Abstract

This study presents a methodology to develop suitable control-oriented thermal RC network models for optimized HVAC load management in typical electrically heated single-family detached houses. Using recurring parameter identification and model-reset, the building dynamics are represented by an explicit discrete time-varying state-space formulation.
Next, these models are applied in a predictive control framework in which the objective is to enhance energy efficiency and energy flexibility of the building by prioritizing the import from the most efficient energy source(s), storing energy in the building's thermal mass and/or a battery, and shifting the HVAC load to lower the stress on the local grid.
Finally, the benefits of predictive control strategies for HVAC load management, both for the building owners and the local grid, are studied through a seasonal simulation where the performance of the building subject to a reference reactive controller and a predictive controller are compared. Applying the predictive controller, the results show an average of 12.1% reduction in the daily heating load, 19.8% reduction in the total daily import, 68.1% reduction in the peak demand, 67.0% reduction in the daily energy cost, and 13.4% increase in the self-consumption of on-site generated solar electricity for the duration of January 1st to March 31st (90 days), compared to the reactive controller.