| An efficient new class of implicit algorithms for the computation of steady, two- and three-dimensional, compressible Navier-Stokes flowfields is presented. A nonlinear relaxation strategy based on a fully-coupled sequence of line or planar Gauss-Seidel sweeps is used to drive the solution toward a steady state. Quasi-Newton techniques are used to accelerate the convergence of the 2-D algorithm and to reduce approximate factorization errors in the 3-D version. Convergence characteristics are further improved through the addition of a coarse-grid correction procedure. The discretization of the Navier-Stokes set is hybrid in nature, with flux-vector splitting utilized for the streamwise inviscid fluxes and central differences with flux-limited artificial dissipation used for the transverse inviscid fluxes. Viscous fluxes are central-differenced. Both laminar and turbulent cases are considered, with turbulent closure provided by a modification of the Baldwin-Barth one-equation model. Convergence histories and comparisons with experimental data are presented for several 2-D and 3-D shock wave-boundary layer interactions. Several numerical simulations of the Mach 10 flow through a generic 3-D sidewall compression SCramjet inlet configuration are also presented. For the medium-sized (175,000-325,000 mesh points) 3-D validation cases, the algorithm provides steady-state convergence in 15 to 17 Cray Y-MP CPU minutes. Converged solutions for some of the inlet flowfields (375,000-395,000 mesh points) are obtained in about one CPU hour. |
