We study numerically joint mixing of salt and colloids by a chaotic velocity field $mathbf{V}$, and how salt inhomogeneities accelerate or delay colloid mixing by inducing a velocity drift $mathbf{V}_{rm dp}$ between colloids and fluid particles as p
roposed in recent experiments cite{Deseigne2013}. We demonstrate that because the drift velocity is no longer divergence free, small variations to the total velocity field drastically affect the evolution of colloid variance $sigma^2=langle C^2 rangle - langle C rangle^2$. A consequence is that mixing strongly depends on the mutual coherence between colloid and salt concentration fields, the short time evolution of scalar variance being governed by a new variance production term $P=- langle C^2 abla cdot mathbf{V}_{rm dp} rangle/2$ when scalar gradients are not developed yet so that dissipation is weak. Depending on initial conditions, mixing is then delayed or enhanced, and it is possible to find examples for which the two regimes (fast mixing followed by slow mixing) are observed consecutively when the variance source term reverses its sign. This is indeed the case for localized patches modeled as gaussian concentration profiles.
We investigate experimentally the spatial distributions of heavy and neutrally buoyant particles of finite size in a fully turbulent flow. As their Stokes number (i.e. ratio of the particle viscous relaxation time to a typical flow time scale) is clo
se to 1, one may expect both classes of particles to aggregate in specific flow regions. This is not observed. Using a Voronoi analysis we show that neutrally buoyant particles sample turbulence homogeneously, whereas heavy particles do cluster. One implication for the understanding and modeling of particle laden flows, is that the Stokes number cannot be the sole key parameter as soon as the dynamics of finite-size objects is considered.
We investigate the preferential concentration of particles which are neutrally buoyant but with a diameter significantly larger than the dissipation scale of the carrier flow. Such particles are known not to behave as flow tracers (Qureshi et al., Ph
ys. Re. Lett. 2007) but whether they do cluster or not remains an open question. For this purpose, we take advantage of a new turbulence generating apparatus, the Lagrangian Exploration Module which produces homogeneous and isotropic turbulence in a closed water flow. The flow is seeded with neutrally buoyant particles with diameter 700mum, corresponding to 4.4 to 17 times the turbulent dissipation scale when the rotation frequency of the impellers driving the flow goes from 2 Hz to 12 Hz, and spanning a range of Stokes numbers from 1.6 to 24.2. The spatial structuration of these inclusions is then investigated by a Voronoi tesselation analysis, as recently proposed by Monchaux et al. (Phys. Fluids 2010), from images of particle concentration field taken in a laser sheet at the center of the flow. No matter the rotating frequency and subsequently the Reynolds and Stokes numbers, the particles are found not to cluster. The Stokes number by itself is therefore shown to be an insufficient indicator of the clustering trend in particles laden flows.
The dynamics of particles in turbulence when the particle-size is larger than the dissipative scale of the carrier flow is studied. Recent experiments have highlighted signatures of particles finiteness on their statistical properties, namely a decre
ase of their acceleration variance, an increase of correlation times -at increasing the particles size- and an independence of the probability density function of the acceleration once normalized to their variance. These effects are not captured by point particle models. By means of a detailed comparison between numerical simulations and experimental data, we show that a more accurate model is obtained once Faxen corrections are included.
We use an extended laser Doppler technique to track optically the velocity of individual particles in a high Reynolds number turbulent flow. The particle sizes are of the order of the Kolmogorov scale and the time resolution, 30 microseconds, resolve
s the fastest scales of the fluid motion. Particles are tracked for mean durations of the order of 10 Kolmogorov time scales. The fastest scales of the particle motion are resolved and the particle acceleration is measured. For neutrally buoyant particles, our measurement matches the performance of the silicon strip detector technique introduced at Cornell University cite{Voth,MordantCornell}. This reference dynamics is then compared to that of slightly heavier solid particles (density 1.4) and to air bubbles. We observe that the acceleration variance strongly depends on the particle density: bubbles experience higher accelerations than fluid particles, while heavier particles have lower accelerations. We find that the probability distribution functions of accelerations normalized to the variance are very close although the air bubbles have a much faster dynamics.
