URANS and DNS of a cough-induced aerosol-laden jet flow interacting with a large-scale circulation

Abstract

This study investigates the dispersion of a cough-induced particle cloud in the presence of a large-scale circulation (LSC) predicted in an unsteady Reynolds-averaged Navier-Stokes simulation (URANS), and compares the results with those of a direct numerical simulation (DNS). Both simulations are performed for 22 s and share identical initial conditions and domain sizes, but they differ in spatial and temporal resolution and in their approach to turbulence. Immediately after the jet phase initiated by the cough, it is found that the particles in the URANS have not advanced as far in the horizontal direction as those in the DNS. In addition, the shape of the URANS particle cloud is more symmetric in the horizontal inlet midplane, whereas in the DNS the particles appear to be quite irregularly distributed. In the well-developed puff phase, the particles of URANS have moved less far in horizontal direction than those of DNS. Although the lateral dispersion of the URANS particles is similar to that of the DNS particles, the top view shows a cone-shaped pattern for the URANS particles, whereas those of the DNS have a uniform distribution in the shape of a rectangle. The major difference between the URANS and DNS results in the late puff phase is the thicker and stronger boundary layer near the upper wall of the URANS, which significantly impedes the vertical ascent of the particles, resulting in a stronger overall circulation intensity throughout the domain. Additionally, the URANS and DNS predictions of the vertical probability density distribution (PDF) compare well in the jet phase, but the URANS underestimates the spreading range in the well-developed and late puff phases.