Fast solution of incompressible flow problems with two-level pressure approximation
The authors propose efficient preconditioned iterative solvers for incompressible flow problems discretized by an enriched Taylor-Hood mixed approximation. They start with the Stokes problem: \(-\Delta \overrightarrow{u }+\nabla p=\overrightarrow{0}\), \(\nabla \cdot \overrightarrow{u}=0\), in \( \Omega \), \(\overrightarrow{u}=\overrightarrow{g}\), on \(\partial \Omega \), where \(\overrightarrow{u}\) is the fluid velocity, \(p\) the pressure, and \( \Omega \subset \mathbb{R}^{d}\), \(d=2,3\), is a polygonal or polyhedral domain. Introducing \(V_{h}\subset H_{0}^{1}(\Omega )^{d}\) and \(Q_{h}\subset L_{0}^{2}(\Omega )=\{q\in L^{2}:\int_{\Omega }q=0\}\), an inf-sup stable pair of finite element spaces, the authors write the finite element approximation problem as: find \((\overrightarrow{u}_{h},p_{h})\in V_{h}\times Q_{h}\) such that \(\int_{\Omega }\nabla \overrightarrow{u}_{h}:\nabla \overrightarrow{v} _{h}-\int_{\Omega }p_{h}\nabla \cdot \overrightarrow{v}_{h}=\int_{\Omega } \overrightarrow{f}_{h}\cdot \overrightarrow{v}_{h}\), for all \( \overrightarrow{v}_{h}\in V_{h}\), \(\int_{\Omega }q_{h}\nabla \cdot \overrightarrow{u}_{h}=\int_{\Omega }g_{h}q_{h}\), for all \(q_{h}\in Q_{h}\), where \(\overrightarrow{f}_{h}\) and \(g_{h}\) are associated with the specified boundary velocity. The authors write the associated linear system: \(\left[ \begin{array}{cc} A & B^{T} \\\NB & 0 \end{array} \right] \left[ \begin{array}{c} u \\\Np \end{array} \right] =\left[ \begin{array}{c} f \\\Ng \end{array} \right] \) to be solved, where \(A\in \mathbb{R}^{n_{u}\times n_{u}}\) is symmetric positive definite and \(B\in \mathbb{R}^{n_{p}\times n_{u}}\). An ideal preconditioner for the matrix of this system is \(\mathcal{P}_{ideal}= \left[ \begin{array}{cc} A & \\\N& BA^{-1}B^{T} \end{array} \right] \) and an efficient preconditioner is \(\mathcal{P}=\left[ \begin{array}{cc} M & \\\N& M_{S} \end{array} \right] \), where \(M\in \mathbb{R}^{n_{u}\times n_{u}}\) is equal to \(A\) or an approximation, and \(M_{S}\in \mathbb{R}^{n_{p}\times n_{p}}\) is equal to the pressure matrix \(M_{Q}\) or an approximation. The authors define the enriched Taylor-Hood finite element spaces \(V_{h}^{TH}\times Q_{h}^{TH}\), they describe the null space of the pressure mass matrix \(M_{Q}\) for the enriched Taylor-Hood pressure space, and they build effective preconditioners through approximations of the two-level matrix \(M_{Q}\) of the form \(\left[ \begin{array}{cc} Q_{k} & R^{T} \\\NR & Q_{0} \end{array} \right] \), where \(Q_{k}\) is the standard Taylor-Hood pressure mass matrix, \( Q_{0}\) is the standard discontinuous pressure mass matrix, and \(R\) represents cross terms between the spaces \(Q_{h}^{TH}\) and \(Q_{h}^{0}\). They test two preconditioners on two examples of enclosed flows: a driven-cavity flow in the square domain \(D=[-1,1]^{2}\) and a 3D version of driven-cavity flow. In each case, they compute the inf-sup constant. They then consider an Oseen flow whose equations are: \(-\nu \Delta \overrightarrow{u}+ \overrightarrow{w}\cdot \nabla \overrightarrow{u}+\nabla p=\overrightarrow{0} \), \(\nabla \cdot \overrightarrow{u}=0\), in \(\Omega \), \(\overrightarrow{u}= \overrightarrow{g}\), on \(\partial \Omega \), where \(\overrightarrow{w}\) is a divergence-free convection field. They repeat the construction of preconditioners of the type \(\mathcal{M}=\left[ \begin{array}{cc} M & B^{T} \\\N0 & -M_{S} \end{array} \right] \), where \(M\) is an optimal complexity operator effecting the action of the inverse of the non-symmetric matrix \(F\) associated with the discretization of the preceding equations and \(M_{S}\) is an optimal complexity approximation of the Schur complement matrix \(BF^{-1}B^{T}\). The authors test on two examples: a 2D flow over a step and a 2D enclosed flow. They consider a pressure convection-diffusion preconditioning for the Oseen flow and they finally propose a second way of approximating the matrix \( BF^{-1}B^{T}\) in the case where \(B^{T}\) is generated by a two-field pressure approximation.
- A locally conservative enriched Galerkin approximation and efficient solver for elliptic and parabolic problems
- A locally conservative finite element method based on piecewise constant enrichment of the continuous Galerkin method
- A Note on Preconditioning for Indefinite Linear Systems
- A stable enriched Galerkin element for the Stokes problem
- An enriched Galerkin method for the Stokes equations
- An optimal iterative solver for symmetric indefinite systems stemming from mixed approximation
- Chebyshev semi-iteration in preconditioning for problems including the mass matrix
- Deflation and projection methods applied to symmetric positive semi-definite systems
- Efficient iterative solvers for elliptic finite element problems on nonmatching grids
- Finite elements and fast iterative solvers. With applications in incompressible fluid dynamics
- scientific article; zbMATH DE number 3700440 (Why is no real title available?)
- scientific article; zbMATH DE number 37941 (Why is no real title available?)
- scientific article; zbMATH DE number 3799204 (Why is no real title available?)
- IFISS3D: a computational laboratory for investigating finite element approximation in three dimensions
- Local mass conservation of Stokes finite elements
- Locally mass-conserving Taylor-Hood elements for two- and three-dimensional flow
- Mixed finite element methods and applications
- Numerical solution of saddle point problems
- Pressure-robust enriched Galerkin methods for the Stokes equations
- Realistic Eigenvalue Bounds for the Galerkin Mass Matrix
- Solution of Sparse Indefinite Systems of Linear Equations
- Solving Singular Generalized Eigenvalue Problems by a Rank-Completing Perturbation
- T-IFISS: a toolbox for adaptive FEM computation
- The Convergence of Linear Stationary Iterative Processes for Solving Singular Unstructured Systems of Linear Equations
This page was built for publication: Fast solution of incompressible flow problems with two-level pressure approximation
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6586808)
