Physics-informed Neural-operator Predictive Control for Drag Reduction in Turbulent Flows
This work addresses the problem of expensive turbulence control simulations for fluid dynamics applications, offering a more efficient solution with significant performance gains.
The paper tackles the challenge of efficiently modeling and controlling turbulent flows for drag reduction by proposing a deep reinforcement learning framework with Physics Informed Neural Operators, achieving a 39.0% drag reduction at a Reynolds number of 15,000 and outperforming prior methods by over 32%.
Assessing turbulence control effects for wall friction numerically is a significant challenge since it requires expensive simulations of turbulent fluid dynamics. We instead propose an efficient deep reinforcement learning (RL) framework for modeling and control of turbulent flows. It is model-based RL for predictive control (PC), where both the policy and the observer models for turbulence control are learned jointly using Physics Informed Neural Operators (PINO), which are discretization invariant and can capture fine scales in turbulent flows accurately. Our PINO-PC outperforms prior model-free reinforcement learning methods in various challenging scenarios where the flows are of high Reynolds numbers and unseen, i.e., not provided during model training. We find that PINO-PC achieves a drag reduction of 39.0\% under a bulk-velocity Reynolds number of 15,000, outperforming previous fluid control methods by more than 32\%.