Improvement of a particle-eddy interaction model for predicting the dispersion of breath-induced particles

Abstract

Particle dispersion models are essential to capture the influence of unresolved turbulent flow structures in computational fluid dynamics (CFD) simulations, but their validation remains challenging. In our previous study we developed the so-called limited particle-eddy interaction (LPI) model using time-averaged velocity fields from Direct Numerical Simulation (DNS) and demonstrated that it is able to mitigate the breath-induced particle overdispersion predicted by the mean particle interaction (MPI) model implemented in OpenFOAM®. In the present study the performance of the LPI model is further improved by introducing a relaxation factor Cr on the limiter of the eddy interaction distance to increase the particle-eddy interaction time and thus improve its accuracy. Our results show that for the smaller particles (≤ 8µm) selecting Cr values of 16 and 4 yields the best particle dispersion predictions for the Convex Hull Volume (CHV) and D2 methods, respectively, while for the larger particles (≥ 16µm), the MPI model performs comparably to the LPI models.