Pasar al contenido principal

CUANDO HPC SE ENCUENTRA CON DRL Y CFD: FLUJOS DE TRABAJO NOVEDOSOS PARA SOLUCIONADORES DE FLUJO Y MODELOS BASADOS EN IA.

PID2023-150408OB-C21

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal para Impulsar la Investigación Científico-Técnica y su Transferencia
Subprograma Subprograma Estatal de Generación de Conocimiento
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2023
Plan Estatal 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario BARCELONA SUPERCOMPUTING CENTER CENTRO NACIONAL DE SUPERCOMPUTACION

Resultados relacionados

Resultados totales (Incluyendo duplicados): 4
Encontrada(s) 1 página(s)

UPCommons. Portal del coneixement obert de la UPC
Publicación Contribución de congreso (ConferenceOutput)

Discovering flow separation control strategies in 3d wings via deep reinforcement learning

  • Montalà Sales, Ricard|||0009-0000-7911-7102
  • Font García, Bernat
  • Suárez Morales, Pol
  • Rabault, Jean
  • Lehmkuhl Barba, Oriol|||0000-0002-2670-1871
  • Vinuesa Motilva, Ricardo
  • Rodríguez Pérez, Ivette María|||0000-0002-3749-277X
In this work, deep reinforcement learning (DRL) is applied to active flow control (AFC) over a threedimensional SD7003 wing at a Reynolds number of Re = 60,000 and angle of attack of AoA = 14◦. In the uncontrolled baseline case, the flow exhibits massive separation and a fully turbulent wake. Using a GPU-accelerated CFD solver and multi-agent training, DRL discovers control strategies that enhance lift (79%), reduce drag (65%), and improve aerodynamic efficiency (408%). Flow visualizations confirm reattachment of the separated shear layer, demonstrating the potential of DRL for complex and turbulent flows., This research was partially funded by the Ministerio de Ciencia e Innovación of Spain (PID2023- 150408OB-C21/C22). Simulations were supported by the Red Española de Supercomputación (IM-2024-2-0004 and IM-2024-3-0005) and EuroHPC JU (EHPCREG-2024R01-038), granting access to MareNostrum V at the Barcelona Supercomputing Center. We also acknowledge AGAUR for supporting the LS/CFD (2021 SGR 00902) and TUAREG (2021 SGR 01051) research groups. R. M. thanks AGAUR for the FISDUR grant (2022 FISDU 00066) and R. V. acknowledges financial support from ERC grant no. 2021- CoG-101043998, DEEPCONTROL., Peer Reviewed, 9 - Indústria, Innovació i Infraestructura, 13 - Acció per al Clima
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

On the integration of geometry agnostic variational-autoencoders into large-scale SVD based models

  • Eiximeno Franch, Benet
  • Miró Jané, Arnau|||0000-0002-2772-6050
  • Kutz, Nathan
  • Rodríguez Pérez, Ivette María|||0000-0002-3749-277X
  • Lehmkuhl, Oriol
This manuscript introduces GAVI (Geometry Agnostic Variational-autoencoder Integration), a scalable technique for model order reduction of terabyte-level state spaces. Such methodology is agnostic to the spatial location of the data points, hence, it can handle any complex geometry regardless of the grid type in which it is represented. The dimensionality reduction of GAVI is performed on two separate steps. First, a parallel QR factorization is applied to the snapshot matrix to decompose, then a variational autoencoder learns a latent representation of the reduced R matrix. The final step can be done at laptop level regardless of the size of the original data. The methodology is used to compress the streamwise velocity component of the flow around a circular cylinder at 𝑅𝑒𝐷 = 𝑈∞𝐷∕𝜈 = 104 , the pressure coefficient of the flow around the Windsor body at 𝑅𝑒𝐿 = 𝑈∞𝐿∕𝜈 = 2.9 × 106 and the streamwise velocity of the realistic urban flow in the Zona Universitària neighborhood located in Barcelona. The latter example is the most demanding one as the 1032 snapshots, which are represented on an unstructured grid of 335 million points, have a total size of 2.05 Tb. GAVI can compress the full dataset into 6 latent vectors that recover up to the 95.29% of the energy., The research leading to this work has been partially funded by the project TIFON with reference PLEC2023-010251/ AEI/10.13039/ 501100011033 and by the Ministerio de Ciencia e Innovación of Spain (PID2023-150408OB-C21/C22). B. Eiximeno’s work was funded by a contract from the Subprograma de Ayudas Predoctorales given by the Ministerio de Ciencia e Innovación (PRE2021-096927). Oriol Lehmkuhl has been partially supported by a Ramon y Cajal postdoctoral contract (Ref: RYC2018-025949-I). The authors acknowledge the support of Departament de Recerca i Universitats de la Generalitat de Catalunya to the Research Group Large-scale Computational Fluid Dynamics (Code: 2021 SGR 00902) and the Turbulence and Aerodynamics Research Group (Code: 2021 SGR 01051). Nathan Kutz acknowledges the NSF AI Institute in Dynamics System., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Capítulo o parte de un libro (BookPart)

Towards active flow control strategies through deep reinforcement learning

  • Montalà Sales, Ricard|||0009-0000-7911-7102
  • Font García, Bernat
  • Suárez Morales, Pol
  • Rabault, Jean
  • Lehmkuhl Barba, Oriol|||0000-0002-2670-1871
  • Vinuesa Motilva, Ricardo
  • Rodríguez Pérez, Ivette María|||0000-0002-3749-277X
This paper presents a deep reinforcement learning (DRL) framework for active flow control (AFC) to reduce drag in aerodynamic bodies. Tested on a 3D cylin-der at Re=100 Re = 100, the DRL approach achieved a 9.32% 9.32% drag reduction and a 78.4% 78.4% decrease in lift oscillations by learning advanced actuation strategies. The method-ology integrates a CFD solver with a DRL model using an in-memory database for efficient communication between the two instances, making it scalable to more complex flows and higher Reynolds numbers., This research has received financial support from the Ministerio de Ciencia e Innovacion of Spain (PID2023-150408OB-C21 and PID2023-150408OB-C22). Authors also extend their gratitude to AGAUR for supporting the research group Large-scale Computational Fluid Dynamics (2021 SGR 00902) and the Turbulence and Aerodynamics Research Group (2021 SGR 01051). Ricard Montalà also expresses gratitude to AGAUR for awarding the FI-SDUR grant (2022 FISDU 00066). Ricardo Vinuesa acknowledges financial support from ERC grant no. 2021-CoG-101043998, DEEPCONTROL., Peer Reviewed, 13 - Acció per al Clima, 9 - Indústria, Innovació i Infraestructura
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

Discovering flow separation control strategies in 3d wings via deep reinforcement learning

  • Montalà Sales, Ricard
  • Font García, Bernat
  • Suárez Morales, Pol
  • Rabault, Jean
  • Lehmkuhl Barba, Oriol
  • Vinuesa Motilva, Ricardo
  • Rodríguez Pérez, Ivette María
In this work, deep reinforcement learning (DRL) is applied to active flow control (AFC) over a threedimensional SD7003 wing at a Reynolds number of Re = 60,000 and angle of attack of AoA = 14◦. In the uncontrolled baseline case, the flow exhibits massive separation and a fully turbulent wake. Using a GPU-accelerated CFD solver and multi-agent training, DRL discovers control strategies that enhance lift (79%), reduce drag (65%), and improve aerodynamic efficiency (408%). Flow visualizations confirm reattachment of the separated shear layer, demonstrating the potential of DRL for complex and turbulent flows., This research was partially funded by the Ministerio de Ciencia e Innovación of Spain (PID2023- 150408OB-C21/C22). Simulations were supported by the Red Española de Supercomputación (IM-2024-2-0004 and IM-2024-3-0005) and EuroHPC JU (EHPCREG-2024R01-038), granting access to MareNostrum V at the Barcelona Supercomputing Center. We also acknowledge AGAUR for supporting the LS/CFD (2021 SGR 00902) and TUAREG (2021 SGR 01051) research groups. R. M. thanks AGAUR for the FISDUR grant (2022 FISDU 00066) and R. V. acknowledges financial support from ERC grant no. 2021- CoG-101043998, DEEPCONTROL., Peer Reviewed, 9 - Indústria, Innovació i Infraestructura, 13 - Acció per al Clima, Postprint (published version)