TY - EJOU AU - Dumbser, M. AU - Balsara, D.S. TI - High-Order Unstructured One-Step PNPMSchemes for the Viscous and Resistive MHD Equations T2 - Computer Modeling in Engineering \& Sciences PY - 2009 VL - 54 IS - 3 SN - 1526-1506 AB - In this article we use the new, unified framework of high order one-step PNPM schemes recently proposed for inviscid hyperbolic conservation laws by Dumbser, Balsara, Toro, and Munz (2008) in order to solve the viscous and resistive magnetohydrodynamics (MHD) equations in two and three space dimensions on unstructured triangular and tetrahedral meshes. The PNPM framework uses piecewise polynomials of degree N to represent data in each cell and piecewise polynomials of degree M ≥ N to compute the fluxes and source terms. This new general machinery contains usual high order finite volume schemes (N = 0) and discontinuous Galerkin finite element methods (N = M) just as special cases of a more general unified class of numerical methods. The new time discretization approach used in our schemes is of the one-step type and is obtained by a local space-time Galerkin predictor-corrector method, where in the predictor step a local weak form of the governing PDE is solved in the small. The formal order of accuracy of our one-step time discretization automatically matches the spatial discretization order.
We perform a numerical convergence study for an unsteady test case on unstructured meshes with PNPM methods from third to eighth order of accuracy in space and time and show detailed CPU times to allow the reader a critical performance assessment. We present numerical results for the viscous Orszag-Tang vortex system, for the first problem of Stokes and for a steady laminar boundary layer at high Reynolds number. We furthermore compute the unsteady flow past a circular cylinder and the Kelvin-Helmholtz instability in two and three space dimensions. KW - Viscous and resistive MHD equations KW - unstructured meshes KW - high order finite volume and discontinuous Galerkin methods KW - PNPMschemes KW - Petrov-Galerkin method DO - 10.3970/cmes.2009.054.301