English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Deep artificial neural networks (ANNs) used together with deep reinforcementlearning (DRL) are receiving growing attention due to their capabilities tocontrol complex problems. This technique has been recently used to solveproblems related to flow control. In this work, an ANN trained through a DRLagent is used to perform active flow control. Two-dimensional simulations ofthe flow around a cylinder are conducted and an active control based on twojets located on the walls of the cylinder is considered. By gatheringinformation from the flow surrounding the cylinder, the ANN agent is able tolearn effective control strategies for the jets, leading to a significant dragreduction. In the present work, a Reynolds-number range beyond those previouslyconsidered is studied and compared with results obtained using classicalflow-control methods. Significantly different nature in the control strategiesis identified by the DRL as the Reynolds number Re increases. For Re <= 1000the classical control strategy based on an opposition control relative to thewake oscillation is obtained. For Re = 2000 the new strategy consists of anenergisation of the boundary layers and the separation area, which modulate theflow separation and reduce drag in a fashion similar to that of the dragcrisis, through a high frequency actuation. A cross-application of agents isperformed for a flow at Re = 2000, obtaining similar results in terms of dragreduction with the agents trained at Re = 1000 and 2000. The fact that twodifferent strategies yield the same performance make us question whether thisReynolds number regime (Re = 2000) belongs to a transition towards anature-different flow which would only admit a high-frequency actuationstrategy to obtain drag reduction. This finding allows the application of ANNstrained at lower Reynolds numbers but comparable in nature, savingcomputational resources.

Deep reinforcement learning for flow control exploits different physics for increasing Reynolds-number regimes