Implicit LES of turbulent, separated flow: wall-mounted hump configuration

Abstract

Direct simulations (ILES) of turbulent, separated flow over the wall-mounted hump configuration is conducted to investigate the physics of separated flows. A chord-based Reynolds number of Rec = 47,500 is set up, with a turbulent inflow of Re θ = 1,400 (θ/c = 3%). FDL3DI, a code that solves the compressible Navier-Stokes equations using highorder compact-difference scheme and filter, with the standard recycling/rescaling method of turbulence generation, is used. Two different configurations of the upper-wall are analyzed, and results are compared with both a higher Rec (= 936,000, Re θ = 7,200, θ/c = 0.77%) experiment for major flow features, and RANS (k-ω SST) results. A lower Rec allows for DNS-like mesh resolution, and an adequately wide span. Both ILES and RANS show earlier separation and delayed reattachment compared to experiment, and significantly higher skin friction in the forebody of the hump, as expected. The upper-wall shape only influences the pressure distribution over the hump. Results from this study are being used to setup higher Rec (lower θ/c) ILES.

A Numerical and Experimental Investigation of Flow Separation Control over a Wall-Mounted Hump Model

“…In the flow deceleration region and x/c < 0.1 regions, the simulations agree well with the experiments but overpredict the skin friction coefficient in the favorable pressure gradient region (0.1 < x/c < 0.5). Skin friction prediction in the favorable gradient region was problematic for most of the computational methods (Large Eddy Simulation, 12 Implicit Large Eddy Simulation and Reynolds Averaged Navier-Stokes Simulations, 13 Lattice Boltzmann Method, 14 and see Ref [15] for other CFD methods) that computed the reference case, predicting considerably higher skin friction coefficient. There are Cf anomalies near x/c = -0.1, 0.45, 0.9, and, 1.1.…”

Wall--modeled Large Eddy Simulation of Flow Over a Wall--mounted Hump

“…The LES simulations of You, Wang & Moin 5 yielded good predictions of the reattachment location and Reynolds stresses. Some of the computational studies for this configuration were performed by Saric et al, 6 Morgan et al, 7 Rumsey, 8 Balakumar 9 and Krishnan et al 10 More recent computations include those by Avdis et al, 11 Franck & Colonius, 12 Yeh et al, 13 Sekhar et al, 14 Woodruff, 15 Duda & Fares 16 and Kalsi & Tucker. 17 Park 18 performed WMLES using an equilibrium and non-equilibrium wall model for the same NASA hump configuration and obtained reasonable agreement with experiment.…”