Convergence of Two Numerical Schemes for Turbulent Boundary Layer Computations

The accuracy of numerical solutions for the incompressible turbulent boundary layer past a flat plate is addressed by means of a prior and a posteriori analysis. The former is carried out by computing the exact solutions of the Reynolds averaged Navier-Stokes equations with the Baldwin-Lomax and Spalart-Allmaras models; these solutions are used for the truncation error estimate in the modified equations. The latter is performed by applying a generalized Richardson extrapolation to the numerical solutions computed by two different finite volume techniques, namely a centered scheme with artificial dissipation and a ENO-type scheme. The results of the a priori theoretical analysis were confirmed by a posteriori analysis of numerical solutions.