AIP Publishing

The changes in the configuration of a FENE bead-spring chain in a direct numerical simulation of turbulent channel flow and in some simple rheological flows are examined. Unraveling occurs both in uniaxial and shear flows, but the uniaxial flow is more effective. A vortex with a large rotation rate perpendicular to the principal strain of a uniaxial flow has only a minor retarding effect while a small rotation rate delays the unraveling substantially. In a turbulent flow, the chain unravels the most in the viscous sublayer, to about 90% of its fully extended length. It aligns at a 7° angle with the direction of mean flow. In the buffer zone, it unravels and coils up and takes different orientations at different times. Outside the wall region, the chain assumes a coiled configuration. The unraveling of the chain strongly depends on the relaxation time of the chain normalized with the wall shear rate, λ+. A value of λ+=10 exhibits strong unraveling while very weak unraveling is observed below λ+=1.

The subgrid-scale closure method developed by Germano et al. is modified by use of a least squares technique to minimize the difference between the closure assumption and the resolved stresses. This modification removes a source of singularity and is believed to improve the method’s applicability.

The hydrodynamic force on a spherical drop that undergoes a translational acceleration in an unbounded fluid at low Reynolds number is considered. The force involves a memory-integral contribution that is not of the familiar form for a solid sphere. This result, in conjunction with the prior results of Lawrence and Weinbaum [J. Fluid Mech. 171, 208 (1986)] for a nonspherical particle, suggest that the form of the force law for a solid sphere is a very special case that is invalidated if there are any departures in either rigidity or shape from a solid sphere. In this Brief Communication the force on a spherical drop is evaluated for a number of limiting cases, after transforming the result from the Fourier-transform domain in which it is derived to the time domain.

A free viscous film is subject to van der Waals attractions that lead to film rupture. Long-wave asymptotics is used to derive approximate equations that govern the unstable evolution of the film. The solution of the nonlinear evolution equation is then considered using bifurcation techniques leading to an estimate for the nonlinear rupture time.  

Several situations have been studied in which a fluid–vapor or fluid–fluid interface ruptures, using molecular dynamics simulations of 3000 to 20 000 Lennard-Jones molecules in three dimensions. The cases studied are the Rayleigh instability of a liquid thread, the burst of a liquid drop immersed in a second liquid undergoing shear, and the rupture of a liquid sheet in an extensional flow. The late stages of the rupture process involve the gradual withdrawal of molecules from a thinning neck, or the appearance and growth of holes in a sheet. In all cases, it is found that despite the small size of the systems studied, tens of angstroms, the dynamics is in at least qualitative accord with the behavior expected from continuum calculations, and in some cases the agreement is to within tens of percent. Remarkably, this agreement occurs even though the Eulerian velocity and stress fields are essentially unmeasurable—dominated by thermal noise. The limitations and prospects for such molecular simulation techniques are assessed.

Based on the observation of vortex splitting in the laminar wake of thin flat plates placed parallel to the flow, an investigation on the consequences of such events for the von Kármán vortex street in the wake of circular cylinders was carried out. It was found that a ‘‘design break line’’ of vortex axes can lead to the decoupling of a wake flow from the always present disturbances deriving from the ends. The decoupling gives rise to parallel vortex shedding of a slightly higher frequency, instead of the oblique or slanted vortex shedding at a lower frequency usually observed.

Experimental measurements of the frequency spectra and frequency cross-spectra of the wall pressure fluctuations beneath a turbulent boundary layer were made in a low-noise flow facility. The data, taken over a range of flow speeds, clearly display a dimensionless frequency (ωδ/uτ=50) at which the spectra achieve a maximum and a low-frequency range with an approximately ω2 rolloff. The scaling laws for the low-, mid-, and high-frequency regions of the spectrum are established. The cross-spectral data, obtained over a range of streamwise separations (0.21≤ξ/δ≤16.4), allow for the computations of the decay Γ(ξ,ω) and convection velocity Uc(ξ,ω) of the wall pressure field. These data show the existence of two distinct wave number groups: a high wave number group that scales on the similarity variable k1ξ=ωξ/Uc(ξ,ω) associated with turbulent sources in the log region of the boundary layer, in which eddies decay in proportion to their size, and a low wave number group that defines the cutoff for the large-scale turbulence contributors in the outer region of the boundary layer. The convection velocity data support the conjecture that the major turbulent contributions to the low and high wave number groups come from the outer and inner layers, respectively. These new results, when examined collectively, firmly establish the spectral features of the wall pressure fluctuations, including the low-frequency range, which is highly sensitive to (passive) structures in the outer flow. The locations for the turbulent sources of the wall pressure field are proposed.

Detailed computations were made of energy transfer among the scales of motion in incompressible turbulent fields at low Reynolds numbers generated by direct numerical simulations. It was observed that although the transfer resulted from triad interactions that were nonlocal in k space, the energy always transferred locally. The energy transfer calculated from the eddy-damped quasinormal Markovian (EDQNM) theory of turbulence at low Reynolds numbers is in excellent agreement with the results of the numerical simulations. At high Reynolds numbers the EDQNM theory predicts the same transfer mechanism in the inertial range that is observed at low Reynolds numbers, i.e., predominantly local transfer caused by nonlocal triads. The weaker, nonlocal energy transfer is from large to small scales at high Reynolds numbers and from small to large scales at low Reynolds numbers.

Most subgrid-scale (SGS) models for large-eddy simulations (LES) are absolutely dissipative (that is, they remove energy from the large scales at each point in the physical space). The actual SGS stresses, however, may transfer energy to the large scales (backscatter) at a given location. Recent work on the LES of transitional flows [Piomelli et al., Phys. Fluids A 2, 257 (1990)] has shown that failure to account for this phenomenon can cause inaccurate prediction of the growth of the perturbations. Direct numerical simulations of transitional and turbulent channel flow and compressible isotropic turbulence are used to study the backscatter phenomenon. In all flows considered roughly 50% of the grid points were experiencing backscatter when a Fourier cutoff filter was used. The backscatter fraction was less with a Gaussian filter, and intermediate with a box filter in physical space. Moreover, the backscatter and forward scatter contributions to the SGS dissipation were comparable, and each was often much larger than the total SGS dissipation. The SGS dissipation (normalized by total dissipation) increased with filter width almost independently of filter type. The amount of backscatter showed an increasing trend with Reynolds number. In the near-wall region of the channel, events characterized by strong Reynolds shear stress correlated fairly well with areas of high SGS dissipation (both forward and backward). In compressible isotropic turbulence similar results were obtained, independent of fluctuation Mach number.

The dynamic subgrid-scale eddy viscosity model of Germano et al. [Phys. Fluids A 3, 1760 (1991)] (DSM) is modified by employing the mixed model of Bardina et al. [Ph.D dissertation, Stanford University (1983)] as the base model. The new dynamic mixed model explicitly calculates the modified Leonard term and only models the cross term and the SGS Reynolds stress. It retains the favorable features of DSM and, at the same time, does not require that the principal axes of the stress tensor be aligned with those of the strain rate tensor. The model coefficient is computed using local flow variables. The new model is incorporated in a finite-volume solution method and large-eddy simulations of flows in a lid-driven cavity at Reynolds numbers of 3200, 7500, and 10 000 show excellent agreement with the experimental data. Better agreement is achieved by using the new model compared to the DSM. The magnitude of the dynamically computed model coefficient of the new model is significantly smaller than that from DSM.