Simulation-based occupancy estimation in office buildings using CO2 sensors

Occupancy proles are required to estimate internal gains in buildings for a model predictive controller. Literature discusses many approaches but the practical usability of these approaches is unclear. A simple dynamic model using only existing sensors from a real oce building with 23 zones was tested. A validated Modelica air flow model that computes the mass flow rates required for the occupancy estimation algorithm is also demonstrated. The variable air volume model is published open-source. Limitations of automatic baseline calibration were identifed and an alternative CO2 sensor calibration approach is presented. The occupancy estimation algorithm is validated using measurement data.