Computation of Turbulent Boundary Layer Flow with Separation — A Critical Evaluation of Parameters Influencing the Numerical Results

Abstract

An exact calculation of turbulent boundary-layer flow with separation generally requires the use of the complete Navier-Stokes equations. In order to determine the turbulence-transport mechanism in the boundary layer, the grid coordinates must correspond to the flow field. Very fine computational grids are required, and even today these grids cause significant difficulties in solving the Navier-Stokes equations for high Reynolds numbers. In practical applications, attached boundary layers can often be computed successfully using the boundary-layer equations. However, in the case of boundary layers with closed separation bubbles, these equations have a singularity at the points of separation and reattachment. This problem can be avoided [l] by using the inverse formulation in which the displacement thickness, rather than the edge velocity, is a prescribed boundary condition.