Pseudo-turbulence models for compressible flow through random particle suspensions

Andreas Nygård Osnes, Magnus Vartdal · 2025

Read Pseudo-turbulence models for compressible flow through random particle suspensions by Andreas Nygård Osnes, Magnus Vartdal (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-05013.txt.

Introduction

Andreas Osnes and Magnus Vartdal built models for turbulent-like stress in flow through particle clouds in two thousand twenty-five. Even without true turbulence, particles stir the flow in ways that act like extra stress, called pseudo-turbulence. Models are needed to close the averaged equations for such flows.

Their main model extends earlier work for incompressible flow to compressible flow below the speed of sound. It is built from data of fully resolved simulations of particle clouds. A second, local model ties the stress near a particle to the drag force on it, for use in simulations that track particles one by one. They also give a first model for the related energy transport terms.

Catalog record

CREATORAndreas Nygård Osnes, Magnus Vartdal
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE28 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-05013

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Pseudo-turbulence models for compressible flow through random particle suspensions Andreas Nygård Osnes, Magnus Vartdal Abstract A model for the pseudo-turbulent Reynolds stress tensor in compressible flows through monodisperse particle clouds is developed based on data from particle resolved numerical simulations. This model extends previous models for the incompressible regime to subsonic bulk Mach numbers. A Lagrangian model for the local pseudo-turbulent Reynolds stress around a particle, based on the correlation of a local estimate of the Reynolds stress and the drag force on the particle, is also presented. This model can be employed in Euler-Lagrange type simulations. Additionally, corresponding models for the pseudo-turbulent transport of turbulence kinetic energy, the velocity triple-correlation in the volume averaged total energy equation, is also introduced as a first effort aimed at closing the energy equation. Journal: International Journal of Multiphase Flow † † highlights: Models for the pseudo-turbulent velocity moments in the volume-averaged momentum and energy equations are developed for finite particle volume fraction, Reynolds number and Mach number. Lagrangian forms of the pseudo-turbulent closure models are developed, which incorporate the correlation between drag and pseudo-turbulence. 1 Introduction Flows through particle suspensions are strongly inhomogeneous at the particle scale, and a bulk description of such systems is incomplete without considering the statistics of the particle scale fluctuations. The particle-scale fluctuations are referred to as pseudo-turbulent fluctuations [ 9 ] . Meso- and macroscale simulations of such flows must, in general, account for pseudo-turbulence, as it may be dynamically important. In this work, we present correlations for pseudo-turbulent Reynolds stresses and pseudo-turbulent transport of pseudo-turbulent kinetic energy, applicable for compressible flows through particle suspensions. We provide two forms of the models - a conventional Eulerian algebraic form, and a Lagrangian form where we associate the particle-induced fluid phase fluctuations with the particles. At finite Mach number, local flow contraction/expansion can cause rapid changes in the flow state, which can have a substantial effect on the interphase momentum and energy transfer, mixing characteristics, chemical reaction rates, etc. Models that capture these compressibility effects, and can be applied in meso/macroscale simulations, are crucial for development of technological applications of multiphase flows, as well as for understanding the vast number of natural phenomena in which multiphase flows play an important role. Examples include spacecraft landing [ 4 ] , jet noise suppression [ 7 ] , fuel injection systems [ 16 ] and pyroclastic currents [ 21 ] . Across engineering applications and scientific studies, simulations are performed at various scales, which necessitates the development of appropriate closure models for each scale. When the information on the exact distribution of particles is available, techniques for reconstructing the forces on individual particles [ 1 ] as well as particle scale fluid field fluctuations [ 10 , 19 ] have been developed for the incompressible regime, and similar methods may potentially be used for compressible flows. For macroscale simulations, where the exact distribution of particles is unknown or the number of particles is too large, the abovementioned techniques are inapplicable and alternative modeling approaches must be used. These latter simulations require computationally inexpensive closures, which motivates the development of algebraic or stochastic models. The most dynamically important of the unclosed terms in the macroscale fluid phase equations is the pseudo-turbulent Reynolds stress tensor, \widetilde{R}_{ij} . Mehrabadi et al. [9] presented an algebraic model for \widetilde{R}_{ij} in the incompressible regime, but for finite Mach numbers, the only algebraic model available is the relatively crude approximation of Osnes et al. [13] . An alternative approach is to introduce additional transport equations for the pseudo-turbulence, as presented for compressible flows in [ 22 , 18 ] . However, these equations introduce their own closure problems. The importance of including models for \widetilde{R}_{ij} has been demonstrated for simulations of shocked particle curtains [ 18 , 11 ] , and we expect the same to hold for other compressible particle-laden flows. The goal of the present paper is to develop algebraic models for the pseudo-turbulent Reynolds stresses that capture the effects of volume fraction, Reynolds number, and Mach number. We construct models for the mean value as well as a Lagrangian model for the variation in pseudo-turbulent Reynolds stresses that correlates with drag on individual particles. The model for the mean pseudo-turbulent Reynolds stresses reduces to the model of Mehrabadi et al. [9] in the incompressible limit. Furthermore, we present corresponding models for the pseudo-turbulent transport of pseudo-turbulent kinetic energy, i.e. a triple-velocity correlation, which appears in the total energy equation. The present model development is based on flow data from particle-resolved simulations of statistically steady flows through monodisperse, random, fixed particle beds. Particle-resolved simulation is a well-established tool for study of gas-solid flows, which has been widely used in both the compressible and incompressible regimes; see for instance [ 20 , 1 , 17 , 13 ] . 2 Problem set-up The datasets used for model development in this work were produced with particle-resolved simulations, and have previously been used to develop models for particle forces [ 12 ] . The computational domains are triply periodic boxes filled with random distributions of fixed rigid particles, occupying a mean volume fraction, \alpha_{\mathrm{p}} . The dimensions of the boxes are 30D_{\mathrm{p}}\times 10D_{\mathrm{p}}\times 10D_{\mathrm{p}} for \alpha_{\mathrm{p}}\leq 0.2 , and 15D_{\mathrm{p}}\times 10D_{\mathrm{p}}\times 10D_{\mathrm{p}} for \alpha_{\mathrm{p}}=0.3 , where D_{\mathrm{p}} is the diameter of a particle. These domain sizes are larger than those suggested by Elmestikawy et al. [6] to avoid excessive correlations of the forces on the particles. The governing equations of the fluid flow between the particles are the compressible Navier–Stokes equations. The equations are closed using a perfect gas equation of state, with an adiabatic index of 1.4, along with a power-law dependence of the viscosity, with an exponent of 0.75, and a constant Prandtl number of 0.7. The boundary conditions on the particle surfaces are isothermal, no-slip, and no-penetration conditions. The fluid flow is driven by imposing a constant acceleration in the x -direction. The resulting equations were solved using the compressible flow solver CharLES [ 3 ] . The simulations were run until a steady-state flow was achieved, at which point the particle forces and heat flux across the particle surfaces balance the imposed force and the work done by it. The mean flow through the system can be described in terms of the particle volume fraction, Reynolds number, and Mach number \alpha_{\mathrm{p}}=\frac{N_{p}\pi D_{p}^{3}}{6V}\,,\quad\mathrm{Re}_{\mathrm{p}}=\frac{\langle\rho\rangle\tilde{u}D_{\mathrm{p}}}{\langle\mu\rangle},\quad\mathrm{Ma}_{\mathrm{p}}=\frac{\tilde{u}}{\langle c\rangle}, (1) where N_{p} is the number of particles, V is the domain volume, c is the speed of sound, u the mean streamwise velocity, and \rho and \mu are the fluid density and dynamic viscosity, respectively. Here, \langle\cdot\rangle denotes the average of a quantity over the volume occupied by the fluid phase, and \tilde{\cdot}=\langle\rho\cdot\rangle/\langle\rho\rangle denotes the corresponding Favre-average. The dataset consists of 42 simulations from the range \alpha_{\mathrm{p}}\in[0.05,0.3] , \mathrm{Ma}_{\mathrm{p}}\in[0.1,0.87] and \mathrm{Re}_{\mathrm{p}}\in[30,266] . However, since the relation between the forcing function and the resulting Reynolds and Mach numbers is unknown, the resulting sampling of the parameter space is irregular. A table containing all the simulated configurations is found in [ 12 ] . Figure 1: Distribution of streamwise velocity, (a, b, c), spanwise velocity, (d, e, f), and density, (g, h, i) at \mathrm{Re}_{\mathrm{p}}\approx 100 , \mathrm{Ma}_{\mathrm{p}}\approx 0.4 for varying

Read the complete text