TY - JOUR
T1 - Transient turbulent flow in a liquid-metal model of continuous casting, including comparison of six different methods
AU - Chaudhary, R.
AU - Ji, C.
AU - Thomas, B. G.
AU - Vanka, S. P.
N1 - Funding Information:
The authors are grateful to K. Timmel, S. Eckert, and G. Gerbeth from MHD Department, Forschungs-zentrum Dresden-Rossendorf (Dresden, Germany) for providing the velocity measurement data in the GaInSn model. This work was supported by the Continuous Casting Consortium, Department of Mechanical Science & Engineering, University of Illinois at Urbana-Champaign, IL. ANSYS, Inc. is acknowledged for providing FLUENT. Also, we would like to thank Silky Arora for helping us with data extraction codes for the POD analysis.
PY - 2011/10
Y1 - 2011/10
N2 - Computational modeling is an important tool to understand and stabilize transient turbulent fluid flow in the continuous casting of steel to minimize defects. The current work combines the predictions of two steady Reynolds-averaged Navier-Stokes (RANS) models, a "filtered" unsteady RANS model, and two large eddy simulation (LES) models with ultrasonic Doppler velocimetry (UDV) measurements in a small-scale liquid GaInSn model of the continuous casting mold region fed by a bifurcated well-bottom nozzle with horizontal ports. Both mean and transient features of the turbulent flow are investigated. LES outperformed all models while matching the measurements, except in locations where measurement problems are suspected. The LES model also captured high-frequency fluctuations, which the measurements could not detect. Steady RANS models were the least accurate methods. Turbulent velocity variation frequencies and energies decreased with distance from the nozzle port regions. Proper orthogonal decomposition analysis, instantaneous velocity patterns, and Reynolds stresses reveal that velocity fluctuations and flow structures associated with the alternating-direction swirl in the nozzle bottom lead to a wobbling jet exiting the ports into the mold. These turbulent flow structures are responsible for patterns observed in both the time average flow and the statistics of their fluctuations.
AB - Computational modeling is an important tool to understand and stabilize transient turbulent fluid flow in the continuous casting of steel to minimize defects. The current work combines the predictions of two steady Reynolds-averaged Navier-Stokes (RANS) models, a "filtered" unsteady RANS model, and two large eddy simulation (LES) models with ultrasonic Doppler velocimetry (UDV) measurements in a small-scale liquid GaInSn model of the continuous casting mold region fed by a bifurcated well-bottom nozzle with horizontal ports. Both mean and transient features of the turbulent flow are investigated. LES outperformed all models while matching the measurements, except in locations where measurement problems are suspected. The LES model also captured high-frequency fluctuations, which the measurements could not detect. Steady RANS models were the least accurate methods. Turbulent velocity variation frequencies and energies decreased with distance from the nozzle port regions. Proper orthogonal decomposition analysis, instantaneous velocity patterns, and Reynolds stresses reveal that velocity fluctuations and flow structures associated with the alternating-direction swirl in the nozzle bottom lead to a wobbling jet exiting the ports into the mold. These turbulent flow structures are responsible for patterns observed in both the time average flow and the statistics of their fluctuations.
UR - http://www.scopus.com/inward/record.url?scp=80054942802&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=80054942802&partnerID=8YFLogxK
U2 - 10.1007/s11663-011-9526-1
DO - 10.1007/s11663-011-9526-1
M3 - Article
AN - SCOPUS:80054942802
SN - 1073-5615
VL - 42
SP - 987
EP - 1007
JO - Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science
JF - Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science
IS - 5
ER -