A multi-block ADI finite-volume method for incompressible Navier-Stokes equations in complex geometries

  • Authors:

  • Satbir Singh;Donghyun You

  • Affiliations:

  • Department of Mechanical Engineering, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, United States;Department of Mechanical Engineering, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, United States

  • Venue:
  • Journal of Computational Physics
  • Year:
  • 2011

Quantified Score

Hi-index 31.45

Visualization

Abstract

An efficient second-order accurate finite-volume method is developed for a solution of the incompressible Navier-Stokes equations on complex multi-block structured curvilinear grids. Unlike in the finite-volume or finite-difference-based alternating-direction-implicit (ADI) methods, where factorization of the coordinate transformed governing equations is performed along generalized coordinate directions, in the proposed method, the discretized Cartesian form Navier-Stokes equations are factored along curvilinear grid lines. The new ADI finite-volume method is also extended for simulations on multi-block structured curvilinear grids with which complex geometries can be efficiently resolved. The numerical method is first developed for an unsteady convection-diffusion equation, then is extended for the incompressible Navier-Stokes equations. The order of accuracy and stability characteristics of the present method are analyzed in simulations of an unsteady convection-diffusion problem, decaying vortices, flow in a lid-driven cavity, flow over a circular cylinder, and turbulent flow through a planar channel. Numerical solutions predicted by the proposed ADI finite-volume method are found to be in good agreement with experimental and other numerical data, while the solutions are obtained at much lower computational cost than those required by other iterative methods without factorization. For a simulation on a grid with O(10^5) cells, the computational time required by the present ADI-based method for a solution of momentum equations is found to be less than 20% of that required by a method employing a biconjugate-gradient-stabilized scheme.