Implementation of a Forecast-Based Predictive Controller in a Residential Building

To support the European Union’s decarbonization pathway for buildings – phasing out fossil heating and cooling by 2040 and achieving zero-emission standards by 2050, this work presents a forecast-based model predictive control (MPC) for thermally activated building components (TAB). TAB provides large-area, low-temperature heat transfer and substantial thermal storage, enabling the controller to use environmental inputs such as solar irradiation to maintain comfort. The proposed MPC reduces complexity by embedding a lightweight, physics-based building model directly in the controller source code and avoiding heavy machine learning pipelines. A 48-hour forecast horizon with hourly updates leverages weather data (temperature, wind, cloudiness) and on-sitebuilding measurements to compute additional heating/cooling actions. Optimization is formulated first via least-squares error minimization for temperature tracking, then extended to a cost-based objective that includes dynamic tariffs, local renewables, CO2 price signals, and a season- dependent comfort band to enable short-term load shifting. Demonstrations in residential buildings in Vienna and Lower Austria show feasibility, high thermal comfort, and improved operability for non-expert users. The results indicate that a complexity-reduced MPC can deliver cost-aware, climate-aligned control with practical implementation effort, supporting broader adoption in the residential sector.