An Asymptotically Fitted Finite Element Method for Convection Dominated Convection-Diffusion-Reaction Problems (Q4033978)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 166692
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | An Asymptotically Fitted Finite Element Method for Convection Dominated Convection-Diffusion-Reaction Problems |
scientific article; zbMATH DE number 166692 |
Statements
An Asymptotically Fitted Finite Element Method for Convection Dominated Convection-Diffusion-Reaction Problems (English)
0 references
16 May 1993
0 references
error estimates
0 references
finite element
0 references
streamline-diffusion
0 references
convection- dominated convection-diffusion-reaction problems
0 references
boundary layers
0 references
numerical examples
0 references
mesh refinement
0 references
exponentially fited method
0 references
0.9374397
0 references
0.92819965
0 references
0.9273093
0 references
0.9262405
0 references
0.9240264
0 references
0.9147085
0 references
The author presents convergence of an asymptotically fitted variant SDFEM-A of the finite element method of streamline-diffusion type (SDFEM) for singularly perturbed elliptic boundary value problems modelling convection-dominated convection-diffusion-reaction problems. The method is based on the observation that for \(\varepsilon \leq C \cdot h\) and \(\varepsilon \leq C \cdot h^{3/2}\), respectively, any unrefined mesh cannot resolve the downstream and characteristic boundary layers, respectively [cf. \textit{C. Johnson}; \textit{A. H. Schatz, L. B. Wahlbin}: Math. Comput. 49, 25-38 (1987; Zbl 0629.65111)]. The idea consists of replacing the sharp layers by smooth layers. As explained by \textit{O. Axelsson} [I.M.A. J. Numer. Anal. 1, 329-345 (1981; Zbl 0508.76069)], this method can he viewed, in some sense as a limit case (\(\varepsilon \ll h\)) of using exponentially weighted functions. Without perturbing the simple finite element shape structure and desirable linearization properties of SDFEM, the SDFEM-A allows for global error estimates in \(L_ 2\)-norm and sometimes in a weighted \(W'_ 2\)-norm which are uniformly valid with respect to \(\varepsilon\). Such global results are not valid for SDFEM. As a result it is concluded that boundary layers are better approximated by the SDFEM-A. Theoretical superiority of SDFEM-A is demonstrated through four numerical examples with the help of tables and graphs. Computed results establish that local oscillations of SDFEM-solutions in boundary layers are suppressed by SDFEM-A. Consequently SDFEM-A is in some situations an alternative to mesh refinement methods or exponentially fitted methods to resolve the layers. The question of optimal local \(L_ \infty\)-estimates for SDFEM is still open. The paper by \textit{C. Johnson} et al. (loc. cit.) concerning a modified streamline diffusion method SDFEM-C is a step forward to solve the problem. The streamline diffusion schemes are now a common method for solving as well transport-dominated problems as more complicated convection- dominated flow problems, in particular incompressible and compressible Euler or Navier-Stokes equations.
0 references