Scalable and adaptive multi-stage hourly calibration of simulation models with high usage fluctuations : case studies for multi-purpose buildings

Accurate and computationally efficient building performance simulation models are necessary to support digital twinning and management of energy systems. Motivated by the perceived need to increase understanding on the scalable and adaptable calibration of energy simulation models, particularly in terms of multi-purpose buildings, this study implements a multi-stage hourly calibration scheme that integrates a trigger-based recalibration mechanism for real-time applications. To verify the performance of the proposed scheme with respect to scalability and adaptability, two case studies with high usage fluctuations in Finland and Norway are investigated. In both cases, all objective estimations satisfied the standard hourly accuracy criteria (e.g. CVRMSE < 30%). In Case_Finland, calibrating variables improved hourly CVRMSE for all objectives by up to 77% over the annual-calibrated model and 88% over the baseline, while in Case_Norway, improvements reached 16% and 22%, respectively. When exposed to a real-world disruption test, the adaptation mechanism enhanced model accuracy by 56%.