Energy characterization of a PASLINK test cell with a gravel covered roof using a novel methodology: Sensitivity analysis and Bayesian calibration

The aim of this study is to accurately simulate the thermal performance of a PASLINK test cell with a gravel covered roof test component using a methodology that combines dynamic simulation and mathematical techniques. The main contribution of this work is a new method that improves the characterization of building models used in simulation software for the evaluation of energy demands and savings. The methodology consists on TRNSYS dynamic simulation, MOAT (Morris one-at-a-time) and BACCO (Bayesian analysis of computer code output) sensitivity analysis and a Bayesian calibration. These techniques are applied to a test cell with high quality monitoring data that validate the simulation performance. The results demonstrate the benefits of these techniques in reducing the coefficient of variation of the root mean square error, CV(RMSE), between the simulated internal temperature and the real temperature to a value less than 3.38% under different load conditions. Therefore, the calibrated models obtained with this method can considerably reduce the gap between the simulation and real performance of buildings.