EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Multi-agent Reinforcement Learning for the Control of Three-Dimensional Rayleigh–Bénard Convection

Joel Vasanth, Jean Rabault, Francisco Alcántara-Ávila, Mikael Mortensen, Ricardo Vinuesa

Flow, Turbulence and Combustion · 2025-09 · 115 (3) · 1319-1355

DOI: 10.1007/s10494-024-00619-2 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Flow, Turbulence and Combustion, online erschienen am 14.12.2024. Verfasst von Joel Vasanth und 4 weiteren Autor:innen. Beteiligte Einrichtung: FLOW, Engineering Mechanics, KTH Royal Institute of Technology, Stockholm, Sweden. Die Arbeit wurde bisher 16-mal zitiert (Crossref) und nennt 70 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: European Research Council, Royal Institute of Technology.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep reinforcement learning (DRL) has found application in numerous use-cases pertaining to flow control. Multi-agent RL (MARL), a variant of DRL, has shown to be more effective than single-agent RL in controlling flows exhibiting locality and translational invariance. We present, for the first time, an implementation of MARL-based control of three-dimensional Rayleigh–Bénard convection (RBC). Control is executed by modifying the temperature distribution along the bottom wall divided into multiple control segments, each of which acts as an independent agent. Two regimes of RBC are considered at Rayleigh numbers $$\textrm{Ra}=500$$ Ra = 500 and 750. Evaluation of the learned control policy reveals a reduction in convection intensity by $$23.5\%$$ 23.5 % and $$8.7\%$$ 8.7 % at $$\textrm{Ra}=500$$ Ra = 500 and 750, respectively. The MARL controller converts irregularly shaped convective patterns to regular straight rolls with lower convection that resemble flow in a relatively more stable regime. We draw comparisons with proportional control at both $$\textrm{Ra}$$ Ra and show that MARL is able to outperform the proportional controller. The learned control strategy is complex, featuring different non-linear segment-wise actuator delays and actuation magnitudes. We also perform successful evaluations on a larger domain than used for training, demonstrating that the invariant property of MARL allows direct transfer of the learnt policy.

Schlagwörter

Stichwörter der Autor:innen

Active flow controlMachine learningMulti-agent reinforcement learningRayleigh–Bénard convectionReinforcement learning

Autor:innen und Institutionen

  • FLOW, Engineering Mechanics, KTH Royal Institute of Technology, Stockholm, Sweden.
  • Oslo, Norway.
  • Department of Mathematics, University of Oslo, Oslo, Norway.

Förderung

  • European Research Council
  • Royal Institute of Technology

Zitieren

APA

Vasanth, J., Rabault, J., Alcántara-Ávila, F., Mortensen, M., & Vinuesa, R. (2025). Multi-agent Reinforcement Learning for the Control of Three-Dimensional Rayleigh–Bénard Convection. Flow, Turbulence and Combustion, 115(3), 1319-1355. https://doi.org/10.1007/s10494-024-00619-2

Vancouver

Vasanth J, Rabault J, Alcántara-Ávila F, Mortensen M, Vinuesa R. Multi-agent Reinforcement Learning for the Control of Three-Dimensional Rayleigh–Bénard Convection. Flow, Turbulence and Combustion. 2025;115(3):1319-1355. doi:10.1007/s10494-024-00619-2

BibTeX
@article{Vasanth2025,
  author = {Vasanth, Joel and Rabault, Jean and Alcántara-Ávila, Francisco and Mortensen, Mikael and Vinuesa, Ricardo},
  title = {Multi-agent Reinforcement Learning for the Control of Three-Dimensional Rayleigh–Bénard Convection},
  journal = {Flow, Turbulence and Combustion},
  year = {2025},
  volume = {115},
  number = {3},
  pages = {1319--1355},
  doi = {10.1007/s10494-024-00619-2}
}
Kostenlos starten →