A Framework for Digital and Sustainable Classroom Management: Non-Intrusive CO2-Based Occupancy Estimation for Demand-Controlled Ventilation

Classrooms often present low Indoor Air Quality (IAQ) conditions, including in mechanically ventilated educational buildings where Constant Air Volume (CAV) HVAC systems operate on fixed schedules that do not account for real-time occupancy. This rigidity creates a critical trade-off: inadequate IAQ during peak usage and unnecessary energy consumption during low-occupancy periods. Addressing this issue requires accurate, real-time occupancy data, which is often unavailable due to cost or privacy constraints. This study analyses and validates a non-intrusive method for estimating real-time occupant counts using CO2 concentration data. Measurements were collected in two adjacent classrooms with similar HVAC configurations during a long-term monitoring campaign in which indoor CO2 levels were continuously recorded. A CO2 mass-balance model was developed to incorporate sensor data and the constant air exchange rate of the HVAC system, enabling dynamic estimation of occupant-generated CO2 and, consequently, occupant count. Model outputs were validated against scheduled occupancy data obtained from surveys. Results indicate that the method predicts occupancy patterns with an accuracy of approximately ±5%. With appropriate sensor placement, spatial occupancy distribution can also be inferred. Continuous monitoring revealed frequent CO2 peaks above recommended thresholds during teaching hours, underscoring the limitations of static ventilation strategies in educational environments. The CO2-based estimation model effectively captured these events, offering a cost-efficient and privacy-preserving proxy for real-time occupancy.