Investigation Of Inertial Particle Phenomena In Homogeneous Isotropic Turbulence
This work is concerned with the study of inertial particles in homogeneous isotropic turbulent flows. In the first part we present the first detailed comparisons between experiments and direct numerical simulations (DNS) of inertial particle clustering, measured through the radial distribution function (RDF) and the correlation dimension. In a comparison of near-perfect parameter overlap, we observe good agreement between the RDF predictions of the DNS and the experimental observations. Our results provide important guidance on ways to improve future measurements. In the second part we use a high-resolution DNS at a low and a moderate Reynolds number to investigate inertial particle relative velocity statistics within the dissipation range and the inertial range of turbulence. Within the dissipation range we highlight the second-order longitudinal velocity structure function and study scaling properties as a function of Stokes number, a non-dimensional measure of particle inertia. We find clear evidence of so-called "caustics" and also find support for the existence of a critical Stokes number, S tc , below which the influence of caustics is negligible, in agreement with recently published work. In the inertial range we calculate the scaling exponents of the velocity structure function of order 2 through 8 in the longitudinal and transverse directions. We find that with increasing inertia, the longitudinal structure function scaling exponents become more intermittent-like, in contrast to well known examples of single-point inertial particle statistics such as acceleration. In addition, we find that the effect of filtering is primarily responsible for the observed behavior. In the third and final part we use DNS to investigate acceleration statistics of inertial particles. Specifically, we address the importance of biased sampling and biased filtering on the tails of the acceleration probability density function as a function of Stokes number. Our findings show that while biased sampling is the dominant effect, biased filtering is still relevant even at Stokes numbers as low as 0.2. Further, we attempt to uncover what aspects of the underlying flow are controlling particle acceleration statistics by studying the relationship between flow topology and inertial particle accelerations. This work highlights the interesting array of phenomena induced by inertia.
dissertation or thesis