Adaptively reconstructed spectral eddy-viscosity in large eddy simulations of particle-laden isotropic turbulence, Part II: Number density spectra and inertial range clustering

Abstract

Large eddy simulations (LES) of isotropic turbulence coupled with the Lagrangian particle tracking have been consistently disregarded as a means of exploring the physics underlying turbulent dispersed two-phase flows with a fully developed inertial subrange. In the present work, we determine the impact of our recently developed adaptively reconstructed spectral eddy-viscosity on the dynamics of small, heavy inertial particles at high Reynolds numbers. We use the particle number density spectrum to assess the ability of LES to predict particle clustering at distances exceeding the Kolmogorov length scale. We demonstrate that the functional form of the spectral eddy-viscosity has in general a moderate impact on the quantitative prediction of this phenomenon, while preserving qualitative agreement between LES and reference direct numerical simulations (DNS). By comparing the results against a state-of-the-art point-particle DNS, we demonstrate that the adaptively reconstructed closure enhances the predictive capabilities of LES for a wide range of Stokes and Reynolds numbers, providing the opportunity to explore the inertial-range clustering of dispersed particles over a broad spectrum of length scales. We point out that, assuming sufficient spatial resolution, the LES enriched with the proposed spectral eddy-viscosity becomes a reliable method for exploring the influence of large, energy-containing flow scales on the dynamics of inertial particles suspended in isotropic turbulence, particularly at Reynolds numbers that are currently unachievable in DNS. We further argue that this approach can support ongoing efforts to develop theories concerning the turbulent transport of dispersed particles.