On the use of deep learning and computational fluid dynamics for the estimation of uniform momentum source components of propellers

This article proposes a novel method based on Deep Learning for the resolution of uniform momentum source terms in the Reynolds-Averaged Navier-Stokes equations. These source terms can represent several industrial devices (propellers, wind turbines, and so forth) in Computational Fluid Dynamics simu...

Descripción completa

Detalles Bibliográficos
Autores: Martínez Cuenca, Raúl, Luis Gómez, Jaume, Iserte, Sergio, Chiva, Sergio
Tipo de recurso: artículo
Fecha de publicación:2023
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/396458
Acceso en línea:https://hdl.handle.net/2117/396458
https://dx.doi.org/10.1016/j.isci.2023.108297
Access Level:acceso abierto
Palabra clave:Deep learning (Machine learning)
Computational fluid dynamics
Artificial intelligence
Industrial engineering
Simulació per ordinador
Àrees temàtiques de la UPC::Informàtica::Aplicacions de la informàtica::Aplicacions informàtiques a la física i l‘enginyeria
Descripción
Sumario:This article proposes a novel method based on Deep Learning for the resolution of uniform momentum source terms in the Reynolds-Averaged Navier-Stokes equations. These source terms can represent several industrial devices (propellers, wind turbines, and so forth) in Computational Fluid Dynamics simulations. Current simulation methods require huge computational power, rely on strong assumptions or need additional information about the device that is being simulated. In this first approach to the new method, a Deep Learning system is trained with hundreds of Computational Fluid Dynamics simulations with uniform momemtum sources so that it can compute the one representing a given propeller from a reduced set of flow velocity measurements near it. Results show an overall relative error below the for momentum sources for uniform sources and a moderate error when describing real propellers. This work will allow to simulate more accurately industrial devices with less computational cost.