Predicting the CO2 levels in buildings using deterministic and identified models

Abstract There has been extensive research in the field of building modelling for adaptive and, increasingly, for predictive control. The inclusion of predictive control in building energy management and control systems shifts the calculation of optimum control strategies to the future and allows the prompt evaluation of various control scenarios. The energy saving potential of CO 2 based demand controlled ventilation is well established. Although several studies have investigated the potential of incorporating CO 2 in building energy and indoor air quality management, in most cases the applications are reactive rather than predictive. Incorporating CO 2 concentration as a factor in predictive models may unlock further optimisation opportunities in controller applications, especially in buildings with highly varied occupancy, such as institutional buildings. The objective of this study is to investigate the potential of creating predictive models tailored to specific spaces in order to estimate future CO 2 concentrations. For this purpose, both measured and simulated data will be combined. The simulated data will be provided by a detailed building model, while the final identified model of CO 2 concentration will be a state space representation derived from a prediction error minimisation method. The results suggest that there is indeed potential for at least short term prediction using a very simple identification procedure.