SearcharxivSearch

arXiv subjects

Vassilios Dallas

Publications and source records attributed to Vassilios Dallas.

At least 19 recordsLinked to original sources

Relaxation and statistical equilibria in generalised two-dimensional flows

We study relaxation toward statistical equilibrium states of inviscid generalised two-dimensional flows, where the generalised vorticity $q$ is related to the streamfunction $ψ$ via $q=(-\nabla^2)^{\fracα{2}}ψ$, with the parameter $α$ controlling the strength of the nonlinear interactions. The equilibrium solutions exhibit an $α\mapsto -α$ symmetry, under which generalised energy $E_G$ and enstrophy $Ω_G$ are interchanged. For initial conditions that produce condensates, we find long-lived quasi-equilibrium states far from the thermalised solutions we derive using canonical ensemble theory. Using numerical simulations we find that in the limit of vanishing nonlinearity, as $α\to 0$, the time required for partial thermalisation $τ_{th}$ scales like $1/α$. So, the relaxation of the system toward equilibrium becomes increasingly slow as the system approaches the weakly nonlinear limit. This behaviour is also captured by a reduced model we derive using multiple scale asymptotics. These findings highlight the role of nonlinearity in controlling the relaxation toward equilibrium and that the inherent symmetry of the statistical equilibria determines the direction of the turbulent cascades.

physics.flu-dyn

Cascades transition in generalised two-dimensional turbulence

Generalised two-dimensional (2D) fluid dynamics is characterised by a relationship between a scalar field $q$, called generalised vorticity, and the stream function $ψ$, namely $q = (-\nabla^2)^\fracα{2} ψ$. We study the transition of cascades in generalised 2D turbulence by systematically varying the parameter $α$ and investigating its influential role in determining the directionality (inverse, forward, or bidirectional) of these cascades. We derive upper bounds for the dimensionless dissipation rates of generalised energy $E_G$ and enstrophy $Ω_G$ as the Reynolds number tends to infinity. These findings corroborate numerical simulations, illustrating the inverse cascade of $E_G$ and forward cascade of $Ω_G$ for $α> 0$, contrasting with the reverse behaviour for $α< 0$. The dependence of dissipation rates on system parameters reinforces these observed transitions, substantiated by spectral fluxes and energy spectra, which hint at Kolmogorov-like scalings at large scales but discrepancies at smaller scales between numerical and theoretical estimates. These discrepancies are possibly due to nonlocal transfers, which dominate the dynamics as we go from positive to negative values of $α$. Intriguingly, the forward cascade of $E_G$ for $α< 0$ reveals similarities to three-dimensional turbulence, notably the emergence of vortex filaments within a 2D framework, marking a unique feature of this generalised model.

physics.flu-dyn

Two-dimensional Rayleigh-Bénard convection without boundaries

We study the effects of Prandtl number $Pr$ and Rayleigh number $Ra$ in two-dimensional Rayleigh-Bénard convection without boundaries, i.e. with periodic boundary conditions. In the limits of $Pr \to 0$ and $\infty$, we find that the dynamics are dominated by vertically oriented elevator modes that grow without bound, even at high Rayleigh numbers and with large scale dissipation. For finite Prandtl number in the range $10^{-3} \leq Pr \leq 10^2$, the Nusselt number tends to follow the `ultimate' scaling $Nu \propto Pr^{1/2} Ra^{1/2}$, and the viscous dissipation scales as $ε_ν\propto Pr^{1/2} Ra^{-1/4}$. The latter scaling is based on the observation that enstrophy $\langle ω^2 \rangle \propto Pr^0 Ra^{1/4}$. The inverse cascade of kinetic energy forms the power-law spectrum $\hat E_u(k) \propto k^{-2.3}$, while the direct cascade of potential energy forms the power-law spectrum $\hat E_θ(k) \propto k^{-1.2}$, with the exponents and the turbulent convective dynamics in the inertial range found to be independent of Prandtl number. Finally, the kinetic and potential energy fluxes are not constant in the inertial range, invalidating one of the assumptions underlying Bolgiano-Obukhov phenomenology.

physics.flu-dyn

Bifurcation analysis of two-dimensional Rayleigh--Bénard convection using deflation

We perform a bifurcation analysis of the steady states of Rayleigh--Bénard convection with no-slip boundary conditions in two dimensions using a numerical method called deflated continuation. By combining this method with an initialisation strategy based on the eigenmodes of the conducting state, we are able to discover multiple solutions to this non-linear problem, including disconnected branches of the bifurcation diagram, without the need for any prior knowledge of the solutions. One of the disconnected branches we find contains an S-shaped curve with hysteresis, which is the origin of a flow pattern that may be related to the dynamics of flow reversals in the turbulent regime. Linear stability analysis is also performed to analyse the steady and unsteady regimes of the solutions in the parameter space and to characterise the type of instabilities.

physics.flu-dyn

The onset of zonal modes in two-dimensional Rayleigh-Bénard convection

We study the stability of steady convection rolls in 2D Rayleigh--Bénard convection with free-slip boundaries and horizontal periodicity over twelve orders of magnitude in the Prandtl number $(10^{-6} \leq Pr \leq 10^6)$ and five orders of magnitude in the Rayleigh number $(8π^4 < Ra \leq 3 \times 10^7)$. The analysis is facilitated by partitioning our modal expansion into so-called even and odd modes. With aspect ratio $Γ= 2$, we observe that zonal modes (with horizontal wavenumber equal to zero) can emerge only once the steady convection roll state consisting of even modes only becomes unstable to odd perturbations. We determine the stability boundary in the $(Pr,Ra)$-plane and observe remarkably intricate features corresponding to qualitative changes in the solution, as well as three regions where the steady convection rolls lose and subsequently regain stability as the Rayleigh number is increased. We study the asymptotic limit $\Pr \to 0$ and find that the steady convection rolls become unstable almost instantaneously, eventually leading to non-linear relaxation osculations and bursts, which we can explain with a weakly non-linear analysis. In the complementary large-$\Pr$ limit, we observe that the stability boundary reaches an asymptotic value $Ra = 2.54 \times 10^7$ and that the zonal modes at the instability switch off abruptly at a large, but finite, Prandtl number.

physics.flu-dyn

Zonal flow reversals in two-dimensional Rayleigh-Bénard convection

We analyse the nonlinear dynamics of the large scale flow in Rayleigh-Bénard convection in a two-dimensional, rectangular geometry of aspect ratio $Γ$. We impose periodic and free-slip boundary conditions in the streamwise and spanwise directions, respectively. As Rayleigh number Ra increases, a large scale zonal flow dominates the dynamics of a moderate Prandtl number fluid. At high Ra, in the turbulent regime, transitions are seen in the probability density function (PDF) of the largest scale mode. For $Γ= 2$, the PDF first transitions from a Gaussian to a trimodal behaviour, signifying the emergence of reversals of the zonal flow where the flow fluctuates between three distinct turbulent states: two states in which the zonal flow travels in opposite directions and one state with no zonal mean flow. Further increase in Ra leads to a transition from a trimodal to a unimodal PDF which demonstrates the disappearance of the zonal flow reversals. On the other hand, for $Γ= 1$ the zonal flow reversals are characterised by a bimodal PDF of the largest scale mode, where the flow fluctuates only between two distinct turbulent states with zonal flow travelling in opposite directions.

physics.flu-dyn

Transitions between turbulent states in a two-dimensional shear flow

We study the bifurcations of the large scale jets in the turbulent regime of a forced shear flow using direct numerical simulations of the Navier-Stokes equations. The bifurcations are seen in the probability density function (PDF) of the largest scale mode with the control parameter being the Reynolds number based on the friction coefficient denoted as $Rh$. As one increases $Rh$ in the turbulent regime, the PDF of the large scale mode first bifurcates from a Gaussian to a bimodal behaviour, signifying the emergence of reversals of the large scale flow where the flow fluctuates between two distinct turbulent states. Further increase in $Rh$ leads to a bifurcation from bimodal to unimodal PDF which denotes the disappearance of the reversals of the largest scale mode. We attribute the latter transition to the long-time memory that the large scale flow exhibits related to low frequency $1/f^α$ type of noise with $0 < α< 2$. We also demonstrate that a minimal model with 15 modes, obtained from the truncated Euler equation, is able to capture the bifurcations of the large scale jets exhibited by the Navier-Stokes equations.

physics.flu-dyn

Bifurcations of a plane parallel flow with Kolmogorov forcing

We study the primary bifurcations of a two-dimensional Kolmogorov flow in a channel subject to boundary conditions chosen to mimic a parallel flow, i.e. periodic and free-slip boundary conditions in the streamwise and spanwise directions, respectively. The control parameter is the Reynolds number based on the friction coefficient, denoted as $Rh$. We find that as we increase $Rh$ the laminar steady flow goes through a degenerate Hopf bifurcation with both the oscillation frequency and the amplitude of the growing mode being zero at the threshold. A reduced four-mode model captures the scalings that are obtained from the numerical simulations. As we increase $Rh$ further we observe a secondary instability which excites the largest mode in the domain. The saturated amplitude of the largest mode is found to scale as a $3/2$ power-law of the distance to the threshold which is also explained using a low-dimensional model.

physics.flu-dyn

Classification of chaotic time series with deep learning

We use standard deep neural networks to classify univariate time series generated by discrete and continuous dynamical systems based on their chaotic or non-chaotic behaviour. Our approach to circumvent the lack of precise models for some of the most challenging real-life applications is to train different neural networks on a data set from a dynamical system with a basic or low-dimensional phase space and then use these networks to classify univariate time series of a dynamical system with more intricate or high-dimensional phase space. We illustrate this generalisation approach using the logistic map, the sine-circle map, the Lorenz system, and the Kuramoto--Sivashinsky equation. We observe that a convolutional neural network without batch normalization layers outperforms state-of-the-art neural networks for time series classification and is able to generalise and classify time series as chaotic or not with high accuracy.

eess.SP

Rotationally induced coherence in turbulent kinematic dynamos

We consider rotating, kinematic dynamos at low magnetic Prandtl number $Pm$. We show that the inclusion of rotation leads to an increase in spatio-temporal coherence and a modification of the turbulent spectrum. These effects make the flow more efficient in driving the dynamo, in the sense that the energy injection rate required to reach the critical value of the magnetic Reynolds number $Rm_c$ is reduced in comparison with a non-rotating dynamo (Seshasayanan et al. 2017). For random dynamos it is known that the growth-rate would largely be determined by the spectral index of the flow at the resistive scale. Here, however, we demonstrate that the dynamo growth-rate in rotating flows is increased by the rotationally induced long-lived large scale eddies with a coherence time greater than the local turnover time. These eddies play the major role in determining the dynamo growth-rate.

physics.flu-dyn

Large-scale dynamics of magnetic helicity

In this paper we investigate the dynamics of magnetic helicity in magnetohydrodynamic (MHD) turbulent flows focusing at scales larger than the forcing scale. Our results show a nonlocal inverse cascade of magnetic helicity, which occurs directly from the forcing scale into the largest scales of the magnetic field. We also observe that no magnetic helicity and no energy is transferred to an intermediate range of scales sufficiently smaller than the container size and larger than the forcing scale. Thus, the statistical properties of this range of scales, which increases with scale separation, is shown to be described to a large extent by the zero flux solutions of the absolute statistical equilibrium theory exhibited by the truncated ideal MHD equations.

physics.flu-dyn

Triad interactions and the bidirectional turbulent cascade of magnetic helicity

Using direct numerical simulations we demonstrate that magnetic helicity exhibits a bidirectional turbulent cascade at high but finite magnetic Reynolds numbers. Despite the injection of positive magnetic helicity in the flow, we observe that magnetic helicity of opposite signs is generated between large and small scales. We explain these observations by carrying out an analysis of the magnetohydrodynamic equations reduced to triad interactions using the Fourier helical decomposition. Within this framework, the direct cascade of positive magnetic helicity arises through triad interactions that are associated with small scale dynamo action, while the occurrence of negative magnetic helicity at large scales is explained through triad interactions that are related to stretch-twist-fold dynamics and small scale dynamo action, which compete with the inverse cascade of positive magnetic helicity. Our analytical and numerical results suggest that the direct cascade of magnetic helicity is a finite magnetic Reynolds number $Rm$ effect that will vanish in the limit $Rm \to \infty$.

physics.flu-dyn

The onset of turbulent rotating dynamos at the low $Pm$ limit

We demonstrate that the critical magnetic Reynolds number $Rm_c$ for a turbulent non-helical dynamo in the low magnetic Prandtl number $Pm$ limit (i.e. $Pm = Rm/Re \ll 1$) can be significantly reduced if the flow is submitted to global rotation. Even for moderate rotation rates the required energy injection rate can be reduced by a factor more than $10^3$. This strong decrease of the onset is attributed to the reduction of the turbulent fluctuations that makes the flow to have a much larger cut-off length-scale compared to a non-rotating flow of the same Reynolds number. The dynamo thus behaves as if it is driven by laminar behaviour (i.e. high $Pm$ behaviour) even at high values of the Reynolds number (i.e. at low values of $Pm$). Our finding thus points into a new paradigm for the design of new liquid metal dynamo experiments.

physics.flu-dyn

Spectral imbalance in the inertial range dynamics of decaying rotating turbulence

Direct numerical simulations of homogeneous decaying turbulence with mild background rotation show the existence of a systematic and significant imbalance between the non-linear energy cascade to small scales and its dissipation. By starting the decay from a statistically stationary and fully developed rotating turbulence state, where the dissipation and the energy flux are approximately equal, the data shows a growing imbalance between the two until a maximum is reached when the dissipation is about twice the energy flux. This dichotomy of behaviours during decay is reminiscent of the non-equilibrium and the equilibrium regions previously reported for non-rotating turbulence [P.C. Valente, J.C. Vassilicos, Phys. Rev. Lett. {\bf 108} 214503 (2012)]. Note, however, that for decaying rotating turbulence the classical scaling of the dissipation rate $ε\propto u'^3/L$ (where $u'$ and $L$ are the root mean square fluctuating velocity and the integral length scale, respectively) does not appear to hold during decay, which may be attributed to the effect of the background rotation on the energy cascade. On the other hand, the maximum energy flux holds the scaling $Π_{max} \propto u'^3/L$ in the initial stage of the decay until the maximum imbalance is reached.

physics.flu-dyn

Forcing-dependent dynamics and emergence of helicity in rotating turbulence

The effects of large scale mechanical forcing on the dynamics of rotating turbulent flows are studied by means of numerical simulations, varying systematically the nature of the mechanical force in time. We demonstrate that the statistically stationary solutions of these flows depend on the nature of the forcing mechanism. Rapidly enough rotating flows with a forcing that has a persistent direction relatively to the axis of rotation bifurcate from a non-helical state to a helical state despite the fact that the forcing is non-helical. We find that the nature of the mechanical force in time and the emergence of helicity have direct implications on the cascade dynamics of these flows, determining the anisotropy in the flow, the energy condensation at large scales and the power-law energy spectra that are consistent with previous findings and phenomenologies under strong and weak-wave turbulent conditions.

physics.flu-dyn

Statistical equilibria of large scales in dissipative hydrodynamic turbulence

We present a numerical study of the statistical properties of three-dimensional dissipative turbulent flows at scales larger than the forcing scale. Our results indicate that the large scale flow can be described to a large degree by the truncated Euler equations with the predictions of the zero flux solutions given by absolute equilibrium theory, both for helical and non-helical flows. Thus, the functional shape of the large scale spectra can be predicted provided that scales sufficiently larger than the forcing length scale but also sufficiently smaller than the box size are examined. Deviations from the predictions of absolute equilibrium are discussed.

physics.flu-dyn

Self-organisation and non-linear dynamics in driven magnetohydrodynamic turbulent flows

Magnetohydrodynamic turbulent flows driven by random mechanical and electromagnetic external forces of zero helicities are investigated by means of direct numerical simulations. It is shown that despite the absence of helicities in the forcing, the system is attracted to self-organized helical states that exhibit laminar behaviour despite the large value of the Reynolds numbers examined. We demonstrate that the correlation time of the external forces is controlling the time spent on these states, i.e. for short correlation times the system remains in the turbulent state while as the correlation time is increased the system spends more and more time in the self-organised states. As a result, time averaged statistics can significantly be affected by the time spent on these states. These results have important theoretical implications for the understanding of the suppression of non-linearities in plasma fusion devises as well as in astrophysical observations.

physics.plasm-ph

The signature of initial conditions on magnetohydrodynamic turbulence

We demonstrate that special correlations in the initial conditions of freely evolving, homogeneous magnetohydrodynamic (MHD) turbulence can lead to the formation of enormous current sheets. These coherent structures are observed at the peak of the energy dissipation rate and are the carriers of long-range correlations despite all the non-linear interactions during the formation of turbulence. Even though the largest scale separation has been achieved at this point, these structures are coherent with a size that spans our computational domain dominating the scaling of the energy spectrum, which follows a $E \propto k^{-2}$ power law. As Reynolds number increases curling of the current sheets, due to Kelvin-Helmhotlz type instabilities and reconnection, modifies the scaling of the energy spectrum from $k^{-2}$ towards $k^{-5/3}$. This transition occurs at the highest Reynolds numbers of direct numerical simulations with resolutions up to $2048^3$ grid points. Finite Reynolds number behaviour is observed due to the initial correlations without reaching a finite asymptote for the energy dissipation rate and with an unexpected scaling between the Taylor and the integral scale Reynolds numbers, i.e. $Re_λ\propto Re^{2/3}$. Our results, therefore, demonstrate that even state-of-the-art numerical simulations of the highest resolution can be influenced by the choice of initial conditions and consequently they are inadequate to deduce unequivocally the fate of universality in MHD turbulence.

physics.flu-dyn