Fast Prediction of Comprehensive Human Infection Risk Using Computational Fluid Dynamics and Machine Learning

HYEONJUN LEE, Donghyun Rim, Pennsylvania State University

     Abstract Number: 770
     Working Group: Aerosol Exposure

Abstract
Previous studies performed Computational Fluid Dynamics (CFD) to examine local infection risk within occupied spaces. Although these studies provided valuable insights, it has been challenging to predict the infection risks considering all possible cases such as building operating conditions, student density, number of infectors and location, and the airflow pattern, resulting in increased computational time and cost. This study presents a framework to assess infection risk in a learning environment using CFD and Artificial Intelligence (AI). The objectives of the study are to (1) predict comprehensive risk assessments with multiple parameters by creating an AI model based on CFD results; (2) evaluate the effect of occupant configuration on infection risk.

The study investigates a classroom with seven occupants by varying parameters, including ventilation modes, air change rates, occupant arrangements source locations, source strengths, and particle sizes. A total of 448 scenarios will be trained using an Artificial Neural Network (ANN) and Long Short-Term Memory (LSTM) model to predict infection risk.

The results reveal that a space with displacement ventilation at a minimum ventilation rate require a number of inhalable 10 μm particles twice as mixing ventilation, and four times for 1 μm particles. It is also shown that, for both ventilation strategies beyond a certain ventilation rate, the number of inhalable particles increased for occupants in a back row implying that increasing the ventilation rate does not necessarily guarantee preventing spread of infectious particles. Based on the results generated by CFD, the AI model could predict the particle emitted number with less than 10% of error. The results can be used to inform HVAC system design and operation strategies to reduce the risk of infection in indoor environments.