Speaker
Description
The increasing reliability of Computational Fluid Dynamics (CFD) simulations has established them as essential tools for aerodynamic analysis and wind-resistant design in many engineering fields. Simultaneous advancements in exascale computing have facilitated the development of large-scale models producing massive datasets. While offering significant advantages, storing and processing these massive datasets, particularly those produced by Large Eddy Simulations (LES), poses a marked challenge for conventional methods. With its ability to directly solve large-scale turbulent structures and model small-scale turbulence, LES simulations provide a highly accurate but computationally prohibitive representation of turbulence.
This work presents a novel compression method specifically designed to address this challenge. This approach focuses on critical flow region compression by employing Gaussian Fourier Features and the proposed Signed Distance Function (SDF) biased flow importance sampling (BiFIS), enabling accurate flow detail reproduction. Additionally, a new and efficient neural network architecture tailored for 3D and spatiotemporal compression efficiency is introduced to further enhance the compression process. The efficacy of this methodology is tested through a large-scale 3D LES simulation of a bridge deck resembling the Sunshine Skyway Bridge in Tampa, Florida, USA. Results reported in the extended version of the abstract will demonstrate the capabilities of the proposed compression methodology, which includes more efficient data storage, facilitates flow processing, and flow features visualization.
