Direct numerical simulations of two-dimensional (2D) and 3-D unsteady flow around a square cylinder for moderate Reynolds numbers (Re=150–500) are performed, employing an implicit fractional step method finite-volume code with second-order accuracy in space and time. The simulations, which are carried out with a blockage ratio of 5.6%, indicate a transition from 2-D to 3-D shedding flow between Re=150 and Re=200. Both spanwise instability modes, A and B, are present in the wake transitional process, similar to the flow around a circular cylinder. However, seemingly in contrast to a circular cylinder, the transitional flow around a square cylinder exhibits a phenomenon of distinct low-frequency force pulsations (Re=200–300). For 3-D simulations, the Strouhal number and the mean drag coefficient are in general agreement with existing experiments. Between Re=300 and 500, an increase in the spanwise coupling of fluctuating forces is indicated. The influence of the spanwise aspect ratio using periodic boundary conditions, a finer grid, and a finer time step is also investigated.

Large eddy simulation of flow past a rigid prism of a square cross section with one side facing the oncoming flow at Re=2.2×104 is performed. An incompressible code is used employing an implicit fractional step method finite volume with second-order accuracy in space and time. Three different subgrid scale models: the Smagorinsky, the standard dynamic, and a dynamic one-equation model, are applied. The influence of finer grid, shorter time step, and larger computational spanwise dimension is investigated. Some global quantities, such as the Strouhal number and the mean and rms values of lift and drag, are computed. A scheme for correcting the global results for blockage effects is presented. By comparison with experiments, the results produced by the dynamic one-equation one give better agreement with experiments than the other two subgrid models. [S0098-2202(00)01001-4]

No abstract

No abstract

SUMMARYFlows over two tandem cylinders were analysed using the newly developed collocated unstructured computational uid dynamics (CUCFD) code, which is capable of handling complex geometries. A Reynolds number of 100, based on cylinder diameter, was used to ensure that the ow remained laminar. The validity of the code was tested through comparisons with benchmark solutions for ow in a lid-friven cavity and ow around a single cylinder. For the tandem cylinder ow, also mesh convergence was demonstrated, to within a couple of percent for the RMS lift coe cient.The mean and uctuating lift and drag coe cients were recorded for centre-to-centre cylinder spacings between 2 and 10 diameters. A critical cylinder spacing was found between 3.75 and 4 diameters. The uctuating forces jumped appreciably at the critical spacing. It was found that there exists only one reattachment and one separation point on the downstream cylinder for spacings greater than the critical spacing.The mean and the uctuating surface pressure distributions were compared as a function of the cylinder spacing. The mean and the uctuating pressures were signiÿcantly di erent between the upstream and the downstream cylinders. These pressures also di ered with the cylinder spacing.

Int. J. of Heat and Fluid Flow, Vol. 30(5), pp. 1016-1025, 2009 The present paper gives an analysis of fully developed channel flow at Reynolds number of Re = u τ δ/ν = 4000 based on the friction velocity, u τ , and half the channel height, δ. Since the Reynolds number is high, the LES is coupled to a URANS model near the wall (hybrid LES-RANS) which acts as a wall model. It it found that the energy spectra is not a good measure of LES resolution; neither is the ratio of the resolved turbulent kinetic energy to the total one (i.e. resolved plus modelled turbulent kinetic energy). It is suggested that two-point correlations are the best measures for estimating LES resolution. It is commonly assumed that SGS dissipation takes place at high wavenumbers. Energy spectra of the fluctuating velocity gradients show that this is not true; the major part of the SGS dissipation takes place at low to midrange wavenumbers. Furthermore, the energy spectra of the fluctuating velocity gradients reveals that the accuracy of the predicted velocity gradients at the highest resolved wavenumbers is very poor.

SUMMARYA hybrid LES-RANS modelling approach is proposed. RANS is used in the near wall regions (y + . 60), and the turbulence is modelled with a k-! model. LES is used in the remaining part of the ow, and the SGS turbulence is modelled with a one-equation ksgs model. The same continuity and momentum equations are solved throughout the domain, the only di erence being that the turbulent viscosity is taken from the k-! model in the RANS region, and from the one-equation ksgs model in the LES region. The new modelling approach is applied to two incompressible ow test cases. They are fully developed ow in a plane channel and the ow over a 2D-hill in a channel.

No abstract