Computational requirements for three-dimensional flows

Abstract

The computational requirements needed for predicting steady viscous flow over complex configurations are considered. The desired predictions must be made at reasonable expense, require a reasonable amount of storage space, and result in solutions that are sufficiently accurate. The data needed to estimate the cost of Navier-Stokes solutions is not available; therefore, experience with the solution of the three-dimensional boundary layers equations are used to illustrate the needed information and what can be expected for the Navier-Stokes solutions.