LARGE EDDY SIMULATIONS OF TRANSIENT TURBULENT FLOW DURING CONTINUOUS SLAB CASTING OF STEEL

Abstract

The quality of continuous cast steel is greatly affected by the turbulent flow in the mold region, especially for transient operation and transport of inclusion particles. During the continuous casting process, shown in Figure 1, the superheated molten steel flows into the mold region from the tundish through the nozzle ports. The molten steel freezes against the water-cooled mold walls to form solidified slab shells, which are continuously withdrawn from the bottom at the casting speed. The jets entering the mold region carry inclusion particles (e.g. alumina) and argon bubbles, which will either be safely transported to the top surface and removed by the slag layer or get entrapped in the final product, resulting in costly defects (e.g. internal cracks, blisters etc.). The flow in the mold is turbulent with Reynolds numbers in excess of 10 5 (based on the nozzle port hydraulic diameter). Plant observations found that the transient nature of this turbulent flow greatly influences the transport of the inclusions and bubbles, causing intermittent defects. This study is part of a larger ongoing research project to investigate the transient structures of this mold flow with an objective of minimizing the defects. Several previous studies have used Reynolds averaged turbulence models (mainly k-e model) [1] to understand this flow. However, the k-e model only predicts the timeaveraged velocities and cannot predict the detailed turbulent dynamics. Large Eddy Simulation (LES) is a more realistic and accurate method for resolving the evolution and dynamics of the large-scale turbulence structures, which are crucial to estimating heat, mass and momentum transport, and transport of inclusions. LES has been applied in many previous studies to simulate model turbulent flows [2, 3]. The application of LES to the present turbulent flow, however, leads to many challenges, including the prescription of inlet conditions, resolution of velocity and thermal boundary layers, the moving solidifying front and the long term transients. Thus the simulations require large computer memory and CPU time. Because of nearly equal kinematic viscosities of the molten steel and water, flow in the steel casting mold region has been studied using scaled water models, which are easier to operate and allow flow visualization. Recently, we applied Particle Image Velocimetry (PIV), to study the flow in a 0.4 scaled water model [4]. This study provides data to validate the numerical code. The water model differs from the real caster. First the side walls, which represent the moving solidifying shell, are non-porous and stationary. Further, the water model has a flat bottom with outlet ports to represent the tapering molten pool. This study presents results from three simulations. First, the LES approach is validated by comparing its results with the PIV data on the 0.4-scale water model. LES was then employed to simulate the flow in a full-scale water model including the nozzle and the mold. Finally, a simulation of the turbulent flow and inclusion transport in the full-scale