Further development of the change-point model – Differentiating thermal power characteristics for a residential district in a cold climate.

• This paper presents a unique change-point model based on selected time periods. • Time-dependent variations in climate and user behavior are central in the selection of time periods. • The model allows for differentiation between thermal power characteristics. • The selected time period for prediction of Q total corresponds to the highest R2 value, 0.70. The building and service sector accounts for nearly 40% of total energy use in Sweden. The existing, historic building stock accounts for large part of this energy use and comprises an important part of the national pursuit to increase energy efficiency. One main problem for decreasing energy use in the existing building stock is the lack of data describing thermal performance characteristics. This paper presents a novel development of the change-point model for predicting the thermal performance of buildings using selected time periods based on time-dependent variations in climate and user behavior. The predicted thermal power characteristics include total specific heat losses (Q total), energy use for hot water circulation (HWC) and hot tap water (HTW), and balance temperature. A residential district with 73 historic buildings in Linköping, Sweden, has been used as the study object. The developed model is shown to be effective and robust for describing building thermal performance. The average R2 was 0.70 for predictions of specific heat losses. The sensitivity analyses conclude that the selected time steps and months correspond to the highest R2 value. The average variation width for prediction of the balance temperature is 0.9 °C for buildings in the interquartile range based on a three-year comparison of hourly heating power supply data. Moreover, from a property owner perspective, the model is shown to be useful for identifying deviating thermal power characteristics and can easily be used to get an overview of a district. [ABSTRACT FROM AUTHOR]