Numerical simulation (DNS and LES) of manipulated turbulent boundary layerflow over a surface-mounted fence

Citation
A. Orellano et H. Wengle, Numerical simulation (DNS and LES) of manipulated turbulent boundary layerflow over a surface-mounted fence, EUR J MEC B, 19(5), 2000, pp. 765-788
Citations number
46
Categorie Soggetti
Apllied Physucs/Condensed Matter/Materiales Science","Mechanical Engineering
Journal title
EUROPEAN JOURNAL OF MECHANICS B-FLUIDS
ISSN journal
09977546 → ACNP
Volume
19
Issue
5
Year of publication
2000
Pages
765 - 788
Database
ISI
SICI code
0997-7546(200009/10)19:5<765:NS(ALO>2.0.ZU;2-M
Abstract
Results from numerical simulations are presented for manipulated turbulent boundary layer flow over a surface-mounted fence, for a Reynolds number of Re-h = 3000 (based on fence height, h, and maximum inflow velocity, U-infin ity) First, a reference data set was provided from a Direct Numerical Simul ation (DNS) using 51.6 million grid points to resolve all the relevant spat ial scales of the flow. A Large-Eddy Simulation (LES), using 1.67 million g rid points, was validated with this reference solution and compared with ex perimental data for the same Reynolds number. Then, manipulated flow cases were investigated applying time-periodic forcing through a narrow slot upst ream of the Row obstacle. High-frequency forcing, with Str(1) = f(1)h/U-inf inity = 0.60, leads to about 10% reduction of the mean re-attachment length . A much stronger reduction of about 36% could be achieved by low-frequency forcing with Str(2) = f(2)h/U-infinity = 0.08. In the latter case, large-s cale coherent structures are created between the location of the disturbanc e and the fence, they roll over the flow obstacle (nearly unaffected) and i n rolling downstream they still grow in size until they fill out the entire height of the separation zone behind the fence. In agreement with correspo nding experiments of Siller and Fernholz in 1997 for a higher Reynolds numb er (Re-h = 10500) the optimum forcing Strouhal number seems to be related t o the low-frequency movement of the entire separation bubble and not to the instability mode of the separating shear layer. (C) 2000 Editions scientif iques et medicales Elsevier SAS (C) 2000 Editions scientifiques et medicale s Elsevier SAS.