A Spectrally Filtered, Least-Squares Projection Method for Stokes Flow Problems in Driven Cavities

Abstract

A least-squares, weighted residual projection method is presented forcomputingStokes flow solutions to driven cavity problems in rectangular andcylindrical geometries.In this procedure, thevelocity field components are first defined by eigenfunction expansionsolutions to the Stokes flow problem in terms of an unknown pressure fieldwhich is subsequentlycomputed by minimizing the continuity equation residual norm by theleast-squaresprojection. The role of spectral filtering methods for improving pointwisesolutionconvergence is also discussed.