Evaluation of Uncertainties of Using CO2 for Studying Ventilation Performance and Indoor Airborne Contaminant Transmissions

The COVID-19 pandemic has raised concerns about indoor ventilation conditions worldwide. Monitoring CO2 concentrations in rooms has been widely used, but its relationship with outdoor air ventilation rates and ventilation performance is uncertain. Several uncertainties must be quantified, including the location and rate of CO2 sources, sensor locations, and the dynamics of the surroundings, as well as limitations of existing simulation models, such as well-mixing assumptions. This paper presents field measurements, stochastic modeling, calibrations, and aerodynamics analysis within rooms and contaminant dispersal. Several CO2 tracer gas tests were conducted in classrooms. Two test setups were used, one for uniformity testing and the other for evaluating ventilation performance. A proposed uniformity index (Ui) is integrated into the tracer decay method to address its limitation due to the well-mixing assumption, thereby improving the air change rate estimation by 22%. As a general rule, the outlet sampling location may represent the average of all locations in mixed-ventilated spaces. Given the small difference in peak CO2 concentrations (2.6%) and decay periods (15%), 60% of the ventilation capacity should be used instead of the full capacity. As opposed to the instructor's location, the room midpoint yields a 7 percent higher peak CO2 concentration, which is recommended as a dosing source to estimate air change rates using the tracer decay method. Additionally, novel simulation models have been developed for estimating ventilation air change rates in indoor environments since deterministic approaches cannot incorporate system uncertainties. It has been found that stochastic models, which combine the physical principles of a system with data collected from field measurements, are effective for resolving uncertainties, but they have not been extensively explored in terms of estimating air change rates. Therefore, we also examined the integration of stochastic differential equations (SDEs) and a Bayesian calibration model to evaluate indoor air quality and ventilation conditions in rooms.