Uncertainty Quantification for Infection Risk Assessment of Different Passenger Compartment Ventilation Concepts

Abstract

Infection risk from airborne transmission is investigated using a forward risk prediction model in two different ventilation concepts for the passenger compartment of a long-distance train. By performing quasi-Monte Carlo simulations, where selected model parameters are sampled from distributions, systematic evaluation of the concepts over a whole range of plausible inputs is possible. This modifies the risk prediction from a singular risk value to a distribution of predicted risks corresponding to the set of input distributions. Using global variance-based sensitivity analysis via polynomial chaos expansion then extracts information on the relative importance of the separate inputs on the overall risk. Detailed knowledge of the risk distribution allows for a more granular assessment of the potential impact of the ventilation concept on the average per-passenger risk via percentile-based improvement metrics.