LFPB Seminar: Multiscale analysis of inertial particledynamics in turbulence

Multiscale statistical analyses of inertial particle distributions are presented to investigate the statistical signature of clustering
and void regions in particle-laden incompressible isotropic turbulence. Three-dimensional direct numerical simulations of
homogeneous isotropic turbulence at high Reynolds number (Re_λ≳200) with up to 10^9 inertial particles are performed for
Stokes numbers ranging from 0.05 to 5.0.
A finite-time measure to quantify divergence and the rotation of the particle velocity by determining respectively the volume
change rate of the Voronoi cells and their rotation is proposed. For inertial particles the probability distribution functions (PDF)
of the divergence and of the curl deviate from that for fluid particles. Joint PDFs of the divergence and the Voronoi volume
illustrate that the divergence is most prominent in cluster regions and less pronounced in void regions. For larger volumes the
results show negative divergence values which represent cluster formation and for small volumes the results show positive
divergence values which represents cluster destruction/void formation. Moreover, when the Stokes number increases the
divergence takes larger values, which gives some evidence why fine clusters are less observed for large Stokes numbers. Finally,
the PDFs of the particle vorticity have much heavier tails compared to the fluid vorticity, and the extreme values increase
significantly with the Stokes number.
Orthogonal wavelet analysis is then applied to the computed particle number density fields. Scale-dependent skewness and
flatness values of the particle number density distributions are calculated and the influence of Reynolds number Re_λ and
Stokes number St is assessed. For St∼1.0, both the scale-dependent skewness and flatness values become larger as the scale
decreases, suggesting intermittent clustering at small scales. For St≤0.2, the flatness at intermediate scales, i.e. for scales larger
than the Kolmogorov scale and smaller than the integral scale of the flow, increases as St increases, and the skewness exhibits
negative values at the intermediate scales. The negative values of the skewness are attributed to void regions. These results
indicate that void regions at the intermediate sales are pronounced and intermittently distributed for such small Stokes
numbers.
Some perspectives for data-driven machine learning techniques are presented and first results for synthesizing preferential
concentration fields of heavy point particles are shown.
This is joint work with Thibault Oujia (Aix-Marseille U, France), Keigo Matsuda (JAMSTEC, Japan) and Katsunori Yoshimatsu (Nagoya U, Japan).
References.:
T. Oujia, K. Matsuda and K. Schneider. Divergence and convergence of inertial particles in high Reynolds number turbulence. J. Fluid Mech., 905, A14, 2020.
K. Matsuda, K. Schneider and K. Yoshimatsu. Scale-dependent statistics of inertial particle distribution in high Reynolds number turbulence. Phys. Rev. Fluids, 6, 064304, 2021.
T. Oujia, K. Matsuda and K. Schneider. Computing differential operators of the particle velocity in moving particle clouds using tessellations. Preprint, 12/2022. arXiv:2212.03580
T. Oujia, S. S. Jain, K. Matsuda, K. Schneider, J. West and K. Maeda. Neural networks for synthesizing preferential concentration of particles in isotropic turbulence. Center for Turbulence Research, Proceedings of the Summer Program 2022, Stanford University, 153-162, 2022.