate variations in the SEN subject to the upstream actuator position change. The complete model system is applied to investigate the effects of slide-gate dithering on transient single- and multi- phase flows in the caster mold. Mold sloshing is identified by both plant experiments and numerical simulations when the dithering frequency matches with the mold natural frequency determined by its geometry. Mechanism for the liquid steel flow variation to activate this standing wave (mold sloshing) is discussed. Multiphase flow pattern and top surface evolution under a low-frequency dithering trial is studied via numerical simulations. Mold level fluctuations are computed from the dithering simulations are compared in favor with the measurements. Up to this point, the model system has been demonstrated a powerful computational tool to resolve complicated multiphase flows during essentially transient events in continuous steel casting subjected to flow rate (both liquid steel and argon gas) variations