SURF

Aaron Kavaler

CFD Data Compression on High-Order Unstructured Meshes

As processing power available to high performance computers has increased, the overall performance has bottlenecked due to the limits of data transfer and storage. For small and medium sized computational fluid dynamics (CFD) researchers, the size of generated data is becoming problematic. In particular, CFD data faces a large resistance to general compression algorithms such as Lempel-Ziv encoding found in .zip files.

Applications in CFD including wind-turbine design and aeronautics demand accurate capture of fluid dynamics phenomena on complex geometries such as turbine blades or airplane wings. For these complex geometries, grid-like straight-edged meshes prove ineffective. Instead meshes which are both curved (referred to as “high-order”) and non-gridlike (referred to as “unstructured”) are gaining popularity. As such, research into a novel compression algorithm of CFD data on high-order unstructured meshes will help small scale CFD researchers efficiently store their data.

Message to Sponsor

I am really grateful to the Anselm fund, for sponsoring my summer research project. The stipend allowed me to dedicate my full attention to research. The resulting experience was unexpectedly enjoyable and satisfying -- nothing at all like taking classes and filling out homework. I was having doubts about whether to go to graduate school, since I didn't want to sign on for a repeat of my undergraduate career. Actually doing research reaffirmed my interest in graduate school, and as a direct result of the SURF program.
  • Major: Applied Mathematics
  • Sponsor: Anselm Fund
  • Mentor: Per-Olof Persson