Hybrid modelling of the continuous casting of steel

To prevent surface defects and enhance product quality, controlled cooling is essential. Ensuring the surface temperature remains above the “Leidenfrost-Temperature” is critical to form a stable vapor layer on the slab surface in order to maintain a consistent heat dissipation rate. This condition is influenced by several factors, including surface temperature and condition, water droplet size and velocity, spray properties, and water distribution, which varies depending on operating parameters. Moreover, sustainable management of cooling water and air reduces costs and minimizes the environmental impact of steel production.

To investigate the effects of nozzle types and operating parameters on cooling intensity, the “Nozzle Measuring Stand” (NMS) at the Chair of Ferrous Metallurgy was established. The setup allows the adjustment of the following parameters to study their influence on cooling performance:

• Water flow rate V(H2O) / water pressure p(H2O)
• Air flow rate V(AIR) / air pressure p(AIR)
• Sample velocity v (< 6 m/min)
• Surface start temperature TStart (<1200 °C)
• Distance nozzle to surface Nz
• In case of two nozzles: Distance nozzle to nozzle Nx

Key measurements from the Nozzle Measuring Stand include water distribution (WD) or water impact density (WID) and the heat transfer coefficient (HTC). The WD is determined using seven measuring grids, which collect water over a defined time period. A photo-optical analysis processes these measurements to create a 3D simulation of the spray pattern. Furthermore, it is also possible to measure the HTC with a moving pre-heated specimen. Through the heat drop, caused by the water spray, the surface temperature and the HTC of the sample can be calculated by an inverse heat conduction model.

Data obtained from these experiments lead to “data triples”, because the measured WD and HTC as well as the surface temperature values are corresponding. These datasets are incorporated into the in-house-developed simulation software called “m²CAST”. The aim of the research focuses on improving the simulation by several aspects, like the implementation of regression models as statistical prediction tools. These models forecast the time- and location-dependent HTC for an accurate solidification modeling.