Statistical approach to turbulent dispersal of aerosols for accurate prediction of concentration and associated uncertainties

Aerosol dispersal in indoor spaces is a two-point, two-time stochastic process, defined by a mean behavior and uncertainties from turbulent fluctuations and spatial inhomogeneity. While existing models capture the mean accurately, few address the associated variances. We develop a theoretical model to quantify the turbulence-related uncertainty. Euler--Lagrange large-eddy simulations (LES) coupled with a novel statistical overloading technique are employed to construct a data-driven model capable of predicting the mean aerosol concentration and the total variance. The proposed model is compared against experimental data.