Numerical studies of particle clustering in circular cylinder wake flows
Abstract
The emphasis of this thesis is to investigate how inertial point particles are clustered and dispersed in the wake of a circular cylinder with an uniform free stream. The numerical simulations that directly solved Navier-Stokes equations were carried out to achieve the single-phase carrier flows in a range of Reynolds number (Re). The movement of spherical particle is one-way coupled to the fluid element and solely subjected to the viscous drag force.
The continuous study of particle concentration starts from the 2D unsteady laminar wake. The particle-cylinder impaction at the front-side cylinder induces a peculiar bow shock cluster convected downstream. The radial component of viscous drag force, as balancing the centrifugal force, alters the direction around the inflection point along particle trajectory. This physical mechanism indicates a convergent tendency that contributes to the dense concentration. In the near wake, the path memory effect along particle trajectories was found to result in the smooth clusters encompassing the local vortex cells at the upstream side. The aiding and opposing dynamics are different for particles emanating from the opposite sides of the cylinder, and lead to different concentration patterns.
As Re increases to the transition-in-wake regime, the mode A instability gives rise to the streamwise vortical braids. The presence of the secondary vortices shapes the particular clustering topology and tends to attenuate the space-averaged particle velocity. A new criterion of defining clusters and voids was proposed based on the co-variance of Voronoï volumes and local ow quantities. This approach is suitable for the present flow where inertial particles are preferentially sampled by velocity gradients. The effect of particle inertia varies non-monotonically in Stokes number (Sk) with the strongest preferential concentration at Sk = 1, as the majority of particles are located in high-strain/low-vorticity regions.
In a shear layer instability-induced turbulent wake, the finer vortices are likely to increase the mixing thus weaken the preferential concentration. The less asymmetric sampling of rotation- and strain-dominant regions indicates the non-negligible role of other mechanisms yet unknown. Another manifestation of an enhanced mixing is the smaller physics-based threshold for void scale than the conventional probability-density-based counterpart. A further investigation of the potential mechanism is possible via a scale-filtering analysis.
Has parts
Paper 1: Shi, Zhaoyu; Jiang, Fengjian; Andersson, Helge Ingolf; Strandenes, Håkon. On simulation of particle-laden wake flow. I: MekIT’19 - 10th National Conference on Computational Mechanics (Trondheim, Norway, 3-4 June 2019). International Center for Numerical Methods in Engineering (CIMNE) 2019 ISBN 978-84-949194-9-7. s. 407-418Paper 2: Shi, Zhaoyu; Jiang, Fengjian; Strandenes, Håkon; Zhao, Lihao; Andersson, Helge Ingolf. Bow Shock Clustering in Particle-laden Wetted Cylinder Flow. International Journal of Multiphase Flow 2020 ;Volum 130. s. -
Paper 3: Shi, Zhaoyu; Jiang, Fengjian; Strandenes, Håkon; Zhao, Lihao; Andersson, Helge Ingolf. Bow Shock Clustering in Particle-laden Wetted Cylinder Flow. International Journal of Multiphase Flow 2020 ;Volum 130. s. -
Paper 4: Shi, Zhaoyu; Jiang, Fengjian; Zhao, Lihao; Andersson, Helge Ingolf. Scale-dependent particle clustering in transitional wake flow. This article is awaiting publication and is therefore not included.
Paper 5: Shi, Zhaoyu; Jiang, Fengjian; Zhao, Lihao; Andersson, Helge Ingolf. Particle concentration in turbulent cylinder wake flow. This article is awaiting publication and is therefore not included.
Paper 6: Tian, Cai; Shi, Zhaoyu; Jiang, Fengjian; Andersson, Helge Ingolf. Different topologies of natural vortex dislocations in mode-A wake This article is awaiting publication and is therefore not included.