Conference Papers Year : 2023

Numerical Analysis of Stefan Problems for Embedded Computation of Moving Internal Boundaries

Abstract

Classical finite element methods used to model problems with internal boundaries rely on body fitted computational grids. However, those methods encounter computational challenges when the boundaries are deformed or moved substantially. In this direction, embedded methods do not require the use ofboundary fitted grids in favor of immersing the boundary in a pre-existing fixed grid. In this talk we are interested in the shifted boundary method, where a surrogate boundary is added to the physical one. For simplicity, we focus on a Stefan Problem written in its mixed form. In the corresponding variational formulation, the moving boundary evolves at a speed determined by the normal flux jump. To obtain an accurate prediction of the temperature field on both sides of the discontinuity, as well as the position of the discontinuity itself, we propose an enhanced variant of the shifted boundary method based on an enriched stabilized mixed form (see [1], [2]). Note that, since the boundary is moving inside the domain, some instabilities can appear. It is then necessary to perform a linear stability analysis (see [3]). We also present some numerical computations, which confirm the expected overall second order accuracy of the method and its ability to properly simulate de-icing problems.
Fichier principal
Vignette du fichier
ENUMATH_lisbonSeptembre2023.pdf (65.35 Ko) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-04359844 , version 1 (21-12-2023)

Licence

Public Domain

Identifiers

  • HAL Id : hal-04359844 , version 1

Cite

Tiffanie Carlier, Brian T Helenbrook, Heloise Beaugendre, Mathieu Colin, Léo Nouveau. Numerical Analysis of Stefan Problems for Embedded Computation of Moving Internal Boundaries. ENUMATH 2023 - The European Conference on Numerical Mathematics and Advanced Applications, Sep 2023, Lisbonne, Portugal. ⟨hal-04359844⟩
39 View
29 Download

Share

More