Assessing the Prediction of Human CO2 Emissions for IAQ Applications

The field of building ventilation and indoor air quality (IAQ) often employs indoor CO2 concentrations as an indicator of outdoor air ventilation rates and, in some cases, as a contaminant impacting human health and comfort. Many of these applications require CO2 emission rates from building occupants (VCO2), which can be predicted based on occupant characteristics (e.g., body mass, sex, age) and activity level (e.g., sleeping, exercise, resting). In some applications, this information is fairly well known. However, in many other cases, the occupant characteristics and activity levels that impact the values of VCO2 are difficult to know with much precision, thereby requiring assumptions about the occupants and their activity levels. The ability to use literature-based values for the required input is particularly important during the building design phase when there are no actual occupants to characterize or when considering occupied spaces where it is not practical to characterize the occupants. Whether these inputs are known or not, it is important to characterize the predictive accuracy of calculated VCO2 values to aid in interpreting indoor CO2 concentrations.  
This study leverages data from whole-room indirect calorimeter chamber measurements of VCO2 to evaluate the accuracy of two VCO2 estimation approaches—ASHRAE Fundamentals Handbook (2021) and Persily and De Jonge (2017). The chamber experiments involved 50 healthy, non-smoking individuals aged 20 to 64 years, engaged in activities such as sleeping, stationary cycling, and sitting (performing sedentary tasks like reading or watching television). Metabolic parameters such as VCO2, rate of oxygen consumption (VO2), basal metabolic rate (BMR), respiratory quotient (RQ), and energy expenditure (EE) were collected during these activities. The validation exercise is performed using two types of input values to estimate VCO2— measured data from the chamber experiments and data from the literature. The results indicate that whether using either type of input values, the Persily and de Jonge (PdJ) predictive values are closer to the measured VCO2 than the values calculated using the ASHRAE approach. PdJ predictions exhibit an absolute mean error of 6 % when using measured inputs, smaller than the ASHRAE predictive error of 28 %. When utilizing literature inputs, the mean predictive error of PdJ is 19 %, comparable to 28 % for the ASHRAE approach.