Computation of Incompressible Flows Using Turbulence Models

In order to predict turbulent flows by finite difference solutions to the Reynolds equations, it becomes necessary to make closure assumptions regarding the apparent turbulent stress. Direct numerical simulations (DNS) avoid the closure problem by solving for the full, time-dependent Navier-Stokes equations. But, high resolution and large-computer capabilities are required to capture the turbulence structure. The recent expansion of computer capabilities makes DNS an important subject. Turbulence modeling of the Reynolds-averaged equation, however, is still considered the practical engineering tool. All presently known turbulence models have limitations; the ultimate turbulence model has yet to be developed. It is important to remember that turbulence models must be verified by comparing predictions with experimental measurements. Care must be taken in interpreting predictions obtained from models outside the range of conditions over which they have been verified by comparisons with experimental data.