TU Darmstadt / ULB / TUbiblio

Non-conforming Galerkin finite element methods for symmetric local absorbing boundary conditions

Schmidt, Kersten and Diaz, Julien and Heier, Christian (2015):
Non-conforming Galerkin finite element methods for symmetric local absorbing boundary conditions.
In: Comput. Math. Appl., pp. 2252-2269, 70, (9), DOI: 10.1016/j.camwa.2015.08.034,
[Article]

Abstract

A new non-conforming finite element discretization methodology for second order elliptic partial differential equations involving higher order local absorbing boundary conditions in 2D and 3D is proposed. The novelty of the approach lies in the application of C⁰-continuous finite element spaces, which is the standard discretization of second order operators, to the discretization of boundary differential operators of order four and higher. For each of these boundary operators, additional terms appear on the boundary nodes in 2D and on the boundary edges in 3D, similarly to interior penalty discontinuous Galerkin methods, which leads to a stable and consistent formulation. In this way, no auxiliary variables on the boundary have to be introduced and trial and test functions of higher smoothness along the boundary are not required. As a consequence, the method leads to lower computational costs for discretizations with higher order elements and is easily integrated in high-order finite element libraries. \em A priori \em h-conv\-er\-gen\-ce error estimates show that the method does not reduce the order of convergence compared to usual Dirichlet, Neumann or Robin boundary conditions if the polynomial degree on the boundary is increased simultaneously. A series of numerical experiments illustrates the utility of the method and validates the theoretical convergence results.

Item Type: Article
Erschienen: 2015
Creators: Schmidt, Kersten and Diaz, Julien and Heier, Christian
Title: Non-conforming Galerkin finite element methods for symmetric local absorbing boundary conditions
Language: English
Abstract:

A new non-conforming finite element discretization methodology for second order elliptic partial differential equations involving higher order local absorbing boundary conditions in 2D and 3D is proposed. The novelty of the approach lies in the application of C⁰-continuous finite element spaces, which is the standard discretization of second order operators, to the discretization of boundary differential operators of order four and higher. For each of these boundary operators, additional terms appear on the boundary nodes in 2D and on the boundary edges in 3D, similarly to interior penalty discontinuous Galerkin methods, which leads to a stable and consistent formulation. In this way, no auxiliary variables on the boundary have to be introduced and trial and test functions of higher smoothness along the boundary are not required. As a consequence, the method leads to lower computational costs for discretizations with higher order elements and is easily integrated in high-order finite element libraries. \em A priori \em h-conv\-er\-gen\-ce error estimates show that the method does not reduce the order of convergence compared to usual Dirichlet, Neumann or Robin boundary conditions if the polynomial degree on the boundary is increased simultaneously. A series of numerical experiments illustrates the utility of the method and validates the theoretical convergence results.

Journal or Publication Title: Comput. Math. Appl.
Volume: 70
Number: 9
Divisions: 04 Department of Mathematics
04 Department of Mathematics > Numerical Analysis and Scientific Computing
Date Deposited: 19 Nov 2018 21:15
DOI: 10.1016/j.camwa.2015.08.034
Export:

Optionen (nur für Redakteure)

View Item View Item