The Computation of Turbulent Engineering Flows with Turbulence-Transport Closures
The paper discusses aspects of modelling complex turbulent flows, placing particular emphasis on second-moment closure and non-linear eddy-viscosity formulations. Principal features of turbulence, viewed mainly in statistical terms, are highlighted first.
- PDF / 7,938,215 Bytes
- 69 Pages / 481.89 x 691.654 pts Page_size
- 20 Downloads / 234 Views
THE COMPUTATION OF TURBULENT ENGINEERING FLOWS WITH TURBULENCE-TRANSPORT CLOSURES
M.A. Leschziner UMIST, Manchester, UK
ABSTRACT
The paper discusses aspects of modelling complex turbulent flows, placing particular emphasis on second-moment closure and non-linear eddy-viscosity formulations. Principal features of turbulence, viewed mainly in statistical terms, are highlighted first. This is followed by considerations directed, principally, towards processes which arise from the interaction between the Reynolds stresses and mean-flow features. Attention focuses, in particular, on the interaction, as expressed through the exact Reynolds-stress generation
terms, between turbulence and curvature, normal straining, system rotation, body forces and heat transfer. This exposition provides the background against which the use of anisotropyresolving closures is advocated. Following a review of simpler approaches, based on the isotropic eddy-viscosity concept, the current status of second-moment and non-linear eddyviscosity modelling is summarised. Consideration is then given to the performance of alternative models by reference to computational solutions for eight flows, both twodimensional and three-dimensional, some incompressible and others compressible. In presenting and discussing representative results, emphasis is placed on fundamental flow features and on assessing the predictive capabilities of alternative models by reference to experimental data. The results are argued to offer support for the use of anisotropyresolving closure, but also serve to highlight model weaknesses and uncertainties which require further research.
R. Peyret et al. (eds.), Advanced Turbulent Flow Computations © Springer-Verlag Wien 2000
210
M.A. Leschziner
1. INTRODUCTION
In 1991, D R Chapman, a prominent US aerodynamicist, boldly predicted (perhaps only in jest) that CFD was en-route to replacing experimental testing, and that 'wind-tunnels were destined to become storage cabinets for computer output'. This prediction might be regarded as entirely reasonable, when viewed from the vantage point of high-speed external aerodynamics and against the background of rapid advances in grid-generation, numerical methods, computer technology, visualisation and the advertising skills of CFD software vendors. However, Chapman appears to have ignored one formidable obstacle, namely turbulence - a phenomenon referred to by Bradshaw [1] as "the invention of the Devil on the 7th day of creation". It is an inescapable fact that the overwhelming majority of engineering and environmental
flows are turbulent, or at least contain influential regions which are turbulent. High-speed flows - the type Chapman probably had in mind - might be dominated by a balance between convection and pressure gradient, but even these can be crucially affected by turbulence through frictional drag, heat transfer, shock/boundary-layer interaction, streamwise vorticity induced by 3D boundary-layer distortion, and separation provoked by strong adverse pressure gradient. At
Data Loading...
