Procedure for modelling and calibrating operating thermal systems in buildings

Abstract

Before applying energy improvement strategies to real energy supply facilities, it is necessary to have an accurate model of their dynamic energy behaviour in order to predict and optimize the results. However, contrary to experimental facilities, in real operating facilities, budget constraints and fast access to recorded data limit the number of installed sensors to the minimum required for control and maintenance purposes. The aim of this work is to validate an energy model based on a real thermal facility with the ASHRAE validation criteria. The database analysis reveals that the recorded data are limited and do not have the same sampling interval, varying from 15 min to 24 h. Then, the facility's energy consumption is simulated for typical days in TRNSYS and a novel application combining TRNSYS and Matlab software is used to iteratively calibrate the model. The calibration process applied reduces the uncertainty of the model from the initial values of 26.60 ± 0.78% and 60.76 ± 19.53% for the statistical indices, mean bias error (MBE) and coefficient of variation of the root mean square error (CV(RMSEP)), to the final values −1.89 ± 0.61% and 55.92 ± 17.89% respectively. The methodology developed in this work overcomes the difficulties derived from the limited data in the model design and calibration processes, resulting in a procedure that balances the little available building information and the computational effort, with satisfactory results. Therefore, it could be applicable in models with also high uncertainty and any facilities configuration.