Assessing the accuracy of predicted values of human CO2 generation rates for IAQ applications

The fields of building ventilation and indoor air quality (IAQ) often use indoor CO 2 concentrations as indicators of ventilation and IAQ, though many of these applications involve technical misinterpretation. Many of these applications require estimates of CO 2 emission rates from building occupants (VCO 2 ), which can be predicted from body mass, sex, age, and activity level. In many cases, these parameters are difficult to determine, necessitating assumptions about the occupants and their activities. The determination of these values can be particularly challenging during the building design phase when no occupants are present or in occupied spaces where characterizing occupants is often impractical. This study utilizes data from experiments conducted by four research laboratories employing whole-room indirect calorimeter measurements of VCO 2 to evaluate the accuracy of two estimation approaches, ASHRAE and Persily and de Jonge. The experiments involved healthy adults performing various activities, including sleeping, cycling, and sedentary tasks such as reading. This validation exercise of VCO 2 predictions was conducted using two types of input values, measured data from the experiments and data from the literature. The results indicate that the ASHRAE approach consistently underestimated VCO 2 , with absolute mean prediction errors ranging from 29% to 58%. The Persily and de Jonge approach exhibited lower prediction errors, particularly when measured inputs were used, with absolute mean differences ranging from 6% to 21%. These findings highlight the critical importance of accurate input data. When measured inputs are unavailable, literature-derived values should be used with an understanding of their uncertainty. • Measured values of human generation rates of CO 2 are compared with predictions. • The measurements are from four laboratories using whole-room indirect calorimetry. • The predictions are made using measured inputs and inputs based on literature values. • A 2017 methodology yields smaller predictive errors than the older ASHRAE approach.