Deep reinforcement learning for active flow control around a three-dimensional flow-separated wing at Re = 1, 000

Abstract

This study explores the use of deep reinforcement learning (DRL) for active flow control (AFC) to reduce flow separation on wings at high angles of attack. Concretely, here the DRL agent controls the flow over the three-dimensional NACA0012 wing section at the Reynolds number Re = 1, 000 and angle of attack AoA= 20◦, autonomously identifying optimal control actions through real-time flow data and a reward function focused on improving aerodynamic performance. The framework integrates the GPUaccelerated computational fluid dynamics (CFD) solver SOD2D with the TF-Agents DRL library via a Redis in-memory database, enabling rapid training. This work builds on previous DRL flow-control studies, demonstrating DRL’s potential to address complex aerodynamic challenges and push the boundaries of traditional AFC methods.

This research has received financial support from the Ministerio de Ciencia e Innovacion of Spain (PID2023-150408OB-C21 and PID2023-150408OB-C22). Simulations were conducted with the assistance of the Red Española de Supercomputación (RES) and the EuroHPC JU, who granted us computational resources at the HPC facilities of MareNostrum V at Barcelona Supercomputing Cente (IM-2024-2-0004 and EHPC-REG-2024R01- 038, respectively). Authors also extend their gratitute to the Agència de Gestió d’Ajuts Universitaris i de Recerca (AGAUR) for supporting the research group Large-scale Computational Fluid Dynamics (2021 SGR 00902) and the Turbulence and Aerodynamics Research Group (2021 SGR 01051). Ricard Montalà express also gratitude to AGAUR for awarding the FI-SDUR grant (2022 FISDU 00066), which supports his doctoral studies. Finally, Ricardo Vinuesa acknowledges financial support from ERC grant no.2021-CoG101043998, DEEPCONTROL.

Peer Reviewed

Postprint (author's final draft)