Session Q08: Multiphase Flows: Particle-Laden Flows II

Flexible fibers in turbulent channel flow

In this paper we investigate the dynamics of small flexible fibers in turbulent channel flow. We aim at examining the translational and rotational behavior of fibers with different elongation (parameterized by the aspect ratio) and inertia (parameterized by the Stokes number) with and without fiber-fluid coupling. We use a Eulerian-Lagrangian approach based on direct numerical simulation of turbulence to describe fiber dynamics. Fibers, which are longer than the Kolmogorov length scale, are modelled as chains of sub-Kolmogorov rods connected through ball-and-socket joints that enable bending and twisting. Different mass and volume fractions are considered to investigate two-way coupling effects. Velocity, orientation and concentration statistics, extracted from simulations at shear Reynolds number Re_τ=300 are presented to give insights into the complex fibers-turbulence interactions that arise when non-sphericity and deformability add to fiber inertia. Flexible fibers in the dilute regime appear to undergo the same inertia-driven mechanisms that govern preferential concentration of spherical particles in bounded flows. We show that momentum coupling provides an additional bias, which may produce significant quantitative modifications in the statistics of both phases.   

The effect of the particle mass loading in particle dispersed multiphase turbulence channel flow

In particle dispersed multiphase channel flow, the interaction between particles and fluid is extremely complex. Many factors such as particle size, particle density, Stokes number and particle mass loading play assignable role in modulation of the flow statistics and coherent structure. The effect of the particle mass loading has been explored in two-way coupled direct numerical simulations of turbulent channel flow. Some fluid statistics such as stream-wise Reynolds stress show non-monotonicity with particle mass loading increasing. Two main mechanisms account for the phenomenon: More particles means stronger interaction between particles and fluid; With particle mass loading increasing, the characteristic time scale of the fluid increases, which means the Stokes number of the particles decreases , the particles become “lighter” and weaken the effect of the particles.   

Effect of Reynolds number and system size on turbulence modification in particle-laden channel flow

Turbulent particle-laden flows find many applications, starting from pneumatic transport of solid to separation of particulate from industrial exhaust. The presence of particles in particle-laden turbulent flows can augment or attenuate the turbulence level of the gas phase. Earlier studies have shown that modification of turbulence intensity varies with mass loading, Stokes number, Reynolds number and ratio of channel width to particle diameter (L/dp) [Tanaka and Eaton, PRL, 2008]. The phenomenon of turbulent modulation is complex and cannot be defined with a single parameter. Coupled LES-DEM simulations have been carried out for different volume fractions and Reynolds numbers with different L/dp ratios to investigate the turbulence modulation in the gas phase. In the present work, we have quantified the effect of L/dp and channel Reynolds number on discontinuous attenuation of turbulence intensity as a function of particle volume loading in a vertical channel flow. It is observed that the extent of attenuation decreases with decrease in L/dp ratio. Also, an increase in Reynolds number subside the phenomena of particle-induced attenuation.
  

Analysis of a Small-Scale Pulsed-Fluidized Bed

Cyclical fluidization of solid particles with a gas phase can produce structured bubbling patterns; these “pulsed-fluidized beds” have recently attracted considerable attention due to efficient mixing of particles and their tunability. Previous studies have shown that bubble patterns are controlled by factors including mean velocity at the inlet, pulsing frequency, amplitude of oscillation and bed dimensions. However, experimental investigation of the effect of bubbles on granular rheology is lacking. We present a Particle Tracking Velocimetry study of a quasi-two-dimensional pulsed-fluidized bed. Using a Proper Orthogonal Decomposition, we show that a change in bubble pattern redistributes energy among the superimposed dynamic scales of particle-phase motion. Furthermore, we explore the diffusion characteristics of particles using Lagrangian data. Finally, we highlight the predictive capability of Eulerian-Lagrangian simulations using the Discrete Element Modeling code, MFiX-DEM.   

A dynamic spectrally-enriched subgrid-scale model for preferential concentration of inertial particles in turbulence

In this study, a new subgrid-scale (SGS) model for turbulent velocity fluctuations is proposed for large-eddy simulations (LES) of multiphase flows. The modeled velocity contains scales smaller than the LES grid resolution thereby enabling, in principle, the calculation of small-scale phenomena such as particle preferential concentration and interfacial corrugation dynamics. The deterministic construction of the spectrally-enriched velocity in physical space is based on 1) the modeling of the smallest-resolved eddies of sizes comparable to the LES grid size via approximate deconvolution and 2) the reconstruction of SGS fluctuations via non-linear synthesis of small-scale turbulence. The new model does not contain tunable parameters, can be deployed in non-uniform grids, and is applicable to inhomogeneous flows subject to arbitrary boundary conditions. The performance of the model is assessed in LES of isotropic turbulence laden with inertial particles, where improved agreement with direct numerical simulation results is obtained in the dispersed-phase statistics. Application to wall-modeled LES of particle-laden channel flow will be presented.   

Ambient fluid entrainment and basal drag in turbidity currents

Turbidity current are sediment laden flows that move over inclined or horizontal surfaces. These flows are driven by buoyancy resulting from the density difference between the current carrying sediment and the clear ambient fluid above. A strong coupling between turbulence and suspended sediment particles influences the flow dynamics. Here we focus our attention on the process of ambient fluid entrainment happening at the interface between the current and the deep ambient layer. In this work, we study the dependence of the entrainment coefficient with the bulk Richardson number and the settling velocity of the sediment in the flow. We focus our study on the regime where the bulk properties of the flow vary slowly, usually called normal flow conditions. We employ direct numerical simulations (DNS) of temporally evolving, streamwise- spanwise- homogeneous turbidity currents. A new model for entrainment is proposed. Furthermore, the dependence of basal drag on Richardson number and settling velocity is analyzed. In addition, the turbulence structures of the flow and its relation to turbulence production and dissipation is studied for both subcritical and supercritical flows.   

Euler-Lagrange Simulations of Shock-Particle Cloud Interaction

Euler-Lagrange (EL) simulations of a shockwave interacting with a cloud of fixed and a cloud of movable particles at various volume fractions (10%-30%) and shock Mach numbers (1.22-3.0) are performed. Results from the particle-resolved simulations of fixed particles and experiments performed at the Multiphase Shock Tube (MST) facility at Sandia National Laboratories (SNL) are used to validate the fixed cloud and the moving cloud EL simulations respectively. In the fixed cloud simulations, the effect of the wave drag at supersonic flow Mach numbers is modeled as an inviscid drag. The mean thermodynamic and kinematic gas quantities and the forces experienced by the particles at very short time-scales are compared against the particle-resolved simulations. In case of the movable particle simulations, the effects of initial particle curtain profile on the particle cloud trajectory is studied. For this, results from a top-hat profile and a curve fit (obtained from the experimental data) are compared against the experimental results.