CFD Benchmark for a Single Strand Tundish (Part II) H.-J. Odenthal 1) * , M. Javurek 2) , M. Kirschen 3) , and N. Vogl 1) 1) SMS Siemag AG, R&D Central Department, Du ¨sseldorf, Germany 2) Johannes Kepler University Linz, Institute for Fluid Dynamics and Heat Transfer, Linz, Austria 3) RHI AG, Corporate R&D, Leoben, Austria *Corresponding author, e-mail: hans-juergen.odenthal@sms-siemag.com In a comparative benchmark, nine participants of the German Steel Institute VDEh working group ‘‘Fluid Mechanics and Fluid Simulation’’ studied the melt flow in a 16-t single-strand tundish. Starting with a steady-state simulation of the melt flow, the transient flow behaviour was simulated for an idealized ladle change involving a sudden jump in temperature and concentration. In addition, the separation of non-metallic particles to the melt surface was examined. No guidelines and limitations were made regarding the simulation strategy. The predicted flow profiles, temperature and concentration distributions coincide with each other within a good approximation. Systematic differences in the transient temperature and turbulence fields are explained by the selection of the boundary condition at the free surface. All CFD programs reproduce the fundamental flow structure with a good degree of accuracy. The separation rate for non-metallic particles calculated on the basis of the Lagrange Method are greater than would be expected according to theory and measurement results obtained on the water model. Keywords: computational fluid dynamics, tundish flow, CFD benchmark, turbulence model, CFD boundary condition Submitted on 23 November 2009, accepted on 16 April 2010 Introduction and Problem Definition This paper is an extension of the CFD benchmark study on the steady-state flow in a water model (scale 1:1.7) of a 16-t single-strand tundish [9]. The water flow was examined by ten participants from four universities and six industrial firms; the latter include a CFD software manufacturer. The numerical results were compared with high-accurate laser Doppler anemometer measurements for the water model and the correlation was very good. In the present part of the CFD benchmark study, the numerical models are applied to the original full-scale tundish filled with liquid metal. In addition to the liquid melt flow, the melt temperature T and the carbon concentration c c should be calculated during an idealized ladle change. The tundish geometry with the shroud tube (index sh) and the submerged entry nozzle (index SEN) are described in the first part of the publication [9]. For the second benchmark, all material data of the stainless steel X5CrNi18-10 (material no. 1.4301) as well as the fluidic boundary conditions for the steady-state casting sequence are fixed and shown in Table 1. As in the first part of the benchmark, no limitations are made regarding the numerical grid, e.g. the grid type, or the number of grid cells. All numerical constraints, e.g. solution strategy, turbulence model including respective constants, consideration of the covering slag layer (single-phase or multi-phase, e.g. steel/ slag), relaxation factors, and other model constants (Pr tur , Sc tur ), could be set at the participants own discretion. The extension of the tundish domain (shroud, SEN), the geometrical symmetry - the tundish has a mirror plane, thus only one half could be modelled using a symmetry boundary condition - as well as the type of boundary condition at the melt interface could be chosen. This basic strategy was discussed in detail and decided by all members of the working group prior to the benchmark, and should accentuate the competence and experience of the partic- ipants. However, no experimental reference data for the velocity, turbulence, temperature, or concentration distri- bution of the tundish melt flow are available because an adequate measurement technique does still not exist. For this reason, the current paper must be seen as modelling study and can support other CFD users. All simulations which are to be carried out in the second part of the benchmark relate to steady-state casting conditions with the mass flow _ m sh ¼ _ m SEN , though with T sh 6¼ T SEN and c c,sh 6¼ c c,SEN . Three cases are to be dealt with by all participants. Figure 1 and Table 1 indicate the boundary conditions and material data required for this. Case 1 - Steady-state simulation. For the first section of the tasks, the steady-state flow and turbulence structure in the continuous-casting tundish are to be calculated for a case where the melt from the ladle enters the tundish with _ m sh ¼ 38 kg=s and T sh ¼ 15008C. The heat losses via the bottom and side-walls of the tundish as well as via the covering flux (slag) have been previously estimated at _ q ¼7:6 kW=m 2 and _ q sl ¼22 kW=m 2 . Heat losses via the shroud tube and submerged entry nozzle are neglected. As a result of the heat losses, a lower mean temperature is obtained in the tundish melt. From the energy balance _ m sh;SEN cðT SEN T sh Þ¼ X _ q i A i (1) the temperature difference obtained between the outlet and inlet of the tundish becomes DT SEN-sh ¼4.64 K. In DOI: 10.1002/srin.201000079 steel research int. 81 (2010) No. 7 www.steelresearch-journal.com ß 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 529