ERCOFTAC Bulletin Archive

Controlling bluff body flows with deep-reinforcement- learning

Abstract

Deep-reinforcement-learning (DRL) agents were trained in situ to drive trailing-edge pulsed jets on a square-back road-vehicle model at Re ≈ 105, achieving up to 10 % base-pressure drag reduction within forty 20s episodes of wind-tunnel time.

Keywords

bluff body, flow control

How to Cite

APA: E. Amico, J. Serpieri, G. Iuso and G. Cafiero. (2025). Controlling bluff body flows with deep-reinforcement- learning. ERCOFTAC Bulletin, 144.