Building Energy Model Calibration as a Service

This presentation explores the use of building energy model (BEM) together with machine learning calibration using actual building data to create mini tuned energy models of existing buildings. BEMs are widely used to evaluate energy improvements in commercial buildings, yet their accuracy is often constrained by data availability and model uncertainty in operation. To address this, a flexible calibration workflow was developed, integrating Monte Carlo sampling, surrogate modelling, and Bayesian optimisation.