In this study, the framework for a transferable model-based predictive controlled shading solution is laid out. The analysis began with a numerical investigation into thermal model training using a Bayesian approach — namely the Ensemble Kalman Filter — for calibrating a low-order control model of the space. The trained model had its effectiveness demonstrated and was successfully utilized within the EnergyPlus environment to control the shades of single zone office and provide total electricity savings of 35% in a complete building automation system.
Later, these methodologies were
adapted and utilized in a demonstrative setting built within a research facility to attempt and identify the challenges associated with the scaling of the approach. The results showed an environment which effectively managed occupant needs both visually and thermally and which ultimately was found to save energy in comparison the previously existing system in the building.