Abstract

Turbulence models based on the Spalart-Allmaras Detached-Eddy Simulation (DES) approach are used to compute the turbulent flow over a two-dimensional dune geometry. DES was developed for massively separated flows, but has been applied as a wall model to attached flows as well. In attached shear layers, however, the lack of resolved eddies in the region where the model switches from a turbulence model to a Subfilter-Scale (SFS) one, results in an underprediction of the wall stress, and a shift in the logarithmic layer. The dune studied here is neither a fully attached flow nor a massively separated one, and allows us to investigate the accuracy of DES wall-models in intermediate cases of this type. Results are compared to a well-validated Large-Eddy Simulation (LES) database. DES based methods are found to be more accurate in this application, compared to attached boundary layers. All the methods required approximately 3% of the CPU time of the wall-resolved LES simulations. All methods gave similar results, but the Improved Delayed Detached Eddy Simulation seemed preferable because of the consistency of the trends it predicted.

Publication Type:	Article
Additional Information:	© Elsevier 2018. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/
Publisher Keywords:	River dunes, Detached-Eddy Simulation, Large-Eddy Simulation, Wall-Modelled LES
Subjects:	G Geography. Anthropology. Recreation > GB Physical geography Q Science > QA Mathematics > QA75 Electronic computers. Computer science T Technology > TL Motor vehicles. Aeronautics. Astronautics
Departments:	School of Science & Technology > Engineering

Preview

Text - Accepted Version Download (2MB) | Preview

Export

Downloads