scientific article; zbMATH DE number 1176590
characteristic Galerkin methodcharacteristic streamline diffusion methodLagrange-Galerkin methodnon-conforming finite element spacesoriented grid generation
Research exposition (monographs, survey articles) pertaining to fluid mechanics (76-02) Navier-Stokes equations for incompressible viscous fluids (76D05) Finite element methods applied to problems in fluid mechanics (76M10) Finite difference methods applied to problems in fluid mechanics (76M20) Diffusion and convection (76R99)
Our purpose is a comparative study of several transport-oriented approaches, with respect to each other but also to Eulerian discretizations for which well-tuned codes are available. The two main strategies are the characteristic Galerkin method or the characteristic streamline diffusion method combining an oriented grid generation process with a standard Galerkin or Petrov-Galerkin discretization, and the Lagrange-Galerkin method which incorporates the Lagrangian perspective directly into the discretization. The investigations are organized in three parts. In chapter one, we start from characteristic Galerkin approaches which are based on a relatively rough approximation of fluid paths that serves for the construction of space-time meshes. Theoretically, they promise improved accuracy as well as stability properties for linear solvers than realizable by discretizations with non-oriented meshes. In practice, these advantages are substantially reduced: the speed-up of linear solvers is mainly outweighed by the additional effort for matrix computation, and more complicated how structures need high-order approximations of the characteristics in order to guarantee well-defined grid generation. Therefore, we come to Lagrange-Galerkin methods which directly incorporate the approximation of characteristics into the discretization. In the context of transport-dominated model problems, not involving any reactive source terms or special physical constraints like for flow problems in porous media, we achieve a comparison between a Lagrangian and an Eulerian discretization, both based on non-conforming finite element spaces. Since Lagrangian methods theoretically require relatively high regularity assumptions not to be fulfilled by non-conforming functions, chapter two confronts, for linear test cases, this approach with its conforming counterpart. In chapter three, we are then able to obtain fairly general conclusions on the pros and cons of Lagrangian methods applied to the incompressible Navier-Stokes equations, set into contrast with a well-tuned Eulerian solver.
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4398487)
