Mirko Javurek

Numeric simulations of a liquid metal model of a bloom caster under the effect of rotary electromagnetic stirring

At the voestalpine Stahl Donawitz GmbH the continuous casting of round steel blooms is commonly supported by electromagnetically induced stirring of the liquid steel flow. A number of beneficial effects are attributed to electromagnetic stirring in the mould region (M-EMS), e.g. the enhanced transition from columnar to equiaxed solidification, the homogenization of the liquid steel flow or the reduction of surface and subsurface defects. Although the positive effects of M-EMS can be seen on the blooms (e.g. in etchings), the link between electromagnetic stirring of the steel melt and the quality of the solidified bloom is not sufficiently understood. Theoretical considerations are often limited to general cases and their results are therefore not directly applicable to real continuous casting geometries. On the other hand, plant measurements can only be performed to a limited extent due to the harsh conditions and other restrictions (e.g. safety regulations). In this work an alternative approach is used to investigate the steel flow in a round bloom caster under the influence of M-EMS. In a 1:3 scale Perspex model of a round bloom strand, measurements of the flow under the influence of a rotating magnetic field can be conducted. These measurements provide a validation benchmark for the numeric simulations. A numeric model of the before mentioned 1:3 scale model is implemented, encompassing the strand, the submerged entry nozzle as well as the M-EMS device. In the modelling approach, the bidirectional coupling between liquid steel flow and the electromagnetic field/forces has to be considered because otherwise the resulting tangential velocities will be overestimated. With the validated modelling approach, simulations of real casting machines can then be conducted, stirring parameter influences can be shown and conclusions for the real casting process can be drawn.

Numerical, Experimental and Analytical Investigation of the Mass Outflow From a Pickling Tank

The process of pickling is an important intermediate step in the production line of steel processing. The strip surface is cleaned from grease and scales before further processing by immersion into an acid bath. Problems arising at higher process speeds with increasing inclination of the free surface are reduced strip immersion length and increased mass outflow. In this paper a differential equation is derived describing the influence of the bath depth on the local surface inclination for a simplified two-dimensional case. Since it can only be solved analytically for trivial boundary conditions a numerical solution has been computed giving an estimation for the order of growth of the bath inclination and mass outflow with the strip velocity. Further, a series of CFD simulations of the complete three-dimensional geometry at different strip velocities have been carried out calibrating the formulas of mass outflow. In the course of the CFD simulations the deformation of the free surface was calculated by a VOF model with explicit reconstruction of the interface. A standard κ–e turbulence model was applied and special considerations have been made regarding the boundary conditions. Finally the resulting formula has been verified making use of data from a small scale model. It was found that the overflow at the far end of the tank is the dominant mass transfer mechanism at process velocities of the current generation of pickling tanks. Still, due to the superior order of growth, mass drag-out via the upper side of the strip becomes important for process velocities of 8 to 10 m/s. The good accordance of the analytical solution, CFD simulation and experiment indicate that the formula derived in the first part of the paper is a good estimation for the mass outflow from the pickling tank, hence making time and resource consuming CFD simulations obsolete for the design layout. Further the validity of geometrically non similar small scale models could be showed.Copyright

Driftflussmodell für unstrukturierte Netze mit Partikelagglomeration

Ein Driftflussmodell (oder auch algebraisches Mischungsmodell genannt) kann in Mehrphasen-CFD-Simulationen eingesetzt werden, wenn die Sekundarphasen Partikel, Tropfen oder Blasen sind. Das Geschwindigkeitsfeld der Sekund arphase wird berechnet, indem die Driftgeschwindigkeit (die von der Partikelgrose abhangt) zum Geschwindigkeitsfeld der Primarphase addiert wird. Fur jede Sekundarphase wird nur eine zusatzliche Transportgleichung gelost. Deswegen ist das Driftlussmodell im Vergleich mit Euler-Euler-Modellen ein sehr robustes, stabiles und schnelles Modell um Mehrphasenstromungen zu berechnen. Bei Implementierungen des Driftlussmodells fur unstrukturierte Gitter konnen Probleme an Wanden auftreten, an denen die Partikel das Rechengebiet nicht verlassen konnen und sich dort ansammeln (Agglomeration). Diese Probleme und eine Losungsidee sind in dieser Arbeit geschildert.