Computing Turbulent Flow Dynamics With Implicit Large Eddy Simulation

The Navier-Stokes equations NSE can be solved directly for aminar flows, but the wide range of eddy scales to be captured rohibits direct numerical simulation for the high Reynoldsumber turbulent flows of technological interest. The prevalent emedy to this resolution problem has traditionally involved the eynolds-averaged Navier-Stokes RANS approach, with averging typically carried out over time or across an ensemble of quivalent flows. The applicability of RANS typically requires hat time scales associated with organized unsteady motion be ubstantially larger than those of turbulent motion. Such statistially steady flow assumptions can be satisfied in many e.g., lowrequency dominated unsteady flow applications, but most turbuent flows of interest do not fall into this category. A more viable approach is the large eddy simulation LES pproach 1 . LES is based on the expectation that the physically eaningful scales of turbulence can be split into two groups: one onsisting of the resolved geometry and regime specific scales, nd the other associated with the unresolved smallest eddies in the ow, for which the presumably more-universal dynamics is repesented with subgrid scale SGS closure models. Scale separaion is achieved by solving the low-pass filtered NSE, and using xplicit SGS models introduced for closure prior to discretization. n the absence of an accepted universal theory of turbulence to olve the problem of SGS modeling, the development and imrovement of such models has been driven by pragmatic practice ependent on the rational use of empirical information. In addition o the physics based difficulties in developing and validating SGS odels, one is faced with simulations where contributions from umerical truncation terms can be as significant as those from GS models in typical LES strategies. LES resolution requireents can thus become prohibitively expensive for practical flows nd regimes of interest. Implicit LES ILES effectively addresses he seemingly insurmountable issues posed to LES by underesolution, by relying on the use of SGS modeling and filtering rovided implicitly by physics-capturing numerics of a broad lass of high-resolution, non-linear finite-volume methods. Increasing interest in ILES techniques is reflected in recent edicated sections in archival journals 2,3 , dedicated chapters in FD textbooks 4 , and by the publication of the first comprehenive book synthesizing our current understanding of the theoretial basis and accomplishments of ILES 5 . Timely ILES issues ere addressed in two very-well attended minisymposia on Computing Turbulent Flow Dynamics with ILES” at the 2006 CCOMAS CFD conference at Egmond aan Zee, Netherlands, eptember 5–8, 2006. Five papers selected from the invited sesions and two other separately invited papers were assembled in he present special issue of JFE to provide a broad state-of-the-art erspective. The first four papers deal with fundamental studies using modied equation analysis and tests in basic cases for which good eference data are available. Grinstein and Fureby discuss a class f flux-limiting methods used for ILES in both, incompressible nd compressible regimes, focusing on their commonalities and asic performance. Comparative verification tests of ILES and