Search results

(1 - 1 of 1)

Title

Computing transition and flow development in turbulent channel flows

Creator

Sallot, Thomas

Date

2012-12-05, 2012-12

Description

A zero pressure gradient (ZPG) boundary layer in a wind-tunnel test section was computed using two-dimensional CFD simulations utilizing two...
Show moreA zero pressure gradient (ZPG) boundary layer in a wind-tunnel test section was computed using two-dimensional CFD simulations utilizing two RANS models, Spalart-Allmaras (SA) and k􀀀! (SST), and two di erent software suites; i.e., Fluent and STAR-CCM+. These computations have been compared and they revealed that the SST model yields better agreement with experimental data than the SA. They also demonstrate that STAR-CCM+ has very little sensitivity to small pressure gradients, and that is why its computation for the ZPG are not exactly matching the reference data (experiment and DNS). Further analysis of the two softwares leads to the conclusion that the transition is dealt with di erently. These di erent approaches can also contribute to the discrepancies observed in the results. The second part of this study is aimed at analyzing the SA model and proposing an approach to improve its results. Tests on the various options proposed in STAR-CCM+ led us to understand how crucial it is to know the e ect of each setup. In addition, several DNS pro les have been used as in ow conditions for the ZPG problem in order to show that this model can be signi cantly improved by using the appropriate in ow conditions past the transition of the boundary layer. Alternatively, incorporating a transition model into the SA computations based on an approach like the en model, or more advanced transition codes, can result in much better agreement with experiments and recent DNS results. The third part of this work is focused on one of the most computed ows in turbulence: the channel ow. The rst goal was to con rm that an implementation of transition similar to that in the ZPG case can be e ectively used for the channel. The second focus was to nd a criterion to establish the required fetch for the fully developed region in a channel. The criterion found here is based on the asymptotic trend of the centerline velocity and its result indicates a longer distance than the usual xi 100 to 150 full channel heights found in the literature, especially at low Reynolds numbers. Finally, this study reveals the three dimensionality of the channel which appears to be impossible to avoid even for extremely high aspect ratios. Using the RSM model, the results document how the channel ow contains secondary motions, sidewall boundary layers and streamwise vortices that exchange energy and momentum between the center and the side walls. Therefore, channel ows need to be calculated as a 3D problem in order to be modeled properly or to be representative of any physical laboratory experiment.
M.S. in Mechanical and Aerospace Engineering, December 2012
Show less