| taille du texte : S-M-L |
| impression | intranet

Towards a statistically accurate wall-model for large-eddy simulation

[ jeudi 18-10-2012 15:00 | français ]Towards a statistically accurate wall-model for large-eddy simulation Olivier Cabrit Université de Melbourne
Résumé :
Despite many decades of study, modelling and predicting the wall shear stress in wall-bounded flows remains an active topic of research. The advent of the standard law-of-the-wall considerably improved this field of research leading to accurate models for the mean friction. However, many configurations of interest (fatigue/rupture issues on air-plane wings, wind gusts in meteorology and environmental flows) would benefit from an accurate prediction of the fluctuating wall shear stress.
The use of large-eddy simulation (LES) is very promising for this purpose, but from the early work of Schumann [JCP, 1975] this technique failed to predict reliably the fluctuating wall-shear stress, mainly because most of the modelling approaches developed to date rely on the standard log-law. Recently, Marusic et al. [Science, 2010] proposed a novel approach to develop new models well-posed for LES. It is based on a superposition and amplitude modulation mechanism [Mathis et al., JFM, 2011] in which the large-scale components of the velocity (i.e. the filtered velocity in LES) is the only input.
The present study discusses the implementation of this theoretical model in a LES solver. Contrarily to the standard log-law, this novel approach is not only able to reproduce the high-order moments of the fluctuating wall-shear-stress with a good level of fidelity, but also their Reynolds number dependency. This study makes use of an experimental/numerical database covering two decades of Reynolds numbers to address the question of potential/robustness/efficiency of this new approach.