AN ADAPTIVE AGGLOMERATION METHOD FOR ADDITIVE CORRECTION MULTIGRID

In the computational simulation of fluid flow and scalar transport, multigrid iterative solution techniques often fail or stall when the discrete linearized equations have strongly anisotropic coefficients. In the present work, an adaptive agglomeration algorithm for forming coarse grids is presented that allows multigrid techniques to work efficiently for equation sets with anisotropic coefficients. The adaptive agglomeration is defined by two rules and several guidelines that follow from a physical interpretation of the performance of iterative solvers like Gauss–Seidel. The effectiveness of the adaptive agglomeration algorithm is demonstrated for a wide range of test cases. © 1997 by John Wiley & Sons, Ltd.