RANS and LES of Particle Dispersion in Turbulent Pipe Flow: Simulations Versus Experimental Results

Abstract

Turbulent transport and dispersion of inertial particles in fully-developed turbulent vertical pipe flow has been investigated (Reτ = 2,200, based on the friction velocity and the pipe diameter) using two approaches: large-eddy simulation (LES) and Reynolds-averaged Navier-Stokes (RANS) both employing Lagrangian tracking of a dilute suspension of particles (glass beads in air with different Stokes’ numbers, namely 0.022 and 2.8). Detailed numerical simulations are performed in order to: (a) assess the capabilities of these two approaches to match the experimental measurements of Arnason and Stock [1, 2]; and (b) validate the extension of the stochastic approach based on Langevin modeling used in a RANS framework to the generation of sub grid-scale fluctuating velocities as seen by solid particles in LES. Results for the particle dispersion coefficient and preferential distribution of particles in different sections of the vertical pipe as well as streamwise and radial particle velocities, are computed and compared to the results of the experimental measurements. The following conclusions are drawn. (a) Both RANS and LES, using stochastic modeling for the fluid velocity, are seen to predict reasonably well the dispersion of solid particles with different Stokes’ numbers in a high Reynolds number, nonhomogeneous and anisotropic turbulent flow. (b) The extension of stochastic modeling based on the Langevin equation to the construction of the subgrid-scale fluctuating velocity field as seen by the particles is successful; it contributes to the better results obtained, compared to RANS results, especially for those predicted for the small particles. (c) As shown in experimental results [1, 2] and demonstrated by theoretical studies [3, 4], the numerical predictions supported the conclusions that large inertia particles can disperse faster than small inertia particles, depending on the combined effects of inertia and drift parameters.