betaFeedback

WorksPhysics and Astronomy

TURB-Hel: an open-access database of helically forced homogeneous and isotropic turbulence

Abstract

We present TURB-Hel, a database formed by two datasets of incompressible homogeneous and isotropic turbulence, maintained in a statistically stationary state by fully helical forcing. The aim is to provide a dataset that clearly exhibits the phenomenon of the helicity cascade from the large to the small scales generated by a large-scale forcing that breaks the mirror symmetry. This database offers the possibility to realize a wide variety of analyses of fully developed turbulence from the sub-grid scale filtering up to the validation of an a posteriori LES. TURB-Hel is available for download using the SMART-Turb portal http://smart-turb.roma2.infn.it.

Introduction

Kinetic helicity, the correlation between velocity fluctuations, uu, and vorticity fluctuations, ω\omega, is known to influence turbulent dynamics and coherent structures formation Waleffe [10] (other references) and to be connected to the breaking of the symmetry under inversion of all axes in a rotationally invariant ensemble. High levels of helicity have been associated with: a strong depletion of the kinetic energy fluxes across scales Kraichnan [7] as a consequence of the nonlinearity reduction caused by the alignment of velocity and vorticity and with low dissipation regions Moffatt [8]. Like energy, helicity is a quadratic invariant of the Euler equations but, unlike energy, it is not sign definite. In this respect, the interactions between helical Fourier modes of opposite signs (heterochiral sector) contribute to the energy transfer from large to small scales Waleffe [10] Alexakis & Biferale [2] Alexakis [1] while those with the same sign (homochiral sector) are responsible for the inverse energy transfer across scales in the inertial range Biferale et al. [4] Biferale et al. [5] Sahoo et al. [9].

Any solenoidal vector field can be decomposed into positively and negatively helical components Waleffe [10]. That is, the velocity field can we written as

u(x,t)=u+(x,t)+u−(x,t).(1)u(x,t) = u^{+}(x,t) + u^{-}(x,t) . \tag*{(1)}

The fields u+u^{+} and u−u^{-} are obtained by projecting the Fourier coefficients u^(k,t)\widehat{u}(k,t) onto basis vectors which are eigen-functions of the curl operator in Fourier space

u^±(k,t)=u±(k,t)h±(k),(2)\widehat{u}^{\pm}(k,t) = u^{\pm}(k,t)h^{\pm}(k) , \tag*{(2)}

where ik×kh±(k)=±h±(k)i k \times k h^{\pm}(k) = \pm h^{\pm}(k). The total kinetic helicity can then be calculated as

⟨u(x,t)⋅ω(x,t)⟩V=∑ku^(k,t)⋅ω^(k,t)=∑kk(u+(k,t)2−u−(k,t)2).(3)\langle u(x,t)\cdot\omega(x,t)\rangle_{V} = \sum_{k}\widehat{u}(k,t)\cdot\widehat{\omega}(k,t) = \sum_{k} k\left(u^{+}(k,t)^{2} - u^{-}(k,t)^{2}\right) . \tag*{(3)}

where ⟨⋅⟩V\langle\cdot\rangle_V denotes a spatial average.

Numerical simulations

Data has been generated by direct numerical simulation of the 3D Navier-Stokes equations on a triply periodic domain of size Lbox=2πL_{\mathrm{box}}=2\pi in each spatial direction

∂tu+u⋅∇u=−∇p+νΔu+f,(4)\partial_t\mathbf{u}+\mathbf{u}\cdot\nabla\mathbf{u}=-\nabla p+\nu\Delta\mathbf{u}+\mathbf{f}, \tag*{(4)}
∇⋅u=0,(5)\nabla\cdot\mathbf{u}=0, \tag*{(5)}

where u\mathbf{u} denotes the velocity field, pp is the pressure divided by the fluid density, which is constant, and ν\nu the kinematic viscosity. The forcing f\mathbf{f} is a random Gaussian process with zero mean active in the wavenumber band k∈[0.5,2.4]k\in[0.5,2.4] for the dataset HL2 while for HL1 the forcing ranges in the interval k∈[1.0,2.0]k\in[1.0,2.0]. It satisfies ∇⋅f=0\nabla\cdot\mathbf{f}=0, and is fully helical, f=f+\mathbf{f}=\mathbf{f}^{+}. Equations (4)-(5) were stepped forwards in time using a second-order Adams-Bashforth scheme, with the viscous term treated implicitly using an integrating factor. The spatial discretisation was implemented via the standard pseudospectral method with complete dealiasing by the two-thirds rule. Further details and mean values of key observables are summarised in Table 1.

idNNEEν\nuε\varepsilonLLτ\tauReλ\mathrm{Re}_{\lambda}η/10−3\eta/10^{-3}kmax⁡k_{\max}kmax⁡ηk_{\max}\etaΔt/τ\Delta t/\tau#
HL15126.320.0022.881.140.502407.411691.250.6056
HL210247.260.0013.331.120.503274.203401.430.6039

Table 1. Simulation parameters and key observables, where NN denotes the number of collocation points in each coordinate, EE the total kinetic energy per unit volume, ν\nu the kinematic viscosity, ε\varepsilon the dissipation rate, L=(3π/4E)∑k=1kmax⁡E(k)/k ΔkL=(3\pi/4E)\sum_{k=1}^{k_{\max}}E(k)/k\,\Delta k the integral scale, τ=L/2E/3\tau=L/\sqrt{2E/3} the large-eddy turnover time, Reλ\mathrm{Re}_{\lambda} the Taylor-scale Reynolds number, η=(ν3/ε)1/4\eta=(\nu^3/\varepsilon)^{1/4} the Kolmogorov microscale, kmax⁡k_{\max} the largest wave number after de-aliasing, Δt\Delta t the sampling interval which is calculated from the length of the averaging interval divided by the number of equispaced snapshots, and # the number of snapshots. The higher-resolved data corresponds to run 22 of Sahoo et al. [9]. Both datasets have been discussed in Capocci et al. [6] and are available for download using the SMART-Turb portal http://smart-turb.roma2.infn.it.

Time series of the total kinetic energy per unit volume, E(t)=12⟨∣u(x,t)∣2⟩VE(t)=\frac{1}{2}\langle|\mathbf{u}(\mathbf{x},t)|^2\rangle_V, and the total dissipation, ε(t)=⟨∣∇u(x,t)∣2⟩V\varepsilon(t)=\langle|\nabla\mathbf{u}(\mathbf{x},t)|^2\rangle_V, for datasets HL1 and HL2, are presented in Figure 1. The red points correspond to instances in time where full data cubes have been sampled.

Figure 1. Time evolution of the total energy (left) and the total dissipation (right). The red dots correspond to the sampled velocity-field configurations. Top panel: dataset HL2. Bottom panel: dataset HL1.

Figure 2 shows the energy spectrum E(k)=12∑k≤∣k∣<k+1∣u^(k)∣2E(k)=\frac{1}{2}\sum_{k\leq|\mathbf{k}|<k+1}|\hat{\mathbf{u}}(\mathbf{k})|^2 averaged over the ensemble of 56 and 39 velocity-field configurations that comprise the datasets of HL1 and HL2 respectively, in raw and Kolmogorov-compensated form. The energy spectrum can be decomposed into contributions from the positively and negatively helical components of the velocity field, E(k)=E+(k)+E−(k)E(k)=E^{+}(k)+E^{-}(k). In both panels of the figure, the black lines correspond to E(k)E(k), the red lines to the energy spectrum of the positively helical velocity-field fluctuations, E+(k)E^{+}(k) and the blue lines to that of the negatively helical velocity-field fluctuations, E−(k)E^{-}(k). The grey-shaded area identifies the forcing range, and the horizontal dashed line, in the compensated spectra of both panels, indicates a Kolmogorov constant CK=1.6C_K=1.6. As the forcing is positively helical, the negatively helical fluctuations must be maintained by nonlinear coupling to the positively helical fluctuations. As can be seen from the date presented in Figure 2, the velocity field is positively helical at and near the forcing scales. In both datasets, the mirror symmetry is restored towards the end of the inertial range, around k=80k=80. Only the E(k)E(k) and E+(k)E^{+}(k) scale approximately as k−5/3k^{-5/3}, while E−(k)E^{-}(k) has a shallower slope.

Figure 2. Time-averaged energy spectra, in raw (left) and Kolmogorov-compensated (right) form. The grey-shaded area indicates the forcing range, the positively helical component is shown in red, the negatively helical component in blue and the total energy spectrum in black. Top panel: dataset HL2. Bottom panel: dataset HL1: a fewer amount of wavenumbers is the consequence of the reduction in the grid point. In both datasets the dashed lines in the right panels indicate a Kolmogorov constant CK≈1.6C_K \approx1.6.

DataBase Description

TURB-Hel database is made of files extracted from the 5123512^3 HL1 and 102431024^3 HL2 simulation described in the previous section as follows:

  • During the simulation we dumped the vector potential of the velocity field in fourier space every 3000dt3000dt simulation steps. In total we stored 56 (HL1) and 39 (HL2) configurations, we call them ”cb” (from Complex B-field);

  • In order to recover the velocity fields, every configuration should be read using the HDF5 library, then a curl operation is needed to compute the velocity field in Fourier space.

  • If the velocity field is needed in real space, a backward FFT is needed.

  • In the support materials, there is a C program read_cb which performs all the above steps, and an accompanying ReadME.pdf for the details on how to compile it.

The database TURB-Hel is available for download using the SMART-Turb portal at http://smart-turb.roma2.infn.it.

Other data-sets concerning rotating turbulence (TURB-Rot Biferale et al. (2020)) and Lagrangian particle under turbulence (TURB-Lagr) are already available.

Acknowledgements

This work received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 882340).

References

  1. [1]Alexakis, A. 2017 Helic ally decomposed turbulence. Journal of Fluid Mechanics 812, 752–770.DOI
  2. [2]Alexakis, A. & Biferale, L. 2018 Cascades and transitions in turbulent flows. Physics Reports 767, 1–101.DOI
  3. [4]Biferale, L., Musacchio, S. & Toschi, F. 2012 Inverse energy cascade in three-dimensional isotropic turbulence. Physical review letters 108 (16), 164501.pubmed.ncbi.nlm.nih.gov/22680722
  4. [5]Biferale, L, Musacchio, Stefano & Toschi, F 2013 Split energy–helicity cascades in three-dimensional homogeneous and isotropic turbulence. Journal of Fluid Mechanics 730, 309–327.DOI
  5. [6]Capocci, D., Johnson, P. L., Oughton, S., Biferale, L. & Linkmann, M. 2023 New exact betchov-like relation for the helicity flux in homogeneous turbulence. Journal of Fluid Mechanics 963, R1.DOI
  6. [7]Kraichnan, Robert H 1973 Helical turbulence and absolute equilibrium. Journal of Fluid Mechanics 59 (4), 745–752.DOI
  7. [8]Moffatt, H K. 2014 Helicity and singular structures in fluid dynamics. Proceedings of the National Academy of Sciences 111 (10), 3663–3670.DOI
  8. [9]Sahoo, G., Bonaccorso, F. & Biferale, L. 2015 Role of helicity for large- and small-scale turbulent fluctuations. Phys. Rev. E 92, 051002.DOI
  9. [10]Waleffe, F. 1992 The nature of triad interactions in homogeneous turbulence. Physics of Fluids A: Fluid Dynamics 4 (2), 350–363.DOI

Paper details

vlnc:W_4ne7hctd
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents