Hostname: page-component-7c8c6479df-xxrs7 Total loading time: 0 Render date: 2024-03-28T08:11:10.777Z Has data issue: false hasContentIssue false

Extreme events in turbulent flow

Published online by Cambridge University Press:  05 March 2021

H.K. Moffatt*
Affiliation:
Department of Applied Mathematics and Theoretical Physics, University of Cambridge, Wilberforce Road, CambridgeCB3 0WA, UK
*
Email address for correspondence: hkm2@damtp.cam.ac.uk

Abstract

Extreme events in turbulent flow are associated with intense stretching of concentrated vortices, intermittent in both space and time. The occurrence of such events has been investigated in a turbulent flow driven by counter-rotating propellors (Debue et al., J. Fluid Mech., 2021), and local flow structures have been identified. Interesting theoretical problems arise in relation to this work; these are briefly considered in this focus paper.

Type
Focus on Fluids
Copyright
© The Author(s), 2021. Published by Cambridge University Press

1. Introduction

A central challenge in the theory of turbulence is to resolve the precise mechanism by which energy is dissipated in the limit of very high Reynolds number ${{Re}}\gg 1$. In homogeneous turbulence, the mean rate of dissipation of energy $\epsilon$ is given by $\epsilon = \nu \left \langle \boldsymbol {\omega } ^{2}\right \rangle$. Here, $\nu$ is the kinematic viscosity of the fluid, $\boldsymbol {\omega } (\boldsymbol {x}, t)=\boldsymbol {\nabla } \times \boldsymbol {u}(\boldsymbol {x}, t)$ is the vorticity field, u is the velocity field and the angular brackets $\left \langle \cdots \right \rangle$ denote a space average. In the widely accepted, although simplistic, scenario of Kolmogorov (Reference Kolmogorov1941), the turbulence acquires its mean kinetic energy $\left \langle \boldsymbol {u}^{2}\right \rangle /2= u_{0}^{2}/2$ on a scale $\ell _{0}$ at a rate $\epsilon \sim u_{0}^{3}/\ell _{0}$; this energy cascades down through the ‘inertial range’ of scales to the Kolmogorov scale $\ell _{\nu }\sim (\epsilon /\nu ^{3})^{1/4}\sim {{Re}} ^{-3/4}\ell _{0}$, below which it is dissipated by viscosity. On dimensional grounds, the all-important parameter $\epsilon$ determines the energy spectrum $E(k)=C \epsilon ^{2/3}k^{-5/3}$ in the inertial range, where $C$ is supposedly a universal constant and k is the wavenumber. The vorticity spectrum $k^{2}E(k)$ thus rises like $k^{1/3}$ through the inertial range, peaking at a wavenumber k of order $k_{\nu }=\ell _{\nu }^{-1}$, consistent with $\left \langle \boldsymbol {\omega } ^{2}\right \rangle \sim \epsilon /\nu$. The fundamental process of vortex stretching is responsible for the cumulative intensification of vorticity on ever-decreasing length scales.

Following Kolmogorov (Reference Kolmogorov1962), it has been known for some time from direct numerical simulations that this process of vorticity intensification is extremely intermittent (see, for example, Ishihara et al. Reference Ishihara, Kaneda, Yokokawa, Itakura and Uno2007). With decreasing scale, the vorticity distribution becomes more and more concentrated in singular structures that look like highly distorted sheets or filaments of vorticity. The sheets have a natural tendency to break up into filaments by elliptic and/or Kelvin–Helmholtz instabilities (McKeown et al. Reference McKeown, Ostilla-Monico, Pumir, Brenner and Rubinstein2018). Such vortex filaments were first detected experimentally by Douady, Couder & Brachet (Reference Douady, Couder and Brachet1991), intermittent in both space and time (see also Rusaouën, Rousset & Roche (Reference Rusaouën, Rousset and Roche2017) for similar observations in liquid helium).

2. Most extreme events of local energy transfer

An important advance has now been made in the detection of near-singular structures in a turbulent flow that is driven by counter-rotating propellors in the ‘von Kármán’ configuration (Debue et al. Reference Debue, Valori, Cuvier, Daviaud, Foucaut, Laval, Wiertel, Padilla and Dubrulle2021; hereafter Reference Debue, Valori, Cuvier, Daviaud, Foucaut, Laval, Wiertel, Padilla and DubrulleDVC). These propellors drive a mean flow with a non-zero helicity that is presumably inherited by the turbulence. By the use of tomographic particle velocimetry, the authors have identified ‘extreme events of local energy transfer’, and have determined the structure of the local velocity and vorticity fields in each case. The Reynolds number based on the propellor geometry and rotation rate varied over the range $6.3\times 10^{3}$ to $3.1\times 10^{5}$; at the lower end of this range, it was possible to resolve structures on the dissipative length scale $\ell _{\nu } \sim 1.4\ \textrm {mm}$, whereas at the upper end only the inertial range was accessible.

Classification of extreme events has been based by Reference Debue, Valori, Cuvier, Daviaud, Foucaut, Laval, Wiertel, Padilla and DubrulleDVC on the non-zero invariants of the velocity-gradient tensor $S_{ij}=\partial u_{i}/\partial x_{j}$, defined by $Q(\boldsymbol {x})=- (S_{ij}S_{ji})/2, R(\boldsymbol {x})=-\text {det}[S_{ij}]$. If $M(R,Q)\equiv 27 R^{2}+4 Q^{3}<0$, the three eigenvalues of $S_{ij}$ are real, and local irrotational strain dominates over vorticity. If $M>0$, one eigenvalue is real and the other two are complex conjugates, so vorticity dominates and streamlines are locally spiral or helical in character. Figure 1(a) reproduces figure 4(a) from Reference Debue, Valori, Cuvier, Daviaud, Foucaut, Laval, Wiertel, Padilla and DubrulleDVC: in this, the tear-drop region shows where, in the $\{R,Q\}$ plane, most extreme events are found (see also their figure 5(a), not shown here). Figure 1(b) shows the basic structure of this figure, in which this plane is separated into four regions in which topologically distinct structures were identified: vortex stretching; vortex compressing; sheets; and filaments.

Figure 1. (a) Reproduction of figure 4(a) from Reference Debue, Valori, Cuvier, Daviaud, Foucaut, Laval, Wiertel, Padilla and DubrulleDVC showing the joint p.d.f. (probability density function) of $R$ and $Q$, with dark red representing the maximum probability; the tear-drop region is the location where most extreme events are found, as shown in their figure 5(a) (not shown here); (b) the four regions of the same plane in which topologically distinct structures, as indicated, were identified; the cusped curve is $M(R,Q)\equiv 27 R^{2}+4 Q^{3}=0$; (c) corresponding divisions of the $\{\beta ,\omega _{0}\}$ plane, when the local velocity field takes the idealised form $\boldsymbol {u}=(-\alpha x- (\omega _{0} y)/2,-\beta y + (\omega _{0} x)/2,(\alpha +\beta )z)$ (with $\alpha =1$), as on the axis of a Burgers-type vortex.

Reference Debue, Valori, Cuvier, Daviaud, Foucaut, Laval, Wiertel, Padilla and DubrulleDVC found that the most extreme events of local energy transfer occurred in the vortex-stretching and vortex-compressing regions. It may be helpful to interpret this finding with reference to a simple vortical flow of the form

(2.1a,b)\begin{equation} \boldsymbol{u}_{v}(\boldsymbol{x})=\frac{\omega_{0}}{2\, r^{2}}(1-\text{e}^{- r^{2}}) (-y, x, 0),\quad \boldsymbol{\omega} =\boldsymbol{\nabla} \times \boldsymbol{u}_{v}(\boldsymbol{x})= (0, 0, \omega_{0} \,\text{e}^{- r^{2}}), \end{equation}

where $r^{2} =x^{2} +y^{2}$, this vortex being stretched by the non-axisymmetric strain field

(2.2)\begin{equation} \boldsymbol{U}=(-\alpha x, -\beta y, (\alpha +\beta)z),\quad \alpha+\beta>0. \end{equation}

At high vortex Reynolds number, such a vortex remains axisymmetric at leading order (Moffatt, Kida & Ohkitani Reference Moffatt, Kida and Ohkitani1994). On the vortex axis $r=0$, the components of the matrix $\{S_{ij}\}$ are

(2.3)\begin{equation} S_{ij} =\left( \begin{matrix} -\alpha & -\omega_{0}/2 & 0 \\ \omega_{0}/2 & -\beta & 0 \\ 0 & 0 & \alpha +\beta \end{matrix}\right), \end{equation}

and the corresponding expressions for $Q$, $R$ and $M$ take the simplified form

(2.4a,b)\begin{align} Q(\alpha,\beta,\omega_{0})= -(\alpha^{2}+\beta^{2}+\alpha\beta) +\omega_{0}^{2}/4, \quad R(\alpha,\beta,\omega_{0})=-(\alpha+\beta)(\alpha\beta +\omega_{0}^{2}/4) \end{align}

and

(2.5)\begin{align} M(\alpha,\beta,\omega_{0}) =27 R^{2}+4 Q^{3}=-{{{\tfrac{1}{16}}}} [((\alpha-\beta)^{2}-\omega_{0}^{2})(8\alpha^{2}+20 \alpha\beta +8\beta^{2} +\omega_{0}^{2})]. \end{align}

Taking $\alpha =1$, figure 1(c) shows the subdivisions of the $\{\beta ,\omega _{0}\}$ plane corresponding to those of figure 1(b). As might be expected, the ‘vortex-stretching’ region is a sub-region of the half-plane $\alpha +\beta >0$.

3. Three topological structures – or three in one?

Reference Debue, Valori, Cuvier, Daviaud, Foucaut, Laval, Wiertel, Padilla and DubrulleDVC found further that their extreme events were associated with three apparently different topological structures, which they describe as ‘screw vortex’, ‘roll vortex’ and ‘U-turn’; sample streamlines are shown in their figures 7(a,b) and 8(a), respectively. They recognise that these structures ‘may correspond to a single structure seen at different times or in different frames of reference’. This is an issue that can again be probed through consideration of an explicit vortex-stretching flow. We first replace (2.2) by the modified strain field

(3.1)\begin{equation} \boldsymbol{U}_{s}=(-\alpha x f'(z), -\beta y f'(z), (\alpha +\beta)f(z)), \end{equation}

with

(3.2a,b)\begin{equation} f'(z)=1-2z^{2}/(1+z^{2}),\quad f(z)= -z +2 \tan^{-1}z. \end{equation}

This flow, which satisfies $\boldsymbol {\nabla }\boldsymbol {\cdot } \boldsymbol {U}_{s}=0$, $\boldsymbol {\nabla }\times \boldsymbol {U}_{s}\ne 0$, could be produced by a system of secondary vortices near the ‘primary vortex’ (2.1a,b). When $\alpha +\beta >0$, it gives positive stretching for $|z| < 2.33$, negative (i.e. compression) for $|z| > 2.33$. (In this way, vortex stretching may always be coupled with adjacent vortex compression, thus explaining the surprisingly high probability of ‘vortex compressing’ in the p.d.f. plot of figure 1a.) Particle paths (i.e. instantaneous streamlines) of this flow combined with the primary vortex flow (2.1a,b) starting from any given point $\boldsymbol {X}(0)$ can be computed from the associated dynamical system $\text {d} \boldsymbol {X}/\text {d}t =\boldsymbol {u}_{v}(\boldsymbol {X})+\boldsymbol {U}_{s}(\boldsymbol {X})$. Their structure depends on the chosen point $\boldsymbol {X}(0)$ and on the frame of reference.

Three examples are shown in figure 2. Here, I have chosen $\alpha =0.001, \beta =0.0005, \omega _{0}=0.2$, so that

(3.3ac)\begin{align} Q(\alpha, \beta,\omega_{0})\approx 0.01,\quad R(\alpha, \beta,\omega_{0})\approx -1.5\times 10^{-5},\quad M(\alpha, \beta,\omega_{0})\approx 4\times 10^{-6}, \end{align}

and (for $|z|<2.33$) we are indeed in the vortex-stretching regime $R<0, M>0$. Although computed from the same velocity field in the same vortex neighbourhood, these streamline patterns nevertheless look quite different: figures 2(a), 2(b) and 2(c) have structures comparable with those of Reference Debue, Valori, Cuvier, Daviaud, Foucaut, Laval, Wiertel, Padilla and DubrulleDVC's screw vortex, roll vortex and U-turn, respectively. Thus care is certainly needed in classifying such observed structures, for which vorticity is presumably a more robust topological feature than velocity.

Figure 2. Streamlines computed from the dynamical system $\text {d} \boldsymbol {X}/\text {d}t =\boldsymbol {u}_{v}(\boldsymbol {x})+\boldsymbol {U}_{s}(\boldsymbol {x})$; $\alpha =0.001, \beta =0.0005, \omega _{0}=0.2$; the vorticity contour $\omega /\omega _{0}=0.3$ is shown in brown, and, as in Reference Debue, Valori, Cuvier, Daviaud, Foucaut, Laval, Wiertel, Padilla and DubrulleDVC, streamlines are shown in blue when approaching the vortex, black when leaving it; (a) $\boldsymbol {X}(0)=(1, 0, {\pm }0.01)$ and $(1, 0, {\pm }0.2)$; (b) the same streamlines viewed in a frame of reference moving with velocity $(0, 0, -0.01); \boldsymbol {X}(0)=(1, 0, {\pm }0.2)$; (c) eight streamlines starting from points close to $\boldsymbol {X}(0)=(2.87, 0, 0)$ and plotted for dimensionless time $1000\le t \le 4000$ when they first ‘encounter’ the vortex.

Declaration of interest

The author reports no conflict of interest.

References

REFERENCES

Debue, P., Valori, V., Cuvier, C., Daviaud, F., Foucaut, J.-M., Laval, J.-P., Wiertel, C., Padilla, V. & Dubrulle, B. 2021 3D analysis of precursors to non-viscous dissipation in an experimental turbulent flow. J. Fluid Mech.CrossRefGoogle Scholar
Douady, S., Couder, Y. & Brachet, M.E. 1991 Direct observation of the intermittency of intense vorticity filaments in turbulence. Phys. Rev. Lett. 67, 983986.CrossRefGoogle ScholarPubMed
Ishihara, T., Kaneda, Y., Yokokawa, M., Itakura, K. & Uno, A. 2007 Small-scale statistics in high-resolution direct numerical simulation of turbulence: Reynolds number dependence of one-point velocity gradient statistics. J. Fluid Mech. 592, 335366.CrossRefGoogle Scholar
Kolmogorov, A.N. 1941 The local structure of turbulence in incompressible viscous fluid for very large Reynolds numbers. Dokl. Akad. Nauk SSSR 30, 299303.Google Scholar
Kolmogorov, A.N. 1962 A refinement of previous hypotheses concerning the local structure of turbulence in a viscous incompressible fluid at high Reynolds number. J. Fluid Mech. 13, 8285.CrossRefGoogle Scholar
McKeown, R., Ostilla-Monico, R., Pumir, A., Brenner, M.P. & Rubinstein, S.M. 2018 A cascade leading to the emergence of small structures in vortex ring collisions. Phys. Rev. Fluids 3, 124702.CrossRefGoogle Scholar
Moffatt, H.K., Kida, S. & Ohkitani, K. 1994 Stretched vortices – the sinews of turbulence; large-Reynolds-number asymptotics. J. Fluid Mech. 259, 241264 and 266, 371.CrossRefGoogle Scholar
Rusaouën, E., Rousset, B. & Roche, P.-E. 2017 Detection of vortex coherent structures in superfluid turbulence. Europhys. Lett. 118 (1), 14005.CrossRefGoogle Scholar
Figure 0

Figure 1. (a) Reproduction of figure 4(a) from DVC showing the joint p.d.f. (probability density function) of $R$ and $Q$, with dark red representing the maximum probability; the tear-drop region is the location where most extreme events are found, as shown in their figure 5(a) (not shown here); (b) the four regions of the same plane in which topologically distinct structures, as indicated, were identified; the cusped curve is $M(R,Q)\equiv 27 R^{2}+4 Q^{3}=0$; (c) corresponding divisions of the $\{\beta ,\omega _{0}\}$ plane, when the local velocity field takes the idealised form $\boldsymbol {u}=(-\alpha x- (\omega _{0} y)/2,-\beta y + (\omega _{0} x)/2,(\alpha +\beta )z)$ (with $\alpha =1$), as on the axis of a Burgers-type vortex.

Figure 1

Figure 2. Streamlines computed from the dynamical system $\text {d} \boldsymbol {X}/\text {d}t =\boldsymbol {u}_{v}(\boldsymbol {x})+\boldsymbol {U}_{s}(\boldsymbol {x})$; $\alpha =0.001, \beta =0.0005, \omega _{0}=0.2$; the vorticity contour $\omega /\omega _{0}=0.3$ is shown in brown, and, as in DVC, streamlines are shown in blue when approaching the vortex, black when leaving it; (a) $\boldsymbol {X}(0)=(1, 0, {\pm }0.01)$ and $(1, 0, {\pm }0.2)$; (b) the same streamlines viewed in a frame of reference moving with velocity $(0, 0, -0.01); \boldsymbol {X}(0)=(1, 0, {\pm }0.2)$; (c) eight streamlines starting from points close to $\boldsymbol {X}(0)=(2.87, 0, 0)$ and plotted for dimensionless time $1000\le t \le 4000$ when they first ‘encounter’ the vortex.