Challenges with the calibration of building energy simulation models

Abstract

In order to correctly understand and predict the energy use of a building, simulation tools is often utilized in the industry. These simulation tools can predict the estimated energy use in buildings using inserted parameters; however, studies shows that these estimates often differs from the actual energy use. To adjust these deviations, calibration of the parameters is a well proven technique which corrects the simulated energy use.In this report an office building is used to analyze and identify challenges with the calibration of building energy simulation models. The building was modeled in IDA-ICE, and energy simulations where performed. Detailed data of the building have been collected to enable a realistic model. Full year energy simulations of the building was performed and compared to the measured energy use. The calibration was performed in three parts: the first calibration was done using building defaults, then a calibration using equipment settings, and finally a calibration using operation settings.An initial simulation was performed to identify the differences in measured and simulated energy use. The energy use was compared in four subgroups, light and appliances, pumps and fans, heating, and cooling. The initial simulation showed discrepancy in all predefined groupings. After the calibration three of the four subgroups achieved a result with acceptable similarity to the measured energy use. The difference in total yearly energy use between measured and simulated was reduced from 17% to 11%. The report failed to calibrate the cooling energy use. Reviewing the measured and simulated energy use with the cooling energy use excluded, yields a difference below 0,5% with a full year simulation. Several challenges were identified during the calibration process. Obtaining precise and thorough data on the building used in the calibration is vital, however this can be challenging, as there are usually many parties involved in constructing a building. The influence one system has with other systems also results in challenges when adjusting parameters. A seemingly calibrated model can achieve approximately similar energy use as the measured energy use, but can still give huge discrepancies when comparing the result in a shorter time resolution. This challenge is apparent in the results produced in this report, and is a challenge when calibrating building energy simulation models.