Mercredi 4 Avril 2012
14 h 00 salle de réunion Branche Fluides
Aerospace Computing Laboratory
Stanford University
Invité par Eric Lamballais
During the last fifteen years computational fluid dynamics (CFD) has been on a plateau with numerous production CFD codes with a well validated capability to solve the Reynolds-Averaged Navier-Stokes (RANS) equations for complex configurations over a wide range of Mach numbers. Advances in computer hardware will enable Large-Eddy Simulation (LES) for industrial problems of interest within the foreseeable future. In order to enable LES for complex configurations, there is a need for methods using unstructured meshes.
Given the current computational capabilities, it is recognized that developing convergence acceleration and hybrid RANS-LES methodologies is crucial in order for LES to become a viable option in a very broad range of applications, including high Reynolds number massively detached flows and noise prediction. There is also general consensus within the scientific community about the necessity to direct research efforts toward high-order methods rather than the more common, and excessively dissipative, low-order upwind biased schemes (Fujii, 2005, Prog. Aerosp. Sci. 41; Georgiadis, et al., 2010, AIAA J. 48:8). Hence it is necessary to combine high-order unstructured numerical schemes with advanced SGS modeling techniques in order for LES to become a valuable and reliable tool for fundamental flow physics and industrial applications.
High-order numerical schemes for solving the compressible Navier-Stokes equations on unstructured grids have been widely studied during the last decade. By far the most mature and widely used of these schemes are based on the Discontinuous Galerkin (DG) method. Recently, however, several alternative high-order methods have been proposed, including Spectral Difference (SD) and Flux Reconstruction (FR) type schemes, which potentially offer increased efficiency compared with DG methods. Within the framework of the SD and FR methods, the Aerospace Computing Laboratory (ACL) at Stanford University have focused extensive research in the past years and successfully addressed viscous compressible flows with shocks, implicit LES of turbulent channel flow and flow around circular cylinders, as well as, transitional turbulent flows. Other relevant applications include the implementation of adaptive refinement and implicit time integration techniques and moving and deformable meshes.
More recently, the implementation of explicit filtering structural SGS models for LES has been addressed.
The object of the present talk is to give a short overview on the actual status of the development of the high-order SD scheme and its capabilities in a relatively broad range of applications.
Liste des séminaires ayant déjà eu lieu en 2009, 2010, 2011 et 2012
Pour toute proposition de séminaire, contactez Laurent Cordier. Merci de suivre les règles d'organisation correspondantes.