Large-Eddy Simulations

Abstract

As already stressed in the previous chapters, there is a priori no difficulty in envisaging a numerical solution of the unstationary Navier Stokes equations for rotational flows: the various operators are represented by discrete systems relating the values taken by the velocity or vorticity components, pressure, density, temperature, etc... on a space time grid. This grid may be spatially regular or irregular, with finite-difference, finite-volume or finite-element methods¹. Often an orthogonal decomposition of the flow allows a spectral method to be used (see e.g. Canuto et al., 1987). For incompressible two-dimensional flows, the use of the stream function permits the elimination of the pressure. It is not the aim of the present monograph to describe the various numerical methods used in the so-called Computational Fluid Dynamics. We will insist rather on the physical limitations which arise when such a simulation is performed on a turbulent flow.