Toward the De-Mystification of LES

Abstract : Some 35 years have elapsed since large-eddy simulation (LES) was introduced as a computational tool for weather modeling (Lilly, 1966). Because direct numerical simulation (DNS) will remain prohibitively expensive well into the foreseeable future for high-Reynolds-number flow, LES continues to hold great promise for simulating flows of engineering interest. However, despite the attention the method has received over the past decades, and despite important recent developments (e.g., the notions of "explicit" filtering and localized or "dynamic' modeling (Germano et al., 1992)), LES has been slow to mature as a predictive tool. In contrast, parabolized stability equation (PSE) methodology, for example, which originated in the mid 1980's (Bertolotti et al., 1992), matured quickly and is now ready for use by the aerospace industry for transition prediction. In the author s view, the relatively slow adaptation of LES as a predictive tool arises not from any fundamental flaw in the idea itself but most likely from misconceptions that widely permeate the practice of LES. The current paper addresses three pervasive misconceptions, each of which reveals a lack of clarity regarding the properties of digital filters and the relationship between the grid filter and the subgrid-scale (SGS) model.