This volume contains a selection of the papers presented at the Eighth Symposium on Turbulent Shear Flows held at the Technical University of Munich, 9-11 September 1991. The first of these biennial international symposia was held at the Pennsylvania State Uni versity, USA, in 1977; subsequent symposia have been held at Imperial College, London, England; the University of California, Davis, USA; the University of Karlsruhe, Ger many; Cornell University, Ithaca, USA; the Paul Sabatier University, Toulouse, France; and Stanford University, California, USA. The purpose of this series of symposia is to provide a forum for the presentation and discussion of new developments in the field of turbulence, especially as related to shear flows of importance in engineering and geo physics. From the 330 extended abstracts submitted for this symposium, 145 papers were presented orally and 60 as posters. Out of these, we have selected twenty-four papers for inclusion in this volume, each of which has been revised and extended in accordance with the editors' recommendations. The following four theme areas were selected after consideration of the quality of the contributions, the importance of the area, and the selection made in earlier volumes: - wall flows, - separated flows, - compressibility effects, - buoyancy, rotation, and curvature effects. As in the past, each section corresponding to the above areas begins with an introduction by an authority in the field that places the individual contributions in context with one another and with related research.

The LES-method is rapidly developing in many practical applications in engineering The mathematical background is presented here for the first time in book form by one of the leaders in the field

Large-Eddy Simulations of Turbulence is a reference for LES, direct numerical simulation and Reynolds-averaged Navier-Stokes simulation.

The issue of modeling the instantaneous viscous and pressure stresses at the wall is addressed via a review of the OLES simulation of turbulent channel flow performed by Das (2004). In this approach, a buffer region with zero velocity is attached adjacent to the wall, and the extended velocity field is filtered using a Fourier-cutoff filter in all directions. The instantaneous wall stresses are then obtained using a "no leakage" condition, where the energy in the buffer-region is minimized at every time-step. Some changes are introduced to the previous formulation---the nonlinear and subgrid terms in the LES equation are obtained using a nonlocal re-filtering approach and a "matched buffer" condition is used to obtain the wall stresses. Simulations performed using both the "no leakage" and "matched buffer" conditions yield statistics which compare well with DNS data. Some numerical experiments involving the linear OLES kernel are also performed, where it is shown that the positive eigenvalues in the kernel and the skew-symmetric part of the kernel are important.

This volume contains papers presented to a EUROMECH-Colloquium held in Munich, September 30 to October 2, 1985. The Colloquium is number 199 in a series of colloquia inaugurated by the European Mechanics Committee. The meeting was jointly organized by the 'Lehrstuhl fur Stromungsmechanik' at the 'Technische Universitat Munchen' and the 'Institut fur Physik der Atmosphare' of the 'Deutsche Forschungs- und Versuchsanstalt fur Luft- und Raumfahrt' (DFVLR) in Oberpfaffenhofen. 'Direct' and 'large eddy simulation' are terms which denote two closely con nected methods of turbulence research. In a 'direct simulation' (DS), turbu lent motion is simulated by numerically integrating the Navier-Stokes equations in three-dimensional space and as a function of time. Besides ini tial and boundary conditions no physical simplifications are involved. Com puter resources limit the resolution in time and space, though simulations with an order of one million discrete points in space are feasible. The simu lated flow fields can be considered as true realizations of turbulent flow fields and analysed to answer questions on the basic behaviour of turbulence. Direct simulations are valid as long as all the excited scales remain within the band of resolved scales. This means that viscosity must be strong enough to damp out the not resolved scales or the simulation is restricted to a lim ited integration-time interval only. In summary, DS provides a tool to investigate turbulent motions from first principles at least for a finite band of scales.