Artículo
Super resolution generative adversarial network for velocity fields in Large Eddy Simulations
Fecha
2023Registro en:
Bove, M., Nesmachnow, S. y Draper, M. "Super resolution generative adversarial network for velocity fields in Large Eddy Simulations". Communications in Computer and Information Science (CCIS). [en línea]. 2023, vol. 1706, pp. 60-74.
1865-0929
1865-0937
Autor
Bove, Maximiliano
Nesmachnow, Sergio
Draper, Martín
Institución
Resumen
This article presents an approach for generating synthetic velocity fields in Large Eddy Simulations. This is a relevant problem, considering the high computational effort required to simulate turbulent flows with fine resolution. The proposed approach applies a Generative Adversarial Network, considering relevant information about horizontal slices of turbulent velocity fields. The approach is evaluated on a realworld case study: augmenting the resolution of horizontal velocity fields downstream of a wind turbine. The main results indicate that the proposed approach is able to generate high resolution images of horizontal velocity fields given a low resolution counterpart, without the need for explicitly performing computationally expensive Large Eddy Simulations.