CONTENT A Study of Particle-Laden Flow of a Tundish Water-Model Dipl.-Ing. (FH) Eike Runschke M. Sc. Gerold Haskamp Dr. rer. nat. Zeljko Cancarevic MBA Dr.-Ing. Henning Schliephake
Content CONTENT Company Market and Products Technology at Georgsmarienhütte GmbH
History COMPANY 1856 Founded as an iron production plant located in the South of Osnabrück. Its name derived from the last rules of the Hannover dynasty, King George V. and Queen Marie. 1923 Klöckner acquires Georgsmarienhütte. 1993 Dr. Großmann and Drueker & Co. GmbH buy and incorporate GMH. 1994 The blast furnace-converter was replaced with a DC electric arc furnace. 1997 Georgsmarienhütte Holding GmbH was created. 2006 A new walking-beam furnace replaces both of the 40+ year old walking-hearth furnaces. 2007 A second ladle furnace is installed to expand secondary metallurgy. 2009 Various modernization of furnace, rolling mill and finishing shop.
Company COMPANY Manufacturer of quality and engineering steels Market leader in Germany Among Europe s top manufacturers Key Data 2012: 644 mil. Euro Turnover 644.000 to rolled products 1,334 employees
International offices COMPANY
Our Vision COMPANY BEING PROMPT Melt and treat in the morning Cast at noon Roll in the evening Finish over night MEANS REACTING WITH MAXIMUM PRECISION IN THE SHORTEST TIME POSSIBLE. CHALLENGE US! Ship the next day!
Market Our steel drives you forward MARKET Powertrain, engine, transmission, steering and chassis components of cars and trucks End use of 80% of our products in the automotive industry
References MARKET
DC electric arc furnace 130 MW TECHNOLOGY
TECHNOLOGY
at Georgsmarienhütte GmbH Wide field of application due to different production processes since August 2011 Until now: Heat transfer, air/water flow as well as particle-laden multi-phase VOF Hardware & OS: Workstation: Windows 7 64bit, 8 cores, 3.6 GHz, 48GB RAM Workstation: Windows 7 64bit, 4 cores, 4.9 GHz, 32GB RAM Cluster: Windows Server 2008 R2, 192 cores, 2.4 GHz, 24GB RAM / node Cluster: Linux Red Hat, 192 cores, 3.3 GHz, 64GB RAM / node HPC Cluster Manager
Air flow Analysis of an axial fan Steady and unsteady, rigid body motion Comparison with measured data
Air flow Simulation vs. measured data 10 [m/s] 15 [m/s] 20 [m/s] 25 [m/s] [ ] 30 [m/s]
Multiphase VOF Tundish Water Model vs. Particle-laden flow of 6-strand tundish Comparison of simulation with scaled water model Simulation: VOF (water and air), non-metallic inclusions = lagrangian particles of different size and density Water model: 1:4, 20 C, Froude-similarity, particles = glass hollow spheres Investigated time = 20 minutes
Water Model ball valve shroud, adjustable in height 1:4 scaled tundish model made of sheet acrylic glass filter, connected to submerged nozzle movable support frame
Water Model usb-hub power distribution surge drum with pump frequency controlled pump flow sensor
Water Model injection device for dye and particles point of injection, in front of the shroud connection to in-tank pump
Water Model Example of the distribution of dye Example of the size of fractionated particles
Water Model Calibration nominal effective error a known mass of glass-sheres were put into the filters and the system was flushed analysis of difference values analysis of influencing factors reduction of error sources fractionation of the glass-spheres sealing the filters with Vaseline increase of inserted particles
Water Model Calibration error sources leaking filters accuracy of laboratory scales mass loss of filter element after drying @ 100 C residue of Vaseline mass of a single glas-sphere unattended flow rate of the strands ± 0.0001 g - 0.0062 g + 0.0041 g + + U = uncertainty K = correction factor K = 2 assumption that 95% of measured data is correct
Water Model influence of sealant at the filter elements and increace of mass of particles no sealant at filter element
-Simulation k- SST Mesh: Trimmer, Prism Layer, 3 mio. volume cells Fluid: Eulerian Multiphase, Particle: Lagrangian Multiphase Particles: 100 to 200µm, density 0.65 to 1.24 g/cm³ Strand 1 to 6: mass flow inlet with identical flow Solution time 0 300s: creation of fluid flow Solution time 300s, 600s, 900s: injection of 1000 particles of every size Solution time 1500s: end of simulation Particles that left the system are counted
-Simulation Time step = 0.025s point of injection 8 inner iterations definition of injected particles velocity of particles @ point of injection
-Simulation definition of water and air definition of mass flow @ outlet definition of particles definition of mass flow @ inlet
-Simulation Reports of Lagrangian Phases in order to count all particles within the system Reports of Particle Tracks in order to detect all particles which left the system These Reports will later be used in order to count particles at the six outlets
-Simulation Sum of Report: Element Count The previously created reports are used in order to count particles at the six outlets
-Simulation vs. Water Model Solution Time = 1500s Density [g/cm³] Diameter [µm] 0.65 200 0.70 150 0.75 100 1.24 110
-Simulation vs. Water Model same pattern big difference of relative value Ø = 110µm = 1.24 g/cm³
-Simulation vs. Water Model different pattern huge difference of relative value Ø = 200µm = 0.65 g/cm³
Conclusion Simple equipment can generate huge knowledge Density and diameter of glass-spheres are hard to determine Glass-spheres are not homogeneous and tend to break to pieces Significant difference of experimental data compared to : huge influence of density and diameter of the particles : sensitivity analysis is needed : fine-tuning is needed in order to a closer match to the water model
Outlook Tundish vs. Water Model better measurement technique study on built-in components study on casting-speed fine-tuning of -Model comparison to real tundish improvement of real tundish
Outlook Tundish vs. Water Model Ladle
Outlook Tundish vs. Water Model Ladle Ingot Casting
Outlook Tundish vs. Water Model Ladle Ingot Casting Continuous Casting Source: Brian G. Thomas, University of Illinois at Urbana-Champaign