Reduced-order modeling of transport of infectious aerosols in ventilated rooms

Abstract

A new approach to numerical modeling of airborne transmission of respiratory infections, such as COVID-19, influenza, or those caused by common rhinoviruses, is presented. The focus is on the long-range transport of infectious aerosol particles by air flows in indoor environments. The approach is based on the Eulerian description of the aerosol field and the reduced-order modeling (ROM) applied to reduce the computational cost of analysis. The ROM is based on the projection of a computational fluid dynamics (CFD) solution onto a Krylov subspace by an Arnoldi-type algorithm. The algorithm does not require access to the original discretization matrix and, therefore, can be applied to solutions of Eulerian transport problems by general-purpose CFD software, in which such a matrix is often unavailable. The model is validated for a realistic setting via direct comparison of its predictions with the results of the full-order CFD solution based on the Eulerian model and the data of Lagrangian tracking of aerosol particles. Applicability of the ROM to simulation of long-term evolution of the aerosol field and to assessment of infection hazard is demonstrated. Computational tests show that use of ROM reduces the computational cost of analysis by a factor of about 103 without a significant loss in the accuracy of the results.