Date of Award
Doctor of Philosophy (PhD)
Jeffrey S. Jeffrey Marshall
Particle motion, clustering and agglomeration play an important role in natural phenomena and industrial processes. In classical computational fluid dynamics (CFD), there are three major methods which can be used to predict the flow field and consequently the behavior of particles in flow-fields: 1) direct numerical simulation (DNS) which is very expensive and time consuming, 2) large eddy simulation (LES) which resolves the large scale but not the small scale fluctuations, and 3) Reynolds-Averaged Navier-Stokes (RANS) which can only predict the mean flow. In order to make LES and RANS usable for studying the behavior of small suspended particles, we need to introduce small scale fluctuations to these models, since these small scales have a huge impact on the particle behavior.
The first part of this dissertation both extends and critically examines a new method for the generation of small scale fluctuations for use with RANS simulations. This method, called the stochastic vortex structure (SVS) method, uses a series of randomly positioned and oriented vortex tubes to induce the small-scale fluctuating flow. We first use SVS in isotropic homogenous turbulence and validate the predicted flow characteristics and collision and agglomeration of particles from the SVS model with full DNS computations. The calculation speed for the induced velocity from the vortex structures is improved by about two orders of magnitude using a combination of the fast multiple method and a local Taylor series expansion. Next we turn to the problem of extension of the SVS method to more general turbulent flows. We propose an inverse method by which the initial vortex orientation can be specified to generate a specific anisotropic Reynolds stress field. The proposed method is validated for turbulence measures and colliding particle transport in comparison to DNS for turbulent jet flow.
The second part of the dissertation uses DNS to examine in more detail two issues raised during developing the SVS model. The first issue concerns the effect of two-way coupling on the agglomeration of adhesive particles. The SVS model as developed to date does not account for the effect of particles on the flow-field (one-way coupling). We focused on examination of the local flow around agglomerates and the effect of agglomeration on modulation of the turbulence. The second issue examines the microphysics of turbulent agglomeration by examining breakup and collision of agglomerates in a shear flow. DNS results are reported both for one agglomerate in shear and for collision of two agglomerates, with a focus on the physics and role of the particle-induced flow field on the particle dynamics.
Number of Pages
Farajidizaji, Farzad, "Numerical Modeling Of Collision And Agglomeration Of Adhesive Particles In Turbulent Flows" (2018). Graduate College Dissertations and Theses. 926.