1: \documentclass[aps,prl,twocolumn,superscriptaddress,showpacs,floatfix]{revtex4}
2: %\documentclass[aps,prl,superscriptaddress,showpacs,floatfix,preprint]{revtex4}
3:
4: \newcommand{\St}{\mathcal{S}}
5: \usepackage{graphicx}
6: \usepackage{bm}
7:
8: \begin{document}
9:
10:
11: \title{Heavy particle concentration in turbulence at dissipative and
12: inertial scales}
13:
14: \author{J.\ Bec} \affiliation{CNRS UMR6202, Observatoire de la C\^ote
15: d'Azur, BP4229, 06304 Nice Cedex 4, France.}
16:
17: \author{L.\ Biferale} \affiliation{Dip.\ Fisica and INFN, Universit\`a
18: ``Tor Vergata'', Via Ricerca Scientifica~1, 00133 Roma, Italy.}
19:
20: \author{M.\ Cencini} \affiliation{INFM-CNR, SMC Dept.\ of Physics,
21: Universit\`a ``La Sapienza", P.zzle A.~Moro~2, 00185 Roma, Italy,
22: and\\ CNR-ISC, Via dei Taurini 19, 00185 Roma, Italy.}
23:
24: \author{A.\ Lanotte} \affiliation{CNR-ISAC and INFN, Sezione di Lecce, Str.\
25: Prov.\ Lecce-Monteroni, 73100 Lecce, Italy.}
26:
27: \author{S.\ Musacchio} \affiliation{Weizmann Institute of Science,
28: Department of Complex Systems, 76100 Rehovot, Israel.}
29:
30: \author{F.\ Toschi} \affiliation{CNR-IAC, Viale del Policlinico 137,
31: 00161 Roma, Italy, and\\ INFN, Sezione di Ferrara, Via G.\
32: Saragat 1, 44100 Ferrara, Italy.}
33:
34: \begin{abstract}
35: Spatial distributions of heavy particles suspended in an
36: incompressible isotropic and homogeneous turbulent flow are
37: investigated by means of high resolution direct numerical
38: simulations. In the dissipative range, it is shown that particles
39: form fractal clusters with properties independent of the Reynolds
40: number. Clustering is there optimal when the particle response time
41: is of the order of the Kolmogorov time scale $\tau_\eta$. In the
42: inertial range, the particle distribution is no longer
43: scale-invariant. It is however shown that deviations from uniformity
44: depend on a rescaled contraction rate, which is different from the
45: local Stokes number given by dimensional analysis. Particle
46: distribution is characterized by voids spanning all scales of the
47: turbulent flow; their signature in the coarse-grained mass
48: probability distribution is an algebraic behavior at small
49: densities.
50: \end{abstract}
51: \pacs{47.27.-i, 47.10.-g}
52: \date{\today}
53:
54: \maketitle
55:
56: Understanding the spatial distribution of finite-size massive
57: impurities, such as droplets, dust or bubbles suspended in
58: incompressible flows is a crucial issue in engineering~\cite{cst98},
59: planetology~\cite{pl01} and cloud physics~\cite{ks01}. Such particles
60: possess inertia, and generally distribute in a strongly inhomogeneous
61: manner. The common understanding of this long known but remarkable
62: phenomenon of \emph{preferential concentrations} relies on the idea
63: that, in a turbulent flow, vortexes act as centrifuges ejecting
64: particles heavier than the fluid and entrapping lighter
65: ones~\cite{ef94}. This picture was successfully used (see, e.g.\
66: \cite{ef94,ck04}) to describe the small-scale particle distribution
67: and to show that it depends only on the Stokes number
68: $\St_\eta=\tau/\tau_\eta$, which is obtained by non-dimensionalizing
69: the particle response time $\tau$ with the characteristic time
70: $\tau_\eta$ of the small turbulent eddies.
71:
72: In this Letter, we confirm that this mechanism is relevant to describe
73: the particle distribution at length-scales which are smaller than the
74: dissipative scale of turbulence, $\eta$. In particular, maximal
75: clustering is found for Stokes numbers of the order of unity. However,
76: we show that stationary particle concentration experiences also very
77: strong fluctuations in the inertial range of turbulence. In analogy
78: with small-scale clustering, it is expected that for $r\gg \eta$ the
79: relevant parameter is the local Stokes number $\St_r=\tau/\tau_r$,
80: where $\tau_r$ is the characteristic time of eddies of size
81: $r$~\cite{ffs03}. Surprisingly, we present evidences that such
82: a dimensional argument does not apply to describe how particles organize
83: in the inertial range of turbulence. We show that the way they
84: distribute depends on a scale-dependent rate at which volumes are
85: contracted.
86:
87: In very dilute suspensions, the trajectory $\bm X(t)$ of small
88: spherical particles much heavier than the fluid evolve according to
89: the Newton equation~\cite{m87}
90: \begin{equation}
91: \tau \ddot{\bm X} ={\bm u(\bm X,t) - \dot{\bm X}}\,.
92: \label{eq:newton}
93: \end{equation}
94: The response time $\tau$ is proportional to the square of the
95: particles size and to their density contrast with the fluid. Here, we
96: neglect buoyancy and particle Brownian diffusion. As stressed
97: in~\cite{bff01,eklrs02}, diffusion may affect concentration of
98: particles. However we assume that the typical lengthscale below which
99: particle diffusion becomes dominant is much smaller than the
100: Kolmogorov scale $\eta$ of the fluid flow, and than any observation
101: scale considered here. In most situations, particles are so massive
102: that such an approximation is fully justified.
103:
104: The fluid velocity $\bm u$ satisfies the incompressible Navier-Stokes
105: equation
106: \begin{equation}
107: \partial_t\bm u+ \bm u\cdot\bm \nabla\bm u=-\bm \nabla
108: p+\nu\nabla^2\bm u+\bm F \;\; {\mbox {with}} \;\; \bm \nabla \cdot
109: \bm u=0\,,
110: \label{eq:ns}
111: \end{equation}
112: $p$ being the pressure and $\nu$ the viscosity; $\bm F$ is an external
113: homogeneous and isotropic force that injects energy at large scales
114: $L$ with a rate $\varepsilon=\langle\bm F\cdot\bm u \rangle$. We
115: performed direct numerical simulations of (\ref{eq:ns}) by means of a
116: pseudo-spectral code on a triply periodic box of size
117: $\mathcal{L}=2\pi$ with $128^3$, $256^3$ and $512^3$ collocation
118: points corresponding to Reynolds numbers (at the Taylor micro-scale)
119: $R_\lambda \approx 65,\,105$ and $185$, respectively. Once the flow
120: $\bm u$ is statistically stationary, particles with 15 different
121: values of the Stokes number in the range $\St_\eta \in [0.16,3.5]$
122: ($N\!=\!7.5$ millions of particles per $\St_\eta$), are seeded
123: homogeneously in space with velocity equal to that of the fluid, and
124: are evolved according to (\ref{eq:newton}) for about two large-scale
125: eddy turnover times. After this time, particle mass distribution
126: reaches a statistical steady state too. Measurements are then
127: performed. Details on the numerics and on the transient are reported in
128: \cite{noi1}.
129:
130: \begin{figure}[t!]
131: \centerline{\includegraphics[width=0.4\textwidth]{correldim}}
132: \caption{(Color online) The correlation dimension $\mathcal{D}_2$ vs
133: $\St_\eta$ for the three different $R_\lambda$. Also shown the
134: probability $\mathcal{P}$ to find particles in non-hyperbolic
135: (rotating) regions of the flow, for $Re_\lambda=185$ (multiplied
136: by an arbitrary factor for plotting purposes). $\mathcal{D}_2$
137: has been estimated taking into account also subleading terms, as
138: described in~\protect\cite{bch06}.}
139: \label{correldim}
140: \end{figure}
141: Below the Kolmogorov length-scale $\eta$ where the velocity field is
142: differentiable, the motion of inertial particles is governed by the
143: fluid strain and the dissipative dynamics leads their trajectories to
144: converge to a dynamically evolving attractor. For any given response
145: time of the particles, their mass distribution is singular and
146: generically scale-invariant with fractal properties at small
147: scales~\cite{ekr96,bff01}. In order to characterize particle clusters
148: at these scales, we measured the correlation dimension
149: $\mathcal{D}_2$, which is estimated through the small-scale algebraic
150: behavior of the probability to find two particles at a distance less
151: than a given $r$: $P_2(r)\sim r^{\mathcal{D}_2}$. The dependence of
152: $\mathcal{D}_2$ on $\St_\eta$ and $Re_\lambda$ is shown in
153: Fig.~\ref{correldim}. Notice that $\mathcal{D}_2$ depends very weakly
154: on $Re_\lambda$ in the range of Reynolds numbers here explored. A
155: similar observation was done in ~\cite{ck04}, where particle
156: clustering was equivalently characterized in terms of the radial
157: distribution function. This indicates that $\tau_\eta$, which varies
158: by more than a factor 2 between the smallest and the largest Reynolds
159: numbers considered here, is the relevant time scale to characterize
160: clustering below the Kolmogorov scale $\eta$. For all values of
161: $Re_\lambda$, a maximum of clustering (minimum of $\mathcal{D}_2$) is
162: observed for $\St_\eta\approx 0.6$. For values of $\St_\eta$ larger
163: than those investigated here, $\mathcal{D}_2$ is expected to saturate
164: to the space dimension~\cite{bch06}. For small values of $\St_\eta$,
165: particle positions strongly correlate with the local structure of the
166: fluid velocity field. A quantitative measure is also given in
167: Fig.~\ref{correldim} by the probability $\mathcal{P}$ to find
168: particles in non-hyperbolic regions of the flow, i.e.\ at those points
169: where the strain matrix has two complex conjugate eigenvalues.
170: \begin{figure}[t]
171: \centerline{\includegraphics[width=0.3\textwidth]{snapshot}}
172: \vspace{-3pt}
173: \caption{(Color online) (a) The modulus of the pressure gradient,
174: giving the main contribution to fluid acceleration, on a slice
175: $512\!\times\! 512\!\times\! 4$. B/W code low and high intensity,
176: respectively. Particle positions in the same slice are shown for
177: (b) $\St_\eta\!=\!0.16$, (c) $0.80$ and (d) $3.30$. Note the
178: presence of voids with sizes much larger than the dissipative
179: scale.}
180: \label{snapshot}
181: \end{figure}
182: Note that $\mathcal{P}$ attains its minimum for values of $\St_\eta$
183: close to the minimum of $\mathcal{D}_2$. This supports the traditional
184: view relating particle clustering to vortex ejection. As qualitatively
185: evidenced from Fig.~\ref{snapshot}, particle positions correlate with
186: low values of the fluid acceleration. This phenomenon was already
187: evidenced in \cite{noi1} where a statistical analysis of the fluid
188: acceleration at particle positions was performed, and also
189: in~\cite{cgv06} for 2D turbulent flows.
190:
191: From Fig.~\ref{snapshot}, it is clear that fluctuations in the
192: particle spatial distribution extend to scales far inside the inertial
193: range; this confirms the experimental measurements
194: of~\cite{achl02}. Moreover, for Stokes numbers of the order of unity
195: (Figs.~\ref{snapshot}c and \ref{snapshot}d), we note that the sizes of
196: voids span all spatial scales of the fluid turbulent flow, similarly
197: to what observed in 2D inverse cascade turbulence~\cite{bdg04,cgv06}.
198: To gain a quantitative insight, we consider the Probability Density
199: Function (PDF) $P_{r,\tau}(\rho)$ of the particle density
200: coarse-grained on a scale $r$ inside the inertial range, that is the
201: probability distribution of the fraction of particles in a cube of
202: size $r$, normalized by the volume $r^3$ of the cube. For tracers,
203: which are uniformly distributed, and for an infinite number of
204: particles $N \to \infty$, this PDF is a delta function on $\rho = 1$.
205: For finite $N$, the probability to have a number $n$ of particles in a
206: box of size $r$ is given by the binomial distribution and leads to a
207: closed form for the PDF of the coarse-grained density $\rho =
208: n\mathcal{L}^3/(Nr^3)$. In our settings, the typical number of
209: particles in cells inside the inertial range is several thousands.
210: Thus, finite number effects are not expected to affect the core of the
211: mass distribution $\rho\! = \!\mathcal{O}(1)$, but they may be
212: particularly severe when $\rho\ll 1$ because of the presence of
213: voids. To reduce this spurious effect, we computed the
214: quasi-Lagrangian (QL) mass density PDF $P^{(QL)}_{r,\tau}(\rho)$,
215: which corresponds to a Lagrangian average with respect to the natural
216: measure~\cite{bgh04}, and is obtained by weighting each cell with the
217: mass it contains. For statistically homogeneous distribution, QL
218: statistics can be related to the Eulerian ones by noticing that
219: $\langle \rho_r^p\rangle_{QL} = \langle \rho_r^{p+1} \rangle$
220: (see, e.g.,~\cite{bgh04}).
221:
222: Figure~\ref{pdf_rho_QLag} shows $P^{(QL)}_{r,\tau}(\rho)$, for various
223: response times $\tau$ and at a given scale ($r=\mathcal{L}/16$) within
224: the inertial range. Deviations from a uniform distribution can be
225: clearly observed, they become stronger and stronger as $\tau$
226: increases. This means that concentration fluctuations are important
227: not only at dissipative scales but also in the inertial range. A
228: noticeable observation is that the low-density tail of the PDF (which
229: is related to voids) displays an algebraic behavior
230: $P^{(QL)}_{r,\tau}(\rho)\sim \rho^{\alpha(r,\tau)}$. This means that
231: the Eulerian PDF of the coarse-grained mass density has also a
232: power-law tail for $\rho\ll 1$, with exponent $\alpha-1$. The
233: dependence of this exponent for fixed $r$ and varying the Stokes
234: number $\mathcal{S}_\eta$ is shown in the inset. For low inertia
235: ($\mathcal{S}_\eta\to0$), it tends to infinity in order to recover the
236: non-algebraic behavior of tracers. At the largest available Stokes
237: numbers, we observe $\alpha \to 1$, indicating a non-zero
238: probability for totally empty areas. Note however that $\alpha$ is
239: expected to go to infinity when $\mathcal{S}_\eta\to\infty$ with $r$
240: fixed, since a uniform distribution is expected for infinite
241: inertia. Algebraic tails are relevant to various physical/chemical
242: processes involving heavy particles. The particle distribution in low
243: density regions is an important effect to be accounted for, e.g.,
244: modelling the growth of liquid droplets by condensation of vapor onto
245: aerosol particles~\cite{bff01} and of aerosol scavenging~\cite{ks01}.
246:
247: \begin{figure}[t!]
248: \centerline{\includegraphics[width=0.47\textwidth]{pdf_rho_QLag}}
249: \vspace{-12pt}
250: \caption{(Color online) The quasi-Lagrangian PDF of the
251: coarse-grained mass density $\rho_r$ in log-log scale for
252: $\St_\eta=0.27$, $0.37$, $0.58$, $0.80$, $1.0$, $1.33$, $2.03$,
253: $3.31$ (from bottom to top) at scale $r =
254: \mathcal{L}/16=0.39$. The dashed line represents a uniform
255: distribution. Inset: exponent of the power law left tail $\alpha$
256: vs $\St_\eta$. It has been estimated from a best fit of the
257: cumulative probability which is far less noisy. Data refer to
258: $Re_\lambda=185$.}
259: \label{pdf_rho_QLag}
260: \end{figure}
261: Fixing the response time $\tau$ and increasing the observation scale
262: $r$ reproduces the same qualitative picture as fixing $r$ and
263: decreasing $\tau$. A uniform distribution is recovered in both limits
264: $r\to\infty$ or $\tau\to 0$. These two limits are actually
265: equivalent. At length-scales $r\gg\eta$ within the inertial range,
266: the fluid velocity field is not smooth: according to Kolmogorov 1941
267: theory (K41), velocity increments behave as $\delta_r u \sim
268: (\varepsilon r)^{1/3}$. K41 theory suggests that fluctuations at
269: scale $r$ are associated to time scales of the order of the turnover
270: time $\tau_r = r/\delta_r u\sim \varepsilon^{-1/3}r^{2/3}$. This
271: implies that for any finite particle response time $\tau$ and at
272: sufficiently large scales $r$, the local inertia measured by
273: $\St_r=\tau/\tau_r$ becomes so small that particles should behave as
274: tracers and distribute uniformly in space~\cite{ffs03}. Deviations
275: from uniformity for finite $\St_r$ are expected not to be
276: scale-invariant~\cite{bff01}. In particular, observations from random
277: $\delta$-correlated in time flows \cite{bch06b} suggest that particle
278: distribution should depend only on the local Stokes number
279: $\St_r$. However, as explained below, this argument does not seem to
280: apply to realistic flows.
281:
282: The presence of inhomogeneities in the spatial distribution of
283: particles is due to a dynamical competition between stretching,
284: folding, and contraction due to their dissipative dynamics. At small
285: scales where the flow is spatially smooth, these mechanisms are
286: described by the Lyapunov exponents and their finite-time
287: fluctuations, associated to the growth and contraction rates of
288: infinitesimal volumes. At larger scales, competition between particle
289: spreading and concentration is also at equilibrium. The spatial
290: distribution of particle at a given scale $r\gg\eta$ should only
291: depend on the time scale given by the inverse of the contraction rate
292: $\gamma_{r,\tau}$ of a size-$r$ blob $\mathcal{B}_r$ of particles with
293: response time $\tau$. However in the inertial range, the flow is not
294: differentiable and an approach based on Lyapunov exponents cannot be
295: used. In particular, the rate $\gamma_{r,\tau}$ should depend on
296: $r$. To estimate this contraction rate in the limit of small $\tau$,
297: we make use of Maxey's approximation~\footnote{For describing scales
298: $r>\eta$ the approximation is expected to be valid whenever
299: $\St_r\ll 1$, i.e. for essentially all the explored range of
300: $\tau$'s.} of the particle dynamics~\cite{m87}
301: \begin{equation}
302: \dot{\bm X} \approx \bm v(\bm X,t)\, \;\; {\mbox {with}} \;\; \bm v
303: = \bm u -\tau \left(\partial_t \bm u + \bm u\cdot\nabla\bm
304: u\right)\,,
305: \label{eq:maxeyapprox}
306: \end{equation}
307: meaning that particle trajectories can be approximated by those of
308: tracers in the synthetic compressible velocity field $\bm v$. The
309: inertial correction is proportional to the fluid acceleration that, in
310: turbulent flows, is itself dominated by pressure gradient. The
311: contraction rate of the blob $\mathcal{B}_r$ with volume $r^3$ is
312: $\gamma_{r,\tau}= (1/r^3)\! \int_{\mathcal{B}_r} \!\! d^3x\, \nabla
313: \!\cdot\! {\bm v}(\bm x)$. As $\nabla \!\cdot\! {\bm v} \!\simeq\!
314: \tau \nabla^2 p$, we have $\gamma_{r,\tau} \sim (\tau / r^2)\,
315: \delta_r p$, where $\delta_r p$ denotes the typical pressure increment
316: at scale $r$. This means that the scale dependence of the contraction
317: rate is directly related to the scaling properties of the pressure
318: field. Dimensional arguments suggest that $\delta_r p \sim
319: (\varepsilon r)^{2/3}$, so that the contraction rate scales as
320: $\gamma_{r,\tau} \sim \tau \varepsilon^{2/3} / r^{4/3}$. However, as
321: stressed in~\cite{gf01-ti03}, K41 scaling for pressure is observable
322: only when the Reynolds number is tremendously large (typically for
323: $Re_\lambda \gtrsim 600$). In our simulations, where $Re_\lambda < 200
324: $, pressure scaling is actually dominated by random
325: sweeping~\cite{t75}, and we observe $\delta_r p \sim U\,(\varepsilon
326: r)^{1/3}$ (as in other simulations at comparable Reynolds,
327: see~\cite{gf01-ti03}). This implies that the contraction rate is
328: $\gamma_{r,\tau} \sim \tau \varepsilon^{1/3} U / r^{5/3}$.
329:
330: Figure~\ref{collapseQL} shows $P_{r,\tau}^{(QL)}(\rho)$ for three
331: choices of $\gamma_{r,\tau}$ obtained from various sets of values of
332: $r$ and $\tau$. The collapse of the different curves strongly supports
333: that the distribution of the coarse-grained mass density depends only
334: upon the scale-dependent contraction rate $\gamma_{r,\tau}$. In
335: particular, as represented in the inset of Fig.~\ref{collapseQL}, the
336: deviations from unity of the first moment of the distribution collapse
337: for all $\St_\eta$ investigated and all scales inside the inertial
338: range of our simulation. This quantity is the same as the Eulerian
339: 2nd-order moment, giving the probability to have two particles within
340: a distance $r$. We note that particle distribution recovers
341: uniformity at large scales very slowly, much slower than if particle
342: were distributed as Poisson point-like clusters for which $\langle
343: \rho^2\rangle\!-\!1 \!\propto\! r^{-3}\! \propto
344: \gamma_{r,\tau}^{9/5}$ (also shown in the inset for comparison).
345: \begin{figure}[t!]
346: \centerline{\includegraphics[width=0.5\textwidth]{collapseQL}}
347: \vspace{-12pt}
348: \caption{(Color online) Pdf of the coarse-grained mass in the
349: inertial range for three values of the non-dimensional contraction
350: rate $\gamma_{r,\tau} \tau_\eta$. From bottom to top:
351: $\gamma_{r,\tau}=4.8\times 10^{-4}/\tau_\eta$ (different curves
352: refer to $\St_\eta = 0.16$, $0.27$, $0.37$, $0.48$),
353: $\gamma_{r,\tau}=2.1\times 10^{-3}/\tau_\eta$ (for $\St_\eta
354: =0.58$, $0.69$, $0.80$, $0.91$, $1.0$), $\gamma_{r,\tau}=7.9\times
355: 10^{-3}/\tau_\eta$ (for $\St_\eta =1.60$, $2.03$, $2.67$,
356: $3.31$). Inset: deviation from unity of the first-order QL moment
357: for $r$ within the inertial range. For comparison, the behavior
358: $\propto\gamma_{r,\tau}^{9/5}$ obtained when assuming uniformly
359: distributed pointlike clusters of particles is shown as a solid
360: line. Data refer to $Re_\lambda=185$.}
361: \label{collapseQL}
362: \end{figure}
363:
364: We presented nu\-me\-ri\-cal evi\-den\-ce that at mo\-de\-ra\-te
365: Reynolds number the distribution of heavy particles at lengthscales
366: within the inertial range is fully described in terms of a
367: scale-dependent volume contraction rate $\gamma_{r,\tau}\propto\tau /
368: r^{5/3}$. However we expect that $\gamma_{r,\tau}\propto\tau /
369: r^{4/3}$ at sufficiently large Reynolds numbers ($Re_\lambda >
370: 600$). State-of-the-art numerical simulations can not currently attain
371: these turbulent regimes, so that only experimental measurements of
372: particle distribution can help to confirm such a prediction. We have
373: seen that particle dynamics in the inertial range can be directly
374: related to the structure of the pressure field (and thus of
375: acceleration). Characterizing the distribution of acceleration is
376: thus crucial to understand particle clusters, and conversely, particle
377: distribution could be used as an experimental tool to probe the
378: spatial structure of turbulent flows.
379:
380: We acknowledge useful discussions with G.\ Boffetta, A.\ Celani and
381: A.\ Fouxon. This work has been partially supported by the EU under
382: contract HPRN-CT-2002-00300 and the Galileo programme on Trasport and
383: dispersion of impurities suspended in turbulent flows. Simulations
384: were performed at CINECA (Italy) and IDRIS (France) under the
385: HPC-Europa programme (RII3-CT-2003-506079). Unprocessed data of this
386: study are freely available from http://cfd.cineca.it.
387:
388: \begin{thebibliography}{18}
389: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
390: \expandafter\ifx\csname bibnamefont\endcsname\relax
391: \def\bibnamefont#1{#1}\fi
392: \expandafter\ifx\csname bibfnamefont\endcsname\relax
393: \def\bibfnamefont#1{#1}\fi
394: \expandafter\ifx\csname citenamefont\endcsname\relax
395: \def\citenamefont#1{#1}\fi
396: \expandafter\ifx\csname url\endcsname\relax
397: \def\url#1{\texttt{#1}}\fi
398: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
399: \providecommand{\bibinfo}[2]{#2}
400: \providecommand{\eprint}[2][]{\url{#2}}
401:
402: \bibitem{cst98}
403: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Crowe}},
404: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Sommerfeld}},
405: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Tsuji}},
406: \emph{\bibinfo{title}{Multiphase Flows with Particles and Droplets}}
407: (\bibinfo{publisher}{CRC Press}, \bibinfo{address}{New York},
408: \bibinfo{year}{1998}).
409:
410: \bibitem{pl01}
411: \bibinfo{author}{\bibfnamefont{I.}~\bibnamefont{de~Pater}} \bibnamefont{and}
412: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Lissauer}},
413: \emph{\bibinfo{title}{Planetary Science}} (\bibinfo{publisher}{Cambridge
414: University Press}, \bibinfo{address}{Cambridge}, \bibinfo{year}{2001}).
415:
416: \bibitem{ks01}
417: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Kostinski}}
418: \bibnamefont{and}
419: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Shaw}},
420: \bibinfo{journal}{J.\ Fluid Mech.} \textbf{\bibinfo{volume}{434}},
421: \bibinfo{pages}{389} (\bibinfo{year}{2001}).
422:
423: \bibitem{ef94} \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Eaton}}
424: \bibnamefont{and}
425: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Fessler}},
426: \bibinfo{journal}{Int.\ J.\ Multiphase Flow}
427: \textbf{\bibinfo{volume}{20}}, \bibinfo{pages}{169}
428: (\bibinfo{year}{1994}).
429:
430: \bibitem{ck04}
431: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Collins}}
432: \bibnamefont{and}
433: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Keswani}},
434: \bibinfo{journal}{New J.\ Phys.} \textbf{\bibinfo{volume}{6}},
435: \bibinfo{pages}{119} (\bibinfo{year}{2004}).
436:
437: \bibitem{ffs03}
438: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Falkovich}},
439: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Fouxon}},
440: \bibnamefont{and}
441: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Stepanov}}, in
442: \emph{\bibinfo{booktitle}{Sedimentation and Sediment Transport}},
443: edited by \bibinfo{editor}{\bibfnamefont{A.}~\bibnamefont{Gyr}}
444: \bibnamefont{and}
445: \bibinfo{editor}{\bibfnamefont{W.}~\bibnamefont{Kinzelbach}}
446: (\bibinfo{publisher}{Kluwer Academic Publishers},
447: \bibinfo{address}{Dordrecht}, \bibinfo{year}{2003}), pp.
448: \bibinfo{pages}{155--158}.
449:
450: \bibitem{m87}
451: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Maxey}}, \bibinfo{journal}{J.\
452: Fluid\ Mech.} \textbf{\bibinfo{volume}{174}}, \bibinfo{pages}{441}
453: (\bibinfo{year}{1987}).
454:
455: \bibitem{bff01}
456: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Balkovsky}},
457: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Falkovich}},
458: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Fouxon}},
459: \bibinfo{journal}{Phys.\ Rev.\ Lett.} \textbf{\bibinfo{volume}{86}},
460: \bibinfo{pages}{2790} (\bibinfo{year}{2001}).
461:
462: \bibitem{eklrs02}
463: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Elperin}},
464: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Kleeorin}},
465: \bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{L'vov}},
466: \bibinfo{author}{\bibfnamefont{I.}~\bibnamefont{Rogachevskii}},
467: \bibnamefont{and}
468: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Sokoloff}},
469: \bibinfo{journal}{Phys.\ Rev.} E \textbf{\bibinfo{volume}{66}},
470: \bibinfo{pages}{036302} (\bibinfo{year}{2002}).
471:
472: \bibitem{noi1} \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Bec}},
473: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Biferale}},
474: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Boffetta}},
475: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Celani}},
476: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Cencini}},
477: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Lanotte}},
478: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Musacchio}},
479: \bibnamefont{and}
480: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Toschi}},
481: \bibinfo{journal}{J.\ Fluid\ Mech.} \textbf{\bibinfo{volume}{550}},
482: \bibinfo{pages}{349} (\bibinfo{year}{2006});
483: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Cencini}},
484: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Bec}},
485: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Biferale}},
486: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Boffetta}},
487: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Celani}},
488: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Lanotte}},
489: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Musacchio}},
490: \bibnamefont{and}
491: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Toschi}},
492: \bibinfo{journal}{Journ.\ Turb.} \textbf{\bibinfo{volume}{7}},
493: \bibinfo{pages}{1} (\bibinfo{year}{2006}).
494:
495:
496:
497: \bibitem{ekr96}
498: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Elperin}},
499: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Kleeorin}},
500: \bibnamefont{and}
501: \bibinfo{author}{\bibfnamefont{I.}~\bibnamefont{Rogachevskii}}, ,
502: \bibinfo{journal}{Phys.\ Rev.\ Lett.} \textbf{\bibinfo{volume}{77}},
503: \bibinfo{pages}{5373} (\bibinfo{year}{1996}).
504:
505: \bibitem{bch06} \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Bec}},
506: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Cencini}},
507: \bibnamefont{and}
508: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Hillerbrand}}, Physica
509: D, in press (2006).
510:
511: \bibitem{cgv06}
512: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Chen}},
513: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Goto}}, \bibnamefont{and}
514: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Vassilicos}},
515: \bibinfo{journal}{J.\ Fluid\ Mech.} \textbf{\bibinfo{volume}{553}},
516: \bibinfo{pages}{143} (\bibinfo{year}{2006}).
517:
518: \bibitem{achl02}
519: \bibinfo{author}{\bibfnamefont{A.}~\bibfnamefont{Aliseda}},
520: \bibinfo{author}{\bibfnamefont{A.}~\bibfnamefont{Cartellier}},
521: \bibinfo{author}{\bibfnamefont{F.}~\bibfnamefont{Hainaux}},
522: \bibnamefont{and}
523: \bibinfo{author}{\bibfnamefont{J.C.}~\bibnamefont{Lasheras}},
524: \bibinfo{journal}{J.\ Fluid\ Mech.} \textbf{\bibinfo{volume}{468}},
525: \bibinfo{pages}{77} (\bibinfo{year}{2002}).
526:
527: \bibitem{bdg04}
528: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Boffetta}},
529: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{De~Lillo}}, \bibnamefont{and}
530: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Gamba}},
531: \bibinfo{journal}{Phys.\ Fluids} \textbf{\bibinfo{volume}{16}},
532: \bibinfo{pages}{L20} (\bibinfo{year}{2004}).
533:
534: \bibitem{bgh04}
535: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Bec}},
536: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Gaw\c{e}dzki}},
537: \bibnamefont{and}
538: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Horvai}},
539: \bibinfo{journal}{Phys.\ Rev.\ Lett.} \textbf{\bibinfo{volume}{92}},
540: \bibinfo{pages}{224501} (\bibinfo{year}{2004}).
541:
542: \bibitem{bch06b}
543: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Bec}},
544: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Cencini}}, \bibnamefont{and}
545: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Hillerbrand}}
546: (\bibinfo{year}{2006}{\natexlab{c}}), \bibinfo{note}{preprint
547: nlin.CD/0606038}.
548:
549: \bibitem{wm05}
550: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Wilkinson}}
551: \bibnamefont{and}
552: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Mehlig}},
553: \bibinfo{journal}{Europhys.\ Lett.} \textbf{\bibinfo{volume}{71}},
554: \bibinfo{pages}{186} (\bibinfo{year}{2005}).
555:
556: \bibitem{gf01-ti03} T.~Gotoh and D.~Fukayama, Phys.\ Rev.\
557: Lett.~\textbf{86}, 3775 (2001); Y.~Tsuji and T.~Ishihara, Phys.\
558: Rev.~E~\textbf{68}, 026309 (2003).
559:
560: \bibitem{t75} H.\ Tennekes, J.\ Fluid Mech.~\textbf{67}, 561 (1975).
561: \end{thebibliography}
562:
563: \end{document}
564:
565:
566:
567:
568:
569:
570:
571:
572:
573:
574:
575:
576:
577:
578:
579:
580:
581: