Simulations of dilute sedimenting suspensions at finite-particle Reynolds numbers

Abstract

An alternative numerical method for suspension flows with application to sedimenting suspensions at finite-particle Reynolds numbers Rep is presented. The method consists of an extended lattice-Boltzmann scheme for discretizing the locally averaged conservation equations and a Lagrangian particle tracking model for tracking the trajectories of individual particles. The method is able to capture the main features of the sedimenting suspensions with reasonable computational expenses. Experimental observations from the literature have been correctly reproduced. It is numerically demonstrated that, at finite Rep, there exists a range of domain sizes in which particle velocity fluctuation amplitudes ⟨ΔV∥, ⊥⟩ have a strong domain size dependence, and above which the fluctuation amplitudes become weakly dependent. The size range strongly relates with Rep and the particle volume fraction ϕp. Furthermore, a transition in the fluctuation amplitudes is found at Rep around 0.08. The magnitude and length scale dependence of the fluctuation amplitudes at finite Rep are well represented by introducing new fluctuation amplitude scaling functions C1, (∥, ⊥)(Rep, ϕp) and characteristic length scaling function C2(Rep, ϕp) in the correlation derived by Segre et al. from their experiments at low Rep [“Long-range correlations in sedimentation,” Phys. Rev. Lett. 79, 2574–2577 (1997)10.1103/PhysRevLett.79.2574] in the form \documentclass[12pt]{minimal}\begin{document}$\langle \Delta V_{\parallel , \perp } \rangle = \langle V_{\parallel } \rangle C_{1, ( \parallel , \perp )} ( Re_{p},\phi _{p} ) \phi _{p}^{1/3} \lbrace 1 - \text{exp} [ -L / ( C_{2} ( Re_{p}, \phi _{p} ) r_{p} \phi _{p}^{-1/3} )] \rbrace$\end{document}⟨ΔV∥,⊥⟩=⟨V∥⟩C1,(∥,⊥)(Rep,ϕp)ϕp1/3{1−exp[−L/(C2(Rep,ϕp)rpϕp−1/3)]}.

Eulerian‐Lagrangian simulations of settling and agitated dense solid‐liquid suspensions – achieving grid convergence

“…Where the settling speed is steady in a significant part of the time window of a simulation (see Figure ), the per‐particle variability in the velocity (expressed in a root‐mean‐square value) is a transient as shown in Figure . The root‐mean‐square (rms) values are—as expected—larger for the vertical velocity component than for the horizontal components (by approximately a factor of 2) . The dependency of the rms particle velocity values with respect to the width of the mapping function follow the same trend as in the (fully periodic) simulations in Ref.…”

“…We then summarize the simulation procedure and refer to the literature (e.g., Refs. ) for further details. In discussing the hindered settling results, we focus on the impact of model choices on the settling speed.…”

Simulations of scalar dispersion in fluidized solid–liquid suspensions

“…The recent review by Guazzelli and Hinch summarizes current understanding of sedimentation and hints at outstanding issues and challenges. Important topics—specifically, as here we study scalar dispersion in the continuous phase, a process brought about by random particle motion—are particle velocity fluctuations and hydrodynamic (or self) diffusion of particles and their dependency on solids volume fraction, Reynolds number, and domain size . Self‐diffusion can be thought of as the particle counterpart of scalar dispersion in the liquid phase.…”

Numerical and experimental analysis of the sedimentation of spherical colloidal suspensions under centrifugal force

“…14 Other power-law expressions have been proposed based on the Richardson-Zaki relationship 15 and implemented in sedimentation models. 16 However, for monodisperse suspensions, in which the maximum volume fraction is not one, Richardson-Zaki power-law expressions can become inaccurate by allowing unphysical volume fractions. 17 Alternatively, particle hindrance can be taken into account by using a relationship based on a suspension viscosity model.…”