Heavy particles in turbulent flows Alessandra Lanotte CNR ISAC Lecce (Italy) [email protected] with: J. Bec, L. Biferale, G. Boffetta, A. Celani, M. Cencini, S. Musacchio, F. Toschi Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Outline • Introduction Physical systems Observations Model Details of numerical simulations What we measured Short summary of some results • Core of the talk Small scales clustering Inertial scales clustering Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Where do we find heavy particles? Formation of planetesimals in the solar system (A. Bracco et al. Phys. Fluids 2002) In clouds, dust storms, fires volcano eruption.. (see e.g. K. Sassen, Nature 2005) Control of combustion processes in diesel engines (see T.Elperin et al. nlin.CD/0305017) Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte What can we observe/measure in a lab? Lagrangian turbulence has always suffered the lack of accurate space & time measurements now particles can be accurately tracked ! QuickTime™ and a Video decompressor are needed to see this picture. State-of-the-art Lagrangian experiments (tracers) From Cornell group: frame rate : 1000fps; 4x4 cm area. Istituto di Scienze dell’Atmosfera e del Clima Ott & Mann exp. at Risø, 3D PTV - Re 300 Pinton exp. at ENS, Doppler track. Re =740 Bodenshatz exp. at Cornell, fast CCD Re =1000 Alessandra Lanotte Heavy particles in wind tunnel turbulence Z. Warhaft experiment at Cornell Re 250 water droplets <d> = 20 micron High-speed camera: 2D frames Sampling time 1/100 then also other experiments in complex geometries: e.g. channel flows,.. Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte The Model finite size impurities of size much a much heavier than the fluid p f smaller than the flow dissipative scale particle Reynolds number low Re a a | V - u | 1 very dilute suspension : no role of collisions no back reaction on the flow Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Simplified equations Under previous assumption we can simplify original eqs: (M. Maxey & J. Riley, Phys Fluids 1983) X Parameters: dragnumber Stokesonly time Stokes --> Stokes (water in air b=0.001) Density ratio Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Something we know about inertia… (since Maxey, Eaton, Fessler, Squires, …) 1. Ejection of heavy particles from vortices --> experience smaller acceleration 2. Particle have finite response time to fluid fluctuations --> smoothing and filtering of fast time scales 3. Very strong concentration fluctuation --> particle distribute on clusters Try to understand physical mechanisms and identify relevant parameters for statistical description… Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte A large numerical “experiment” To start with the simplest situation To have good statistics To build up a database for common use The lab Istituto di Scienze dell’Atmosfera e del Clima particles in the flow box Alessandra Lanotte Details about the DNS Lagrangian Particles with 15 Lagrangian Tracers Initial conditions particles and tracers injected randomly & homogeneously with initial veloc. = fluid veloc. Re= 65, 105, 185 Pseudo Spectral Code, MPI Normal viscosity Istituto di Scienze dell’Atmosfera e del Clima STATISTICS TRANSIENT (1-2 T)+BULK ( 3-4 T) 3 3 3 3 N 512 256 128 Tot #particles 120Millions 32Millions 4Millions Fast 0.1 500.000 250.000 32.000 Slow 10 7.5Millions 2Millions 250.000 Stoke/Lyap (15+1)/(32+1) (15+1)/(32+1) 15+1 Traject. Length 900 +2100 756 +1744 600+ 1200 Disk usage 1TB 400GB 70GB Alessandra Lanotte How long do we wait for the stationary mass distribution? St=0.9 St=1.6 St=0.48 St=3.3 St=0.27 St=0.16 St=0 Coarse-grained mass in the j-th cell of side l=2x Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Just a quick overview about few things: Acceleration Conditioned analysis Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Why study acceleration ? Urban reshape, Old Shangai Steel factory, Taranto Acceleration is relevant for Lagrangian Stochastic Models for relative dispersion (see e.g. Sawford, Ann. Rev. Fluid Mech. 2001) Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Acceleration for tracers Tracers acceleration can be very well described in terms of the multifractal model Phenomenological model for small scale fluctuations a P(a) 1024^3 DNS Multifractal K41 prediction (Biferale, Boffetta , Celani, Devenish, AL, Toschi 2004) Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Acceleration for heavy particles Increase Re Increase St Two coexisting effects No simple phenomenological model for particles at varying preferential concentration low Reynolds St filtered dynamics at higher St Stokesatand numbers ! Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Comparison with experiments St=0.15 St=0.09 Water drops in air: clearly polydisperse flow ! A. Gylfason, S. Ayyalasomayajula, E. Bodenschatz, Z. Warhaft, PRL submitted 2006 Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Particles and 3d flow structures hyperbolic non-hyperbolic Such effect is clearly evident by looking St =0.16 at the fluid acceleration St = 0.8 conditioned on particle positions a(X,t) (Bec, Biferale, Boffetta, Celani, Cencini, AL, Pnonhyper Musacchio & Toschi 2006) St = 3.3 White: non-hyper regions Black: hyperbolic regions Particles preferentially concentrate in hyperbolic regions Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Summarising Acceleration statistics depends on two mechanisms: 1. Preferential concentrat. of particles effective at small St 2. Filtering due to particles response time effective at large St A very small amount of inertia expels particles out of intense structures: strong correlation with flow at small St; at larger St, because of filtering, particles can not follow the flow: no correlation with flow at larger St Can we better understand clustering ? Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte One motivation Strong particle concentration fluctuations have an impact on climate in different ways Desert dusts are particularly active ice-forming agents. They can affect clouds formation. Reflective power of the atmosphere due to aerosols scattering and absorption is crucial for climatological models Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Rain droplets formation due to clustering (warm) cloud large scale L=100m; Rain drop size 2mm coalescence dissipative scale = 1mm; Re=107 Droplet size 0.02mm condensation preferential concentration + gravity CCN size 0.2-2micron nucleation Enhanced collision rates may explain rapid rain formation Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Only a small scale feature? Particle clusters & voids are observed both in the dissipative and in inertial range Slice of width ≈ 2.5. Particles with St = 0.58; R = 185 Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Observables at small scales r < Space density of particles pairs (useful for collisions, pair dynamics) Probability to find 2 particles at a distance smaller than r r is the correlation dimension (Grassberger 1983 ; Hentschel Procaccia, 1983) Another common observable is the radial distribution function g(r) It is O(1) for tracers, it diverges as r--> 0 for inertial particles (or in compressible flows). Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Probability and D2 • Velocity is smooth: we expect fractal distribution (with power law tails) • At these scales, the only relevant time scale is thus everything should depend on St & Re only Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Shape of correlation dimension D2 • Optimal Stokes number for maximal clusterization • No Reynolds dependence (as in Collins & Keswani 2004) • Similar behaviour at higher order Dq • Particles positions correlate with low values of acceleration (for 2d flows Chen, Goto, Vassilicos 2006) Maximum of clustering seems to be connected to preferential concentration, confirming classical scenario Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte What happens at larger scales < r < L? Can particles of Stokes time feel effects of time scales tr>> ? How do particles distribute out of vortical regions? What are the proper parameters to describe the mass distribution? Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Inertial range observables Probability Distribution Function of the coarse-grained particle density: r Given N particles, we compute number density of particles within a cell of scale r, weighting each cell with the mass it contains: Quasi-Lagrangian measure a natural measure to reduce finite N effects at <<1 due to voids Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Quasi-Lagrangian mass density r=L/16 Tracers behave according to uniform Poisson distribution Particle show deviations, already there for very small such deviations become stronger with Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Algebraic tails at low density <<1 we have (tracers limit, uniform) St (non zero prob. to have empty areas) These empty regions can play a relevant role in many physical issues Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte How do we understand this PDFs? Particles should not distribute self-similarly i.e. Deviations from a uniform distribution are not scale-invariant (Balkovsky, Falkovich & Fouxon 2001) No simple rescaling of the mass distributions We note however that for the mass PDF these two limits are equivalent: • fixed and r • fixed r and ∞ (large observation scale) small inertia) Both limits give a uniform particle distribution. So… Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte So there could be a parameter, rescaling the mass distribution , which relates Stokes times and observations scales r At scale r, the eddy-turn-over time scale is r=-1/3r2/3, in analogy with dissipative scales, we could define: Is this time scale relevant particle clustering in the inertial range? Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Unfortunately not so simple! This simple analogy works in synthetic flows: e.g. Kraichnan flows • no time correlation • no spatial structures • no large scale-sweeping (Bec, Cencini & Hillerbrand 2006) But it does not work in real turbulence where all these features are present… X Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte A different observation Effective compressibility good for r<< for St<<1 [Maxey (1987)] particle flow [Balkovsky, Falkovich & Fouxon (2001)] Suppose the argument remains valid also for finite r & This is the contraction- rate of a particle volume of of size r and Stokes time Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Numerical Results Collapse of the coarse-grained mass PDF for different values of Uniformity is recovered going to the large scales But very slowly Non-dimensional contraction rate Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Deviation from uniformity: 2nd moment can give some better information Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Conclusions We gave a description of particle clustering for moderate St and moderate Re200 numbers 1. scales r <scales 2. clustering clustering at at small inertial range < r < L The only relevant number for particle dynamics is St=/ concentration fluctuations are relevant also for the inertial range scales Particles concentrate onto a multi-fractal set, whose dimension depends on the Stokes number only (or uniformity mass distribution is recovered very slowly just veryofweakly depends on Reynolds) at large scale Optimal finite Stokes number for clusterization: St ~ 0.6 if the(unpredictable..) contraction rate , and not Str, is the proper number to rescale mass statistics ----> sweeping is important This global picture is the same as in smooth random flow (Bec, Biferale, Cencini, AL, Musacchio, Toschi PRL submitted 2006) (see Bec 2005; Bec, Celani, Cencini,Musacchio 2005) Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Perspectives A better understanding of the statistics of fluid acceleration (rather than vorticity) seems crucial to understand clustering Conversely inertial particles can be used as probes for acceleration properties Larger Re studies are necessary to confirm the picture Currently performing DNS to study rain drops growth Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte A common database : http://cfd.cineca.it/ END Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Particles with different inertia inside a vortex St=3.31 St=0 Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte Clusters & voids 2d slice (512x512x4) at Stokes 0.16 (blue) 0.8 (red) 1.33 (green) Istituto di Scienze dell’Atmosfera e del Clima Alessandra Lanotte