Data-driven Methods for Reliable Energy Performance Characterisation of Occupied Buildings

Abstract

Global and increased awareness of climate changes are apparent, and within the EU, strengthened energy policies are put in force to reduce greenhouse gas emissions. By accounting for 40 % of the total energy end-use, buildings are the single most energy-intensive consumer, and the residential sector alone, accounts for 25 % with space heating as the dominant share. With the current EU Energy Performance in Buildings Directive (EPBD), each member state is instructed to establish a national building renovation plan and conduct an energy performance screening of the building stock. The aim is to increase the building renovation rate and the effect of the renovations. Despite all the initiatives, there exists a well documented and rather large discrepancy between anticipated and actual energy consumption in buildings. Furthermore, the gap can hardly be quantified by evaluating the energy consumption alone. The reason is that occupants, weather, and the building quality affect it. The focus of this PhD thesis is therefore to develop methods to quantify the performance of buildings and to separate the reasons for the energy use. Earlier research has shown promising results in terms of identifying thermal building performance characteristics based on data-driven quasi-stationary and dynamical mathematical models. A special focus has been on the transition of modelling techniques applied on unoccupied and thermally controlled test buildings, to reliable modelling techniques applied on occupied buildings. It has been shown inPaper Cthat methods relying on only heat consumption measurements and weather data are capable of quantifying the heat loss coefficient, solar transmittance, influence of wind, transition period etc. Indications of the occupants’ effect on energy use have been estimated as well. Paper D showed that time constants and the building’s energy flexibility (i.e. energy demand-shifting capabilities) can be obtained from measurements of only the indoor and outdoor temperature. For more detailed dynamical grey-box modelling techniques, a novel approach to sun position dependent solar gain estimation has been proposed in Paper B. The thesis discusses the importance of handling the disturbances caused by the occupants’ interaction with the building. In Paper A, a method for estimating the occupancy status in dwellings has been proposed, with the intention of describing model noise in grey-box models in a more detailed manner. The work of this thesis outlines new scalable approaches for documentation and screening of thermal building performance and energy flexibility. New modelling techniques have been presented and discussed in order to increase the reliability of data-driven models. The combination of the intensified energy data collection within the EU and reliable methods for thermal building performance assessments is believed to bring us closer to targeted building renovation strategies, and evidence-based energy performance documentation of buildings.