Multi-stage calibration of the simulation model of a school building through short-term monitoring
MetadataShow full item record
The increasing attention on the improvement of new and existing buildings' performance is emphasizing the importance of the reliability of the simulation models in predicting the complexity of the building behaviour and, consequently, in some advanced applications of building simulation, such as the optimization of the choice of different Energy Efficiency Measures (EEMs) or the adoption of model predictive control strategies. The reliability of the energy model does not depend only on the quality and details of the model itself, but also on the uncertainty related to many input values, such as the physical properties of materials and components, the information on the building management and occupation, and the boundary conditions considered for the simulation. Especially for the existing buildings, this kind of data is often missing or characterized by high uncertainty, and only very simplified behavioural models of occupancy are available. This could compromise the optimization process and undermine the potential of building simulation. In this context, the calibration of the simulation model by means of on-site monitoring is of crucial importance to increase the reliability of the predictions, and to take better decisions, even though this process can be time consuming. This work presents a multi-stage methodology to calibrate the building energy simulation by means of low-cost monitoring and short-term measurements. This approach is applied to a Primary School in the North-East of Italy, which has been monitored from December 2012 to April 2014. Four monitoring periods have been selected to calibrate different sets of variables at a time, while the validation has been carried out on two different periods. The results show that even if less than 8 weeks have been considered in the proposed calibration approach, the maximum error in the estimation of the temperature is less than +/- 0.5 in 77.3% of the timesteps in the validation period.