We study the turbulent flow of air in contact with a hot, moving surface when subjected to various external disturbances. In particular, we are interested in changes due to different levels of surface roughness. How much do the results change due to a small deviation in the form of added surface roughness or other surface inhomogeneities? This project is the academic component of a joint project with an industrial partner. The results will be compared with experimental findings in "hot rolling mills." To this end, we will start with computational fluid dynamics (CFD) simulations using Ansys Academic Research CFD. One of the initial challenges of this project was securing suitable computational resources. Our Ansys license permits two parallel tasks, each of which can use up to 34 computer cores simultaneously. We are grateful for the opportunity to use the de.NBI Ccloud infrastructure for this two-year project. Therefore, we requested and obtained two virtual machines, each with 28 cores and 240 GB of RAM. Setting up the virtual machines was straightforward. Among other root disk images, a preconfigured browser-based remote desktop application (Guacamole with the Ubuntu Linux operating system) was offered. This made for a very user-friendly start. The Ansys software runs smoothly. However, the CFD simulations require significant computing power. Increasing the number of finite elements (e.g., by employing a higher grid density) easily results in computational times of a few days. Currently, we are optimizing simulation parameters to gain experience in this area. In any case, such large simulations would simply not be possible with the other computational hardware in our group.
