The dual variable method for finite element discretizations of Navier/Stokes equations

The numerical solution of two-dimensional, transient, incompressible Navier–Stokes problems is considered. The dual variable method, originally developed in the context of a finite difference discretization, is a technique to considerably reduce the size of the linear system to be solved at each time step. The steps involved in the method are (1) the determination of the rank of the discrete divergence operator, A, (2) the determination of a basis for the null space of A, N(A), and (3) the calculation of a particular solution of the discrete continuity equation. A finite element implementation of the method is presented using quadrilateral piecewise bilinear velocity/constant pressure elements. Algorithms for the determination of a basis for N(A) and a particular solution are presented. Numerical comparisons of primitive versus dual variable formulations on several problems demonstrate the advantage of the dual variable method, in terms of both execution speed and memory requirements.