A hybrid numerical simulation of isotropic compressible turbulence

A novel hybrid numerical scheme with built-in hyperviscosity has been developed to address the accuracy and numerical instability in numerical simulation of isotropic compressible turbulence in a periodic domain at high turbulent Mach number. The hybrid scheme utilizes a 7th-order WENO (Weighted Essentially Non-Oscillatory) scheme for highly compressive regions (i.e., shocklet regions) and an 8th-order compact central finite difference scheme for smooth regions outside shocklets. A flux-based conservative and formally consistent formulation is developed to optimize the connection between the two schemes at the interface and to achieve a higher computational efficiency. In addition, a novel numerical hyperviscosity formulation is proposed within the context of compact finite difference scheme for the smooth regions to improve numerical stability of the hybrid method. A thorough and insightful analysis of the hyperviscosity formulation in both Fourier space and physical space is presented to show the effectiveness of the formulation in improving numerical stability, without compromising the accuracy of the hybrid method. A conservative implementation of the hyperviscosity formulation is also developed. Combining the analysis and test simulations, we have also developed a criterion to guide the specification of a numerical hyperviscosity coefficient (the only adjustable coefficient in the formulation). A series of test simulations are used to demonstrate the accuracy and numerical stability of the scheme for both decaying and forced compressible turbulence. Preliminary results for a high-resolution simulation at turbulent Mach number of 1.08 are shown. The sensitivity of the simulated flow to the detail of thermal forcing method is also briefly discussed.     

Three-dimensional boundary conditions for numerical simulations of reactive compressible flows with complex thermochemistry

The Navier–Stokes characteristic boundary conditions (NSCBC) is a very efficient numerical strategy to treat boundary conditions in fully compressible solvers. The present work is an extension of the 3D-NSCBC method proposed by Yoo et al. and Lodato et al. in order to account for multi-component reactive flows with detailed chemistry and complex transport. A new approach is proposed for the outflow boundary conditions which enables clean exit of non-normal flows, and the specific treatment of all kinds of edges and corners is carefully addressed. The proposed methodology is successfully validated on various challenging multi-component reactive flow configurations.     

Accurate numerical methods for micromagnetics simulations with general geometries

In current FFT-based algorithms for micromagnetics simulations, the boundary is typically replaced by a staircase approximation along the grid lines, either eliminating the incomplete cells or replacing them by complete cells. Sometimes the magnetizations at the boundary cells are weighted by the volume of the sample in the corresponding cell. We show that this leads to large errors in the computed exchange and stray fields. One consequence of this is that the predicted switching mechanism depends sensitively on the orientation of the numerical grid. We present a boundary-corrected algorithm to efficiently and accurately handle the incomplete cells at the boundary. We show that this boundary-corrected algorithm greatly improves the accuracy in micromagnetics simulations. We demonstrate by using A. Arrott’s example of a hexagonal element that the switching mechanism is predicted independently of the grid orientation.     

Direct numerical simulation of passive scalar in decaying compressible turbulence

1IntroductionDirectnumericalsimulation(DNS)becomesanimportanttoolinrecentresearchofturbulence[1].DNSofcompressibleturbulenceismoredifficultthanthatoftheincompressibleturbulence.WhentheturbulentMachnumberisgreaterthan0.3theshockletsmayappearinthecompressibleturbulentflowfields.Thereasonandmechanismofshockletsexistencearenotclearyet.TheturbulentMachnumberinDNScannotbeveryhighwiththepresentexistingnumericalmethodsandcomputerresource.Fortheproblemofcompressibleisotropicturbulencewiththeinitia…     

Vortex locking in direct numerical simulations of quantum turbulence

Direct numerical simulations are used to examine the locking of quantized superfluid vortices and normal fluid vorticity in evolving turbulent flows. The superfluid is driven by the normal fluid, which undergoes either a decaying Taylor-Green flow or a linearly forced homogeneous isotropic turbulent flow, although the back reaction of the superfluid on the normal fluid flow is omitted. Using correlation functions and wavelet transforms, we present numerical and visual evidence for vortex locking on length scales above the intervortex spacing.     

On the generation and maintenance of waves and turbulence in simulations of free-surface turbulence

One of the challenges in numerical simulation of wave–turbulence interaction is the precise setup and maintenance of wave and turbulence fields. In this paper, we investigate techniques for the generation and suppression of specific surface wave modes, the generation of turbulence in an inhomogeneous physical domain with a wavy boundary-fitted grid, and the generation and maintenance of waves and turbulence during the complex wave–turbulence interaction process. We apply surface pressure to generate and suppress waves. Based on the solution of linearized Cauchy–Poisson problem, we derive three pressure expressions, which lead to a δ-function method, a time-segment method, and a gradual method. Numerical experiments show that these methods generate waves as specified and eliminate spurious waves effectively. The nonlinear wave effect is accounted for with a time-relaxation method. For turbulence generation, we extend the linear forcing method to an inhomogeneous physical domain with a curvilinear computational grid. Effects of force distribution and computational grid distortion are examined. For wave–turbulence interaction, we develop an algorithm to instantaneously identify specific progressive and standing waves. To precisely control the wave amplitude in a complex turbulent flow field, we further develop an energy controlling method. Finally, a simulation example of wave–turbulence interaction is presented. Results show that turbulence has unique features in the presence of waves. Velocity fluctuations are found to be strongly dependent on the wave phase; variations of these fluctuations are explained by the pressure–strain correlation associated with the wave-induced strain field.     

On the use of the discontinuous Galerkin method for numerical simulation of two-dimensional compressible turbulence with shocks

In this paper, the discontinuous Galerkin (DG) method combined with localized artificial diffusivity is investigated in the context of numerical simulation of broadband compressible turbulent flows with shocks for under-resolved cases. Firstly, the spectral property of the DG method is analyzed using the approximate dispersion relation (ADR) method and compared with typical finite difference methods, which reveals quantitatively that significantly less grid points can be used with DG for comparable numerical error. Then several typical test cases relevant to problems of compressible turbulence are simulated, including one-dimensional shock/entropy wave interaction, two-dimensional decaying isotropic turbulence, and two-dimensional temporal mixing layers. Numerical results indicate that higher numerical accuracy can be achieved on the same number of degrees of freedom with DG than high order finite difference schemes. Furthermore, shocks are also well captured using the localized artificial diffusivity method. The results in this work can provide useful guidance for further applications of DG to direct and large eddy simulation of compressible turbulent flows.     

New scaling for compressible wall turbulence

Classical Mach-number(M) scaling in compressible wall turbulence was suggested by van Driest(Van Driest E R.Turbulent boundary layers in compressible fluids.J Aerodynamics Science,1951,18(3):145-160) and Huang et al.(Huang P G,Coleman G N,Bradshaw P.Compressible turbulent channel flows:DNS results and modeling.J Fluid Mech,1995,305:185-218).Using a concept of velocity-vorticity correlation structure(VVCS),defined by high correlation regions in a field of two-point cross-correlation coefficient between a velocity and a vorticity component,we have discovered a limiting VVCS as the closest streamwise vortex structure to the wall,which provides a concrete Morkovin scaling summarizing all compressibility effects.Specifically,when the height and mean velocity of the limiting VVCS are used as the units for the length scale and the velocity,all geometrical measures in the spanwise and normal directions,as well as the mean velocity and fluctuation(r.m.s) profiles become M-independent.The results are validated by direct numerical simulations(DNS) of compressible channel flows with M up to 3.Furthermore,a quantitative model is found for the M-scaling in terms of the wall density,which is also validated by the DNS data.These findings yield a geometrical interpretation of the semi-local transformation(Huang et al.,1995),and a conclusion that the location and the thermodynamic properties associated with the limiting VVCS determine the M-effects on supersonic wall-bounded flows.     

Direct numerical simulation of a fully developed compressible wall turbulence over a wavy wall

Compressible turbulent channel flow over a wavy surface is investigated by direct numerical simulations using high-resolution finite difference schemes. The Reynolds number considered in the present paper is 3380 based on the bulk velocity, the channel half-width and the kinetic viscosity at the wall. Four test cases are simulated and analysed at Ma_m = 0.33, 0.8, 1.2, 1.5 based on the bulk velocity and the speed of sound at the wall. We mainly focus on the curvature and the Mach number effects on the compressible turbulent flows. Numerical results show that although the wavy wall has effects on the mean and fluctuation quantities, log law still exists in the distribution of the wave-averaged streamwise velocity if the roughness effects are taken into consideration in the scaling of it. Near-wall streaks are broken by the wavy surface and near-wall quasi-streamwise vortices mostly begin at the upslope of the wave and pass over the crest of it. The wavy wall makes the turbulence more active and the flow easier to be blended. From the viewpoint of turbulent kinetic budgets, curvature effects strengthen both the diffusion terms and the dissipation terms. At the same time, they change the properties of the compressibility-related terms and promote more inner energy transferring into turbulent kinetic energy. As the Mach number increases, the reattachment of the mean flow is delayed, which indicates the mean separation bubble becomes larger. Concerning the near-wall coherent structures, the vortices are more sparsely distributed with the increasing of the Mach number. For the supersonic cases, shock waves appear. Though they have little effects on the mean turbulent quantities, they change the structures of the flow fields and induce local separations at the upper wall of the channel.     

具有声激波的跨音流管道中声传播的数值方法

本文采用四阶MacCormack格式和附加四阶粘性项方法,求解具有声激波的跨音流变差分格壁管中的声传播问题,比前人结果有明显改善。本文详细介绍了这种差分方法,特别是关于截面硬式,人工粘性项和计算可靠性判据。这种方法省内存,省机时,可以在微机上实现计算。     

Conformal variables in the numerical simulations of long-crested rogue waves

The recently suggested theoretical model for highly nonlinear potential long-crested water waves is discussed, where weak three-dimensional effects are included as small corrections to exact two-dimensional equations written in terms of the conformal variables [V.P. Ruban, Phys. Rev. E 71, 055303(R) (2005)]. Some numerical results based on this theory are presented, which describe spontaneous formation of rogue waves on the deep water for different initial conditions. In particular, the given numerical examples describe: (i) nonlinear stage of the modulational instability, (ii) breathing rogue wave in a random wave field, and (iii) freak wave in a weakly crossing sea state.     

Three-dimensional finite-difference lattice Boltzmann model and its application to inviscid compressible flows with shock waves

In this paper, a three-dimensional (3D) finite-difference lattice Boltzmann model for simulating compressible flows with shock waves is developed in the framework of the double-distribution-function approach. In the model, a density distribution function is adopted to model the flow field, while a total energy distribution function is adopted to model the temperature field. The discrete equilibrium density and total energy distribution functions are derived from the Hermite expansions of the continuous equilibrium distribution functions. The discrete velocity set is obtained by choosing the abscissae of a suitable Gauss–Hermite quadrature with sufficient accuracy. In order to capture the shock waves in compressible flows and improve the numerical accuracy and stability, an implicit–explicit finite-difference numerical technique based on the total variation diminishing flux limitation is introduced to solve the discrete kinetic equations. The model is tested by numerical simulations of some typical compressible flows with shock waves ranging from 1D to 3D. The numerical results are found to be in good agreement with the analytical solutions and/or other numerical results reported in the literature.     

A turbulence characteristic length scale for compressible flows

The current RANS models are generally established and calibrated under incompressible condition and these kinds of models could succeed in predicting many features of incompressible flows. However, these models extended to the high-speed, compressible flows are always less accurate. In the paper, a compressible von Kármán length scale is proposed for compressible flows considering the variable densities. It contains no empirical coefficients and is based on phenomenological theory. In the turbulent kinetic equation, the extra unclosed terms induced by non-constant densities are treated as dissipation terms and the equation is closed algebraically via the introduction of the von Kármán length scale. The original and the proposed von Kármán length scale lead to two different kinds of SAS (scale adaption simulation) models, KDO (turbulence kinetic energy dependent only) and CKDO (compressible KDO), respectively. Compressible mixing layer with significant compressibility is studied within standard k–?, k–ω, KDO turbulence models and their compressible versions. The compressibility effects such as the reduced mixing layer thickness, growth rate and turbulence intensity can be reproduced by CKDO. The new length scale can improve the performances of the model in predicting the mixing layer thickness, stream-wise velocity and Reynolds shear stresses when the convective Mach number is 0.8. Besides, the new length scale also leads to accurate computed growth rate when the convective Mach number ranges from 0.1 to 1.0.     

A unified wall function for compressible turbulence modelling

Turbulence modelling near the wall often requires a high mesh density clustered around the wall and the first cells adjacent to the wall to be placed in the viscous sublayer. As a result, the numerical stability is constrained by the smallest cell size and hence requires high computational overhead. In the present study, a unified wall function is developed which is valid for viscous sublayer, buffer sublayer and inertial sublayer, as well as including effects of compressibility, heat transfer and pressure gradient. The resulting wall function applies to compressible turbulence modelling for both isothermal and adiabatic wall boundary conditions with the non-zero pressure gradient. Two simple wall function algorithms are implemented for practical computation of isothermal and adiabatic wall boundary conditions. The numerical results show that the wall function evaluates the wall shear stress and turbulent quantities of wall adjacent cells at wide range of non-dimensional wall distance and alleviate the number and size of cells required.     

Drag reduction of compressible wall turbulence with active dimples

Direct numerical simulations are carried out to assess the potential drag reduction of compressible turbulent flow between isothermal walls.For the sake of achieving drag reduction,the flow is actively controlled by deformable dimples lying on the bottom wall of the channel.The first stage of the procedure consists in assessing the optimum geometry of the dimples.In this regard,the lower wall is allowed to freely deform itself according to the loop of control.This method is called the smart wall approach in...     

Resolution effects and scaling in numerical simulations of passive scalar mixing in turbulence

The effects of finite grid resolution on the statistics of small scales in direct numerical simulations of turbulent mixing of passive scalars are addressed in this paper. Simulations at up to 2048³ grid points with grid spacing Δx varied from about 2 to 1/2 Batchelor scales (η_B) show that most conclusions on Schmidt number (Sc) dependence from prior work at less stringent resolution remain qualitatively correct, although simulations at resolution Δx≈η_B are preferred and will give adequate results for many important quantities including the scalar dissipation intermittency exponent and structure functions at moderately high orders. For Sc≥1, since η_B=ηSc^−1/2 (where η is the Kolmogorov scale), the requirement Δx≈η_B is more stringent than the corresponding criterion Δx≈η for the velocity field, which is thus well resolved in simulations aimed at high Schmidt number mixing. A simple argument is given to help interpret the effects of Schmidt and Reynolds numbers on trends towards local isotropy and saturation of intermittency at high Schmidt number. The present results also provide evidence for a trend to isotropy at high Reynolds number with fixed Sc=1.0. This is a new observation apparently not detected in less well resolved simulations in the past, and will require further investigation in the future.