Grad-div stablilization for Stokes equations

From MaRDI portal





The positive effect of the term \(\xi (\text{ div}\;u, \text{ div}\;v)\) in the variational formulation of the Stokes problem is analyzed. Here \(\xi\geq 0\), \(u\) is the unique solution, \(v\) denotes any test function in \((H_0^1)^d\), and \((\cdot,\cdot)\) stands for the \(L_2\) scalar product. In the incompressible case, the presence of that term does not change the continuous solution of the problem. Nevertheless, its inclusion has a positive effect in the performance of some numerical methods. This is called \textit{grad-div stabilization}. The main result presented concerns finite element error bounds. Ladyshenskaya-Babuška-Brezzi stable conforming finite element spaces are employed. For \(\nu\downarrow 0\), where \(\nu\) is the viscosity of the fluid, it is proved that the discretization error bounds become significantly better if grad-div stabilization is used, and through numerical experiments the authors show that the theoretical upper bounds predict the correct behaviour. It is also analyzed an Uzawa type iterative method for solving the discrete problem, showing that the rate of convergence of the outer iteration for solving the Schur complement equation for the pressure in general increases if grad-div stabilization is employed. But, as the authors observe, the inner velocity problem becomes stiffer for \(\nu\downarrow 0\). The conclusion is that by adding the grad-div term the Stokes problem with \(\nu\ll 1\) can be solved up to a prescribed accuracy with significantly lower arithmetic costs. Similar effects occur when solving the linearized Navier-Stokes equations with high Reynolds numbers. For the continuous variational problem, it is proved uniform (with respect to \(\nu\)) well-posedness in the natural energy norm when \(\xi>0\). This is not true when \(\xi=0\).



Cites work


Cited in
(only showing first 100 items - show all)








This page was built for publication: Grad-div stablilization for Stokes equations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4813605)