Multiple Linear Regression Model Approach for Aerosol Dispersion in Ventilated Spaces Using Computational Fluid Dynamics and Dimensional Analysis

Abstract

Aerosol dispersion in living spaces especially bioaerosols, due to accidents or deliberate acts, is of significant current interest. Computational fluid dynamics (CFD) provides an accurate and detailed platform to study the influence of different parameters on aerosol distribution in indoor spaces. The simulations however are time consuming and site-specific. The work here introduces an approach toward addressing this challenge. During emergencies, an accurate, quicker, and more general model is required to give rapid answers to first responders. Significant parameters influencing aerosol behavior in an office room were identified and through dimensional analysis, nine dimensionless groups were developed. Fractional factorial design was used to build sixteen scenarios to explore the design space. These scenarios were then simulated using a comprehensive CFD model. Large eddy simulation with the Smagorinsky subgrid scale model was applied to compute the airflow. Aerosols were modeled as a dispersed solid phase using the Lagrangian treatment. The influence of the dimensionless groups on the temporal variation of the number of aerosols in the room and the spatial distribution of the particles in the room was analyzed. The results showed that all the identified dimensionless groups were significant. Multiple linear regression models were developed for the prediction of the number of aerosols in the room and their spatial distribution as a function of the significant parameters influencing aerosol transport. The linear models accurately predicted the data on which they were based but did not predict the results of the independent tests as well. The limited predictive ability of the model showed that the relationships between the dimensionless groups are nonlinear and a higher level of experimental design will have to be applied to better explore the design space.