Efficient Solution Methods for Strand Grid Applications

摘要

The strand-Cartesian grid approach offers many advantages in terms of automation, efficiency, and accuracy for complex moving-body problems, such as rotorcraft. In this study, the solution procedures for strand grids are investigated by testing a variety of explicit, implicit, and hybrid methods on canonical aerodynamics test cases involving inviscid and viscous flows in three dimensions. A novel multigrid algorithm is formulated that acts at both the nonlinear pseudotime level and the linear level using line Gauss-Seidel sweeps. Various approximations are tested for the Jacobian, and the impacts on memory and convergence are quantified. The major result is that implicit schemes with first-order approximate Jacobians employing few Gauss-Seidel sweeps are most efficient for strand grids. The improvement by using the multigrid algorithm is significant for all cases tested and indicates (O)(n) convergence.