Towards Active Flow Control Strategies Through Deep Reinforcement Learning

TL;DR

A deep reinforcement learning (DRL) framework for active flow control (AFC) to reduce drag in aerodynamic bodies using an in-memory database for efficient communication between CFD solver and DRL model.

Abstract

This paper presents a deep reinforcement learning (DRL) framework for active flow control (AFC) to reduce drag in aerodynamic bodies. Tested on a 3D cylinder at Re = 100, the DRL approach achieved a 9.32% drag reduction and a 78.4% decrease in lift oscillations by learning advanced actuation strategies. The methodology integrates a CFD solver with a DRL model using an in-memory database for efficient communication between

Figures (7)

• Figure 1: DRL-CFD setup
• Figure 2: Case configuration
• Figure 3: Evolution of the averaged reward $\overline{r_i}$ across the 40 pseudo-environments during the training
• Figure 4: Lift $C_l$ (a) and drag $C_d$ (b) coefficients before and after the DRL control is applied
• Figure 5: Applied mass flow rate $Q$