An inquiry into the accuracy of the energy model calibration process

A calibrated physics-based energy model is mostly assumed to be an adequate representation of the actual building when it complies with a measurement and verification standard's criteria for fitness. This paper aims at quantifying the uncertainty inherent in the building energy model (BEM) calibration process and calibrated model's predictive performance on operational decisions. In this regard, a generic building is simulated with three different occupancy and envelope scenarios using EnergyPlus. Subsequently, seven parameters are assumed unknown, and a custom optimization script is employed to search for these parameters. Results demonstrated that calibration with hourly meter data yields more accurate and stable parameter estimates than monthly data. However, accuracy is increased up to 58% by repeating the monthly calibration a few times, averaging the parameter estimates, and adopting them as final calibrated model parameters. Results indicated that the poorly calibrated models might not be sufficient to support operational decisions.