12.09.2024 I If you missed Prof. Brians Vermeires from Michigan University Sci-Fi-Turbo Seminar on Optimizing Chaos: Aerodynamic Design using High-Fidelity Scale Resolving Simulations, it it now online available! 

Current industry-standard aerodynamic shape optimization is performed using a combination of Reynolds Averaged Navier-Stokes (RANS) solvers and adjoint-based optimization. However, despite decades of development RANS is often deficient, particularly for separated and transitional flows. High-fidelity scale-resolving techniques, such as Large Eddy Simulation (LES) and Direct Numerical Simulation (DNS), have been demonstrably more accurate in these flow regimes. Over the past decade enabling technologies for efficient high-fidelity simulations, including high-order spatial discretizations, temporal discretizations, and many-core hardware architectures, have significantly reduced their computational cost. However, significantly less attention has been dedicated to the development of suitable optimization frameworks for LES/DNS. This talk will focus on two recently proposed optimization frameworks for LES/DNS. The first is a gradient-based approach, which uses a combination of reduced order modelling, least squares shadowing, and the adjoint. The second is a gradient-free approach using Mesh Adaptive Direct Search (MADS). It will be demonstrated that both of these frameworks are suitable for the fundamental chaotic behavior of scale resolving simulations. The utility of these frameworks will then be demonstrated for general chaotic systems, aerodynamic optimization, and aeroacoustic optimization including low-pressure turbine cascades.

Click here for the Sci-Fi-Turbo Seminar Optimizing Chaos: Aerodynamic Design using High-Fidelity Scale Resolving Simulations.

________________________________________________________________________________________________________________________________