URANS Simulation of Continuous Casting Mold Flow: Assessment of Revised Turbulence Models. Issue 4 (8th April 2015)

• Record Type: Journal Article
• Authors: Kratzsch, Christoph; Timmel, Klaus; Eckert, Sven; Schwarze, Rüdiger
• Is Part Of: Steel research international. Volume 86:Issue 4(2015:Apr.)
• Page Start: 400
• Page End: 410
• Publication Date: 2015-04-08
• Subjects: mini‐LIMMCAST -- mold flow -- OpenFOAM -- RANS -- unsteady -- URANS; Steel -- Periodicals; Steel -- Metallurgy -- Periodicals; 669.142
• DOI: 10.1002/srin.201400097
• Languages: English
• ISSNs: 1611-3683

Abstract

Abstract : A time resolving numerical model of single phase mold flows is formulated in the frame of Reynolds‐averaged Navier–Stokes (RANS) simulations. Three revised eddy viscosity turbulence models besides the often used Standard k – ϵ model for steady and unsteady RANS simulations are evaluated with the computational fluid dynamics (CFD) software tool OpenFOAM. The performance of the numerical model and the different turbulence models is tested with help of the mini‐LIMMCAST facility. Numerical data are compared with experimental results focusing especially on the jet flow region. The different turbulence models lead to slight deviations from the experimental data in the case of steady state calculation. For the time‐averaged unsteady RANS equations (URANS) results, a crucial influence by the discretization order is observed. In the case of unsteady results a mix between first and higher order accuracy schemes is necessary for representative results. The use of the Standard k – ϵ model and the RLZ k – ϵ model can lead to unstable behavior in the case of URANS simulations. The study concludes, that in the case of qualitative, temporally resolved informations, and temporally averaged results, an unsteady RANS simulation can keep up with a large eddy simulations (LES). Abstract : A time resolving numerical model of a single‐phase mold is formulated in the frame of Reynolds‐averaged Navier–Stokes (RANS) simulations. Thereby the performance of various turbulence models is evaluated focusing the jet flow. The results show, that it is necessary to have higher order discretization with either the RNG k – ϵ model or the k – ω SST model.