Data-Driven Models Applied to Heating System Control Design Based on Weather Forecast

Primary energy consumption for domestic purposes represents the 46% of overall energy consumption in Italy1 and thus reducing inefficiencies and waste in this field may have important and beneficial effects on a global scale. Since 1997 when the Kyoto protocol was approved by 169 nations committing to reduce greenhouse gas emissions to address climate change, many steps have been made to promote sustainable architecture. The aim of this project is to implement a boiler control system, regulating with respect to the outdoor temperature (and not to the indoor one), which is cheap, efficient and easy to use. This control system allows boilers to decide when to come on and off and how long for depending on outdoor temperature and considering the thermal capacity of the building. This means creating a control program capable of predicting, using weather forecasts, the outlet temperature profile of the water taking into consideration the current temperature of the building (and so that the building needs to be heated just for the difference between the indoor and the comfort temperatures). In order to present the methodology we used to obtain this control system, we wrote a program for the specific case of an existing building (built in 1939) in Turin. Obviously are of that time also the criteria the heating system was built according to, and so they were not studied for energy savings. In order to improve the efficiency of the heating system the boiler was changed in 2010 before the heating season, but we wanted to see if it was possible to save even more gas by implementing the control system described before. To implement this control system it was decided not to use any thermodynamic equations, but black box correlations only. We thought that this approach would be more effective, albeit less general, for the specific building.