A machine learning strategy for computing interface curvature in Front-Tracking methods

Autor(es)

Franca, Hugo L. ; Oishi, Cassio M.

• Tipo de documento: Artigo Científico
• Fonte: Journal of Computational Physics; v. 450, p. 9-pg., 2022-02-01.

Resumo

In this work we have described the application of a machine learning strategy to compute the interface curvature in the context of a Front-Tracking framework. Based on angular information of normal and tangential vectors between marker points, the interface curvature is predicted using a neural network. The Front-Tracking-Machine-Learning method is validated using a sine wave and then applied in combination with a Marker-And-Cell method for solving a complex free surface flow. Our results indicate that it is feasible to employ machine learning concepts as an alternative approach for computing curvatures in Front-Tracking schemes. (C) 2021 The Authors. Published by Elsevier Inc. (AU)

• Processo FAPESP: 13/07375-0 - CeMEAI - Centro de Ciências Matemáticas Aplicadas à Indústria
• Beneficiário: Francisco Louzada Neto
• Modalidade de apoio: Auxílio à Pesquisa - Centros de Pesquisa, Inovação e Difusão - CEPIDs
• Processo FAPESP: 19/01811-9 - Simulação numérica de escoamentos de fluidos complexos com superfícies livres
• Beneficiário: Hugo Leonardo França
• Modalidade de apoio: Bolsas no Brasil - Doutorado